Skip to main content
European Commission logo print header

Peptide-based diodes for solar cells

Final Report Summary - PEPDIODE (Peptide-based diodes for solar cells)

Executive Summary:
Most of the objectives and milestones due within the PEDIODE project have been achieved. WP1: The consortium considerably advanced their solid-material-based synthesis method. Now, KIT and PPP can produce arrays with 10,000 peptides per cm(+2) with their patented novel cLIFT procedure for just a few Euros chemical costs. SME PEPperPRINT plans to commercialize this procedure soon. Moreover, KIT managed to transfer up to 10,000 peptides per cm(+2) to a recipient surface with only very limited lateral diffusion. All the deliverables and milestones related to WP1 have been achieved. Quite recently, KIT also achieved the proof of principle (and a patent application) in synthesizing very high density peptide arrays (up to 25 Million per cm(+2)) in small cavities. This approach has the advantage that we might be able to profit from very small (<10µm) cavities that could be sealed from neighbouring cavities with other peptides, thereby allowing to assemble dimers, trimers, or tetramers from synthesized peptides. These might be screened e.g. for novel reaction centres in the future. WP2: Although AMU couldn’t synthesize all the compounds as planned originally, they managed to synthesize closely related artificial catalytic centres and chromophores, and more of these than originally envisioned. Moreover, KIT’s novel cLIFT method obviated the need to synthesize these very expensive artificial amino acids in the gram scale. WP3 & WP5: The consortium could use IMS’ measuring chip to determine voltage / current characteristics for single pixel electrodes, thereby validating the concept to link the nano-world (many different peptides) to the macro-world (reading out their electronic properties on a chip’s pixel electrode) as such. However, the first versions of IMS’ measuring chip showed intolerable internal leakage currents that ruled out its use for screening purposes. As contingency measures the consortium started to develop alternative screening assays that should directly screen for LHCs and RCs, while IMS designed and manufactured a new version of their screening chip that has only been tested yet at IMS due to time constraints. WP4 & WP6: MIGAL and CUNY employed recombinantly expressed small proteins to find modules that relay excitation energy to nearby chromophores (artificial light harvesting complexes), and they investigated a small protein that could coordinate a pair of chlorophylls similar to the special pair in the photosystem’s reaction centres. However, so far we failed to identify an artificial reaction centre, and we could not assemble these building blocks into a novel kind of biomimetic solar cell as originally planned within WP7.
Project Context and Objectives:
Objectives for the whole project: We wanted to lay the foundations of a bioinspired solar cell that – similar to the photosystem – uses self-assembled building blocks to efficiently convert a photon’s energy into electrical energy. To do this, nature uses “light harvesting complexes” (LHCs) that – similar to a ball path – direct the absorbed light through an array of closely spaced (~1nm) chlorophylls to a “reaction centre” (RC) where charge separation and build-up of electrostatic potential takes place. Unfortunately, nature’s invention generates a gradient of protons over a membrane, while a solar cell in a technical world should generate a gradient of electrons. Our goal was to develop screening procedures that would allow us to screen for novel small-protein-based LHCs and RCs, and to assemble them into a bioinspired solar cell. Moreover, these screening procedures should allow us – in the long run and beyond the PEPDIODE project – to screen in an evolutionary approach for improved building blocks that could be produced at low cost in bacteria.
Objectives broken down to work packages: We wanted to further develop our tools and technologies in order to achieve the following goals:
(I.) Within WP1 participants KIT and PPP wanted to advance different variants of their solid-material-based-synthesis-method (Breitling et al., 2009, Molecular BioSystems 5, 224; Stadler et al., 2008, Angew. Chem. Int. Ed. 47, 7132) up to a level that should allow us to synthesize 100.000 different peptides on a glass slide for chemical costs of ~100€ (deliverable D12; 10.000 peptides per cm(+2)). In doing that, we always first pattern the surface – where peptide synthesis takes place – with amino acid derivatives that are embedded in a solid matrix material. Next, the patterned synthesis slide is heated up in an oven, which frees hitherto immobilized amino acids to diffuse to the surface of the synthesis slide where they couple to free amino groups. This solid-material-based combinatorial synthesis procedure is identical to the 40-years-old Merrifield synthesis, with the only difference being an intermittent “freezing” of the coupling reaction at room temperature. Moreover, we wanted to advance this technology up to a level of an automated machine, which should allow us to commercialize this technology within PPP (with customers mainly from the life sciences).
(II.) Nature’s photosystems rely on a panel of specialized light-harvesting modules, e.g. chlorophylls and carotenes. In light-harvesting complexes (LHCs) these building blocks are exactly arranged at a nanometre-scale distance from each other. Therefore, within WP2 participant AMU wanted to synthesize a panel of light-absorbing and electron-relaying building blocks, e.g. porphyrins (deliverable D21) that could be coupled at arbitrarily chosen positions within a peptide’s sequence, which – in principle – allows us to position these building blocks in 0,35 nm steps within the peptide’s sequence. When combining the synthesis capacity from WP1 with WP2’s artificial amino acid building blocks, we should then be able to synthesize a whole panel of small molecules that, eventually, fold into 3D structures with exactly defined positions of their light-absorbing or electron-relaying building blocks. When we started the PEPDIODE project, we assumed that AMU’s specialized artificial amino acid building blocks were needed in gram amounts, and, in addition, with a sometimes difficult-to-achieve C-terminal chemical activation, which is needed when using printed amino acid particles for peptide synthesis. However, two unexpected hallmarks of KIT’ and PPP’s novel cLIFT procedure for the synthesis of very high density peptides arrays – that was developed during the PEPDIODE project – obviate the need for large amounts of artificial amino acids: (1) the cLIFT method has a frugal consumption of chemicals, and (2) it can use also use non-activated building blocks for peptide synthesis. Therefore, AMU tried to synthesize more building blocks than originally planned, but in smaller amounts, and, if chemically difficult, without C-terminal chemical activation.
(III.) The capacity to synthesize many different small proteins (= peptides; WP1) with built-in light-absorbing porphyrins and electron –relaying ferrocens (WP2) is certainly not enough to achieve our goal of a self-assembled, bio-inspired solar cell. IMS therefore strived to develop within WP3 a screening chip that would allow us to individually and parallel measure the electric properties of up to 10.000 different peptides per cm(+2) that are linked to individual pixel electrodes. Eventually, we could then screen for peptides that conduct a current in one preferred direction (“peptide diode”; see WP5), which is also a feature from nature’s reaction centres that separate an energy-rich electron from its hole. Starting with a CMOS chip that stores for every pixel the electrons that are mobilized by light (i.e. a chip for a digital camera), IMS wanted to design and manufacture several generations of chips that are then tested for suitable properties, e.g. leakage current, sensitivity, background noise etc.. Please note, that similar chips in digital cameras – the “ancestors” of our screening chips – are only used in a dry environment without any interface to nasty electrolyte solutions, which adds to the difficulties of this formidable task.
(IV.) Within WP4, and starting from recombinantly expressed protein-scaffolds, CUNY and MIGAL wanted to assemble RC- and LHC-like (small)protein-cofactor modules. The main task within this work package was to design, prepare, and characterize simple and robust water-soluble protein scaffolds that can specifically bind and organize multiple redox cofactors and pigments at well-defined stoichiometry and geometry. These, eventually, should allow relaying excitation energy to neighbouring choromophores (i.e. a light-harvesting-complex; LHC), and, eventually, facilitate charge separation as an artificial reaction centre (RC).
(V.) The overall aim of WP5 was to develop methods that should allow the PEPDIODE consortium to link the nano-world (many different nanoscale peptides) to the macro-world (a measuring chip that would report the electrical properties of many different peptides in parallel). In order to achieve this overall goal, the consortium wanted to develop (a) a method to transfer up to 10,000 peptides per cm(+2) to a rigid gold covered support (D53), (b) to interface the peptides sitting on “their” pixel electrodes on the measuring chip with a computer (D52), (c) to readout in good sensitivity pixel-specific currents in parallel for the different pixel electrodes from IMS measuring chip, and (d) to find a first molecular diode.
(VI.) Within WP6, we wanted to advance the completely new concept of linking a peptide diode to artificial RC-LHC proteins. We intended to reproduce Nature’s architecture with artificial RC and LHC protein-pigment complexes and then couple these to a peptide-based diode that will serve as an interface for electron transfer to the electrode. At the first stage, we planned to test in solution the assembly of protein-cofactor complexes, photo-induced charge separation and ET through the redox chains of RC modules, excited state dynamics in LHC modules, and LHC to RC excitation energy transfer. Next, we wanted to integrate successful modules with peptide diodes on the measurement chip and test them for photocurrent generation using similar methodology as in WP5. The basic idea behind this approach was that due to the inherent modularity in our design we should then be able to test the functional components independently. Thus, we planned to first test RC-peptide diode coupling on the chip, while RC-LHC coupling may be tested independently in solution. Then, we wanted to evaluate the best LHC-RC-peptide diode combinations – in a kind of evolutionary approach – for most effective photocurrent generation.
(VII.) Finally, within WP7 we wanted to assemble the different parts (reaction centre, peptide diode, light harvesting complexes) into a biomimetic solar cell.
Project Results:
Summary work progress for the whole project: Most of the objectives and milestones due within the PEDIODE project have been achieved. WP1: The consortium considerably advanced their solid-material-based synthesis method. Now, KIT and PPP can produce arrays with 10,000 peptides per cm(+2) with their patented novel cLIFT procedure for just a few Euros chemical costs. SME PEPperPRINT plans to commercialize this procedure soon. Moreover, KIT managed to transfer up to 10,000 peptides per cm(+2) to a recipient surface with only very limited lateral diffusion. All the deliverables and milestones related to WP1 have been achieved. Quite recently, KIT also achieved the proof of principle (and a patent application) in synthesizing very high density peptide arrays (up to 25 Million per cm(+2)) in small cavities. This approach has the advantage that we might be able to profit from very small (<10µm) cavities that could be sealed from neighbouring cavities with other peptides, thereby allowing to assemble dimers, trimers, or tetramers from synthesized peptides. These might be screened e.g. for novel reaction centres in the future. WP2: Although AMU couldn’t synthesize all the compounds as planned originally, they managed to synthesize closely related artificial catalytic centres and chromophores, and more of these than originally envisioned. Moreover, KIT’s novel cLIFT method obviated the need to synthesize these very expensive artificial amino acids in the gram scale. WP3 & WP5: The consortium could use IMS measuring chip to determine voltage / current characteristics for single pixel electrodes, thereby validating the concept to link the nano-world (many different peptides) to the macro-world (reading out their electronic properties on a chip’s pixel electrode) as such. However, the first versions of IMS’ measuring chip showed intolerable internal leakage currents that ruled out its use for screening purposes. As contingency measures the consortium started to develop alternative screening assays that should directly screen for LHCs and RCs, while IMS designed and manufactured a new version of their screening chip that hasn’t been tested yet due to time constraints. WP4 & WP6: MIGAL and CUNY employed recombinantly expressed small proteins to find modules that relay excitation energy to nearby chromophores (artificial light harvesting complexes), and they investigated a small protein that could coordinate a pair of chlorophylls similar to the special pair in the photosystem’s reaction centres. However, so far we failed to identify an artificial reaction centre, and we could not assemble these building blocks into a novel kind of biomimetic solar cell.


