Skip to main content
European Commission logo print header

Nanosystems for the early Diagnosis of Neurodegenerative diseases

Final Report Summary - NADINE (Nanosystems for the early Diagnosis of Neurodegenerative diseases)

Executive Summary:
The consortium of 18 partners from 9 countries has successfully finalized work within the NaDiNe (Nanosystems for early Diagnosis of Neurodegenerative diseases) project. The project’s main goal was to provide diagnostic tools to determine the onset of a neurodegenerative diseases (such as Alzheimer’s disease or Parkinson’s diseases) by using blood or serum samples, a procedure that is potentially amenable to regular screening of a certain segment of the population, as it is very little invasive, works on a small sample volume and can be performed at a cost level that is not prohibitive for public health systems. This was an extremely ambitious goal, as the concentration level of biomarker molecules indicative of these neurodegenerative diseases can be expected to be very low in the early (onset) stages of the diseases. Furthermore, to be able to distinguish between different types of neurodegenerative diseases and to improve the fidelity of the diagnosis, it is not sufficient to rely on a single biomarker, but on a group of biomarker, providing a signature or fingerprint.

During the project period, a wealth of approaches utilizing miniaturization and novel nanotechnologies have been developed and tested in combination with advanced biochemical protocols. It was a chosen strategy from the start to pursue several avenues, both to be able to explore different possibilities, but especially also to reduce the risk of failure of one single approach. In the project period, some of the chosen technologies (e.g. evanescent wave detection, self-assembled bar-coded particles) were demonstrated, but deemed to not provide sufficient performance or sensitivity to be useful for NaDiNe’s main applications, but could certainly be used for other (medical) applications. In the end, novel magnetic nanoparticles with custom-made antibodies and fine-tuned assay protocols were used for an immunoprecipitation-based enrichment step, to be followed by either a protein array with fluorescence detection or a mass spectrometric approach with nanopillar arrays for further up-concentration. Also, in parallel, an electrochemical assay again using antibody-coated magnetic microparticles was used as a further risk-reducing measure. Further refinements are possible by incorporating miniaturized separation systems as well to aid in the fingerprinting. Microdevices were fabricated from polymer materials and operated using sophisticated fluidic control hardware and software, established by one of the project partners. This led ultimately to integrated lab demonstrator devices for the above-mentioned processes. These were fully functional, albeit not yet so user-friendly as to be used directly by the medical partners.

Instead, validation of the developed systems was performed in the labs of the developing partners, using non-clinical and clinical samples provided by the medical partners. Four developed technologies (microarray, micro droplet array, electrochemical assay, and mass spectrometric analysis) were tasked to determine diagnostic levels in these medical samples. The medical partners provided guidance for handling and preparation of the samples, and also joined in critically evaluating the data, together with bioinformatics experts. In particular, data from clinically relevant samples was compared to previously obtained results with the current gold standard method. Overall, while on a limited data set, the data produced by NaDiNe technology aligns very well with established methods, both in terms of (diagnostically relevant) limits of detection achieved as well as showing the ability to clearly distinguish between different diagnostics groups, and thus ultimately shows the potential to provide diagnostic answers with less sample, at earlier times in the disease stage and at significantly reduced costs. In conclusion, the results of the NaDiNe underline the significance of using sophisticated biochemistry combined with nanotechnology to provide next-generation medical diagnosis tools.

Project Context and Objectives:
From its inception, the NaDiNe project was designed to follow a structure, where the initial three years were earmarked to explore a variety of innovative nanotechnological approaches for the sensitive detection of relevant biomarkers for the early diagnosis of neurodegenerative diseases. The following (4th) year was then planned to be used for consolidating the most promising technological bits and bringing them together in a prototype device, which, then, in the last (5th) year, would be tested and validated in close collaboration with the medical partners on clinically relevant samples.
It was also already envisioned from the start, that not all technological tracks that the consortium initially had suggested would prove successful (because of the inherent risks associated with developing novel techniques and diagnostics approaches), but that, through iterative processes, technological solutions not sensitive enough or not robust enough would be pruned away, while others were further developed, improved upon and potentially used in conjunction with other developed bits (i.e. the combination of a novel sample preparation method with a highly sensitive detection or sensing method). This strategy of parallel development with regular checkpoints and decision points had always been deemed absolutely necessary, not least as a risk reducing effort as several of the initially suggested routes looked extremely promising (if successful) but also were associated with a high risk of failure.
Furthermore, apart from the overall time plan (development – consolidation and prototyping – testing and evaluation) and the parallel, risk-reducing strategy in the development phase, the final important design element for the project was to have a clear idea of the typically necessary steps in the workflow of a successful analysis and particularly a successful diagnosis. This entails a solid strategy for sample preparation and pre-treatment, highly sensitive detection or sensing approaches that are compatible with the sample preparation methods, instruments and instrumental control to automate processes and minimize operator mistakes, but, very importantly, also to involve bioinformatics tools to process data such that clinically meaningful results can be extracted and statistically validated. All these three “design elements” were reflected in the way the work packages for the NaDiNe project were put together, where they were placed in the project timeline, and how they were meant to provide input to each other.
The objectives of WP 1 were thus to focus in the preparation, characterisation, modification and testing of novel micro and nanoparticles. The use of such particles was important for both sample preparation purposes as well as detection and sensing purposes. In order to meet one of the key requirements of the NaDiNe project, namely to detect minute concentrations of biomarkers in the onset phase of the neurodegenerative disease, it was necessary to include sample pre-concentrations techniques into the overall workflow. Here, immune-labelled micro and nanoparticles provide a large surface when incubated with the sample, a high specificity when labelled with the best available antibodies, and the possibility to be manipulated by, e.g. magnetic forces inside microfluidic channel networks. Nanoparticles can also be made with special optical and electrochemical properties (e.g. quantum dots) making them very attractive for sensing platforms.
In WP 2, technologies for fabricating microfluidic devices and preparing them for use with biological matrices (blood, plasma, cerebrospinal fluid, ...) were investigated along with ways to provide the basic fluidic handling elements (pumping, valving, connections, ...) as well as strategies for integration and automation. Here, considerations for what parts of a final system should be re-usable and what parts should be disposable (because of direct contact with patient samples) were important. A lot of emphasis was put on working with polymer substrates to make the actual chips, but other materials were also investigated if necessary from, e.g. a detection point of view. As with the actual detection approaches themselves, material choices were tested, evaluated and adjusted during the course of the project.
WP 3 then focused on a range of different nanotechnological approaches to facilitate ultrasensensitive detection of selected biomarkers in biological matrices. The original selection included novel optical, electrochemical and mass spectrometry-based approaches. Early on, it became obvious that the waveguide-based techniques, while inherently very elegant, would not provide the necessary sensitivity. The same became clear, albeit a bit later, for the quantum-dot approaches. The latter would probably have worked in combination with the sample pre-concentration approaches, but they were not selected for further development, as other techniques proved more promising. During the course of the project, a set of criteria (figures of merit, compatibilities, price, ...) had been established to test and compare the various techniques, and this was used during the regular evaluation and decision meetings. Finally, an additional approach that had been developed in parallel for – initially – a different application, was adopted late in the project and proved successful for the detection of biomarkers for neurodegenerative diseases.
Further important techniques, which are not directly detection methods, were developed in WP4. This included in particular sample prep approaches, but also other techniques which were intended to be combined or hyphenated with detection, such as labelling, separation, compartmentalization, and immunocapturing. This WP also included some high-risk developments, such as the convective self-assembly of bar-coded particles, that were included as they potentially could provide highly novel functionalities, but it was expected that their potential to be fully developed to be included in a prototype device within the project period would be rather limited. One of the techniques planned in this WP was abandoned fairly early on, as it became clear – through discussions with the medical partners - that a clean-up of whole blood would not really be necessary as most clinical samples are available as plasma samples already.
Something that was discussed heavily already in the proposal writing phase of the project, but which was not possible to pin down clearly enough from the start was the selection of the actual biomarkers to be included. The choice of biomarkers – apart from being crucial for diagnostic success – influences almost all steps of the anticipated workflow: the choice and production of antibodies, the choices for sample enrichment, the design of the assay and all its protocols, the choice of materials and chemicals, the detection parameters and so on. Finding biomarkers for neurodegenerative diseases was not the objective of the NaDiNe project, but it was part of WP 5 to determine a roster of possible biomarkers that are clinically relevant and can help detecting both the onset of neurodegenerative diseases and also help in subtyping them. Alas, all throughout the project, it became apparent that the biomarker field was in constant flux and new potential biomarkers were described in the literature all the time. The temptation to use “better” biomarkers was obvious from earlier struggles with proteins and peptides that were very tedious to work with, but the amount of time necessary to, again and again, develop antibodies and optimize assays and detection protocols was prohibitive and this caused some frustration among the partners. It was probably here where the consortium should have made a clear decision early and stuck with it, instead of hoping for a “better” set of biomarkers to appear. This caused some of the delays that took from the prototyping and testing phase of the project.
Further tasks in WP 5 pertained to the optimization and comparison of the developed technologies, and to provide input for the prototype to be built in WP 6. In view of the obtained results, it became necessary to realistically assess and adjust the objectives after year 3, and to focus on a reduced and easier problem first, namely the detection of Abeta-peptides and ApoE in mainly cerebrospinal fluid samples. A panel of criteria was established and the most promising techniques were tested against these criteria for possible inclusion in the prototype. A final task in WP 5 was to develop and provide bioinformatics tools to a) improve biomarker selection through datamining of already published results, and b) statistically assess data provided by using the prototype for clinical samples. The latter had to be adjusted drastically as not enough data was produced in the end to require a more concerted biostatistical treatment.
The prototype itself was to be developed (and, potentially, iterated) in WP 6. Due to the delay in optimizing some of the most promising nanotechnologies, a true prototype device was not manufactured. Instead, several demonstrator units were provided demonstrating the usefulness of the chosen nanotechnologies. The distinction here being that a prototype would be tested at the clinic by trained, but non-technical operators, whereas a demonstrator, while fully functioning, is typically not very user-friendly yet and often only can be operated in a fully equipped laboratory. In the end, hybrids of these two cases were made available, already with a fairly high degree of user-friendliness and automation as well as commercialization potential, but still being tested in a technical lab rather than a hospital setting. The main reasons for deviating from the original objectives are lying in longer development times for the nanotechnologies as breakthroughs are always “just around the corner”, and challenges with the biochemistry, where antibodies, functionalization protocols or assays just would not work reliably and/or with too much batch-to-batch variation.
WP 7 was the workpackage where everything was supposed to come together, i.e. where clinical samples provided by the medical partners could be tested on the prototype developed in WP 6, and where these tests could be validated by, e.g. comparison with gold standard methods. This WP was reduced in time caused by the delays of the earlier WPs and the amount of testing done was consequently much less than anticipated. Also, instead of one prototype, four workflows including nanotechnological solutions from the partners were tested. Still, as described further below, excellent results were obtained that demonstrated that clinically relevant diagnostic results could be achieved.
Finally, WP 8 and 9 were predominantly concerned with administrative tasks and tasks related to dissemination and exploitation of results obtained throughout the project. It is noteworthy to mention that the inclusion of a partner almost exclusively in charge of designing and implementing risk management is something to be recommended for other projects. This both provided some valuable guidance in running the project and also forced the coordinator and the steering group to regularly take stock and adjust objectives and strategies as necessary. While not all issues could be anticipated and prevented, the awareness of potential risks and ways to react clearly contributed to the overall success of the project.

