Poster: Aachen

M.A.R.S. - Magnetic ATP Recycling System

Salim Atakhanov • Kim Büttgen • Selma Busch • Daniel Costard • Maximilian Dreimann • Raphael Egging • Pia Ergezinger • Julia Gehrmann • Florian Kroh • Sven Leisen • Vladyslav Los • Lina Schmidt • Shlok Szatkowski • Marcel Wittmund

Project Supervisors
Prof. Dr. Lars Blank • Prof. Dr. Ulrich Schwaneberg • Prof. Dr. Wolfgang Wiechert

Project Advisors
Dr. Yu Ji • Volkan Besirlioglu

Poster Design
Dr. Verena Resch

About M.A.R.S.

The complex regeneration of biochemical energy sources represents a cost-intensive hurdle for many production and research processes. With M.A.R.S., we want to establish an innovative strategy to create light-powered, mitochondrion-like protocells and a bioreactor that will recycle those cells by magnetism. Through the design of our reusable recycling system it will be able to power every ATP-driven enzyme cascade, making M.A.R.S. universally applicable. By extracting Bacteriorhodopsin out of Halobacterium salinarum, a phototrophic archaea species, and combining it with an ATP synthase from Saccharomyces cerevisiae in self-produced polymersomes and liposomes, we get simple but effective chassis, which make it possible to cover the energy requirement of any enzyme reaction cascade. Binding those chassis to magnet particles via anchor peptides enables the reuse of the entire protocell system within the reactor by means of magnetic purification, whereby they can be fed directly into enzyme cascades, without depending on living cells.

The Problem
Cofactor-dependent enzymes are extremely useful and often absolutely necessary in catalyzing reactions like dehydrogenation [1], but the high cost and low availability is a limit for many industrial applications [2]. For example, enzyme-catalyzed carboxylate reductions pose a big challenge, presumably because they require expensive cofactors. A sustainable and efficient cofactor regeneration system would make production paths for many biochemical products cheaper and environmentally friendly [3]. Those products include important precursors for pharmaceuticals. The system would not only be influential for the pharmaceutical industry but also for others like the food industry.

There are many enzyme systems which use adenosine triphosphate as a cofactor, but there is no efficient ATP regeneration method yet. Acetyl phosphate, phosphoenolpyruvate and creatine phosphate are expensive phosphate donors used for conventional ATP regeneration systems, thus not ideal for industrial application [4]. A common method is using polyphosphate to regenerate ATP, but this can slow down some reactions [5]. What is missing is a method to regenerate ATP like cells do, without using expensive metabolites or lowering the efficiency of the main reaction.
The Idea
To generate ATP, we need the enzyme ATP synthase. In vivo it is located in the inner mitochondrial membrane in eukaryotes or the cytoplasmatic membrane of prokaryotes. The ATP synthase uses a proton gradient and converts an electrochemical potential into chemical energy. For the proton gradient, we chose Bacteriorhodopsin. Bacteriorhodopsin is a light-dependent ion channel and requires light energy as a power source. Integrating these two membrane proteins into a polymersome can generate ATP from ADP and phosphate, just by using light energy, which is the easiest and cleanest energy available.

After figuring out the basic design, we wanted to optimize it by making it reusable. Doing so was possible by connecting these powerplants with magnetic particles using anchor peptides. After the use of M.A.R.S, our system can be separated from the product and can be used again.

We need two anchor peptides and a spacer domain, forming one complex. The first anchor peptide binds to the polymersome; the second anchor peptide binds polystyrene coated magnetic particles. A spacer domain connects both anchor peptides, which enables the whole construct to link the ATP factories to the magnetic particle.

We also wanted to provide an easy to assemble and cheap photobioreactor, with the functionality to recycle the chassis by using magnets.

The Modeling
To support the work of our wet lab team, we relied heavily on the use of modeling techniques for our complex system. Especially in the times of limited laboratory access, we could greatly profit from creating and molecular dynamics simulations of components, experiments or the whole processes and stoichiometry of M.A.R.S.

