Skip to main content

Synthetic Biology for the production of functional peptides

Final Report Summary - SYNPEPTIDE (Synthetic Biology for the production of functional peptides)

Executive Summary:
Peptides are an immensely rich source of natural products whose bioactivities have been harvested by humankind over and over again in the past. Their richness in chemical structure and functionality makes them – in principle – ideal molecules to interact with biological functions inside the body or on or in microorganisms. Hence their potential as drug molecules, for example as receptor (ant)agonists or as antibiotics.
However, their similarity with proteins as a major class of biological polymers makes their use also very challenging, as humans as well as bacteria have developed a number of measures to deal with alien peptides in their interior or environment.
These two overarching factors are the background against which the SYNPEPTIDE project wanted to develop novel methods for developing peptide-based medicines. Two targets stood out: On the one hand the development of new antibiotics against the increasing number of clinical bacterial isolates that are resistant to the standard antibiotics in clinical treatment; on the other hand the use of peptides as agonists against G-coupled protein receptors (GPCRs).
In order to efficiently generate novel molecules that can be evaluated for biological function, we used ribosomally produced and posttranslationally modified peptides (RiPPs), specifically lantibiotics. As these molecules are gene-encoded, they are amenable to the well-established methods of DNA-synthesis and genetic engineering. As they are posttranslationally modified (in this case including intramolecular cyclizations by thioether bridge-formation) , for example to insert ring structures, they come typically with a stabilized secondary structure which increases affinity in the interaction with the biological target and stability against standard defense mechanisms (such as proteolytic digestion).
The downside of using such directed evolution schemes for the generation of large population of novel molecules, most of which are either non-functional or at least not improvements on the activity of the parent molecule. This requires the development of efficient methods to analyze th effect of such different molecules and then their rapid characterization afterwards.
In the SYNPEPTIDE project, we successfully developed such methods, on the one hand for the development of novel antibiotics, and on the other hand for the evaluation of lanthipeptides as agonists against GPCRs. We furthermore identified novel promising posttranslational modifications (PTMs) beyond thioether bridge formation that will help in the future to further functionalize peptides in screening approaches, and implemented the insertion of non-canonical amino acids (ncAAs) in standard peptide secretion hosts such as Lactococcus lactis. As a specific result, we will continue to pursue the development of 2 novel antimicrobial peptides.

Project Context and Objectives:
SYNPEPTIDE’s overriding objective is to contribute to the development of novel functional peptides for applications for the benefit of humankind. Specifically, we were looking for novel peptides for the following applications:

a) Novel antimicrobial leads
The problem of increasing resistance to clinically administered antibiotics has emerged as one of the major challenges to the public health system of the developed world. Specifically, strains such as methicillin-resistant Staphylococcus aureus (MRSA) or vancomycin resistant Enterococcus faecalis (VRE) and other strains from the so-called ESKAPE-list are threatening to become major problems in hospital settings and are responsible for a large number of deaths already, at least 50’000 annually in the US and Europe alone. We argue that novel variants of naturally produced peptides, such as the RiPPs lantibiotics and polyteonamides, could be a promising reservoir for novel leads for antibiotic development.

b) Novel, exceptional powerful leads for GPCR agonists
In the treatment of human disease, GPCR-related drugs occupy 25 – 30% of all molecules. In general, receptor agonists have greater therapeutic potential than antagonists. However the discovery of highly specific agonists is much more difficult, because antagonists just have to bind and block a site, whereas agonists not only have to bind to a specific site but also to transduce a signal. Peptides that contain modifications to increase their rigidity, such as lanthipeptides featuring intramolecular thioether bridges, are a new class of drugs with great potential for the discovery of specific agonists. The introduction in known therapeutic peptides of various intramolecular rings of different sizes and in different positions is feasible, although prediction of functionality of a specific rationally designed structured peptide is less easy. The activity of such peptides could be optimized by randomization of one or more cyclization positions within the peptide or even by screening “naïve” large combinatorial peptide spaces.

c) Novel biotechnological production pathways to vitamins
Industrial biotechnology is one of the most attractive options to increase the sustainability of the European chemical industry. In this context, SYNPEPTIDE aims to develop a novel production route to the essential vitamin biotin, which currently is still produced by chemical means. This involves developing a way to exploit the non-conventional biochemistry that is employed in the various posttranslational reactions, for example in lantibiotic production, as a means for biochemical pathway design.

In order to work towards its objectives, SYNPEPTIDE needs to address a number of conceptional challenges:

1) To generate novel leads for antimicrobials or drugs, we need to generate a large diversity of molecules and then screen through this population to identify the few interesting candidates. The generation of novel ribosomally produced peptides is easy, as the methods of DNA manipulation allow the introduction of a broad variety of sequence changes. However, the peptides that are required for the objectives mentioned above require secondary structure for activity, affinity and stability. Therefore, we need to do directed evolution in the context of the required PTMs. This is an entirely different proposition.

2) As a reference PTM, we focused on the formation of intramolecular thioether bridges (generating so-called lanthionine amino acid residues). In the group of the corresponding natural products, the lantibiotics, one can find a broad variety of peptides with antimicrobial function already. However, many other PTMs are known to contribute to important structural features, such as activity, secondary structure and stability. Such additional modifications include epimerizations, methylations, chemical modifications of standard amino acids residues, etc. Finally, the recent advances in the manipulation of the genetic code, leading to the expansion of the amino acid repertoire that is used for making proteins by non-canonical amino acids pose the opportunity to even further open the realm of innovative functionalization of peptide structures.