Project Results:
Main scientific and technological results (max 25 pages)
Task 1.1 - Innovative magnetic nanoparticles
We have developed superparamagnetic maghemite nanoparticles (γ-Fe2O3), 9 nm in size according to TEM, by coprecipitation of Fe2+ and Fe3+ chlorides with aqueous ammonia; the resulting magnetite was then oxidized with sodium hypochlorite to chemically stable maghemite (Figure 1a). The particles were subsequently coated with a shell of hydrophilic and biocompatible poly(N,N-dimethylacrylamide) (PDMAAm). The PDMAAm-γ-Fe2O3 nanoparticles were shown to be non-cytotoxic and they were intensively phagocytosized by the mammalian macrophages. High potential of PDMAAm-γ-Fe2O3 nanoparticles in diagnosis of phagocytary activity as well as in delivery of various biomolecules, such as specific proteins, can be predicted. Optionally, the γ-Fe2O3 nanoparticles were coated with D-mannose, poly(L-lysine), silica and its derivatives, poly(vinylpyrrolidone) (PVP), poly(ethylene glycol) (PEG) and poly(2-hydroxyethyl methacrylate) (PHEMA). The superparamagnetic γ-Fe2O3 nanoparticles are promising in cell applications as the cells labelled with the nanoparticles can be non-invasively monitored and easily magnetically separated and redispersed in water solutions on removing the external magnetic field.

a
b
c
Figure 1. (a, b) TEM micrographs of (a) γ-Fe2O3 nanoparticles synthesized by coprecipitation and (b) Fe3O4 particles prepared by thermal decompositon of Fe3+ oleate in 1-octadecene. (c) SEM micrographs of magnetic PHEMA microspheres.
Another approach for synthesis of monodisperse superparamagnetic iron oxide nanoparticles of controlled size consisted of thermal decomposition of organic iron compounds in different high-boiling solvents (1-octadecene, eicosane and trioctylamine) in the presence of oleic acid and/or oleylamine (Figure 1b). The compounds included Fe(III) oleate and mandelate, formed from FeCl3 and the respective acids. The size of the nanoparticles was easily tuned to 8–27 nm by varying the experimental conditions. The nanoparticles were characterized using X-ray diffraction, transmission electron microscopy, and magnetization measurements. The hydrophobic coating of the particles was analyzed by thermogravimetric analysis and atomic absorption spectroscopy. To make the particles biocompatible and water dispersible, nontoxic hydrophilic poly(ethylene glycol) derivatives containing phosphonic and hydroxamic groups were synthesized and used for phase transfer of hydrophobic particles into water using a ligand-exchange procedure. Saturation magnetization of nanoparticles was between 40 and 95 Am2/kg.
Our second main achievement during the project consisted of development of magnetic macroporous poly(glycidyl methacrylate) (PGMA) and poly(2-hydroxyethyl methacrylate) (PHEMA) microspheres containing carboxyl and amino groups by multi-step swelling and polymerization followed by precipitation of iron oxide inside the pores (Figure 1c). The functional groups enabled bioactive ligands of various sizes and chemical structures to couple covalently. Microspheres containing amino and/or carboxyl groups were then coated with α,ω-bis-carboxy PEG and amino-terminated PEG or α-methoxy-ω-amino PEG, optionally with sulfobetain. Adsorption of bovine serum albumin (BSA), γ-globulin, 125I-BSA, pepsin, and chymotrypsin on neat and PEGylated microspheres was determined by UV–VIS spectroscopy of supernatants and eluates or by measurement of radioactivity in an ionization chamber. The carboxyl groups of the magnetic PGMA microspheres were also conjugated with primary amino groups of mouse monoclonal DO-1 antibody using conventional carbodiimide chemistry. The efficiency of protein p53 capture and the degree of nonspecific adsorption on neat and PEG-coated magnetic microspheres were determined by western blot analysis. The applicability of these monodisperse magnetic microspheres in biospecific catalysis and bioaffinity separation was confirmed by coupling with the enzyme trypsin and IgG. The indisputable advantage of magnetic carriers consists in the ability to control their transport by applying a magnetic field. Magnetic microspheres modified by the above-described procedure can thus be utilized in standard immunoprecipitation techniques.
A nanobiosensor based on the use of porous magnetic microspheres as efficient capturing/pre-concentrating platform was presented for detection of Alzheimer's disease (AD) biomarkers (by partner ICN). These microspheres prepared by a multistep swelling polymerization combined with iron oxide precipitation afford carboxyl functional groups suitable for immobilization of antibodies on the particle surface allowing an enhanced efficiency in the capturing of AD biomarkers from human serum samples. The AD biomarker signalling was generated by gold nanoparticle (AuNP) tags monitored through their electrocatalytic effect towards hydrogen evolution reaction. Novel properties of porous magnetic microspheres in terms of high functionality and high active area available for enhanced catalytic activity of the captured AuNPs electrocatalytic tags were exploited for the first time. A thorough characterization by scanning transmission electron microscope in high angle annular dark field mode demonstrated the enhanced ability of porous magnetic microspheres to capture a higher quantity of analyte and consequently of electrocatalytic label, when compared with commercially available microspheres. The optimized and characterized porous magnetic microspheres were also applied for the first time for the detection of beta amyloid and ApoE at clinical relevant levels in cerebrospinal fluid, serum and plasma samples of patients suffering from AD.
Task 1.2 - Nanoparticles for signal amplification
Several types of novel NPs were synthesized: (i) different-sized gold nanoparticles (AuNPs) (5-80 nm) usable as labels in immunoassays and as blockers in nanochannels-based sensing systems, (ii) CdS QDs NPs (2 nm) with improved characteristics (size, stability and fluorescent/electrochemical properties) for electro-chemical/optical applications, (iii) iridium oxide nanoparticles (IrO2NPs) (12 nm) with novel electrocatalytic detection method at pH 7 (through water oxidation reaction - WOR), and (iv) Prussian blue nanoparticles (PBNPs) (4 nm) usable as red-ox indicator in nanochannel-based sensing systems (Fig. 2). All these new nanoparticles and nanoparticle/biomolecule conjugates have been characterized using different optical techniques (TEM, Z-potential, UV-Vis spectroscopy) and their electrochemical behavior has also been successfully evaluated.

A new method for oriented functionalization of gold nanoparticles (AuNPs) with antibodies has been developed. It consists in taking advantage of oriented ionic attraction and strong covalent bond formation using EDC chemistry. The result is an oriented and strong functionalization, which creates an excellent label to be used in many different types of immunosensors.
Novel magnetic poly(2-hydroxyethyl methacrylate) (PHEMA) microspheres containing carboxylic groups (synthesized by partner MACRO) were used as platforms for the covalent and oriented immobilization of antibodies in a magnetosandwich immunoassay for the detection of ApoE, one of the AD biomarkers, using AuNPs (20 nm sized) as electrocatalytic labels.

Figure 2. (A) TEM images of the synthesized CdS Quantum dots and the corresponding optical characterizations. (B) TEM images of the synthesized IrO2 nanoparticles and summary of their electrocatalytic activity on the water oxidation reaction in both voltammetric and chronoamperometric measurements. (C) TEM images of the synthesized Prussian Blue nanoparticles and the corresponding optical and electrochemical characterizations.

ApoE detection by using porous magnetic microspheres and electrocatalytic gold nanoparticles were compared with detection of another AD biomarker, Beta amyloid. High porosity of magnetic microspheres (PMMs) allows a more efficient capturing of AD biomarkers in magnetoimmunoassays. AuNP tags and electrocatalytic hydrogen evolution (HER) were successfully used for biomarkers detection. Beta amyloid and ApoE in CSF, serum and plasma samples of patients suffering from AD were evaluated with achieving performances of clinical relevance.
The detailed study of the analytical parameters related to the new catalytic nanoparticles (LOD, linear range, etc.) and their application for the detection of AD biomarkers (i.e. ApoE) were performed also. ApoE detection through electrocatalytic water oxidation induced by iridium oxide nanoparticles was one such application. High catalytic effect of IrO2 NPs tags towards the Water Oxidation Reaction (WOR) enables sensitive chronoamperometric quantification at pH 7. Sensitive method for the ApoE detection in human plasma (LOD: 68 ng/mL) following a magnetoimmunoassay format was developed. Both the IrO2 NPs tags and electrocatalytic detection method present many advantages (simplicity, cost, etc.) compared with previously reported ones based on quantum dots (QDs) or Hydrogen Evolution Reaction (HER).
Task 1.3 – Biofunctionalisation
The objectives of the task have been successfully accomplished. Common and advanced procedures for surface modification of magnetic micro/nanoparticles with subsequent immobilization of specific IgG were applied to provide final functional immunosorbents to requesting partners (CI, DS, UPS). With the aim to minimize the rate of nonspecific adsorption and tendency of particles to self-aggregate and adhere to the walls of a microfluidic devices we successfully modified the surface of particles by poly(ethylene glycol) (PEG) derivatives and hyaluronic acid (HA) fragments. The list of modified magnetic micro/nanoparticles (commercial or newly developed (MACRO) and list of polymers used for modifications is given below (Tab. 1). Testing the behavior of modified particles in magnetic field and their ability to create self-organized columns in microfluidic systems were performed in cooperation with CI partner.

Table 1. List of tested carriers and polymers used for surface modification.
Various methods such as zeta potential measurement, infrared spectroscopy, scanning electron microscopy, laser diffraction, anti-PEG ELISA assay successfully proved the coating of particles. Grafting of specific antibodies (different clones of anti-Aβ, anti-Tau, anti-ApoE, anti-ubiquitin IgG molecules) in a proper way on magnetic micro/nanoparticles without adversely affected binding function was the major goal in the construction of an efficient immunosorbents. Different immobilization approaches were optimized and tested. Prepared carriers have been tested not only in batch, but mainly in the microfluidic systems prepared by partners, their colloidal stability was confirmed. Experiments with real biological samples (CSF, whole blood) were also successfully performed. Comparison of specific IgG immobilization efficiency using various particles is shown in Figure 3.

Figure 3. The immobilization efficiency of IgG molecules to the various commercial and particles newly developed by partner MACRO
Prepared immunosorbents with desired specificity (directed against Tau protein, Aβ peptides, ApoE, ubiquitin, etc.) were applied for immunocapturing of target proteins associated with AD from spiked/real complex biological materials such as human plasma, serum or CSF. Optimization of immunocapturing protocol was necessary (batch vs. μIP binding/elution conditions, subsequent detection method). As an example, we successfully detected Aβ peptides in CSF (Fig. 4). For affinity and specificity evaluation of applied antibodies we also developed dot-blot affinity test using NH4SCN as chaotropic agent and magnetic beads-based epitope extraction technique.