Work progress WP1:
One very early goal within WP1 was the routine availability of peptide arrays right from the start of the PEPDIODE project. Indeed, PPP advanced their 2nd generation peptide laser printer (Fig. WP1-1) significantly to a user-friendly machine that synthesizes peptide arrays for the PEPDIODE project and also for PPP’s customers (D11; M11). KIT and PPP advanced this synthesis capacity further up to the level of 10,000 peptides per cm(+2) (D12, D13; M12). In order to routinely synthesize such high density peptide arrays, KIT assembled, and in 2016 together with PEPperPRINT will commercialize a prototype machine that punches out and transfers tiny material spots from foils that are covered with a thin layer of our amino acid derivatives for combinatorial synthesis – a recently patented method that was not foreseen when writing the PEPDIODE grant proposal (Figs. WP1-2, WP1-3). One highlight certainly is the unexpected high quality of the arrays synthesized with this cLIFT method (Fig. WP1-4). Moreover, when advancing the cLIFT method, KIT realized that the laser very reliably transfers very thin (adjustable from 1nm to 50nm) material spots onto the solid support that is used for peptide synthesis (Fig. WP1-6). Recently, we proved that this feature can be exploited to also couple non-activated amino acid building blocks onto the surface by simply adding an activation layer underneath the layer with the non-activated amino acid building block (Fig. WP1-7), which is another commercially very interesting highlight (there is a big market for peptide arrays with post-translationally modified peptides). Thereby, we can employ less demanding chemical synthesis (without C-terminal activation) to generate arrays with expensive artificial building blocks, which further eases the task of our chemical partner AMU. Finally, in the last year of the PEPDIODE project yet another very simple method emerged to synthesize in small cavities up to 25 Million peptides per cm(+2): When spreading a mixture of quantum dot coded amino acid particles with a diameter of exactly 9µm over a microstructured glass slide with cavities of 10µm in diameter, only one particle fits into one cavity, while the colour code tells which amino acids were deposited in the different cavities (Fig. WP1-8). In the future, this approach should allow us to synthesize millions of peptides (each with known sequences), then detach them from the solid support post synthesis (e.g. by UV), and then assemble them into larger protein like structures which might have the properties of a reaction centre. Another ambitious goal that KIT achieved was the transfer of up to 10,000 peptides per cm(+2) in the array format from a support used for peptide synthesis to a recipient support (D14, D15; M13). Indeed, KIT achieved this transfer of 10,000 peptides per cm(+2) to an adjacent solid support with very little lateral diffusion observable (Fig. WP1-11). Except for the achieved density of peptides that are synthesized with the “peptide laser printer” (we achieved ~800 per cm(+2) vs. originally planned 1,000 peptides per cm(+2)) all objectives due within WP1 have been achieved.

Deliverable D11 and milestone M11 (Second) prototype peptide laser printer (Fig. WP1-1): Deliverable D11 and milestone M11 have both been achieved. Our “peptide laser printer” is used by profitable SME PPP to routinely produce peptide arrays at a density pf 800 peptides per cm(+2), and on a format of 20 x 20 cm(+2) (Fig. WP-2). The solid-particle-based procedure is shielded by a broad patent (Breitling, F., Poustka, A., Groß, K.H. Dübel, S., and Saffrich, R. (1999). Verfahren und Vorrichtungen zum Aufbringen von Substanzen auf einen Träger, insbesondere von Monomeren für die kombinatorische Synthese von Molekülbibliotheken. Patents AU773048B (issued 2004), EP1140977B1 (issued 2005), US20020006672A1. Status: Licensed to PEPperPRINT GmbH.).

Deliverable D12 Synthesis of 10.000 peptides per cm(+2)
KIT and PPP wanted to develop a method that allows for routine synthesis of arrays with 10,000 different peptides per cm(+2) and with only mg amounts of very expensive artificial amino acid building blocks. Another goal was to allow for routine synthesis in order to supply the consortium with high density arrays, and, beyond the PEPDIODE project, to commercialize it for routine peptide synthesis within SME PEPperPRINT.
“Chip printer”: We first constructed a prototype chip printer (see 2nd periodic report). Using this machine, we could produce high density peptide arrays and we could publish these results in a very good journal (Löffler et al., Advanced Functional Materials, 2012; impact factor ~10). Unfortunately, we had to realize that far too many chips were destroyed by a short-circuit fault when printing the particles from the chip’s surface to a glass slide that was used for peptide synthesis. We therefore abandoned this line of development.
Combinatorial Laser Induced Forward Transfer method (cLIFT): In a 2nd trial, KIT and PPP invented and patented (Maerkle F, Nesterov-Müller A, Breitling F, Löffler F, Schillo S, Bykovskaya V, von Bojnicic-Kninski C, Leibe K. filed on 17th 04.2013. Verfahren zur kombinatorischen Partikelablagerung zur Herstellung von hochdichten Molekülarrays, insbesondere von Peptidarrays, und damit erhältliche Molekülarrays. PCT/EP2013/001141) yet another solid-material-based structuring method that meanwhile also allows us to achieve the objectives of deliverable D12, i.e. to routinely synthesize 10,000 different peptides on an area of one cm(+2) in high quality, and for ~5€ chemical costs. Within the 3rd reporting period, KIT advanced this machine up to the level of a completely automated synthesis robot that will soon produce arrays in a commercial setting (Fig. WP1-2). Recently, we submitted a manuscript that describes this method (Loeffler F, Foertsch TC, Popov R, Schlageter M, Sedlmayr M, Ridder B, Althuon D, von Bojničić-Kninski C, Weber LK, Fischer A, Bykovskaya V, Buliev I, Hahn L, Meier MAR, Braese S, Powell AK, Balaban S, Breitling F, Nesterov-Mueller A. High-flexibility combinatorial peptide synthesis with laser-based transfer of monomers in solid matrix material. submitted).
The core part of our machine is made of a 2D laser scanning arm (Fig. WP1-2) that punches out tiny pieces of solid material from a thin layer of solid material where the chemically activated amino acid building blocks are embedded within (see Fig. WP1-3 for a schematic explanation of the method). Thereby, a pattern of small material spots with a pitch of ~100µm that all comprise a first amino acid is transferred to a glass slide that is used for peptide synthesis. Afterwards, a slide loader positions additional material layers with other amino acid derivatives on top of the synthesis slide that are all transferred in a similar, completely flexible way by the laser to discrete positions – a kind of multicolour printing process. Please note, that this loading process doesn’t employ any sophisticated alignment procedures (as in the peptide laser printer, and also in the chip printer): positioning is built into the 2D laser scanning arm at an accuracy of +/-2µm (the laser “visits” from time to time a reference point). Surprisingly, the cLIFT method generates arrays of strikingly good quality (Fig. WP1-4). Moreover, we think that the method should be scalable to spot densities of 1 million per cm(+2) (when using a very thin light absorbing layer in a dust free environment). A first part of the cLIFT procedure was published in a high ranking journal (Maerkle et al., Advanced Materials, 2014; impact factor ~17).
In addition, when developing the cLIFT machine, KIT realized that the cLIFT method very reliably transfers very thin (1nm to 50nm) pan-cake-like material layers from the donor slide to the acceptor slide (where synthesis takes place; Fig. WP1-6). We used this feature to deposit very thin material layers of non-activated amino acid building blocks upon an activation layer (made of DIC and HoBt, Fig. WP1-7), and observed bright signals of coupled amino acids. This feature greatly reduces the work load that is needed within work package 2 – AMU’s task to develop artificial amino acid building blocks: It is no longer necessary to synthesize these very expensive building blocks in gram amounts, and with a C-terminal activation (e.g. OPfP esters) which proved to be very difficult.
Doing additional chemical reactions: As mentioned above, the very thin pan-cake-like material spots that are deposited with the cLIFT method (Fig. WP1-6) allow us (1) to synthesize >10.000 peptides per cm(+2) in high quality (should be scalable to spot densities of 1 million per cm(+2) or more), (2) to do that with minute amounts of very expensive material, e.g. the porphyrin-amino-acids from AMU (mg vs. gram), (3) to do it with non-activated amino acid building blocks (because very thin layers of different materials perfectly mix when heated; see Fig. WP1-7d), and (4) to do it routinely – hopefully soon in 2016 – also in a commercial setting (which would shift the difficult task of quality controls to SME PEPperPRINT). Currently, KIT systematically builds into our peptide arrays all kinds of non-activated artificial amino acids (from AMU and commercially available) to validate this point (the work is delayed because the PhD student was ill). In addition, we realized that the very same feature might also allow us to do different kinds of chemistry beyond peptide synthesis (add layers with different building blocks, catalysts, bases, acids, scavengers ...) and monitor the success of synthesis spot by spot by scanning with a mass spectrometer (we started a collaboration with Prof. Bernhard Spengler, Giessen; www.uni-giessen.de/cms/faculties/f08/departments/iaac/spengler/research/projects/imaging-mass-spectrometry).

