Final Report Summary - SICTEC (Single Cell Technologies for SMEs)
Executive Summary:
SICTEC focussed on the development of technologies for separation and manipulation of single biological cells for life science research and medical applications. The single cell manipulation technology (SCM technology) developed within the a former EC-funded project (PASCA, Platform for Advanced Single Cell Manipulation and Analysis) bases on inkjet-like printing of single biological cells confined in free flying micro droplets. It constitutes a universal platform for single cell analysis with proven potential for many life science applications. The objective of this project is to support the participating SMEs and companies to take up the SCM technology, to realize their own applications and to develop them into innovative products for the medical, biomedical and pharmaceutical markets. Central element of the research is the SCM prototype instrument and the single cell dispenser as presented in numerous publications.
Based on the technology, a new spinoff (Cytena GmbH, Germany) was established.
Project Context and Objectives:
The SICTEC project aimed to provide validation and extension of the use of the developed prototype instrument and to deal with necessary improvements and modifications of pre-production prototypes towards the specific needs and applications of the SME partners. In particular also topics affecting commercial exploitation like e.g. application development, design for manufacturability, reliability issues, cost and throughput optimization, extension of technical specifications, and preparation of CE IVD labeling are being investigated. Applications targeted by the individual SMEs are considering fast pathogen detection for clinical use by combining the SCM technology with MALDI TOF mass spectroscopy of single bacteria, instruments for single cell cancer and stem cell research, methods for monoclonal cell line development and other in-vitro medical diagnostic applications. The anticipated innovations stemming from this research will be exploited by the involved SMEs individually as well as jointly through several innovative products targeted for different markets and applications.
Project Results:
The core technology development (WP1) included that the specifications for the dispenser and detection unit needed to be elaborated, taking into account the different needs from the different applications and users. Based on that, the design and implementation of different print heads needed to be started: On the one hand for a print head which is capable of dispensing bacteria which has a lot smaller diameter (~ 1 µm) than all objects printed so far (~ 10 µm), on the other hand on a print head which apart from the standard optical detection for cells includes a fluorescence detection option. Furthermore, a better chip supply chain was about to be initially developed with more optimizations and yield increases to come in RP2. The final goal here is to have an industry-scale fabrication process for the disposables, including a quality control mechanism. Furthermore, this includes the design and fabrication of dispensing chips which are able to dispense smaller droplets 10 times smaller than the current chips (200 pl → 20 pl) to cope with the limitations which might arise with bacteria dispensing. Finally, the cell detection software is enhanced so that geometrical properties, number and statistical distribution of cells can be monitored to provide cytometry data of the liquid sample and to be able to extract further cell parameters such as roundness. A data-management software is to be developed which manages all this collected data. For the SCM instrument for cell research (WP2), the works performed in the reporting period includes finding and fixing the specifications of the single cell printer (SCP) and the initial and detailed design of such a machine. Based on the design, orders will need to be placed at different suppliers to be able to assemble the device in the next reporting period. In parallel to the design, the documentation (CAD drawings, electrical sketches, CE documentation, manual ...) will be started with more to come with the instrument fabrication and characterization. Last but not least, initial development of the control software will be conducted. For the MALDI MS for pathogen detection (WP3), also specifications need to be established and first spectral responses of lice cycle phases are about to be extracted. For that, bacterial colonies where cell populations are all in the same life cycle phase will need to be grown and their spectra are recorded using a “traditional” MALDI TOF instrument. Based on that, “biological noise” calibration curves to correct for difference in life cycle phase of bacteria in clinical samples will be created. Furthermore, the test panel for the Antibiotic Susceptibility Test needs to be selected and first designs of an multiplexed “just-in-time” mixing will be discussed. Also for the IVD handling platform (WP4), the specifications need to be defined, in parallel the performance of and existing platform is characterized and optimized. Based on these findings and the specs, the design of an IVD certifiable liquid handling device is started while keeping cost reduction and optimization in mind. During the design phase, but also while the device is assembled, the documentation for this machine is preliminary written.Within WP5 (monoclonal cell culture), the clonal cell line production needs to be optimized. For that, an existing single cell handling machine (SCM) from the PASCA project will be used and will be checked for improvements which arise especially from WP1: Higher throughput through the use of better chips and better detection software, improved sorting capabilities, or higher viability of the cells. WP6 (Automated single cell patch clamping) starts by setting up some preliminary specification based on which the stroboscopic imaging module for detection of the droplets in flight will be improved. Together with optimizations on the dispenser module which should consist of the cell dispenser and a top-view camera which can be used to find fiducial marks on the targets to be dispensed on, software optimisations will be initiated which will then allow automatically setup the dispenser and finding the right dispensing parameters without user interaction. Initial work will be performed to enhance the control software to allow it to be remote controlled without a graphical user interface, so that in the end, the user only utilized one single software: The SOPHION machine control software.The validation work package (WP7) aims in acquiring pilot partners during both reporting period and to validate their applications for feasibility with the single cell dispensing technology. For this, various ways to get in touch with such will be elaborated, be it the website, fairs or application workshops which will be held with a set of potential interested persons. WP8 (Exploitation, IP Management and Dissemination) has the task to setup and maintain the project website and to address IP management related questions. Furthermore, and in strong link with WP7 exhibitions and publications are planned, conducted or launched. Apart from that, updating the plan for the use and dissemination of the foreground is a repetitive task of the work package. Project management (WP9) finally deals with all administrative issues within the project, be it the technical coordination, submission of deliverables and reports or to hold internal meetings. This also includes setting up the internal communications structure e.g. by use of a cloud-based internal website. The technical specifications of the Chip Supply Chain are determined. Focus was put on fabrication scale-up, cost reduction and improved handling of cartridges. To gather the specifications all relevant partners were sent a questionnaire which afterwards were compared and a common base was established. This common base was finally sent around to the partners for final agreement. The resulting final technical specifications for dispenser chips and cartridge fabrication are listed in Table 1.
Table 1: Technical Specifications of the Chip Supply Chain
Dispensing chip supply chain development - Scope of the work is on new design of the dispenser unit and cartridge fabrication. On the one hand the changes of the P9 Dispenser and on the other hand the new design of the mechanical cartridge attachment. Cartridge fabrication is ramped up. Detailed failure analysis and shelf life study is performed. In the course of the changed P9 Dispenser package, develop and provided by BioFluidix, and for the new design of the cartridge attachment new and adapted parts have to be constructed, fabricated and assembled. The new assembled P9 dispenser is fully compatible to the old versions and demonstrates reliable and robust performance. The new P9 is now protected against corrosion, which was a known problem in first version when being operated with salt containing liquids. New material compositions are used and the piston was miniaturized. Housing and electrical safety was reworked to meet industrial standards. In Figure 1 the old and the new P9 dispenser designs are illustrated.
Figure 1 (ALU): Former P9 design (left) and new, improved P9 next generation (right).
The new mechanical Cartridge attachment is redesigned towards a clip mechanism, instead of screws (cf. Figure 2). The pneumatic shutter system is reworked towards more space-saving. Alignment structures are geometrically redesigned and the clip mechanism is positioned on top and on bottom. All the specifications are also in WP 1.1 scheduler listed. These changes allow a better handling of the system. P9 Dispenser and piston guide now build a combined unit. The material is changed to steel or brass, which increases the abrasion resistance of the piston guide. The clip mechanism allows for controlled positioning of the cartridge and at the same time features ease-of-handling. Rapid prototyping methods like micro-milling and 3D-printing were used to perform first design and material studies on various clip designs and piston guides. All designs are compatible to future injections molding fabrication processes. The reworked sample reservoir has the same dimensions as a 200 µl pipette tip and gives the opportunity to increase the sample volume from former 80 to now 200 µl. Figure 2 exemplarily shows one evaluated design of the new clip mechanism compared to the old design using screws.
Figure 2 (ALU): (Left) Design drawing of the old Cartridge attachment. (Right) Design drawing of the mechanical cartridge attachment (clip mechanism) at the piston guide and chip holder.
Figure 3 shows photographs of the new clip cartridges made of different materials (left) and the brass piston guide (right) that are all fabricated by rapid prototyping methods. Multiple design iterations had been necessary to optimize geometry and tolerances.
Figure 3 (ALU): New designed cartridges fabricated as micro-milling and 3D-printing parts in different materials (left, green). 3D-printed piston guide from brass in front and rear view (right).
Figure 4 shows a photograph of the first prototype with new clip design cartridge and new piston guide mounted on the second generation P9 NG.
Figure 4 (ALU): New designed dispenser unit existing from P9 dispenser, piston guide and cartridge.
First 3D-printed models of the clip design already demonstrated promising mechanical stability and clipping performance. However further design adaptions have to be done to optimize clipping performance and evaluate influence of production tolerances. During preliminary printing experiments it turned out that the new designed parts did not perform as well as the old design in some respects: Clipping the cartridge to the piston guide needs practice and it was observed that during clipping the piston scratches the silicon chip, which can influence the dispensing performance and thus must be avoided. Additionally, the lower clip breaks from time to time, due to poor stability. Furthermore, the filling of the chip is more difficult and the reservoir agitation system doesn’t perform well with this reservoir shape. Therefore a new design was developed and tested. The second design is also based on a clip mechanism. It now features four geometrically identical clips, which allow clipping from the top and avoid that during clipping the piston scratches the silicon chip. The reservoir is formed like the old one, so that agitation should work. The new cartridge design is also compatible with future fabrication by injection molding. The piston guide was redesigned for the four clip mechanism. Furthermore, the inner region of the piston guide will be produced from a non-metallic material to prevent abrasion between the piston and the guiding structure. Alignment structures, pneumatic shutter system, and attachment to the P9 dispenser are the same as for the first design. Different configurations of the clips are tested to optimize the clipping performance and stability of the clips. Design drawings of the new and final designed parts are shown in figure 5.
Figure 5 (ALU): Design drawing of the second design. (Left) Piston guide assembled from two different materials. (Right) Mechanical cartridge attachment (clip mechanism) at the piston guide and chip holder.
Figure 6 shows photographs of the new clip cartridge made of prime gray (left) and the brass piston guide with the plastic inlet (right) that were all fabricated by rapid prototyping methods.
Figure 6 (ALU): (A) New designed cartridges fabricated as 3D-printing part in prime gray. (B and C) 3D-printed piston guide from brass in front (B) and back (C) view.
Figure 7 shows a photograph of the second prototype with the new clip design cartridge and the new piston guide mounted to the second generation P9 NG.
Figure 7 (ALU): New designed dispenser unit existing from P9 dispenser, piston guide and cartridge.
These 3D-printed models of the second generation clip design demonstrated promising mechanical stability and clipping performance. Also, the first experiments to test the dispensing performance, handling, and agitation system showed good results. Clipping of the cartridge to the piston guide works very well and handling is much easier than it was with the first generation clip design. Filling of the sample and the use of the agitation system show the same good performance as the old design. The preferred future material of the cartridge is the polymer polypropylene (PP). It is superior for injection molding and can be sterilized by autoclaving, which is a convenient and standardized process. The 3D-printing material prime grey has the same flexural modulus as PP. Hence, the mechanical performance can be evaluated with the printed parts. This cartridge design was already reworked to fulfill the requirements for injection molding, such as same wall thickness, 1° draft angle, and the use of maximum two grafters. Figure 8 shows the cartridge design as injection molding part.
Figure 8 (ALU): (A-C) Design drawing of the cartridge as injection molding part in side (B) and back (C) view.
Offers from different companies were requested and documented in the following list. The first three companies (HSG-IMAT, Braunform and MDX Devices) are injection molders, and the tools are designed for large amounts up to 1.000.000 parts. The tools are fabricated from steel and thus have a high manufacturing cost, except MDX Devices. This company works with microinjection molding machine, which allow a cheaper fabrication of the tool. They need for the tool less material and fabrication time. The other three companies (Wehl&Partner, Kaiser, and 1zu1) are rapid prototyping companies. These tools are fabricated from aluminum and are meant to produce between 1.000 to 80.000 parts.
List 1: Injection molding companies
The final design is based on a clip mechanism. It now features four geometrically identical clips, which allow clipping from the top and avoid that during clipping the piston scratches the silicon chip. The reservoir is formed like the old one and supports the agitation system. The final cartridge design is compatible with fabrication by injection molding. The piston guide was designed for the four clip mechanism. Furthermore, the inner region of the piston guide will be produced from a non-metallic material to prevent abrasion between the piston and the guiding structure. Design drawings of the final designed parts are shown in figure 9.
Figure 9 (ALU): Design drawing of the final design. (Left) Piston guide assembled from two different materials. (Right) Mechanical cartridge attachment (clip mechanism) between piston guide and chip holder.
MDX Devices was chosen for fabrication of the chip holder as injecting molding part. The chosen material of the cartridge is polymer polypropylene (PP). It is superior for injection molding and can be sterilized by different methods. The first injection molded parts were produced steel safe, which means that the injection molding tool can be adapted after the first molds if necessary. The 3D-printed Piston Guides didn’t show high enough precision. The tolerance was too high with about ± 300 µm deviation. Therefore the printed parts will be reworked by milling at the important structures. Figure 10 shows photographs of the new clip cartridge made by injection molding (left) and the brass piston guide with the polymer inlet (right) that were fabricated by rapid prototyping methods and reworked by milling.
Figure 10(ALU): (A) New designed cartridges fabricated as injection molding part in PP. (B and C) 3D-printed and milled piston guide from brass in front (B) and back (C) view.
Figure 11 shows a photograph of the new clip design cartridge and the new piston guide mounted to the second generation P9 NG.
Figure 11 (ALU): New designed dispenser unit existing from P9 dispenser, piston guide and cartridge.
These injection molded models of the final clip design demonstrated promising mechanical stability and clipping performance. Also, the first experiments in order to test the dispensing performance (67% of tested cartridges dispensed stable droplet at parameters in a range of 10 µm and 160 µm/ms to 12 µm and 200 µm/ms), handling, and agitation system showed good results (see figure 12). Clipping of the cartridge to the piston guide works very well and handling is much easier than it was with the first generation clip design. Filling of the sample and the use of the agitation system showed the same good performance as the old design. Furthermore, different sterilization methods (Plasma sterilization, E-beam sterilization and Gamma sterilization) were tested. Cartridges show no differences in performance before and after sterilization.
Figure 12 (ALU): Dispensing performance of the injection molded parts at parameters of stroke length 10 µm and downstroke velocity 180 µm/ms. (Left) Nozzle of chip. (Right) Dispensed droplet.
The dispensing experiments show that the first injection molded parts need some additional reworks. The dispensed droplets are very small and need higher parameters than the screwed cartridges. The clip length is too big, a gap of 0.07 mm is given between cartridge and piston guide. Also the chip cavity is displaced by 0.3 mm. Because of that the piston doesn’t hit the chip at the right position (see figure 13). Both problems cause the generation of the small droplets.Figure 13 (ALU): Front view of the piston position relative to the chip reservoir. (Left) Current status, piston displaced by 0.3 mm. (Right) Ideally piston hits the chip reservoir in the middle.
Additional reworks have to be done. At the side of the chip holder, the length of the clips was shortened. Furthermore the width and the depth of the chip cavity, as well as the outlet diameter had to be decreased for more accurate chip gluing. At the side of the piston guide the clipping structure will be displaced by 0.3 mm, so that the piston hits the chip reservoir in the middle (figure 13, right). Drawings of the designed parts with the additional reworks are shown in figure 14.
Figure 14 (ALU): Design drawing of the final design with the additional reworks. (Left) Piston guide assembled from two different materials. (Right) Mechanical cartridge attachment (clip mechanism) between the piston guide and chip holder.
Finally, chip holder and piston guides with the changed dimensions were ordered and have been tested. 22 cartridges were fabricated using the final injection molded chip holder. While 2 failed, 20 performed very well and allowed to generate stable droplets.
The amount of fabricated cartridges increased since January 2014 from 10 to 40 cartridges per week. Through the introduction of some parallelization steps the fabrication could be speed up. Additionally the chips are analyzed under the microscope to check if any incorrectness is visible before gluing. Since this step was included chip performance increased. Since January 2014 all fabrication and quality control steps are stored in a detailed database. Cartridge usage and failure causes are also tracked within the database. With this data comprehensive statistical analysis was done to gather detailed information about reasons for failure and yields for different chip types and assembly methods. The left graph in Figure 15 shows the yield of the fabricated cartridges depending on the nozzle size since January. The average yield of both cartridge types is 81 %. The right graph in Figure 5 shows the failure categories. It illustrates that more than 60 % of the failure cases belong to impurities like particles inside the chip found in final quality control. Process control describes failure cases during parts fabrication like dicing quality of the chip nozzles and during cartridge assembly like tilted chips or glue having entered the chip.
Figure 15 (ALU): Statistical analysis about the yield of fabricated cartridges depending on the nozzle size (left). Statistical analysis about the failure cases (right).
The mentioned failure cases are further investigated and appropriate measures are taken.
Dispensing chip supply chain development - The cartridges are generally packaged in plastic containers, specifically in Corning® microarray slide mailers/storage boxes (Sigma-Aldrich Co. LLC., St Louis, MO, USA), under nitrogen atmosphere. Thereby one container is carrying 2 x 5 cartridges separated by a glass slide. In order to enable individual packaging of the slides in prospect of potential sterilization processes, a sealing film (MELAfol, MELAG Medizintechnik, Berlin, Germany) was tested. MELAfol is impermeable to bacteria, heat sealable, and transparent in accordance with the EN 868-5 standard. In addition, the film has peelable sealing seams and a handling indicator that changes color from blue to brown with steam sterilization. The cartridges were packaged individually in MELAfol film and sealed with an appropriate foil sealing device. In May 2014 a shelf life study for the cartridges was started. Cartridges were stored in batches in containers and individually packed in film under various conditions over durations of 1-3 months, for certain conditions up to 12 months. All tested cartridges were quality controlled before and after storage. Batches of 10-24 cartridges each were tested in
● two different packaging formats
• container carrying 10 cartridges each
• individually packed cartridges in MELAfol film
● three different storage conditions
• room temperature (laboratory condition)
• 5°C (fridge)
• 70°C (oven, accelerated aging)
● at seven time points
• 1 month
• 2 months
• 3 months
• 4 months
• 5 months
• 6 months
• 12 months
Figure 16 summarizes the results of the shelf life study attesting the storability of the cartridges for up to one year without appreciable reduction of their functionality.
Figure 16 (ALU): Shelf-life study of cartridges. Operability of cartridges was determined by standardized quality control (QC) before and after applying indicated packaging and storage conditions at different time points and is expressed as the percentage of cartridges passing the QC (n=10-24). Note: 3 out of 10 cartridges packaged in film did fail the QC after 6 months at room temperature, but passed it successfully when re-investigated at a later time point (1 year).
On a long run, the cartridges should be provided as sterilized consumables to enable a contamination-free single-cell printing. Common sterilization methods such as heat (e.g. steam), chemicals (e.g. ethylene oxide), irradiation (e.g. UV, gamma- and beta-radiation), and plasma are investigated. To address the issue of cartridge sterility a study is being conducted in 2015. Therefore the most common and established standards in sterilization are being evaluated. As first method steam autoclaving was tested. Here cartridges were exposed to standard programs using a laboratory standard autoclave working with 121°C water steam. The standard cartridge material PMMA does not survive this process. PEEK as alternative survives. The microchip is also not compromised. The cartridge as a composition of chip (Si-glass), chip holder (PEEK) and glue (2K epoxy) does not survive in sufficient numbers and quality. As critical issue the connectivity between Si, epoxy and PEEK was found. The different thermal extension coefficients of all cartridge materials seem to introduce mechanical stress resulting in leakage. Current failure rates around 60% making this process unsuitable. Therefore, steam sterilization does not appear as a method of choice for sterilization of the cartridges, at least when using the thermally labile PMMA as material for the chip holder. Currently, further sterilization methods like UV, gamma-, beta-irradiation and plasma are under investigation in consideration of their compatibility with the PMMA cartridges. Preliminary results indicate that irradiation of the cartridges at 20 – 25 kGy with gamma- and beta-rays, respectively, and plasma sterilization with hydrogen peroxide, do not impair the operability of the cartridges and could therefore be suitable sterilization methods for the consumables of the single-cell printer.
Optical detection system for bacteria - In order to reliably print single bacteria cells with the SCP several modifications on the detection optics were calculated and tested. New dispenser chips with increased optical clarity are being developed.
The following optical setups were tested with beads and bacteria cells:
1. The standard SCP optics with the 3.1x TV lens
2. Modified SCP optics with a plug & play 5.0x TV lens
3. Modified optics with a 10x long working distance microscope objective
While the standard SCP optics (1) is not capable of detecting 1 µm sized beads, the slightly modified version (2) works fine with these beads. Although setup (2) has a decreased working distance of d = 40 mm compared to the standard setup (d = 77mm) it proved to be beneficial for our work with smaller objects as it is easily interchangeable with setup (1) (plug & play). Although most bacteria such as E.coli are larger than 1 µm, they could not be reliably detected by setup (2). Therefore we assembled a modified optic that comprises a 20x microscope objective featuring extra-long working distance. Nevertheless, the higher magnification has to be paid off by a decreased field of depth and working distance (for setup 3, d = 17.6 mm). While the flexibility of the standard SCP suffers from such a low working distance between the dispenser chip and the objective, this is acceptable for the MALDI system where the dispenser is fixed and not mounted on a movable axis. A first detection and dispensing system was deployed to the MALDI-TOF device at BiosparQ for initial integration and characterization (see also Deliverable D1.1 Pre-production prototypes). IMTEKs work on characterizing the optical system for detection of bacteria is still ongoing towards an improved 2nd prototype, including the development of dispenser chips with increased optical clarity. Further, we will also work on the necessary software optimizations regarding the detection algorithm for bacteria cells. The initial tests with the new print head look promising: The integrated platform has been tested with latex microspheres and well known E.coli and Serratia strains delivered by TOPLab (University of Leiden). With latex spheres it was discovered that droplets are charged, resulting in the particles being repelled by the electrical field of the ion source in MS. Implementing an earthwire seems to solve this issue. This result was fed back to ALU-FR to update the design of the dispenser or the disposable chip. Experiments with E.coli strains have revealed that coating of the bacterial cells is the most crucial step for the BSQ platform. More specifically, the morphology of the resulting particles (bacterial cell + crystallized matrix material+ interface) determines the information content of the individual spectra. The morphology is influenced by
● Matrix material (HCCA, SA, DMT, ...)
● Additives (surfactants, acids ...)
● Droplet formation / condition of the nozzle surface (contamination, ...)
● Drying regime (matrix co-crystallization)
Exemplary images of “bare” and matrix-coated cells are shown in Figure 17
Figure 17 : REM image of “bare” cells (left) and matrix coated ones (right).
As described above we evaluated several optical setups regarding their applicability for single bacteria cell detection. It became clear that to facilitate robust single bacteria cell printing the following points need to be addressed:
● Resolution of the optical detection system
● Reduced dosage chamber depth (i.e. reduced nozzle size) due to smaller depth of focus and improved optical clarity of the dispenser chips
● Image processing algorithm optimization for bacteria detection
In order to evaluate and test the setup we worked with two setups: One setup was integrated into a printhead that was attached to the Single Cell Printer, in order to evaluate the bacteria cell printing capability. The other optical setup is designed for the MALDI-TOF prototype and was assembled tested “offline”, i.e. without a robotic stage.
Dispenser chips - Usually the depth of focus is inversely related to the magnification (and resolution) of objectives. Therefore, a smaller nozzle size of the dispenser chips are not only beneficial for the MALDI-TOF instrument as they produce smaller droplets, but also because they reduce the depth of focus that is necessary to image the small cells with sufficient sharpness and contrast. As stated in an earlier report we produced 20x20 µm² sized nozzles, i.e. dispenser chips that have a dosage chamber that is 20 µm deep. These work very well and were tested for their capability to print single bacteria cells, which requires an efficient detection. Due to their small size and the low contrast it is crucial that the dispenser chips are optically clear and provide a bright and uniform background. By optimizing the silicon etch protocol during the micro-chip fabrication we could significantly improve the situation as depicted in Figure 18.
Figure 18 (ALU): Dispenser chips with 40 µm x 40 µm dispenser chip (left) and new 20 µm x 20 µm orifice opening. 20µm chips have reduced droplet volume (right). The new chips show significantly reduced surface roughness on the silicon.
Optical detection system - The general design of the optical detection system for printing single bacteria cells is very similar to the standard Single-Cell Printer (SCP) detection setup with two important modifications: First, for printing bacteria cells we use a special high magnification objective with extra-long working distance. And second, we placed two apertures in front of the illumination LED which allow collimating and focusing the blue light illuminating the microchip. The apertures in the illuminating light path help to significantly increase the contrast and sharpness of bacteria cells in the microchip as depicted in Figure 19.
Figure 19 (ALU): The coaxial illumination system was equipped with two apertures allowing focusing and collimating the incoming light. The sharpness and contrast of bacteria cells can be significantly increased by closing the 2nd aperture (20 µm dispenser chip, E.Coli cells).
Software optimization - The SCP software was modified in the following aspects:
● The different cell detection camera types are now supported.
● The object detection algorithm was modified to enable for detection of low contrast objects 1-3 µm in size by increasing the detection sensitivity and optimizing the object classification.
Evaluation of the prototype - In order to evaluate the performance of the setup we printed fluorescently labeled Escherichia Coli cells (a GFP expressing strain). We printed three arrays with 100 single cells each on a glass slide. The cells were suspended in DI-water and Glycerol (10 % v/v) which allowed us to image and count the cells after printing. The resulting spots were classified according to the number of cells: No cell, single cell, or multiple cells (see table 2). Typical fluorescent images can be seen in Figure 20. The total single cell efficiency yielded 66.7 %. We think that the single cell efficiency can be further improved by reducing the vibrations of the printhead induced by the robotic stage. This could be achieved by using a fixed printhead and moving the substrate only. Since the MALDI-TOF MS setup is not mounted to a moving stage we believe that here the single cell efficiency will be significantly higher once the setup is assembled to the MALDI-TOF MS instrument.
Figure 20 (ALU): Typical fluorescent/bright field overlay images of printed droplets allow to verify whether the droplet contains a single cell (left), no cells (center), or multiple cells (right).
Table 2: To analyze the performance of the instrument, we printed three arrays (10x10) of a GFP-expressing E. Coli sample on a glass slide and counted the number of E.Coli cells in each droplet. The total single cell efficiency yielded 66.7 %.
Prototype instruments - The optical detection system for bacteria was integrated into a Single-Cell Printer prototype instrument at IMTEK for optimization and evaluation of the printing performance. A prototype Single-Bacteria printhead for the BSQ´s MALDI-TOF instrument was designed, assembled and tested.
Single-Cell Printing efficiency and chip-to-chip variation - Single-cell printing efficiency was evaluated by printing GFP expressing E.coli BL21 cells on glass slides. Each spot of a 10x10 array was classified via fluorescent microscopy either as void droplet, droplet containing a single cell, or droplet containing multiple cells. The assay was performed three times for each of three different cartridges. In total, 699 out of 900 spots contained a single-cell, resulting in an averaged single-cell printing efficiency of 79 %. The results are summarized in Fig. 21 (ALU) Single-cell printing efficiency was evaluated by printing GFP expressing E. coli cells on glass slides. Each spot of a 10x10 array was classified via fluorescent microscopy either as void droplet, droplet containing a single cell, or droplet containing multiple cells. The assay was performed three times for each of three different cartridges. In total, 699 out of 900 spots contained a single-cell, resulting in an averaged single-cell printing efficiency of 78 %.
Droplet size reduction - For many applications including MALDI MS single cell pathogen detection (WP 3) smaller droplet volumes are required. Our current dispensing chips allow to reliably produce droplets with volumes as low as 150 pl. Here (WP 1.5) we aim to produce dispenser chips that can consistently produce droplets with a volume of 20 pl. Therefore we manufactured dispenser chips with smaller nozzles which are characterized by gravimetric and image based methods. So far we are able to produce water droplets with 35 pl with the new dispenser chips in a reliable manner. As depicted in Figure 22, our current dispenser chips with a 40 µm x 40 µm nozzle allow to reliably produce droplets with volumes as low as 150 pl. In order to be able to produce smaller droplet volumes we manufactured a set of new dispenser chips with various nozzle sizes of 20 µm x 40 µm, 20 µm x 30 µm, 20 µm x 20 µm, and 20 µm x 10 µm as outlined in Figure 23. In WP 3 (MALDI MS for pathogen detection) we are aiming for very small droplet volumes of 25 pl. As the chips with the rectangular 20 µm x 10 µm nozzle did not reproducibly generate droplets, we therefore focused on characterizing the chips with the quadratic 20 µm x 20 µm nozzle.
Figure 22 (ALU): REM pictures of our current dispenser chip with a 40 µm x 40 µm nozzle (left) and a 20 µm x 20 µm nozzle (right).
Figure 23 (ALU): Current dispenser chip with a 40 µm x 40 µm nozzle is able to produce droplet volumes down to 150 pL which optimal for dispensing mammalian cells. However, some applications such as real time MALDI MS for pathogen detection require much smaller droplet volumes.
In order to measure the droplet volumes of the new dispenser chips (20 µm x 20 µm nozzle, silane coated) we used a highly precise gravimetric method as presented by Dong et al [1]. The balance of our setup (XP2U/M, Mettler Toledo) has a resolution of 0.1 µg. However, we are dealing with droplets volumes down to 20 pl that therefore only weight 0.02 µg each. In order to precisely determine the (average) droplet volume we therefore dispense n = 600 droplets for each data point. Hence, with the gravimetric method we cannot address the drop-to-drop variation which is a separate issue that will be discussed at the end of the report. First we determined the parameter space that allows for generation of stable droplets by varying both the piezo downstroke velocity and the stroke length. The droplet stability and the presence of satellite drops was observed by dispensing on a glass slide on top of a camera (see Action Report WP 6.4). We found that for downstroke velocities of 70 µm/s we obtain stable droplets that can be tuned in size by adjusting the piezo stroke length. Figure 24 illustrates such a characteristic dosage curve for a fixed downstroke velocity. Stable droplets with 35-60 pL volumes were obtained at stroke lengths of 2-4 µm (green data points). This regime could be reproduced by all our dispenser chips (n=7) tested so far. At stroke lengths < 2 µm some of the dispenser chips were able to produce 25 pL droplets, others did not shoot at all. At higher stroke lengths > 4 µm we occasionally observed satellite droplets.
Figure 24 (ALU): The droplet volume as function of the piezo stroke length at a constant downstroke velocity of 70 µm/s. The green data points represent a stable droplet generation regime which could be reproduced by all the dispensing chips tested so far. At higher stroke lengths (red) satellites occurred occasionally. Droplet volumes as low as 25 pL could be achieved but stability issues were observed (yellow).
We also investigated whether the dispensing frequency influences the droplet volume. Figure 25 illustrates that we can dispense with frequencies up to 200 Hz without affecting the droplet size. The stroke length was fixed to 2 µm and the downstroke velocity to 70 µm/s. Note: we have not tested dispensing at even higher frequencies, which might as well be possible.
Figure 25 (ALU): The droplet volume is constant for dispensing frequencies between 50 and 200 Hz. Here the stroke length and the downstroke velocity were set to 2 µm and 70 µm/s, respectively.