We used two different approaches: Mathematical modeling and molecular dynamics simulation

In our mathematical model, we simulated the implementation of our complete system in the context of a real industrial application. We could show that a set-up of our concept exceeds the efficiency of existing ATP-Regeneration systems and can compete with the conversion rates of whole cell approaches while preventing any formation of secondary metabolites. We ran several in silico experiments with our mathematical model, compared thousands of different sets of parameters, to find the optimal parameters for M.A.R.S. like the mean chassis diameter or the initial CAR, ADP and Pi concentration.

With our extensive molecular dynamics simulations, we went much further than just rendering our models in simple solvents and gathering data plots with the standard GROMACS workflow. We assisted the wet lab team by creating valuable predictions in silico, like the integration behavior of our proteins into the different lipid batches. Furthermore, we used molecular dynamics to create comprehensive molecular movies of the components of M.A.R.S. like the main ATP Recycling-Mechanism of our chassis as well as the coupling to magnetic particles.

The Chassis
In our experiments, we decided to work with liposomes instead of polymersomes, because it is easier to perform a protein integration on membranes made of natural phospholipids. However, the structure and handling of polymersomes and liposomes isquite similar. However, polymersomes would be preferred in a final concept of M.A.R.S. as they are more stable than liposomes.

We choose a lipid mixture of POPC, DOTAP and DOPC to have slightly positively charged liposomes for better bacteriorhodopsin integration.

For the liposome formation, we decided to use the film rehydration technique as a proven method [6]. It consists of the rehydration of a dried lipid film with a buffer solution which creates small unilamellar vesicles. To make the liposomes easier to detect under the microscope, we want to form giant unilamellar vesicles. This was achieved by creating an oil-surfactant mix where an aqueous solution which contains the lipid batch is added.

The desired mixture of 30% DOTAP, 35% DOPC, and 35% POPC did not work; there were no liposomes visible. We then tried liposomes consisting of only POPC first. POPC worked with different buffers. DOTAP and DOPC also worked for themselves. After that, we tried a mixture of 80% POPC and 20% DOTAP to get closer to the desired positively charged composition, which also worked.

The Enzymes
For the implementation of our system, we selected Bacteriorhodopsin (BR) from Halobacterium salinarum and ATP synthase from Saccharomyces cerevisiae. H. salinarum is a halophilic marine obligate aerobic archaeon. When treated with an oxygen deficit and constant exposure of light, it produces large amounts of the so-called purple membrane, which contains Bacteriorhodopsin. This leads to a cell membrane, which contains up to 75% BR, so we decided to use the purified purple membrane due to its high BR concentration.

Bacteriorhodopsin ATP synthase

After cultivation, we used osmotic shock for cell lysis and ran five rounds of centrifugation at 50,000 g and resuspension, finishing with sucrose-density gradient centrifugation at 280,000 g for eight hours. After these purification steps, we aimed to have three different fractions of which fractions one and two should have been the ones with the highest purple membrane purity. Due to complications with the rotor of the ultracentrifuge, the last step was not possible to implement.

ATP synthase was produced by the commonly known „baker’s yeast“ Saccharomyces cerevisiae. First, we isolated mitochondria in which the ATP synthase is located. For a higher yield and easier purification, we decided to use a specific inhibitor protein for the ATP synthase called IF1. The IF1 protein inhibits the hydrolyzation of ATP by blocking the rotation of the F1-subunit in one direction [7]. IF1 was modified with a His-tag for easier purification and expressed in Escherichia coli.

Before continuing with further purification, we lysed the mitochondria, following it up with a gel-electrophoresis and a western blot, including antibody immunodetection, to ensure that we purified mitochondria.

After successful implementation, we wanted to continue with the inhibition of the ATP synthase and further purification with affinity chromatography. However, because of time issues, we were not able to complete this final step.