Deliverable D13 and milestone M12 Small scale formulation (10g batches) of amino acid particles:
Deliverable D13 and milestone M12 have both been achieved in time (see 2nd periodic report and deliverable report D13). However, we focused on the cLIFT method to synthesize our peptide arrays, and, therefore, don’t need these particles for this purpose anymore. However, this line of research resulted in yet another method to synthesize very high density peptide arrays:
One-cavity-one-peptide method: Recently, KIT (Alexander Nesterov-Müller) came across the idea that when spreading amino acid particles with a very narrow size distribution over glass slides that are microstructured with exactly defined cavities, we might find conditions where exactly one particle fits in one of these cavities. If these amino acid particles are differently labelled, e.g. with quantum dots (Fig. WP1-8) we could especially easily synthesize > 1 million peptides – each with a known sequence – and without employing any complicated machinery for just a few Euro per glass slide (just spread the particles over the surface and gently rub them; Fig. WP1-8). KIT achieved the proof of principle with this method at a density of 25 million peptides per cm(+2) (data not shown). KIT used its proof of principle experiments to patent this procedure (Patent application 102015117567.3: D. Althuon, F. Breitling, V. Bykovskaya, F.F. Loeffler, A. Nesterov-Mueller, R. Popov, B. Ridder, C. v. Bojnicic-Kninski. „Ultra-hochdichte Oligomerarrays und Verfahren zu deren Herstellung“. Filed October 15, 2015. Patent application submitted), and to win the KIT internal UpCat competition that aims to train participants to spin out a company.
(Future) screening for reaction centres with the one-cavity-one-peptide method: The one-cavity-one-peptide method is well on track to routinely yield very high density peptide arrays early in 2016 (it is planned to commercialize this technology within a spin out company; first applications will be in the life sciences, e.g. readout serum samples from autoimmune patients). Each peptide of these arrays will be synthesized in a small cavity of some 10µm in diameter by standard Merrifield synthesis, and employing our solid material based approach. These cavities should allow us to process the peptides also in solution (e.g. cleaving deprotected peptides from a UV linker), and, thereby, e.g. circularize or oligomerize them. Since one of the goals of the PEPDIODE project is to screen for novel reaction centres and light harvesting complexes, KIT will dimerize, trimerize or, eventually tetramerize synthesized peptides in solution and within individual cavities (post synthesis and post deprotection, fill in dimerization solution; seal neighbouring cavities from each other; then cleave peptides from a UV-sensitive linker; then oligomerize the synthesized peptides; then analyse them with spectrometer for energy transfer). This will be done by coupling a mixture of two or three different artificial amino acids to the N-terminal end of synthesized peptides that will fluoresce only when assembled into oligomerized protein-like structures (3x 40 amino acid long peptides would yield a molecule of some 12.000d). Currently, KIT tries to identify suitable artificial amino acid building blocks that could serve as a fluorescent oligomerization organizer. The rationale behind this approach is that proteins obviously control the flow of electrons by incorporating “stepping stones” for electrons into the protein (e.g. a porphyrin), while otherwise shielding the electrons in the catalytic centre by a cloud of amino acids with >1nm in diameter from uncontrolled “jumping”.

Deliverable D14 and milestone M13 In situ purification of 1,000 peptides per cm(+2) in the array format: Deliverable D14 and milestone M13 have both been achieved considerably ahead of time (see 1st periodic report and deliverable report D14).

Deliverable D15 and milestone M13 In situ purification of 10,000 peptides per cm(+2) in the array format: In deliverable report D14 KIT successfully immobilized the peptides in the array format on gold-coated membranes down to a resolution of 10.000 peptides per cm(+2) (Fig. WP1-11; see also deliverable report D14), and we investigated the transfer of peptides in the array format also onto rigid surfaces, whereby we had to confine lateral diffusion during these transfers. Most interestingly, however, was the previous finding (see deliverable report D14) that we could cleave the peptides from their normal support that was used for peptide synthesis simply by incubating them in a vapour of conc. ammonia. Within D15 we investigated this finding in more detail (see also the 2nd periodic report).

Conclusions WP1:
All the deliverables and milestones due in WP1 have been achieved. Much better than the original plan (>100x more peptides per area synthesized than anticipated), the consortium managed to develop methods that – in the future – should allow us to routinely synthesize up to 1 million peptides per cm(+2) (compared to 25 peptides per cm(+2) in the previous state of the art; Frank, R. 1991, Tetrahedron 48, 9217–9232), to copy these arrays (doesn’t work routinely yet), and to transfer them under mild buffer conditions to any flat surface employing gold / SH-groups or the biotin streptavidin system. Especially interesting with respect to the goals of the PEPDIODE project will be the one-cavity-one-peptide-method (Fig. WP1-8): It should allow us (1) to synthesize peptides at a density (>1 million per cm(+2)) and (2) at a quality (we can use a vapour of normal solvents) that was not imaginable before the start of the PEPDIODE project. Moreover, (3) we should be able to use these cavities to assemble an array of protein-like structures (with >100 amino acids per molecule that shield electrons from the catalytic centre from uncontrolled “jumping”) that could be screened directly for artificial reaction centres or light harvesting complexes. This will be done by first synthesizing high density peptide arrays (incorporating e.g. porphyrine moieties from AMU), then deprotect the peptides in the array format, then fill them with an assay solution and seal different cavities from each other with a seal before peptides are cleaved by UV from their solid support, then allow them to assemble to dimers or even tetramers, and, finally, analyse them spectroscopically for energy transfer within the protein-like-structures.