We have manufactured new dispenser chips with various nozzle sized. The smallest nozzle with a quadratic orifice (20 µm x 20 µm) performed very well, and we can consistently produce droplet volumes as low as 35 pl as measured on our gravimetric setup. At lower volumes we observed a strong chip-to-chip variation in terms of dispensing parameters. Therefore we will continue to investigate the small volume dispenser chips with special focus on the following points:
● Detailed study of the chip-to-chip variation by correlating the droplet velocity with the actual dimensions and the quality of the nozzle (which can vary due to the dicing process in the chip fabrication).
● We will investigate the influence of the medium viscosity on the droplet volumes.
● In order to determine the drop-to-drop variation, we are currently working on a new setup based on stroboscopic imaging using a flashed LED and high magnification optics. (Note, that it is not an easy task to monitor the 20 pl droplets in flight with a decent resolution that allows to calculate the droplet volumes).
We think that with these chips even lower volumes down to 25 pl might be possible once we have gained more insight into the chip-to-chip droplet volume variation. Further, we are a considering manufacturing chips with even smaller nozzle sizes of 10 µm x 10 µm. As stated above we manufactured and tested new dispenser chips with smaller nozzle sizes. The smallest nozzle with a quadratic orifice (20 µm x 20 µm) performed very well and we started to integrate this chip into our regular cartridge fabrication process. Right now, we are producing about 60 cartridges with 20x20 µm² nozzle each month (out of 160 in total). After one of the earlier bacteria printing experiments a failure analysis was carried out. Figure 26 below summarizes the reasons for all “no-cell” or “multiple-cell” events out of 200 printed bacteria cells. “ROI too small” and “no cell” accounted for almost 19 % of the failures. Meanwhile these can be avoided to a large extend by choosing the optimal ROI size, a stiffer printhead design, and a more sensitive cell detection algorithm. After these improvements we frequently achieve single-cell printing efficiencies of close to 80 % as discussed above. Given the small size of the cells this is a reasonable performance, although slightly lower than what is achieved with the eukaryotic cell printer. In order to reach printing efficiency of > 90 % with bacteria cells, the two remaining reasons for failure have to be addressed:
● In almost 4% of the events bacteria cells cannot be detected due to the edge shadows of the dosage chamber.
● In 6% of the events bacteria cells cannot be imaged with sufficient contrast as they are not close enough to the focal plane of the objective.
Figure 26 (ALU): A failure analysis was performed on 200 printing events by evaluating the SCP images and the printing results. The figure summarizes the reasons for “no-cell” or “multiple-cell” events.
Decreasing the depth of the microfluidic chamber addresses the two remaining issues. Therefore we fabricated dispenser chips with only 5 µm and 10 µm etch depth. As these chips are based on the previous design, the major challenge was the dicing step that determines the final width of the nozzle. Using our current process, this step could not be performed with the necessary precision, resulting in chips with non-uniform nozzle sizes ranging from 2 µm to 20 µm. Nevertheless, this allowed us to pick a dispenser chip with a nozzle size of 10x10 µm for testing. We found that this chip can generate droplets with a volume of approximately 4 pL.
Figure 27 (ALU): Chips with a nozzle size of 10x10 µm² were produced and evaluated via stroboscopic imaging.
Fluorescence detection - The optical detection system of the SCP prototype is further developed towards integration of single channel fluorescence detection. Focus is on selection of hardware components and construction of a specific print head. The prototype system being available for proof-of-concept was reworked in terms of reconstruction of the optical train including improved filter and beam-splitter elements as well as compactness and robustness of laser beam adjustment. Main part of the work thereby was construction using SolidWorks and investigations on optical elements. An improved 3axis adjustment system for the dispenser and an advanced illumination concept for bright-field imaging are work in progress. The setup would allow for precise free-beam coupling of the laser. Such accurate positioning of the laser beam spot in the focus area on the chip will be possible. Figure 28 shows the construction in Solidworks and the final prototype system under assembly. The system is currently still under construction and has not been tested or characterized yet. More details will be provided in Deliverable D1.1 “Pre-production instruments”.
Figure 28 (ALU): SolidWorks construction of compact fluorescence print head featuring laser and high resolution high transmission optical elements and a 3-axis adjustment system for correct focusing of the dispenser chip. Left: photograph of the print head under assembly.
The development of the fluorescence printhead has been finalized. It is mounted to a SCP platform which was electronically upgraded and reworked in terms of electrical safety. An emergency shutdown for the built-in laser (class 3B) setup was included to guarantee user safety. Further a reworked substrate holder was developed to meet the special requirements of the printhead. A protective cover shields the entire laser optical system.
Figure 29 shows a photograph of the final printhead version under operation.
29 (ALU): Photo of the final printhead including protective cover and safety instruction stickers according to EC norms.
The functional set of different batches (array printing, well plate printing, etc.) has been programmed and tested successfully using polystyrene microbeads and fluorescent cells ( Figure 30).
Figure 30 (ALU): Screenshot of the system under operation. Fluorescent microspheres illustrate the dimension of the laser illuminated area and the intensity of fluorophore excitation.
Besides protective covers and laser safety the usability of the system has been further improved. The magnifying objective with its high numerical aperture requires a reduced working distance to capture sufficient amounts of fluorescent light from the cells inside the chip. Reduced working distance has two major drawbacks that need to be compensated: accessibility of dispenser chip and accessibility of well plate format substrates. Well plate access is – due to a specially developed substrate carrier – still possible by filling one half of the plate, then rotating the plate and filling the second half. Although this workaround requires one additional operator step, it does not compromise the laboratory workflow of the SME partners. Access to the dispenser chip is required to change chips / cartridges between different experiments. Working distance between chip and objective is very limited in this fluorescence setup (< 25 mm). Both a proper screwing with the current cartridge models as well as clipping with future models is not easily done in this setup. Therefore a special, user friendly system featuring a swivel arm was developed. It allows for sideways movement of the entire dispensing system giving much space and free access to the dispenser chip. Figure 31 shows the workflow.
Figure 31 (ALU): New innovative chip mounting system applied in the fluorescence printhead. A two stage swivel arm is used to swing the dispensing unit sideways away from the objective. Such, the user has free access to the montage screws and can easily change the chip / cartridge.
The single-channel fluorescence detection system was implemented into a Single-Cell Printer prototype. The prototype was delivered to Innoprot for beta-testing (production of fluorescent cell lines). Transport, on-site installation and training by an IMTEK staff member was provided in April 2015. Prior delivery of the prototype, evaluation experiments were performed at IMTEK using a GFP transfected cancer cell line. Figure 32 shows a U2OS-GFP cell inside the ROI of the dispenser chip.
Figure 32 (ALU): U2OS cell inside the ROI of the dispenser chip.
To evaluate the performance we spiked U2OS-GFP cells into a non-fluorescent cell sample (HEK293) in a 1:10 and 1:100 ratio. We then used the SCP to sort and isolate the U2OS cells based on their fluorescence. Averaged over four 96-well plates the single-cell efficiency (single U2OS cells printed into well) yielded 93 %. The results are summarized in figure 33.
Figure 33 (ALU): GFP expressing U2OS Cells were spiked into a non-fluorescent cell sample (HEK293). The fluorescent SCP prototype was used to sort and isolate the U2OS cells based on fluorescence. The cells were printed into 96 well plates for clonal culturing. Post priting imaging was performed with a fluorescent microscope.
Throughput optimization – This section covers software and hardware actions to enhance the overall throughput of the system regarding detected and printed cells. Hardware has been improved on:
● Enhancement of camera frames by faster sensor and active binning.
Starting value: 60 fps
Actual limit: 150 fps
Target: 200 fps
● Redesign of vacuum shutter system towards faster shutter times.
Starting value: 500 ms
Actual limit: 50 ms
Software has been improved on:
● Reworked detection algorithm with faster object recognition.
Starting value: 6 ms/image
Actual limit: 4 ms/image
Target: 5 ms/image
The actual software/hardware configuration has been tested successfully on a SCP pre-production instrument. Further enhancements towards the specified limits are under development.
Data-management software - Several scripting functions have been programmed and embedded into the CORE image processing and cell detection algorithm. Both live data displaying as well as data readout via comma separated value based text files was realized. Comma separated value format thereby is the leading standard interchange format for feeding data collections into database constructs. The data sets can actually be loaded in table based programs like MS Excel and filtered by specific data. Further visual data like dot plots (well-known from flow cytometry analysis) can be created and displayed. Data storage of image sets is further optimized by reworking the storage serial system and assigning comma separated values to the specific image sets on the hard drive they belong to. Both data collection text file as well as image sets are stored together in one folder. An automated user and experimental settings file currently is under development. The file contains all necessary information about experimental conditions regarding hardware settings and stores all user input registered prior to start of the experiment. This settings file also will be automatically created and stored in each experimental folder. Further it can be loaded in case of performing the experiment again under identical conditions at a later time.
Image storage optimization - To consistently store images on the hard drive, the software GUI allows for setting individual folder names (e.g. experimental name or name of the user + experiment, etc.). To avoid overwriting a respective routine has been implemented to protect existing folders from being overwritten. Such, no experimental information is lost. Image naming has been optimized. Each image is named after a serial key which is assembled according to the following steps:
● Well coordinate / array coordinate
● Number of cell being dispensed at this coordinate (typically 1 -> single-cell)
● Number of track image[1]
● Date-time stamp
The date-time stamp guarantees that image names are always unique and cannot be confused. Each image sequence can be assigned to the specific experiment and the specific location on the used substrate (typically well plates).
User input log file - To preserve the user input and assign it to the respective experiment, a user log file was integrated. It reads out the entire set of user input required to set up an experiment and stores it in a text file. The file is stored in the respective folder, where the images are stored.
The data from the log file can be read out via any text editor and can even be converted to other formats (like MS Excel) since data is stored as comma-separated value format.
Data log file - Each event (cell, particle, etc.) during an experimental run is preserved by a second log file. The file stores mean diameter, area, perimeter, aspect ratio, center of mass, and classification of the object. Objects being identified as single cells are specially marked. Once one of these single cells has actually being dispensed to the substrate the respective coordinate (well or array) is added to the event. The file is stored in the respective folder, where the images are stored. Such the image sequence can directly be correlated to log file data. The user can post-print extract the required information on specific events. Further the log file format is also comma-separated values and can such be easily loaded to excel. Here the entire set of events from a single experiment can be visually displayed in form of dot plots (point clouds) which is the common method to analyze cellular data. Within the point clouds different classes of objects (e.g. single cells, cell clusters, cell fragments, dirt particles, etc.) can be distinguished. When loading multiple experiments direct comparison of point clouds is possible. The data management processing stores are relevant data in the same folder. Unique file naming convention and overwrite protection routines prevent mixing up and loss of data. Universally accessible file formats allow for easy post-processing of data.
[1] Each single-cell event consists of a series of five subsequent images to track the single-cell before being dispensed and to verify if the cell has actually being correctly dispensed
Furthermore, an integrated machine for cell research is developed which on the one hand utilizes the core technology as developed in work-package 1, on the other hand is a good base for a device which can then readily be sold to customers after the project. However, this work-package also suffered from the startup delays at ALU-FR and is currently delayed by about one month. It is expected that by end of the year, the work-package as a whole will be back in time. For supporting ALU-FR in designing a device which can later on be CE-IVD certified, an external consultant was hired: Manfred Augstein from velixx GmbH (http://www.velixx.com) has been designing and market-launching CE-IVD machines for more than 20 years at Roche and thus has a strong knowledge in what to do and especially what better not to do. His mandate includes reviewing the process from time to time and providing hints to improve the development process. The technical specifications of the pre-production prototypes to be developed can be found in table 3.
Table 3: Specifications of the basis platform
Design for manufacturability and system integration - The design and the mechanics of the existing Prototype have been reviewed and partially optimized. The axis of the Prototypes is improved in order to enable the different printheads. An EMV test was performed at the existing SCP prototype. As a result, grounding of the PC casing and the stroboscopic setup needs to be improved. The anodization of the housing has been removed at specific points to ensure electrical conductivity throughout the complete casing. Additional ground wiring was mounted to the stroboscopic setup. Another issue addressed was heat exchange inside the electronic box. This problem has been solved with stronger fans and improved cable management to enhance free space for air guidance. Serious problems with USB 3.0 driver settings and chip set drivers leading to continuous crashes of the camera frame grabber were solved. Further the routing of the USB 3.0 and the camera trigger cables was shifted to outside the cable duct to prevent cross-talk that had been identified as a possible reason for hardware instabilities. The named changes had positive effects on system overall stability. On the mechanical side the accuracy of the axes system had to be improved. In the new prototype the implementation of HIWIN ball screw spindles was favored. The spindles are high end class and can be easily exchanged in terms of required precision without the necessity to change gears, ball screws or bearings. This system is especially attractive towards the different requirements in applications from the SME partners. Pipetting inside a 96-well plate format may require less precision than non-contact printing of dense micro-arrays. With the exchangeable spindles a significant cost-reduction is possible for lower precision applications. The linear guides of the existing SCP prototype instrument have good accuracy but their long term robustness is questionable. It has already been observed that accuracy decreases over number of use cycles. In the next prototype thus linear ball guides will be implemented to guarantee a constant accuracy even at high numbers of working cycles. In addition to the accuracy issues the axes coupling had been reviewed. In the existing prototype the motor is in line with the axis and the coupling. This means, that the axis loses traveling distance and has no loose bearing (See Figure 34). Such the axes is only fixed at one end which makes it susceptible to vibrations and mechanical stress (e.g. longitudinal torsion).
Figure 34: Scheme of the actual axis approach. Motor and axes are in line.
In the approach for the new design we implemented a belt which decouples the axis from the motor and added a loose bearing. This new approach has also a larger traveling distance, is more defined and less susceptible to vibrations (cf. Figure 35).
Figure 35: Scheme of the new axis approach. Motor and axis are next to each other and connected by a belt.
The implementation of the novel components is work in progress. A briefer view on what has been achieved so far is provided in Deliverable D2.1 Finalized prototype design. As above, the existing printheads have been adapted in order to enable fluorescence detection and the detection of bacteria. These two prototypes are heavier than the existing prototypes. The new pre-production prototype has more accurate linear rails, which can also take higher loads and accept higher torsion forces. Several adapters between the prototype and the printheads have been manufactured. This allows for montage of fluorescence and bacteria detection printheads. The additional functionality of both printheads made modifications inside the instrument necessary. New illumination and – in case of fluorescence – the implementation of a laser source required additional modules as well as driver and control electronics to be embedded into the corpus of the instrument. Therefore additional space has been generated by re-designed and re-arrangement of internal component as well as introducing different decks for certain component classes. The corpus height was slightly increased. Further additional safety was embedded, specifically regarding laser safety and emergency shutdown.
Documentation (CAD, CE-documentation, user manual) - The CE documentation of the device was started for this work package. Test criteria for the specifications were defined. Initialy some general CE relevant points were added to the specification list (see Table 4). References for these points are the CE directive 2006/95/EC on low voltage devices and the CE directive 2004/108/EC on Electromagnetic compatibility. In addition to these two, there is the CE IVD directive 98/79/EG. For the technical specifications were some test criteria defined. The accuracy of the base platform for example is tested with two different Tests. The first test is based on the VDE Guideline 3712 and is used to check the general dispensing performance. A second test was implemented especially for our base system and can measure the accuracy of the device positioning with a higher resolution. The position of the device is here checked optically several times with an USAF Resolution Chart. These Tests are explained in detail in WP 4.2. In Order to convert the specifications into a written format, an external engineering consultant, called Manfred Augstein (from velixX GmbH ), was hired. He gives us advises on CE and business related questions and helps to set up the CE documentation. At the moment we got a Project Specification Document, a Project Documentation Plan and a Project Guide. The construction of the first Prototype is nearly finished. After the finalization of the construction, the design is frozen. Following changes of the design need to be documented with reason, explanation, date and signature. This approach is part of the CE conform documentation.
Table 4 (ALU): CE relevant terms of the Specification
In parallel to the assembly of the pre-production prototype, an assembly manual has been written. This manual includes mechanical and electronic assembly as well as cable interconnections. For electronic and power supply connectivity a routing plan has been written. This plan includes all wiring, all connectors (internal and external) and all circuit board I/O’s. Further a user manual has been written. This manual includes an overview of all parts of the SCP (nomenclature) and a detailed guide for hardware and software. The manual was revised and improved internally by the team of ALU-FR as well as externally through the application partners. The pre-prdocution instrument platforms meanwhile have approached towards CE marking. CE conformal document structure has been established and prior documentation is already partly transferred into the conformal formats. The BioFluidix platform instrument has passed the industrial CE conformal EMC test for laboratory devices.
Pre-production instrument fabrication - Assembling of the prototype started with the mechanical parts. The base was assembled first alongside with the necessary wires for the printhead (see Figure 2.5.1). Next was the bridge with the X-movement (see Figure 2.5.2) followed by the Z-axis and the printhead. The next step was to grease and calibrate the axis. After the mechanical, the electrical assembly was done. This included power supply, PC, motor controllers, dispensing electronics and additional electronics for cell detection. These parts have been routed together and with the motors, end switches and finally the printhead (see Figure 36). Software has been installed at the end. Software in this context features the operating system, the camera and electronics drivers and control software, the motor DLL’s and the actual single cell printing software (GUI). The latter is the interface (GUI) which actually interacts with the user and controls the cell printing process as well as all hardware parts. The following photographs were taken during assembly and functional testing in Sept./Oct. 2014.
Figure 36 (ALU): Base of the prototype with routed cables.
Figure 37 (ALU): Assembled base and Bridge.
Figure 38 (ALU): The electronics of the prototype during the routing of the electronics.
Figure 39 (ALU): The assembled prototype.
Pre-production instrument characterization - In this period the pre-production Prototype has been assembled and characterized. Both the mechanical specifications of the base platform and the performance of the printhead were evaluated.
Base platform - The pre-production prototype has a working area of 280x380x75 mm³ (X,Y,Z). The working distance in X of the base platform is regularly 338 mm large, but has to be limited according to the size of the cell printing printhead. The range of the Z-Axis is only 75 mm instead of the specified 100 mm , but can be adapted by taking a different axis. This can be easily done given the modularity by design. After the assembly of the pre-production Prototype, the axis performance has been tested and optimized. The velocity of the x- and y- Axis was initially set to 75 mm/s. After the first tests of the axes movement the velocities were increased to 150 mm/s. For all of the following axis tests the latter velocity has been used. Characterization of the accuracy was performed with an USAF resolution chart. The test has already been described in the action report WP 4.2. These measurements showed that the X and Y-Axis have an absolute accuracy of about ±15 µm. Together with the absolute positioning accuracy, the backlash has been measured. This test showed, that the X-Axis has no backlash, but the Y-Axis has about 100 µm backlash. The relative positioning accuracy still needs to be measured
Single cell printhead – A single cell printhead was mounted on the base platform and tested with beads. To fully exploit the whole range of single cell printing application we work with two optical detection setups. Setup A is our standard setup that is used for most single cell printing applications. It can be used to detect and sort particles (or cells) ranging from 4 to 40 µm in size. Setup B is based on a high resolution microscopy optic and is used for smaller particles such as bacteria cells. The detailed technical specifications are given in the table below. Droplet size was measured gravimetrically, as described in previous reports. The particle size refers to printing of polystyrene beads. The dispensing frequency is about 30 Hz but can be increased to 60 Hz by using the binning function of the cell detection camera.
Table 5. (ALU): technical specifications of the SCP
MALDI MS for pathogen detection - A MALDI mass spectrometry device for pathogen detection incorporating the single cell printing technology from WP1 is developed and evaluated. There have been some delays in starting all clinical related work-packages as there has been a change in the partner structure at EMC medical: Dr. Willem B. van Leeuwen has left EMC and has been acquired as a lecturer at Leiden university with Dr. René te Witt taking over the lead of his former group. It is planned to integrate Willem van Leeuwen’s group in Leiden as a third RTD performer after the end of RP1 to increase the development speed by using his knowledge and by the fact that BiosparQ and his group are located in the same building in Leiden which will shorten learning and adoption curves. Table 6 to 8 show the specifications for the MALDI application.
Table 6: Specifications for the dispenser chip
Table 7: Specifications for the optical detection system
Table 8: Specifications for material compatibility
SCM-technology integration & verification
BiosparQ aspires to analyze samples comprising mixtures of micro-organisms in a background of biological particles. To enable this capability, BiosparQ generates a MALDI mass spectrum of individual cells. For this purpose, microbial cells are to be suspended in an appropriate fluid phase containing all additives required for MALDI (including the MALDI matrix). This suspension is subsequently dispensed into droplets sufficiently small to prevent presence of more than a single cell. Upon drying the MALDI matrix crystalizes, resulting in an individually coated cell suited for MALDI analysis. So, the key challenge for BSQ in the SICTEC project is to interface the SCM technology with the Single Cell MALDI TOF platform. Prior to the SICTEC project BSQ has demonstrated that, using pico-liter droplets that contain one bacterial cell and the MALDI matrix material dissolved (a so-called pre-mix approach), it is possible to generate spectral information from which identification of the bacteria is possible. These droplets were dispensed by means of a Microdrop dispenser, which had to be cleaned after each test. Clearly, this cannot be used in high throughput clinical applications, an approach using disposable dispenser heads should be used instead. For that reason the IMTEK SCM technology was selected to interface with the BSQ platform. The scope of work of this work package was to integrate the SCM printhead on the BSQ ‘drying tower’ sub-assembly, learn how to work with this sub-assembly, integrate the sub-assembly on the BSQ instrument and verify its functional performance. Completion of this work package was the starting point for testing well known bacterial strains and actual clinical samples as described in the subsequent work packages. The SCM hardware and software delivered by IMTEK was integrated on the ‘drying tower’ of the BSQ platform and initially tested off-line for BSQ to become acquainted with the SCM technology. Different experiments were performed to explore the different settings of the printhead. Parker ink was used to demonstrate performance and to check the printing and drying behavior of the sub-assembly.
Figure 40: Off-line test set-up with SC dispenser integrated on top of the BSQ ‘drying tower’. The SC dispenser is not visible and is located in the box on top of the drying tower.
After that, the SCM print head - drying tower sub-assembly was integrated onto the BSQ platform for system verification. As planned, initial test using 1 µm Polystyrene Latex (PSL) spheres were done. The purpose of this test was twofold: firstly it should be verified that the PSL spheres could be detected by the detection lasers prior to firing the ionization laser and secondly it should be verified that a polystyrene spectrum could be built after ionization of the resulting particle.
Figure 41: SC dispenser with drying tower integrated on BSQ MALDI MS prototype in UAS-L Toplab. The left side computer screen is the output of the data acquisition / analysis system for real time monitoring of the experiments.
The first part of the verification was initially troublesome as no detections were observed, indicating that somewhere droplets/particles got lost on their way from the SC print head towards the ionization chamber. Only after shutting down the High Voltage on the ionization source it was discovered that static electricity was the cause of this problem. Charged particles were deflected by the electrical field generated by the ion-source, see below. Apparently this charging of droplets was something that is typical for this type of SC dispensing technology, as it was never observed before by BSQ using the Microdrop dispenser technology.
Figure 42: Simplified model of the trajectory of charged particles entering the ions source HV electric field when leaving the stage 3 skimmer. The trajectories explain the phenomena experienced regarding the lack of detection / ionization.
To remove charge generated during dispensing, the analyte was grounded. This appeared to be very effective, albeit that many times during subsequent testing droplets seemed to be (somewhat) charged again and became deflected. In addition, a solution was implemented where the droplets were uncharged by induction upon leaving the dispenser chip. For this purpose a ‘charger’ was constructed (aluminum foil) close to the nozzle of the chip in order to ‘push back’ charge on the droplets just before separating from the stream.
Figure 43: Picture showing the ‘spring like’ solution to ground the analyte. The spring fits in the reservoir of the chip-cartridge assembly (not mounted here) and transports charge towards ground. The aluminum foil is the ‘charger’ on which a voltage can be applied to ‘push back’ charge by induction on droplets being formed at the nozzle.
Although this way of de-charging seemed to work, it application during actual testing appeared to be too complex and time consuming. Typical test results, where the charged particles are deflected and are being slowed down in the electrical field generated by the ion source are shown below.
Figure 44: Typical example of histogram which illustrates an experiment where a voltage is applied on the analyte (not on the aluminum foil). Note that the Time of Flight (horizontal axis) between the detection lasers can be controlled by the voltage applied. Also the detection rate and the number of mis-firings of the UV laser changes with voltage.
As conductivity of the analyte, potential obstructions in the micro fluidic chips and other uncontrollable phenomena could be the source of this, it was discussed with BFX to change the design of the dispenser chip to prevent charging of the analyte, see WP 1. After it was shown that PSL spheres could be detected after dispensing, albeit still with rather low efficiency (which indicates that part of the PSL spheres is lost on their way to ionization), It was demonstrated that from these detected PSL spheres spectra could be obtained. This proved that these spheres were coated indeed and could be analyzed individually by the BSQ system. The spectral data contained the typical polystyrene spectrum with equidistant peaks at the correct M/Z locations.
Figure 45: Histogram of particles, showing biomodal partcile distribution. The left peak represents particles resulting from empty droplets (only non-volatile MALDI matrix material, diameter 1.2 µm), the right peak represents particles from droplets that contain a PSL 2 µm sphere.
Figure 46: Mean spectra for particles that result from empty droplets (‘empty partícles’, green curve) and for particles that contain a 2 µm PSL sphere (blue curve). Note the typical polymer spectrum in the blue curve, in this case corresponding to polystyrene.
Next, PSL tests were done, where small amounts of insulin was added to the analyte. The PSL spheres absorb the insulin from the analyte, which could be shown by the results of the test: only particles that contained a PSL sphere showed insulin peaks. BSQ considered such analyte where insulin is absorbed by PSL spheres as a simple cell model for end-to-end verification of the BSQ instrument.
Figure 47: Mean spectra for particles that result from empty droplets (‘empty partícles’, green curve) and for particles that contain a 2 µm PSL sphere (blue curve). Note the typical insulin peak at 6000 Dalton in the blue spectrum, illustrating that all insulin is absorbed by the PSL spheres.
MALDI MS spectral responses cell life cycle phases
At the start of the SICTEC project it was believed that for identification of bacteria from individual spectra so-called ‘biological noise’ calibration curve was needed. It was decided however, that a statistical approach for the data analysis is needed to deal with this type of noise. The scope of the work in this work package was initially focused on cell populations that were all in the same life cycle phase. After many discussions with potential end-users it appeared that there was much more interest in what the performance was of the instrument to discriminate between large sets of clinically relevant micro-organisms, see WP’s 3.7 and 3.8
AST development
The purpose of AST (Antibiotic SusceptibilityTesting) is to supply data in addition to the identity of a microorganism that describes for which antibiotics the organism is not susceptible. This is to allow targeted treatment with antibiotics, thus reducing the mis-use of antibiotics as a result of ‘blind treatment’. Currently, a lot of research is going on what spectral biomarkers can be used to perform AST using MALDI TOF mass spectrometry. One approach is to look for specific M/Z peaks in the spectra that represent beta lactamase, where the bacteria breaks down the antibiotics structure by hydrolysis. Unfortunately, this is just one mechanism of resistance, so more biomarkers should be identified to have a clincially revelant solution for AST using MALDI MS. In the SICTEC project we selected an approach where a database was to be constructed that could be used to discriminate between susceptible and resistant strains of one species. The BSQ single-cell MALDI-TOF MS system employs an alternative matrix (2-mercapto-4,5-dimethylthiazole) compared to the matrices traditionally used for identification of microorganisms with MALDI-TOF MS systems (e.g. α-Cyano-4-hydroxycinnamic acid). Using this matrix it is possible to discriminate between bacteria of different species. However, it was not clear which taxonomical resolution can be achieved using the BSQ 2-mercapto-4,5-dimethylthiazole matrix. To investigate the taxonomical resolution potential of this matrix, a collection of 96 wild-type Staphylococcus aureus strains were analyzed on a (classical) target-plate MALDI-TOF mass spectrometer. Fortunately, it appeared that the BSQ MALDI matrix offered the potential for sub-species characterization. With the potential of the BSQ MALDI matrix confirmed, the next step was to confirm this potential using the single-cell MALDI-MS instrument (see WP 3.8). Prior to the start of SICTEC project it was believed that producing meaningful Single Cell MALDI spectra needed ‘just in time mixing’ of additives to the analyte. The task of such additives was to perforate the bacterial cell walls in order to make proteomic molecules available for ionization. Fortunately, this pretreatment appeared not to be necessary. Good spectra could be produced without it, see for example a spectrum for E.coli. The quality of this spectrum is comparable to that of classical MALDI TOF instruments.. Moreover, in case spectra showed insufficient detail (see above) the solution always appeared to be another morphology of the co-crystallized bacteria - MALDI matrix.
Data-processor development
As already mentioned, the classification of single-particle spectra is based on a (BSQ patented) statistical approach, not on the foreseen ‘biological noise’calibration curve. So, for each single-particle spectrum a ‘score’ is calculated for a limited number of ‘spectral features’. Based on these scores a measure is derived for the probability that the spectrum results from the MS analysis of each of the organisms that comprise the reference library. The scope of this work package covers the design, test and evaluation for the BSQ data analysis module. In the Figure below an illustration is given of the stochastic character of single-particle spectra. In this figure the spectrum on the bottom line is an E-coli single-particle spectrum. On the top line the accumulated spectrum of 200 spectra is shown. In between, the number of accumulated spectra grows from 5 to 45 in steps of 5 spectra. Note that despite large variations on single-particle basis, the peak location is recognizable in the single-particle spectrum.
Figure 48: Accumulation of single cell spectra. Not that a lot of clutter has disappeared already after just 10 - 20 spectra.
The design of the data analysis module is based on the Bayes’ theorem. This is briefly explained.
When we know the identity of the bacterial cell, we know the probability of finding a large ion intensity (a peak) at a certain mass, based on measurements of spectra of different species in advance. An example of such a probability density function (PDF) is shown in the figure below for a simple binary mixture. For a certain feature (intensity at location x) the PDF for particles that contain the feature-inducing substance (blue) and the PDF for particles without the feature-inducing substance (green) are shown. The latter shows a sharp peak at 0 (no peak at location x), whereas the first PDF shows elevated values for normalized feature intensities up to 1.5.
Figure 49: Feature probability density function
Unfortunately this is not what we need. Instead of knowing the probability of finding a certain intensity when we know a substance is present, we need to know what the probability that a substance is present when we measure a certain intensity. This is where Bayes’ theorem comes in handy; because it performs exact this translation of the probability of finding a certain intensity when we know a substance is present into the probability that a substance is present when we measure a certain intensity. To use Bayes’ theorem, we need the probability density functions for all reference spectra: the contents of the ‘reference library’. To test and validate the data analysis approach, an experiment was done where a set of one E-coli strain and two Serratia m. strains were used as reference collection of bacteria. The Serratia m. strains were harvested in a nearby hospital during a suspected outbreak.
To demonstrate the performance of the data processing a two-stage approach was followed:
1. Construction of a reference library for the E-coli and two Serratia m. strains
2. Classification of the single particle spectra of these strains
In this case we constructed a mixture in the computer using single shot spectra of the E.coli and Serratia strains recorded earlier. This procedure allows constructing an artificial ‘mixture’ with known composition. For all of the necessary steps of the data analysis Matlab routines have been written and verified. In a first step, features (peaks) in spectra of the E-coli and two Serratia m. strains (reference organisms) were identified. The picture shows the accumulated spectra for the E-coli and Serratia m. strains, where the features (peaks) are indicated by “o” symbols.