3) Such engineering campaigns benefit immensely from assays that allow the rapid evaluation of function of molecules in high-throughput mode (directed evolution). SYNPEPTIDE is to contribute to the development of such assays in various ways:
i) There are currently no assays available that allow the testing of antimicrobial activity of compounds in high-throughput mode.
ii) There are currently no strategies to do directed evolution of RiPPs – on the one hand, their requirement for PTM is considered as a limitation in the sequence space that is meaningfully accessible (i.e. where PTMs can still be introduced), on the other hand as there is – see above – no high throughput method to read out function.
iii) There is currently no strategy to assay the effect of lanthipeptides on GPCRs in high-throughput moode.
iv) While it is easy to implement a gain-of-function screen for biotin synthesis (complementation of a biotin auxotrophy), it is difficult to do so in a way that allows selecting for increasing concentrations of biotin, simply as any give cell needs only minute amounts of the molecule (a few dozen molecules per cell). Furthermore, rather than screening for biotin production, an event that can be expected at the end of route development, it is more opportune to implement assays that allow to improve single steps. As any proposed biosynthetic route to biotin will qua necessity invade uncharted territory, such assays need to be developed – in high-throughput format – from scratch.

Next to the different technological developments that are required to implement such assays, SYNPEPTIDE proposes a specific approach to the directed evolution of lanthipeptides. Specifically, we propose to shuffle entire sections of a molecule rather than exchange amino acids. Specifically, we argue that lanthipeptides, specifically lantibiotics, can be seen as an assembly of functional segments, equivalents of which can be found in many different natural lantibiotics and which therefore could be randomly recombined in order to generate novel molecules built from segments that have been proven functional in nature. Such segments coincide typically with – in the case of lantibiotics – intramolecular rings, so that one could argue we recombine structured segments.
Such an approach bears an obvious risk, which is that such molecules are no longer posttranslationally modified by the PTM machinery of any given host. We plan to deal with this risk by “modularizing” lanthipeptide synthesis. Here, a module is defined as a functionally autonomous collection of information that enables proper modification of a given peptide sequence. Usually, a module thus contains the peptide sequence that can be modified by a specific PTM machinery, the leader peptide that is required to successfully interact with the desired PTM enzyme, and any other information (e.g. incompatibility with another module) that is pertinent to design. This way, it might be possible to recruit different PTM machineries in one cell and exploit them to apply different modifications to a single peptide sequence.
Importantly, this opens the scope of modifications that can be introduced, from only thioether bridges to other modifications of the RiPPs world, provided we can provide all the information required for PTM.

From these considerations, we devised the following structure of work packages for SYNPEPTIDE:

WP1: PTM-modules
In order to broaden the accessible space of PTMs, we proposed to recruit a broad variety of thioether-forming PTM enzymes to the experimental approach. In addition, this collection of modules recruited from nature is to be complemented by modification from other RiPPs, specifically from the chemically extremely diverse group of polytheonamides. Ideally, this work package should allow introducing alternative PTMs in addition to thioether-bridge formation into a high-throughput screening approach for novel antimicrobial peptides.

WP2: NC-modules
In addition to PTMs, we proposed to additionally functionalize peptides by insertion of ncAAs that are amenable to click chemistry and thus can be functionalized in completely orthogonal ways once that have been synthesized and secreted into the medium. To achieve this, one needs to implement the insertion of ncAAs in the bacterium that is used for synthesizing the different lantibiotic variants. While insertion of ncAAs in general has become feasible over the last years, in particular with non-standard hosts this is far from trivial. We propose to explore different methods and then to technically integrate click-chemistry into the novel screening work-flows that we intend to develop (see below).

WP3: Combinatorial methods
In order to enable efficient screening through potentially very large combinatorial and evolutionary spaces, some prerequisites need to be implemented. First of all, methods need to be implemented that allow integrating the concept of the “design module” effectively into the molecular evolution process. Due to its unique possibilities of combining diversity and control on the DNA-sequence level, we propose to resort to chemical DNA synthesis and develop the appropriate methods to generate libraries that allow, concomitantly, recombination of modules and diversifying DNA-sequence outside the modules. As the modules are typically defined by short amino acid sequences, an efficient design strategy for such libraries is non-trivial and needs to be carefully investigated. Next, we will implement the technical requirements to assay variant peptides for antimicrobial and GPCR-agonistic activity.

WP4: Modular recombination and evolution
Once design modules have been carefully defined and technical prerequisites implemented, we plan to conduct a number of screening campaigns to actually screen for novel functional peptides. By using the information of modules assembled in WPs 1 and 2, we propose to DNA-synthesize molecular libraries that can be synthesized in suitable hosts and then be tested in suitable high-throughput assays.

WP5: PTMs for metabolic engineering
Next to their crucial role in implementing secondary structure in peptides, the PTM enzymes of lantibiotics represent an interesting source of non-conventional sulfur-chemistry This makes them potentially attractive for the construction of the thiophane ring in biotin. We therefore propose to investigate PTM enzymes recruited from the domain of lantibiotics in a new biosynthetic route to biotin.

Additional work packages of the project include the activities in dissemination and management.