Work progress WP2:
The objective of WP2 was to synthesize artificial amino acid building blocks (or click partners) that could be incorporated into growing peptides (Fig. WP2-1) and in the array format. During the PEPDIODE project, KIT developed novel combinatorial synthesis methods, the cLIFT method (Figs. WP1-2 to WP1-4), and the one cavity-one-peptide method (Fig. WP1-8) that allow to synthesize very high density peptide arrays (1) with very frugal consumption of amino acid building blocks (only mg amounts vs. gram amounts), and (2) without the need to activate these building blocks at their C-terminal end (which is sometimes difficult). Therefore, AMU somewhat changed the focus of their work in producing more building blocks than originally planned, but synthesize them in mg quantities (instead of gram quantities) and without C-terminal activation.
In doing that, AMU developed a novel synthetic method for diversely substituted porphyrins (Fig. WP2-2). Because of the synthetic challenges encountered in the preparation of porphyrinic compounds in WP2 we searched for a better yielding synthesis that avoids complicated mixtures of closely related porphyrins which are thus very difficult to separate. The results were published jointly with the group of Prof. Daniel Gryko from Warsaw in Chemistry - A European Journal (Nowak-Król A. Plamont R, Canard G, Andeme Edzang J, Gryko DT, Balaban TS (2015) An Efficient Synthesis of Porphyrins with Different meso-Substituents that Avoids Scrambling in Aqueous Media. Chem. Eur. J. 21, 1488-1498.; DOI: 10.1002/chem.201403677). We also turned our attention to other closely related tetrapyrrolic chromophores – which are similar to porphyrins, namely corroles and phthalocyanines. Several publications resulted which describe the syntheses and properties of these interesting novel compounds (Gao D, Andeme Edzang AJ, Diallo AK, Dutronc T, Balaban TS, Videlot-Ackermann C, Terazzi E, Canard G (2015) Light Absorption and Hole-transport Properties of Copper Corroles: From Aggregates to a Liquid Crystal Mesophases. New J. Chem. 39, 7140-7146.¸ DOI: 10.1039/c5nj01268f; Jin H-G, Balaban MC, Chevallier-Michaud S, Righezza M, Balaban TS (2015) Biomimetic Self-assembling Acylphthalocyanines. Chem. Commun. 51, 11884-11887.; DOI: 10.1039/c5cc04602e; Gao D, Canard G, Giorgi M, Vanloot P, Balaban TS (2014) Electronic and Steric Effects of the Peripheral Substitution in Deca- and Undecaaryl Metallocorroles. Eur. J. Inorg. Chem. 279-287.; DOI: 10.1002/ejic.201301314).
As planned, we could obtain the targeted styrylpyridinium compound with a Fmoc protecting group and an OPfp activating ester group on the lysine fragment in excellent purity and in sufficient amount (Figs. WP2-4, WP2-8, WP1-9). This compound was sent to our partners at KIT that incorporated it into their peptide arrays (Fig. WP1-9). The functionalization proved that the initial idea was correct as this fluorophore could be efficiently coupled by the KIT techniques.
The goal of the PEPDIODE project to elongate by on-chip synthesis peptides at various positions with a chromophoric building block could be unequivocally proven with the targeted styrylpyridinium construct having an Fmoc protecting group and a pentafluorophenyl (OPfp) activating group which could be synthesized by AMU in Marseille in gram quantities: The modular strategy in which we could synthesize this building block enabled us to obtain similar styrylpyridinium salts with a tuneable absorption and fluorescence wavelength which are suitable in other applications. Thus, we could devise inhibitors for carbonic anhydrase having a brilliant fluorophore. This could have applications in precise imaging of malignant tumours in view of their resection. Similarly, we could attach via thiol linkers such styrylpyridinium salts to gold nanoparticles for direct current rectification. A patent application regarding such compounds is currently under consideration. These subsequent developments of the initial compounds developed for the PEPDIODE project were pursued after the EU-funding in Marseille was terminated with the help of a locally funded project by the A*MIDEX initiative. The PhD student involved, Mr. F-X Dang will defend his thesis in December 2015.

Conclusions WP2:
The objective of WP2 was to synthesize artificial amino acid building blocks (or click partners) that could be incorporated at will and in 0,35 nm steps into individual peptides, and, moreover, in high density array format. This overarching objective has been achieved for a whole panel of artificial amino acid building blocks, albeit in less quantity and without C-terminal chemical activation (which is not needed any more when employing, e.g. the cLIFT method to synthesize high density peptide arrays). In the near future AMU and KIT will systematically explore the influence of amino acid side chains (natural ones and artificial ones) on the spectroscopic properties of built-in chromophores. As planned originally AMU could synthesize a whole panel of porphyrin and styrylpyridinium based compounds that are easily integrated into peptides either directly during synthesis (i.e. as an artificial amino acid derivative), or by click chemistry or similar (e.g. using orthogonally deprotected cysteine in combination with a maleiminido function). The next step will be to analyse many different peptides with integrated specialized building blocks in their optical properties with a specialized scanner to find a whole panel of peptides with fine-tuned redox potential..
We expect from experiments of this kind a much deeper understanding and we think that such a finding should be publishable in a high ranking journal, especially if we could indeed pinpoint those amino acids within the peptide moiety that are responsible for observed subtle changes in the optical properties (and in the redox potential). This will be done by systematically varying every amino acid position in eventually found interesting peptide candidates.


Work progress WP3:
The final goal of WP3 was to supply to the consortium a specialized CMOS-based chip that could measure fast and very sensitive the current / voltage characteristics of ~10,000 peptides that are linked to as many gold covered pixel-electrodes of the chip, and on an area of one cm(+2). Within the first two reporting periods, IMS successfully tackled two major problems we encountered when we manufactured previous versions of our measurement chip. Within deliverable D31 we succeeded to implement a manufacturing process that resulted into flat and very smooth gold surfaces on individual pixel-electrodes, with good electrical contact to the CMOS part, while withstanding aggressive chemicals used in the peptide processing (Fig. WP3-2). These surfaces should easily accommodate self-assembled monolayers of alkane thiols and peptides. Moreover, and given the fact that gold is not tolerated in CMOS processing lines, this achievement from the 1st reporting period is certainly a major highlight. A second highlight was the fabrication of Measurement Chip 1, which is a major step in the design and fabrication of a universal vehicle for highly parallelized assays. With the implementation of an active current integrator, a very flexible approach was taken that should bring us one step closer to our goal, namely using the measuring chip as a universal tool to readout electric properties of different nanoscaled molecules on a small pixel area. However, the next version of our screening chip showed quite high leakage currents, which forced us to redesign the chip (please note, that chips of this type are not used in a liquid environment of electrolytes). Due to the delay in fabrication of Measurement Chip 1 (see D33) and the availability of powerful and easy to use microcontroller systems like the Arduino platform (www.arduino.cc) the rather inflexible approach of integrating the analogue-to-digital converters into the Measurement Chip 2 was given up. Instead a platform base on Arduino Mega 2560 consisting of a hardware part (shield) and software modules were implemented and tested. The inexpensive platform was duplicated so that all partners can use it with only a standard Windows/Linux/MacOsX PC as a control station

Deliverable D31 and milestone MS31 Passive chip containing gold pads and appropriate electrode for peptide transfer: To briefly summarize:
a) IMS successfully implemented a lift-off process to structure the gold electrodes. This was a major achievement, given the fact that gold is usually not tolerated in CMOS processing lines. The gold surfaces are needed to couple peptides to the chip’s electrodes via SH- chemistry.
b) IMS verified electrical functioning of this - still quite primitive – chip version.
c) IMS and KIT checked the chip for stability towards different chemicals.

Deliverable D32 and milestone MS32 Design of 1st measurement chip: To briefly summarize:
MeasurementChip1 was designed to comprise 10,000 individually addressable gold electrodes (also called pixels) with an internal circuit that records the current flowing in/out the electrode. This recorded current is then read out of the pixel cells and fed into an amplifier and following analogue to digital conversion. In principle, Measurement Chip1 can be seen as a specialized image sensor with 100 x 100 pixels, where the analogue of the light sensing element is the peptide deposited on the electrode.

Deliverable D33 and milestone MS33 Prototypes of 1st measurement chip: Deliverable report D33 has been accepted in the 1st reporting period. To briefly summarize:
Measurement Chip1 was fabricated in the IMS CMOS line. Due to a combination of layout problem and stretching the processing limits of the gold layer, the first process lot showed a massive short between the power lines that prevented the chip from functioning. After extensive analysis the above mentioned problems could be identified. With a Laser Cutting process the short could be repaired on some selected chips. A second process lot consisting of four wafers each with 88 Measurement Chip1 was processed with a corrected layout allowing enough space for process fluctuations.
Testing at KIT: Due to the delayed availability of Measurement Chip1 KIT performed first cyclovoltametric tests with Test chip version 2. KIT could address single pixel electrodes of 100 x 100 µm2, and read cyclovoltagrams with single electrodes (see deliverable report D52), which is quite an achievement. However, the pixel electrodes showed an internal leakage current that severely hampered the obtainable sensitivity. Due to the different architecture of Measurement Chip1 this problem is solved and very low leakage currents (see D34) were shown at IMS.

Deliverable D34 Prototypes of 2nd measurement chip: Measurement Chip1 from D33 integrates the electrode charge into a voltage, amplifies it and makes it available to be measured with a simple DVM or a discrete A/D converter. Measurement Chip2 from D34 was planned to add dedicated high precision, parallel operating analogue to digital converters on the chip. Due to the delay in fabrication of Measurement Chip1 (see D32) and the availability of powerful and easy to use microcontroller systems like the Arduino platform the rather unflexible approach of integrating the analogue-to-digital converters into the Measurement Chip2 was given up before the final layout of Measurement Chip2 was started. Instead a platform based on Arduino Mega 2560 consisting of the controller boards, a custom hardware part (shield) and software modules were implemented and tested. The inexpensive platform was duplicated so that all partners can use it with only a standard PC as a control station. Deliverable report D34 gives a detailed description of the implemented concept. Figure WP3-5 shows the setup with the custom shield on top of the Arduino board connected to the Measurement Chip1 testboard.

Deliverable D35 Series quantities of 2nd measurement chip:
Due to the change of concept in D34, Measurement Chip1 was produced in series quantities to perform routine screening measurements. Figure WP3-5 shows the result of the wafer test (so called wafer map) on one of the processed wafers. About 50% of all chips on the wafers were functional and can be used for further investigations. For IMS purposes about 10 Measurement Chip1 were packaged onto the board shown in fig. WP3-4, left.