Figure 4. Immunocapturing of Aβ peptides confirmed by Western blot.
In cooperation with partner ICN the quantum dots were coupled with specific antibodies to prepare conjugates for amplification of the detection signal and increasing the sensitivity of the analysis. Apolipoprotein E (ApoE) antigen (BioVision, USA) and specific polyclonal anti-ApoE antibody (partner MB) were used as a model system. Various strategies were tested and optimized. One of the innovative approaches is based on fixation of the antibody in the immunocomplex with specific antigen covalently coupled on the surface of magnetic particles wit subsequent QDs conjugation based on carbodiimide chemistry and elution of formed QDs-conjugate.
Task 1.4 - Nanoparticle characterization and safety
Nanostructural characterization, i.e. morphology, size and particle size distribution in the dry state were characterized by SEM and TEM microscopy; micrographs of cross-section of the microspheres provided information about interior of the particles. Dynamic light scattering determined hydrodynamic size (polydispersity) of particles in water and also the zeta-potential, values of which generally predict if the particles aggregate in water or not. ATR-FTIR spectra confirmed success of modification reactions, i.e. presence of functional groups on surface of the particles. Elemental analysis quantified the amount of functional groups, atomic absorption spectroscopy and thermogravimetric analysis then analyzed Fe content in the magnetic microspheres. Magnetic measurements, in particular the determination of hysteresis loops and saturation magnetization using magnetometer and measurement of temperature dependence of magnetic susceptibility, confirmed magnetic character of particles. Moreover, X-ray powder analysis enabled to differentiate between forms of iron oxides, Fe3O4 vs. γ-Fe2O3. Toxicity of the magnetic nanoparticles was evaluated in biological experiments, in particular, during engulfment of iron oxide nanoparticles by macrophages or mesenchymal stem cells. Non-toxicity (biocompatibility) of surface-modified iron oxide nanoparticles was thus confirmed if concentration of the particles was moderate. As the particles are contained in a colloidal fluid, there is no risk of inhalation. Hence, the potentially hazardous effects are effectively avoided.
Newly developed particles were delivered to the following partners: Univerzita Pardubice (UPCE), Institut Curie (CI), Institut Catalá de Nanotecnologia (ICN), Université Paris-Sud (UPS), Moravian-Biotechnology (MB), Diagnoswiss (DS), and Instituto di Chimica del Riconosciato Moleculare (ICRM).
Task 2.1 - Microfluidic flow control
The task consisted in developing an external fluid transport controller able to monitor reliably and accurately the transport of fluids in and between the modules. We have developed three solutions (combining hardware and software) to control transport of fluids in modules.
1. A valving platform enabling the sequential injection of fluid in microfluidic system (Figure 5): useful in WP4 and WP6 for partners CI and ICRM with the following specifications:
- Low internal volume to limit loss of samples
- Compatible with a large range of pressures
- Response time < 1s
- Compact, biocompatible, compatible with the software

Figure 5: Hardware and software of the valving platform.

2. Integrated electrodes in pressurized reservoirs to apply electric fields in the module of pre-concentration (WP4, partner CI). The system of electrodes has been developed but the pre-treatment module has been abandoned in M12.
3. Flow-rate control module: useful in WP4 and WP6 for partners CI and ICRM
- Precise control of the flow: control of the flow-rate and/or the volume
- High stability of the flow
- Ability to perform long experiments
The flow-rate control module, patented during the project, is an algorithm that adjusts automatically the pressures to reach the flow-rate set points even in complex microsystems. It minimizes cross-talk between channels, compensates partial clogging, and allows stable and pulse-less flow (Figure 6).

Figure 6: Principle and software of the flow-rate control module.
The contribution of the developments for the different partners and WPs are listed below:
- Fluidised bed (Institut Curie):
o Integration of 3 valves (2way/3port): biocompatibility, low internal volume, allowing integration and automation
o Use of flow-rate control module coupled with the pressure controller: feedback on flow-rate, preserve magnetic bead structure during protein capture, control of the volume injected in the system combining high flow stability and short response time.
- Microarray (ICRM):
o Integration of a rotative valve: biocompatibility, low internal volume, allowing the injection of 7 different solutions automatically, without cross contamination
o Use of flow-rate control module coupled with the pressure controller: feedback on flow-rate, control of the volume injected in the system combining high flow stability and short response time.
Task 2.2 - Low-cost microfabrication
For medical applications it is essential to work with materials that do not need to be cleaned after having been in contact with medical (patient) samples. Considering such factors as base cost of the material, ease of fabrication, fabrication costs, biocompatibility, and disposal, this points to polymer-based materials. A number of polymer materials have been considered, investigated and tested for use in this project.

Our first standard choice of material for fabrication needs in this project was poly methyl methacrylate (PMMA), as it is cheap, readily available and easy to machine. Certain parts of the prototype, which might come in contact with the sample were made in either polycarbonate (PC) or aluminum to facilitate more efficient cleaning.
It was decided to fabricate the microfluidic chip itself from poly dimethoxy siloxane (PDMS). That way DTU could simply supply the mould for the users of the platform to make their own chips on a daily basis. Furthermore, since PDMS is a soft rubber it is not necessary to bond the microfluidic part to the silicon chip, a simple clamping feature is sufficient to maintain leak proof operation.
It was immediately decided to go for a modular approach when fabricating the prototype. That way each individual part could easily be replaced and improved without affecting the other parts. The underlying intent was to have a setup consisting of several simple parts, which each could easily be adapted and refabricated to improve performance and handling.
Since the fluidic chip was to be made of PDMS, It was necessary to create a mould. The mould consists of three parts: bottom, frame and top. Each of these components serves a specific and separate purpose. The bottom part of the mould is the most delicate, as it carries the features of the microfluidic chamber and channels. The frame is simply a spacer used to define the thickness of the PDMS chip. The top part of the mould is where the in- and outlet positions are defined. Hence, in line with the modular approach, each part can be adapted individually and replaced if needed.
In order to be able to easily place the square silicon chip in the slide scanner for detection, a custom made aluminum slide was developed. The aluminum slide has similar dimensions as a regular glass slide except for a recess in one end, where the silicon chip fits in. The combined thickness of the slide and chip is just below 1 mm, which allows it to fit into the slide scanner. Aluminum was chosen as material for the slide due to its relatively high stiffness and chemical resistance, which allows the slide to be reused.
It is not enough to just place the PDMS chip on top of the silicon chip; in order to prevent leakage, a slight clamping force must be provided. Hence the chip holder platform and the chip holder top were developed. The platform has a simple recess to fit the aluminum slide, and a cavity to help removing the slide from the platform again. The top part of the holder has a window to prevent the clamping force from collapsing the microfluidic chamber. The top part also has three holes for metallic pins, which provides fool proof alignment between the window position and the chamber in the PDMS chip. The top part also features a threaded hole for connecting the reused fluidic tubings. The non-reusable tubings, such as the sample inlet and the outlet, are connected by squeezing the tubes into the PDMS. The top part is made from polycarbonate rather than PMMA as it needs cleaning with ethanol.
Task 2.3 - Surface treatment and characterization
The hydrophobic nature of disposable plastic, PDMS and thiol-ene devices presents a considerable drawback to microfluidics applications. Much effort has been devoted to modifying the surface of polymeric materials to adjust their wettability, adhesion and biocompatibility.
A simple approach to reduce dramatically the hydrophobicity for a wide group of materials including cyclic olefin copolymer (COC), polyethylene terephthalate (PET), polycarbonate (PC) and polytetrafluoroethylene (PTFE), polydimethylsiloxane (PDMS) and the newly introduced thiol-ene polymers was developed, to avoid protein unspecific binding and suppress electroosmotic flow in microchannel for microchip electrophoresis.
The process, easily implementable in any laboratory, including microfabrication clean room facilities, consists in coatings prepared via the “dip and rinse” approach by immersing the plasma-oxidized materials into an aqueous solution containing two different poly(dimethylacrylamide) based copolymers. The polymers, comprising a segment of poly(dimethylacrylamide) and incorporating a silane co-monomer, were functionalized with either N-acryloyloxysuccinimide (NAS), poly-(DMA-co-NAS-co-MAPS), or glycidyl methacrylate (GMA), (poly(DMA-co-GMA-co-MAPS). An extensive characterization of the hydrophilicity, binding capacity and resistance to unspecific protein adsorption of these coatings was performed. Polymers were distributed among the partners for use in their specific applications.
Task 2.4 - Disposable flow through microfluidic and nanoparticle based platforms
A flexible hybrid polydimethylsiloxane (PDMS)–polycarbonate (PC) microfluidic chip with integrated screen printed electrodes (SPE) was fabricated and applied for electrochemical quantum dots (QDs) detection. The developed device combines the advantages of flexible microfluidic chips, such as their low cost, the possibility to be disposable and amenable to mass production, with the advantages of electrochemistry for its facility of integration and the possibility to miniaturize the analytical device. Due to the interest in biosensing applications in general and particularly the great demand for labelling alternatives in affinity biosensors, the electrochemistry of cadmium sulfide quantum dots (CdS QDs) was evaluated. Square wave anodic stripping voltammetry (SWASV) was the technique used due to its sensitivity and low detection limits that can be achieved. The electrochemical as well as the microfluidic parameters of the developed system were optimized. The detection of CdS QDs in the range between 50 to 8000 ng/mL with a sensitivity of 0,0009 µA/(ng/mL) has been achieved.

Other significant results are related to the electrochemical detection of cadmium-selenide/zinc-sulfide (CdSe@ZnS) quantum dots (QDs) used as labeling carriers in an assay for apolipoprotein E (ApoE) detection. The immunocomplex was performed by using tosylactivated magnetic beads (2.8µm of diameter) as preconcentration platform into the same device. All the conjugation steps were performed in flow mode. ApoE was evaluated for its potential as biomarker for Alzheimer’s disease detection. The limit of detection (LOD) achieved was ~12.5ng mL-1 with a linear range between 0 to 200ng mL-1. Finally, dilutions from human plasma were assayed with high accuracy with respect to the calibration curve.
Task 3.1 - Electrochemical detection
At the start of the project, the technology concept for electrochemical detection in this task was already gathered in a semi-automatic prototype. During the Nadine project, this prototype was further developed to a more sensitive and robust approach and amenable to clinical routine. This objective has been reached through the development and production of two new electrochemical chips, a microwell (ImmuDrop-Chip) and a microchannel based sensor (ImmuChip), with dedicated manual (the ImmuDrop) and automatic (ImmuSpeed) operated systems, as well as specific detection methods and software (see deliverable 3.1 for details).
The production of industrial chips has been outsourced to local printed circuit board companies having the infrastructure and know-how required for the manufacture of these particular devices. The quality of the chips was verified and was in good agreement with the manufacturers specifications. Industrial prototypes of both ImmuSpeed and ImmuDrop systems have been designed and successfully fabricated. In addition they have been conceived so as to render them readily amenable to clinical practice, with control and traceability features (bar codes, hardware key...). The two approaches showed to be applicable to immunoassays. Regarding assay protocols, the ImmuDrop system requires larger amount of reagents and the detection takes longer before stabilization of the signal compared to ImmuSpeed system. In addition, the ImmuDrop-chips cannot be regenerated which is the case for the ImmuChips. The ImmuSpeed system has been therefore used for the assay development and the dosage of Aolipoprotein E and Amyloid beta 1-42 AD biomarkers (see WP 6&7).
Task 3.2 - Integrated optical waveguides for various detection schemes
The objective for task 3.2 was “to develop waveguides, integrated with separation chips for sensitive on-line detection”. Because of the many promising properties of thiol-ene, including the initial assumption that the refractive index would change proportional to the sulfur concentration, it was decided that the device would be based on a thiol-ene chip. During the early work in the NADINE project it had already been published that thiolene chips support an electroosmotic flow (EOF). One of the advantages of using thiol-ene is that during the chip fabrication the same material can be used as both substrate, channel/waveguide layer, and lid. The design of the channel/waveguide layer is shown in Fig. 7. The chip features a standard injection cross in one end and a set of 200 µm wide waveguides just before the outlet in the other end. The channel is 30 µm wide and measures 48 mm from injection cross to detection region.