Project Results:
In the course of SYNPEPTIDE, we could make critical contributions to a number of fields. Most importantly, we successfully conducted screening campaigns for the discovery of novel antimicrobial lantibiotic molecules. More specifically, we could demonstrate discovery of novel PTMs, the design of hybrid leader peptides that allow the utilization of multiple PTM machineries in one cell, the implementation of ncAAs in the non-conventional bacterial host Lactococcus lactis, the development of a novel high-throughput (HTP) assay to assess antimicrobial activity of secreted peptides (the “nanoFleming”), implemented a HTP assay for GPCR-agonistic activity, and established a variety of assays to identify novel enzymes for implementation of a novel biosynthetic route to biotin.

I) Screening for novel antimicrobial peptides

Fig. 1 illustrates the molecular design logic applied to one of the peptide libraries that we screened for novel peptide antimicrobials. The structure of a model lantibiotic, nisin, is dissected into different segments (Fig. 1A), for which we find equivalents in other natural lantibiotics (Fig. 1B). These segments largely coincide with intramolecular rings, and therefore one could conceptually approach the segment shuffling as a shuffling of secondary structure elements. Importantly, the ring formation is dependent on the proper PTM machinery (Fig. 1C) which contributes two factors: First, the formation of thioether bridges (lanthionine formation) is the result of the two PTM enzymes NisBC. The modified peptide is exported by the exporter NisT, and the concomitantly the leader peptide that directed the peptide translated from nisA in the first place to NisBC is cleaved off, this activating the peptide.
It is known that the NisBC PTM machinery possesses considerable promiscuity, and therefore it seemed prudent to use the host of the nis system, L. lactis, as the expression host for the shuffled library indicated in Fig. 1B.

Fig. 2 illustrates the workflow of the developed “nanoFleming” assay that was subsequently used for screening of the peptide library. Briefly, a cell containing a peptide candidate is incubated together with sensor cells in an alginate bead of nL volume. The cells are expanded into microcolonies (II) and the sensor cell starts to secrete (or not) the peptide candidate. The peptide is activated (lower panel) by externally added protease NisT. Depending on the activity of the candidate, the sensor cells are inhibited in their growth or not (III). This difference in sensor growth can be easily read out by flow cytometry (e.g. the library cells express a red fluorescent protein and the sensor cells are dyed with a suitable fluorescent dye).

A crucial step in this process is the library construction step. As the library is prepared by a recombination of short DNA segments, the synthesis methodology is non-trivial and required careful optimization of numerous steps. The principle of the method is shown in Fig. 3. Importantly, as shown in fig. 3C, we succeeded in ensuring that a large fraction of theoretically possible combinations is actually included in the library, and the required diversity of modules is actually achieved at all positions (Fig. 3D).

The results from the screening of the library in question (SYNPEPTIDE 1) are summarized in Fig. 4. In Fig 4A, a clear cut case of secretion of an active peptide is shown. Note the absence of fluorescence, reflecting the absence of stainable sensor cells, in one of the nL-reactors of the right panel. After analysis of the population of reactors in a specialized flow cytometer (Fig. 4B), the reactors were isolated, opened, the and library strain was isolated and the lantibiotic gene was sequenced. After removal of duplicates, the activity of the secreted peptides was confirmed, and then characterized (Fig. 4C). For this, strains were grown and the petide in the supernatant concentrated and treated with NisP protease. The amount of peptide was approximately quantified via MS of the released leader peptide and the activity was assessed by inhibition zone assays. Also, the diversity of modules at this stage was analyzed (Fig. 4D).
The large number of active peptides allowed at this stage to take a detailed look at the underlying logic of generating antimicrobial peptides (Fig. 5). Clearly, flexibility in molecular design increases from N- to C-terminus and we can assign segments that are allowed and forbidden at specific positions.
For a more detailed characterization, the genes of the lantibiotic variants were re-cloned into a vector that allowed expression of the gene fused to a leader peptide with a His6-tag. This allowed purification of the molecules via affinity chromatography and, after activation, a careful analysis of antimicrobial properties against a panel of relevant pathogenic bacteria (Fig. 6). Note the large number of active peptides, which already indicates that a) the method generates a large variety of active molecules and b) that we had to develop an entire set of novel downstream processing methods to deal with the large number of peptides. Importantly, the peptides show a variety of properties (Fig. 6A), which is also reflected in the identification of peptide variants which are more active against single representatives of a panel of clinically relevant Gram-positive bacteria (Fig. 6B), as well as against strains that possess increased resistance against the reference lantibiotic nisin, either because they carry the natural immunity system, (Fig. 6C) or because they carry a protease known to inactivate nisin (Fig. 6D). In summary, we successfully established a workflow that allows the large-scale generation of peptide variants and their HTP-analysis for antimicrobial activity. Furthermore, the segment-shuffling strategy allows isolation of lantibiotics with novel antimicrobial properties.

II) Novel PTMs
Careful analysis of the biosynthetic systems of different polytheonamides resulted in a variety of novel PTMs of potentially considerable value for the generation of novel peptide molecules. Please note that, as long as the determinants of molecular recognition are not fully clarified, we cannot yet use these novel PTMs in library design. Still, their biochemical novelty sets them apart from known peptides and makes them particularly interesting. Two prominent examples are detailed below. They go back to the discovery of a novel group of natural product, the polytheonamides. These are cytotoxic, highly posttranslationally modified pore forming RiPPs from the marine sponge Theonella swinhoei encoded in the poy gene cluster. A similarity search for specific enzymes indicated that some of the enzymes encoded in the original poy cluster are actually quite broadly distributed. One such enzyme, OspD from Oscillatoria sp. PCC6506, was particularly amenable to handling in vitro and found to introduce D-amino acids into the peptide scaffold shown below. A mentioned before, such epimerizations would in the longer term be highly useful in order to increase peptide stability for example in bodily fluids.