Conclusions WP3:
Most of the deliverables and milestones due in WP3 within the reporting period have been achieved, but the manufacturing of the latest version of our screening chip was delayed (largely due to building activities within IMS that generated a new cleanroom). IMS developed and characterized a CMOS compatible add-on process that yields flat and smooth gold surfaces with good electrical contact to the CMOS part (D31), while withstanding aggressive chemicals (e.g. used during peptide synthesis). In the final version of its screening chip, IMS integrated an active current integrator within the chip that allows for a very flexible readout of the parallelized cyclovoltametric measurements (D32) avoiding the intolerable leakage currents of Test Chip2. Near future work will test this feature again, as already shown in deliverable report D52.


Work progress WP4:
The goal of WP4 was to find protein scaffolds and peptide sequences that could be used to “fix” chromophores and redox centres at defined distances from each other. These could be used beyond WP4, to self-assemble artificial RCs, LHCs, and peptide based diodes at defined sub-nanoscale distances from each other, bringing us nearer to our goal of cheap self-assembled, modular solar cells with most parts of the modules being built in E. coli. In a first step towards this final goal, MIGAL analysed different de novo designed pigment complexes in their excitation energy transfer behaviour. Moreover, MIGAL was able to assemble hydrophobic native Chls and BChls with water-soluble proteins using water-in-oil emulsion, which we consider to be a major achievement (Fig. WP4-1b; Bednarczyk, D., Takahashi, S., Satoh, H., & Noy, D. (2015). Biochimica Et Biophysica Acta Bioenergetics, 1847(3), 307–313. http://doi.org/10.1016/j.bbabio.2014.12.003) The role of charge-transfer states in energy transfer and dissipation within natural and artificial bacteriochlorophyll-proteins. (Wahadoszamen, M., Margalit, I., Ara, A. M., van Grondelle, R., & Noy, D. (2014). Nature Comm. 5(5287), 1–8. http://doi.org/10.1038/ncomms6287). In yet another approach, six new fusion proteins that were expressed in E. coli showed significant exciton energy transfer between modified natural LH proteins phycobiliproteins and de novo designed Chl and heme binding proteins. In parallel, CUNY used their expertise to find small proteins that bind to chlorophyll dimers in very high affinity, which might be a first step to an artificial RC with a special chromophor pair.

Deliverable D41 A de novo designed protein cofactor complex with RC functionality: We aimed at producing proteins with three or more closely spaced redox cofactors, of which at least one is a photoactive chlorophyll (Chl) or porphyrin derivative. A “performance/research indicator” for this part would be inducing a multi-step electron transfer process along the protein-bound redox chain by photoexcitation of the Chl/porphyrin derivative thereby leading to charge separation that will be stable for a few microseconds or longer.
Prototype RC protein: MIGAL chose the artificial four-helix bundle protein HP7 as a starting point, where we could show that this very small protein is capable of binding Chl and heme derivatives (Cohen et al. 2011 J Am Chem Soc). We could show that the recombinantly expressed HP7 molecule could assemble three zinc-bacteriochlorophyllide (ZnBC) molecules, which is a first prerequisite when the goal is to design a self-assembled path of redox centres the electrons could travel along. Interestingly, when analysing spectra of these molecules, we got strong hints that two of these ZnBC dimer in HP7 accumulate significant charge transfer character upon excitation (Fig. WP4-1). This implies that in principle this dimer could serve as the primary electron donor analogous to the special pair in photosynthetic RCs. A manuscript that describes this work was publicatished (Wahadoszamen M et al., 2014).
Problems: So far, we found no evidence for charge transfer to the monomeric ZnBC in HP7. The most probable reason for this is that the cofactor is located too far away from the ZnBC dimer, which slows down electron transport to the monomer such that it cannot compete with the very fast charge recombination to the ground state that occurs in the dimer.

Deliverable D42 A de novo designed RC-like protein with a high-affinity binding site to a peptide-diode building block: We aimed at producing proteins with three or more closely spaced redox cofactors, of which at least one is a photoactive Chl or porphyrin derivative. A “performance/research indicator” for this part would be inducing a multi-step electron transfer process along the protein-bound redox chain by photoexcitation of the Chl/porphyrin derivative thereby leading to charge separation that will be stable for a few microseconds or longer. This will provide deliverable D41, namely a de novo designed minimal functional RC analogue.
Protein scaffold selection: Based on our previous success in three- and four-helix bundle design (Reedy and Gibney, 2004, Chem Rev.; Westerlund et al., 2008, Proteins) and mutli-cofactor protein design (Gibney et al., 1996, PNAS; Gibney et al., 2000, Biochemistry) as well as the functional RC protein design specifications set forth by Nature, we felt a four-helix bundle architecture was the ideal RC protein scaffold. We expressed the small single-chain four-helix bundle scaffold of cytochrome b562 in E. coli and attached a heme redox cofactor through self-assembly to it.
Heme binding of cytochrome b562: Next, we measured heme affinity of cytochrome b562 in both the ferric and ferrous states by developing a new method (see deliverable report D42). We determined a ferric heme affinity of 1 x 1010 M-1 (-13.6 kcal/mol) for cytochrome b562, and a ferrous heme affinity of 2 x 1013 M-1 (-18.1 kcal/mol) for cytochrome b562. These binding constant values indicate a very strong affinity for natural heme (Fig. WP4-2), which will be published in the near future. We consider this measured high binding affinity a major achievement because thereby we should be able to dock a variant of the cytochrome b562 protein to a synthetic heme moiety that is tethered to the pepdiode building block.
Problems: We also produced the M7C and M7Y methionine ligand mutants of cytochrome b562 in order to provide the heme iron an anionic ligand which should stabilize the ferric heme binding and lower the heme reduction potential. Both proteins were expressed, purified and subjected to biophysical studies. However, heme binding studied demonstrated that M7C bound ferric heme well, but lacked affinity for ferrous heme. Conversely, M7Y bound ferrous heme but not ferric heme. These mutants proved unsuitable for the purpose of RC design as oxidation/reduction of the bound heme would result in heme dissociation.
Near Future works: We will be testing the binding of the synthetic heme provided by partner 2 (AMU) to evaluate the ability of the cytochrome b562 scaffold to bind to peptide-tethered heme. Please note, that the same synthetic heme could also be incorporated into peptides in the array format. These studies will lay the ground work for screening for peptide diodes using the screening chip in D14. We will also be working with partner 3 (Migal) to fuse the cytochrome b562 scaffold to the WSCP light-harvesting protein for the purposes of generating functional RC designs.

Deliverable D43 A de novo designed or modified natural protein cofactor complex with LHC functionality, namely, high absorption cross section and long lived excited state lifetime: We used the chlorophyll binding protein of brasicacea (WSCP) for constructing a light-harvesting unit of the PEPDIODE cell (Bednarczyk et al., manuscript in preparation). WSCP is a small (~ 20 KDa) protein that can self-assemble with Chls into tetrameric complexes. The challenge in producing WSCP-Chl complexes in vitro is to get the highly hydrophobic native Chls into the aqueous environment of WSCP. MIGAL was able to overcome this challenge by developing a novel assembly method whereby aqueous solutions of the WSCP apo-protein are mixed with mineral oil and surfactants to form water-in-oil emulsions. When adding Chls to the emulsion these dissolve in the mineral oil phase from where they are eventually taken up by the WSCP within the aqueous microdroplets in the emulsions. By now we have produced four recombinant variants of WSCP based on sequences of natural WSCP from different plant species. All four variants have the same ratio of Chl to protein as indicated by their similar ratios of light absorption at 280 nm and 660 nm (Fig. WP4-3).
Moreover, spectra shown in Fig. WP4-3 indicate for all four variants strongly interacting four Chls within the protein complex as indicated by the typical excitonic features in the visible CD spectra. Most importantly, the strong excitonic interactions between the protein-bound Chls do not affect significantly their excited state lifetime. The fluorescence quantum yield of the Chls in WSCP is about 20% which is comparable to that of Chl in acetone (Fig. WP4-4). This implies that the fluorescence lifetime of Chls in WSCP is a few nanoseconds. Thus, the WSCP variants can be used as functional LHC modules. A manuscript describing this work was published (Bednarczyk et al., 2015).

Conclusions WP4:
Some of the objectives within WP4 have been only partially achieved, namely deliverables D41 and D42. We worked on some candidate reaction centres, but couldn’t prove that they indeed function as an artificial reaction centre. When we scrutinized our artificial four-helix bundle protein HP7 as detailed in deliverable report D41, we realized that this particular artificial protein cannot support a sufficiently stable charge separation in solution. In that sense we failed to achieve deliverable D41. However, the HP7-ZnBC complex still may serve as a promising start for an RC-like protein because it isolates a functional analogue of the special pair in natural RCs. However, we have been able to demonstrate tight binding of ferric and ferrous heme to the cytochrome b562 scaffold and modulation of the bound heme electrochemistry. The tight binding of heme to the scaffold, and the ability to alter its electrochemistry are necessary properties for achieving deliverable D42 and for screening for peptide diode behaviour on the screening chip.
Quite encouraging, we could indeed show the functionality of an artificial light harvesting complex (D43). Certainly a highlight was a new screening method that MIGAL developed and published in a high impact journal (Bednarczyk et al., 2015): They mixed aqueous solutions of the WSCP apo-protein with mineral oil and surfactants to form water-in-oil microemulsions. When adding Chls to the emulsion these dissolve in the mineral oil phase from where they are eventually taken up by the WSCP within the aqueous microdroplets in the emulsions. By now we have produced four recombinant variants of WSCP based on sequences of natural WSCP from different plant species. All four variants have the same ratio of Chl to protein as indicated by their similar ratios of light absorption at 280 nm and 660 nm (Fig. WP4-3). Two of them were suitable for use as light-harvesting modules in PEPDIODE. Particularly, our new reconstitution method makes it possible to add any protein extension or tag to the WSCP building block since the protein is prepared by recombinant DNA techniques and is assembled with Chls in-vitro.