Figure 50: Accumulated M/Z spectra for E-coli and two Serratia m. strains
Note that E-coli peaks are easy discernible from Serratia m. peaks, whereas Serratia m peaks are hardly discernible from each other. The classification result and performance using all features is summarized in the table below:
Table 9: Classification result and performance using all features
The table indicates that good discrimination between E-coli and Serratia strains can be obtained. Mutual discrimination between Serratia strains must be improved, which can probably be achieved by a selection of features. When we use the single particle spectra that were classified from the mixture as E-coli to construct a mean spectrum, it could be demonstrated that this reconstructed mean spectrum is mimics the E.coli reference spectrum to a high degree of accuracy.
Figure 51: Reference spectrum E-coli (top) and E-coli spectrum retrieved from mixture (bottom)
Preclinical testing - The scope of this work package comprises a first pre-clinical test series to evaluate the functionality of the BSQ instrument. Though the number of organisms used during this test exceeded 100 isolates representing 10 species, this number is far too low to consider this test as a full preclinical test. Hence, the test executed in this work package should be considered as a first step towards a full verification and quantification of the BiosparQ functionality. All tests were performed using organisms supplied by University of Applied Sciences in Leiden. The organisms used were chosen based on their clinical relevance, see table. For each species 1 to 30 strains were used.
Table 10: List of clinical relevance
The bacteria were cultured overnight in a liquid broth. From each culture, 250 ml was taken washed in demi and pelletized by centrifuging the suspension at 10000 g for 5 minutes. Finally the pellets were frozen at -20o C, for use at a later moment. Just before use, the pellets were suspended in the BSQ matrix solution and approximately 50 ml was transferred to the BFX CS printer. As indicated above the classification algorithms are based on a statistical approach. Input for the algorithms are normalized probability density functions (pdf’s) for the ion count at the masses corresponding with characteristic features (peaks) of the spectra. For each of the above organisms characterizing peak locations were determined, and at all peak locations for all organisms pdf’s were recorded. At least 1000 single particle spectra were used to characterize the pdf’s. To evaluate that the BSQ instrument is able to identify micro-organisms, accumulated spectra were produced for all organisms mentioned in the table, including a number of repeats, using at least 500 single-particle spectra. Minimum requirement for identification is that the spectra stemming from different organisms from the same species are stronger related than spectra stemming from organisms of different species. In other words, in a hierarchical clustering scheme, the accumulated spectra should cluster according to their species. In the figure the functionality of the BSQ system for a subset of the organisms of the table is illustrated in the form of a combined dendrogram (left side) and heatmap of the spectra (middle).
Figure 52: dendrogram (left, each line representing a single strain) and heatmap (middle, the red lines indicating the peaks in the spectra ) of the accumulated spectra for 5 species (indicated on the right) of the organisms listed in table.
As the figure shows, for this set of organisms, the different spectra are clustered according to their species, indicating that identification is possible. However, the figure also shows that the inter-test variability, resulting from repeats with organisms stemming from the same culture, is relatively large compared to species variation. Hence, for species more closely related than those present in the current set of micro-organisms, or for sub-species, the identification capability is not yet self-evident. At this moment it is not certain what the cause of the large inter-test variability is. Since the samples used for the repeats stem from the same culture, differences can only be introduced during the washing step, the resuspension step or the dispensing step. Hence, the current research is aimed at improving and quantification of these steps.
Clinical benchmarking
Like in the case with the pre-clinical test, it was not possible to perform full clinical testing within the scope of the SICTEC project. However, a big step forward could be made if the clinical test demonstrates significant clinical value of the combined BFX-BSQ system. Showing this to potential customers and/or (corporate) investors is a necessary step to bring the company to the next stage. To create the most value from a limited clinical test, a concrete clinical application had to be selected with great care. To demonstrate the capability of the instrument to the market, the application should require one or more of the unique selling points of the platform, where at the same time this application should be technologically “low hanging fruit” aimed to mitigate the risk of failure. With the SICTEC partner UAS-L and some clinical end-users in its network (Erasmus MC - Rotterdam, Maasstad Hospital - Rotterdam, Diaconessenhuis - Leiden and Bronovo Hospital - The Hague), workshops were organized to select the clinical application(s) to be tested. The outcome of the workshops was that the potential clinical applications for the platform are twofold, ie (i) diagnostics of infectious diseases and (ii) infection prevention / outbreak management. Based on the outcome of the workshop first a test plan was drafted in cooperation with the SICTEC partner UAS-L and its network to demonstrate the capabilities of the combined BFX-BSQ system to provide the infection control team with relevant clonal information of bacterial strains (so-called “wild type strains”) isolated from patient samples collected during actual (potential) outbreaks from 3 different hospitals. With the potential of 2-mercapto-4,5-dimethylthiazole as a BSQ MALDI matrix confirmed see WP 3.5) the next step was to confirm this potential using the single-cell MALDI-MS instrument. Firstly, the identification capability of the single-cell instrument had to be confirmed, using clinical samples. For this purpose, the same wild-type MRSA samples were used as for WP 3.5 and a set of 20 wild-type Saramis marcesens (received from the Diaconessenhuis Leiden Hospital) were analyzed. Minimum requirement is that it is possible to identify these samples using a set of previously recorded spectra of reference strains (containing Staphylococcus haemolyticus, Staphylococcus aureus, Enterococcus faecium, Enterococcus faecalis, Pseudomonas aeruginosa, Enterobacter cloacae, Klebsiella pneumoniae, Escherichia coli and Serratia marcescens strains). It appeared that both the MRSA wild-type sample spectra as well as the Serratia wild-type sample spectra cluster in the group with accompanying reference-strain spectra. Thus it appears that identification of unknown wild-type samples using the single-cell MALDI-MS technology is indeed feasible. Secondly, the taxonomical resolution potential is tested. For this purpose two sets of samples are used, one consisting of the MRSA samples used above and one consisting of 10 Serratia marcescens samples retrieved during a suspected outbreak in the Diaconessenhuis Leiden Hospital. In contrast to the results presented above, this time no clear clustering was obtained. Hence, it appeared that the spectra either contain too much clutter to enable clustering at a level below species level or that the instrument was not stable enough. Indeed, while repeats of this analysis yield the correct clustering of the MRSA strains within the aureus cluster, the inner cluster hierarchy of the MRSA strains was not constant. When wild-type Serratia marcescens samples were analysed the spectra again just like with MRSA a mix of conserved peaks and sample specific peaks, see figure below.
Figure 53: Mass spectra of ten Serratia marcescens samples, retrieved during a suspected outbreak and one Escherichia coli spectrum (for reference)
In contrast to the clustering of the MRSA set, in this case a meaningful clustering might be possible. A parallel analysis of the Serratia marcescens using AFLP indeed yielded results comparable to those obtained with the BSQ instrument. Since the experiments performed earlier on a traditional MALDI-MS instrument did yield meaningful typing results, the inconsistent typing results obtained with the single-cell MALDI instrument are not caused by the matrix used. Apparently, the variability of the particle morphology currently obtained is not sufficient for deeper analysis than identification on species level.
CE IVD preparation / planning
A detailed plan for CE certification according to Directive 98/79/EC on in vitro diagnostic medical devices was drafted. As BSQ was initially not familiar with this topic, help was requested from a consultant who was expert in the filed of IVD CE certification. This consultant, Quality Made Esay (QME) set up the structure of the document which was subsequently detailed by BSQ. Although the BiosparQ IVD device is not subjected to list A or B of the IVDD 98/79/EC, BSQ decided to set up an ISO 13485 Quality management system which will be assessed by a Notified Body. This Notified Body then annually assesses the quality system for continuing conformance. The following table shows a complete set of harmonized standards that BSQ has already identified and will certainly be considered.
Table 11: complete set of harmonized standards
In addition to the plans for CE IVD certification, the document was completed with exploitation planning. CE IVD certification is an important step towards exploitation for BSQ, it will open the possibility to market the spectrometer as an In Vitro Diagnostics medical device. Before CE certification, BSQ’s instrument will be marketed as a device “for research use only (RUO)” according to paragraph 5 of MEDDEV 2.14/2. This market is substantially smaller than the IVD market, however this market is important to generate and disseminate data that can be used in future commercial marketing activities. The overall timeline and milestones for CE IVD certification and exploitation are shown in the table below:
Figure 54: BSQ overall time line and milstones for exploitation
Typical sales numbers are as shown below.
Figure 55: BSQ projected sales numbers
IVD liquid handling platform - A CE-IVD certifiable liquid handling platform will be designed, built and evaluated both mechanically but also by comparing it to other existing CE-IVD devices. With an existing device from PASCA/PDG forming the base for this work, the first tasks were to identify the critical points in the machine design and based on that to optimize it in special regard to manufacturability, robustness, and costs. As already mentioned in WP2, the help of an external consultant (Manfred Augstein from velixx GmbH) is also used in this work-package. His knowledge helps ALU-FR in CE-IVD related design and implementation questions. The PASCA projects allowed to finalise a CAD design of an optimized and customer- attractive platform that could open the route to an industrial and CE IVD certified equipment. The work carried out in WP4 therefore aimed to start from finalized CAD, in order to establish the related 2D drawings, to produce the associated parts, to supply the required components and to build a prototype device, which would be functionally validated and then use for a clinical study. The aim of the clinical study being ultimately to benchmark the performance of the platform against the internal workflow as implemented at Erasmus MC.
Defining the robotic base from the PASCA CAD design - The CAD design obtained at the end of PASCA by PrimaDiag was reviewed by IMTEK which led to a series of minor changes. These were integrated in the initial CAD design, which was then use to establish the related 2D drawings. All standard components which exists already were bought. Non standard part (such as mechanical part and electronic wiring) were either subcontracted to specialized factories or made in house.
Figure 56 : Assembled base of the WP4 prototype.
Defining the required modules - The required modules to be added to the robotic base were selected in accordance with Erasmus MC. The major problem that arises was that the reference interlocutor changed every 9 months, obligating to modify the specs and the protocol that was requested to be automated. The interlocutors have been sequencially prof. Willem Van Leeuwen, Dr Rene Then you Wit, Then Wil Goessens).
In March 2015, two protocols were finally proposed by Erasmus. One directly related to Roche Magnapure system (which does automated DNA extraction from diverse starting materials : blood, urine, culture broth...). The main problem with this being that the reagent are proprietary and included in pre-packed cartridges which can hardly been used on the "WP4" instrument.
The second proposal was to use XNAR reagent kits from Sigma (ref : E7526, D5688, E3004) which are available as manual kits, and hence are more easily automatable.
Protocol derived from Roche Magnapure requires a pipette, a magnetic bloc and an 25-90°C incubator.
Protocol derived from XNAR reagent kits required no DNA purification by magnetic beads, and hence can be done with the pipette and the incubator only.
Figure 57: Selection of the modules based on the analyses of the biological protocol
Figure 58: Assembled prototype WP4 with its 3 modules.
Specs deriving from the analysis of the regulation - In addition to the talk which was given by Mr Manfred Augstein from Vellix, which was mostly focused on methodological approach, PrimaDiag followed some additional support from C-Reg Medical with gave more practical advices on the development that were required on the instrument in order to comply the 98/79/CE directive (CE IVD).
In particular, it was requested that :
1 ) the samples should be protected from contamination from the outside. Reversely, the operator should be protected against possible contamination from the samples (virus, bacteria, etc.). At a minimum, the system should then include a enclosing cabinet. With the possibility to include an HEPA filter.
2) cross contamination between experiments made in separate runs should be avoided. Hence a UV decontamination system should be included.
3) proper traceability should be insured. Hence, the final system should at a minimum include a barcode reader, if not a 2D matrix reader.
4) Pipette should include some level sensing capabilities in order to control the presence of the sample and the reagents when the process is running. Ultimately, the system should also be able to monitor the volume of liquid that will be / will have been handled during the runs.
Performance optimization - In this period has the positioning accuracy and the modularity of the WP2.2 prototype been optimized. In parallel, the prototype assembled in WP4 was functionally validated. As already mentioned in WP 2.2 the accuracy of the device has been improved by using alternative ball screws. The ball screws from HIWIN are available in different quality grades. The highest precision ball screws have at a travelling distance of 300 mm a maximal difference of only 3,5 µm. Another advantage is that the accuracy of the device can be adapted without changes in the design. After the manufacturing of the device the position accuracy will be tested with two different tests as mentioned in WP 2.4. One general test is oriented at the VDI/VDE guideline 3712. The scope of this test is the dispensing of a set of droplets at defined positions. Based on VDI/VDE guideline 3712 was a dispenser test plate designed (see Figure 59) which is going to be milled in aluminum.
Figure 59: Constructed dispenser test plate after VDI/VDE guideline 3712
For the dispensing test the test plate is placed on the working plate of the SCP and aligned optically by three alignment structures. On this plate then 50 droplets are spotted with ink at defined positions. These positions force the prototype to move different distances and into different directions. Afterwards the positions of the spotted droplets will be checked with an PC- DMIS measuring microscope. The differences to the intended positions show the accuracy of the whole dispensing system. The second test focuses on device positioning and uses a USAF Resolution Chart to check the position optically. This chart has several lines with different distances in between. Here the USAF resolution Chart is fixed to the working plate and the position of the Print head is checked with the camera of the SCP. The reticle of the camera is directed between two line pairs and the coordinates are noted (see Figure 60). The SCP now is moved ten times to a different position and back to the coordinates between the lines, where the camera takes a picture. On the pictures is afterwards checked, if the reticle is between the lines. Because of the different distances of the line pairs, it is possible to check the accuracy on different size levels. With the implemented optics, it is possible to see the line pairs down to a distance of 20 µm. The test was already performed at an existing dispenser of BioFluidix.
Figure 60: Reticle of the SCP camera placed between the 30 µm lines of the USAF Resolution Chart
In the new prototype are several points modular. The design is modular in the way, that it has defined interfaces, where the look can be changed. For example the sidebars can be exchanged or the look of the base platform can be adapted. The number of z-axis can also be changed between one and two.
The next modular point of the prototype is the type of Print head. As Print head either a commercial pipet, the PASCA single cell dispenser or a PipeJet dispenser device of Biofluidix can be mounted. The layout of external plugs is also adaptable, by changing a cover plate at the backside and the working plate. For example if no stroboscopic setup is needed, a blank cover plate can be used on the working plate. On its side. PrimaDiag validated the WP4 prototype that was built in WP4.1 which was obtained by assemblying the mechanical parts and components described upper. Thanks to the improvement already obtained with PASCA, the initial assembly was almost 100% satisfying. After a set of minor weaknesses being identified and corrected, the system was ready to accept the modules required to fulfill the Erasmus protocol (ie. Pipette, Magnetic Bloc and Incubator).
IVD specific design - For this we have started to check the mechanical and the electronic parts for compliance with the CE IVD directives on the prototype developed within WP2. For what concerns the prototype which was built within WP4, the works are related to the development of the enclosure cabinet, the UV lightning and the barcode reader. Ultimately, it would have been related to the integration of the level sensing in the pipette and on the HEPA filter. For what concerns the robotic base developed within WP2, the axis system has been reworked compared to the existing prototype. As already mentioned in WP 2.3 in the next prototype very accurate ball screws from HIWIN are used. The axis suspension is also changed. The motor and the screw are separated from each other and the forces are transferred from the motor to the axis by a belt. The belt decouples the motor and the screw and the axis has more traveling distance and is less susceptible to vibrations and torsion. Along CE IVD directives it is mandatory to prevent any harm from the user when operating the device. This especially refers to mechanical moving parts and electrical safety. Sharp edges are removed and bellows are used at the X-Axis to cover the linear ball guide. Cable routing was optimized and partly reconstructed to avoid damage by mechanical stress, moving axes and also to waive any external cable ducts. The cables are guided by drag chains and rooted inside the side bars (see Figure 61).
Under the electronic aspects of CE IVD, we implemented an emergency switch and all internal DC voltages are under 48 V. It was also paid attention that every part is grounded correctly, in order to ensure, that the device can pass the EMV Test. Further information on the design is also provided in Deliverable D2.1 “Finalized prototype design”.
Figure 61: Cable routing in the sidebars and drag chain. Internal cable management using drag chains to proper guide cable bunches besides moving axes parts. To improve visual appearance of the device all cables from Y-bridge and print head are internally routed through the side panels.
For what concerns the prototype built within WP4. The enclosing cabinet was designed, produced and assembled in a similar way to what is described earlier (assembly of the robotic base). The UV decontamination system was integrated using components available commercially. Proper measurement of the UV flux was made in order to insure the efficiency of the system. Reversely, absence of leaking UV outside the enclosure cabinet was also checked.
Fabrication cost optimisation – During this period the prototype has been designed with and we paid attention on lowering the cost. A quality test for the accuracy of the prototype has also been prepared. The design of the first prototype was done with special attention of cost reduction. This includes that standard parts were implemented if possible. For example, most of the design is based on item profile bars or alternative standard aluminum profiles, like L- or U- profiles. The plates are - if possible - manufactured only from the front or back and have a maximum thickness of 10 mm. Under these circumstances the plates can either be milled by a front panel milling company or manufactured via Laser/ Waterjet cutting. Most of the smaller design elements are planned as 3D- printed parts because of the low costs and the high variety of possibilities. Bending parts are recommended due to high fabrication costs and non-optimal precision. Cost reduction is also applied for the electronic parts. The motor controllers will be changed to PiBot from Repetier. This motor controller is used for self-made 3D- Printers and costs, including three stepper motor drivers and three optical end switches, less than 100 €. Special attention was paid on assembly of the final prototype. With regard to easy exchange of spindles and motors the Y-bridge was reconstructed. Electronics and embedded computer are now easily accessible and exchangeable through a double bottom design with plug and play connectors. For the quality test of the final equipment, the two accuracy tests introduced will be used. These tests are explained in WP 2.4 and will be used to confirm the technical specifications of the accuracy. One test is used for the accuracy of a whole set of dispensing steps and the other for the accuracy of the device positioning. On its side, PrimaDiag worked on the prototype developed within WP4. Each component and milled parts was reviewed in order to optimized the cost. The following table shows the price variation before and after cost optimization :
Table 12: cost optimization summary
Hence, the cost reduction process succeeded in dropping the platform price by 39%.
Documentation - CE IVD regulation is very strict and requires a precise definition of the product, how it is built, how it is used and what kind of results it can provide when running in routine. Also, this stage requires heavy preparation of a large range of technical files, user manuals and reports related to the systematic or standard tests. Obtaining the CE marking is the ultimate goal of this sub-WP. Writing and compiling the required documentation was not technically difficult, but is only a tedious process. When achieved, the compilation of the whole documents led to the writing of deliverable D4.1 ; and the machine was passed to BT and CEM normalized testing. Thanks to the preliminary works done already within PASCA, the WP4 prototype passed the norms without difficulties. The CE mark was then obtained on July 2014, just as RP2 started.
Pre-production instrument fabrication - After the WP4 prototype being produced and validated, two additional pre-industrial series had to be built, and then used respectively by PrimaDiag for additional characterization and development ; and by Erasmus MC for biological pre-validation and then clinical trial. The amount of fabricated instruments has been reduced. The instrument for BFX is no longer needed. Instead, one prototype based on the Biospot was. In advance to the assembly of the prototype, Dirk Buselmeier went to Primadiag in Paris and had a look on an already assembled device. During the visit, an assembling plan was created in accordance with the PDG team. This plan was created, in order that Dirk can assemble the prototypes on his own in Freiburg. For the communication and organization of the project, a Telephone conference between ALU FR and PDG was held every two weeks. After the arrival of the parts, the fabrication of the instruments started. The bridge of the instruments was assembled first (see Figure 62), followed by the base. This two parts have then been merged together (see Figure 63). To the bridge was added the X carrier with the Z-axis. The last mechanical part was to implement the working plate with the heating module. After the completion of the mechanics have the motor controllers and the power supplies been implemented. The electronic devices have then been connected and the cover plates been added. After the assembly were the instruments shipped back to PrimaDiag where the software was added. The plans were that one instrument would have stayed in Paris, on which further development will be made. The second instrument will have the CE IVD modules added (eg. the hood, the UV and the barcode)before it would be shipped to EMC for the benchmark.
Figure 62: (ALU): Assembled bridge.
Figure 63: (ALU): Bridge and base combined.
Figure 64: (ALU): Two assembled prototypes.
Unfortunately, the two units which arrived in PrimaDiag were not functional. And both units needed to be fully dismounted and remounted before being usable. The two units were then only available on end of April 2015. And this made the development of the HEPA filter and on the pipette (level sensing) difficult to achieve properly in the remaining time (2-4 months max).
Pre-production instrument characterization - Validating the prototype, and then the two pre-industrial series before launching the clinical trials. With the periodic changes of principal investigator at Erasmus during RP1, PrimaDiag took the liberty to contact some hospitals in its neighborhood (in the Paris region) and start the biological validation of the prototype on the field. Taking into account device capabilities (incubation, DNA purification and pipetting), it was proposed by the academic partners to use the DNA library preparation protocols upstream the NGS sequencers that would be more innovative and interesting when compared with the simple, 90's old fashion, DNA extraction protocols foreseen initially. The idea of using PrimaDiag system for such protocols was already identified as a good opportunity since the end of 2013. For review, one can refer to the final report of the PASCA project. Anyhow, this work around automating the NGS protocols started in June 2014 and lasted for the first campaign up to September 2014. This works were done in partnership with Genomics Department at the Brain and Spinal Cord Institute (La pitié Salpétrière Hospital) which confirmed the feasibility of the approach. A second campaign was then started with IntegraGen, a service provision company with sequence complete genomes and exomes for thirds parties. With them, several kits for preparing DNA and RNA library preps coming for some major reagent manufacturer (Illumina, Kapa, New England Biotech) were tested and validated. From October 2014 until January 2015, PrimaDiag then had the chance to collaborate wirth Roche Diagnostic which also has some reagent for the NGS sector. PrimaDiag system showed excellent performance for automating these kits, and this led Roche to propose to PrimaDiag to enter into a commercial agreement by April 2015 (see details in WP8 – dissemination). Taking into account the delay required to set up the partnership between Primadiag and Erasmus on RP1, and then the problem arising with the non functional pre-industrial series, it was not possible to prepare the plateform in time in order to program Erasmus prototol and then send it for biological pre-validation, and then the clinical trial.
IVD panel selection - Define the strains on which the biological prevalidation would be run. Then define the strain on which the benchmark and the clinical trial would be made. A first set of 8 strains were selected by Rene te Witt before he left on November 2014. But much information was lost with his depart
Mono-clonal cell culture - Following the same strategy used in the first period, INNOPROT has continued developing cell based products. As already commented the cell lines/models developed through this methodology have a very clear competitive advantage; they are truly monoclonal. This characteristic makes them more homogeneous in their response and the performance of the assays would be better. The SCM technology has been incorporated to INNOPROT production protocol and has been also included to improve existing cell products by “cleaning” them in order to make them monoclonal. As a result we have launched 22 new cell lines Out of these, 15 cell lines belong to the new proprietary technology developed by INNOPROT; NOMAD technology. NOMAD is a new family of biosensors designed to determine the GPCR activation through the β-arrestin activity or measuring changes in the concentration of second messengers within the cell. This new technology seems very promising and many potential customers have shown their interest. This is the complete list of cell lines developed during this second period.
List 2: developed cell lines
Apart from these cell lines, Innoprot has used SCM prototype in another 23 cell lines developments that have not come to an end due to not accomplish our quality standards, 20 cell lines produced for internal R&D projects and up to 50 cell based models required on-demand by our customers or included in other public funded projects. The data sheets of most of these products are available through the website www.innoprot.com. The improvements carried out in the SCM prototype included; a new printer head (component P9), and improved chip manufacture and reliability and an increased software stability. Overall this has represented a 5.4 fold increase in the cell lines produced per year with respect to the previous PASCA project. However this would be even better if would have had the fluorescent capabilities in the SCM because this would allow us to improve the positive clones sorting yield.
Develop an in chip drug screening assay - To use the fluorescent capabilities of the SCM to develop an “in chip” assay. This would be very useful in the validation phase of the cell based model production, saving time developing sub-optimal clones and saving the misuse of resources (the use of the screening platform for clone validation).. Moreover this could be also used for selecting clones with different degree of response to a given stimuli (ie. drugs, agonists or antagonist). Finally we would explore this feature for pre-screening purposes. This could be used for preliminary efficacy-toxicity assays in order to identify promising drug candidates or to discard molecules in labs lacking the resources of an expensive drug screening platform, or even to screen molecules in rare or low abundant cells. Unfortunately, the SCM with the fluorescent capabilities suffered several delays and currently it still does not work as expected. Thus, we could not use this interesting feature yet and therefore, the In-Chip drug screening development was not accomplished. The fluorescent excitation and detection was successful with both tGFP expressing U2OS cells (53% fluorescence intensity compared to the beads) and with fluorescent beads. However we wanted to use cell lines carrying a recently launched proprietary technology (NOMAD Biosensors Technology) and these cells are much less fluorescent (0.96% fluorescence intensity compared to the beads). It was not possible to detect the NOMAD cells with the SCM using fluorescent detection and therefore, we had to limit our tests to a high signal cell lines. On the other hand, we could not dispense cells detected by fluorescence. After trying all the suggestions and recommendations made by IMTEK, we were not able to dispense cells on a regular basis. The system excites certain spot with the laser and then detects cells compatible with given parameters (that is, cell size, roundness, fluorescence and so on) looking to a specific region determined previously (Region of interest, ROI). What was happening (and it is still unsolved) is that these two regions need to be aligned in order to decide whether a fluorescent particle within the ROI fulfils the parameters. The problem is that when you start a run, the printer head moves to the first well position (A1 in a 96 well plate) and starts to shoot drops, somehow this alignment is lost and therefore the ROI is not illuminated by the laser anymore and usually the ROI goes to somewhere else within the chip (many times even outside the reservoir in the plastic part). As a consequence the system does not find any particle fulfilling the given parameters and the dispensation does not take place. The figure 5.1 shows one example of this misalignment. This big problem is still ongoing and IMTEK could not addressed it due to the delay in the project. In addition, we needed a modification in the core.exe file in order to be able to measure the cell’s response to a given treatment. The first two modification of the original file did not work in the system and again most likely due to time constrains IMTEK did not provide us more versions to try. Nevertheless, it seems promising and when the last pitfalls would be overcome, the proposed assay could be developed in relative short time.
Upgrade of SCM machine - The original plan included a SCM prototype developed during the previous FP7-PASCA project, modified in order to be improved with fluorescence measuring characteristics. This feature to be developed and improved throughout SICTEC, was central for INNOPROTs activities. However, as already discussed previously in the PR1, this task was delayed due to difficulties in their development. Therefore we received the prototype in our facilities at month M20. On top of this, due to several problems (laser spot-ROI alignment, red led knob replacement, core modifications) we could not achieve some of our goals such as increasing the positive clones yield by selecting the fluorescent cells, sorting clones with different response levels, and developing an “in chip” assay.
Figure 65: Dispensation of one fluorescent nanoparticle. Picture sequence showing a single event dispensation of one fluorescent nanoparticle. In the panel 3, is clearly visible the misalignment of the ROI respect to the laser spot. ROI (Region of interest) is highlighted as a white circle line. The dispensed nanoparticle is highlighted within a white line contour. Panel 1; the particle is approaching the nozzle, Panel 2; the nanoparticle enters in the laser spot and is excited showing a green fluorescence, Panel 3; the system detect the fluorescent nanoparticle and verifies that meets all the set up criteria, Panel 4; the particle has been dispensed and is not visible in the image anymore.
Automated Single Cell Patch Clamping - The scope of this work-package is the integration of the single cell printing technology into Sophions single cell patch clamping platform. This was already planned in the former PASCA project, but could not be shown to work as a complete workflow: All individual steps worked from time to time, but not as a whole. Therefore, the work-package aims in increasing the overall robustness of the chips, the dispensers and detectors and the software protocols involved. Both the original PASCA SCM/Patch clamp platform and the new SICTEC Patch clamp platform require tight integration of IMTEK and Sophion hardware and software to work. Under work package 6 significant development efforts have been made not only at IMTEK but also at Sophion where new amplifier hardware and multiple new software components were required in order to allow integration with and evaluation of the IMTEK provided modules and upgrades in subsequent application work to validate these modules. This chapter therefore also contains brief descriptions of Sophions development and integration efforts as well as preliminary application evaluation. Apart from initial specification work - the Sophion SICTEC effort has been focused in the periods August-Sept 2014 and May-Nov. 2015. A comprehensive set of experiments on a first Single Cell Patch Clamping prototype (developed in PASCA and extended in SICTEC) are scheduled for August/September 2014. Therefore SPH, ALU-FR, and BXF agreed on a preliminary list of specifications that should enable a successful proof of concept. The results will be then used to further specify the next SICTEC prototype that will be designed.
List 3: Preliminary list of specifications
Remote controlled dispenser software - The SCM/Patch clamp application requires that the cell dispenser w. robot is operated in tight coordination with Sophion amplifiers and pressure systems to deliver, catch and characterize individual cells. The coordination and control is performed by Sophion software. The IMTEK software controlling the dispenser is completely separate from the Sophion software - so the execution coordination is achieved through Internet protocol based communication. The precise support of the protocols and the stability of both software throughout experiments lasting 30-60 minutes under a variety of usage patterns is critical to the success of the application which involve many complicated steps in addition to the required precise SCM cell delivery. The Internet protocol (IP) based protocols enabling the coordination between IMTEK and Sophion technology were upgraded in two rounds. In August 2014, new commands were added to the Biofludix PASCA/Biospot Software IP interface and the support for and usage for these commands were implemented in the Sophion Software. Those upgrades did improve the interoperability significantly - but could not solve some fundamental architectural flaws in the PASCA/Biospot software IP-communication design that caused it to crash often. In the second WP6.2 deliverable September 2015 the dispenser side a changed version Protocol was handled by a embedded process called Xcore - which controls the dispenser and the cell camera - while management of the XYZ-robot (holding the dispenser) and the TopView camera was taken over by Sophion software. This design was based on the assumption that the two software components could run on the same PC. The initial version of the new protocol was documented on August 14th and the final revisions were made on Sept. 10th. Development and testing of the new versions of the Sophion software interface layer and experiment execution coordination support for the new protocol was ongoing throughout August and September in addition to the new code required for Sophion to take over software control of the robot axis motion and TopView camera operation. This work has required a significant effort in man-months. Currently (early November 2015) the Sophion software and the IMTEK Xcore software and their respective USB and TCP/IP connected devices appears to have some apparently hardware problems co-inhabiting the same PC. In the currently most stable configuration the Xcore & cell camera connection crashes each time during the second cell dispense event. Measures to improve communication is from the Xcore to better allow Sophion to analyse and eliminate sources of instability. Significant improvements has been made to the IMTEK software’s ability to support the tight coordination of experiment execution through IP protocol communication - but the stability of the inter-operation has not yet reached a stability suited for mature product integration.
