Cluster Selection

The Basis of Our Antibiotic Discovery Platform

Graphical Abstract

Introduction

Native NRPS Producers

Nonribosomal peptide synthetases (NRPS) are widespread across all domains of life, particularly among microbes such as bacteria and fungi, where they play a crucial role in the biosynthesis of diverse secondary metabolites^[1][2]. These metabolites are often essential for survival and competition within microbial communities.

A wide range of bacteria are known to produce nonribosomal peptides (NRPs), but their native producers, as well as the NRPS clusters, are often genetically difficult to access and manipulate. Streptomycetes, for example, produce a wide range of diverse nonribosomal peptides, yet their exceptionally high GC content makes genetic engineering particularly challenging^[3].

At the start of our project, we chose E. coli as our engineering platform. It is one of the best-established heterologous expression hosts, offers reliable systems for DNA manipulation, and is widely accessible to iGEM teams around the world — making it an ideal foundation for our NRPS engineering efforts.

To increase the likelihood of having functional biosynthetic gene clusters (BGCs) we aimed to express in our host, we focused on NRPS BGCs from the same class of gamma proteobacteria, more specifically on the closely related genera Photorhabdus and Xenorhabdus and Pseudomonas^[4][5]. This choice was guided by key considerations such as GC content, codon usage, and the availability of metabolite precursors required for efficient peptide biosynthesis.

Choosing the Clusters

Within the genera Photorhabdus, Xenorhabdus, and Pseudomonas, a plethora of different NRPS clusters can be found. To guide our decision on which cluster, and therefore which peptide, to prioritize, we consulted Prof. Dr. Felix Hausch.

He explained that cyclic peptides often outperform linear ones. The interaction of the peptides through multiple contact points result in antibody-like affinity and specificity for a broader range of drug receptors. Based on this insight, we chose to focus on five biosynthetic gene clusters (BGCs) encoding cyclic peptides: Gacamide (Pseudomonas), Photoditritide (Xenorhabdus), Chaiyaphumine (Xenorhabdus), Szentiamide (Xenorhabdus), and Xentrivalpeptide (Xenorhabdus) (Fig. 1).

🔍

Plasmid Design

Cloning NRPS clusters is highly challenging due to their enormous size. In general, each NRPS module is encoded by about 3 kb of DNA (NRPS Engineering). Since the Garcamidide cluster extends up to 11 modules, this corresponds to roughly 33 kb in total. Introducing such large inserts into plasmids is possible, but they are very challenging to clone and transform, and therefore not an ideal basis for our platform.

To make the expression of larger native BGCs feasible, we split the NRPS BGCs across three separate, orthogonal plasmids, each capable of carrying up to 15 kb inserts (Fig. 2).

🔍

For the backbones we have chosen to design our plasmids according to the Standard European Vector Architecture ( SEVA) to allow for interoperability and a standardized format. We decided on pACYC, pCOLA and pCDF as they are optimized for protein expression with multi-plasmid systems in E. coli ^[6]. To enable simultaneous transformation of all three plasmids and achieve full NRPS expression, each backbone was equipped with a distinct antibiotic resistance marker for selection and orthogonal origins of replication (ori) (Fig. 3). p15A, ColA, and CloDF13 were selected due to their relatively low copy numbers, which makes them suitable for expressing large proteins as they place less metabolic burden on the host cell^[7][8].

🔍

The NRPS expression cassette comprises an inducible araBAD promoter, allowing NRPS expression to be regulated by the addition of arabinose. Arabinose import is ensured by expressing the arabinose transporter AraE and the araC regulator, using a constitutive promoter. Followed by the native araBAD ribosomal binding site (RBS). Once divided at the DNA level and co-expressed in E. coli, the challenge was to ensure that the separately expressed NRPS fragments could reassemble into a functional protein complex. To achieve this, we looked at different NPRS engineering tools and we decided to use split inteins — a powerful tool in protein engineering that enables specific benefits compared to other engineering tools (NRPS Engineering).

Split inteins provide post-translational covalent linkage of two independently expressed proteins. We utilized two orthogonal pairs of split inteins, that were recently introduced as tools for NRPS engineering, to ensure efficient and specific reassembly of the NRPS fragments^[9]. Depending on the specific NRPS fragment, split inteins are positioned at the N-terminus, C-terminus, or both. The cassette is terminated by a T0 terminator.

🔍

NRPS Expression

After splitting the NRPS cluster across three plasmids, we verified functionality through heterologous expression in E. coli (Fig. 4). The bacterium is transformed with all three plasmids encoding a single NRPS cluster which then translates them into the NRPS proteins and inteins. These inteins mediate self-splicing and reassemble the NRPS into its functional form. The reconstituted NRPS subsequently loads amino acids and synthesizes the peptide. Since peptide production itself indicates successful NRPS functionality, we assessed only the abundance of the compound as a proxy for functional expression (Measurement).