In fact, the epimerization is only one element of a much more complex sequence of PTMs, as shown in Fig. 8 for the original polytheonamide PTMs. A cluster of only 7 enzymes is responsible for introducing the approximately 50 modifications observed in polytheonamides, including eperimerization, C-methylation and hydroxylations. Here, we have the case of a variety of PTMs that are accessible via one PTM machinery and therefore probably also accessible via the same leader peptide. In the longer term, including such PTMs into library generation as discussed in the previous section would much increase accessible functional space.


III) Hybrid leader peptides

As part of our efforts to unite different PTM machineries in one cell, we undertook to generate hybrid leaders that would route peptides to two PTM machineries, specifically from the nis system and the osp system (epimerization, see section II). We used the well-characterized NisBC module for the introduction of lantibiotic modifications and epimerase OspD for introduction of D-amino acids into designed peptides. To address the feasibility of leader peptides to be recognized by different modification modules the set of six hybrid leaders was designed (Table 1). Leaders were designed in truncating manner to evaluate the space for substituting amino acids in the sequences. However, regions known for their requirement to be conserved from previous analyses (FNLD/FDLD and LAIKALKD) were retained during the designing process. Every leader in the set was fused to the natural core peptides (encoded by nisA, ospA, and ospD). In the majority of the constructs protease sites were introduced for convenient leader removal (Table 1).

All the peptides were expressed in E. coli co-synthesizing the corresponding modification enzymes module to select the most processed prepeptides. Leader peptides 1/2 and 11/12 (further referred to as H1 and H11) were selected for further studies as they showed the highest level of thioether fomation, dehydrations and epimerizations in the attached core peptides. In the next step a set of hybrid peptides, containing also hybrid core peptides fused to the hybrid leaders H1 and H11, was designed (Table 2).

All the peptides listed in Table 2 were co-expressed with lantibiotic modification NisBC and epimerizations introducing modules. The production and dehydration levels were the best for the H1N1O, H1N2O and H1N3O variants. Also epimerizations were observed within the core peptides. A summary of the data is presented in Table 3.

In summary, we could clearly show that single PTM enzymes from different systems specific for different leader peptides can act in response to rationally designed hybrid leader peptides and also chimeric substrates. The assumption that both, D-amino acids and lanthionines, can be introduced at once is highly possible as the data shows high promiscuity and substrate flexibilities for the employed modules. Currently, the expression platform for simultaneous module activity is in the final steps to be utilized for the combined introduction of both types of modifications in peptides of choice.


IV) Lantibiotics with non-canonical amino acids

The insertion of ncAAs into therapeutically relevant peptides holds much promise. Reasons for that include an increase in the therapeutic options available (e.g. photoactivation of previously “silent” molecules), an increase in chemical flexibility due to the integration of new chemical handles to insert additional functions influencing optical properties or targeting or life-time properties, or even an improvement of the desired biological function (McKay & Finn, Chem. Biol. 21 (2014):1075, Kuriakose et al., Angew Chemie Intl Ed 52 (2013):9664). Therefore, we considered integration of ncAAs into novel antimicrobial peptides an important opportunity to explore in SYNPEPTIDE. Unfortunately, we had to learn that the use ncAA-insertion in non-standard expression hosts can be non-trivial. In the course of SYNPEPTIDE, we tried different methods in the bacterium that we had selected as the expression host for our SYNPEPTIDE libraries L lactis (see section I). Specifically, we investigated the use of a force feeding method and of STOP-codon suppression (see Fig. 9) in order to position ncAAs in the reference lantibiotic nisin, which we had selected as a reference. Even though some promising preliminary results could be obtained, we had to learn that as a rule the effect of such ncAA insertion schemes had drastic effects on the production level of the resulting peptides, to an extent that the system could no longer be used for screening purposes. This general observation remained true despite extensive efforts in optimizing various aspects of the system, such as details of the genetic set-up as well as the position of the stop codon that was to be suppressed, which in the end allowed us to confirm successful production of nisin containing boc-lysine (Fig. 10).

We therefore focused on using E. coli as a host. This has the disadvantage that every assay becomes more complicated as the produced peptides are not secreted and therefore need to be liberated by perforating the cell envelope. As a result, production levels are also low in most cases. However, the molecular tools for ncAA insertions are much better developed and therefore allow more flexibility in terms of insertions of different ncAAs.

In E. coli, we could indeed insert via SPI a number of ncAAs that are interesting from a pharmaceutical point of view (Fig. 11). We targeted the proline residue in the core peptide core peptide for replacement, which is crucial for activity (XXX Rink etal., 2007). Besides structure of the pre- and/or propeptide and ultimately antimicrobial activity, replacement by proline analogs should affect the rigidity of ring B. The corresponding peptides were produced in a strain auxotrophic for proline and co-synthesizing NisBC. Evidently, growth inhibition could be observed for the peptides produced with most of the chosen proline analogs (Fig. 11). Low ncAA toxicity is evident from the production strain reaching high culture densities. Without ncAA addition, E. coli cell densities stay low after induction and no activity can be observed (data not shown). With NisB and NisC activity evident from this assay, it should be noted that both enzymes carry 23 and 5 proline residues, respectively, which become likewise modified during SPI. For the two most active compounds, trans-4-hydroxyproline and trans-4-fluoroproline, peptides were affinity-purified. ESI-MS analysis confirmed ncAA installation. As before (Shietal., 2011), multiple dehydration extents are observed. Consequently, our data show that recombinant production of ncAA-modified bioactive nisin is feasible.