Automatic dispense parameter tuning - In order to allow robust automatic single cell patch clamping (WP 6) stable droplet generation is one of the key requirements. To minimize the necessary user interactions we will therefore implement an automatic dispense parameter tuning capability. For single cell patch clamping this should be done with respect to an optimal droplet placement precision. The basis for automatic dispense parameter tuning is a comprehensive picture of the dispensing process and the influence of the piezo parameters on the droplet formation – especially with respect to robustness and placement precision. As described in action report 6.3 we started to systematically evaluate the placement precision as function of the piezo parameters using a special test setup. We investigated the influence of the piezo downstroke velocity on the droplet placement precision. Although the other three parameters are important for the droplet formation they have a minor influence on the droplet placement performance once stable droplets are created. Hence they were set as follows:
Table 13: Piezo parameters used for stable droplet generation. The stroke length was set such that a stable droplet without satellites was created.
We found that the droplet placement precision improves with increasing downstroke velocity as long as a stable droplet without satellites is formed. This results in a reduced scattering radius as illustrated in Figure 66.
Figure 66: The scattering radius of the droplet placement depends on the piezo downstroke velocity. With increasing velocity the droplet placement precision improves. The graph shows data from two chips which were studied in detail. Medium was deionized water.
This trend was also observed when PBS was dispensed (using cartridges with a silane coated nozzle). We observed (and recorded) the droplet position data for 7 cartridges with PBS using high downstroke velocity values. In all cases a very good droplet placement precision as for deionized was found. As the optimal downstroke velocity varies between the chips we shifted our focus on the droplet velocity, which can be extracted from the image data provided by the stroboscopic camera.It turned out that all cartridges tested (with water and with PBS) showed a high droplet placement precision for droplet velocities of 1.5 – 1.85 m/s.
In a first investigation of the droplet placement precision we identified the relation of a sufficiently high droplet velocity and the placement precision. The droplet velocity can be extracted from image data from the stroboscopic camera, which is already part of the standard SCM. Therefore the droplet velocity might be an ideal state variable in terms of piezo parameter auto-tuning. So far we implemented the capability to automatically measure the droplet velocity into the SCM software. As the space of influencing parameters (medium, cell type, and environment) is large further evaluation of the placement precision is necessary.Finally we will introduce a feedback loop that allows automatic parameter tuning. Therefore it is further necessary to reliably detect droplet instabilities such as the formation of satellites with the stroboscopic camera. For full integration of the code we finally have to implement the auto-tuning module together with Biofluidix as they are in charge of the modified software for Sophion.
Sophion results - In SCM based patch clamping, reliable dispenser performance is a necessary - but far from sufficient condition for a even moderately successful experiment. Lack of chip droplet placement reliability is only manifested in the middle of the experiment execution - and typically means that 1-2 hours of preparation is partially wasted. On the other hand failures elsewhere in the setup (like software crashes) may prevent a flawless chip from proving its worth. We have tested multiple batches of silane coated chips since August 2014 with both versions of the SICTEC setup in various stages of development and instability. Is is hard to make statistical evaluations under such variable conditions - but generally the silane coated chips have been performing better than the chips available during the late PASCA work. Chips from some batches persistently performed well when given the chance - while the chips from other batches had a very mixed performance leading to many failed experiments due to unreliable droplet delivery. The not-well performing chips were often associated with priming issues and a narrow down stroke velocity window between droplet formation onset and satellite droplet formation. In the patch clamp application we will eventually both good droplet placement reliability for some chips - but also reliable performance from most chips and batches. This highlights the importance of automation of the dispenser chip production effort in WP1. No automatic dispense parameter tuning functionality or poor chip performance features were demonstrated to us or made available for evaluation at Sophion during the SICTEC project (or since). Thus - while it is good to read that IMTEK & Biofluidix is making progress in this field - we are not in a position to comment on the performance or SCM patch clamp application suitability of these features (which is the focus of WP6). Currently dispense parameter tuning is still being done fully manually in a user interface developed at Sophion since the Xcore software just provides a simple text-based IP protocol interface for basic parameter value changes. The improved reliability of the silane coated chips have meant that manual dispense parameter tuning has generally been easier than before.
Integrateable, contained dispenser module - Key for a functioning print head for single cell patch clamping are dispenser chips that allow depositing single living cells on patch clamp holes in a robust manner. Therefore, we first focused on improving the dispenser chips in terms of droplet placement precision and long term stability. This report summarized our recent work. The chips (and therefore also the nozzles) are rendered hydrophilic during the production process by an oxygen plasma treatment to allow reliable filling with media. During dispensing the hydrophilic surface properties leads to wetting of the silicon/pyrex chip around the nozzle. There are no problems when deionized water is used as medium for dispensing. However, if phosphate saline solutions are used, salt precipitates around the nozzle during dispensing. This results in a significant decrease of the droplet placement precision after 2-10 min of dispensing and often leads to complete failure of the chip. Therefore we developed a protocol for a hydrophobic coating off the nozzle while the hydrophilicity of the interior parts of the chips is preserved. We then investigated the effect on the chip performance (with focus on droplet placement precision). Further we started to improve the coating step in order to allow for robust integration in our chip supply line. Coating is achieved by vapour phase deposition of an organo-silane. Chip performance was studied by dispensing PBS on a glass slide. A camera is placed below the slide which allows to observe the dispensed droplets on the slide. 15 Cartridges with silane coated nozzles were tested for at least 60 minutes. Both the total failure and the droplet positioning accuracy were qualitatively compared with untreated chips. So far all of the 15 silane coated cartridges that were tested performed very well with PBS in terms of dispensing. All of them were used for at least 60 minutes of dispensing and none of them failed. Liquid film formation around the nozzle and therefore the precipitation of salt could be suppressed. The effect of the surface coating can be also seen in Figure 67 which shows a cartridge with an untreated (left) and a silane coated chip nozzle (right): After gently applying pressure on the water-filled reservoir, the uncoated chip is immediately wetted while on the coated chip a round droplet can be observed (high contact angle).
Figure 67: The effect of the surface treatment can be examined by a quick test: If pressure is applied on the water filled reservoirs, an untreated chip will be rapidly wetted by the water (left) while a droplet forms on the hydrophobic nozzle with coating (right).
At the moment we are working on the full integration of the coating step into our chip supply line. Next, we have to further investigate the shelf life, mechanical, and chemical stability of the silane coating.
In most of the applications (including single cell patch clamping) dispenser chips are required to work reliable for extended time durations (> 30 minutes). Therefore we performed long term dispensing experiments with 10 µm fluorescently labeled beads and PBS using three uncoated and three silane coated dispenser chips. Arrays of 34x70 beads were printed on microscope slides which were subsequently analyzed with a microarray scanner. All three uncoated chips failed completely within one hour and thereby only about one half of a 34x70 array could be printed. Further, in some cases the placement precision rapidly deteriorated as can be seen in Figure 68 (left). In contrast, the silane coated chips allowed for extended printing times: We dispensed three arrays with each of them within 5 hours and none of them failed.
Figure 68: The non-coated chips all failed after printing about one half of the 34x70 array and in some cases the precision deteriorated rapidly (left). The silane coated chips dispensed beads reliably for 5 hours without failure. Here one of the full 34x70 arrays of single beads is shown (right).
In order to investigate the performance of the printer in terms of droplet placement precision water and phosphate buffered saline (PBS) were dispensed on a glass slide. Three cameras were setup such that (i) the position of the droplet on the substrate after dispensing can be recorded (bottom camera), (ii) the nozzle of the chip can be observed during dispensing (nozzle camera), and (iii) the droplet can be monitored by the stroboscopic camera. Both the bottom camera and the nozzle camera are triggered each time a droplet was dispensed and allowed to pass by the vacuum suction hole. Note, that the trigger was coupled to the vacuum valve opening event. The SCM software was modified such that the printer can continuously dispense at a fixed frequency (specified in the GUI). In order to evaluate the images produced by the bottom camera, the frequency of droplets arriving on the glass slide must not exceed 0.5 Hz to provide enough time for droplet evaporation. Therefore the standard experimental procedure was as following:
● Droplets were produced by the printer at a frequency of 5 Hz.
(Piezo shooting frequency = 5 Hz )
● Only every 20th droplet passes the vacuum suction valve.
(Vacuum valve actuation frequency = 0.25 Hz)
● Each time a droplet passed the vacuum suction valve images of the substrate and the nozzle were taken. The stroboscopic camera was used before and after the experiment in order to calculate the droplet velocity and to adjust the piezo parameters.
● For each parameter set 1000 droplets were dispensed. Of these, 50 droplets reached the glass slide and were monitored by the bottom camera.
● The working distance between the dispenser and the substrate was 5 mm.
The image data produced by the bottom camera are evaluated with Matlab. Figure 69 shows a droplet detected by the algorithm. The graph shows the x and y-deviation of 50 droplets. The droplet placement precision can be expressed in terms of the standard deviation , calculated from the positions of the 50 droplets. In order to obtain an intuitive measurement tool, we define the scattering radius
,
which indicates the radius of the circle that encloses 95 % of the dispensed droplets given a normal distribution. The scattering radius is illustrated by the red circle in Figure 69 (right).
Figure 69: The image data from the bottom camera is processed by a Matlab sript: For each experiment 50 droplets are detected (left) and their positions and the resulting scattering radius are plotted (right).
We evaluated the image data of 3 cartridges (representative for the 7 we tested). The resulting scattering radii are given in Table 14.
Table 14: Results of Droplet Placement Precision Measurements of three chips. Both with water and PBS a placement precision < 60 µm is reached repeatedly if the droplet velocity is set between 1.5 and 1.85 m/s.
We therefore think, that by using silanized nozzles we can reach a droplet placement precision of for continuous dispensing of PBS over extended times. Although we have not yet investigated the influence off idle time on the placement precision in great detail with the new test setup, we saw that the placement precision can be recovered after dispensing breaks of 1-2 minutes.
We will further examine idle times and use-cases reflecting Sophions workflow for patch clamping in the future. There has been the issue of sedimentation of cells in the reservoir of the cartridge, especially if cell dispensing was performed for extended time periods with the SCM. This led to a decrease in cell concentration over time. To fix this issue, an agitation module has been designed and tested at IMTEK. It consists of a microcontroller and a roller pump which agitates the reservoir via a disposable pipette tip To test the agitation module we dispensed microbeads for 30 minutes. Figure 70 shows that with the automated agitation module it was possible to dispense single beads with a constant frequency over time. The module will be integrated into the existing PASCA hardware and into future print head designs for patch clamping.
Figure 70: Number of single micro bead dispenses over time. A single column in the diagram corresponds to 1 minute. The concentration of beads arriving at the nozzle is constant over time.
Results at Sophion - No design changes to the PASCA dispenser module were made during the first SICTEC upgrade in August 2014. The redesigned dispenser module from Sept 2015 is discussed under redesigned WP6. The agitation module upgrade mounted on the enhanced PASCA SCM/Patch clamp setup worked in principle - but a mismatch between the rolling pump and the control electronics meant that it either sucked the cell reservoir dry or had to be run “in reverse” making correctly timed insertion critical. This lead to a lot of failures and meant that the module became a risk factor and was unmounted for many experiments. Extended software and hardware problems with the SICTEC 2nd generation setup has meant that we have limited experience with using the 2nd generation agitation feature.
Development of the next generation SICTEC patch clamping prototype - Major part of WP 6 as defined in the SICTEC DOW is a next generation single cell dispenser module (printhead) that will be specifically designed for patch clamping (see WP6.4 and WP 6.5) based on the experiences with the first prototype (PASCA machine) . Although proof-of-concept could not be fully achieved, the experiments using the upgraded prototype helped to define a set of major issues that will be addressed with the design of the next SICTEC WP6 patch clamping prototype:
● Software complexity must be further reduced, such that the user has to interact with a single GUI only.
● The necessary user interactions with the software must be further reduced as another step towards a fully automated solution.
● The complexity of the hardware must be reduced and the stability increased. In detail:
• A more robust amplifier system that does not need to be controlled from an additional PC.
• A more integrated printhead, that supports further automation such as dispenser parameter tuning, offset compensation and a fast & smooth single cell dispensing.
• A new robotic system that is mechanically more stable and can be controlled easier than the old PASCA setup which was reconstructed several times (as stated by Sophion).
Given this basis for the design of the next generation SICTEC patch clamping system and the remaining time of the project, WP6 was redefined by ALU-FR and Sophion. In summary, the actions in the reporting period regarding the development of the next generation SICTEC patch clamping prototype included:
● Redefinition of WP6 together with Sophion.
● Preparation of formalities for the EU with respect to the WP redefinition (ongoing).
● Specification of the next generation SICTEC patch clamping system (ongoing).
The Sophion evaluation of the first SICTEC delivery revealed that the upgraded PASCA platform based on the complex ino teroparation of 3 PCs had become too complex and unstable to be usable in practice: The 1st generation Robot SCM05 had precision and wiring stability problems - the latter due to multiple repairs and fixes made during PASCA. Swapping it with a more stable and precise PASCA 2nd generation robot was not an option and Sophion had no other suitable candidate for a a XYZ robot that could carry the dispenser module under development at IMTEK. It was agreed that IMTEK would find a tabletop type XYZ robot controllable by Internet Protocol (IP) and that Sophion would take care Software
control for the robot, the topview camera and ensure the development of a tabletop friendly
amplifier system. These development tasks required significant development efforts at Sophion in the period May-September 2015. It could be realized due to synergies with another Sophion EU project (HiMicro). The Festo XYZ axis sold to IMTEK as IO controllable turned out not to be and resulted in 25+ man-days of wasted development effort. Some specifications came rather late but Sophion had almost all components ready for the 6 day integration effort with the robot&dispenser components delivered and installed by Julian Riba Sept. 1st-10th 2015. Some adjustments were required but after 7 workdays all individual components and control mechanisms seemed to be in working order. The full setup was now running from a single PC by the Sophion QTracks platform for experiment execution.
Figure 71: Integrated Sictec g2 SCM/Patch clamp prototype - showing new integrated despenser mounted on Festo XYZ axis setup - mounted over the new “flat” Sophion amplifier box. The 16 channel pressure system is placed under the “robot” and the single controlling PC can be seen next to it.
The stobo-lighting failed within a few hours of Julian Ribas departure and left the platform unusable for close to two weeks. After Sept. 24 the stroboscope light worked again - but since then lab work has been hampered by driver and possible USB comminication conflicts on communication controller PC causing frequent software crashes. Hardware upgrades have helped a little - but the IMTEK XCore process currently (Nov 6th) still crashes every time we try to dispense the second cell on a 16 well PASCA plate. Extended debugging efforts and the inability to do experiments with more than one site/cell at the time has so far limited our ability to achieve proof of concept with the new setup.
Figure 72: One of multiple SCM dispensed cells caught on the 2um patch hole. The top graph shows the usual ~2MOhm resistance of an un-occupied patch hole - while the second plot shows resistances of ~160MOhm indicating successful positioning of a cell.
Preliminary colclusion regarding the new SICTEC g2 setup are:
● Elimination of separate strobo station with extra camera and slow robot roundtrip is a significant improvement in setup complexity and experimental procedure (except illumination point source stablility).
● Mirror-based cell camera observation of cell window make cell observation in cell dispensing position possible and removes need for z-motion during cell delivery is a significant progress. The safer placement of cell camera optics is also a clear advantage.
● Sophion’s takeover of top-view and Robot XYZ motion control has worked well despite some initial problems.
● Direct communication with the XCore process controlling the dispenser and cell camera generally works well but the absence of User interface and event logging - and our lack of access to the source code makes it very hard to identify and fix problems when the Xcore needs to work together with other software.
● In the Sophion SCM05 PASCA setup the TopView camera could directly observe the droplet delivery site without need for horizontal motion. This was very benefital given the cruzial role of optical feedback based positioning and performance evaluation in the SCM patch clamp appliction. The redesigned cell print module delivered in Sept 2015 did not follow the angeled delivery proposed in WP 6.4 left no space for such a topview camera placement: The topview camera was relocated making a 2x13cm X-axis roundtrip motion necessary in droplet delivery evaluation feedback loop. Since the X-axis carries the Y-axis and the full dispenser module the accelerations of this roundtrip introduces significant vibrations in the whole setup and no meaningful camera observation or cell delivery can be performed before they die out. This is so far a major downside of the new dispenser module.
Potential Impact:
The ownership of project results and the management of intellectual property - whether foreground or background - are handled following the recommendations of the EC as described in the “Guide to intellectual properties rules for FP7 projects”, and according to the consortium agreement. The wording used in the following (e.g. “foreground”, “background”, “IPR”, etc.) corresponds to the definitions described therein in detail.
One of the basic principles the participants have previously agreed on and still agree on is that ownership of background is not affected by participation in the project, i.e. all proprietary information, technologies, know-how and IPR remain property of the project partner that had owned it prior to the project. In order to be able to use such background of all participants, the participants additionally agree to grant each other mutually a time limited, non-exclusive, royalty free licence to use all available background (IPR included) for the purpose of the project, during the official duration of the project, on a “need to use“-basis (i.e. project partners that do / do not have to use specific background to carry out their research tasks are / are not granted a licence). This includes that the RTD perfomers gain access to the knowledge/the background of the SME partners or of the background of other RTD performers when required, e.g. to setup the system to fit into the SME’s apparatus or when tuning the dispensing parameters onto a specific application. This will also be done on a “need to use” basis. Thus, the full set of background can be applied and considered for the purpose of the project, while no commercial opportunity is lost for the background owning partner to generate revenues through products, service and licences outside of the project. Major background and IPR of the individual participants relevant to the project are listed below. The list which was provided in the DoW still holds true and is provided in table 1 for reference.
Table 15: Access rights to important background information. No access rights are abbreviated by N.A. Every partner has access to those background and IPR that is required for its specific exploitation route and product vision.
In summary it can be stated, that the current IPR situation and the regulations for accessing background and exploiting foreground are appropriate to enable all SMEs a commercially viable exploitation within their own application. Furthermore, this scheme allows the RTD performers to access the background of the other partners (be it SME or RTD) they need to fulfil their contractual research.
As far as foreground is concerned that is developed as part of the R&D activities carried out within the project, the general principle is applied that foreground resulting from the project is owned by the participant generating it. Such foreground might be related towards a complete product like for example considered in WP2 and WP4 with respect to design, manufacture, software and application, or it can be related to components such as specific optical detections systems developed in WP1, to software like developed in WP1 and WP2 or to processes and applications like developed for example in WP1 and WP5. When such foreground is generated jointly (i.e. where the separate parts of some result cannot be attributed to different participants), it is jointly owned like already drafted in the DoW. In this case the participants share benefits and burdens of the foreground (e.g. in terms of IPR rights and expenses for IPR protection) according to their shares in the invention (e.g. as documented on the patent application forms). The only exception to this principle concerns the RTD partners: The foreground generated by a RTD partner (whether alone or as part of a joint invention with an SME partner) belongs completely to the SME paying for the research work (referred to as “SME Client”), since the RTD partners are fully financed through the R&D subcontracts. This regulation is the default regulation for RTD partners proposed for this funding scheme in the guide for applications. In the rare case where more than one SME is paying for a specific R&D subcontract to a RTD partner (e.g. WP1.2 or WP1.7) joint ownership of the foreground is acquired according to the shares by which the individual SMEs contribute to the financing of the R&D subcontract (i.e. 30% of financing of the R&D contract will earn a SME 30% ownership of the foreground / IPR generated within the R&D contract), see also table 2. The commercial exploitation by sales or sub-licencing of any jointly owned foreground is decided by the owners in a democratic way according to their shares. Also revenues from such licences or sales are split as well according to the shares of ownership. RTD performers are free to use the results for further research (non-commercial exploitation) if those results are not identified as confidential. Besides the RTD performers can publish the results after obtaining the agreement of the SMEPs.
Table 16: Overview on the project results (including knowledge) to be acquired by the participants and their type of exploitation
The main principle governing the above regulations is that no SME should be excluded from the right of a royalty free use of foreground, even if this foreground is later IPR protected by other members of the consortium. In order to extend this principle even further and to enable an efficient exploitation of the foreground generated within the project, the SME project partners furthermore agree that all non-tangible foregrounds generated through any R&D activity should be exploitable by all other participating SMEs for their own business. Therefore, all owners of foreground (incl. IPR) are granting to all other SMEs involved in this project from the start (namely: BFX, BSQ, PDG, INPT, SPH) a non-recoverable, non-exclusive, non-transferable, royalty free, permanent licence for using any non-tangible foreground generated within the project for their own commercial purposes of any kind, excluding the right to grant sub-licences to third parties. After the end of the project this free license can either be prolonged or will be changed to a non-free license which then bases on fair and reasonable conditions.
Tangible foreground (e.g. pre-production prototypes) however, belongs to those SME(s) paying for the development of this foreground. Therefore, other SMEs have to ask the owner(s) for using such foreground, whereby the permission to do so should not be unreasonably be withheld. This regulation enables all SMEs to benefit equally from all parts of the R&D work-plan, creates synergies and alignment in the individual developments and hopefully fosters cooperation and joint exploitation. Nevertheless, the ownership of foreground (incl. IPR) includes the right to grant sub-licences or to sell IPR, still vests with the one SME (or group of SMEs for joint inventions) being the owner of a specific foreground or IPR. Thus, also further commercial exploitation through cooperation with third parties remains possible for the owner(s).
In order to separate background clearly from foreground the project partners claim ownership of background as listed in detail in section 4. This list is going to be updated prior to project start. In summary, each of the participants contributes with a unique proprietary technology to the project and requires certain background from other project partners. This relationship is summarized within table 1 describing the access rights of the individual partners to background. As far as the central background related to the core SCM technology is concerned, the project partner ALU-FR has declared to be willing to enter into good faith negotiations about licensing of background and IPR with all project partners (see letter of intent (LOI) / in Annex B ) after the project and to give a free license during the project. As it can be seen in the LOI, this licensing will either be free or base on fair and reasonable conditions. Thus, the core SCM technology is directly accessible for all SME partners at fair conditions and their freedom to operate is maintained as far as the SCM technology itself is concerned. The same applies to all other background (as listed in B3.2.1) from the other partners: A free license is granted during the project and after the end of the project this free license can either be prolonged or will be changed to a non-free license which then bases on fair and reasonable conditions. An overview on access rights to important background is provided in the following table 1. As can be seen from this table, each SME has ownership of the specific technology required for their envisioned product and can access the background related to droplet dispensing technology and SCM technology through licensing or by direct cooperation e.g. through joint fabrication or hardware / software purchase. Thus, freedom to operate with respect to the background of the project partners is established individually as well as collectively.
Freedom to operate and third party IPR - As far as IPR of third parties are concerned, there is currently no evidence that the freedom to operate and the commercial exploitation of background and foreground considered within this project might be limited. This opinion is based on several reasons:
1. The freedom-to-operate analysis performed within the PASCA project (Del. 10.1) did not identify any IPR that is potentially in conflict with the SCM technology (i.e. inkjet-like drop-on-demand printing of single cells) like applied in this project.
2. For BSQ a freedom-to-operate analysis was performed in 2011 by a 3rd party (Arnold + Siedsma, Amsterdam) in the framework of the Financing Round A due diligence. Also here no IP was identified that could be in conflict with the single cell MALDI TOF technology.
3. The positive outcome of the patent examination of the PCT EP application of ALU-FR (PCT/EP2010/058170). Following patents: US2008/02086751, EP 0421406, US 4318480, GB 1555091A have been considered by the European patent office during examination of PCT/EP2010/058170. The fact that this application was preliminarily approved on 06.09.2012 is a strong indication for the SCM technology being commercially exploitable for the SMEs under licence of ALU-FR.
There are no existing anticipated business agreements which may impose limitations on the subsequent exploitation or information or inventions generated as a result of the project.
Background of the project partners - This list was provided in the DoW and has been checked for completeness during the project runtime. By filing this deliverable the SCM core EP patent was finally granted and also granted in other countries. Furthermore, this patent has been licensed to the ALU-FR spinoff Cytena GmbH which takes over all obligations resulting from previous LOI’s or contracts. As the founders of Cytena were also active in the former PASCA and in the SICTEC project, they are well known to the SICTEC partners and a good and productive cooperation has been established.
BioFluidix (BFX)
• NanoJet dispensing technology: Simple, non-exclusive licence to use following IPR :
o EP 0961655 (and derived national patents)
o EP 1049538 (and derived national patents)
o DE 198 02 367
• Technical design and fabrication of SCM automation system (as developed within the PASCA project, joint ownership with PrimaDiag)
• IP rights for the software developed within the PASCA project to control the SCM prototype instrument (software version Biospot BS12.33.13 Core V1.3.15)
BiosparQ B.V. (BSQ)
Patents and patent applications:
• Method and device for detecting and identifying bio-aerosol particles in the air (WO 02/052246 (WO246)) and patents derived therefrom
• MALDI matrix and MALDI method (WO 2009/064180A1 (WO 180)) and patents derived therefrom
• Method and apparatus for identification of biological material (WO 2010/021548 (WO 548)) and patents derived therefrom
• Method for classification of a sample on the basis of spectral data, method creating a database and method for using this database, and corresponding computer program, data storage medium and system (2009015) and patents derived therefrom
Software
• IP rights for all BSQ data processor software (Matlab scripts), system controller software, GUI (source codes)
• IP rights for all BSQ reference database software and contents (source codes, classifiers)
• Models (CFD, performance, etc.)
Data
• IP rights for all experience gathered during the testing and the development of BSQ technology as reported in TNO project N02/30, DV2009 A 417 and TNO DV2009-418 and all BSQ-TECH-RPT’s up to kick-off of SICTEC project.
• IP rights for all descriptions of the current BSQ instrument design up to kick-off of SICTEC project.
PrimaDiag SAS (PDG)
• Versatile robotic Platform (design partly shared with BFX)
• Pipettes in the ml - µl range
• Controlling software dedicated to routine biology works (Liquid Handling)
• DNA extraction and DNA purification methods (patent pending)
• Robotic configuration able in performing sample preparation for PCR and qPCR analysis (for research use only).
Innoprot (INPT)
• Fluorescent bio-sensor technology (Patent pending)
• Single cell based production method for monoclonal cell lines developed within the PASCA project
Sophion Bioscience A/S (SPH)
Patents and patent applications
• System for Electrophys. Measurements (WO 0229492A3/US6932893B2)
• Substrate and method for measuring the electrophysiological properties of cell membranes (Protruding Rim : EP1495105B1/US2004005696A1)
Software/ hardware
• IP rights to QTracks platform for planning, execution, supervision and result management for multi-device electrophysiology instruments
• IP rights to software for structured, large scale electrophysiology data analysis and presentation
• IP rights to multichannel patch clamp amplifier design and control software
• IP rights to design and production techniques of multi-well patch clamp chips - including Si-die integration in plastic substrates.
• background to techniques for combining automated patch clamp platform operation with single cell dispenser operation developed within the PASCA project
The Institute of Microsystem Technology, University of Freiburg (ALU-FR)
• SCM core technology for drop-on-demand printing of single cells: Granted EP patent EP10724847.8 US Patent US13/707,086 and, WO2011/154042 A1 → Licensed to Cytena GmbH with Cytena taking over all obligations arising from previous letter of intents (LOIs) or other contracts such as the SICTEC consortium agreement.
Erasmus Medical Center (EMC)
• The bacterial strain collections and the associated data that are used in the SICTEC project are considered EMC IP and can only be used for research purposes. EMC is willing to negotiate a license in case BSQ wants to commercialize the reference database in which the EMC bacterial strains are stried.
Dissemination activities
Exhibitions - Most external partner or customer attraction during the project was either done through the website or direct communication by the partners. This lead to more than 20 contacts, e.g. BioSparq talked to all major IVD industries during the project (e.g. bioMérieux, Becton Dickinson, Biocartis, Bruker, Accelerate Diagnostics). However, the technology was showcased at the following exhibition and meetings:
• MicroTas 2013 conference and exhibition (October 2013, Freiburg, Germany): Presentation of the prototype device and live-lab demonstration on single cell printing.
• Sophion user meeting (August 2015, Kopenhagen, Denmark). Presentation of the current status and potential applications.
Furthermore, BioFluidix attended BioJapan from October 15th to 17th in Yokohama, Japan with its distribution partner MicroJet. As MicroJet also has a background in cell dispensing, various talks were given to interested parties based on which further development is currently being discussed.
Publications - The following publication were submitted to journals or presented at conferences with reference to the SICTEC project:
● Gross A, Schoendube J, Zimmermann S, Steeb M, Zengerle R, Koltay P. “Technologies for Single-Cell Isolation”. Int J Mol Sci. 2015 Jul 24;16(8):16897-919. doi: 10.3390/ijms160816897.
● 16-18 September 2014: Single Cell Genomics meeting 2014 – Utrecht, The Netherlands. Riba R, Niemöller JC, Zimmermann S, Schoendube J, Bleul S, Koltay P, Zengerle R, Claus R, Becker H, Gross “A. Single-cell printing for the genetic analysis of cancer cells”. (Poster: P074)
● 25-28 April 2015: the 25th European Congress of Clinical Microbiology and Infectious Diseases, Copenhagen, Denmark. Dekter, Hinke. “Single cell MALDI-TOF based identification of strains obtained from hospitalized patients” (Talk)
● 19-22 July 2015: 6th International Conference on Analysis of Microbial Cells at the Single Cell Level. Retz, Austria. Riba J, Gleichmann T, Zengerle R, Koltay P. “Label-free sorting and deposition of single bacterial cells using the Single-Cell Printer technology” (Poster and Talk)
● Upcoming: 11-14 October 2015: New Approaches and Concepts in Microbiology. EMBL Heidelberg, Germany. Riba J, Gleichmann T, Zengerle R, Koltay P, Zimmermann S. “A new tool for label-free isolation and deposition of bacterial cells” (Poster)
Available patent portfolio - As the main intension of the works performed in the SICTEC project was to increase the maturity and robustness of the technologies involved to enable the SME performers for a fast adoption to the markets, no new patents applications were created during the project. Nevertheless, some patents were granted, increasing the possibility to enter the markets quickly. Thus, the following patents have been granted or licensed to the project partners.
BioFluidix (BFX)
• NanoJet dispensing technology: Simple, non-exclusive licence to use following IPR :
o EP 0961655 (and derived national patents)
o EP 1049538 (and derived national patents)
o DE 198 02 367
BiosparQ B.V. (BSQ)
Patents and patent applications:
• Method and device for detecting and identifying bio-aerosol particles in the air (WO 02/052246 (WO246)) and patents derived therefrom
• MALDI matrix and MALDI method (WO 2009/064180A1 (WO 180)) and patents derived therefrom
• Method and apparatus for identification of biological material (WO 2010/021548 (WO 548)) and patents derived therefrom
• Method for classification of a sample on the basis of spectral data, method creating a database and method for using this database, and corresponding computer program, data storage medium and system (WO 2013154425) and patents derived therefrom
PrimaDiag SAS (PDG)
• Magnetic attraction module, robot including such a module, and method for using such a module or such a robot on magnetic beads (WO 2014083165)
Sophion Bioscience A/S (SPH)
• System for Electrophys. Measurements (WO 0229492A3/US6932893B2)
• Substrate and method for measuring the electrophysiological properties of cell membranes (Protruding Rim : EP1495105B1/US2004005696A1)
The Institute of Microsystem Technology, University of Freiburg (ALU-FR)
• SCM core technology for drop-on-demand printing of single cells: : EP patent EP10724847.8 US Patent US13/707,086 and, WO2011/154042 A1 and patents derived therefrom, now licensed to Cytena GmbH (see D8.3 for reference)
List of Websites:
will be available with all contact data also after the project
SICTEC focussed on the development of technologies for separation and manipulation of single biological cells for life science research and medical applications. The single cell manipulation technology (SCM technology) developed within the a former EC-funded project (PASCA, Platform for Advanced Single Cell Manipulation and Analysis) bases on inkjet-like printing of single biological cells confined in free flying micro droplets. It constitutes a universal platform for single cell analysis with proven potential for many life science applications. The objective of this project is to support the participating SMEs and companies to take up the SCM technology, to realize their own applications and to develop them into innovative products for the medical, biomedical and pharmaceutical markets. Central element of the research is the SCM prototype instrument and the single cell dispenser as presented in numerous publications.