The plasmids were introduced simultaneously via heat shock transformation into competent E. coli DH10B::mtaA (Fig. 6). This strain was engineered for NRPS expression by integrating the mtaA gene, which encodes a phosphopantetheinyl transferase (PPTase) responsible for attaching the 4′-phosphopantetheine (Ppant) arm to the thiolation (T) domains (NRPS Engineering)^[10].

Results

Cluster Cloning

All five candidate clusters were split, and their fragments were cloned into the respective backbones using Gibson Assembly. However, cloning of the Gacamide cluster from Pseudomonas fluorescens was unsuccessful, underscoring the challenges of NRPS cloning due to its highly repetitive sequence (Fig. 5).

🔍

Cluster Expression

After cloning the clusters into their respective backbones, we proceeded to test NRPS and peptide production for Chaiyaphumine, Szentiamide, Xentrivalpeptide, and Photoditritide.

NRPS expression was induced by adding arabinose to the production cultures (Fig. 6). Cultures were grown in XPP3 medium - originally developed as “Xenorhabdus/Photorhabdus Production Medium” - to enhance the production of secondary metabolites, including nonribosomal peptides and polyketides. Additional details on the medium composition and supplements are provided in the protocol section.

🔍

Successful Expression of Xenorhabdus NRPS Clusters in E. coli

Photoditritide failed to produce a detectable peptide in E. coli (Fig. 7).

🔍

This left clusters - Chaiyaphumine, Szentiamide, and Xentrivalpeptide (all from Xenorhabdus) - that successfully yielded the expected cyclic peptides (Fig. 8).

🔍

Notably, Chaiyaphumine and Xentrivalpeptide were detected at high intensities, demonstrating that splitting NRPS clusters across three plasmids and reassembling them with inteins is an effective expression strategy. Using our calibration curve (Measurement), we determined production titers of 70 mg/L for Chaiyaphumine D and 17 mg/L for Chaiyaphumine A. These results also validate our backbone design as a robust system for expressing large BGCs, enabling high production yields.

Localization of Nonribosomal Peptides in E. coli Production Cultures

To determine whether the NRPs remain in the cell lysate, are secreted into the medium, or bind to added beads, we conducted a targeted experiment (Fig. 9).

🔍

The results revealed that the NRPs are primarily secreted into the culture and subsequently attach to the beads, confirming that our method effectively captures NRP production.

Outlook

Building the NRPieceS library on these clusters

The native clusters Chaiyaphumine, Szentiamid and Xentrivalpeptide build the very basis of the NRPieceS library. How the library was constructed out of these clusters is described in the 'Designing & Building our Parts' section below.

Part Collection

All parts described on this page are registered and described on the iGEM Parts Registry. Learn more about our parts collection here.

Designing & Building our Parts

Key Points

• establishing 9 acceptor vectors from 3 native NRPS clusters, laying the foundational framework of the NRPieceS plasmid collection.

• Construction of 35 donor modules enabling modular peptide derivatization.

• Cloning a total of 105 NRPS hybrid plasmids.

Graphical Abstract

Introduction

The Principle of the NRPieceS Library

The purpose of the NRPieceS library is to enable targeted exchange of NRPS modules, thereby altering the amino acid composition of the final peptide (Fig. 10). For the creation of our library we chose the XUT^I site (NRPS Engineering) for defining our exchange units as it was reported in literature to be a reliable engineering site^[11].

🔍

To efficiently engineer NRPSs, we developed a system that enables the exchange of NRPS XUT^I modules using Golden Gate. Our approach is based on a set of acceptor and donor vectors designed for integrating DNA fragments (Fig. 11). The acceptor vectors contain the native biosynthetic gene clusters, split onto three plasmids (Cluster Selection). In the acceptor vectors the insertion position is pre-defined, while the donor vectors provide the corresponding XUT^I modules ready for insertion. After inserting a donor module into an acceptor vector, the plasmid is co-expressed with two other plasmids. The flexibility in choosing the two plasmids for co-expression allows further recombination (NRPS Engineering).

🔍

Results

Designing the Acceptor Vectors

The NRPieceS library is derived from three NRPS clusters identified during the cluster selection process, with serving as the primary focus of our engineering efforts due to its high production.

First, XUT^I sites were identified in each native NRPS. Two XUT^I sites per NRPS were selected to introduce inteins for post-translational protein reassembly (Fig. 12). Because all three NRPS produce depsipeptides, the threonine module was excluded from exchange, as this amino acid is essential for the cyclization. Furthermore, the final module was left unchanged, as the TE domain may exhibit specificity for upstream amino acids, and swapping it could compromise function.