V) An agonist screen for lanthipeptides

Goal of this project within SYNPEPTIDE was to develop a screening system for lanthipeptide GPCR agonists. This system would be composed of on the one hand a lanthipeptide bacterial display system and on the other hand a eukaryotic GPCR expressing reporter cell (Fig. 12). The concept involved lanthipeptide display system induced GPCR activation leading to a fluorescent reporter cell which then could be sorted by a FACS. It was hypothesized that the activating display system would remain attached to the reporter cell, thus allowing identification of the activating displayed lanthipeptide via the encoding DNA present within the display system in this case the bacterial lanthipeptide display system. The development of the bacterial lanthipeptide display system (Bosma et al. Appl Environm Microbiol 77 (2011):6794) led to the conceptual basis of this project. An unknown factor was: how many GPCRs need to be stimulated within the reporter cell to give rise to sufficient fluorescence to allow for sorting of the activated cells? Phages only display 3-5 peptides per cell and it was thought that this would be insufficient to detectable stimulate the GPCRs on the reporter cell. Based on other Gram-positive bacterial display systems it was thought that around 4’000-10’000 lanthipeptides might be displayed by the bacterial lanthipeptide display system. It was hypothesized initially that this was enough to detectably stimulate reporter cells allowing for cell sorting. In the course of the project proof of principle of bacterial displayed peptide stimulating eukaryote reporter cell was reached, the importance of the site of display N- or C-terminal and leader peptide removal, the importance of the position of the activating part of the agonistic peptide, and successful agonist display on phages that agonistically stimulated the reporter cell at high sensitivity.

Bacterially displayed galanin after leader peptide removal was able to slightly elevate a fluorescence signal in reporter cells. However, significant background signal was measured as well. Since the signal was clearly not distinct (Fig. 13, left panel) no sorting was possible. Bacterial display used in the different experiments within SYNPEPTIDE without exception concerned C-terminal display. We suspected that the position in the fusion protein could be the problem and therefore switched to phage display, which would allow for N-terminal display of galanin. Indeed, N-terminally displayed galanin was clearly able to elicit distinct fluorescence (Fig. 13, right panel) allowing FACS-mediated sorting.

N-terminally phage-displayed galanin and neurotensin, however failed to be modified by lanthionine-installing modification enzymes. This could have several causes such as insufficient time prior to SEC-mediated export, too high bulkiness for SEC-export, of aspecifically crosslinking cysteines precluding export.
As a consequence C-terminal phage display, developed within our organization in the course of another project, was applied and was demonstrated to successfully allow the introduction of lanthionines, the display and selection of lanthipeptides. C-terminally phage-displayed neurotensin stimulated the neurotensin receptor in CHO reporter cells (Fig. 14). The stimulation was dose dependent and absent in control experiments (Fig. 14). To verify that phage-displayed neurotensin indeed is capable of activating the neurotensin receptor, a DiscoverX assay based on chemiluminescence reporting stimulation of the G-protein-independent pathway was applied. While control peptides did not elicit any signal, the phage-displayed neurotensin gave a dose- dependent increase in chemiluminescence. These data prove that neurotensin displayed on phages indeed stimulate the neurotensin receptor.

We conclude that FACS-detected activity of phage displayed peptide is confirmed by classical assays in microwell plates. This indicates that in principle also via large scale robotics a screening of phage displayed agonists might be performed. However, screening of phage-displayed lanthipeptides via FACS as intended and achieved in this project is obviously overwhelmingly more efficient.


VI) Assays for novel enzymes for implementation of a novel biosynthetic route to biotin.

The replacement of chemical production routes by possibly economically and environmentally more sustainable biosynthetic production routes is one of the cornerstones of industrial biotechnology. Interestingly, even though vitamins by definition are produced biochemically, only a few of them are produced by biotechnological means (e.g. riboflavin). Biotin is not among them, and one of the reasons for that is that it is only required in small amounts per cell and therefore evolution apparently never produced enough selective pressure to optimize what is one of the worst catalyzed reactions known in biochemistry, the final step in the standard biosynthesis to biotin. In the past, we drew up a number of alternative metabolic routes based on intermediates that are either available in standard metabolism of can be easily imagined to be obtained from it. In our analysis, the crucial hurdle to overcome was the introduction of the formation of the thioether-bridge of the thiophane ring of biotin. As thioether formation is one of the crucial activities of the PTM chemistry of lantibiotics, we decided to scout in this class of enzymes for possible enzymes that could help us in biotin synthesis.

In Fig. 15 below, the overall strategy is shown. The critical thio-ether introducing step is labelled ω-1, and in order to select for improvements in enzymes that could catalyze it, two prerequisites had to be fulfilled: first, compound C was needed to be available inside the cell (either fed from outside or produced from metabolism) and then reaction product D would need to be converted to biotin for selection either by growth or via a regulatory circuit sensitive to biotin.

The identification of a suitable enzyme to form biotin from D could be achieved without major problems. Suitable enzymes are available in the standard metabolism of E. coli, but we settled for an alternative version from Pseudomonas taetrolens. The more difficult problem resulted, as expected the catalysis to produce D in vivo.