Based on the technology, a new spinoff (Cytena GmbH, Germany) was established.
Project Context and Objectives:
The SICTEC project aimed to provide validation and extension of the use of the developed prototype instrument and to deal with necessary improvements and modifications of pre-production prototypes towards the specific needs and applications of the SME partners. In particular also topics affecting commercial exploitation like e.g. application development, design for manufacturability, reliability issues, cost and throughput optimization, extension of technical specifications, and preparation of CE IVD labeling are being investigated. Applications targeted by the individual SMEs are considering fast pathogen detection for clinical use by combining the SCM technology with MALDI TOF mass spectroscopy of single bacteria, instruments for single cell cancer and stem cell research, methods for monoclonal cell line development and other in-vitro medical diagnostic applications. The anticipated innovations stemming from this research will be exploited by the involved SMEs individually as well as jointly through several innovative products targeted for different markets and applications.
Project Results:
The core technology development (WP1) included that the specifications for the dispenser and detection unit needed to be elaborated, taking into account the different needs from the different applications and users. Based on that, the design and implementation of different print heads needed to be started: On the one hand for a print head which is capable of dispensing bacteria which has a lot smaller diameter (~ 1 µm) than all objects printed so far (~ 10 µm), on the other hand on a print head which apart from the standard optical detection for cells includes a fluorescence detection option. Furthermore, a better chip supply chain was about to be initially developed with more optimizations and yield increases to come in RP2. The final goal here is to have an industry-scale fabrication process for the disposables, including a quality control mechanism. Furthermore, this includes the design and fabrication of dispensing chips which are able to dispense smaller droplets 10 times smaller than the current chips (200 pl → 20 pl) to cope with the limitations which might arise with bacteria dispensing. Finally, the cell detection software is enhanced so that geometrical properties, number and statistical distribution of cells can be monitored to provide cytometry data of the liquid sample and to be able to extract further cell parameters such as roundness. A data-management software is to be developed which manages all this collected data. For the SCM instrument for cell research (WP2), the works performed in the reporting period includes finding and fixing the specifications of the single cell printer (SCP) and the initial and detailed design of such a machine. Based on the design, orders will need to be placed at different suppliers to be able to assemble the device in the next reporting period. In parallel to the design, the documentation (CAD drawings, electrical sketches, CE documentation, manual ...) will be started with more to come with the instrument fabrication and characterization. Last but not least, initial development of the control software will be conducted. For the MALDI MS for pathogen detection (WP3), also specifications need to be established and first spectral responses of lice cycle phases are about to be extracted. For that, bacterial colonies where cell populations are all in the same life cycle phase will need to be grown and their spectra are recorded using a “traditional” MALDI TOF instrument. Based on that, “biological noise” calibration curves to correct for difference in life cycle phase of bacteria in clinical samples will be created. Furthermore, the test panel for the Antibiotic Susceptibility Test needs to be selected and first designs of an multiplexed “just-in-time” mixing will be discussed. Also for the IVD handling platform (WP4), the specifications need to be defined, in parallel the performance of and existing platform is characterized and optimized. Based on these findings and the specs, the design of an IVD certifiable liquid handling device is started while keeping cost reduction and optimization in mind. During the design phase, but also while the device is assembled, the documentation for this machine is preliminary written.Within WP5 (monoclonal cell culture), the clonal cell line production needs to be optimized. For that, an existing single cell handling machine (SCM) from the PASCA project will be used and will be checked for improvements which arise especially from WP1: Higher throughput through the use of better chips and better detection software, improved sorting capabilities, or higher viability of the cells. WP6 (Automated single cell patch clamping) starts by setting up some preliminary specification based on which the stroboscopic imaging module for detection of the droplets in flight will be improved. Together with optimizations on the dispenser module which should consist of the cell dispenser and a top-view camera which can be used to find fiducial marks on the targets to be dispensed on, software optimisations will be initiated which will then allow automatically setup the dispenser and finding the right dispensing parameters without user interaction. Initial work will be performed to enhance the control software to allow it to be remote controlled without a graphical user interface, so that in the end, the user only utilized one single software: The SOPHION machine control software.The validation work package (WP7) aims in acquiring pilot partners during both reporting period and to validate their applications for feasibility with the single cell dispensing technology. For this, various ways to get in touch with such will be elaborated, be it the website, fairs or application workshops which will be held with a set of potential interested persons. WP8 (Exploitation, IP Management and Dissemination) has the task to setup and maintain the project website and to address IP management related questions. Furthermore, and in strong link with WP7 exhibitions and publications are planned, conducted or launched. Apart from that, updating the plan for the use and dissemination of the foreground is a repetitive task of the work package. Project management (WP9) finally deals with all administrative issues within the project, be it the technical coordination, submission of deliverables and reports or to hold internal meetings. This also includes setting up the internal communications structure e.g. by use of a cloud-based internal website. The technical specifications of the Chip Supply Chain are determined. Focus was put on fabrication scale-up, cost reduction and improved handling of cartridges. To gather the specifications all relevant partners were sent a questionnaire which afterwards were compared and a common base was established. This common base was finally sent around to the partners for final agreement. The resulting final technical specifications for dispenser chips and cartridge fabrication are listed in Table 1.
Table 1: Technical Specifications of the Chip Supply Chain
Dispensing chip supply chain development - Scope of the work is on new design of the dispenser unit and cartridge fabrication. On the one hand the changes of the P9 Dispenser and on the other hand the new design of the mechanical cartridge attachment. Cartridge fabrication is ramped up. Detailed failure analysis and shelf life study is performed. In the course of the changed P9 Dispenser package, develop and provided by BioFluidix, and for the new design of the cartridge attachment new and adapted parts have to be constructed, fabricated and assembled. The new assembled P9 dispenser is fully compatible to the old versions and demonstrates reliable and robust performance. The new P9 is now protected against corrosion, which was a known problem in first version when being operated with salt containing liquids. New material compositions are used and the piston was miniaturized. Housing and electrical safety was reworked to meet industrial standards. In Figure 1 the old and the new P9 dispenser designs are illustrated.
Figure 1 (ALU): Former P9 design (left) and new, improved P9 next generation (right).
The new mechanical Cartridge attachment is redesigned towards a clip mechanism, instead of screws (cf. Figure 2). The pneumatic shutter system is reworked towards more space-saving. Alignment structures are geometrically redesigned and the clip mechanism is positioned on top and on bottom. All the specifications are also in WP 1.1 scheduler listed. These changes allow a better handling of the system. P9 Dispenser and piston guide now build a combined unit. The material is changed to steel or brass, which increases the abrasion resistance of the piston guide. The clip mechanism allows for controlled positioning of the cartridge and at the same time features ease-of-handling. Rapid prototyping methods like micro-milling and 3D-printing were used to perform first design and material studies on various clip designs and piston guides. All designs are compatible to future injections molding fabrication processes. The reworked sample reservoir has the same dimensions as a 200 µl pipette tip and gives the opportunity to increase the sample volume from former 80 to now 200 µl. Figure 2 exemplarily shows one evaluated design of the new clip mechanism compared to the old design using screws.
Figure 2 (ALU): (Left) Design drawing of the old Cartridge attachment. (Right) Design drawing of the mechanical cartridge attachment (clip mechanism) at the piston guide and chip holder.
Figure 3 shows photographs of the new clip cartridges made of different materials (left) and the brass piston guide (right) that are all fabricated by rapid prototyping methods. Multiple design iterations had been necessary to optimize geometry and tolerances.
Figure 3 (ALU): New designed cartridges fabricated as micro-milling and 3D-printing parts in different materials (left, green). 3D-printed piston guide from brass in front and rear view (right).
Figure 4 shows a photograph of the first prototype with new clip design cartridge and new piston guide mounted on the second generation P9 NG.
Figure 4 (ALU): New designed dispenser unit existing from P9 dispenser, piston guide and cartridge.
First 3D-printed models of the clip design already demonstrated promising mechanical stability and clipping performance. However further design adaptions have to be done to optimize clipping performance and evaluate influence of production tolerances. During preliminary printing experiments it turned out that the new designed parts did not perform as well as the old design in some respects: Clipping the cartridge to the piston guide needs practice and it was observed that during clipping the piston scratches the silicon chip, which can influence the dispensing performance and thus must be avoided. Additionally, the lower clip breaks from time to time, due to poor stability. Furthermore, the filling of the chip is more difficult and the reservoir agitation system doesn’t perform well with this reservoir shape. Therefore a new design was developed and tested. The second design is also based on a clip mechanism. It now features four geometrically identical clips, which allow clipping from the top and avoid that during clipping the piston scratches the silicon chip. The reservoir is formed like the old one, so that agitation should work. The new cartridge design is also compatible with future fabrication by injection molding. The piston guide was redesigned for the four clip mechanism. Furthermore, the inner region of the piston guide will be produced from a non-metallic material to prevent abrasion between the piston and the guiding structure. Alignment structures, pneumatic shutter system, and attachment to the P9 dispenser are the same as for the first design. Different configurations of the clips are tested to optimize the clipping performance and stability of the clips. Design drawings of the new and final designed parts are shown in figure 5.
Figure 5 (ALU): Design drawing of the second design. (Left) Piston guide assembled from two different materials. (Right) Mechanical cartridge attachment (clip mechanism) at the piston guide and chip holder.
Figure 6 shows photographs of the new clip cartridge made of prime gray (left) and the brass piston guide with the plastic inlet (right) that were all fabricated by rapid prototyping methods.
Figure 6 (ALU): (A) New designed cartridges fabricated as 3D-printing part in prime gray. (B and C) 3D-printed piston guide from brass in front (B) and back (C) view.
Figure 7 shows a photograph of the second prototype with the new clip design cartridge and the new piston guide mounted to the second generation P9 NG.
Figure 7 (ALU): New designed dispenser unit existing from P9 dispenser, piston guide and cartridge.
These 3D-printed models of the second generation clip design demonstrated promising mechanical stability and clipping performance. Also, the first experiments to test the dispensing performance, handling, and agitation system showed good results. Clipping of the cartridge to the piston guide works very well and handling is much easier than it was with the first generation clip design. Filling of the sample and the use of the agitation system show the same good performance as the old design. The preferred future material of the cartridge is the polymer polypropylene (PP). It is superior for injection molding and can be sterilized by autoclaving, which is a convenient and standardized process. The 3D-printing material prime grey has the same flexural modulus as PP. Hence, the mechanical performance can be evaluated with the printed parts. This cartridge design was already reworked to fulfill the requirements for injection molding, such as same wall thickness, 1° draft angle, and the use of maximum two grafters. Figure 8 shows the cartridge design as injection molding part.
Figure 8 (ALU): (A-C) Design drawing of the cartridge as injection molding part in side (B) and back (C) view.
Offers from different companies were requested and documented in the following list. The first three companies (HSG-IMAT, Braunform and MDX Devices) are injection molders, and the tools are designed for large amounts up to 1.000.000 parts. The tools are fabricated from steel and thus have a high manufacturing cost, except MDX Devices. This company works with microinjection molding machine, which allow a cheaper fabrication of the tool. They need for the tool less material and fabrication time. The other three companies (Wehl&Partner, Kaiser, and 1zu1) are rapid prototyping companies. These tools are fabricated from aluminum and are meant to produce between 1.000 to 80.000 parts.
List 1: Injection molding companies
The final design is based on a clip mechanism. It now features four geometrically identical clips, which allow clipping from the top and avoid that during clipping the piston scratches the silicon chip. The reservoir is formed like the old one and supports the agitation system. The final cartridge design is compatible with fabrication by injection molding. The piston guide was designed for the four clip mechanism. Furthermore, the inner region of the piston guide will be produced from a non-metallic material to prevent abrasion between the piston and the guiding structure. Design drawings of the final designed parts are shown in figure 9.
Figure 9 (ALU): Design drawing of the final design. (Left) Piston guide assembled from two different materials. (Right) Mechanical cartridge attachment (clip mechanism) between piston guide and chip holder.
MDX Devices was chosen for fabrication of the chip holder as injecting molding part. The chosen material of the cartridge is polymer polypropylene (PP). It is superior for injection molding and can be sterilized by different methods. The first injection molded parts were produced steel safe, which means that the injection molding tool can be adapted after the first molds if necessary. The 3D-printed Piston Guides didn’t show high enough precision. The tolerance was too high with about ± 300 µm deviation. Therefore the printed parts will be reworked by milling at the important structures. Figure 10 shows photographs of the new clip cartridge made by injection molding (left) and the brass piston guide with the polymer inlet (right) that were fabricated by rapid prototyping methods and reworked by milling.
Figure 10(ALU): (A) New designed cartridges fabricated as injection molding part in PP. (B and C) 3D-printed and milled piston guide from brass in front (B) and back (C) view.
Figure 11 shows a photograph of the new clip design cartridge and the new piston guide mounted to the second generation P9 NG.
Figure 11 (ALU): New designed dispenser unit existing from P9 dispenser, piston guide and cartridge.
These injection molded models of the final clip design demonstrated promising mechanical stability and clipping performance. Also, the first experiments in order to test the dispensing performance (67% of tested cartridges dispensed stable droplet at parameters in a range of 10 µm and 160 µm/ms to 12 µm and 200 µm/ms), handling, and agitation system showed good results (see figure 12). Clipping of the cartridge to the piston guide works very well and handling is much easier than it was with the first generation clip design. Filling of the sample and the use of the agitation system showed the same good performance as the old design. Furthermore, different sterilization methods (Plasma sterilization, E-beam sterilization and Gamma sterilization) were tested. Cartridges show no differences in performance before and after sterilization.
Figure 12 (ALU): Dispensing performance of the injection molded parts at parameters of stroke length 10 µm and downstroke velocity 180 µm/ms. (Left) Nozzle of chip. (Right) Dispensed droplet.
The dispensing experiments show that the first injection molded parts need some additional reworks. The dispensed droplets are very small and need higher parameters than the screwed cartridges. The clip length is too big, a gap of 0.07 mm is given between cartridge and piston guide. Also the chip cavity is displaced by 0.3 mm. Because of that the piston doesn’t hit the chip at the right position (see figure 13). Both problems cause the generation of the small droplets.Figure 13 (ALU): Front view of the piston position relative to the chip reservoir. (Left) Current status, piston displaced by 0.3 mm. (Right) Ideally piston hits the chip reservoir in the middle.
Additional reworks have to be done. At the side of the chip holder, the length of the clips was shortened. Furthermore the width and the depth of the chip cavity, as well as the outlet diameter had to be decreased for more accurate chip gluing. At the side of the piston guide the clipping structure will be displaced by 0.3 mm, so that the piston hits the chip reservoir in the middle (figure 13, right). Drawings of the designed parts with the additional reworks are shown in figure 14.
Figure 14 (ALU): Design drawing of the final design with the additional reworks. (Left) Piston guide assembled from two different materials. (Right) Mechanical cartridge attachment (clip mechanism) between the piston guide and chip holder.
Finally, chip holder and piston guides with the changed dimensions were ordered and have been tested. 22 cartridges were fabricated using the final injection molded chip holder. While 2 failed, 20 performed very well and allowed to generate stable droplets.
The amount of fabricated cartridges increased since January 2014 from 10 to 40 cartridges per week. Through the introduction of some parallelization steps the fabrication could be speed up. Additionally the chips are analyzed under the microscope to check if any incorrectness is visible before gluing. Since this step was included chip performance increased. Since January 2014 all fabrication and quality control steps are stored in a detailed database. Cartridge usage and failure causes are also tracked within the database. With this data comprehensive statistical analysis was done to gather detailed information about reasons for failure and yields for different chip types and assembly methods. The left graph in Figure 15 shows the yield of the fabricated cartridges depending on the nozzle size since January. The average yield of both cartridge types is 81 %. The right graph in Figure 5 shows the failure categories. It illustrates that more than 60 % of the failure cases belong to impurities like particles inside the chip found in final quality control. Process control describes failure cases during parts fabrication like dicing quality of the chip nozzles and during cartridge assembly like tilted chips or glue having entered the chip.
Figure 15 (ALU): Statistical analysis about the yield of fabricated cartridges depending on the nozzle size (left). Statistical analysis about the failure cases (right).
The mentioned failure cases are further investigated and appropriate measures are taken.
Dispensing chip supply chain development - The cartridges are generally packaged in plastic containers, specifically in Corning® microarray slide mailers/storage boxes (Sigma-Aldrich Co. LLC., St Louis, MO, USA), under nitrogen atmosphere. Thereby one container is carrying 2 x 5 cartridges separated by a glass slide. In order to enable individual packaging of the slides in prospect of potential sterilization processes, a sealing film (MELAfol, MELAG Medizintechnik, Berlin, Germany) was tested. MELAfol is impermeable to bacteria, heat sealable, and transparent in accordance with the EN 868-5 standard. In addition, the film has peelable sealing seams and a handling indicator that changes color from blue to brown with steam sterilization. The cartridges were packaged individually in MELAfol film and sealed with an appropriate foil sealing device. In May 2014 a shelf life study for the cartridges was started. Cartridges were stored in batches in containers and individually packed in film under various conditions over durations of 1-3 months, for certain conditions up to 12 months. All tested cartridges were quality controlled before and after storage. Batches of 10-24 cartridges each were tested in
● two different packaging formats
• container carrying 10 cartridges each
• individually packed cartridges in MELAfol film
● three different storage conditions
• room temperature (laboratory condition)
• 5°C (fridge)
• 70°C (oven, accelerated aging)
● at seven time points
• 1 month
• 2 months
• 3 months
• 4 months
• 5 months
• 6 months
• 12 months
Figure 16 summarizes the results of the shelf life study attesting the storability of the cartridges for up to one year without appreciable reduction of their functionality.
Figure 16 (ALU): Shelf-life study of cartridges. Operability of cartridges was determined by standardized quality control (QC) before and after applying indicated packaging and storage conditions at different time points and is expressed as the percentage of cartridges passing the QC (n=10-24). Note: 3 out of 10 cartridges packaged in film did fail the QC after 6 months at room temperature, but passed it successfully when re-investigated at a later time point (1 year).
On a long run, the cartridges should be provided as sterilized consumables to enable a contamination-free single-cell printing. Common sterilization methods such as heat (e.g. steam), chemicals (e.g. ethylene oxide), irradiation (e.g. UV, gamma- and beta-radiation), and plasma are investigated. To address the issue of cartridge sterility a study is being conducted in 2015. Therefore the most common and established standards in sterilization are being evaluated. As first method steam autoclaving was tested. Here cartridges were exposed to standard programs using a laboratory standard autoclave working with 121°C water steam. The standard cartridge material PMMA does not survive this process. PEEK as alternative survives. The microchip is also not compromised. The cartridge as a composition of chip (Si-glass), chip holder (PEEK) and glue (2K epoxy) does not survive in sufficient numbers and quality. As critical issue the connectivity between Si, epoxy and PEEK was found. The different thermal extension coefficients of all cartridge materials seem to introduce mechanical stress resulting in leakage. Current failure rates around 60% making this process unsuitable. Therefore, steam sterilization does not appear as a method of choice for sterilization of the cartridges, at least when using the thermally labile PMMA as material for the chip holder. Currently, further sterilization methods like UV, gamma-, beta-irradiation and plasma are under investigation in consideration of their compatibility with the PMMA cartridges. Preliminary results indicate that irradiation of the cartridges at 20 – 25 kGy with gamma- and beta-rays, respectively, and plasma sterilization with hydrogen peroxide, do not impair the operability of the cartridges and could therefore be suitable sterilization methods for the consumables of the single-cell printer.
Optical detection system for bacteria - In order to reliably print single bacteria cells with the SCP several modifications on the detection optics were calculated and tested. New dispenser chips with increased optical clarity are being developed.
The following optical setups were tested with beads and bacteria cells:
1. The standard SCP optics with the 3.1x TV lens
2. Modified SCP optics with a plug & play 5.0x TV lens
3. Modified optics with a 10x long working distance microscope objective
While the standard SCP optics (1) is not capable of detecting 1 µm sized beads, the slightly modified version (2) works fine with these beads. Although setup (2) has a decreased working distance of d = 40 mm compared to the standard setup (d = 77mm) it proved to be beneficial for our work with smaller objects as it is easily interchangeable with setup (1) (plug & play). Although most bacteria such as E.coli are larger than 1 µm, they could not be reliably detected by setup (2). Therefore we assembled a modified optic that comprises a 20x microscope objective featuring extra-long working distance. Nevertheless, the higher magnification has to be paid off by a decreased field of depth and working distance (for setup 3, d = 17.6 mm). While the flexibility of the standard SCP suffers from such a low working distance between the dispenser chip and the objective, this is acceptable for the MALDI system where the dispenser is fixed and not mounted on a movable axis. A first detection and dispensing system was deployed to the MALDI-TOF device at BiosparQ for initial integration and characterization (see also Deliverable D1.1 Pre-production prototypes). IMTEKs work on characterizing the optical system for detection of bacteria is still ongoing towards an improved 2nd prototype, including the development of dispenser chips with increased optical clarity. Further, we will also work on the necessary software optimizations regarding the detection algorithm for bacteria cells. The initial tests with the new print head look promising: The integrated platform has been tested with latex microspheres and well known E.coli and Serratia strains delivered by TOPLab (University of Leiden). With latex spheres it was discovered that droplets are charged, resulting in the particles being repelled by the electrical field of the ion source in MS. Implementing an earthwire seems to solve this issue. This result was fed back to ALU-FR to update the design of the dispenser or the disposable chip. Experiments with E.coli strains have revealed that coating of the bacterial cells is the most crucial step for the BSQ platform. More specifically, the morphology of the resulting particles (bacterial cell + crystallized matrix material+ interface) determines the information content of the individual spectra. The morphology is influenced by
● Matrix material (HCCA, SA, DMT, ...)
● Additives (surfactants, acids ...)
● Droplet formation / condition of the nozzle surface (contamination, ...)
● Drying regime (matrix co-crystallization)
Exemplary images of “bare” and matrix-coated cells are shown in Figure 17
Figure 17 : REM image of “bare” cells (left) and matrix coated ones (right).
As described above we evaluated several optical setups regarding their applicability for single bacteria cell detection. It became clear that to facilitate robust single bacteria cell printing the following points need to be addressed:
● Resolution of the optical detection system
● Reduced dosage chamber depth (i.e. reduced nozzle size) due to smaller depth of focus and improved optical clarity of the dispenser chips
● Image processing algorithm optimization for bacteria detection
In order to evaluate and test the setup we worked with two setups: One setup was integrated into a printhead that was attached to the Single Cell Printer, in order to evaluate the bacteria cell printing capability. The other optical setup is designed for the MALDI-TOF prototype and was assembled tested “offline”, i.e. without a robotic stage.
Dispenser chips - Usually the depth of focus is inversely related to the magnification (and resolution) of objectives. Therefore, a smaller nozzle size of the dispenser chips are not only beneficial for the MALDI-TOF instrument as they produce smaller droplets, but also because they reduce the depth of focus that is necessary to image the small cells with sufficient sharpness and contrast. As stated in an earlier report we produced 20x20 µm² sized nozzles, i.e. dispenser chips that have a dosage chamber that is 20 µm deep. These work very well and were tested for their capability to print single bacteria cells, which requires an efficient detection. Due to their small size and the low contrast it is crucial that the dispenser chips are optically clear and provide a bright and uniform background. By optimizing the silicon etch protocol during the micro-chip fabrication we could significantly improve the situation as depicted in Figure 18.
Figure 18 (ALU): Dispenser chips with 40 µm x 40 µm dispenser chip (left) and new 20 µm x 20 µm orifice opening. 20µm chips have reduced droplet volume (right). The new chips show significantly reduced surface roughness on the silicon.
Optical detection system - The general design of the optical detection system for printing single bacteria cells is very similar to the standard Single-Cell Printer (SCP) detection setup with two important modifications: First, for printing bacteria cells we use a special high magnification objective with extra-long working distance. And second, we placed two apertures in front of the illumination LED which allow collimating and focusing the blue light illuminating the microchip. The apertures in the illuminating light path help to significantly increase the contrast and sharpness of bacteria cells in the microchip as depicted in Figure 19.
Figure 19 (ALU): The coaxial illumination system was equipped with two apertures allowing focusing and collimating the incoming light. The sharpness and contrast of bacteria cells can be significantly increased by closing the 2nd aperture (20 µm dispenser chip, E.Coli cells).
Software optimization - The SCP software was modified in the following aspects:
● The different cell detection camera types are now supported.
● The object detection algorithm was modified to enable for detection of low contrast objects 1-3 µm in size by increasing the detection sensitivity and optimizing the object classification.
Evaluation of the prototype - In order to evaluate the performance of the setup we printed fluorescently labeled Escherichia Coli cells (a GFP expressing strain). We printed three arrays with 100 single cells each on a glass slide. The cells were suspended in DI-water and Glycerol (10 % v/v) which allowed us to image and count the cells after printing. The resulting spots were classified according to the number of cells: No cell, single cell, or multiple cells (see table 2). Typical fluorescent images can be seen in Figure 20. The total single cell efficiency yielded 66.7 %. We think that the single cell efficiency can be further improved by reducing the vibrations of the printhead induced by the robotic stage. This could be achieved by using a fixed printhead and moving the substrate only. Since the MALDI-TOF MS setup is not mounted to a moving stage we believe that here the single cell efficiency will be significantly higher once the setup is assembled to the MALDI-TOF MS instrument.
Figure 20 (ALU): Typical fluorescent/bright field overlay images of printed droplets allow to verify whether the droplet contains a single cell (left), no cells (center), or multiple cells (right).
Table 2: To analyze the performance of the instrument, we printed three arrays (10x10) of a GFP-expressing E. Coli sample on a glass slide and counted the number of E.Coli cells in each droplet. The total single cell efficiency yielded 66.7 %.
Prototype instruments - The optical detection system for bacteria was integrated into a Single-Cell Printer prototype instrument at IMTEK for optimization and evaluation of the printing performance. A prototype Single-Bacteria printhead for the BSQ´s MALDI-TOF instrument was designed, assembled and tested.
Single-Cell Printing efficiency and chip-to-chip variation - Single-cell printing efficiency was evaluated by printing GFP expressing E.coli BL21 cells on glass slides. Each spot of a 10x10 array was classified via fluorescent microscopy either as void droplet, droplet containing a single cell, or droplet containing multiple cells. The assay was performed three times for each of three different cartridges. In total, 699 out of 900 spots contained a single-cell, resulting in an averaged single-cell printing efficiency of 79 %. The results are summarized in Fig. 21 (ALU) Single-cell printing efficiency was evaluated by printing GFP expressing E. coli cells on glass slides. Each spot of a 10x10 array was classified via fluorescent microscopy either as void droplet, droplet containing a single cell, or droplet containing multiple cells. The assay was performed three times for each of three different cartridges. In total, 699 out of 900 spots contained a single-cell, resulting in an averaged single-cell printing efficiency of 78 %.
Droplet size reduction - For many applications including MALDI MS single cell pathogen detection (WP 3) smaller droplet volumes are required. Our current dispensing chips allow to reliably produce droplets with volumes as low as 150 pl. Here (WP 1.5) we aim to produce dispenser chips that can consistently produce droplets with a volume of 20 pl. Therefore we manufactured dispenser chips with smaller nozzles which are characterized by gravimetric and image based methods. So far we are able to produce water droplets with 35 pl with the new dispenser chips in a reliable manner. As depicted in Figure 22, our current dispenser chips with a 40 µm x 40 µm nozzle allow to reliably produce droplets with volumes as low as 150 pl. In order to be able to produce smaller droplet volumes we manufactured a set of new dispenser chips with various nozzle sizes of 20 µm x 40 µm, 20 µm x 30 µm, 20 µm x 20 µm, and 20 µm x 10 µm as outlined in Figure 23. In WP 3 (MALDI MS for pathogen detection) we are aiming for very small droplet volumes of 25 pl. As the chips with the rectangular 20 µm x 10 µm nozzle did not reproducibly generate droplets, we therefore focused on characterizing the chips with the quadratic 20 µm x 20 µm nozzle.
Figure 22 (ALU): REM pictures of our current dispenser chip with a 40 µm x 40 µm nozzle (left) and a 20 µm x 20 µm nozzle (right).
Figure 23 (ALU): Current dispenser chip with a 40 µm x 40 µm nozzle is able to produce droplet volumes down to 150 pL which optimal for dispensing mammalian cells. However, some applications such as real time MALDI MS for pathogen detection require much smaller droplet volumes.
In order to measure the droplet volumes of the new dispenser chips (20 µm x 20 µm nozzle, silane coated) we used a highly precise gravimetric method as presented by Dong et al [1]. The balance of our setup (XP2U/M, Mettler Toledo) has a resolution of 0.1 µg. However, we are dealing with droplets volumes down to 20 pl that therefore only weight 0.02 µg each. In order to precisely determine the (average) droplet volume we therefore dispense n = 600 droplets for each data point. Hence, with the gravimetric method we cannot address the drop-to-drop variation which is a separate issue that will be discussed at the end of the report. First we determined the parameter space that allows for generation of stable droplets by varying both the piezo downstroke velocity and the stroke length. The droplet stability and the presence of satellite drops was observed by dispensing on a glass slide on top of a camera (see Action Report WP 6.4). We found that for downstroke velocities of 70 µm/s we obtain stable droplets that can be tuned in size by adjusting the piezo stroke length. Figure 24 illustrates such a characteristic dosage curve for a fixed downstroke velocity. Stable droplets with 35-60 pL volumes were obtained at stroke lengths of 2-4 µm (green data points). This regime could be reproduced by all our dispenser chips (n=7) tested so far. At stroke lengths < 2 µm some of the dispenser chips were able to produce 25 pL droplets, others did not shoot at all. At higher stroke lengths > 4 µm we occasionally observed satellite droplets.
Figure 24 (ALU): The droplet volume as function of the piezo stroke length at a constant downstroke velocity of 70 µm/s. The green data points represent a stable droplet generation regime which could be reproduced by all the dispensing chips tested so far. At higher stroke lengths (red) satellites occurred occasionally. Droplet volumes as low as 25 pL could be achieved but stability issues were observed (yellow).
We also investigated whether the dispensing frequency influences the droplet volume. Figure 25 illustrates that we can dispense with frequencies up to 200 Hz without affecting the droplet size. The stroke length was fixed to 2 µm and the downstroke velocity to 70 µm/s. Note: we have not tested dispensing at even higher frequencies, which might as well be possible.
Figure 25 (ALU): The droplet volume is constant for dispensing frequencies between 50 and 200 Hz. Here the stroke length and the downstroke velocity were set to 2 µm and 70 µm/s, respectively.