Figure 7: The design of the chip features a curved input waveguide (top left) to prevent unguided light from reaching the detection region. The white circles serve as support beams during the chip moulding process. The small grey circles are the in/outlets to be used during the moulding of the chip. The T-shaped channels serve to guide excess thiol-ene and air bubbles away from the more important parts of the chip.

The standard method for making thiolene chips is to pour the liquid thiolene mixture into a PDMS mould and close it with a lid (i.e. a glass slide). The mould is then exposed to UV light to cure the thiolene (see Fig. 8 - moulding). This method does, however, often leave a residual layer behind after curing. The residual layer is rarely a problem when the chip only contains features such as channels. However, when making waveguides, which are ridges and not recesses like channels, a residual will form. The residual layer adjacent to the waveguide will allow light to escape and thereby lower the quality of the waveguide. So, to avoid creating a residual layer, a new moulding method was developed. In the new method, the PDMS mould is completely assembled prior to adding the thiolene mixture. Using a syringe with a needle the thiolene is then injected into the mould through venting holes, the same holes allow the trapped air to escape (see Fig. 8 - Injecting). When using the injection method no residual layers were created, but the method does leave another unintended feature. At each venting hole there is a chance that excess thiolene will form little pillars. This is potentially critical, but if the mould is designed carefully the venting holes can be positioned in areas where the left over pillars will not interfere with the chip’s performance.

Figure 8: The two different methods developed to fabricate thiolene chips. The moulding method (top) is good when making channels, but leaves a residual layer. The injection method (bottom) is preferred when making waveguides. In cases where waveguides or other high precision features are necessary, the bottom part of the PMMA holder can be replaced by a cleanroom fabricated master.

In order to obtain the highest possible quality of the waveguides the chip was originally designed to be made by the injection method. However, due to the issue with non-changing refractive indices the loss in the waveguides would become too large if the bottom substrate was not changed to a different material (glass). Also, the lid of the chip could not be the same material as the waveguides, but this could easily be fixed by simply cutting holes in the lid above the waveguides, thereby changing the cladding material to air. If the injection method was kept while the bottom substrate was changed to glass, the waveguides would be perfect but the bottom of the channel would now be a completely different material from the remaining three walls. So, by switching to the regular moulding method, a residual layer will cover the entire channel floor, thus restoring the situation. However, now the quality degrading residual layer will also be present next to the waveguides. By relating the 120 µm height of the waveguide to the assumed, less than 10 µm thick, residual layer this was considered to be an acceptable configuration.
The final step in the chip fabrication process is to attach little glass reservoirs with UV-glue to each of the four in/outlets of the chip.
Task 3.3 - Evanescent wave optical detection
Due to a combination of the resulting (low) quality waveguides from T3.1 and the particular chip design, the optical signal from the analytes could not be collected and detected by waveguides as originally planned. However, the result presented below (see Fig. 9) does use the thiol-ene waveguide to excite the fluorophores, while the collection of emitted light was done through a microscope.

Waveguide assisted fluorescence detection of CE-analytes was achieved on a thiol-ene device with the exception that emitted light was collected through a microscope rather than the output waveguide. Excitation through the thiol-ene waveguide was working satisfactorily despite the loss caused be the residual layer. The results shown in Figure 9 demonstrate that the chip was capable of single analyte injections, inverse injection, and separation of multiple analytes.

Figure 9: Examples of detection of CE-analytes. All experiment were conducted at approximately 550 V/cm.

In order to investigate the band broadening in the chips, a series of single analyte injections was performed at field strengths varying from 200 V/cm to 720 V/cm using an injection time of 0.5 seconds. The retention times and the full widths at half height (FWHM) of each peak were determined and used to calculate the “plate height” as a figure of merit to evaluate the quality of the separation system. All data points fall around a linearly increasing trend line between 3 µm and 14 µm. This tells us that in the parameter space investigated the longitudinal diffusion is not significant and does not become large enough to influence the plate height and thus the performance. Since the chip design in this case features rather big channels (30 x 120 µm2) it is reasonable to assume that the fairly steep slope is mainly due to significant Joule heating.

A re-designed chip clearly demonstrated the feasibility of evanescent-based detection, but also pinted to a number of inherent issue with this mode of detection. Results of these experiments were published, but the approach was deemed not sensitive enough for the particular challenges in this project.

A modified and further developed version of this chip has later also been used by partner UPS in WP 4.
Task 3.4 - Nanotarget mass spectrometry
In this task, we have designed and fabricated chip-based pillar-structured targets for MALDI mass spectrometry. The pillars were made in various sizes, ranging from 500 down to 5 µm in diameter. The particular features of the new targets are a strong confinement of the sample volume and a concentration of the analytes onto a very small surface area.
To be able to load the samples onto the pillars, we designed and constructed a robotic system, which made it possible to handle volumes down to 1 picoliter. The new concept was evaluated with a set of model peptides, and a sensitivity of 10 zeptomol was routinely obtained. This is at least a thousand-fold better than results obtained with conventional target structures.

In a subsequent stage, we tested the new concept for the analysis of Aβ-peptides in cerebrospinal (CSF) fluid. After performing an immunoprecipitation (IP) and concentration of the analytes, clear MS signals for a set of Aβ-peptides (including Aβ 1-42) were obtained, even when only 0.1% of the final sample volume was loaded on a micropillar. Using the same IP protocol, we were also able to detect the Aβ-peptides in blood plasma. Additionally, we successfully performed an Aβ- peptide analysis using CSF from mice, where the amount of CSF was only 10 – 15 µL.

The IP protocol was further optimized, and Aβ-peptide analyses, using a set of 40 clinical CSF samples (20 Alzheimer disease (AD) patients and 20 non- AD patients) was performed. From the mass spectra obtained, we calculated the ratio between the integrated peak areas for Aβ-1-40 and Aβ-1-42. It was shown that there was a significant difference for this ratio between the AD patients and the non-AD patients. It is therefore not unlikely, that this method could be utilized as a clinical tool for detecting AD.

Compared to traditional methods, based on single component immunoassays, mass spectrometry has several advantages. Apart from the fact that data for basically all Aβ-peptides are obtained, possible post translational modifications and/or truncated variants can also be detected. Specific antibodies for performing an immunoassay of such components are normally not available. It is believed, that these components can contribute to a better understanding of the cause for onset of AD.

It should be pointed out that it would be feasible to perform the IP process as well as the deposition of the analyte concentrate on the MALDI targets in parallel mode, by using a robotic work station. In comparison with the time-consuming traditional ELISA methods, the actual MALDI-MS analysis only takes a few seconds, and such a setup would be both time-saving and cost-effective, when dealing with a large number of samples.

Task 3.5 - Signal amplifications with nanoparticles
Gold nanoparticles (AuNPs) have been successfully used as signal enhancers in both electrochemical and optical biosensing systems. They have been proposed as blocking agents in nanochannels-based electrochemical approaches for the sensitive (at ng/mL levels) detection of protein biomarkers (CA15-3 breast cancer biomarker and thrombin) in whole human blood (Figure 1, left). On the other hand, AuNPs properties have been approached for their use as carrier of enzymes for the amplified optical detection (at ng/mL levels) of model proteins (human IgG) in lateral-flow immunoassays (Figure 10, right).

Figure 10. Summary of the applications of AuNPs for signal amplification in nanochannels (left) and lateral-flow (right) based biosensing systems.

In parallel, cadmium-selenide/zinc-sulfide (CdSe@ZnS) quantum dots (QDs) have proved to be highly effective reporters in fluorescent microarrays. QDs are five times more sensitive than the Alexa dye in microarray format and seven times more than the corresponding ELISA for ApoE biomarker screening (at pg/mL levels). Real (clinical) samples (serum and plasma) have also been assayed (in collaboration with UULM) with acceptable accuracy and precision (Figure 11).

Figure 11. Summary of the use of QDs for signal amplification in microarray format.

Task 3.6 - Chemically compartmentalized detection
The objective of Task 3.6 was to seek for ultimate sensitivity in protein microarray on chips. This was intended as a contingency strategy, if the sensitivity obtained in the other tasks of WP3 was found not sufficient (at the price of a reasonable increase in complexity, development and fabrication cost).
Different strategies were first considered, but at we finally focused on droplet based microfluidic technology. This new platform aimed at bringing to the clinical world the advantages of droplet microfluidics relevant to the specific challenges of diagnosis, i.e. reliability, robustness to contamination, high multiplexing potential, user-friendliness and cost reduction, while maximizing the compatibility and interoperability with currently validated and used protocols and workflows. The front-end of this platform is a pipetting robot, which allows the direct sampling of samples and reagents from standard microtiter plates (MTP), for full compatibility with current sample storage strategies and equipment. As a main difference as compared to conventional droplet approaches, all droplets are kept equally spaced by oil in a highly confined state (a format also called “plug microfluidics” in literature). This way, inter-droplet contamination risk is reduced by orders of magnitude, and the “identity” (originating well) of each droplet, built in the sequence of droplets all along the protocol, is known without need for internal tags. These differences alleviate several current limitations of droplet microfluidics. The second original feature of the platform is the “magnetic tweezer” concept (Fig. 12). This allows the transfer of functionalized magnetic particles between subsequent droplets, bringing into the droplet microfluidics world the power of solid-state extraction and the flexibility of magnetic beads-based protocols. In particular, we demonstrated the potential of this technology for protein detection (ELISA for TSH analysis) and in the frame of WP8 its ability to be used for nucleic acid analysis (HER2 overexpression).

Figure 12: Magnetic tweezers technology a) Picture of the magnetic tweezers with the capillary (highlighted with a red liquid). A second passive magnetic tip, placed opposite to the first with regards to the capillary contributes to field lines shaping, and to the optimization of the magnetic force. b-h) Sequence of images showing the extraction and the redispersion of magnetic beads from one droplet to another one (colored in orange), switching ON and OFF the electric current in the coil.
Task 3.7 - Digital ELISA towards Abeta and neurofilament biomarkers
A new digital platform for immunoassays like ELISA was developed and applied for Abeta detection. The platform is based on single molecule counting where a sample is automatically divided into millions of femtoliter droplets ordered in an array on a surface. If target molecules are present in a sample, they will be trapped in the droplet and elicit a signal. Quantification is obtained by simply counting the positive droplets. The technology was tested by detecting pure ApoE3 and the results showed that the detection limit was in the upper aM range. This was several orders of magnitude more sensitive than the corresponding ELISA, and enables blood based diagnostics when biomarkers exist.
The platform was used to detect Abeta 1-42. The detection limit was in the fM range and thereby poorer than the corresponding ApoE3 assay. The explanation was probably a relatively poor antibody pair to detect Abeta1-42. However, fM detection limit was sufficient to analyze CFS samples and more detailed results will be reported in WP 7. It was not possible to find suitable antibodies for neurofilaments.
In addition, T3.7 involved prototyping an instrument. The instrument was designed and built and can process semi automatically 16 samples in parallel and detect the positive droplets. The instrument demonstrates that the technology can be packaged into a smaller instrument, less complex than separate large peristaltic pumps and extensive microscopes.
Task 4.1 - Sample pre-treatment
The initial deliverable was the development of an automated system able to get plasma from total blood before analysis. The specifications of the module have been defined and a first design based on tangential filtration was proposed, see Figure 13.