Proposed is the dehydration of compound C to form the five-membered thiophane ring of biotin (reaction ω-2, Figure 1) formed from two chemically synthesized precursor substances compound A and B, respectively. Compound C is a highly functionalized molecule containing an thiol, an amine, and a hydroxyl functionality, all required in order to allow cyclization.

A similar moiety is found in lantiobiotics. Mesarcidine, for instance, contains at the N-terminus a lantionine which is formed from cysteine and serine upon cyclization catalysed by the lanM (see Figure 16). Note the in case of lantibiotics, this thioether bridge is introduced between the thiol of cysteine and the enamine formed form serine.

However, while the nitrogen in the serine present in the peptide chain is part of a peptide bond, in compound C the amine is primary. Unfortunately, primary enamines are unstable in water and readily (time constant on the order of seconds) undergo hydrolysis to the corresponding ketone (Fig. 16) in a proton-catalysed reaction. This is due to the fact that peptide bonds cannot be protonated while primary amines are still rather basic and therefore accept protons from solution. This important difference between the two substrates, i.e. peptide bound nitrogen cysteine and serine or threonine-bound present in lanthionines and the primary amine present in compound C, suggest that cyclization of the compound C after its dehydration requires an active enzyme that efficiently converts the instable amine prior to its hydrolysis. On the other hand, metabolites are frequently of poor stability but can still be converted rather efficiently by microbes provided that appropriate enzyme catalysts are synthesized by the cell which suggests that catalysis of the reaction might still be possible. Therefore, we proceeded to install a screening system that would allow evolving the NisBC machinery for this reaction.
The lantibiotic cyclization machinery is tightly regulated as the enzymes occur to be allosterically regulated by the peptide leader sequence, i.e. only display activity if the leader is bound to them. We demonstrated that processing of the nisin core peptide (native peptide backbone with no leader attached) with the Nis B/C machinery is possible provided that the leader is concomitantly expressed separately in trans. Our model is that the leader transiently binds to the NisB/C machinery and thereby renders the enzymes catalytically active. We successfully expressed this system and achieved sufficiently high conversion rates (in vivo) in order to introduce three of the five rings characteristic for nisin into a core-peptide template in mg-quantities at low biomass concentrations (~1 g / L CDW). The fact that not all rings were introduced points towards subtle differences of allosteric regulation of the enzyme as compared to the conversion of the native substrate of leader and core peptide together.
While the screening system was thus established, we could procure compound C from customized synthesis. Therefore, we proceeded to design a biochemical route to this intermediate. The reaction should be catalysed by an aldolase, more specifically by a threonine aldolases. These enzymes are capable of reacting primary amines with aldehydes to amino alcohols. Unfortunately, the two starting materials we considered for this reaction (compounds A and B) displayed considerable instability in water and readily (minutes) reacted with each other to thiazolidines. This side reaction impeded our efforts to screen directly for the forward reaction, i.e. the formation of compound C from compound A and B. Nevertheless, we cloned a number of enzymes (among others threonine aldolases) after a literature survey and tested them for activity after purification in the hope that catalysis would take place prior to the formation of the unwanted thiazolidine products. Unfortunately, no traces of compound C could be detected by HPLC-MS in any of the cases.
We therefore decided to screen for a proxy of the reverse reaction. Specifically, analogues to phenylserine carrying a substitution pattern similar to compound C (i.e. a thiol in the required position) were sued as a screening substrates and we set out to identify new enzymes from the metagenome capable of catalysing aldole cleavage of amino alcohols. From the mechanistic point of view, there is one of the important structural difference between compound C and phenylserine, which is the carboxyl-function of phenylserine, which is replaced by a slightly bigger, polar, thiolated and amidated substituent. Still, aldolases accosting phenylserine as a binding partner are likely to possess the catalytic promiscuity to also catalyse the aldol condensation reaction of compound A and B to C.
For identification of the desired activities a genetically encoded sensor capable of responding to benzoic acid was developed. This sensor consists of the transcriptional regulator BenR responsive to benzoate as an activator and the reductase XylC which oxidizes benzaldehyde to benzoic acid. Upon aldol cleavage of phenylserine or its analogues benzaldehyde will be liberated, converted to benzoic acid, which then can be detected. The sensor circuit is therefore a generic tool for detection of potentially interesting activities in library screenings.
We first attempted to use this circuit for the identification of aldolases from metagenomic libraries in flow cytometric assays. In other words, the desired reactions were assayed directly in the cells used for expression of the library. However, even though the sensor worked well in cultures and we went through various round of enrichment, we could not identify any positive clone from library screenings. We reasoned that either benzaldehyde or benzoic acid that might have been formed in during screening might have left the cells by either diffusion or even active transport. However, as we also saw that the background signal increased in the course of our enrichments, we reasoned that unspecific induction of the library towards the end of the incubation time is impeding the result. We hence decided to modify the protocol and shortened the incubation times in the hope to further keep the background at lower levels. We also developed a colour assay in order to reconfirm activity after spotting of putative screening positive clones on plates. This assay should allow us to better cope with the limitations of the flow cytometric assay by more generous sampling. However, screening still failed as indicated by the fact that we could not recover clones capable of cleaving phenylserine due to the overexpression of a threonine aldolase from a mock library albeit.
We therefore went into another round of assay development using microcompartments for immobilization of the cells which should allow retaining any reaction product. For this, small agarose carriers (volume of 50 pL at a diameter of 50 µm) were synthesized and a mock library was encapsulated in the microcompartments and allowed to develop into microcolonies. This time, E. coli cells expressing a threonine aldolase could be clearly separated from the background. We expect that this assay will now allow us to produce a larger collection of aminoalcohol aldolases which may than also be used for condensation of compound A and B to compound C.