Bacteriorhodopsin and ATP synthase integrated into the membrane

Anchors and Binding
For the binding of the ATP factories to the magnetic particles, we use different antimicrobial peptides, so-called anchor peptides. We first test their binding abilities on our polystyrene coated magnetic particles to determine the best anchor peptide. We would then use this one as the magnet-binding domain in our binding complex. To evaluate the binding abilities, we first express different anchor peptides with an eGFP label. We then check the remaining fluorescence under a confocal microscope after several washing steps. To determine the polymersome or liposome binding domain, we proceed analogously. After finding the best binding peptides, we can express a biadhesive binding complex using both domains and a stiff linker domain to connect our ATP factories to our magnetic particles.

Higher fluorescence means more anchors are bound to the magnetic particle´s surface. Within the first test with the magnetic particles, we identified MBP-1 to have the best binding. Under the same conditions, MBP-1 had the highest fluorescence. eGFP worked as a negative control and showed no fluorescence, which means no binding.

eGFP before 2nd washing step eGFP after 2nd washing step

LCI before 2nd washing step LCI after 2nd washing step

MBP1 before 2nd washing step MBP1 after 2nd washing step

Cg-Def before 2nd washing step Cg-Def after 2nd washing step

TA2 before 2nd washing step TA2 after 2nd washing step

The tests with the liposomes were more difficult. There is no easy way of separating the liposomes from unbound anchors. We first tried centrifugation, which did not work because the liposomes broke. The next try was to use a gentler method for separation, dialysis. Due to the bigger size of the anchors, we tried a molecular cut off 50 kDa. The liposomes were stable this time, but the anchors did not get through the dialysis tube. Next tries should be performed with a molecular cut off 100 kDa and maybe even 300 kDa.

Because we could not get bigger dialysis tubes fast enough, we tried out our binding candidates with low concentrations. The aim was to see whether the liposomes or the whole background shows fluorescence first. With LCI the liposomes showed a fluorescence first. Therefore, more tests with LCI should be performed.

POPC with Tris 30mM buffer, POPC with Tris 30mM buffer,
0.2 µM eGFP, Brightfield mode 0.2 µM eGFP, eGFP mode

POPC with Tris 30mM buffer, POPC with Tris 30mM buffer,
0.25 µM LcI, eGFP mode 0.25 µM LcI, eGFP mode with zoom:

With these first results, we concluded that a possible biadhesive peptide would be a fusion protein of MBP-1 and LCI. Both anchor peptides are connected through a stiff spacer Domain Z. This construct is able to connect polystyrene coated magnetic particles to POPC liposomes. Further tests should be performed with different lipid compositions, especially with the desired slightly positively charged lipids. The whole fusion protein can then be expressed and binding of both complexes can be checked. This could be done through magnetical separation to see if the liposomes would also be separated.
The Bioreactor
First, we expect our protocells to work very well in a large variety of different bioreactors. As long as the process works within certain parameters, regarding light intensity, sheer stress and magnetic discharge processes, our MARS works with your process. Nevertheless, we envisioned a bioreactor, which works relatively autonomously with our protocells, but also gives you a platform to tailor it to your process needs.

We approached the design and construction of the bioreactor holistically. Therefore, we wanted to mitigate several pain points with current solutions, mainly the reliance of human labour between the tasks.

The main part of the reactor is the orbital shaker. It must mix the reactants reliably. Furthermore, it has to be easy to manufacture and assemble as well as be comparatively cheap. We used Erlenmeyer flasks to change as little of the laboratory parameters as possible. We designed a linker which transforms the rotation of the electric motor into an eccentric motion. The timing belt allows for the seamless translation of torque to the other shafts. Under each flask there is a compartment for specialized equipment. In our project, we use electric magnets to immobilize the magnetic particles at the bottom of the flask.

Orbital shaker

To automate the process of pumping out the products as much as possible, we designed an automatic arm with a mechanical guide for the tubes to drive them into the Erlenmeyer flasks. Therefore, the tubes do not interfere with the shaking motion.