During the first period (M1-M18), the need for the deliverable has changed. Partners have stated that it was more relevant to deal with serum sample rather than raw blood. Indeed, it is easier to manipulate serum than blood: no need for specific agreement, large banks available. Consequently, the pre-treatment module was not necessary anymore.
After the decision of analyzing serum sample, work was done towards integration and automation of the concentration module developed by partner Institut Curie (task 4.2).

Figure 14: Schematic of the microfluidic fluidized bed set-up (left) image of particle in the packed bed regime (no flow) (top) and in the fluidized regime (bottom) accompanied with PIV profiles

Figure 13: Principle and software of the flow-rate control module.

Task 4.2 - Dynamic magnetic immunocapturing
Task 4-2 is aimed at capturing low-abundance peptides from the plasma using micro- or nanoparticles (issued from Task 1.1) and releasing them as a concentrated plug, for downstream analysis by CE on chip (Task 4.3). This preconcentration is mandatory for reaching pM sensitivity, even in combination with online labelling (Task 4.4).
In this task, we developed a completely new microfluidic technology relying on the concept of fluidized bed. Fluidized beds are an appealing strategy to enhance the kinetics of liquid-solid processes in microfluidics, but until recently their implementation was hindered by the weakness of gravitational forces at the microscale. In the context of the NaDiNe project, we introduced magnetically-driven fluidized beds. The design of the microfluidic device was optimized in order to compensate for the decaying intensity of the magnetic field (Fig. 14). A particle image velocimetry (PIV) analysis has shown that beads undergo a continuous recirculation. Indeed, a higher velocity at the centre of the bed is observed, followed by a perpendicular circulation to the borders of the chamber when arriving to the end of the bed, and a backflow near the borders. Playing with the flow rate applied using partner FG’s fluidic control equipment, we have been able to control finely the bed porosity. The microfluidic fluidized bed presents specific features regarding biomarkers extraction and preconcentration: a low backflow pressure allowing to sample a large volume, and a continuous recirculation while maintaining a high beads density.
This new technology was successfully applied to Aβ extraction. We demonstrated that a high preconcentration factor could be achieved with commercial peptides ranging from 700 to 1000 with high selectivity. Besides the potential of this technology for other applications has also been considered especially regarding bacteria extraction and detection.
Task 4.3 - On-line fluorescent labelling
An on line-derivatization method to analyse ubiquitin in CSF has been developed. The method allows ubiquitin to be fluorescently labeled (without the need of antibodies), during the capillary electrophoresis separation process. The linearity of the CE assay was demonstrated in the range of physiological concentrations expected in CSF, as well as the reproducibility of the method. The method has a limit of detection around 15 nM, which is compatible with its detection in CSF with a preconcentration step, which is performed to desalt and preconcentrate three times the sample. The electrophoretically mediated microanalysis method could be applied to the detection and quantification of ubiquitin in CSF samples (Fig. 15).
Figure 15: Quantification of UBI in desalted CSF using electrophoretically mediated microanalysis. Samples: Derivatization blank (a), CSF (b), CSF spiked with 16 nM of UBI (c), standard solution of UBI at 60 nM (d), at 120 nM (e). Before injection, samples are diluted in DMSO (2/1 v/v).
An innovative isotachophoresis preconcentration method compatible with alkaline conditions has been developed for the first time. This strategy allows to achieve the detection of Aβ 1-40 at a concentration of 30nM without derivatization, which is an unprecedented sensitivity for UV detection.
We developed a new methodology called “multiple ITP”, to preconcentrate abeta peptides from CSF. Starting from standard solution, we can reach the limit of quantification (LOQ) of 50 nM with normal UV detection (at 200 nm wavelength), which is much improved compared to the LOQ obtained with CE-UV without electrokinetic preconcentration (2000 nM).
Figure 16: Electropherogram corresponding to electrokinetic separation of a mixture of Aβ 1-38, Aβ 1-40, Aβ 1-42, Aβ 2-42, on EpDMA coated glass microchip. The peptides concentration is around 40 µM in the mixtures. LOD was estimated to 1-10µM.

We also demonstrated that immunocapture on magnetic beads together with derivatization of captured peptides may be performed in one step and lead to both efficient capture and high sensitivity of detection. We have successfully detected A-beta peptides, at 1-38 (1-2 nM), 1-40 (6-8 nM) and 1-42 in CSF (0.5 nM) after this pre-treatment using capillary electrophoresis coupled to LIF detection for the off-line analysis of the eluted fractions. We obtained an enrichment factor of 100.
Task 4.4 - High resolution electrophoretic separation on chip
A highly resolving electrophoretic method for the separation of different Aβ-amyloid peptides on coated glass microchips has been developed by combining polymeric surface treatment (input from partner ICRM) and their (off-line) derivatization by Fluoprobes 488. The method led to good resolution between close migrating species with very satisfactory RSD for EOF, migration times and peak areas. This high resolutive electrophoretic method has been applied to the separation of a mixture of Aβ 1-38, Aβ 1-40, Aβ 1-42, Aβ 2-42 on EpDMA coated glass microchip (Figure 16).
Alternatively, thiolene chips have been designed and microfabricated by partner DTU. We performed a deep investigation on the potential of thiolene microchips (with partner DTU) for the electrokinetic separation of proteins, measuring the EOF, searching for a new coating (with partners ICRM and DTU). We have now found a stable coating over a wide pH range, an efficient coating procedure, demonstrated the efficient hydrophilisation of the channel surface as well as the high ability of this coating to suppress protein adsorption. We have applied this system successfully to the analysis of acidic and alkaline proteins (see Fig. 17).

Figure 17: Electropherograms obtained from the separation of two acidic fluorescently labeled proteins (left) and of Histone-Alexa 488, an alkaline fluorescently labeled protein (right) on thiolene chip (20 % excess of thiols, coated with the terpolymer DMA-PMA-MAPS). The insert in the figures displays electropherograms from five repetitive analyses
Task 4.5 - Spotted proteins and peptide microarrays
Antibody microrarrays for high-sensitivity immunoassays were developed. The microrarray platform consists in polymer coated silicon slides. Due to constructive interference, Si/SiO2 layered slides allow enhancement of the fluorescence intensity on the surface with significant improvements in sensitivity of detection.
Silicon oxide surfaces coated by copoly(DMA-NAS-MAPS) have been extensively characterized by atomic force microscopy (AFM) and dual polarization interferometry (DPI) showing a uniform and smooth polymeric layer with a thickness of 1.89 nm in dry conditions swelling up to 15.7 nm in buffer. Model experiments on the detection of 5 inflammation markers (Interleukin 6 and 10, PTX3, CRP and TNF-alpha) were used to optimize spotting and incubation protocols and demonstrated the feasibility of fluorescence microarrays on copoly(DMA-NAS-MAPS) coated silicon oxide for the detection of a panel of biomarkers with femtomolar sensitivity.
Highly sensitive fluorescence immunoassays for the detection of neurodegenerative diseases biomarkers amyloid-beta 1-42 and 1-40 (Aβ42), (Aβ40), and apolipoprotein E (ApoE) were realized based on a label/label-free microarray platform that utilises silicon/silicon oxide (Si/SiO2) substrates. Limits of detection in the order of pg/mL were achieved.
Integration of the developed microarrays with microfluidics (WP6) allowed a reduction in the assay time and semi-automation of the assay. A validation of the proposed method using human CSF samples was performed within WP7.
Task 4.6 - Microfluidic convective self-assembly for microarrays
This task was planned by partner CI as an alternative approach, in which the different biocapture elements should be prepared in a high throughput fashion in a batch process by stop-flow lithography, and then self-assembled microfluidically in the microfluidic chip.
Figure 18: Successive capillary assembly steps in selective traps from schematics to experiments

The objective we have defined is to develop an alternative approach to overcome the limitations of conventional antibody spotting approaches used to elaborate protein microarray. We have mainly focused our work on the self-assembly by fluidic means of commercial particles or photo-polymerized hydrogel particles for the creation of high-density protein arrays directly integrated in the detection chamber of microfluidic devices. This work is based on a technology pioneered by L. Malaquin that allows a high multiplexing capability of beads self-assembly with high accuracy (Fig. 18). Two approaches have been successfully investigated and applied in the NaDiNe project namely convective and capillary self-assembly that allows either to create 2D dense layers of particles or to induce a deterministic positioning of beads of different nature. We have demonstrated the ability to assemble protein labelled beads with various sizes (ranging from 500 nm to 5 µm) while maintaining the ability of antibodies grafted on the beads surface to bind their targeting antigens. This was demonstrated by capturing cells by antibody-antigen interaction.
Finally, by optimizing the device microstructures as well as the wetting properties of the substrate, we have successfully demonstrated the multiplexing capabilities of the methods: dense arrays composed of four different types of functionalized beads were precisely assembled on PDMS substrate with a placement accuracy <2µm and an assembly yield >95%.
Task 4.7 - Hyphenation with mass spectrometry
We have developed a system for preparative fraction collection from a CE capillary for subsequent matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF-MS). Initially, a direct coupling of CE to a mass spectrometer via an electrospray interface (ESI) was tested, but there is too much incompatibility between the crucial criteria for obtaining a good ESI and the optimal operating parameters for CE of our target molecules (Tau-fragments). An additional advantage of MALDI-MS is that it offers a much higher sensitivity than ESI-MS. A schematic of the developed preparative system is shown in the figure 19.