Potential Impact:
SYNPEPTIDE’s impact comes on three levels: On a societal level, on an economic level and on a conceptual level.

Societal level
The activities carried out in SYNPEPTIDE addressed three contemporary challenges of modern societies, two of them in the pharmaceutical field, and one in the field of industrial biotechnology. First, SYNPEPTIDE contributed to relieving one of the major increasing threats in modern medicine, the rise in antibiotic resistant strains among clinical isolates, which threatens to render one of the most effective tools of modern medicine increasingly ineffective. Next, SYNPEPTIDE targeted the development of improved drugs in the major drug class in the modern pharmaceutical industry, G-protein coupled receptors. And finally, SYNPEPITDE contributed to the transition of the chemical industry from a sector based nearly entirely on fossil raw materials to an industry employing more and more biosynthetic processes and thus based increasingly on renewable resources.

Regarding the first issue, a combination of negligence of antibiotic development in the pharmaceutical field (Fig. 17), inadequate application in the treatment of humans, and excessive use in the farming field has left many of the current antibiotics harmless in clinical settings (Tab. 4). While some of the major pharma companies have re-entered the field, much of the involved difficulties relate to the fact that antibiotic development is only a moderately attractive field of research for a global pharma company, as the revenues generated from novel antibiotics (which quasi by definition become reserve antibiotics reserved for exceptional clinical use) will be limited.
Consequently, much of the responsibility in this domain has shifted from large pharma to SMEs and the academic sector. This has led to the emergence of a flurry of novel concepts in how to deal with pathogenic and resistant bacteria, and in particular to an invigoration of research in the field of natural products, which is still the main source of inspiration for novel small molecule drugs.
SYNPEPITDE has, thanks to the competence of its participants in RiPP-research, assay development, and DNA synthesis developed a novel strategy to evolve antimicrobial molecules. The validity of this strategy has been demonstrated, and novel interesting molecules have been generated. We provide a template for future academic efforts, but more importantly a strategy for how to obtain novel, urgently required molecules.

Regarding the second issue, better drugs for GPCRs, SYNPEPTIDE has pioneered an approach with considerable mid-term potential. Agonists are an underrepresented drug class, simply because the structural requirements for such drugs are very high and can usually only be met by complex molecules such as proteins. Nevertheless, by being able not only to block, but also to activate receptors, the therapeutic potential is, broadly spoken, doubled. Given the eminence of GPCRs in pharma, this can have major implications for future drug development programs.
Here, SYNPEPTIDE contributes a novel high-throughput assay with the potential to drastically accelerate the rate with which molecules are evaluated for such purposes. In fact, chances are that the increased rate effectively enables some approaches in this domain.

Finally, in terms of industrial biotechnology, SYNPEPTIDE has made contributions to the constantly increasing tool set that is provided by academia and industry to drive the development of biosynthetic routes. Even though we did not manage to demonstrate a novel route, we designed a variety of assays that involve enzyme classes of considerable importance in the fine chemical sector (threonine aldolases), where E-factors (ratio of kg waste produced per kg product) are particularly high. We therefore contributed to accelerating the transition to biosynthetic routes and to lessening the reliance on fossil resources and a more sustainable chemical sector in the future.

Economic level

The SYNPEPTIDE participants undertook the project effort with the explicit goal to lay the foundations of a novel company with the focus of developing novel antibiotics. Even though we cannot present molecules in clinical trials yet, we argue that we are on an excellent track. We established a technological basis, provide proof-of-principle, have established a pertinent network, and therefore are set to continue our efforts.

We expect that we will found “Terebra” in the very near future (see Deliverable 6.11) and hope that we can develop this into a thriving drug development company with the corresponding potential for generating employment.

Also the two SMEs with own developmental activities that were part of the SYNPEPTIDE consortium benefit from the obtained results. Lanthiopharma has now available a novel assay in the field of its core competence which hopefully will allow it to address entirely novel drug development trajectories. While the assay alone is of course not enough, it is an important milestone on the way. Therefore, we argue that the SYNPEPTIDE results strengthen the position of Lanthiopharma in its commercial sector and improve their mid-term economic outlook.

Fog FGen, the novel assay fall in the core competence of the company. Next to its efforts of developing its own products (such as biotin), FGen provide services to the chemical and pharmaceutical industries, which consist mainly of high-throughput experimentation. Here, the bottleneck in each new project is assay development that needs to precede each (bio)catalyst development campaign and is critical to the generation of business. Here, additional assays for important enzyme classes are crucial and strengthen the capacity to attract service business.


Conceptual level

We argue that SYNPEPTIDE makes a major impact at the conceptual level. Specifically we argue that thanks to the efforts of the SYNPEPTIDE consortium the concept of peptides as drugs has gained much credibility and as such SYNPEPITDE has made an important contribution to equipping the pharmaceutical industry with strategies to develop novel and efficacious drugs.