Automatization arm

Lastly, we began to work on an IoT Platform, so that the user can decide on shaking intensity, duration, etc. and the process begins autonomously, automating the process as much as possible.
The Proof of Concept
Through our laboratory work and simulation approaches we could proof parts of our system and verified that our system as a whole works in a real industrial context. One crucial part which elevates our project from other cofactor regeneration systems is synergy of magnetic particles and our ATP factories. Therefore, a big part of our strategy consisted of proofing the possibility to connect magnetic particles to membrane like structures.

We found an anchor peptide that binds specifically and has a high affinity to the magnetic particles, that possess a polystyrene surface. The anchor peptide MBP-1 showed the most promising binding. While testing multiple other peptides, we found out that LCI is the equivalent to MBP-1 respectively to the liposomes. The investigated anchor peptides were fused to an eGFP Protein to offer a visual confirmation of possible binding abilities.

Magnet LCI + Liposome

The next step would be the expression of an entire protein, which features an additional linker between those two binding domains, to create a bridge between our ATP Factories and magnetic particles.

Apart from the proof of the different compounds of M.A.R.S. we wanted to come up with a specific example were our system can be used.

To not only show that our system might work in Vitro, we aimed to test our system in a real industrial environment. By collecting Data of a specific enzyme reaction, we simulated how M.A.R.S. would act in the context of the following application. The enzyme carboxylate reductase (CAR) needs NADPH and ATP as cofactors. It´s part of very important enzymatic processes and enzyme cascades to produce pharmaceutical precursors such as R-PAC. Additional applications where our system comes into play will be elaborated in the following abstracts of this poster.

As a first reaction of the specific cascade, on which we set a special focus on, the CAR synthetizes a benzaldehyde out of a carbon acid [8]. This substrate is present in second generation feedstock and therefore offers a sustainable alternative to the common production methods. In a series of experiments, we tested the normally used regeneration system, whole cells as cofactor regeneration approach and did some negative controls to prove the dependency on ATP of this specific enzyme. We could represent the activity of the enzyme under the different conditions in form of kinetic curves.

As a result, we saw less main product conversion with the whole cell system that with the normally used regeneration system. The whole cell approach tends to over reduce the wanted product to an alcohol, which compromises the efficiency of this method.

We simulated the enzymatic activity of the CAR in the reaction where our system provides the ATP as a cofactor. With our system cytotoxic products and products in high concentrations can be synthesised. M.A.R.S. enables a comparable activity after 2 hours to that of the whole cell approach, without production of unwanted secondary metabolites. In the long run, significantly higher product conversion can possibly be achieved.

M.A.R.S. enables the industry to produce any specific relevant chemical, without the ineffectiveness of whole cell approaches or the unsustainable implications of purely chemical regeneration systems.
Application - Fundamental Research
As the universal energy currency in every organism, ATP is crucial in every part of the body, from muscle to brain. To unravel the secrets of the body, we must first understand how its individual components work. This also includes the role of ATP. Looking at the brain, ATP plays an important role as a messenger and activator for pain receptors like rP2X2 [9]. By expressing these receptors in electrophysical experiments in Xenopus laevis oocytes, their behaviour can be studied, for example. The supply of the necessary ATP for these experiments is very expensive, and so far, there is no direct recycling possibility to reuse it.
Application - Pharmaceuticals
Nowadays, the production of pharmaceuticals is possible in a wide variety of ways. The use of enzyme cascades is just one of them, and many of these turnover reactions depend on ATP as a universal energy donor.

One of the many ATP-dependent enzyme families is Carboxylic Acid Reductases (CARs). CARs can be used to catalyze reactions by themselves, yet they are often part of more sophisticated enzyme cascades. Because of their broad spectrum of catalyzed reactions and their high stereoselectivity, they are an ideal tool to produce precursors or complicated pharmaceuticals that could not be easily synthesized before [10]. One example of these important precursors is cinnamyl alcohol, which can be produced from glucose by the combination of CARs with other reductases [11]. A more recent proposal is the usage of CARs to synthesize (S)-2-aminobutanol, an important chiral precursor for the pharmaceutical compound Ethambutol, which is an antibiotic for tuberculosis treatment [12] [13].