🔍

The native NRPS clusters were divided at the XUT^I sites into three parts - initiation, elongation, and termination (Fig. 13). For the reassembly of the NRPS after translation inteins were inserted at the division sites. The inteins pose a great benefit for recombination of NRPS parts but they also constrain the selection of our exchange positions to 3, except Xentrivalpeptide 4, positions. These are the XUT^I modules that incorporate the amino acid in the cyclic part of the peptides.

The native NRPS clusters were divided at the XUT^I sites into three segments - initiation, elongation, and termination (Fig. 13). Inteins were inserted at these sites to enable reassembly of the NRPS after translation. While inteins greatly facilitate the recombination of NRPS parts, they also restrict which XUT^I positions are available for module exchange. In Chaiyaphumine and Szentiamid, three positions are available (XUT^I 2, XUT^I 3, and XUT^I 4), whereas Xentrivalpeptide has an additional position in the elongation module. All of these XUTI modules are responsible for incorporating amino acids into the cyclic part of the peptides.

🔍

After identifying which XUT^I modules we want to exchange in the native NRPS, we next aimed to make these modules accessible for Golden Gate cloning at the DNA level while also establishing a reliable, post-transformation readout for cloning success after transformation. To achieve this, Type IIS restriction sites for the enzyme BsaI were inserted creating A-B overhangs at the XUT^I sites and an mCherry expression cassette was incorporated (Fig. 14). Details on the overhang design can be found on our engineering page.

🔍

🔍

This design provided a simple visual readout in the subsequent Golden Gate cloning. Colonies retaining the mCherry cassette appear pink, indicating unsuccessful donor insertion. In contrast, successful donor insertion removes the mCherry expression cassette, resulting in white colonies and enabling rapid, reliable screening (Fig. 15).

The final design of our acceptor vectors incorporates all elements necessary for modular NRPS engineering: an inducible promoter, inteins for NRPS reassembly, Golden Gate restriction sites, mCherry cassette, terminator, and the remaining NRPS unit within a standardized backbone (Cluster Selection) (Fig. 16). The vectors were assembled using Gibson assembly, creating fully functional constructs ready for recombination experiments.

🔍

Designing the Donor Vectors

After designing and constructing the acceptor vectors for module insertion, the next step was to generate the corresponding donor vectors carrying the modules themselves. As NRPS are widely known for their great variety of building blocks we wanted to make exactly this available with the selection of our donor modules - the NRPieceS^[12]. This selection was guided by two key criteria: the amino acid specificity of the adenylation (A) domain and the overall domain composition of each donor module.

As we wanted to include epimerization for more peptide diversity in our library, three donor module designs are possible (Fig. 17):

• The LCL-type consists of a condensation domain only which means that the condensation reaction leaves the stereochemistry of the upstream amino acid untouched in L-configuration (LCL).

• In the CE-type, the condensation–epimerization (CE) domain converts the upstream amino acid from L- to D-configuration during the condensation reaction.
  In the E-DCL-type, the epimerization (E) domain is separated from the condensation domain, and the upstream amino acid is converted from L- to D-configuration, then the DCL domain condensate the upstream D with the downstream L amino acid.

🔍

🔍

The donor backbone carries a gentamicin resistance marker, distinct from the acceptor vector’s antibiotic marker for selection, and features a moderate-copy-number ColE1 origin of replication (Fig. 18)^[7]. The selected XUT^I modules were inserted into this backbone via Gibson Assembly and flanked with compatible A-B overhangs for Golden Gate assembly with the acceptor vectors.

Proof of Concept of the NRPieceS Library

We validated our Golden Gate system by cloning XUT^I modules from Xenorhabdus and Photorhabdus into donor plasmids and assembling them with the middle Chaiyaphumine acceptor vector (Fig. 19A).

We expressed hybrid NRPS constructs in which the central module was exchanged using our Golden Gate–Intein system. Successful peptide production confirmed that our system enables functional module replacement at defined positions.

However, only three out of eleven constructs produced detectable peptide levels, and the titers were markedly low (Fig. 19B).

🔍

The native Chaiyaphumine control successfully yielded its peptide, confirming the functionality of the Golden Gate–Intein system (Fig. 20). These results suggest that while our system enables modular NRPS recombination, the efficiency of functional peptide production strongly depends on the compatibility between donor and acceptor modules.

🔍

Consequently, following this DBTL cycle, we implemented a more systematic strategy for selecting compatible modules to enhance the efficiency and yield of engineered NRPS systems.

Phylogeny-driven Donor Module Selection for Hybrid NRPS

To find modules that are more likely to be compatible, we wanted to harness the fact that NRPS modules from related clusters are known to work better together To harness that we came up with the idea to use the thioesterase domain as a phylogenetic marker for finding related clusters^[13]. We consulted an expert in evolutionary biochemistry Prof. Dr. Georg Hochberg, who confirmed that the TE domain is particularly well suited for this approach, as it only appears once per cluster, and thus allow comparison between different clusters, without having to deal with the problem, that all other domains appear multiple times within the same cluster.