We have manufactured new dispenser chips with various nozzle sized. The smallest nozzle with a quadratic orifice (20 µm x 20 µm) performed very well, and we can consistently produce droplet volumes as low as 35 pl as measured on our gravimetric setup. At lower volumes we observed a strong chip-to-chip variation in terms of dispensing parameters. Therefore we will continue to investigate the small volume dispenser chips with special focus on the following points:
● Detailed study of the chip-to-chip variation by correlating the droplet velocity with the actual dimensions and the quality of the nozzle (which can vary due to the dicing process in the chip fabrication).
● We will investigate the influence of the medium viscosity on the droplet volumes.
● In order to determine the drop-to-drop variation, we are currently working on a new setup based on stroboscopic imaging using a flashed LED and high magnification optics. (Note, that it is not an easy task to monitor the 20 pl droplets in flight with a decent resolution that allows to calculate the droplet volumes).
We think that with these chips even lower volumes down to 25 pl might be possible once we have gained more insight into the chip-to-chip droplet volume variation. Further, we are a considering manufacturing chips with even smaller nozzle sizes of 10 µm x 10 µm. As stated above we manufactured and tested new dispenser chips with smaller nozzle sizes. The smallest nozzle with a quadratic orifice (20 µm x 20 µm) performed very well and we started to integrate this chip into our regular cartridge fabrication process. Right now, we are producing about 60 cartridges with 20x20 µm² nozzle each month (out of 160 in total). After one of the earlier bacteria printing experiments a failure analysis was carried out. Figure 26 below summarizes the reasons for all “no-cell” or “multiple-cell” events out of 200 printed bacteria cells. “ROI too small” and “no cell” accounted for almost 19 % of the failures. Meanwhile these can be avoided to a large extend by choosing the optimal ROI size, a stiffer printhead design, and a more sensitive cell detection algorithm. After these improvements we frequently achieve single-cell printing efficiencies of close to 80 % as discussed above. Given the small size of the cells this is a reasonable performance, although slightly lower than what is achieved with the eukaryotic cell printer. In order to reach printing efficiency of > 90 % with bacteria cells, the two remaining reasons for failure have to be addressed:
● In almost 4% of the events bacteria cells cannot be detected due to the edge shadows of the dosage chamber.
● In 6% of the events bacteria cells cannot be imaged with sufficient contrast as they are not close enough to the focal plane of the objective.
Figure 26 (ALU): A failure analysis was performed on 200 printing events by evaluating the SCP images and the printing results. The figure summarizes the reasons for “no-cell” or “multiple-cell” events.
Decreasing the depth of the microfluidic chamber addresses the two remaining issues. Therefore we fabricated dispenser chips with only 5 µm and 10 µm etch depth. As these chips are based on the previous design, the major challenge was the dicing step that determines the final width of the nozzle. Using our current process, this step could not be performed with the necessary precision, resulting in chips with non-uniform nozzle sizes ranging from 2 µm to 20 µm. Nevertheless, this allowed us to pick a dispenser chip with a nozzle size of 10x10 µm for testing. We found that this chip can generate droplets with a volume of approximately 4 pL.
Figure 27 (ALU): Chips with a nozzle size of 10x10 µm² were produced and evaluated via stroboscopic imaging.
Fluorescence detection - The optical detection system of the SCP prototype is further developed towards integration of single channel fluorescence detection. Focus is on selection of hardware components and construction of a specific print head. The prototype system being available for proof-of-concept was reworked in terms of reconstruction of the optical train including improved filter and beam-splitter elements as well as compactness and robustness of laser beam adjustment. Main part of the work thereby was construction using SolidWorks and investigations on optical elements. An improved 3axis adjustment system for the dispenser and an advanced illumination concept for bright-field imaging are work in progress. The setup would allow for precise free-beam coupling of the laser. Such accurate positioning of the laser beam spot in the focus area on the chip will be possible. Figure 28 shows the construction in Solidworks and the final prototype system under assembly. The system is currently still under construction and has not been tested or characterized yet. More details will be provided in Deliverable D1.1 “Pre-production instruments”.
Figure 28 (ALU): SolidWorks construction of compact fluorescence print head featuring laser and high resolution high transmission optical elements and a 3-axis adjustment system for correct focusing of the dispenser chip. Left: photograph of the print head under assembly.
The development of the fluorescence printhead has been finalized. It is mounted to a SCP platform which was electronically upgraded and reworked in terms of electrical safety. An emergency shutdown for the built-in laser (class 3B) setup was included to guarantee user safety. Further a reworked substrate holder was developed to meet the special requirements of the printhead. A protective cover shields the entire laser optical system.
Figure 29 shows a photograph of the final printhead version under operation.
29 (ALU): Photo of the final printhead including protective cover and safety instruction stickers according to EC norms.
The functional set of different batches (array printing, well plate printing, etc.) has been programmed and tested successfully using polystyrene microbeads and fluorescent cells ( Figure 30).
Figure 30 (ALU): Screenshot of the system under operation. Fluorescent microspheres illustrate the dimension of the laser illuminated area and the intensity of fluorophore excitation.
Besides protective covers and laser safety the usability of the system has been further improved. The magnifying objective with its high numerical aperture requires a reduced working distance to capture sufficient amounts of fluorescent light from the cells inside the chip. Reduced working distance has two major drawbacks that need to be compensated: accessibility of dispenser chip and accessibility of well plate format substrates. Well plate access is – due to a specially developed substrate carrier – still possible by filling one half of the plate, then rotating the plate and filling the second half. Although this workaround requires one additional operator step, it does not compromise the laboratory workflow of the SME partners. Access to the dispenser chip is required to change chips / cartridges between different experiments. Working distance between chip and objective is very limited in this fluorescence setup (< 25 mm). Both a proper screwing with the current cartridge models as well as clipping with future models is not easily done in this setup. Therefore a special, user friendly system featuring a swivel arm was developed. It allows for sideways movement of the entire dispensing system giving much space and free access to the dispenser chip. Figure 31 shows the workflow.
Figure 31 (ALU): New innovative chip mounting system applied in the fluorescence printhead. A two stage swivel arm is used to swing the dispensing unit sideways away from the objective. Such, the user has free access to the montage screws and can easily change the chip / cartridge.
The single-channel fluorescence detection system was implemented into a Single-Cell Printer prototype. The prototype was delivered to Innoprot for beta-testing (production of fluorescent cell lines). Transport, on-site installation and training by an IMTEK staff member was provided in April 2015. Prior delivery of the prototype, evaluation experiments were performed at IMTEK using a GFP transfected cancer cell line. Figure 32 shows a U2OS-GFP cell inside the ROI of the dispenser chip.
Figure 32 (ALU): U2OS cell inside the ROI of the dispenser chip.
To evaluate the performance we spiked U2OS-GFP cells into a non-fluorescent cell sample (HEK293) in a 1:10 and 1:100 ratio. We then used the SCP to sort and isolate the U2OS cells based on their fluorescence. Averaged over four 96-well plates the single-cell efficiency (single U2OS cells printed into well) yielded 93 %. The results are summarized in figure 33.
Figure 33 (ALU): GFP expressing U2OS Cells were spiked into a non-fluorescent cell sample (HEK293). The fluorescent SCP prototype was used to sort and isolate the U2OS cells based on fluorescence. The cells were printed into 96 well plates for clonal culturing. Post priting imaging was performed with a fluorescent microscope.
Throughput optimization – This section covers software and hardware actions to enhance the overall throughput of the system regarding detected and printed cells. Hardware has been improved on:
● Enhancement of camera frames by faster sensor and active binning.
Starting value: 60 fps
Actual limit: 150 fps
Target: 200 fps
● Redesign of vacuum shutter system towards faster shutter times.
Starting value: 500 ms
Actual limit: 50 ms
Software has been improved on:
● Reworked detection algorithm with faster object recognition.
Starting value: 6 ms/image
Actual limit: 4 ms/image
Target: 5 ms/image
The actual software/hardware configuration has been tested successfully on a SCP pre-production instrument. Further enhancements towards the specified limits are under development.
Data-management software - Several scripting functions have been programmed and embedded into the CORE image processing and cell detection algorithm. Both live data displaying as well as data readout via comma separated value based text files was realized. Comma separated value format thereby is the leading standard interchange format for feeding data collections into database constructs. The data sets can actually be loaded in table based programs like MS Excel and filtered by specific data. Further visual data like dot plots (well-known from flow cytometry analysis) can be created and displayed. Data storage of image sets is further optimized by reworking the storage serial system and assigning comma separated values to the specific image sets on the hard drive they belong to. Both data collection text file as well as image sets are stored together in one folder. An automated user and experimental settings file currently is under development. The file contains all necessary information about experimental conditions regarding hardware settings and stores all user input registered prior to start of the experiment. This settings file also will be automatically created and stored in each experimental folder. Further it can be loaded in case of performing the experiment again under identical conditions at a later time.
Image storage optimization - To consistently store images on the hard drive, the software GUI allows for setting individual folder names (e.g. experimental name or name of the user + experiment, etc.). To avoid overwriting a respective routine has been implemented to protect existing folders from being overwritten. Such, no experimental information is lost. Image naming has been optimized. Each image is named after a serial key which is assembled according to the following steps:
● Well coordinate / array coordinate
● Number of cell being dispensed at this coordinate (typically 1 -> single-cell)
● Number of track image[1]
● Date-time stamp
The date-time stamp guarantees that image names are always unique and cannot be confused. Each image sequence can be assigned to the specific experiment and the specific location on the used substrate (typically well plates).
User input log file - To preserve the user input and assign it to the respective experiment, a user log file was integrated. It reads out the entire set of user input required to set up an experiment and stores it in a text file. The file is stored in the respective folder, where the images are stored.
The data from the log file can be read out via any text editor and can even be converted to other formats (like MS Excel) since data is stored as comma-separated value format.
Data log file - Each event (cell, particle, etc.) during an experimental run is preserved by a second log file. The file stores mean diameter, area, perimeter, aspect ratio, center of mass, and classification of the object. Objects being identified as single cells are specially marked. Once one of these single cells has actually being dispensed to the substrate the respective coordinate (well or array) is added to the event. The file is stored in the respective folder, where the images are stored. Such the image sequence can directly be correlated to log file data. The user can post-print extract the required information on specific events. Further the log file format is also comma-separated values and can such be easily loaded to excel. Here the entire set of events from a single experiment can be visually displayed in form of dot plots (point clouds) which is the common method to analyze cellular data. Within the point clouds different classes of objects (e.g. single cells, cell clusters, cell fragments, dirt particles, etc.) can be distinguished. When loading multiple experiments direct comparison of point clouds is possible. The data management processing stores are relevant data in the same folder. Unique file naming convention and overwrite protection routines prevent mixing up and loss of data. Universally accessible file formats allow for easy post-processing of data.
[1] Each single-cell event consists of a series of five subsequent images to track the single-cell before being dispensed and to verify if the cell has actually being correctly dispensed
Furthermore, an integrated machine for cell research is developed which on the one hand utilizes the core technology as developed in work-package 1, on the other hand is a good base for a device which can then readily be sold to customers after the project. However, this work-package also suffered from the startup delays at ALU-FR and is currently delayed by about one month. It is expected that by end of the year, the work-package as a whole will be back in time. For supporting ALU-FR in designing a device which can later on be CE-IVD certified, an external consultant was hired: Manfred Augstein from velixx GmbH (http://www.velixx.com) has been designing and market-launching CE-IVD machines for more than 20 years at Roche and thus has a strong knowledge in what to do and especially what better not to do. His mandate includes reviewing the process from time to time and providing hints to improve the development process. The technical specifications of the pre-production prototypes to be developed can be found in table 3.
Table 3: Specifications of the basis platform
Design for manufacturability and system integration - The design and the mechanics of the existing Prototype have been reviewed and partially optimized. The axis of the Prototypes is improved in order to enable the different printheads. An EMV test was performed at the existing SCP prototype. As a result, grounding of the PC casing and the stroboscopic setup needs to be improved. The anodization of the housing has been removed at specific points to ensure electrical conductivity throughout the complete casing. Additional ground wiring was mounted to the stroboscopic setup. Another issue addressed was heat exchange inside the electronic box. This problem has been solved with stronger fans and improved cable management to enhance free space for air guidance. Serious problems with USB 3.0 driver settings and chip set drivers leading to continuous crashes of the camera frame grabber were solved. Further the routing of the USB 3.0 and the camera trigger cables was shifted to outside the cable duct to prevent cross-talk that had been identified as a possible reason for hardware instabilities. The named changes had positive effects on system overall stability. On the mechanical side the accuracy of the axes system had to be improved. In the new prototype the implementation of HIWIN ball screw spindles was favored. The spindles are high end class and can be easily exchanged in terms of required precision without the necessity to change gears, ball screws or bearings. This system is especially attractive towards the different requirements in applications from the SME partners. Pipetting inside a 96-well plate format may require less precision than non-contact printing of dense micro-arrays. With the exchangeable spindles a significant cost-reduction is possible for lower precision applications. The linear guides of the existing SCP prototype instrument have good accuracy but their long term robustness is questionable. It has already been observed that accuracy decreases over number of use cycles. In the next prototype thus linear ball guides will be implemented to guarantee a constant accuracy even at high numbers of working cycles. In addition to the accuracy issues the axes coupling had been reviewed. In the existing prototype the motor is in line with the axis and the coupling. This means, that the axis loses traveling distance and has no loose bearing (See Figure 34). Such the axes is only fixed at one end which makes it susceptible to vibrations and mechanical stress (e.g. longitudinal torsion).
Figure 34: Scheme of the actual axis approach. Motor and axes are in line.
In the approach for the new design we implemented a belt which decouples the axis from the motor and added a loose bearing. This new approach has also a larger traveling distance, is more defined and less susceptible to vibrations (cf. Figure 35).
Figure 35: Scheme of the new axis approach. Motor and axis are next to each other and connected by a belt.
The implementation of the novel components is work in progress. A briefer view on what has been achieved so far is provided in Deliverable D2.1 Finalized prototype design. As above, the existing printheads have been adapted in order to enable fluorescence detection and the detection of bacteria. These two prototypes are heavier than the existing prototypes. The new pre-production prototype has more accurate linear rails, which can also take higher loads and accept higher torsion forces. Several adapters between the prototype and the printheads have been manufactured. This allows for montage of fluorescence and bacteria detection printheads. The additional functionality of both printheads made modifications inside the instrument necessary. New illumination and – in case of fluorescence – the implementation of a laser source required additional modules as well as driver and control electronics to be embedded into the corpus of the instrument. Therefore additional space has been generated by re-designed and re-arrangement of internal component as well as introducing different decks for certain component classes. The corpus height was slightly increased. Further additional safety was embedded, specifically regarding laser safety and emergency shutdown.
Documentation (CAD, CE-documentation, user manual) - The CE documentation of the device was started for this work package. Test criteria for the specifications were defined. Initialy some general CE relevant points were added to the specification list (see Table 4). References for these points are the CE directive 2006/95/EC on low voltage devices and the CE directive 2004/108/EC on Electromagnetic compatibility. In addition to these two, there is the CE IVD directive 98/79/EG. For the technical specifications were some test criteria defined. The accuracy of the base platform for example is tested with two different Tests. The first test is based on the VDE Guideline 3712 and is used to check the general dispensing performance. A second test was implemented especially for our base system and can measure the accuracy of the device positioning with a higher resolution. The position of the device is here checked optically several times with an USAF Resolution Chart. These Tests are explained in detail in WP 4.2. In Order to convert the specifications into a written format, an external engineering consultant, called Manfred Augstein (from velixX GmbH ), was hired. He gives us advises on CE and business related questions and helps to set up the CE documentation. At the moment we got a Project Specification Document, a Project Documentation Plan and a Project Guide. The construction of the first Prototype is nearly finished. After the finalization of the construction, the design is frozen. Following changes of the design need to be documented with reason, explanation, date and signature. This approach is part of the CE conform documentation.
Table 4 (ALU): CE relevant terms of the Specification
In parallel to the assembly of the pre-production prototype, an assembly manual has been written. This manual includes mechanical and electronic assembly as well as cable interconnections. For electronic and power supply connectivity a routing plan has been written. This plan includes all wiring, all connectors (internal and external) and all circuit board I/O’s. Further a user manual has been written. This manual includes an overview of all parts of the SCP (nomenclature) and a detailed guide for hardware and software. The manual was revised and improved internally by the team of ALU-FR as well as externally through the application partners. The pre-prdocution instrument platforms meanwhile have approached towards CE marking. CE conformal document structure has been established and prior documentation is already partly transferred into the conformal formats. The BioFluidix platform instrument has passed the industrial CE conformal EMC test for laboratory devices.
Pre-production instrument fabrication - Assembling of the prototype started with the mechanical parts. The base was assembled first alongside with the necessary wires for the printhead (see Figure 2.5.1). Next was the bridge with the X-movement (see Figure 2.5.2) followed by the Z-axis and the printhead. The next step was to grease and calibrate the axis. After the mechanical, the electrical assembly was done. This included power supply, PC, motor controllers, dispensing electronics and additional electronics for cell detection. These parts have been routed together and with the motors, end switches and finally the printhead (see Figure 36). Software has been installed at the end. Software in this context features the operating system, the camera and electronics drivers and control software, the motor DLL’s and the actual single cell printing software (GUI). The latter is the interface (GUI) which actually interacts with the user and controls the cell printing process as well as all hardware parts. The following photographs were taken during assembly and functional testing in Sept./Oct. 2014.
Figure 36 (ALU): Base of the prototype with routed cables.
Figure 37 (ALU): Assembled base and Bridge.
Figure 38 (ALU): The electronics of the prototype during the routing of the electronics.
Figure 39 (ALU): The assembled prototype.
Pre-production instrument characterization - In this period the pre-production Prototype has been assembled and characterized. Both the mechanical specifications of the base platform and the performance of the printhead were evaluated.
Base platform - The pre-production prototype has a working area of 280x380x75 mm³ (X,Y,Z). The working distance in X of the base platform is regularly 338 mm large, but has to be limited according to the size of the cell printing printhead. The range of the Z-Axis is only 75 mm instead of the specified 100 mm , but can be adapted by taking a different axis. This can be easily done given the modularity by design. After the assembly of the pre-production Prototype, the axis performance has been tested and optimized. The velocity of the x- and y- Axis was initially set to 75 mm/s. After the first tests of the axes movement the velocities were increased to 150 mm/s. For all of the following axis tests the latter velocity has been used. Characterization of the accuracy was performed with an USAF resolution chart. The test has already been described in the action report WP 4.2. These measurements showed that the X and Y-Axis have an absolute accuracy of about ±15 µm. Together with the absolute positioning accuracy, the backlash has been measured. This test showed, that the X-Axis has no backlash, but the Y-Axis has about 100 µm backlash. The relative positioning accuracy still needs to be measured
Single cell printhead – A single cell printhead was mounted on the base platform and tested with beads. To fully exploit the whole range of single cell printing application we work with two optical detection setups. Setup A is our standard setup that is used for most single cell printing applications. It can be used to detect and sort particles (or cells) ranging from 4 to 40 µm in size. Setup B is based on a high resolution microscopy optic and is used for smaller particles such as bacteria cells. The detailed technical specifications are given in the table below. Droplet size was measured gravimetrically, as described in previous reports. The particle size refers to printing of polystyrene beads. The dispensing frequency is about 30 Hz but can be increased to 60 Hz by using the binning function of the cell detection camera.
Table 5. (ALU): technical specifications of the SCP
MALDI MS for pathogen detection - A MALDI mass spectrometry device for pathogen detection incorporating the single cell printing technology from WP1 is developed and evaluated. There have been some delays in starting all clinical related work-packages as there has been a change in the partner structure at EMC medical: Dr. Willem B. van Leeuwen has left EMC and has been acquired as a lecturer at Leiden university with Dr. René te Witt taking over the lead of his former group. It is planned to integrate Willem van Leeuwen’s group in Leiden as a third RTD performer after the end of RP1 to increase the development speed by using his knowledge and by the fact that BiosparQ and his group are located in the same building in Leiden which will shorten learning and adoption curves. Table 6 to 8 show the specifications for the MALDI application.
Table 6: Specifications for the dispenser chip
Table 7: Specifications for the optical detection system
Table 8: Specifications for material compatibility
SCM-technology integration & verification
BiosparQ aspires to analyze samples comprising mixtures of micro-organisms in a background of biological particles. To enable this capability, BiosparQ generates a MALDI mass spectrum of individual cells. For this purpose, microbial cells are to be suspended in an appropriate fluid phase containing all additives required for MALDI (including the MALDI matrix). This suspension is subsequently dispensed into droplets sufficiently small to prevent presence of more than a single cell. Upon drying the MALDI matrix crystalizes, resulting in an individually coated cell suited for MALDI analysis. So, the key challenge for BSQ in the SICTEC project is to interface the SCM technology with the Single Cell MALDI TOF platform. Prior to the SICTEC project BSQ has demonstrated that, using pico-liter droplets that contain one bacterial cell and the MALDI matrix material dissolved (a so-called pre-mix approach), it is possible to generate spectral information from which identification of the bacteria is possible. These droplets were dispensed by means of a Microdrop dispenser, which had to be cleaned after each test. Clearly, this cannot be used in high throughput clinical applications, an approach using disposable dispenser heads should be used instead. For that reason the IMTEK SCM technology was selected to interface with the BSQ platform. The scope of work of this work package was to integrate the SCM printhead on the BSQ ‘drying tower’ sub-assembly, learn how to work with this sub-assembly, integrate the sub-assembly on the BSQ instrument and verify its functional performance. Completion of this work package was the starting point for testing well known bacterial strains and actual clinical samples as described in the subsequent work packages. The SCM hardware and software delivered by IMTEK was integrated on the ‘drying tower’ of the BSQ platform and initially tested off-line for BSQ to become acquainted with the SCM technology. Different experiments were performed to explore the different settings of the printhead. Parker ink was used to demonstrate performance and to check the printing and drying behavior of the sub-assembly.
Figure 40: Off-line test set-up with SC dispenser integrated on top of the BSQ ‘drying tower’. The SC dispenser is not visible and is located in the box on top of the drying tower.
After that, the SCM print head - drying tower sub-assembly was integrated onto the BSQ platform for system verification. As planned, initial test using 1 µm Polystyrene Latex (PSL) spheres were done. The purpose of this test was twofold: firstly it should be verified that the PSL spheres could be detected by the detection lasers prior to firing the ionization laser and secondly it should be verified that a polystyrene spectrum could be built after ionization of the resulting particle.
Figure 41: SC dispenser with drying tower integrated on BSQ MALDI MS prototype in UAS-L Toplab. The left side computer screen is the output of the data acquisition / analysis system for real time monitoring of the experiments.
The first part of the verification was initially troublesome as no detections were observed, indicating that somewhere droplets/particles got lost on their way from the SC print head towards the ionization chamber. Only after shutting down the High Voltage on the ionization source it was discovered that static electricity was the cause of this problem. Charged particles were deflected by the electrical field generated by the ion-source, see below. Apparently this charging of droplets was something that is typical for this type of SC dispensing technology, as it was never observed before by BSQ using the Microdrop dispenser technology.
Figure 42: Simplified model of the trajectory of charged particles entering the ions source HV electric field when leaving the stage 3 skimmer. The trajectories explain the phenomena experienced regarding the lack of detection / ionization.
To remove charge generated during dispensing, the analyte was grounded. This appeared to be very effective, albeit that many times during subsequent testing droplets seemed to be (somewhat) charged again and became deflected. In addition, a solution was implemented where the droplets were uncharged by induction upon leaving the dispenser chip. For this purpose a ‘charger’ was constructed (aluminum foil) close to the nozzle of the chip in order to ‘push back’ charge on the droplets just before separating from the stream.
Figure 43: Picture showing the ‘spring like’ solution to ground the analyte. The spring fits in the reservoir of the chip-cartridge assembly (not mounted here) and transports charge towards ground. The aluminum foil is the ‘charger’ on which a voltage can be applied to ‘push back’ charge by induction on droplets being formed at the nozzle.
Although this way of de-charging seemed to work, it application during actual testing appeared to be too complex and time consuming. Typical test results, where the charged particles are deflected and are being slowed down in the electrical field generated by the ion source are shown below.
Figure 44: Typical example of histogram which illustrates an experiment where a voltage is applied on the analyte (not on the aluminum foil). Note that the Time of Flight (horizontal axis) between the detection lasers can be controlled by the voltage applied. Also the detection rate and the number of mis-firings of the UV laser changes with voltage.
As conductivity of the analyte, potential obstructions in the micro fluidic chips and other uncontrollable phenomena could be the source of this, it was discussed with BFX to change the design of the dispenser chip to prevent charging of the analyte, see WP 1. After it was shown that PSL spheres could be detected after dispensing, albeit still with rather low efficiency (which indicates that part of the PSL spheres is lost on their way to ionization), It was demonstrated that from these detected PSL spheres spectra could be obtained. This proved that these spheres were coated indeed and could be analyzed individually by the BSQ system. The spectral data contained the typical polystyrene spectrum with equidistant peaks at the correct M/Z locations.
Figure 45: Histogram of particles, showing biomodal partcile distribution. The left peak represents particles resulting from empty droplets (only non-volatile MALDI matrix material, diameter 1.2 µm), the right peak represents particles from droplets that contain a PSL 2 µm sphere.
Figure 46: Mean spectra for particles that result from empty droplets (‘empty partícles’, green curve) and for particles that contain a 2 µm PSL sphere (blue curve). Note the typical polymer spectrum in the blue curve, in this case corresponding to polystyrene.
Next, PSL tests were done, where small amounts of insulin was added to the analyte. The PSL spheres absorb the insulin from the analyte, which could be shown by the results of the test: only particles that contained a PSL sphere showed insulin peaks. BSQ considered such analyte where insulin is absorbed by PSL spheres as a simple cell model for end-to-end verification of the BSQ instrument.
Figure 47: Mean spectra for particles that result from empty droplets (‘empty partícles’, green curve) and for particles that contain a 2 µm PSL sphere (blue curve). Note the typical insulin peak at 6000 Dalton in the blue spectrum, illustrating that all insulin is absorbed by the PSL spheres.
MALDI MS spectral responses cell life cycle phases
At the start of the SICTEC project it was believed that for identification of bacteria from individual spectra so-called ‘biological noise’ calibration curve was needed. It was decided however, that a statistical approach for the data analysis is needed to deal with this type of noise. The scope of the work in this work package was initially focused on cell populations that were all in the same life cycle phase. After many discussions with potential end-users it appeared that there was much more interest in what the performance was of the instrument to discriminate between large sets of clinically relevant micro-organisms, see WP’s 3.7 and 3.8
AST development
The purpose of AST (Antibiotic SusceptibilityTesting) is to supply data in addition to the identity of a microorganism that describes for which antibiotics the organism is not susceptible. This is to allow targeted treatment with antibiotics, thus reducing the mis-use of antibiotics as a result of ‘blind treatment’. Currently, a lot of research is going on what spectral biomarkers can be used to perform AST using MALDI TOF mass spectrometry. One approach is to look for specific M/Z peaks in the spectra that represent beta lactamase, where the bacteria breaks down the antibiotics structure by hydrolysis. Unfortunately, this is just one mechanism of resistance, so more biomarkers should be identified to have a clincially revelant solution for AST using MALDI MS. In the SICTEC project we selected an approach where a database was to be constructed that could be used to discriminate between susceptible and resistant strains of one species. The BSQ single-cell MALDI-TOF MS system employs an alternative matrix (2-mercapto-4,5-dimethylthiazole) compared to the matrices traditionally used for identification of microorganisms with MALDI-TOF MS systems (e.g. α-Cyano-4-hydroxycinnamic acid). Using this matrix it is possible to discriminate between bacteria of different species. However, it was not clear which taxonomical resolution can be achieved using the BSQ 2-mercapto-4,5-dimethylthiazole matrix. To investigate the taxonomical resolution potential of this matrix, a collection of 96 wild-type Staphylococcus aureus strains were analyzed on a (classical) target-plate MALDI-TOF mass spectrometer. Fortunately, it appeared that the BSQ MALDI matrix offered the potential for sub-species characterization. With the potential of the BSQ MALDI matrix confirmed, the next step was to confirm this potential using the single-cell MALDI-MS instrument (see WP 3.8). Prior to the start of SICTEC project it was believed that producing meaningful Single Cell MALDI spectra needed ‘just in time mixing’ of additives to the analyte. The task of such additives was to perforate the bacterial cell walls in order to make proteomic molecules available for ionization. Fortunately, this pretreatment appeared not to be necessary. Good spectra could be produced without it, see for example a spectrum for E.coli. The quality of this spectrum is comparable to that of classical MALDI TOF instruments.. Moreover, in case spectra showed insufficient detail (see above) the solution always appeared to be another morphology of the co-crystallized bacteria - MALDI matrix.
Data-processor development
As already mentioned, the classification of single-particle spectra is based on a (BSQ patented) statistical approach, not on the foreseen ‘biological noise’calibration curve. So, for each single-particle spectrum a ‘score’ is calculated for a limited number of ‘spectral features’. Based on these scores a measure is derived for the probability that the spectrum results from the MS analysis of each of the organisms that comprise the reference library. The scope of this work package covers the design, test and evaluation for the BSQ data analysis module. In the Figure below an illustration is given of the stochastic character of single-particle spectra. In this figure the spectrum on the bottom line is an E-coli single-particle spectrum. On the top line the accumulated spectrum of 200 spectra is shown. In between, the number of accumulated spectra grows from 5 to 45 in steps of 5 spectra. Note that despite large variations on single-particle basis, the peak location is recognizable in the single-particle spectrum.
Figure 48: Accumulation of single cell spectra. Not that a lot of clutter has disappeared already after just 10 - 20 spectra.
The design of the data analysis module is based on the Bayes’ theorem. This is briefly explained.
When we know the identity of the bacterial cell, we know the probability of finding a large ion intensity (a peak) at a certain mass, based on measurements of spectra of different species in advance. An example of such a probability density function (PDF) is shown in the figure below for a simple binary mixture. For a certain feature (intensity at location x) the PDF for particles that contain the feature-inducing substance (blue) and the PDF for particles without the feature-inducing substance (green) are shown. The latter shows a sharp peak at 0 (no peak at location x), whereas the first PDF shows elevated values for normalized feature intensities up to 1.5.
Figure 49: Feature probability density function
Unfortunately this is not what we need. Instead of knowing the probability of finding a certain intensity when we know a substance is present, we need to know what the probability that a substance is present when we measure a certain intensity. This is where Bayes’ theorem comes in handy; because it performs exact this translation of the probability of finding a certain intensity when we know a substance is present into the probability that a substance is present when we measure a certain intensity. To use Bayes’ theorem, we need the probability density functions for all reference spectra: the contents of the ‘reference library’. To test and validate the data analysis approach, an experiment was done where a set of one E-coli strain and two Serratia m. strains were used as reference collection of bacteria. The Serratia m. strains were harvested in a nearby hospital during a suspected outbreak.
To demonstrate the performance of the data processing a two-stage approach was followed:
1. Construction of a reference library for the E-coli and two Serratia m. strains
2. Classification of the single particle spectra of these strains
In this case we constructed a mixture in the computer using single shot spectra of the E.coli and Serratia strains recorded earlier. This procedure allows constructing an artificial ‘mixture’ with known composition. For all of the necessary steps of the data analysis Matlab routines have been written and verified. In a first step, features (peaks) in spectra of the E-coli and two Serratia m. strains (reference organisms) were identified. The picture shows the accumulated spectra for the E-coli and Serratia m. strains, where the features (peaks) are indicated by “o” symbols.