ATP’s potential as a vital compound in the production of pharmaceuticals is not purely secondary- there is much more to it.

However, the use of ATP in the pharmaceutical sector is not limited to its activity as a cofactor. The direct application of ATP, for example, in the treatment of mitochondrial disease, is just another of many examples of the relevance of ATP recycling systems in the industry [14].
Application - Food Industry
The enzyme family of the Carboxylic Acid Reductases (CARs) shows great potential. Among other things it has broad application in the synthesis of aldehyde components [15]. In the food industry, many aldehydes are relevant as flavor and fragrance compounds like vanillin. The annual production of synthetic vanillin by CARs was estimated to be as high as 16,000 metric tons in 2019 [16].
Application - Chemicals
In addition to the use of CARs in the pharmaceutical and food sector, they can be applied in various ways to produce classic bulk chemicals like alkanes or aliphatic alcohols. Many application pathways were proposed, especially for a chain length from six to 16 carbons for alcohols and seven to 15 carbons for alkanes [17]. These include the most popular and important aliphatic compounds in the whole chemical industry.

There is also a strong dependency on CARs for the production of fine chemicals, for example, in the agrochemical sector, where they find application in simple two-step reaction cascades to synthesize arbitrary secondary amines [18] [19]

Since, as mentioned above, CARs are ATP-dependent enzymes, secure ATP recycling is also essential in the field of chemical production.
References and Acknowledgements
[1] Hanson, R. L., Goldberg, S. L., Brzozowski, D. B., Tully, T. P., Cazzulino, D., Parker, W. L., ... & Patel, R. N. (2007). Preparation of an amino acid intermediate for the dipeptidyl peptidase IV inhibitor, saxagliptin, using a modified phenylalanine dehydrogenase. Advanced Synthesis & Catalysis, 349(8‐9), 1369-1378

[2] H. Keith Chenault, Ethan S. Simon and George M. Whitesides. Cofactor regeneration for enzyme-catalysed synthesis. (1988)

[3] Anna N. Khusnutdinova, Robert Flick, Ana Popovic, Greg Brown, Anatoli Tchigvintsev, Boguslaw Nocek, Kevin Correia, Jeong Chan Joo, Radhakrishnan Mahadevan and Alexander F. Yakunin. Exploring bacterial carboxylate reductases for the reduction of bifunctional carboxylic acids. (2007)

[4] Sol. M. Resnick and Alexander J. B. Zehnder. In Vitro ATP Regeneration from Polyphosphate and AMP by Polyphosphate: AMP Phosphotransferase and Adenylate Kinase from Acinetobacter johnsonii 210A. (2000)

[5] Huimin Zhao and Wilfred A. van der Donk. Regeneration of cofactors for use in biocatalysis. (2003)

[6] Ting F. Zhu, Itay Budin, and Jack W. Szostak. Preparation of fatty acid or phospholipid vesicles by thin-film rehydration. Methods Enzymol. 2013 ; 533: 267–274.

[7] Michael J. Runswick, John V. Bason, Martin G. Montgomery,Graham C. Robinson, Ian M. Fearnley and John E. Walker. The affinity purification andcharacterization of ATP synthasecomplexes from mitochondria (

[8] Strohmeier GA, Eiteljörg IC, Schwarz A, Winkler M. Enzymatic One-Step Reduction of Carboxylates to Aldehydes with Cell-Free Regeneration of ATP and NADPH. Chemistry. 2019;25(24):6119-6123

[9] Bongartz, E. V., Rettinger, J., Hausmann, J. (2010). Aminoglycoside block of P2X2 receptors heterologously expressed in Xenopus laevis oocytes; Purinergic Signalling. 6:393–403 DOI 10.1007/s11302-010-9204-9