Our goal was not only to enhance the functionality of hybrid NRPS through a better donor module selection, but also to expand the chemical diversity accessible via NRPS engineering. To achieve this, we expanded the pool of amino acids covered by the NRPieceS library and included all three types of donor modules (LCL, E-DCL, CE) to increase the variety of possible peptide products. Therefore, all canonical amino acids were selected, along with additional “special” amino acids such as ß-alanine, 2,4-diaminobutyric acid (DAB), and 4-azido-L-phenylalanine (N3-Phe) for use in drug delivery experiments (Fig. 21).

🔍

NRPS clusters from Xenorhabdus and Photorhabdus were screened to identify modules specific for the amino acids we aimed to include. Based on these results, a phylogenetic tree was constructed, and the most closely related module was selected as the donor (Fig. 22). More details on the TE-based donor selection approach can be found here.

🔍

By applying this principle, a total of 50 donor vectors were designed and 35 of them were successfully cloned via Gibson Assembly (Fig. 23). All of them are available as parts of the NRPieceS Collection.

🔍

Using the TE approach, we developed a reliable tool to guide the design of hybrid NRPS. To make this accessible for everyone, our software includes a user-friendly feature for calculating TE domain similarity.

Building the NRPieceS Plasmid Collection

Finally, we combined the 35 donor vectors with the 3 acceptor vectors of our lead cluster, Chaiyaphumine, using Golden Gate cloning.

Alanine

Arginine

Arginine

Arginine

Asparagine

Asparagine

Aspartic acid

beta-Alanine

beta-Alanine

Diaminobutyric acid

Glutamine

Glutamic acid

Glutamic acid

Glycine

Histidine

Isoleucine

Isoleucine

Leucine

Lysine

Lysine

Lysine

Phenylalanine

Phenylalanine

Phenylalanine

Proline

Proline

Serine

Threonine

Threonine

Tryptophan

Tryptophan

Tyrosine

Tyrosine

Valine

Valine

Epimerization:

no

yes

yes

no

yes

no

no

yes

no

yes

no

yes

no

yes

yes

yes

no

yes

yes

yes

no

yes

no

no

yes

no

yes

yes

no

no

no

yes

no

yes

no

Source cluster:

Xenorhabdus khoisanae DSM 25463, Xenoamicin synthetase XUT2

Xenorhabdus mauleonii DSM 17908, Fitayylide derivat synthetase XUT6

Xenorhabdus cabanillasii JM26, Bicornitin derivat synthetase XUT6

Xenorhabdus cabanillasii JM26, Bicornitin derivat synthetase XUT3

Xenorhabdus cabanillasii JM26, Fabclavine synthetase XUT4

Xenorhabdus innexi DSM 16336, Fabclavine synthetase XUT3

Xenorhabdus khoisanae DSM 25463, unknown synthetase XUT6

Xenorhabdus nematophila ATCC 19061, Xenematide synthetase XUT4

Xenorhabdus innexi DSM 16336, Fitayylide synthetase XUT3

Photorhabdus kayaii DSM 15194, unknown synthetase XUT3

Xenorhabdus romanii DSM 17910, Intrazentin derivat synthetase XUT4

Photorhabdus temperata subs. thracensis DSM 15199, unknown synthetase XUT5

Xenorhabdus beddingii DSM 4764, unknown synthetase XUT2

Xenorhabdus szentirmaii DSM 16338, unknown synthetase XUT8

Xenorhabdus miraniensis DSM 17902, unknown synthetase XUT13

Xenorhabdus KK7.4, Xenobactin synthetase XUT5

Xenorhabdus KK7.4, Fiayylide derivat synthetase XUT6

Xenorhabdus cabanillasii JM26, Taxlllaide derivat synthetase XUT6

Photorhabdus luminescense IT4.1, unknown synthetase XUT3

Xenorhabdus hominickii DSM 17903, PAX synthetase XUT6

Xenorhabdus mauleonii DSM 17908, PAX synthetase XUT3

Xenorhabdus mauleonii DSM 17908, Protegomycin derivat synthetase XUT3

Xenorhabdus PB61.4, Chaiyaphumine synthetase XUT2

Photorhabdus luminescens subs. laumondii TT01 DSM_15139, GxpS synthetase XUT3

Xenorhabdus PB61.4, Chaiyaphumine synthetase XUT4

Xenorhabdus nematophila ATCC 19061, Xenoamicin synthetase XUT11

Xenorhabdus KK7.4, Fitayylide derivat synthetase XUT2

Photorhabdus temperata K122, unknown synthetase XUT4

Xenorhabdus KK7.4, Xenobactin synthetase XUT3

Xenorhabdus hominickii DSM 17903, Xenematide/​Taxlllaide synthetase XUT5

Xenorhabdus KJ12.1, Xenoprotide synthetase XUT3

Xenorhabdus innexi DSM 16336, Fitayylide synthetase XUT5

Xenorhabdus stockiae DSM 17904, Xenoprotide derivat synthetase XUT3

Xenorhabdus hominickii DSM 17903, unknown synthetase XUT3

Xenorhabdus KK7.4, Xentivialpeptide synthetase XUT6

Part name: BBa_25GRKS3J BBa_25JQ5Y0E BBa_2546K4QE BBa_25I5QU6P BBa_253F5MN6 BBa_25W6NAAK BBa_25CZ42LP BBa_25ODH20W BBa_25C8C5FV BBa_255Z007P BBa_25DUZZVM BBa_25L885HE BBa_25498JTE BBa_25631CKS BBa_25A33AG4 BBa_25RUHHBG BBa_25SPBXEH BBa_25L2LT9Z BBa_25QM73MD BBa_257L92YL BBa_25E3QF9W BBa_25B6CBPG BBa_25JVF1RU BBa_25P50E3H BBa_256QSWQN BBa_2544KP0W BBa_25NMTD8Y BBa_25E7OHKD BBa_255BCE4G BBa_25SPZFCP BBa_25055HQG BBa_25APRBKS BBa_25KE336Y BBa_25F2MQ35 BBa_25X9PUZ6

A

R

R

R

N

N

D

ßA

ßA

Dab

Q

E

E

G

H

I

I

L

K

K

K

F

F

F_L

P

P

S

T

T

W

W

Y

Y

V

V

2

3

4

Chaiyaphumine
Xenorhabdus sp.
PB61.4

BBa_25O5HCVX BBa_25V5G86W BBa_253DFKWX BBa_25696F6U BBa_251I5HRE BBa_25FFJF3V BBa_251SWGLU BBa_25JQ6UGB BBa_25GX3ZJ4 BBa_2585O7JE BBa_257NQ6XB BBa_25HJOUSX BBa_25M17972 BBa_25CZFJ68 BBa_25MBCAYC BBa_25ZGNUI4 BBa_252KOVW6 BBa_25XRQSSG BBa_25HPRG5E BBa_251OOHAS BBa_259MYIUM BBa_25HSSATK BBa_257TGFX2 BBa_253M7EBF BBa_252ZQMN6 BBa_25YLDO0C BBa_25HDI0R9 BBa_25B9NJO8 BBa_25RNFPM8 BBa_25LMDCTV BBa_25J3CYXN BBa_25ZGVXKU BBa_25TRMRJH BBa_25UH5WSK BBa_25U8NYFH BBa_25P3Q1BQ BBa_25NILA3H BBa_25A661VA BBa_25V2SUON BBa_25NILA3H BBa_25N1JNJU BBa_25L7FO2G BBa_25NILA3H BBa_25DVND4I BBa_25S3RX4R BBa_25D8391D BBa_25HXHLBU BBa_25LZPWWI BBa_25NBWMXC BBa_25G6XSNC BBa_25ATPF1V BBa_255IBIXB BBa_251C3KOQ BBa_25M92T4P BBa_258D3BQX BBa_256CPU7S BBa_2590HRJX BBa_25BPT7FT BBa_25JCZSGT BBa_250BLCIQ BBa_25IY3OJV BBa_25N5CBXT BBa_25TJHBMA BBa_25JQL2V1 BBa_25TBBHMJ BBa_25N8PT1Z BBa_2590ZIEO BBa_258GSVNQ BBa_2508MJ0Y BBa_25WCBOUE BBa_25U0C9DU BBa_25TFXXFO BBa_25TIP463 BBa_25P6EHSR BBa_25Y2DS8P BBa_25T4ZUXN BBa_250NPZCE BBa_25GPAMZF BBa_253IIP3G BBa_25T38H3M BBa_25AXTKU8 BBa_25RVBZA5 BBa_25DR5PII BBa_25V5EA9I BBa_257K6LLL BBa_25AUL0PL BBa_25SC77SH BBa_25XEC79C BBa_25VFHJ3A BBa_25CSBUH7 BBa_25MDJISM BBa_25RYS8II BBa_25HBVKHA BBa_25CCQHIT BBa_25IE7WWB BBa_25ADPKBO BBa_25C47GU4 BBa_25Q44PSG BBa_25FDWYMQ BBa_255GP9OC BBa_259OT3UF BBa_2571R42T BBa_25STMUQJ BBa_25MF1AUK BBa_2525JBI3

Novel
derivatives

This resulted in 105 NRPS expression plasmids, which can be freely recombined to generate up to 35³ (42,875) potential hybrid NRPS constructs for expression (Fig. 24). By integrating inteins with Golden Gate cloning, the NRPieceS Parts collection enables the creation of far more new-to-nature peptides than existing libraries.^[14] All plasmids are available through our parts registry.