Figure 19: Schematic of the preparative systems for sampling from a CE capillary to a MALDI target
It includes an in-line UV detector, to monitor the content of the effluent flow from the CE-capillary, and a XYZ robotics system, which deposits the effluent flow onto an array of MALDI targets (such as silicon pillars).
A successful MALDI analysis of CNBr fragments of the important Alzheimer’s disease biomarker Tau-441 was accomplished. All of the expected (6) fragments were detected. With the new preparative separation system, 5 of the 6 expected fragments could be detected.
Task 5.1 - Biomarker panel elaboration
Potentially interesting biomarker candidates have been evaluated. In addition to the biomarker candidates S-100B, GFAP, ERK1/2, progranulin, ubiquitin, which had been selected already within the initial phases of the project, partners UULM and UKES have started efforts towards the study of charge-isoforms of Serpin and DJ-1 in CSF samples by novel capillary isoelectric focusing immunoassays. These are currently being established as novel applications of the NanoProTM technology (ProteinSimple). Additionally, the application of this technology for the detection of oligoclonal IgG bands in CSF has been successfully tested.
To further address the potential prognostic and diagnostic value of measuring total ERK1/2 in CSF in early stages of Alzheimer’s disease and mild cognitive impairment, partners UKES and UEF have selected a suitable clinical cohort. Pilot experiments have been done successfully so that the clinical study will be launched soon. Partner MB has produced polyclonal and monoclonal antibodies against different phosphorylated and unphosphorylated forms of ERK1/2, which were evaluated by partner UKES by Western blot and capillary isoelectric focusing immunoassay. While reasonably good antibodies were identified against diphosphorylated ERK1/2 (serum 78), the generation of antibodies against monophosphorylated ERK epitopes was apparently not successful. Partner UKES established a capillary isoelectric focusing immunoassay for the measurement of amino-terminal variants of Aβ peptides.
Task 5.2 - Antibody production
We have developed and characterized a number of polyclonal and monoclonal antibody reagents recognizing Aβ-peptides (DAEFRHDSGYE; AEFRHDSGYEVHH; EVHHQKLVFFAED; KLVFFAEDVGSNK) generated from Aβ 1-42 peptide. 19 hybridoma clones (AEF1.1-AEF19.1) originating from the extreme N-terminal part of Aβ 1-42 peptide recognize both Aβ 1-42 and Aβ 2-42 isoforms in different approaches including ELISA. We also finalised the development of five new monoclonal antibodies to the central part of Aβ 1-42 peptide. Three of these monoclonal antibodies recognize the sequence KLVFFAEDVGS within the full length Aβ 1-42 peptide. These antibodies are therefore useful for combination with N-terminal Aβ 1-42 peptide specific antibodies AEF to improve sensitivity and specificity in different applications, including two-site ELISA. The requirement to develop antibodies that will specifically recognize only Aβ 2-42 was achieved using affinity purified polyclonal serum 77 that binds strongly to Aβ 2-42 peptide with very low cross-reactivity with Aβ 1-42 peptide. To remove this low level cross-reaction and improve sensitivity and specificity, we developed a new method for affinity purification of rabbit sera to isolate only antibodies recognizing Aβ 2-42 peptide and not Aβ 1-42 peptide (one amino acid difference). We also developed and affinity purified rabbit serum 76 that specifically recognizes Aβ 1-42 peptide and does not recognize Aβ 2-42. Additionally, we developed a new methodology to prepare Fab fragments of IgG from monoclonal antibodies AEF-4.1 AEF14.1 and AEF-16.1. These fragments were also conjugated with biotin in order to immobilize them on paramagnetic microparticles.
Figure 20: Development of ELISA test for detection of ApoE in human sera. (A) Sandwich ELISA was coated using different mouse monoclonal antibodies and detected by HRP conjugated mouse monoclonal antibody Apo1.1. (B) Sandwich ELISA using combination of mouse monoclonal antibodies and HRP conjugated rabbit polyclonal antibody.
We also concentrated on antibody development for the human ApoE3 protein. We have developed five high quality monoclonal antibodies (Apo1.1 - Apo5.1) and one rabbit polyclonal sera number 33 that recognize both the native and denatured forms of ApoE3, either in purified form or in the endogeneous form in serum. We developed the two site ELISA method using our new set of antibodies, which is able to recognize ApoE3 protein in human serum, supporting their use for NaDiNe partners for development of original two-site ELISA methodology for detection of ApoE3 in serum or other body fluids (Figure 20).
Within the project we additionally developed ubiquitin specific antibodies (UBI3.1 UBI4.1 UBI5.1) recognizing mono-ubiquitin and one affinity purified rabbit polyclonal serum (number 37) specific to mono and poly-ubiquitin and we also developed and characterised new monoclonal antibodies to poly-ubiquitin (peptide CTLEVEPSDTIENVK), three of which also recognize full length poly-ubiquitin.
In summary, we have developed several monoclonal antibodies recognizing different regions of Aβ peptide, ApoE3 and ubiquitin, and sets of polyclonal antibodies recognizing newly discovered isoforms of Aβ peptide, including a method of affinity purification of these isoforms. We have also developed a method for preparation of Fab fragments of these antibodies and their immobilization on solid surfaces such as magnetic microparticles. These Fab fragments will be useful for downstream applications involving either diagnostic kits or analytical techniques based on MALDI-TOF.
Full details of antibody production and the testing of these reagents using a variety of immunochemical methods are available on the Moravian Biotechnology website (http://www.moravian-biotech.com/sql). A presentation of the results is also available on www.moravian-biotech.com.
Task 5.3 - Optimization with model samples and clinical samples
The objective of this task has been to prepare a “catalogue” of potential technologies or combination of technologies for prototyping. The technologies will be optimised and compared by tests with model samples and clinical samples.
During the first 30 months, the NaDiNe partners have been developing technologies based on different approaches (electrokinetic separation, electrochemical detection, microarray, fluidized bed, mass spec...). To achieve a sufficient maturity of the technologies developed in the framework of the NaDiNe project, the decision on the candidate technologies has been slightly delayed. A specific meeting focused on this topic has been held in Paris in January 2013. The objective of this meeting was to select the different technologies for further prototyping. A methodology combined to a set of criteria has been established to compare these different approaches in the most objective way possible. The technologies comparison has been performed based on a set of two biomarkers. The analysis has been, in a first approach, restricted to ApoE and Aβ peptides, in either CSF or serum/plasma. If a technology can provide distinction of isoforms of A-beta, then ratios of certain isoforms should be considered for improved sub-typing. As expected, serum and CSF samples have been provided by partners UULM, UKES and UEF. Based on the results obtained during this comparative study, different technologies have proven their potential mainly for CSF sample analysis and have been identified as candidate technologies for prototyping. Besides the technologies developed by the NaDiNe partners, partner UULM has been working on a commercial apparatus (NanoPro system) that will be considered as a commercial benchmark. During the Paris meeting, the different technologies have been compared and different approaches have been selected ( Figure 21)
Figure 21: Potential workflows and suitable technologies.
Task 5.4 - Data analysis strategies and bioinformatics
To prepare a bioinformatics strategy to build a multi-marker diagnostics platform, we performed a systematic study and benchmark of several statistical and machine learning methods for data-driven feature selection, the goal being to extract a robust and accurate prognostic signature from a limited number of samples. We identified the best strategies and reported the results in the following publication:
A.-C. Haury, P. Gestraud and J.-P. Vert, "The influence of feature selection methods on accuracy, stability and interpretability of molecular signaturess", PLoS ONE, 6(12):e28210, 2011.To help in the selection of biomarkers, we re-analysed the seminal work of Ray et al. (Nature Medicine, 2(9):e4141, 2007) who identified a multi-markers signature of 18 proteins for AD diagnosis, and the later re-analysis of Ravetti et al. (PLoS One, 3(9):e3111, 2008) who shortened the signature to 5 proteins. We found flaws in the second analysis, the results of which should therefore not be trusted for our own selection of markers.
Task 6.1 - Instrument specification and design
A first step here was to determine which techniques would be selected to be integrated in the instrument prototype (based on input from T5.3) in order to draw the instrument specifications and design.
The planned functional diagram for the instrument is shown on Fig. 22.

Figure 22: First version of the instrument functional diagram

The work has then mainly been focused on the automation of the fluidized bed and the “fluidic” microarray approach, and on the achievement of integrated platforms.
Task 6.2 - Prototype instrument development
The work on the development of the prototype instrument was divided into 2 parts: the electronic integration and the fluidic integration. This development was motivated by different aspects: ease of use, volume, robustness, compactness and integration.
In line with the needs of the Nadine project, a compact module has been developed allowing the fluid handling of the fluidized bed and the fluidic microarray. In Figure 23, we see the electronic board integrating all the different parts needed. Further, in Figure 24, we see the fluidic platform developed for the fluidized bed and the microarray. The fluidic platform integrates flow sensor, valves, and reservoirs.

Figure 23: Electronic platform.

Figure 24: Fluidic platform for the sample preparation (left) and for the fluidized bed (right)
Task 6.3 - Software for instrument control
At M36, prototypes of the different selected modules were developed to progress towards the prototype instrument. The software developed in this task enables the control of all the flow-control tools developed in other WPs and a lot of effort was put into the programming of a software architecture based on a modular and powerful approach. This version is adaptive depending on the connected devices: the user can choose which elements he wants to display (Figure 25). This approach makes all the more sense to match precisely the different needs of the integrated modules.
Additionally, a scripting tool to automate the two selected technologies and a SDK (Software Development Kit) to allow the integration in global software has been developed.
Task 6.4 - Software for analysis of results
A prototype of software to learn a predictive multi-marker signature from the device measurements has been implemented in the R language. This prototype, based on preliminary development performed in Task 5.4 systematically compares different methods for feature selection (univariate tests, lasso regression, random forests) and predictive modelling (logistic and linear regression, support vector machines, random forests) and selects the best one, as evaluated by cross-validation, to form the final signature. In addition to automatically learning a predictive signature, the software prototype comes with various visualization options to control the distribution of measurements over a cohort of samples. It also implements functions to quickly visualize the performance of different methods to train a predictive signature.
However, due to unavailability of a fully functional demonstrator tested on a collection of sample, this prototype software has not been implemented yet to directly process the outputs of the demonstrators, which will require additional functionalities to process and normalize the measurements before running the prototype to learn a signature.
Task 6.5 - Optimisation of prototype with model samples
The task involves some optimizations on the prototypes developed earlier in the project. From the realisation of the first prototype until the end of the project some optimisations and improvements have been achieved for the fluidized bed: selection of a suitable immunosorbent, Ab grafting method, selection of microchip material (COC), surface treatment, flow-rate, reduction of the manual steps. Similarly, the microfluidic microarray were optimized on a number of parameters: microfluidics layout, the chip clamping method, the flow-rate platform, the surface treatment, the software. The results show that the optimisations achieved on the prototypes allow a good efficiency compared with the static method, are time saving, and are amenable towards automation.

Figure 25: Software panels
For the parallel approach on the equipment from partner DS, different assay protocols were successfully developed for ApoE and Aβ1-42 for the different matrices. The Apo E assay takes less than 1.5 h and requires low volumes (<1μl) of samples since CSF must be diluted by a factor of 300, and plasma and serum by a factor of 3’000 for the assay. The Aβ1-42 assay takes about 2h and requires 10μl of CSF maximum to be tested. It was shown to be difficult to measure Aβ1-42 in plasma with classical dilution approach. A specific standard-addition method has therefore been developed but required larger amount (~100μl) of samples compare to the classical dilution method. Both assays showed to give comparable results as compared with commercial ELISA but with shorter time to result (2h instead of 4h in general) and with lower reagents and sample consumption (~1/10th).
Task 6.6 - Risk assessment and regulatory issues
Any medical device put into the market needs CE marking and approval, where a full risk assessment will have to follow EN14971, which covers risks through the entire life-cycle of the device. A full risk assessment according to EN14791 has not possible as even the integration of the various steps into an integrated device is still work to come. Hence, an analysis has been carried out at the final stage of the prototypes. The study concerned risk assessment in relation to human safety when handling the LAB prototype. Moreover, the study has provided inputs to a potential future integrated instrument. The analysis has covered all four envisioned prototypes: “Fluidized Bed”, “Automated microfluidic Elisa with electrochemical detection”, “Micro-array for fluorescence detection module” and “Lab prototype microarray”. For each prototype, all steps in the process as relevant for the LAB prototype have been described, and the specific process has been briefly described for each step, and the potential hazards has been identified and listed. Hazard causes have been included where relevant, and barriers reducing the frequency and/or the consequences of the hazard have been listed. The assessment concluded that for all the four prototypes hazards and relevant barriers have been identified, and at the present stage all of the prototypes have an acceptable safety level. Furthermore, there were no indications that the prototypes cannot be taken to commercials products that could not be approved (in accordance with EN14971).
Task 7.1 - Clinical characterization of patients and biobanking
The three clinical partners UEF, UULM and UKES have continued to assemble the patient collective as defined in the NADINE WP7 goals and milestones. Altogether the three clinical partners have collected a sufficient number of CSF and EDTA-plasma samples for each diagnostic subgroup and a control cohort. The total number of available samples is > 1500. CSF and plasma samples can be provided to all partners on request to support the technology development within other NaDiNe work packages.
Task 7.2 - Pre-clinical testing on model samples
The objective of Task 7.2 was to perform a first validation of the project’s prototypes, on samples containing well defined concentrations of analytes, and thus to provide an assessment on a technical performance, rather than clinical, basis. The intention was to provide quantitative references for clinical testing.
This task has been performed with two prototypes technologies:
- The automated version of the microfluidic fluidized bed has been applied to immunoextraction with commercial Ab peptides. The peptides have been spiked at different concentrations in PBS and high preconcentration factors have been achieved (around 1000 depending on eluting buffers).
- The prototype of integrated microarray has been applied to model samples of ApoE and Ab peptides analyses. The efficiency of the prototype has been proven with ApoE (for concentrations between 10 ng/mL and 200 ng/mL) and with Ab-42. The results show that the optimisations achieved on the prototype allows a good efficiency compared with the static method, are time saving, and support automation approaches.
Task 7.3 - Clinical testing on samples from humans (Task 7.4 and 7.5)
• The three clinical partners UEF, UULM, UKES have collected a large number of CSF and blood plasma samples from well characterized patients with neurodegenerative diseases and suitable control groups with other diseases (disease controls). In total, more than 1500 CSF and blood plasma samples are potentially available for method development, assay validation and biomarker research within or related to the NaDiNe project.
• A reference method (gold standard) for measuring Aβ peptides in CSF has been selected and validated.
• Fluorescence microarray immunoassay (partner ICRM), microfluidic ELISA (partner DS) and droplet array (partner DTU) were demonstrated by the respective partners to be able to detect Aβ42 in human CSF and to differentiate preselected diagnostic groups.
• Anti-Aβ immunoprecipitation followed by mass spectrometry was successfully applied by partner KTH to determine Aβ42/40 ratios in human CSF samples and to differentiate preselected diagnostic groups.