[10] France, S.P., Hepworth, L.J., Turner, N.J., Flitsch, S.L. (2017). Constructing biocatalyticcascades: in vitro and in vivo approaches to de novo multi-enzyme pathways. ACSCatal. 7, 710–724. (

[11] Venkatesan, K., Srinivasan, K.V. (2008). Stereoselective synthesis of (S)-dapoxetine starting from trans-cinnamyl alcohol (08-3482EP). (

[12] Dobrikov, G.M., Valcheva, V., Stoilova-Disheva, M., Momekov, G., Tzvetkova, P., Chimov,A., Dimitrov, V. (2012). Synthesis and in vitro antimycobacterial activity of com-pounds derived from (R)- and (S)-2-amino-1-butanol – the crucial role of the con-figuration. Eur. J. Med. Chem. 48, 45–56.(

[13] Schwendenwein, D., Fiume, G., Weber, H., Rudroff, F., Winkler, M. (2016). Selective en-zymatic transformation to aldehydes in vivo by fungal carboxylate reductase from Neurospora crassa. Adv. Synth. Catal. 358, 3414–3421.(

[14] Sperl, W., Prokisch, H., Karall, D., Mayr, J.A., Freisinger, P. (2011). Mitochondriopathien. Monatsschr Kinderheilkd 159, 848. (

[15] France, S.P., Hepworth, L.J., Turner, N.J., Flitsch, S.L., 2017. Constructing biocatalyticcascades: in vitro and in vivo approaches to de novo multi-enzyme pathways. ACSCatal. 7, 710–724

[16] Banerjee, G., Chattopadhyay, P., 2019. Vanillin biotechnology: the perspectives and fu-ture. J. Sci. Food Agric. 99, 499–506.

[17] Akhtar, M.K., Dandapani, H., Thiel, K., Jones, P.R. (2015). Microbial production of 1-octanol: a naturally excreted biofuel with diesel-like properties. Metab. Eng.Commun. 2,1–5. (

[18] Ramsden, J.I., Heath, R.S., Derrington, S., Montgomery, S.L., Mangas-Sanchez, J.,Mulholland, K.R., Turner, N.J. (2019) Biocatalytic N-alkylation of amines using eitherprimary alcohols or carboxylic acids via reductive aminase cascades. J. Am. Chem.Soc. 141, 1201–1206. (

[19] Schaffer, S., Corthals, J., Wessel, M., Hennemann, H.-G., Haeger, H., Volland, M., Roos, M. (2017). Producing amines and diamines from a carboxylic acid or dicarboxylic acidor a monoester thereof. US Patent USO09725746 B2

We want to thank our PIs Prof. Dr. Lars Blank, Prof. Dr. Ulrich Schwaneberg and
Prof. Dr. Wolfgang Wiechert as well as our Team Instructor Dr. Yu Ji and our Team Advisor Volkan Besirlioglu for accompanying us through this project and for always being at our side with help and advice.

In the same way we want to thank Dr. Verena Resch for taking over a large part of the design of this poster.

Last but not least, we want to thank those who have supported us with financial help or by providing rooms and materials:

This project was funded by the "Hans Hermann Voss-Stiftung".

The Outlook
Finally, the ability to build flexible energy regeneration devices that can process a variety of substrates and convert them into suitable forms of energy will be invaluable in future. For instance, a supply of enough energy to stimulate the regrowth of human neurons damaged by injury or treatment the mitochondrial diseases depend on expensive ATP production, our cost-efficient ATP recycling system has an enormous potential to be helpful. It can be utilized in compound screening or disease modeling. Moreover, our system can also indirectly improve prevention medicine, diagnostics, and therapy in the third world countries as the cost-efficient M.A.R.S. can play a crucial role in pharmaceuticals production.

Perhaps, if we manage to let go of comforting familiarity of mitochondrion and embrace seemingly unlimited potential of prokaryotes-inspired energy regeneration, we might one day be able to solve some current energy depending problems and even to energize artificial life.