🔍

The full potential of the NRPieceS plasmid collection has yet to be realized. Beyond Chaiyaphumine, acceptor vectors for Szentiamid and Xentrivalpeptide can also be combined with the existing donor vectors, potentially yielding 1’543’500 (105x140x105) new hybrid NRPS constructs.

To illustrate, adding just one additional donor incorporating a different building block would generate 33,391 new combinations. Even this modest expansion underscores the platform’s vast capacity to explore NRPS chemical space and its tremendous potential for discovering novel peptide antibiotics. While this example demonstrates the impact of a single donor, it is only a fraction of the more than 400 building blocks that NRPS can utilize and can be converted into donor modules^[15].

Outlook

• Explore the NRPieceS Plasmid Collection
  Detailed instructions for using the NRPieceS Plasmid Collection can be found on our parts collection page, and all parts are also available on the iGEM registry.

• Testing the Plasmid Collection
  Designing a library is a major challenge, but it’s only the first step toward advancing antibiotic development. The next stage is characterizing the functionality of our parts and testing the bioactivity of the novel hybrid peptides against relevant pathogens. Learn more in the characterization or screening part.

• Drug delivery
  The modularity of the NRPieceS library allows access to a vast chemical space of nonribosomal peptides and opens new avenues for targeted delivery. Promising peptides identified through expression and activity testing can be conjugated to siderophores, leveraging bacterial iron uptake to enhance delivery into pathogens.

• Expanding the library with more acceptors & donors
  The NRPieceS Plasmid Collection fully open and ready for contributions. By adding new acceptor or donor modules, the community can expand the library, tailor it for specific applications, and unlock even more hybrid NRPS combinations.

Characterizing our Parts

Testing all our parts

Key Points

• Production of 95 new-to nature peptides.

• Characterization of 35 NRPS exchange units in 105 combinations.

• Established protocol for high-throughput screening of peptide production in E. coli DH10B::mtaA using a three-plasmid system.

• Recombination of native clusters into 24 hybrid NRPS expression systems.

Graphical Abstract

Introduction

After constructing our NRPS library, the next step was to confirm whether these parts could be expressed and produce the intended peptides. Characterizing this library was essential to validate our design and ensure that the expression system functioned as expected. This section describes our workflow for expressing the NRPS modules in E. coli, isolating the resulting peptides from culture, and analysing the production with LC-MS.

Results

Shuffling the Native Cluster via Inteins

After identifying three NRPS clusters that were expressible in E. coli (Cluster Selection), we investigated recombination through intein-mediated shuffling. For this, we combined the expression plasmids of Chaiyaphumine, Szentiamid, and Xentrivalpeptide with each other to generate 24 hybrid NRPS variants alongside the 3 native NRPS constructs (Fig. 25).

🔍

Peptide production was assessed using our 10 ml NRP production protocol, with empty backbones and uninduced Chaiyaphumine plasmids serving as a negative control. Extracts were analyzed by LC-MS following expression. Of the 27 NRPS, 21 showed peptide production, which means also 18 new-to-nature peptides were produced. (Fig. 26).

🔍

Intein Shuffling Chromatograms

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

These results demonstrate that the three NRPS clusters can be recombined successfully via inteins to yield new-to-nature peptides with high efficiency. The approach significantly broadens the structural diversity of the peptide library.

High-Throughput Expression and Screening of the Golden Gate Library

The possibilities for hybrid peptide production of our library is not restricted to intein shuffling. Moreover, we can further derivatize the NRPS through exchanging units in the native clusters using Golden Gate. While this strategy offers great potential, testing all possible module combinations through conventional expression would be highly time-consuming and impractical. To address this, we established a high-throughput expression workflow that enables the parallel evaluation of numerous constructs under standardized conditions.

We miniaturized the transformation, expression, and extraction steps into multi-well formats, integrating automated plasmid handling and bead-based peptide capture. Library plasmids were distributed into well plates using an Echo liquid handler for maximum accuracy, co-transformed into E. coli DH10B::mtaA, and cultivated in deep-well plates. Production cultures were carried out in XPP3 medium supplemented with Amberlite XAD-16N beads for in-situ peptide adsorption. Extracts were then analyzed by LC-MS to determine which module combinations produced detectable products. To streamline media preparation and distribution, empty backbone plasmids lacking an NRPS were used as negative controls instead of uninduced expression plasmids.