Microfluidic ELISA (DS): The analytical performances of both ApoE and Aβ1-42 assays was extensively studied and showed high performances with very low limit of detection in the picomolar range (respectively 0.4 and 3 pM) and reproducibility below 15%. In CSF both assays showed to give linear results with recovery percentages within the 80-120% acceptance range. Aβ1-42 was further investigated in a Case/Control and in a blinded study with samples with low and high levels of Aβ1-42. The results were highly correlated with commercial ELISA and a clear separation of clinical CSF samples with either high or low levels of Aβ42 was achieved (statistical significance). We conclude that, as a next step, our electrochemical micro ELISA should be applied to a larger clinical sample under routine laboratory conditions.
In conclusion, we have demonstrated that our approach was able to measure with a high reproducibility and repeatability the targeted biomarker in complex human samples. In addition the assay was applied to the dosage of two sample cohorts and demonstrated statistical power to distinguish between non-AD and AD samples. Finally the ImmuSpeed ™ prototype is directly amenable to clinical routine. This is in line with the objective of DiagnoSwiss’ contribution in the NaDiNe project.
Microarray based analysis (ICRM): The microarray test was performed using standard Amyloid Beta 1-42 peptide freshly dissolved and diluted in ACSF as described in (Gagni et al. 2013). Calibration curves showed a limit of detection (LOD) of 200pg/mL. However, analysing clinical samples required additional optimization and control of CSF dilution (4-fold dilution was optimal) and storing and shipping conditions of samples. Due to unforeseen difficulties with antibodies to detect A Beta 1-42, only five patients were analysed. The results showed however that the method could separate AD patients from controls.
Droplet array digital analysis (DTU): For the work performed by partner DTU: Patient samples were analyzed on the droplet microarray using 20 hCSF samples shipped from UULM. Briefly, 10-fold diluted CSF samples was flushed over droplet array substrates contains spots of Abeta 1-42 capture antibody. After incubation with a biotinylated detection antibodies and horse reddish peroxidase conjugated streptavidin, droplets with detection substrate was created. Fluorescent imaging and image analysis allowed for counting positive droplets where one spot correspond to one capture Abeta 1-42 peptide. The patient samples could be sorted in two groups based on number of positive spots where healthy individual had a mean spot count of 1623 (SD=570) while AD patients had spot counts of 685 (SD=165). The data suggest that the droplet array technology is a convenient and robust method to diagnose AD patients.
Mass spectrometry based analysis (KTH): This report describes the evaluation of the performance of a platform, comprising immunoprecipitation (IP) and MALDI-Mass spectrometry (MS). This platform was developed in task 3.4 and was utilized to determine the area ratio between Amyloid beta (Aβ) 1-42 and 1-40 peptides in the cerebrospinal fluid (CSF) from Alzheimer and non-Alzheimer disease (AD) patient samples. The IP-MALDI platform included a parallel batch immunoprecipitation (IP)-procedure using magnetic beads, which allowed a simultaneous analyte concentration determination from several samples. The concentrates were deposited on microchip-based MALDI-MS targets by means of a computer-controlled robotic device. The mass spectra obtained provided the possibility for an unambiguous identification of all visible Aβ-peptides. Subsequently, ratios between different Aβ-peptides could be calculated. Moreover, post-translational modifications were also detected.
The developed IP-MALDI-MS platform reached a satisfactory analytical performance regarding precision, with coefficients of variation (CV) below 18%. Measurements, using a set (20 + 20) of CSF patient samples showed that our IP-MALDI-MS approach was able to distinguish between samples from AD and non-AD patients, with a clear statistical significance
Golden standard (UKES/UMG): As a reference method, the commercially available electro-chemiluminescence Aβ-multiplex assay (Mesoscale Discovery Aβ Peptide Panel 1 (6E10) V-Plex was selected as a ”gold standard” for comparisons. It allows for the simultaneous immunological measurements of the Aβ peptides Aβ38, Aβ40 and Aβ42 in a 96 well assay. The MSD V-Plex assay was subjected to a partial, “fit for purpose” assay validation program addressing the lower limits of detection, the lower limits of quantification, impact of sample dilution and parallelism, precision and analytical spike recoveries. In general, the assay performed very well in our hands, and it passed defined acceptance criteria regarding intra- and inter-assay reproducibility, parallelism and spike recoveries. The clinical sample under investigation included 62 patients diagnosed as probable dementia due to Alzheimer’ disease (AD) and 90 subjects categorized as improbable AD (non-demented disease controls, nDC). We performed Receiver-Operator-Characteristics (ROC) analyses for Aβ42 and the Aβ42/40 ratio according to the MSD-assay and calculated cutpoints (maximum Youden indices) of < 481.6 pg/ml for CSF Aβ42 and < 0.0635 for the Aβ42/40 ratio for the discrimination between the two diagnostic groups. In summary, we consider the MSD V-Plex assay suitable for routine measurements of Aβ38, Aβ40 and Aβ42.

Potential Impact:
Impact and dissemination
Introduction

Figure 1. Overview of impact of the NaDiNe project. NaDiNe has been firmly planted in the R&D sphere (light blue) with a few activities going into the general societal sphere (commercialization or usage in the clinic). Dashed lines indicate projections with further additional R&D funding.
NaDiNe has addressed areas of technologic development and diseases with continuously increasing market forecasts. The achievements of the project have an important economic impact in: 1) analytical instrumentation; 2) diagnostics and monitoring of dementia patients. The achievement in the NaDiNe project is summarized in Fig. 1 below. The graph shows an intricate network of output in relation to R&D and wider societal impact. Overall the NaDiNe project has mostly worked in the R&D field with small activities in commercialization and end user product development. A number of new techniques developed under NaDiNe are sensitive enough to detect very low expressed biomarkers in blood such as those expected for neurodegenerative disease. Unfortunately, we did not find suitable biomarkers for diagnosing Alzheimer’s disease (AD) in blood. However, neurofilament measured in blood provided diagnostics for amyotrophic lateral sclerosis (ALS). The NaDiNe project has achieved a large knowledge base for various kinds of diagnostics and future advanced diagnostics devices. These techniques can be used for ultrasensitive detection of, e.g. cancer and infections in humans and animals. The technologies are also applicable to environmental control such as analysis of drinking water.
Impact of better diagnostics provided by NaDiNe and spinoffs
Neurodegenerative disease.
According to the World Health Organization, the number of people with dementia worldwide in 2010 is estimated at 35.6 million and is projected to nearly double every 20 years, to 65.7 million in 2030 and 115.4 million in 2050. The total number of new cases of dementia each year worldwide is nearly 7.7 million. Treating and caring for people with dementia currently costs the world more than US$ 604 billion per year. AD is the most common form of dementia and thus one of the main causes of disability and dependence in elderly people, with great socioeconomic impact. On this ground, AD has been recognized as a public health priority by WHO which has recently recommended the development of national programs to address dementia, focused on improving early diagnosis and disease management, and providing better support to caregivers.
NaDiNe supports this societal need fully by proving validated assays, biobanking for future biomarker discovery, and a whole range of viable technologies for detecting biomarkers in CSF and blood. Some of these technologies (microfluidics, nanoparticles, microarray, sensing) can be packages in a point of care device serving developing countries, while developed countries may opt for automated processes like the mass spectroscopy methods developed in NaDiNe. The foundation is thereby laid in NaDiNe to improve disease management of neurodegenerative diseases. Even small improvements of disease management may have large positive economic consequences.
Cancer.
Some of the developed techniques in NaDiNe proved to be so sensitive and powerful that they are better used for cancer diagnostics as compared to neurodegenerative diagnostics. Preliminary tests performed in the NaDiNe project indicated that some of the technologies could be converted to cancer diagnostics tests for breast cancer and leukemia. Oncology is the second fastest growing segment of the molecular diagnostics market (infectious diseases being the fastest growing). The technologies proposed here is equally ground breaking for infectious diseases, as they also typically require long procedures in order to amplify few infectious agents from a large volume of patient samples. Furthermore, other preliminary experiments in the NaDiNe project showed ultra sensitive detection of bacteria in milk using a highly efficient sample preparation method. Hence, the proposed technology under NaDiNe could carve a large share of the molecular diagnostics market after future maturation (Fig 1).
The total economical impact on cancer treatment of NaDiNe achievements is likely much larger than just getting a part of the diagnostic market as cancer-related costs in the EU were about 126 billion Euro in 2009. 40% were directly associated with health care costs while the rest were lost work, care leaves for relatives etc. The cancer burden is projected to increase just like AD over time given EU’s ageing population. Diagnosing disease efficiently leading to more efficient treatment will have a large economic impact. A 10% efficiency boost would give the EU about 12 billion Euro annually in savings (and 120 000 saved lives per year) using 2009 annual cancer cost numbers. NaDiNe has provided state of the art technologies and knowledge to strengthen Europe in a highly competitive and large market, and the potential products are poised to reduce the economic burden of cancer.
Molecular diagnostics.
Digital analysis of samples has a strong future because of its sensitivity and ability to process large sample volumes. Some digital analysis was explored in the NaDiNe project and a so-called droplet array was the most successful. It is expected that digital analysis will have a large impact on molecular diagnostics.