Work progress WP5:
The overall aim of WP5 was to develop methods that should allow the PEPDIODE consortium to link the nano-world (many different nanoscale peptides) to the macro-world (e.g. a measuring chip that would report the electrical properties of many different peptides in parallel). Although a functioning measuring chip was not yet available from IMS (the first measuring chip showed too much internal leakage current, and the manufacturing of next versions of the measuring chip was delayed), we reached nearly all the goals from WP5: (a) KIT developed, published, and patented a method to transfer up to 10,000 peptides per cm(+2) to a rigid gold covered support (D53), (b) KIT could interface a first version of IMS’ measuring chip with a computer (D52), (c) KIT could readout in good sensitivity pixel-specific currents for at least one of the pixel electrodes from IMS’ first measuring chip (however, most pixel electrodes showed intolerable internal leaking current), and (d) KIT used a semi-theoretical and literature based approach to find a first molecular diode. However, we failed to achieve deliverable D56 (First peptide diode with antenna building block). Another, quite exciting development – that was not foreseen at the start of the PEPDIODE project – has been described in the context of WP1: KIT invented the cLIFT-method (Figs. WP1-2 to WP1-4) and the one-cavity-one-peptide method (Fig. WP1-8) that should enable us to synthesize millions of peptides with artificial catalytic centres (e.g. porphyrins from WP2), assemble them into protein-like structures, and then directly screen for reaction centres and light harvesting complexes (employing a scanner to readout energy transfer within the molecules). Therefore, KIT focused its work on advancing these techniques.

Deliverable D52 Read-out electronics and interface to computer system for measurement chip: Within WP5 the PEPDIODE consortium wanted to couple high density peptide arrays onto a measurement chip (see deliverable reports D31 to D34), and, thereby, link the nano-world (our different peptides) to the macro-world (a chip’s pixel electrodes that measures in parallel for 10,000 different peptides the number of transferred electrons). The aim of deliverable D52 was to provide and to test an interface between a computer system and a first measurement chip from IMS, which should allow us to address the chip’s electrodes individually. We achieved this goal (see deliverable report D52), however, first versions of our screening chip had intolerable leakage currents. Therefore, IMS designed and manufactured improved versions of the measurement chip (see deliverable D35). KIT will use them to measure I-V curves for individual pixel electrodes.

Deliverable D53 Coupling peptides to gold pads from measurement chip: In deliverable D53 KIT developed (1) experimental conditions that allow us to transfer peptides in the array format from a synthesis slide to a gold covered recipient surface, whereby the peptides are linked to the gold surface via Au-S bonds (from cysteine), and (2) KIT analysed the recipient slide by fluorescence and TOF-SIMS. The work done for this deliverable has been / will be published (Schirwitz et al., 2013, Advanced Materials; impact factor ~17; Muenster et al. 2015; PhD thesis Bastian Muenster 2014; PhD thesis Jacob Striffler 2014; a third manuscript is in preparation). However, although this procedure works in principle, it is still too tricky to do it routinely. Therefore, KIT will optimize the conditions of the peptide transfer employing ammonia- and UV-pre-cleaved peptides and transferring them to rigid surfaces. The incentive to do that is the commercial benefit that comes along with replicated arrays with – eventually – better quality (if we manage to enrich the “right” peptides during the transfer process; see deliverable reports D14 and D15). We know that this method works well on membranes (Schirwitz et al., 2013, Advanced Materials; impact factor ~17).
b) KIT will employ thinner gold films to avoid quenching problems when using fluorescence based detection methods.

Deliverable D54 Read-out peptide-specific current from measurement chip: Since a measuring chip without current leakage was not available in time from IMS, KIT explored the three different technical steps that are necessary to screen for peptide-based diodes. These experiments were explained in the 2nd periodic report and in deliverable reports D52, D53, and D55. Indeed:
i. We could transfer peptides in the array format onto flat gold surfaces with high precision (see D53).
ii. We implemented read-out electronics and an interface to the computer system for the first version of IMS’ measurement chip (see D52).
iii. We could read-out of the I-V characteristics of individual peptides and molecular diodes on a larger gold electrode (see D55).

Deliverable D55 First peptide diode: The aim of deliverable D55 was to provide a first molecular diode that could be used as a positive control when screening the measurement chip from IMS for improved / other peptide-based diodes. During work done within deliverable D52, KIT learned that current leakage occurred within the chip structures from IMS that obscured pixel-specific electron flux. Therefore, KIT decided to first use a semi-theoretical approach in the search of peptides with interesting conductivity behaviour. We therefore investigated within D55 the current-voltage behaviour of a few different peptide monolayers on larger-area gold surfaces. These are detailed in deliverable report D55.

Deliverable D56 First peptide diode with antenna building block: The aim of deliverable D56 was to provide a first peptide diode that is coupled to an antenna complex. We failed to achieve this deliverable.

Conclusions WP5:
Although a functioning measuring chip was not available in time from IMS, we reached nearly all the goals from WP5:
a.) KIT developed, published, and patented a method to transfer up to 10,000 peptides per cm(+2) to a rigid gold covered support (D53);
b.) KIT could interface a first version of IMS’ measuring chip with a computer (D52);
c.) KIT could readout in good sensitivity pixel-specific currents for at least one of the pixel electrodes from IMS’ first measuring chip (however, most pixel electrodes showed intolerable internal leaking current);
d.) KIT used a semi-theoretical and literature based approach to find a first molecular diode; but
e.) We failed to link this first molecular diode to an antenna complex; and
f.) In the future we will concentrate on our one-cavity-one-peptide-method to directly screen millions of protein-like structures for light harvesting complexes and reaction centres that, in the long run, might self-assemble to bio-inspired solar cells.
g.) All of these alternative screens will use AMU’s chromophores that are incorporated at many different discrete locations within the protein-like structures (oligomerized peptides). Thereby, we might be able to setup e.g. screening assays to identify those peptides that coordinate AMU’s chromophores at a distance of 1nm, and, thereby, serve as especially efficient artificial light harvesting complexes that transfer the energy harvested by a green chromophore to a red fluorophore that serves as a reporter.


Work progress WP6:
In this period we followed the previous MIGAL’s demonstration of different de novo designed pigment complexes that feature significant excitation energy transfer behaviour, i.e. behave as light harvesting complexes. In parallel, CUNY developed and characterized several relatively high potential heme binding redox proteins such as Cyt b562, together with a novel low-potential [4Fe-4S] protein based on a bacterial microcompartment protein. MIGAL and CUNY attempted to construct fusion proteins of Cyt b562 with WSCP and to assemble them with Chls. In addition, MIGAL attempted to fuse a phycobilisome linker protein to His-tagged HP7 in order to construct an anchor for assembling phycobiliprotein light-harvesting subunits. Both directions revealed interesting results but at the bottom line, did not lead to assembly of the desired RC-LHC complexes.

Deliverable D61 RC-LHC combinations selected for effectively LHC to RC excitation energy transfer in solution: Coupling an artificial reaction center (RC) to artificial light harvesting complex proteins (LHC) will provide a functional analog of a natural photosystem, which is the basic solar energy conversion unit in Nature’s photosystem. Within D61 we tested potential combinations of LHCs with RCs.
The first combination attempted was a fusion of WSCP and Cyt b562 as LHC and RC modules, respectively. WSCP can be reconstituted with Chls using water-in-oil emulsions to form homotetrameric protein complexes that bind four Chls in a compact arrangement that do not compromise their energy transfer properties. Cyt B562 is a monomeric four-helix bundle heme-binding protein that may provide a redox link to electrodes and may act as a RC. A fusion protein combining the amino acid sequences of natural Cyt b562 and WSCP from cauliflower was recombinantly expressed in E. coli and purified by nickel affinity chromatography (facilitated by including a six-histidine tag at the protein’s N-terminus). Absorption spectroscopy indicated that Cyt b562 domain in the fusion protein was in its holoprotein form containing a single heme. This is not unexpected based on the tight ferric and ferrous heme binding constants of cyt b562 measured at CUNY. Unfortunately, our attempts to assemble the fusion protein with Chls in order to form a LHC-RC type protein did not succeed. The protein aggregated and precipitated upon addition of Chls and we were unable to isolate a soluble complex containing both Chls and heme.
A second combination for LHC-RC module was based on linking phycobiliproteins to Chl and heme binding proteins. MIGAL has already demonstrated energy transfer between pigment in a modified phycobiliprotein (ApcE) fused to HP7. As an alternative approach, more suitable for assembly with onto electrode surfaces, MIGAL tested a construct whereby a His-tagged single chain version of HP7 was fused to the Lc linker protein of allophycocyanines. This resulted in a small protein that can be specifically and tightly bound to a surface through the His-tag moiety that is in close proximity to the Chl/heme binding sites of HP7. Upon binding to the surface, the Lc module is at the other end of the protein and should serve as a specific anchor for the allophycocynains (Apcs). We tested the concept by binding the HP7-Lc fusion protein to a Ni affinity column followed by introducing Apc subunits to the column. Although there were indications of specific binding of Apcs to the column via the HP7-Lc protein we were unable to isolate sufficient amounts of complexes for spectroscopic characterization. We assume that part of the problem is that the Apcs that were isolated from thermophilic cyanobacterium might contain the native Lc protein, which prevents assembly with the HP7-Lc protein.