With our library theoretically capable of producing 42,875 peptides, we focused on a comparable reference set to characterize the exchange units. For this purpose, we expressed all 105 Golden Gate vectors in combination with the native Chaiyaphumine cluster. LC-MS analysis revealed that 63 of these NRPS constructs produced detectable peptides (Fig. 36).

🔍

Golden Gate Library Chromatograms

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

🔍

Based on our results we observed that there is a clear dependency between position of the exchange unit and NRPS functionality. Modules placed at the XUT^I4 position supported peptide synthesis in 86% of cases (30/35), compared to 54% (19/35) for XUT^I3 and only 40% (14/35) for XUT^I2 (Fig. 72).

🔍

This gradient suggests that substitutions introduced at later positions might reduce unfavorable interactions between non-native residues and condensation (C) domains. Early-position swaps expose substituted residues to multiple C domains that may not have evolved to accommodate these non-cognate intermediates, potentially disrupting catalytic efficiency.

Similarly, the type of condensation domain correlated with success rates: LCL and E-DCL domains were linked to approximately 73% functional constructs, while CE domains showed only 24% success. However, the CE domain clusters also exhibited low phylogenetic relatedness to Chaiyaphumine, confounding direct attribution.

Phylogenetic relatedness, assessed by sequence similarity of combined thioesterase (TE) and thiolation (T) domains, was also correlated with production outcomes (Fig. 73). Modules with over 55% similarity produced peptides in 39 of 51 cases versus 20 of 54 below 26%. Despite these correlations, the approach remains limited because some related clusters lack essential amino acids or TE domains, indicating additional factors likely to influence hybrid NRPS functionality.

🔍

This parallelized setup streamlined the screening of our NRPS library, reducing hands-on time while maintaining reproducibility. It allowed us to quickly pinpoint productive constructs, which is a necessary prerequisite for all further steps like scale-up and detailed activity testing.

Outlook

• Screening the Library for Bioactivity

After confirming small-scale production, the next step toward identifying potential antibiotics is bioactivity screening against multidrug-resistant pathogens. Read more about it in the next section.

• Minimizing Tested Combinations with Dry Lab Approach

The results of our characterisation showed that not all combinations yielded a peptide producing NRPS. While phylogeny-guided module selection improved yields, it limited amino acid diversity. To overcome this, we focused on the key T–C–T domain interaction and developed a software pipeline to model condensation complexes and predict productive combinations. More information is provided on our software page.

• Upscaling the Production

Since larger-scale production is essential for developing and validating potential antibiotics, we upscaled peptide synthesis to obtain purified reference compounds for quantitative analysis (Measurement). Achieving a 100-fold increase, we cultivated litre-scale cultures for four peptides, demonstrating the scalability and biotechnological potential of our platform, which we further explored for industrial applicability in our human practices work (Interview with Prof. Dr. Raphael Reher).

Screening our Parts for Bioactivity

Key Points

• Screening of 93 nonribosomal peptides (NRPs) for antimicrobial activity against seven bacterial strains, including clinically relevant ESKAPE pathogens.

• Identification of four hit compounds exhibiting inhibitory activity.

• Further derivatization of one hit compound, resulting in bioactivity against Staphylococcus aureus (MRSA) and Enterococcus faecialis.

Graphical Abstract

Created in BioRender. Solakoudi, E. (2025) https://BioRender.com/xz91jvy

Introduction

After characterizing our parts, the next step was to determine which of the novel compounds display antimicrobial activity. we adapted and refined a screening method that rapidly identifies the most promising candidates.

Test Organisms

To explore the bioactivity of the produced nonribosomal peptides, we carefully selected a panel of relevant organisms, covering the relevant pathogens. In some cases, however, we substituted safer or more accessible alternatives. Alongside the classical ESKAPE pathogens (excluding Enterobacter), we included methicillin-resistant Staphylococcus aureus (MRSA), Escherichia coli (sometimes considered part of an “extended ESKAPE(E)” grouping^[16], and Bacillus spizizenii as a reference organism for inhibitor substance testing (Fig. 74).

🔍

All organisms used in this study were handled under biosafety level 2 (BSL-2, S2) conditions. The strains are opportunistic or non-pathogenic in laboratory settings and can be safely managed in BSL-2 facilities by following standard microbiological precautions. To learn more about the work with S2 organisms visit our safety page.

Bioactivity Screening Assays

To apply the test compounds onto the bacteria, we evaluated two different approaches, each with its own advantages, disadvantages, and technical challenges (Fig. 75).

• Drop-spot assay

In this method, 2 µL of extract was pipetted onto predefined agar locations and allowed to dry for adherence. The approach is simple and requires minimal sample volume but can be inconsistent due to droplet spreading, uneven agar surfaces, or air bubbles, which may lead to overestimated inhibition zones.