Figure 50: Accumulated M/Z spectra for E-coli and two Serratia m. strains
Note that E-coli peaks are easy discernible from Serratia m. peaks, whereas Serratia m peaks are hardly discernible from each other. The classification result and performance using all features is summarized in the table below:
Table 9: Classification result and performance using all features
The table indicates that good discrimination between E-coli and Serratia strains can be obtained. Mutual discrimination between Serratia strains must be improved, which can probably be achieved by a selection of features. When we use the single particle spectra that were classified from the mixture as E-coli to construct a mean spectrum, it could be demonstrated that this reconstructed mean spectrum is mimics the E.coli reference spectrum to a high degree of accuracy.
Figure 51: Reference spectrum E-coli (top) and E-coli spectrum retrieved from mixture (bottom)
Preclinical testing - The scope of this work package comprises a first pre-clinical test series to evaluate the functionality of the BSQ instrument. Though the number of organisms used during this test exceeded 100 isolates representing 10 species, this number is far too low to consider this test as a full preclinical test. Hence, the test executed in this work package should be considered as a first step towards a full verification and quantification of the BiosparQ functionality. All tests were performed using organisms supplied by University of Applied Sciences in Leiden. The organisms used were chosen based on their clinical relevance, see table. For each species 1 to 30 strains were used.
Table 10: List of clinical relevance
The bacteria were cultured overnight in a liquid broth. From each culture, 250 ml was taken washed in demi and pelletized by centrifuging the suspension at 10000 g for 5 minutes. Finally the pellets were frozen at -20o C, for use at a later moment. Just before use, the pellets were suspended in the BSQ matrix solution and approximately 50 ml was transferred to the BFX CS printer. As indicated above the classification algorithms are based on a statistical approach. Input for the algorithms are normalized probability density functions (pdf’s) for the ion count at the masses corresponding with characteristic features (peaks) of the spectra. For each of the above organisms characterizing peak locations were determined, and at all peak locations for all organisms pdf’s were recorded. At least 1000 single particle spectra were used to characterize the pdf’s. To evaluate that the BSQ instrument is able to identify micro-organisms, accumulated spectra were produced for all organisms mentioned in the table, including a number of repeats, using at least 500 single-particle spectra. Minimum requirement for identification is that the spectra stemming from different organisms from the same species are stronger related than spectra stemming from organisms of different species. In other words, in a hierarchical clustering scheme, the accumulated spectra should cluster according to their species. In the figure the functionality of the BSQ system for a subset of the organisms of the table is illustrated in the form of a combined dendrogram (left side) and heatmap of the spectra (middle).
Figure 52: dendrogram (left, each line representing a single strain) and heatmap (middle, the red lines indicating the peaks in the spectra ) of the accumulated spectra for 5 species (indicated on the right) of the organisms listed in table.
As the figure shows, for this set of organisms, the different spectra are clustered according to their species, indicating that identification is possible. However, the figure also shows that the inter-test variability, resulting from repeats with organisms stemming from the same culture, is relatively large compared to species variation. Hence, for species more closely related than those present in the current set of micro-organisms, or for sub-species, the identification capability is not yet self-evident. At this moment it is not certain what the cause of the large inter-test variability is. Since the samples used for the repeats stem from the same culture, differences can only be introduced during the washing step, the resuspension step or the dispensing step. Hence, the current research is aimed at improving and quantification of these steps.
Clinical benchmarking
Like in the case with the pre-clinical test, it was not possible to perform full clinical testing within the scope of the SICTEC project. However, a big step forward could be made if the clinical test demonstrates significant clinical value of the combined BFX-BSQ system. Showing this to potential customers and/or (corporate) investors is a necessary step to bring the company to the next stage. To create the most value from a limited clinical test, a concrete clinical application had to be selected with great care. To demonstrate the capability of the instrument to the market, the application should require one or more of the unique selling points of the platform, where at the same time this application should be technologically “low hanging fruit” aimed to mitigate the risk of failure. With the SICTEC partner UAS-L and some clinical end-users in its network (Erasmus MC - Rotterdam, Maasstad Hospital - Rotterdam, Diaconessenhuis - Leiden and Bronovo Hospital - The Hague), workshops were organized to select the clinical application(s) to be tested. The outcome of the workshops was that the potential clinical applications for the platform are twofold, ie (i) diagnostics of infectious diseases and (ii) infection prevention / outbreak management. Based on the outcome of the workshop first a test plan was drafted in cooperation with the SICTEC partner UAS-L and its network to demonstrate the capabilities of the combined BFX-BSQ system to provide the infection control team with relevant clonal information of bacterial strains (so-called “wild type strains”) isolated from patient samples collected during actual (potential) outbreaks from 3 different hospitals. With the potential of 2-mercapto-4,5-dimethylthiazole as a BSQ MALDI matrix confirmed see WP 3.5) the next step was to confirm this potential using the single-cell MALDI-MS instrument. Firstly, the identification capability of the single-cell instrument had to be confirmed, using clinical samples. For this purpose, the same wild-type MRSA samples were used as for WP 3.5 and a set of 20 wild-type Saramis marcesens (received from the Diaconessenhuis Leiden Hospital) were analyzed. Minimum requirement is that it is possible to identify these samples using a set of previously recorded spectra of reference strains (containing Staphylococcus haemolyticus, Staphylococcus aureus, Enterococcus faecium, Enterococcus faecalis, Pseudomonas aeruginosa, Enterobacter cloacae, Klebsiella pneumoniae, Escherichia coli and Serratia marcescens strains). It appeared that both the MRSA wild-type sample spectra as well as the Serratia wild-type sample spectra cluster in the group with accompanying reference-strain spectra. Thus it appears that identification of unknown wild-type samples using the single-cell MALDI-MS technology is indeed feasible. Secondly, the taxonomical resolution potential is tested. For this purpose two sets of samples are used, one consisting of the MRSA samples used above and one consisting of 10 Serratia marcescens samples retrieved during a suspected outbreak in the Diaconessenhuis Leiden Hospital. In contrast to the results presented above, this time no clear clustering was obtained. Hence, it appeared that the spectra either contain too much clutter to enable clustering at a level below species level or that the instrument was not stable enough. Indeed, while repeats of this analysis yield the correct clustering of the MRSA strains within the aureus cluster, the inner cluster hierarchy of the MRSA strains was not constant. When wild-type Serratia marcescens samples were analysed the spectra again just like with MRSA a mix of conserved peaks and sample specific peaks, see figure below.
Figure 53: Mass spectra of ten Serratia marcescens samples, retrieved during a suspected outbreak and one Escherichia coli spectrum (for reference)
In contrast to the clustering of the MRSA set, in this case a meaningful clustering might be possible. A parallel analysis of the Serratia marcescens using AFLP indeed yielded results comparable to those obtained with the BSQ instrument. Since the experiments performed earlier on a traditional MALDI-MS instrument did yield meaningful typing results, the inconsistent typing results obtained with the single-cell MALDI instrument are not caused by the matrix used. Apparently, the variability of the particle morphology currently obtained is not sufficient for deeper analysis than identification on species level.
CE IVD preparation / planning
A detailed plan for CE certification according to Directive 98/79/EC on in vitro diagnostic medical devices was drafted. As BSQ was initially not familiar with this topic, help was requested from a consultant who was expert in the filed of IVD CE certification. This consultant, Quality Made Esay (QME) set up the structure of the document which was subsequently detailed by BSQ. Although the BiosparQ IVD device is not subjected to list A or B of the IVDD 98/79/EC, BSQ decided to set up an ISO 13485 Quality management system which will be assessed by a Notified Body. This Notified Body then annually assesses the quality system for continuing conformance. The following table shows a complete set of harmonized standards that BSQ has already identified and will certainly be considered.
Table 11: complete set of harmonized standards
In addition to the plans for CE IVD certification, the document was completed with exploitation planning. CE IVD certification is an important step towards exploitation for BSQ, it will open the possibility to market the spectrometer as an In Vitro Diagnostics medical device. Before CE certification, BSQ’s instrument will be marketed as a device “for research use only (RUO)” according to paragraph 5 of MEDDEV 2.14/2. This market is substantially smaller than the IVD market, however this market is important to generate and disseminate data that can be used in future commercial marketing activities. The overall timeline and milestones for CE IVD certification and exploitation are shown in the table below:
Figure 54: BSQ overall time line and milstones for exploitation
Typical sales numbers are as shown below.
Figure 55: BSQ projected sales numbers
IVD liquid handling platform - A CE-IVD certifiable liquid handling platform will be designed, built and evaluated both mechanically but also by comparing it to other existing CE-IVD devices. With an existing device from PASCA/PDG forming the base for this work, the first tasks were to identify the critical points in the machine design and based on that to optimize it in special regard to manufacturability, robustness, and costs. As already mentioned in WP2, the help of an external consultant (Manfred Augstein from velixx GmbH) is also used in this work-package. His knowledge helps ALU-FR in CE-IVD related design and implementation questions. The PASCA projects allowed to finalise a CAD design of an optimized and customer- attractive platform that could open the route to an industrial and CE IVD certified equipment. The work carried out in WP4 therefore aimed to start from finalized CAD, in order to establish the related 2D drawings, to produce the associated parts, to supply the required components and to build a prototype device, which would be functionally validated and then use for a clinical study. The aim of the clinical study being ultimately to benchmark the performance of the platform against the internal workflow as implemented at Erasmus MC.
Defining the robotic base from the PASCA CAD design - The CAD design obtained at the end of PASCA by PrimaDiag was reviewed by IMTEK which led to a series of minor changes. These were integrated in the initial CAD design, which was then use to establish the related 2D drawings. All standard components which exists already were bought. Non standard part (such as mechanical part and electronic wiring) were either subcontracted to specialized factories or made in house.
Figure 56 : Assembled base of the WP4 prototype.
Defining the required modules - The required modules to be added to the robotic base were selected in accordance with Erasmus MC. The major problem that arises was that the reference interlocutor changed every 9 months, obligating to modify the specs and the protocol that was requested to be automated. The interlocutors have been sequencially prof. Willem Van Leeuwen, Dr Rene Then you Wit, Then Wil Goessens).
In March 2015, two protocols were finally proposed by Erasmus. One directly related to Roche Magnapure system (which does automated DNA extraction from diverse starting materials : blood, urine, culture broth...). The main problem with this being that the reagent are proprietary and included in pre-packed cartridges which can hardly been used on the "WP4" instrument.
The second proposal was to use XNAR reagent kits from Sigma (ref : E7526, D5688, E3004) which are available as manual kits, and hence are more easily automatable.
Protocol derived from Roche Magnapure requires a pipette, a magnetic bloc and an 25-90°C incubator.
Protocol derived from XNAR reagent kits required no DNA purification by magnetic beads, and hence can be done with the pipette and the incubator only.
Figure 57: Selection of the modules based on the analyses of the biological protocol
Figure 58: Assembled prototype WP4 with its 3 modules.
Specs deriving from the analysis of the regulation - In addition to the talk which was given by Mr Manfred Augstein from Vellix, which was mostly focused on methodological approach, PrimaDiag followed some additional support from C-Reg Medical with gave more practical advices on the development that were required on the instrument in order to comply the 98/79/CE directive (CE IVD).
In particular, it was requested that :
1 ) the samples should be protected from contamination from the outside. Reversely, the operator should be protected against possible contamination from the samples (virus, bacteria, etc.). At a minimum, the system should then include a enclosing cabinet. With the possibility to include an HEPA filter.
2) cross contamination between experiments made in separate runs should be avoided. Hence a UV decontamination system should be included.
3) proper traceability should be insured. Hence, the final system should at a minimum include a barcode reader, if not a 2D matrix reader.
4) Pipette should include some level sensing capabilities in order to control the presence of the sample and the reagents when the process is running. Ultimately, the system should also be able to monitor the volume of liquid that will be / will have been handled during the runs.
Performance optimization - In this period has the positioning accuracy and the modularity of the WP2.2 prototype been optimized. In parallel, the prototype assembled in WP4 was functionally validated. As already mentioned in WP 2.2 the accuracy of the device has been improved by using alternative ball screws. The ball screws from HIWIN are available in different quality grades. The highest precision ball screws have at a travelling distance of 300 mm a maximal difference of only 3,5 µm. Another advantage is that the accuracy of the device can be adapted without changes in the design. After the manufacturing of the device the position accuracy will be tested with two different tests as mentioned in WP 2.4. One general test is oriented at the VDI/VDE guideline 3712. The scope of this test is the dispensing of a set of droplets at defined positions. Based on VDI/VDE guideline 3712 was a dispenser test plate designed (see Figure 59) which is going to be milled in aluminum.
Figure 59: Constructed dispenser test plate after VDI/VDE guideline 3712
For the dispensing test the test plate is placed on the working plate of the SCP and aligned optically by three alignment structures. On this plate then 50 droplets are spotted with ink at defined positions. These positions force the prototype to move different distances and into different directions. Afterwards the positions of the spotted droplets will be checked with an PC- DMIS measuring microscope. The differences to the intended positions show the accuracy of the whole dispensing system. The second test focuses on device positioning and uses a USAF Resolution Chart to check the position optically. This chart has several lines with different distances in between. Here the USAF resolution Chart is fixed to the working plate and the position of the Print head is checked with the camera of the SCP. The reticle of the camera is directed between two line pairs and the coordinates are noted (see Figure 60). The SCP now is moved ten times to a different position and back to the coordinates between the lines, where the camera takes a picture. On the pictures is afterwards checked, if the reticle is between the lines. Because of the different distances of the line pairs, it is possible to check the accuracy on different size levels. With the implemented optics, it is possible to see the line pairs down to a distance of 20 µm. The test was already performed at an existing dispenser of BioFluidix.
Figure 60: Reticle of the SCP camera placed between the 30 µm lines of the USAF Resolution Chart
In the new prototype are several points modular. The design is modular in the way, that it has defined interfaces, where the look can be changed. For example the sidebars can be exchanged or the look of the base platform can be adapted. The number of z-axis can also be changed between one and two.
The next modular point of the prototype is the type of Print head. As Print head either a commercial pipet, the PASCA single cell dispenser or a PipeJet dispenser device of Biofluidix can be mounted. The layout of external plugs is also adaptable, by changing a cover plate at the backside and the working plate. For example if no stroboscopic setup is needed, a blank cover plate can be used on the working plate. On its side. PrimaDiag validated the WP4 prototype that was built in WP4.1 which was obtained by assemblying the mechanical parts and components described upper. Thanks to the improvement already obtained with PASCA, the initial assembly was almost 100% satisfying. After a set of minor weaknesses being identified and corrected, the system was ready to accept the modules required to fulfill the Erasmus protocol (ie. Pipette, Magnetic Bloc and Incubator).
IVD specific design - For this we have started to check the mechanical and the electronic parts for compliance with the CE IVD directives on the prototype developed within WP2. For what concerns the prototype which was built within WP4, the works are related to the development of the enclosure cabinet, the UV lightning and the barcode reader. Ultimately, it would have been related to the integration of the level sensing in the pipette and on the HEPA filter. For what concerns the robotic base developed within WP2, the axis system has been reworked compared to the existing prototype. As already mentioned in WP 2.3 in the next prototype very accurate ball screws from HIWIN are used. The axis suspension is also changed. The motor and the screw are separated from each other and the forces are transferred from the motor to the axis by a belt. The belt decouples the motor and the screw and the axis has more traveling distance and is less susceptible to vibrations and torsion. Along CE IVD directives it is mandatory to prevent any harm from the user when operating the device. This especially refers to mechanical moving parts and electrical safety. Sharp edges are removed and bellows are used at the X-Axis to cover the linear ball guide. Cable routing was optimized and partly reconstructed to avoid damage by mechanical stress, moving axes and also to waive any external cable ducts. The cables are guided by drag chains and rooted inside the side bars (see Figure 61).
Under the electronic aspects of CE IVD, we implemented an emergency switch and all internal DC voltages are under 48 V. It was also paid attention that every part is grounded correctly, in order to ensure, that the device can pass the EMV Test. Further information on the design is also provided in Deliverable D2.1 “Finalized prototype design”.
Figure 61: Cable routing in the sidebars and drag chain. Internal cable management using drag chains to proper guide cable bunches besides moving axes parts. To improve visual appearance of the device all cables from Y-bridge and print head are internally routed through the side panels.
For what concerns the prototype built within WP4. The enclosing cabinet was designed, produced and assembled in a similar way to what is described earlier (assembly of the robotic base). The UV decontamination system was integrated using components available commercially. Proper measurement of the UV flux was made in order to insure the efficiency of the system. Reversely, absence of leaking UV outside the enclosure cabinet was also checked.
Fabrication cost optimisation – During this period the prototype has been designed with and we paid attention on lowering the cost. A quality test for the accuracy of the prototype has also been prepared. The design of the first prototype was done with special attention of cost reduction. This includes that standard parts were implemented if possible. For example, most of the design is based on item profile bars or alternative standard aluminum profiles, like L- or U- profiles. The plates are - if possible - manufactured only from the front or back and have a maximum thickness of 10 mm. Under these circumstances the plates can either be milled by a front panel milling company or manufactured via Laser/ Waterjet cutting. Most of the smaller design elements are planned as 3D- printed parts because of the low costs and the high variety of possibilities. Bending parts are recommended due to high fabrication costs and non-optimal precision. Cost reduction is also applied for the electronic parts. The motor controllers will be changed to PiBot from Repetier. This motor controller is used for self-made 3D- Printers and costs, including three stepper motor drivers and three optical end switches, less than 100 €. Special attention was paid on assembly of the final prototype. With regard to easy exchange of spindles and motors the Y-bridge was reconstructed. Electronics and embedded computer are now easily accessible and exchangeable through a double bottom design with plug and play connectors. For the quality test of the final equipment, the two accuracy tests introduced will be used. These tests are explained in WP 2.4 and will be used to confirm the technical specifications of the accuracy. One test is used for the accuracy of a whole set of dispensing steps and the other for the accuracy of the device positioning. On its side, PrimaDiag worked on the prototype developed within WP4. Each component and milled parts was reviewed in order to optimized the cost. The following table shows the price variation before and after cost optimization :
Table 12: cost optimization summary
Hence, the cost reduction process succeeded in dropping the platform price by 39%.
Documentation - CE IVD regulation is very strict and requires a precise definition of the product, how it is built, how it is used and what kind of results it can provide when running in routine. Also, this stage requires heavy preparation of a large range of technical files, user manuals and reports related to the systematic or standard tests. Obtaining the CE marking is the ultimate goal of this sub-WP. Writing and compiling the required documentation was not technically difficult, but is only a tedious process. When achieved, the compilation of the whole documents led to the writing of deliverable D4.1 ; and the machine was passed to BT and CEM normalized testing. Thanks to the preliminary works done already within PASCA, the WP4 prototype passed the norms without difficulties. The CE mark was then obtained on July 2014, just as RP2 started.
Pre-production instrument fabrication - After the WP4 prototype being produced and validated, two additional pre-industrial series had to be built, and then used respectively by PrimaDiag for additional characterization and development ; and by Erasmus MC for biological pre-validation and then clinical trial. The amount of fabricated instruments has been reduced. The instrument for BFX is no longer needed. Instead, one prototype based on the Biospot was. In advance to the assembly of the prototype, Dirk Buselmeier went to Primadiag in Paris and had a look on an already assembled device. During the visit, an assembling plan was created in accordance with the PDG team. This plan was created, in order that Dirk can assemble the prototypes on his own in Freiburg. For the communication and organization of the project, a Telephone conference between ALU FR and PDG was held every two weeks. After the arrival of the parts, the fabrication of the instruments started. The bridge of the instruments was assembled first (see Figure 62), followed by the base. This two parts have then been merged together (see Figure 63). To the bridge was added the X carrier with the Z-axis. The last mechanical part was to implement the working plate with the heating module. After the completion of the mechanics have the motor controllers and the power supplies been implemented. The electronic devices have then been connected and the cover plates been added. After the assembly were the instruments shipped back to PrimaDiag where the software was added. The plans were that one instrument would have stayed in Paris, on which further development will be made. The second instrument will have the CE IVD modules added (eg. the hood, the UV and the barcode)before it would be shipped to EMC for the benchmark.
Figure 62: (ALU): Assembled bridge.
Figure 63: (ALU): Bridge and base combined.
Figure 64: (ALU): Two assembled prototypes.
Unfortunately, the two units which arrived in PrimaDiag were not functional. And both units needed to be fully dismounted and remounted before being usable. The two units were then only available on end of April 2015. And this made the development of the HEPA filter and on the pipette (level sensing) difficult to achieve properly in the remaining time (2-4 months max).
Pre-production instrument characterization - Validating the prototype, and then the two pre-industrial series before launching the clinical trials. With the periodic changes of principal investigator at Erasmus during RP1, PrimaDiag took the liberty to contact some hospitals in its neighborhood (in the Paris region) and start the biological validation of the prototype on the field. Taking into account device capabilities (incubation, DNA purification and pipetting), it was proposed by the academic partners to use the DNA library preparation protocols upstream the NGS sequencers that would be more innovative and interesting when compared with the simple, 90's old fashion, DNA extraction protocols foreseen initially. The idea of using PrimaDiag system for such protocols was already identified as a good opportunity since the end of 2013. For review, one can refer to the final report of the PASCA project. Anyhow, this work around automating the NGS protocols started in June 2014 and lasted for the first campaign up to September 2014. This works were done in partnership with Genomics Department at the Brain and Spinal Cord Institute (La pitié Salpétrière Hospital) which confirmed the feasibility of the approach. A second campaign was then started with IntegraGen, a service provision company with sequence complete genomes and exomes for thirds parties. With them, several kits for preparing DNA and RNA library preps coming for some major reagent manufacturer (Illumina, Kapa, New England Biotech) were tested and validated. From October 2014 until January 2015, PrimaDiag then had the chance to collaborate wirth Roche Diagnostic which also has some reagent for the NGS sector. PrimaDiag system showed excellent performance for automating these kits, and this led Roche to propose to PrimaDiag to enter into a commercial agreement by April 2015 (see details in WP8 – dissemination). Taking into account the delay required to set up the partnership between Primadiag and Erasmus on RP1, and then the problem arising with the non functional pre-industrial series, it was not possible to prepare the plateform in time in order to program Erasmus prototol and then send it for biological pre-validation, and then the clinical trial.
IVD panel selection - Define the strains on which the biological prevalidation would be run. Then define the strain on which the benchmark and the clinical trial would be made. A first set of 8 strains were selected by Rene te Witt before he left on November 2014. But much information was lost with his depart
Mono-clonal cell culture - Following the same strategy used in the first period, INNOPROT has continued developing cell based products. As already commented the cell lines/models developed through this methodology have a very clear competitive advantage; they are truly monoclonal. This characteristic makes them more homogeneous in their response and the performance of the assays would be better. The SCM technology has been incorporated to INNOPROT production protocol and has been also included to improve existing cell products by “cleaning” them in order to make them monoclonal. As a result we have launched 22 new cell lines Out of these, 15 cell lines belong to the new proprietary technology developed by INNOPROT; NOMAD technology. NOMAD is a new family of biosensors designed to determine the GPCR activation through the β-arrestin activity or measuring changes in the concentration of second messengers within the cell. This new technology seems very promising and many potential customers have shown their interest. This is the complete list of cell lines developed during this second period.
List 2: developed cell lines
Apart from these cell lines, Innoprot has used SCM prototype in another 23 cell lines developments that have not come to an end due to not accomplish our quality standards, 20 cell lines produced for internal R&D projects and up to 50 cell based models required on-demand by our customers or included in other public funded projects. The data sheets of most of these products are available through the website www.innoprot.com. The improvements carried out in the SCM prototype included; a new printer head (component P9), and improved chip manufacture and reliability and an increased software stability. Overall this has represented a 5.4 fold increase in the cell lines produced per year with respect to the previous PASCA project. However this would be even better if would have had the fluorescent capabilities in the SCM because this would allow us to improve the positive clones sorting yield.
Develop an in chip drug screening assay - To use the fluorescent capabilities of the SCM to develop an “in chip” assay. This would be very useful in the validation phase of the cell based model production, saving time developing sub-optimal clones and saving the misuse of resources (the use of the screening platform for clone validation).. Moreover this could be also used for selecting clones with different degree of response to a given stimuli (ie. drugs, agonists or antagonist). Finally we would explore this feature for pre-screening purposes. This could be used for preliminary efficacy-toxicity assays in order to identify promising drug candidates or to discard molecules in labs lacking the resources of an expensive drug screening platform, or even to screen molecules in rare or low abundant cells. Unfortunately, the SCM with the fluorescent capabilities suffered several delays and currently it still does not work as expected. Thus, we could not use this interesting feature yet and therefore, the In-Chip drug screening development was not accomplished. The fluorescent excitation and detection was successful with both tGFP expressing U2OS cells (53% fluorescence intensity compared to the beads) and with fluorescent beads. However we wanted to use cell lines carrying a recently launched proprietary technology (NOMAD Biosensors Technology) and these cells are much less fluorescent (0.96% fluorescence intensity compared to the beads). It was not possible to detect the NOMAD cells with the SCM using fluorescent detection and therefore, we had to limit our tests to a high signal cell lines. On the other hand, we could not dispense cells detected by fluorescence. After trying all the suggestions and recommendations made by IMTEK, we were not able to dispense cells on a regular basis. The system excites certain spot with the laser and then detects cells compatible with given parameters (that is, cell size, roundness, fluorescence and so on) looking to a specific region determined previously (Region of interest, ROI). What was happening (and it is still unsolved) is that these two regions need to be aligned in order to decide whether a fluorescent particle within the ROI fulfils the parameters. The problem is that when you start a run, the printer head moves to the first well position (A1 in a 96 well plate) and starts to shoot drops, somehow this alignment is lost and therefore the ROI is not illuminated by the laser anymore and usually the ROI goes to somewhere else within the chip (many times even outside the reservoir in the plastic part). As a consequence the system does not find any particle fulfilling the given parameters and the dispensation does not take place. The figure 5.1 shows one example of this misalignment. This big problem is still ongoing and IMTEK could not addressed it due to the delay in the project. In addition, we needed a modification in the core.exe file in order to be able to measure the cell’s response to a given treatment. The first two modification of the original file did not work in the system and again most likely due to time constrains IMTEK did not provide us more versions to try. Nevertheless, it seems promising and when the last pitfalls would be overcome, the proposed assay could be developed in relative short time.
Upgrade of SCM machine - The original plan included a SCM prototype developed during the previous FP7-PASCA project, modified in order to be improved with fluorescence measuring characteristics. This feature to be developed and improved throughout SICTEC, was central for INNOPROTs activities. However, as already discussed previously in the PR1, this task was delayed due to difficulties in their development. Therefore we received the prototype in our facilities at month M20. On top of this, due to several problems (laser spot-ROI alignment, red led knob replacement, core modifications) we could not achieve some of our goals such as increasing the positive clones yield by selecting the fluorescent cells, sorting clones with different response levels, and developing an “in chip” assay.
Figure 65: Dispensation of one fluorescent nanoparticle. Picture sequence showing a single event dispensation of one fluorescent nanoparticle. In the panel 3, is clearly visible the misalignment of the ROI respect to the laser spot. ROI (Region of interest) is highlighted as a white circle line. The dispensed nanoparticle is highlighted within a white line contour. Panel 1; the particle is approaching the nozzle, Panel 2; the nanoparticle enters in the laser spot and is excited showing a green fluorescence, Panel 3; the system detect the fluorescent nanoparticle and verifies that meets all the set up criteria, Panel 4; the particle has been dispensed and is not visible in the image anymore.
Automated Single Cell Patch Clamping - The scope of this work-package is the integration of the single cell printing technology into Sophions single cell patch clamping platform. This was already planned in the former PASCA project, but could not be shown to work as a complete workflow: All individual steps worked from time to time, but not as a whole. Therefore, the work-package aims in increasing the overall robustness of the chips, the dispensers and detectors and the software protocols involved. Both the original PASCA SCM/Patch clamp platform and the new SICTEC Patch clamp platform require tight integration of IMTEK and Sophion hardware and software to work. Under work package 6 significant development efforts have been made not only at IMTEK but also at Sophion where new amplifier hardware and multiple new software components were required in order to allow integration with and evaluation of the IMTEK provided modules and upgrades in subsequent application work to validate these modules. This chapter therefore also contains brief descriptions of Sophions development and integration efforts as well as preliminary application evaluation. Apart from initial specification work - the Sophion SICTEC effort has been focused in the periods August-Sept 2014 and May-Nov. 2015. A comprehensive set of experiments on a first Single Cell Patch Clamping prototype (developed in PASCA and extended in SICTEC) are scheduled for August/September 2014. Therefore SPH, ALU-FR, and BXF agreed on a preliminary list of specifications that should enable a successful proof of concept. The results will be then used to further specify the next SICTEC prototype that will be designed.
List 3: Preliminary list of specifications
Remote controlled dispenser software - The SCM/Patch clamp application requires that the cell dispenser w. robot is operated in tight coordination with Sophion amplifiers and pressure systems to deliver, catch and characterize individual cells. The coordination and control is performed by Sophion software. The IMTEK software controlling the dispenser is completely separate from the Sophion software - so the execution coordination is achieved through Internet protocol based communication. The precise support of the protocols and the stability of both software throughout experiments lasting 30-60 minutes under a variety of usage patterns is critical to the success of the application which involve many complicated steps in addition to the required precise SCM cell delivery. The Internet protocol (IP) based protocols enabling the coordination between IMTEK and Sophion technology were upgraded in two rounds. In August 2014, new commands were added to the Biofludix PASCA/Biospot Software IP interface and the support for and usage for these commands were implemented in the Sophion Software. Those upgrades did improve the interoperability significantly - but could not solve some fundamental architectural flaws in the PASCA/Biospot software IP-communication design that caused it to crash often. In the second WP6.2 deliverable September 2015 the dispenser side a changed version Protocol was handled by a embedded process called Xcore - which controls the dispenser and the cell camera - while management of the XYZ-robot (holding the dispenser) and the TopView camera was taken over by Sophion software. This design was based on the assumption that the two software components could run on the same PC. The initial version of the new protocol was documented on August 14th and the final revisions were made on Sept. 10th. Development and testing of the new versions of the Sophion software interface layer and experiment execution coordination support for the new protocol was ongoing throughout August and September in addition to the new code required for Sophion to take over software control of the robot axis motion and TopView camera operation. This work has required a significant effort in man-months. Currently (early November 2015) the Sophion software and the IMTEK Xcore software and their respective USB and TCP/IP connected devices appears to have some apparently hardware problems co-inhabiting the same PC. In the currently most stable configuration the Xcore & cell camera connection crashes each time during the second cell dispense event. Measures to improve communication is from the Xcore to better allow Sophion to analyse and eliminate sources of instability. Significant improvements has been made to the IMTEK software’s ability to support the tight coordination of experiment execution through IP protocol communication - but the stability of the inter-operation has not yet reached a stability suited for mature product integration.