Sales of molecular diagnostics totaled $4.1Billion (app. €3B) in 2010 and $4.5B (app. €3.4B) in 2013 (http://www.darkdaily.com/frost-sullivan-report-identifies-molecular-diagnostics-as-fastest-growing-sector-of-clinical-pathology-laboratory-testing-03192012#axzz3Aev8WjxQ) http://www.grandviewresearch.com/industry-analysis/molecular-diagnostics-market http://www.marketsandmarkets.com/PressReleases/molecular-diagnostic.asp). Estimates of annual growth rates vary from 8.7 to 11%. This shows that NaDiNe provides tools and technology into a fast growing field.
Much of molecular diagnostics is driven by polymerase chain reaction (PCR) and protein assays. Due to its ability to screen large sample volumes, the droplet arrays platform can compete efficiently with PCR for a number of applications including cancer diagnostics and infectious disease diagnostics. This indicates that the platform has the possibility to take a significant part of the diagnostics market as droplet arrays have a number of practical, economical and analytical advantages compared to PCR. When super sensitive detection is needed for identification of protein markers in blood, droplet arrays will be useful and have a chance to compete with existing de facto gold standards, such as enzyme linked immune assays (ELISA) as well as existing digital protein analysis platforms (Qanterix). Therefore, we strongly believe that droplet arrays will produce job opportunities in the EU as well as enable better disease management through streamlined and highly sensitive assays.
Point of care diagnostics encompasses many variants and implementation modes, but most techniques are typically based on simple flow control. These devices are not compatible with advanced diagnostics such as those demonstrated in NaDiNe. The prototypes developed under NaDiNe demonstrate that advanced molecular diagnostics can be packaged into devices that can be further developed into point of care devices. We expect that these types of advanced diagnostics devices will enable new types of diagnostics in medicine but also in other areas.

Providing a highly skilled and educated work force to relevant industries
It is without doubt that future molecular diagnostics devices will be very complex, but at the same time, as easy to operate for the users as modern mobile phones. In the NaDiNe project we have had numerous students working with the development of advanced systems and characterization and usability of potential biomarkers. NaDiNe has therefore provided Europe with a number of people that are knowledgeable in instrument development, usage of these instruments, developing molecular diagnostic kits, and their application in the clinical setting.
Bio banking
NaDiNe has assembled a large biobank for the research in the project. However, the biobank is equally usable in the future for further studies. Having a large available biobank is really important for rapid biomarker discovery. The concrete impact of the biobank is not easy to predict and depends very much on the research it is going to enable. There are, however, numerous examples where biobanks provide competitive advantages in research for biomarkers. In this context, it is noteworthy that any new biomarkers together with suitable detection technologies are patentable intellectual properties. We thus see this effort as an investment for the future.
Analytical tools and techniques developed in the project.
The technological development in NaDiNe lays the foundation for further studies and later possible commercialization.
Particles
There has been a large development of innovative nanoparticles in the NaDiNe project. Specifically, we specialized on their biofunctionalization. All the achievements obtained in this project constitute the starting point for further inventions in the field of preparation and application of biofunctionalized magnetic particles. Most of the developed approaches are versatile and applicable in variety of therapeutic and diagnostic contexts. A large part of existing (and, arguably, future) biomedical methods are particle-based, and thus there is a substantial commercial interest. Particles have furthermore large binding capacity and can be extended to other usages, such as, for instance, nano particle cleanups and other environmental applications.
Microfluidics
Many of the microfluidics techniques developed in NaDiNe are modules that are going to be used in compound instruments or work flows. For instance, a - microfluidic fluidized bed was developed to extract and pre-concentrate analytes from large samples volumes (1-1000µL). This is important and useful for a number of down-stream analytical techniques such as mass spectrometry, protein microarrays (both tested in NaDiNe), but also other detection or identification techniques. There is an unmet need for highly automated and miniaturised sample pre- concentration in order to improve detection limits and specificity of a range of analytical instruments existing in labs today. This technology has therefore a large commercial potential if connected to down stream analysis equipment.
Detection technologies
NaDiNe has explored a whole range of different detection technologies including electrochemistry, quantum dots enhanced detection, enzymatic amplifications, fluorescence and absorbance. The detection methods were furthermore combined with various techniques to obtain identity of the biomarker such as on chip capillary electrophoresis, isoelectric focusing and sandwich immune assays. In the end common fluorescent based microarrays, enzymatic amplified droplet array detection were superior combinations. However, the other techniques and combination may have suitable application is other areas where low detection limit is not required. The importance of this exhaustive effort is the knowledge, which are the viable alternatives for future instrumentations.
Mass spectrometry
The new concepts for ultra-sensitive matrix-assisted laser desorption ionization (MALDI) mass spectrometry, as developed in this project, could have a considerable impact for the discovery of unknown biomarkers, not only in relation to Alzheimer’s disease, but also to other diseases such as different forms of cancer for which the number of reliable clinical diagnostic tools is very limited or not available yet.
Therefore, a scenario could be foreseen that the technology platform developed in this project could have a large socio-economic impact, and a great benefit for society as a whole. However, it should be pointed out that thus far, the project has only shown the feasibility of the concepts. A more focused, large effort is required to develop the technology into a clinical tool for routine use (Fig. 1). Apart from further basic improvements, such work should include the development of automated handling of patient samples, as well as a sample work-up system, using a high throughput robotized system.
Biomarkers
Especially with the recently investigated markers Ubiquitin, Neurofilaments and SerpinA1 disease management of an individual patient can be better organized. This is far beyond a proof of principle study, but will have a direct differential diagnostic value. For markers like tau and abeta1-42 the diagnostic value was already established. This was not the case for the recently investigate markers. Ubiquitin will directly help us to define subtypes of AD patients and may help to restratify therapeutic trails. The same hold true for SerpinA1. This is the first neurochemical marker, which helps to predict the development of a dementia in Parkinsons´s disease. The measurement of NF has the impact to be included in the diagnostic criteria for Amytrophic lateralsclerosis (ALS). Instead of waiting until all clinical criteria are fulfilled, patients under the differential diagnosis of an ALS can be directly treated if NF levels are elevated or included in therapeutic trail at a time point which a therapeutic effect is more realistic. The economic impact must be evaluated in further studies and is not within the scope of NaDiNe.
Commercialization
Fluigent – fluidics control instruments.
Three different products have emerged from the NaDiNe project and are now commercialized by Fluigent:
• The Easy Switch SolutionTM platform: a product combining hardware and software to be able to inject sequentially different fluids into a microfluidic system.
• The Flow-Rate Platform: a product enabling the monitoring of flow-rates generated by pressure pumps.
• The Flow-Rate Control Module: an algorithm to be able to control the flows by flow-rate using pressure actuation of fluids.
The different products enable the company to increase its sales revenue, its market share and also to penetrate new markets. One such market is microfluidics driven by syringe pumps, the most popular solution for microfluidics at research labs. The thousands of microfluidics labs constitute a large commercial opportunity. However, Fluigent also developed an application programming interface (API) which significantly broadens Fluigent’s potential customer base.

Already today, almost all instrumentation in the general practitioner’s office is based on microsystems and some are even microfluidics based. With the NaDiNe project, Fluigent is well positioned to go into further collaboration to solve the next step in diagnostics – namely molecular diagnostics that is usually not found in the practitioner’s office (or in bed side diagnostics) yet. In fact, the prototype developed by Fluigent is capable of controlling automated and highly advanced molecular analyses, as was successfully demonstrated in the project. The billion USD molecular diagnostics market will likely increasingly incorporate more miniaturized analysis systems. The NaDiNe project therefore has strengthened Fluigent in particular and Europe in general to address this fast growing market.
AK Diagnostics, a new NaDiNe spinoff company.
One of the techniques developed in the project was the droplet array digital platform where very small concentrations of biomarkers could be detected. The platform has a potentially large commercial value in molecular diagnostics in general. Therefore, we have investigated patenting and commercialization. The result is a Danish spinoff company (AK-Diagnostics), possibly a patent application (still in preparation) and, at a later stage, scientific journal publications. The goal of the company is molecular diagnostics on serum samples. Areas of interest are cancer and infection diagnostics. AK Diagnostics is looking into point of care diagnostics devices but the technology is also highly suitable for single cell proteome or transcriptome analysis. Recently, talks with a medium sized European company have been initiated to explore collaborations to put the technology into research instruments for various purposes such as epigenomics.
Diagnoswiss
Diagoswiss has had the opportunity to develop assays for neurodegenerative diseases, which will be considered to be commercialized. They further improved upon their equipment (both from a hardware and software point of view) and, together with the new assays, are now able to offer their customers advanced solutions for medical diagnostic applications.
Moravian Biotech - Antibodies
In a broad sense, the reagents we have produced are expected to contribute to further understanding of the (neurodegenerative) diseases, allowing for improved diagnostics and potentially identifying novel therapeutics, with the corresponding socio-economic benefits. With specific respect to Moravian Biotech (MB), the reagents that have been produced will be further commercialised to allow their wider use across the scientific community, also providing MB with a stable income. This future impact relies on the dissemination of data relating to the reagents, which is available on the MB website and is accessible through the reports and the website of the NaDiNe project.

Dissemination
The NaDiNe project has the potential to be of larger than average public interest due to the fact that Alzheimer´s disease is widespread and frequently covered in the media. However, the NaDiNe project’s technical and medical results are, while important, challenging to communicate to a wider audience. We have therefore prepared a dissemination manual based on this experience. Instead of a wide outreach, NaDiNe has focused its dissemination mostly to researchers. The NaDiNe project has resulted in >90 peer reviewed journal papers and >150 conference and lecture invitations during the last 5 years indicating a large academic impact. The key results of NADINE are presented on NaDiNe’s home page (http://www.fp7nadine.eu Figure 3 with all partner listed). Extensive consortium networking has taken place and numerous PhDs and postdocs have accessed a large multidisciplinary research community. An outline of NaDiNe’s dissemination activities is given in Figure 2.

Figure 2. Dissemination in NaDiNe.

Networking
The NaDiNe consortium was highly cross-disciplinary with researchers from physics, engineering, chemistry, biology and medicine all working together to reach the goal of the project. This cross-disciplinary consortium also provided large possibilities to learn form each other.
During the project we had extensive flow of people between different labs and countries. Furthermore, we had two in-person project meetings per year where principle investigators as well as students attended, as well as numerous teleconferences with the entire project team, or, more frequently, focused subsets of the project team. The opportunity to collaborate and exchange experiences with many excellent partner institutions within the project enabled researchers to broaden their existing knowledge and skills, which can be applied in further work. Having been part in the FP7 framework NaDiNe project serves as a reference point for future potential partners in similar projects, as well as improving the ability to interact with and provide additional resources to collaborators, both current and coming.
Articles, Conferences and talks
>90 articles and >150
Website and communication plan
DTU has set up a website, www.fp7nadine.eu from the beginning of the project. The webpage has been the project’s public face throughout the project, where publications and conference presentations have been updated regularly. The web page will be updated for the next 6 months to include the last results of the project. The web site will be publically accessible for the next three years.
A communication plan for the partners how to disseminate their results to the broader public. It has been sent to all partners. The plan includes for example information on
• The identification of different target groups such as scientific community, general public and external stake-holders (industry and patients groups).
• Identification of primary and secondary stories
• Different means to promote the results
• Individual story placement and press stories
• Creating attention in the press

Demo events and training.
Three demo events has been organised during the project course
• At the partner meeting on 5-6 March 2015 at DTU, Fluigent organized demo and training (with the help of other partners, ICN/ICRM/Curie). Fluigent brought the flow control demonstrator and operated typical experiments using colored water. In addition, DTU demonstrated the droplet array platform. The objectives were to show the demonstrator to the clinicians and to explain them how they can use it. Number of participants: 30
• The mass spectrometry was demonstrated for the clinicians at a separate meeting in Stockholm on 24 February 2015. Number of participants: 6
• On 3 July 2015, Fluigent organized a workshop in their facilities. This event was divided in lectures and training sessions and was not only addressed to members of the Consortium but also external stakeholders and the pharmaceutical industry. Number of participants: 18

List of Websites:
www.fp7nadine.eu