Deliverable D62 Optimised peptide-diode/RC combination:
Light harvesting module (see D43): As described in deliverable D43, the design of two functional LHC analogs, the native chlorphyll protein WSCP, and the semi artificial ApcE(1-240) phycobiliprotein, has been achieved. Each of these modules has a high absorption cross section and a long-lived excited state required for efficient light-harvesting. Each of these LHC modules has been recombinantly expressed in E. coli which allows for the future fusion to the RC protein module (Cohen-Ofri et al., 2011, J Am Chem Soc; Iris Margali, PhD thesis, 2013; Bednarczyk et al., 2015, BBA-Bioenergetics; Reedy et al., 2008, Nucleic Acids Research).
Charge-separation and electron-transfer chain modules (see D41 and D42): Migal and CUNY have been working on RC modules capable of charge-separation and efficient electron transfer. MIGAL has demonstrated the potential of HP7 as a RC analogue although they were not yet able to demonstrate the required functionality. Working towards deliverable D61, MIGAL fused HP7 to ApcE(1-240) and recombinantly expressed it in E. coli while incorporating the pigment phycoerythrobilin (PE). The fluorescence emission of PE is at 590 nm which overlaps with the Qx absorption band of HP7-bound ZnBC and allows for energy transfer from PEB to ZnBC. While the ZnBC-HP7 module is not yet a functional RC analogue, energy transfer from the ApcE to the HP7 module is clearly observed.
Meanwhile, CUNY has been working on constructing an electron-transfer chain protein with the redox cofactors biased for efficient electron-transfer (Reedy et al., 2008, Nucleic Acids Research). The heme protein cytochrome b562 and two mutants with altered electrochemistry, M7K and M7H, have been produced in E. coli. Their electrochemical values span 250 mV (-91 mV to +166 mV vs. SHE) which provides sufficient driving force for electron-transfer. These form the basis for the construction of an electron-transfer chain with the proper bias for the RC module.
RC module-peptide diode coupling: CUNY has also been exploring the use of the cytochrome b562 scaffold as a module to couple the RC protein to the peptide-based diode on the screening chip surface. The affinity of cytochrome b562 for ferric and ferrous heme was measured to ensure that the heme affinity is adequate to bind to a porphyrin-tethered peptide on the chip surface. Equilibrium studies of heme binding to the cytochrome b562 scaffold in the ferric and ferrous oxidation states demonstrate strong binding affinity. This suggests that attachment of the cytochrome b562 scaffold to the peptide-diode on the chip surface is feasible. In addition, the M7K and M7H mutants of cyt b562 modulate the electron transfer driving force, and while their ferric and ferrous heme affinities are weaker than the wild-type, they may be sufficient for attachment to the chip surface and work is progressing on testing this.

Conclusions WP6:
We didn’t manage to construct and assemble optimized RC-LHC combinations, but we could show efficient energy transfer in a functional LHC analogue, modulation of heme electrochemistry required for efficient electron transfer, and sufficient heme affinity to suggest that binding of the RC module to a peptide-based diode is feasible. Even more important, MIGAL developed and published a novel screening method to find and optimize LHC building blocks (Bednarczyk D., Takahshi S., Satoh H., and Noy D. (2015) Biochim . Biophys Acta Bioenerg, 1847, 307–313; The role of charge-transfer states in energy transfer and dissipation within natural and artificial bacteriochlorophyll-proteins (Wahadoszamen M, Margalit I, Ara AM, van Grondelle R, Noy D. (2014) Nature Communications 5(5287), 1–8). Moreover, as mentioned within WP1, KIT has developed their one-cavity-one-peptide method, that – in the long run – should allow us to directly screen for novel artificial reaction centres.
Thereby, our work to date has poised us to be able to construct a feasible RC-like protein and attach it to the surface-bound peptides for PEPDIODE optimization. Construction of the peptide diodes and the assembly of the proteins on the electrodes for the purposes of measuring light-induced currents remains to be done. The peptide modules mentioned in the preceding paragraphs should provide the building blocks for modular solar cells constructed from artificial RCs, LHCs and peptide based diodes. These modules could then be used to add “docking sites” for artificial LHCs and RCs that were described in D41, D42, and D43, bringing us nearer to our goal of self-assembled, modular solar cells made of artificial RCs, LHCs, and peptide based diodes.

Work progress WP7:
We found artificial light harvesting complexes, and we worked on artificial reaction centres (with encouraging results), but we failed to assemble them into functioning RC-LHC modules. Therefore, we didn’t work on the proof of concept biomimetic solar cell as originally envisioned for WP7.
Potential Impact:
We could considerably advance the completely new concept of a self-assembled solar cell, but the idea as such is certainly still in its infancy stage. The idea to (1) generate many different LHCs and RCs and screen them in an evolutionary approach for better performance, and (2) do it again with self-assembled RC-LHC modules, was and still is still valid, especially if (3) we could rely in the end on cheap recombinantly produced protein-based building blocks.

Although we failed to achieve the proof of principle of such a bio-inspired solar cell – as originally planned in WP7 – work during the PEPDIODE project resulted in several achievements that were patented or published (some of them in excellent journals) that should have an impact when standing alone:

(WP1): KIT and PPP advanced their solid-material-based peptide array synthesis method much further than anticipated when we started the PEPDIODE project. Moreover, please note, the state of the art in peptide array synthesis used to be the SPOT technology with its 25 peptides per cm(+2) (Frank, R. 1991, Tetrahedron 48, 9217–9232), and at chemical costs of roughly 1€ per peptide (www.jpt.com). Mainly KIT developed a fully automated synthesis robot (Maerkle F, Nesterov-Müller A, Breitling F, Löffler F, Schillo S, Bykovskaya V, von Bojnicic-Kninski C, Leibe K. Method for combinatorial particle manipulation for producing high-density molecular arrays, particularly peptide arrays, and molecular arrays which can be obtained therefrom. patent application PCT/EP2013/001141, EP2013/001141) that will soon (presumably in 2016) be commercialized within PPP for the synthesis of very high density peptide arrays. Currently this robot can synthesize some 100.000 peptides per glass slide at a density of 10.000 peptides per cm(+2), and in better quality when compared to the peptide laser printer, which would bring down the chemical cost per synthesized peptide to some 0,1 cent (vs. 1€ for the SPOT synthesis and 1 cent for the “peptide laser printer”). Moreover, by using a thinner light absorbing layer in our donor slides (currently we use a roughly 80µm thick capton foil) it should be straightforward to decrease the size of synthesized peptide spots at least another 10-fold (data not shown), which would bring the cost per synthesized peptide down to 0,01 cent per peptide. All we need for that is to work in a dust free environment. We expect that within the next three years mainly customers from the life sciences will order such arrays (from SME PEPperPRINT).
One especially interesting feature of the cLIFT procedure was completely unexpected: The cLIFT method very reliably transfers extremely thin material spots (depending on the laser power and duration 1 to 50 nm thin). This feature means that two or more material spots can be positioned on top of each other and, it is common sense to assume that – if melted – these thin material layers would nearly perfectly mix, and, e.g. thereby activate a non-activated building block to couple to the surface (it works for non-activated amino acid building blocks; see Fig. WP1-7). If combined with a sensitive method to readout the result of this coupling step, e.g. by synthesizing fluorescent molecules or by mass spectrometry imaging (www.uni-giessen.de/cms/faculties/f08/departments/iaac/spengler/research/projects/imaging-mass-spectrometry) this method might open a completely new field of research in synthetic chemistry. The reason is simple: the cLIFT procedure now allows us to structure a glass slide with >100.000 spots (= different sites of chemical synthesis), and it certainly will allow us in the future to further increase the number of discrete synthesis sites to >>1 million per glass slide. These boundary conditions should allow us to use a variant of our solid material synthetic approach to explore millions of different synthetic routes by employing a few hundred different donor slides with different building blocks, catalysts, bases, acids, scavengers etc., each one embedded into a matrix material that is suitable to solubilize its embedded material if melted (certainly, we will need to explore different kinds of matrix materials, e.g. hydrophobic, amphiphilic, or hydrophilic to solubilize different kinds of building blocks). The success of these millions of different synthesis routes could then be monitored, e.g. by mass spectrometry imaging (we have an agreement with the Spengler group to cooperate; see web link from above). This output compares to a typical PhD thesis in organic chemistry with its 20 – 100 different synthesized molecules. The impact of such a procedure on science and on the society might be immense.