Automatic dispense parameter tuning - In order to allow robust automatic single cell patch clamping (WP 6) stable droplet generation is one of the key requirements. To minimize the necessary user interactions we will therefore implement an automatic dispense parameter tuning capability. For single cell patch clamping this should be done with respect to an optimal droplet placement precision. The basis for automatic dispense parameter tuning is a comprehensive picture of the dispensing process and the influence of the piezo parameters on the droplet formation – especially with respect to robustness and placement precision. As described in action report 6.3 we started to systematically evaluate the placement precision as function of the piezo parameters using a special test setup. We investigated the influence of the piezo downstroke velocity on the droplet placement precision. Although the other three parameters are important for the droplet formation they have a minor influence on the droplet placement performance once stable droplets are created. Hence they were set as follows:
Table 13: Piezo parameters used for stable droplet generation. The stroke length was set such that a stable droplet without satellites was created.
We found that the droplet placement precision improves with increasing downstroke velocity as long as a stable droplet without satellites is formed. This results in a reduced scattering radius as illustrated in Figure 66.
Figure 66: The scattering radius of the droplet placement depends on the piezo downstroke velocity. With increasing velocity the droplet placement precision improves. The graph shows data from two chips which were studied in detail. Medium was deionized water.
This trend was also observed when PBS was dispensed (using cartridges with a silane coated nozzle). We observed (and recorded) the droplet position data for 7 cartridges with PBS using high downstroke velocity values. In all cases a very good droplet placement precision as for deionized was found. As the optimal downstroke velocity varies between the chips we shifted our focus on the droplet velocity, which can be extracted from the image data provided by the stroboscopic camera.It turned out that all cartridges tested (with water and with PBS) showed a high droplet placement precision for droplet velocities of 1.5 – 1.85 m/s.
In a first investigation of the droplet placement precision we identified the relation of a sufficiently high droplet velocity and the placement precision. The droplet velocity can be extracted from image data from the stroboscopic camera, which is already part of the standard SCM. Therefore the droplet velocity might be an ideal state variable in terms of piezo parameter auto-tuning. So far we implemented the capability to automatically measure the droplet velocity into the SCM software. As the space of influencing parameters (medium, cell type, and environment) is large further evaluation of the placement precision is necessary.Finally we will introduce a feedback loop that allows automatic parameter tuning. Therefore it is further necessary to reliably detect droplet instabilities such as the formation of satellites with the stroboscopic camera. For full integration of the code we finally have to implement the auto-tuning module together with Biofluidix as they are in charge of the modified software for Sophion.
Sophion results - In SCM based patch clamping, reliable dispenser performance is a necessary - but far from sufficient condition for a even moderately successful experiment. Lack of chip droplet placement reliability is only manifested in the middle of the experiment execution - and typically means that 1-2 hours of preparation is partially wasted. On the other hand failures elsewhere in the setup (like software crashes) may prevent a flawless chip from proving its worth. We have tested multiple batches of silane coated chips since August 2014 with both versions of the SICTEC setup in various stages of development and instability. Is is hard to make statistical evaluations under such variable conditions - but generally the silane coated chips have been performing better than the chips available during the late PASCA work. Chips from some batches persistently performed well when given the chance - while the chips from other batches had a very mixed performance leading to many failed experiments due to unreliable droplet delivery. The not-well performing chips were often associated with priming issues and a narrow down stroke velocity window between droplet formation onset and satellite droplet formation. In the patch clamp application we will eventually both good droplet placement reliability for some chips - but also reliable performance from most chips and batches. This highlights the importance of automation of the dispenser chip production effort in WP1. No automatic dispense parameter tuning functionality or poor chip performance features were demonstrated to us or made available for evaluation at Sophion during the SICTEC project (or since). Thus - while it is good to read that IMTEK & Biofluidix is making progress in this field - we are not in a position to comment on the performance or SCM patch clamp application suitability of these features (which is the focus of WP6). Currently dispense parameter tuning is still being done fully manually in a user interface developed at Sophion since the Xcore software just provides a simple text-based IP protocol interface for basic parameter value changes. The improved reliability of the silane coated chips have meant that manual dispense parameter tuning has generally been easier than before.
Integrateable, contained dispenser module - Key for a functioning print head for single cell patch clamping are dispenser chips that allow depositing single living cells on patch clamp holes in a robust manner. Therefore, we first focused on improving the dispenser chips in terms of droplet placement precision and long term stability. This report summarized our recent work. The chips (and therefore also the nozzles) are rendered hydrophilic during the production process by an oxygen plasma treatment to allow reliable filling with media. During dispensing the hydrophilic surface properties leads to wetting of the silicon/pyrex chip around the nozzle. There are no problems when deionized water is used as medium for dispensing. However, if phosphate saline solutions are used, salt precipitates around the nozzle during dispensing. This results in a significant decrease of the droplet placement precision after 2-10 min of dispensing and often leads to complete failure of the chip. Therefore we developed a protocol for a hydrophobic coating off the nozzle while the hydrophilicity of the interior parts of the chips is preserved. We then investigated the effect on the chip performance (with focus on droplet placement precision). Further we started to improve the coating step in order to allow for robust integration in our chip supply line. Coating is achieved by vapour phase deposition of an organo-silane. Chip performance was studied by dispensing PBS on a glass slide. A camera is placed below the slide which allows to observe the dispensed droplets on the slide. 15 Cartridges with silane coated nozzles were tested for at least 60 minutes. Both the total failure and the droplet positioning accuracy were qualitatively compared with untreated chips. So far all of the 15 silane coated cartridges that were tested performed very well with PBS in terms of dispensing. All of them were used for at least 60 minutes of dispensing and none of them failed. Liquid film formation around the nozzle and therefore the precipitation of salt could be suppressed. The effect of the surface coating can be also seen in Figure 67 which shows a cartridge with an untreated (left) and a silane coated chip nozzle (right): After gently applying pressure on the water-filled reservoir, the uncoated chip is immediately wetted while on the coated chip a round droplet can be observed (high contact angle).
Figure 67: The effect of the surface treatment can be examined by a quick test: If pressure is applied on the water filled reservoirs, an untreated chip will be rapidly wetted by the water (left) while a droplet forms on the hydrophobic nozzle with coating (right).
At the moment we are working on the full integration of the coating step into our chip supply line. Next, we have to further investigate the shelf life, mechanical, and chemical stability of the silane coating.
In most of the applications (including single cell patch clamping) dispenser chips are required to work reliable for extended time durations (> 30 minutes). Therefore we performed long term dispensing experiments with 10 µm fluorescently labeled beads and PBS using three uncoated and three silane coated dispenser chips. Arrays of 34x70 beads were printed on microscope slides which were subsequently analyzed with a microarray scanner. All three uncoated chips failed completely within one hour and thereby only about one half of a 34x70 array could be printed. Further, in some cases the placement precision rapidly deteriorated as can be seen in Figure 68 (left). In contrast, the silane coated chips allowed for extended printing times: We dispensed three arrays with each of them within 5 hours and none of them failed.
Figure 68: The non-coated chips all failed after printing about one half of the 34x70 array and in some cases the precision deteriorated rapidly (left). The silane coated chips dispensed beads reliably for 5 hours without failure. Here one of the full 34x70 arrays of single beads is shown (right).
In order to investigate the performance of the printer in terms of droplet placement precision water and phosphate buffered saline (PBS) were dispensed on a glass slide. Three cameras were setup such that (i) the position of the droplet on the substrate after dispensing can be recorded (bottom camera), (ii) the nozzle of the chip can be observed during dispensing (nozzle camera), and (iii) the droplet can be monitored by the stroboscopic camera. Both the bottom camera and the nozzle camera are triggered each time a droplet was dispensed and allowed to pass by the vacuum suction hole. Note, that the trigger was coupled to the vacuum valve opening event. The SCM software was modified such that the printer can continuously dispense at a fixed frequency (specified in the GUI). In order to evaluate the images produced by the bottom camera, the frequency of droplets arriving on the glass slide must not exceed 0.5 Hz to provide enough time for droplet evaporation. Therefore the standard experimental procedure was as following:
● Droplets were produced by the printer at a frequency of 5 Hz.
(Piezo shooting frequency = 5 Hz )
● Only every 20th droplet passes the vacuum suction valve.
(Vacuum valve actuation frequency = 0.25 Hz)
● Each time a droplet passed the vacuum suction valve images of the substrate and the nozzle were taken. The stroboscopic camera was used before and after the experiment in order to calculate the droplet velocity and to adjust the piezo parameters.
● For each parameter set 1000 droplets were dispensed. Of these, 50 droplets reached the glass slide and were monitored by the bottom camera.
● The working distance between the dispenser and the substrate was 5 mm.
The image data produced by the bottom camera are evaluated with Matlab. Figure 69 shows a droplet detected by the algorithm. The graph shows the x and y-deviation of 50 droplets. The droplet placement precision can be expressed in terms of the standard deviation , calculated from the positions of the 50 droplets. In order to obtain an intuitive measurement tool, we define the scattering radius
,
which indicates the radius of the circle that encloses 95 % of the dispensed droplets given a normal distribution. The scattering radius is illustrated by the red circle in Figure 69 (right).
Figure 69: The image data from the bottom camera is processed by a Matlab sript: For each experiment 50 droplets are detected (left) and their positions and the resulting scattering radius are plotted (right).
We evaluated the image data of 3 cartridges (representative for the 7 we tested). The resulting scattering radii are given in Table 14.
Table 14: Results of Droplet Placement Precision Measurements of three chips. Both with water and PBS a placement precision < 60 µm is reached repeatedly if the droplet velocity is set between 1.5 and 1.85 m/s.
We therefore think, that by using silanized nozzles we can reach a droplet placement precision of for continuous dispensing of PBS over extended times. Although we have not yet investigated the influence off idle time on the placement precision in great detail with the new test setup, we saw that the placement precision can be recovered after dispensing breaks of 1-2 minutes.
We will further examine idle times and use-cases reflecting Sophions workflow for patch clamping in the future. There has been the issue of sedimentation of cells in the reservoir of the cartridge, especially if cell dispensing was performed for extended time periods with the SCM. This led to a decrease in cell concentration over time. To fix this issue, an agitation module has been designed and tested at IMTEK. It consists of a microcontroller and a roller pump which agitates the reservoir via a disposable pipette tip To test the agitation module we dispensed microbeads for 30 minutes. Figure 70 shows that with the automated agitation module it was possible to dispense single beads with a constant frequency over time. The module will be integrated into the existing PASCA hardware and into future print head designs for patch clamping.
Figure 70: Number of single micro bead dispenses over time. A single column in the diagram corresponds to 1 minute. The concentration of beads arriving at the nozzle is constant over time.
Results at Sophion - No design changes to the PASCA dispenser module were made during the first SICTEC upgrade in August 2014. The redesigned dispenser module from Sept 2015 is discussed under redesigned WP6. The agitation module upgrade mounted on the enhanced PASCA SCM/Patch clamp setup worked in principle - but a mismatch between the rolling pump and the control electronics meant that it either sucked the cell reservoir dry or had to be run “in reverse” making correctly timed insertion critical. This lead to a lot of failures and meant that the module became a risk factor and was unmounted for many experiments. Extended software and hardware problems with the SICTEC 2nd generation setup has meant that we have limited experience with using the 2nd generation agitation feature.
Development of the next generation SICTEC patch clamping prototype - Major part of WP 6 as defined in the SICTEC DOW is a next generation single cell dispenser module (printhead) that will be specifically designed for patch clamping (see WP6.4 and WP 6.5) based on the experiences with the first prototype (PASCA machine) . Although proof-of-concept could not be fully achieved, the experiments using the upgraded prototype helped to define a set of major issues that will be addressed with the design of the next SICTEC WP6 patch clamping prototype:
● Software complexity must be further reduced, such that the user has to interact with a single GUI only.
● The necessary user interactions with the software must be further reduced as another step towards a fully automated solution.
● The complexity of the hardware must be reduced and the stability increased. In detail:
• A more robust amplifier system that does not need to be controlled from an additional PC.
• A more integrated printhead, that supports further automation such as dispenser parameter tuning, offset compensation and a fast & smooth single cell dispensing.
• A new robotic system that is mechanically more stable and can be controlled easier than the old PASCA setup which was reconstructed several times (as stated by Sophion).
Given this basis for the design of the next generation SICTEC patch clamping system and the remaining time of the project, WP6 was redefined by ALU-FR and Sophion. In summary, the actions in the reporting period regarding the development of the next generation SICTEC patch clamping prototype included:
● Redefinition of WP6 together with Sophion.
● Preparation of formalities for the EU with respect to the WP redefinition (ongoing).
● Specification of the next generation SICTEC patch clamping system (ongoing).
The Sophion evaluation of the first SICTEC delivery revealed that the upgraded PASCA platform based on the complex ino teroparation of 3 PCs had become too complex and unstable to be usable in practice: The 1st generation Robot SCM05 had precision and wiring stability problems - the latter due to multiple repairs and fixes made during PASCA. Swapping it with a more stable and precise PASCA 2nd generation robot was not an option and Sophion had no other suitable candidate for a a XYZ robot that could carry the dispenser module under development at IMTEK. It was agreed that IMTEK would find a tabletop type XYZ robot controllable by Internet Protocol (IP) and that Sophion would take care Software
control for the robot, the topview camera and ensure the development of a tabletop friendly
amplifier system. These development tasks required significant development efforts at Sophion in the period May-September 2015. It could be realized due to synergies with another Sophion EU project (HiMicro). The Festo XYZ axis sold to IMTEK as IO controllable turned out not to be and resulted in 25+ man-days of wasted development effort. Some specifications came rather late but Sophion had almost all components ready for the 6 day integration effort with the robot&dispenser components delivered and installed by Julian Riba Sept. 1st-10th 2015. Some adjustments were required but after 7 workdays all individual components and control mechanisms seemed to be in working order. The full setup was now running from a single PC by the Sophion QTracks platform for experiment execution.
Figure 71: Integrated Sictec g2 SCM/Patch clamp prototype - showing new integrated despenser mounted on Festo XYZ axis setup - mounted over the new “flat” Sophion amplifier box. The 16 channel pressure system is placed under the “robot” and the single controlling PC can be seen next to it.
The stobo-lighting failed within a few hours of Julian Ribas departure and left the platform unusable for close to two weeks. After Sept. 24 the stroboscope light worked again - but since then lab work has been hampered by driver and possible USB comminication conflicts on communication controller PC causing frequent software crashes. Hardware upgrades have helped a little - but the IMTEK XCore process currently (Nov 6th) still crashes every time we try to dispense the second cell on a 16 well PASCA plate. Extended debugging efforts and the inability to do experiments with more than one site/cell at the time has so far limited our ability to achieve proof of concept with the new setup.
Figure 72: One of multiple SCM dispensed cells caught on the 2um patch hole. The top graph shows the usual ~2MOhm resistance of an un-occupied patch hole - while the second plot shows resistances of ~160MOhm indicating successful positioning of a cell.
Preliminary colclusion regarding the new SICTEC g2 setup are:
● Elimination of separate strobo station with extra camera and slow robot roundtrip is a significant improvement in setup complexity and experimental procedure (except illumination point source stablility).
● Mirror-based cell camera observation of cell window make cell observation in cell dispensing position possible and removes need for z-motion during cell delivery is a significant progress. The safer placement of cell camera optics is also a clear advantage.
● Sophion’s takeover of top-view and Robot XYZ motion control has worked well despite some initial problems.
● Direct communication with the XCore process controlling the dispenser and cell camera generally works well but the absence of User interface and event logging - and our lack of access to the source code makes it very hard to identify and fix problems when the Xcore needs to work together with other software.
● In the Sophion SCM05 PASCA setup the TopView camera could directly observe the droplet delivery site without need for horizontal motion. This was very benefital given the cruzial role of optical feedback based positioning and performance evaluation in the SCM patch clamp appliction. The redesigned cell print module delivered in Sept 2015 did not follow the angeled delivery proposed in WP 6.4 left no space for such a topview camera placement: The topview camera was relocated making a 2x13cm X-axis roundtrip motion necessary in droplet delivery evaluation feedback loop. Since the X-axis carries the Y-axis and the full dispenser module the accelerations of this roundtrip introduces significant vibrations in the whole setup and no meaningful camera observation or cell delivery can be performed before they die out. This is so far a major downside of the new dispenser module.
Potential Impact:
The ownership of project results and the management of intellectual property - whether foreground or background - are handled following the recommendations of the EC as described in the “Guide to intellectual properties rules for FP7 projects”, and according to the consortium agreement. The wording used in the following (e.g. “foreground”, “background”, “IPR”, etc.) corresponds to the definitions described therein in detail.
One of the basic principles the participants have previously agreed on and still agree on is that ownership of background is not affected by participation in the project, i.e. all proprietary information, technologies, know-how and IPR remain property of the project partner that had owned it prior to the project. In order to be able to use such background of all participants, the participants additionally agree to grant each other mutually a time limited, non-exclusive, royalty free licence to use all available background (IPR included) for the purpose of the project, during the official duration of the project, on a “need to use“-basis (i.e. project partners that do / do not have to use specific background to carry out their research tasks are / are not granted a licence). This includes that the RTD perfomers gain access to the knowledge/the background of the SME partners or of the background of other RTD performers when required, e.g. to setup the system to fit into the SME’s apparatus or when tuning the dispensing parameters onto a specific application. This will also be done on a “need to use” basis. Thus, the full set of background can be applied and considered for the purpose of the project, while no commercial opportunity is lost for the background owning partner to generate revenues through products, service and licences outside of the project. Major background and IPR of the individual participants relevant to the project are listed below. The list which was provided in the DoW still holds true and is provided in table 1 for reference.
Table 15: Access rights to important background information. No access rights are abbreviated by N.A. Every partner has access to those background and IPR that is required for its specific exploitation route and product vision.
In summary it can be stated, that the current IPR situation and the regulations for accessing background and exploiting foreground are appropriate to enable all SMEs a commercially viable exploitation within their own application. Furthermore, this scheme allows the RTD performers to access the background of the other partners (be it SME or RTD) they need to fulfil their contractual research.
As far as foreground is concerned that is developed as part of the R&D activities carried out within the project, the general principle is applied that foreground resulting from the project is owned by the participant generating it. Such foreground might be related towards a complete product like for example considered in WP2 and WP4 with respect to design, manufacture, software and application, or it can be related to components such as specific optical detections systems developed in WP1, to software like developed in WP1 and WP2 or to processes and applications like developed for example in WP1 and WP5. When such foreground is generated jointly (i.e. where the separate parts of some result cannot be attributed to different participants), it is jointly owned like already drafted in the DoW. In this case the participants share benefits and burdens of the foreground (e.g. in terms of IPR rights and expenses for IPR protection) according to their shares in the invention (e.g. as documented on the patent application forms). The only exception to this principle concerns the RTD partners: The foreground generated by a RTD partner (whether alone or as part of a joint invention with an SME partner) belongs completely to the SME paying for the research work (referred to as “SME Client”), since the RTD partners are fully financed through the R&D subcontracts. This regulation is the default regulation for RTD partners proposed for this funding scheme in the guide for applications. In the rare case where more than one SME is paying for a specific R&D subcontract to a RTD partner (e.g. WP1.2 or WP1.7) joint ownership of the foreground is acquired according to the shares by which the individual SMEs contribute to the financing of the R&D subcontract (i.e. 30% of financing of the R&D contract will earn a SME 30% ownership of the foreground / IPR generated within the R&D contract), see also table 2. The commercial exploitation by sales or sub-licencing of any jointly owned foreground is decided by the owners in a democratic way according to their shares. Also revenues from such licences or sales are split as well according to the shares of ownership. RTD performers are free to use the results for further research (non-commercial exploitation) if those results are not identified as confidential. Besides the RTD performers can publish the results after obtaining the agreement of the SMEPs.
Table 16: Overview on the project results (including knowledge) to be acquired by the participants and their type of exploitation
The main principle governing the above regulations is that no SME should be excluded from the right of a royalty free use of foreground, even if this foreground is later IPR protected by other members of the consortium. In order to extend this principle even further and to enable an efficient exploitation of the foreground generated within the project, the SME project partners furthermore agree that all non-tangible foregrounds generated through any R&D activity should be exploitable by all other participating SMEs for their own business. Therefore, all owners of foreground (incl. IPR) are granting to all other SMEs involved in this project from the start (namely: BFX, BSQ, PDG, INPT, SPH) a non-recoverable, non-exclusive, non-transferable, royalty free, permanent licence for using any non-tangible foreground generated within the project for their own commercial purposes of any kind, excluding the right to grant sub-licences to third parties. After the end of the project this free license can either be prolonged or will be changed to a non-free license which then bases on fair and reasonable conditions.
Tangible foreground (e.g. pre-production prototypes) however, belongs to those SME(s) paying for the development of this foreground. Therefore, other SMEs have to ask the owner(s) for using such foreground, whereby the permission to do so should not be unreasonably be withheld. This regulation enables all SMEs to benefit equally from all parts of the R&D work-plan, creates synergies and alignment in the individual developments and hopefully fosters cooperation and joint exploitation. Nevertheless, the ownership of foreground (incl. IPR) includes the right to grant sub-licences or to sell IPR, still vests with the one SME (or group of SMEs for joint inventions) being the owner of a specific foreground or IPR. Thus, also further commercial exploitation through cooperation with third parties remains possible for the owner(s).
In order to separate background clearly from foreground the project partners claim ownership of background as listed in detail in section 4. This list is going to be updated prior to project start. In summary, each of the participants contributes with a unique proprietary technology to the project and requires certain background from other project partners. This relationship is summarized within table 1 describing the access rights of the individual partners to background. As far as the central background related to the core SCM technology is concerned, the project partner ALU-FR has declared to be willing to enter into good faith negotiations about licensing of background and IPR with all project partners (see letter of intent (LOI) / in Annex B ) after the project and to give a free license during the project. As it can be seen in the LOI, this licensing will either be free or base on fair and reasonable conditions. Thus, the core SCM technology is directly accessible for all SME partners at fair conditions and their freedom to operate is maintained as far as the SCM technology itself is concerned. The same applies to all other background (as listed in B3.2.1) from the other partners: A free license is granted during the project and after the end of the project this free license can either be prolonged or will be changed to a non-free license which then bases on fair and reasonable conditions. An overview on access rights to important background is provided in the following table 1. As can be seen from this table, each SME has ownership of the specific technology required for their envisioned product and can access the background related to droplet dispensing technology and SCM technology through licensing or by direct cooperation e.g. through joint fabrication or hardware / software purchase. Thus, freedom to operate with respect to the background of the project partners is established individually as well as collectively.
Freedom to operate and third party IPR - As far as IPR of third parties are concerned, there is currently no evidence that the freedom to operate and the commercial exploitation of background and foreground considered within this project might be limited. This opinion is based on several reasons:
1. The freedom-to-operate analysis performed within the PASCA project (Del. 10.1) did not identify any IPR that is potentially in conflict with the SCM technology (i.e. inkjet-like drop-on-demand printing of single cells) like applied in this project.
2. For BSQ a freedom-to-operate analysis was performed in 2011 by a 3rd party (Arnold + Siedsma, Amsterdam) in the framework of the Financing Round A due diligence. Also here no IP was identified that could be in conflict with the single cell MALDI TOF technology.
3. The positive outcome of the patent examination of the PCT EP application of ALU-FR (PCT/EP2010/058170). Following patents: US2008/02086751, EP 0421406, US 4318480, GB 1555091A have been considered by the European patent office during examination of PCT/EP2010/058170. The fact that this application was preliminarily approved on 06.09.2012 is a strong indication for the SCM technology being commercially exploitable for the SMEs under licence of ALU-FR.
There are no existing anticipated business agreements which may impose limitations on the subsequent exploitation or information or inventions generated as a result of the project.
Background of the project partners - This list was provided in the DoW and has been checked for completeness during the project runtime. By filing this deliverable the SCM core EP patent was finally granted and also granted in other countries. Furthermore, this patent has been licensed to the ALU-FR spinoff Cytena GmbH which takes over all obligations resulting from previous LOI’s or contracts. As the founders of Cytena were also active in the former PASCA and in the SICTEC project, they are well known to the SICTEC partners and a good and productive cooperation has been established.
BioFluidix (BFX)
• NanoJet dispensing technology: Simple, non-exclusive licence to use following IPR :
o EP 0961655 (and derived national patents)
o EP 1049538 (and derived national patents)
o DE 198 02 367
• Technical design and fabrication of SCM automation system (as developed within the PASCA project, joint ownership with PrimaDiag)
• IP rights for the software developed within the PASCA project to control the SCM prototype instrument (software version Biospot BS12.33.13 Core V1.3.15)
BiosparQ B.V. (BSQ)
Patents and patent applications:
• Method and device for detecting and identifying bio-aerosol particles in the air (WO 02/052246 (WO246)) and patents derived therefrom
• MALDI matrix and MALDI method (WO 2009/064180A1 (WO 180)) and patents derived therefrom
• Method and apparatus for identification of biological material (WO 2010/021548 (WO 548)) and patents derived therefrom
• Method for classification of a sample on the basis of spectral data, method creating a database and method for using this database, and corresponding computer program, data storage medium and system (2009015) and patents derived therefrom
Software
• IP rights for all BSQ data processor software (Matlab scripts), system controller software, GUI (source codes)
• IP rights for all BSQ reference database software and contents (source codes, classifiers)
• Models (CFD, performance, etc.)
Data
• IP rights for all experience gathered during the testing and the development of BSQ technology as reported in TNO project N02/30, DV2009 A 417 and TNO DV2009-418 and all BSQ-TECH-RPT’s up to kick-off of SICTEC project.
• IP rights for all descriptions of the current BSQ instrument design up to kick-off of SICTEC project.
PrimaDiag SAS (PDG)
• Versatile robotic Platform (design partly shared with BFX)
• Pipettes in the ml - µl range
• Controlling software dedicated to routine biology works (Liquid Handling)
• DNA extraction and DNA purification methods (patent pending)
• Robotic configuration able in performing sample preparation for PCR and qPCR analysis (for research use only).
Innoprot (INPT)
• Fluorescent bio-sensor technology (Patent pending)
• Single cell based production method for monoclonal cell lines developed within the PASCA project
Sophion Bioscience A/S (SPH)
Patents and patent applications
• System for Electrophys. Measurements (WO 0229492A3/US6932893B2)
• Substrate and method for measuring the electrophysiological properties of cell membranes (Protruding Rim : EP1495105B1/US2004005696A1)
Software/ hardware
• IP rights to QTracks platform for planning, execution, supervision and result management for multi-device electrophysiology instruments
• IP rights to software for structured, large scale electrophysiology data analysis and presentation
• IP rights to multichannel patch clamp amplifier design and control software
• IP rights to design and production techniques of multi-well patch clamp chips - including Si-die integration in plastic substrates.
• background to techniques for combining automated patch clamp platform operation with single cell dispenser operation developed within the PASCA project
The Institute of Microsystem Technology, University of Freiburg (ALU-FR)
• SCM core technology for drop-on-demand printing of single cells: Granted EP patent EP10724847.8 US Patent US13/707,086 and, WO2011/154042 A1 → Licensed to Cytena GmbH with Cytena taking over all obligations arising from previous letter of intents (LOIs) or other contracts such as the SICTEC consortium agreement.
Erasmus Medical Center (EMC)
• The bacterial strain collections and the associated data that are used in the SICTEC project are considered EMC IP and can only be used for research purposes. EMC is willing to negotiate a license in case BSQ wants to commercialize the reference database in which the EMC bacterial strains are stried.
Dissemination activities
Exhibitions - Most external partner or customer attraction during the project was either done through the website or direct communication by the partners. This lead to more than 20 contacts, e.g. BioSparq talked to all major IVD industries during the project (e.g. bioMérieux, Becton Dickinson, Biocartis, Bruker, Accelerate Diagnostics). However, the technology was showcased at the following exhibition and meetings:
• MicroTas 2013 conference and exhibition (October 2013, Freiburg, Germany): Presentation of the prototype device and live-lab demonstration on single cell printing.
• Sophion user meeting (August 2015, Kopenhagen, Denmark). Presentation of the current status and potential applications.
Furthermore, BioFluidix attended BioJapan from October 15th to 17th in Yokohama, Japan with its distribution partner MicroJet. As MicroJet also has a background in cell dispensing, various talks were given to interested parties based on which further development is currently being discussed.
Publications - The following publication were submitted to journals or presented at conferences with reference to the SICTEC project:
● Gross A, Schoendube J, Zimmermann S, Steeb M, Zengerle R, Koltay P. “Technologies for Single-Cell Isolation”. Int J Mol Sci. 2015 Jul 24;16(8):16897-919. doi: 10.3390/ijms160816897.
● 16-18 September 2014: Single Cell Genomics meeting 2014 – Utrecht, The Netherlands. Riba R, Niemöller JC, Zimmermann S, Schoendube J, Bleul S, Koltay P, Zengerle R, Claus R, Becker H, Gross “A. Single-cell printing for the genetic analysis of cancer cells”. (Poster: P074)
● 25-28 April 2015: the 25th European Congress of Clinical Microbiology and Infectious Diseases, Copenhagen, Denmark. Dekter, Hinke. “Single cell MALDI-TOF based identification of strains obtained from hospitalized patients” (Talk)
● 19-22 July 2015: 6th International Conference on Analysis of Microbial Cells at the Single Cell Level. Retz, Austria. Riba J, Gleichmann T, Zengerle R, Koltay P. “Label-free sorting and deposition of single bacterial cells using the Single-Cell Printer technology” (Poster and Talk)
● Upcoming: 11-14 October 2015: New Approaches and Concepts in Microbiology. EMBL Heidelberg, Germany. Riba J, Gleichmann T, Zengerle R, Koltay P, Zimmermann S. “A new tool for label-free isolation and deposition of bacterial cells” (Poster)
Available patent portfolio - As the main intension of the works performed in the SICTEC project was to increase the maturity and robustness of the technologies involved to enable the SME performers for a fast adoption to the markets, no new patents applications were created during the project. Nevertheless, some patents were granted, increasing the possibility to enter the markets quickly. Thus, the following patents have been granted or licensed to the project partners.
BioFluidix (BFX)
• NanoJet dispensing technology: Simple, non-exclusive licence to use following IPR :
o EP 0961655 (and derived national patents)
o EP 1049538 (and derived national patents)
o DE 198 02 367
BiosparQ B.V. (BSQ)
Patents and patent applications:
• Method and device for detecting and identifying bio-aerosol particles in the air (WO 02/052246 (WO246)) and patents derived therefrom
• MALDI matrix and MALDI method (WO 2009/064180A1 (WO 180)) and patents derived therefrom
• Method and apparatus for identification of biological material (WO 2010/021548 (WO 548)) and patents derived therefrom
• Method for classification of a sample on the basis of spectral data, method creating a database and method for using this database, and corresponding computer program, data storage medium and system (WO 2013154425) and patents derived therefrom
PrimaDiag SAS (PDG)
• Magnetic attraction module, robot including such a module, and method for using such a module or such a robot on magnetic beads (WO 2014083165)
Sophion Bioscience A/S (SPH)
• System for Electrophys. Measurements (WO 0229492A3/US6932893B2)
• Substrate and method for measuring the electrophysiological properties of cell membranes (Protruding Rim : EP1495105B1/US2004005696A1)
The Institute of Microsystem Technology, University of Freiburg (ALU-FR)
• SCM core technology for drop-on-demand printing of single cells: : EP patent EP10724847.8 US Patent US13/707,086 and, WO2011/154042 A1 and patents derived therefrom, now licensed to Cytena GmbH (see D8.3 for reference)
List of Websites:
will be available with all contact data also after the project