Even though peptides have a tradition in the pharmaceutical industry (insulin in the treatment of diabetes, vancomycin as an antibiotic), their potential is seen very critically. This includes arguments such as short half-life in bodily fluids, difficult application (injectables), and cost-intensive manufacturing. While each of these arguments is valid, there are ample examples to the contrary (see above). Therefore, we argue that a main hurdle in opening the drug development field for peptides is implementing the methods that are required to come rapidly to novel molecular scaffolds that can serve as a starting point for a worthwhile development. In other words, it has to become easier to generate starting point from existing and novel function space.
Here, SYNPEPTIDE has made breakthrough contributions. By focusing on gene-based, ribosomally produced RiPPs as targets and by providing the corresponding assay capacities, we have shifted molecular engineering from low-throughput, rational approaches to high-throughput molecular evolution (WP4). This is an entirely different proposition in molecular engineering, and has large consequences for the rate at which novel molecules can be developed and improved.
In addition, we have addressed major problems that might in the future limit this approach: Given the importance of posttranslational modifications in order to provide structure to the peptide molecules, we provide methods to comprehensively explore the potential for promiscuity of existing PTM machineries (WP4) as well as templates on how to increase the functional space that can be accessed in screenings (WP1). In addition, we laid a foundation to include “docking stations” (ncAAs with potential for click-chemistry) into the molecular evolution process (WP2).
Finally, we contribute two entirely different assay principles that open two entirely different sectors for development. One the one hand, we use bacterial sensor strains for read out of activity (such as antimicrobial function), on the other hand we use phage display to enable the evaluation of interactions with mammalian cells (WP4). Both enable pharmaceutical developments, yet for entirely different applications.

In summary, we argue that SYNPEPITDE has made substantial contributions on different levels of the chain of innovation from basic science to translation by SMEs.


Main dissemination activities so far

In agreement with the different levels of implications, we also pursued a differentiated dissemination strategy:

Conceptual level
We report the different conceptual advances in peer-reviewed journals and via poster and oral contributions to the relevant community of academic and industrial scientists. We are happy to point out that SYNPEPTIDE efforts have already lead to the publication of 4 articles in high impact journals (Angewandte, Nature Chemistry), and three more are currently in review or about to be submitted (1 in review at Science, one at Nature Communications, and one to be submitted to Science). We think that this adequately reflects the level of conceptual innovation that we aimed at in SYNPEPITDE.

Economic level
Dissemination at the economic level is to a large extent in the hands of our SME participants FGen and Lanthiopharma. We are confident that the breakthrough nature of some of the findings will be put to excellent use in their capable hands.
At the same time, the academic partners in the SYNPPETIDE consortium are committed to develop the identified antimicrobial molecules and future novel molecules to useful medicines in the frame of a newly founded company, Terebra. The activities are ongoing.

Societal level
SYNPE TIDE has invested substantial effort in bringing the overall setting of problems and the different approaches we undertook to address them to the general public. SYNPEPTIDE participants participated in diverse activities including radio shows, newspaper articles, science cafes, etc in order to provide direct contact. Next to these diverse activities, we followed four major communication routes:

a) The SYNPEPTIDE film: We produced a film that summarizes the set of problems and the approaches in a very appealing way. We are confident that the film will find a worthy audience and in the short and mid-term, and possibly even in the long term, will contribute to raise awareness of the increase in antibiotic resistance in clinical settings and increase the appreciation of the solutions that science can deliver.
b) The TEREBRA mobile phone game: We produced an electronic game that captures the essence of the antibiotic resistance problem and the scientific approach that we took to tackle it in gamified form. We hope that this unconventional format will allow addressing entirely new groups of people and again increase their appreciation of problem and potential solution.
c) High school packages: To heighten the awareness for the decisive role of science in the modern society, we produced “ready-to-go” material packages for high school teachers that they can use in their teaching efforts to illustrate their teaching content with a modern and important problem.
d) Artists in residence: Finally, we tried to blur the often very sharp boundary between art and science by inviting two scientists to labs of SYNPPEITDE participants and allowing them to explore artistic ways of dealing with the SYNPPEITDE set of problems. We hoped that this would help addressing the disconnect between scientific thinking and working and artistic representation that many scientists feel when visiting events such as “Biofiction”. Alternatively, we hoped that by the opportunity for intensive exchange the definition of what could realistically be artistically represented would be sharpened. We feel that this communication was implemented very effectively.

For all of these “societal” dissemination activities, it is by definition difficult to estimate their effectiveness. SYNPEPITDE has organized professional distribution during the project, but much of the effect is bound to occur in the future.

In summary, we argue that we have effectively disseminated the insights and developments that took place in the course of SYNPEPTIDE. We have done so at different levels for different purposes, and we also argue that the impact of SYNPEPTIDE will be felt at those levels in the future.

List of Websites:
www.synpeptide.eu.

Alternatively, contact the involved PIs under

Coordinator
Sven Panke, ETH Zurich, sven.panke@bsse.ethz.ch

Participants
Jörn Piel, ETH Zurich, jpiel@ethz.ch
Oscar Kuipers, RU Groningen, o.p.kuipers@rug.nl
Ned Budisa, TU Berlin, budisa@chem.tu-berlin.de
Ralf Wagner, Uniklink Regensburg, ralf.wagner@klinik.uni-regensburg.de
Martin Held, FGen, Basel, held@fgen.ch
Gert Moll, Lantiopharma, Groningen, moll@lanthiopharma.com
Markus Schmidt, Biofaction, Vienna, schmidt@biofaction.com