(WP2): When we started the PEPDIODE project, AMU planned to synthesize several artificial amino acid building blocks (e.g. porphyrins, fluorophores) in gram amounts and with C-terminal activation (OPfP ester or similar) to manufacture with those building blocks the correspondent “amino acid particles”, that are finally printed with the “peptide laser printer” to synthesize peptide arrays. However, KIT’ cLIFT method only requires mg-amounts of these artificial building blocks (vs. gram amounts for the peptide laser printer), and, in addition, we don’t need to activate them anymore, which proved to be difficult for some chemicals (Fig. WP1-7). The scientific impact of such a method might be immense, soon KIT and AMU will synthesize many different peptides in array format to ask the question which amino acid side chains (e.g. artificial ferrocene amino acids) influence the spectroscopic behaviour of a fluorescent moiety that is also build into the peptides. In the short term, we expect an even larger impact in the life sciences, e.g. by synthesizing arrays with post-translationally modified amino acids (many of these are either commercially available in non-activated form, or synthesized from AMU) that might lead to the identification of the antigens that are targeted by autoantibodies in different autoimmune diseases (e.g. rheuma sera stain citrulline containing peptides; sera from multiple sclerosis patients seem to recognize glycopeptides, but nobody can synthesize them in array format). Experiments of this kind will certainly lead to a commercial impact, with the commercialisation either done by SME PEPperPRINT, or by a newly founded company that mainly KIT intends to spin-off in 2016 (mainly involved PhD students Daniela Althuon (chemist), Roman Popov (chemical engineer), and Clemens von Bojnicic-Kninski (physicist) will run this spin off company.

(WP3): When we started the PEPDIODE project, we planned to use a screening chip – designed and produced by IMS – to readout the electrical properties of many different peptides in parallel. Although it proved to be difficult to develop a chip with minimized leakage current (the ancestor chips are sealed and used in insulating dry environment, while we bathe the chips in an electrolyte solution), and although it proved to be difficult to link the peptides in the array format to its tiny pixels (and embed them into an insulating SAM layer), we think that a chip that allows for highly parallelized voltage / current measurements is an achievement in its own that might find its industrial applications. Certainly, IMS will design and manufacture further chip generations to come closer to the goal we envisioned.

(WP4-7): Within the PEPDIODE project, we strived to develop screening procedures that would result into the identification of suitable building blocks (RCs and LHCs) that – in the long run – could self-assemble to a bioinspired solar cell. Although we didn’t reach this final and very ambitious goal, the consortium considerably advanced the concept of screening for novel building blocks. Notably, MIGAL developed a novel general method for assembling hydrophobic pigments and redox cofactors such as natural chlorophylls (Chls) with water-soluble proteins using a template based on the water-soluble Chl protein (WSCP) from Brassicaceae, a natural high-affinity water-soluble Chl binding protein (Bednarczyk, D., Takahashi, S., Satoh, H., & Noy, D. (2015). Biochimica Et Biophysica Acta Bioenergetics, 1847(3), 307–313. http://doi.org/10.1016/j.bbabio.2014.12.003). Their approach allows us to screen for recombinantly expressed LHC variants that could then be tested for better performance, and moreover, they could do it in an evolutionary approach (i.e. vary initially found LHCs and screen for those with better performance). MIGAL and CUNY also developed novel protein-cofactor complexes that may provide the building blocks for artificial photosynthetic systems. Some of these complexes provided new and important insights into the role of charge-transfer states in energy transfer and dissipation within natural and artificial bacteriochlorophyll-proteins which was published in a high ranking journal (Wahadoszamen M, Margalit I, Ara AM, van Grondelle R, Noy D. (2014) Nature Communications 5(5287), 1–8.).
KIT developed and patented (PCT/EP2013/001141) not only the cLIFT procedure (see WP1), but also a novel method that uses microstructured glass slides (Patent application 102015117567.3: D. Althuon, F. Breitling, V. Bykovskaya, F.F. Loeffler, A. Nesterov-Mueller, R. Popov, B. Ridder, C. v. Bojnicic-Kninski. „Ultra-hochdichte Oligomerarrays und Verfahren zu deren Herstellung“. Filed October 15, 2015. Patent application submitted; main inventor Dr. Alexander Nesterov-Müller). Similar to the method that was developed by MIGAL, one of the hallmarks of this novel procedure is that it allows for an evolution style screen that should help to improve the performance of found peptides. The method uses differently labelled amino acid particles with a very narrow size distribution that are spread over a glass slide that is microstructured with exactly defined cavities. This very simple procedure results into arrays with > 1 million peptides for just a few Euro per glass slide (just spread the particles over the surface and gently rub them; Fig. WP1-8). KIT achieved the proof of principle with this method at a density of 25 million peptides per cm(+2) (data not shown). KIT used its proof of principle experiments to patent this procedure, and to win the KIT internal UpCat competition that aims to train participants to spin out a company.

We think that the impact of such a simple and cheap synthesis procedure might be immense. It could be used for many applications in the life sciences:
(1) synthesize several millions of peptides with built-in post translationally modified amino acids, and screen autoimmune sera for disease correlated antibodies (should work without any previous knowledge as to the cause of the disease; if an antibody detects an antigen that needs a post translational modification – as in rheuma patients – we should find it due to the large number of synthesized peptides);
(2) synthesize on a UV cleavable linker several millions of peptides with built-in artificial circularization domains, deprotect the side chains, add a buffer / culture medium that induces circularization of soluble peptides, and, in addition, sustains the growth of reporter bacteria (one to three per cavity); then seal the cavities from each other and cleave the peptides by UV to screen for novel antibiotics (there are many circular peptides that are excellent antibiotics; the cavities allow us to circularize the peptides which doesn’t work when still linked to the surface);
(3) do a similar experimental approach, but this time add a protease together with a reporter peptide that starts to fluoresce when cleaved by the protease in order to screen for novel protease inhibitors (nearly all viruses need a virus-specific protease to infect a cell);

In the long run it could also be used to screen for protein-like structures with catalytic activities:
(4) synthesize on a UV cleavable linker several millions of peptides with built-in artificial electron relaying side chains (e.g. ferrocenes from WP2) and photon absorbing side chains (from AMU), and, eventually add at the N-terminal end a mixture of two different building blocks that lead to dimerization of peptides and at the same time reports this event by a fluorescence, deprotect the side chains, add a buffer that induces dimerization of soluble peptides, then seal the cavities from each other and cleave the peptides by UV to screen for dimerized (oligomerized) protein-like structures that are detected in a scanner by their fluorescent characteristics. Such an array of protein-like structures could then be “energized” by light (which is absorbed by the porphyrin), the energy could be transferred to an acceptor (similar to a reaction centre), and the output could be read with mass spectrometry imaging (e.g. detecting a reduction of a fatty acid to the corresponding aldehyde by photocatalysis).

List of Websites:
List of participants and contact details:

PEPDIODE website: www.PEPDIODE.eu

Coordinator: Dr. Frank Breitling (Karlsruhe Institute of Technology (KIT), Institute for Microstructure Technology (IMT), Germany; www.KIT.edu; email Frank.Breitling@KIT.edu);
Administration at KIT: Dipl. Ing. Sebastian Schillo (Sebastian.Schillo@KIT.edu) Mr. Berndt Kronimus (Berndt.Kronimus@KIT.edu); Ms. Natascha.Kindsvogel (Natascha.Kindsvogel@KIT.edu);

Artificial amino acids: Prof. Dr. Silviu Balaban, (Aix Marseille University, AMU), France; www.univ-amu.fr; email ts.balaban@univ-amu.fr

Peptide synthesis: Dr. Volker Stadler (SME PEPperPRINT (PPP), Germany; www.PEPperPRINT.com; email Volker.Stadler@PEPperPRINT.com) Dr. Ralf Bischoff (www.PEPperPRINT.com; and R.Bischoff@dkfz.de);

Light harvesting complexes: Dr. Dror Noy (former address: Weizmann Institute (WEIZMANN), Israel; old address: www.weizmann.atc.il; email dror.noy@weizmann.ac.il; new address: Migal - Galilee Technology Center (MIGAL), Israel www.migal.org.il/Dror-Noy; e-mail: drorn@migal.org.il);
Dr. Dror Noy moved his laboratory to Migal - Galilee Technology Center" as of 01.01.2013. The administrative procedure to transfer the funds for his tasks with PEPDIODE has been completed.

Artificial reaction centres: Prof. Dr. Brian Gibney (The City University of New York (CUNY), USA; www.cuny.edu; email bgibney@brooklyn.cuny.edu);
Screening chip: Dr. Harald Richter (Institut für Mikroelektronik, Stuttgart (IMS), Germany; www.ims-chips.de; email richter@ims-chips.de);