• Filter disk assay

Here, sterile filter disks were placed on predefined locations, and 10 µL of extract was applied to each. Capillary action ensured uniform distribution and improved consistency, though the method required more sample and time. Filter disks could obscure small inhibition zones or adsorb compounds, but overall, this approach proved more suitable for high-throughput screening.

🔍

While we initially tested both drop-spot and filter disk assays, the drop-spot approach proved inconsistent and prone to artifacts, leading us to discard it for further work. Instead, the filter disk assay emerged as the method of choice: it provided more reproducible inhibition zones and offered the robustness required for large-scale testing of the NRPieceS library.

Results

Based on the results from our parts characterization, we selected the functional NRPS combinations and upscaled the production cultures (50 mL) to increase the extract yield, ensuring sufficient compound concentration for subsequent screening (Expression). The extracts were concentrated 500-fold relative to the original culture volume (100 µL DMSO extract from 50 mL culture) and pipetted onto bacterial lawns using either a dropspot or filter plate assay (Protocol).

Alongside the test compounds, we always included a positive control with gentamicin at 10 mg/mL and negative controls using extracts from uninduced cultures, extracts from empty-vector strains to exclude host-derived peptide production, and DMSO alone to assess solvent toxicity.

False-Positives due to Antibiotic Coextraction

When we conducted the first bioactivity tests with our peptide extracts we observed a surprisingly high number of positive results with nearly all extracts showing activity against certain strains (Fig. 76).

🔍

These observations raised concerns that many apparent hits were false positives. Upon reviewing our experimental design, we hypothesized that antibiotics used in the production cultures to maintain selection pressure were co-extracted with our peptide compounds. These residual antibiotics may have inhibited bacterial growth during testing, thereby confounding the results.

Hence, we designed an experiment testing exactly this by varying the antibiotic concentration in the production medium by 0 %, 33 % and 100 % (Fig. 77). The results clearly confirmed our hypothesis as the samples: samples with 100% antibiotics showed clear inhibitory zones, the effect was reduced at 33%, and absent at 0% antibiotic concentration.

🔍

We also assessed NRPS production and found it was not affected by the absence of selection pressure, leading us to omit antibiotics from the production medium thereafter (Fig. 78).

🔍

Testing our Peptide Library

Bioactivity screening was conducted on our peptide library generated from the intein-shuffled native clusters Chaiyaphumine, Szentiamid, and Xentrivalpeptide. All crude extracts were evaluated using a filter disk diffusion assay against selected bacterial test strains. Although 13 of the 27 nonribosomal peptides (NRPs) were confirmed by LC-MS analysis, none produced detectable growth inhibition zones under the tested conditions (Fig. 79).

🔍

The plate images from the bioactivity screening of the intein shuffled library are provided here.

From our Golden Gate library of 105 hybrid NRPS constructs, 63 successfully produced detectable peptides. We upscaled the production of all 63 variants for bioactivity testing, though two failed to produce peptides under these conditions.

🔍

All extracts were screened for antimicrobial activity against the selected test strains using a filter disk assay. For this screening, we excluded the second S. aureus strain (ATCC 29213) to conserve material, focusing instead on the more resistant and clinically relevant MRSA strain ATCC BAA-1717. The plate images from the bioactivity screening of the Golden Gate library are provided here.

Excitingly, sample 25 showed a clear inhibition zone on Bacillus spizizenii plates. Since B. spizizenii is highly sensitive even to trace antibiotic concentrations, this result strongly suggested genuine antimicrobial activity - marking our first identified hit compound from the NRPieceS peptide library (Fig. 80).

The native Chaiyaphumine showed no detectable inhibitory activity in our bioassays, whereas the derivatized extract 25 produced a clear inhibition zone, suggesting that the observed bioactivity resulted from the derivatization (Fig. 81). In this variant, a LCL XUT^I module was inserted in the termination cassette (BBa_2525JBI3). This led to the substitution of proline with valine and prevented epimerization of the upstream alanine. Whereas the upstream amino acid in native Chaiyaphumine normally undergoes conversion, the derivatized variant retained the L-configuration.

These modifications possibly contributed to the inhibitory zone of the derivatized variant.

🔍

Optimization of Hit Compounds via Biosynthetic Derivatization

🔍

Building on our identified hit and leveraging the tools of the NRPieceS Plasmid Collection, we aimed to further derivatize the compound with the goal of enhancing its bioactivity.

Therefore, we combined the LCL-V Chaiyaphumine terminator (BBa_2525JBI3) with additional modified initiator and elongator modules from our Golden Gate Plasmid Collection. The modules were selected according to three criteria: (1) confirmed peptide production in the library characterization, (2) representation of both canonical amino acids and non-standard residues such as β-alanine, and (3) inclusion of XUT^I modules exhibiting both the presence and absence of epimerization domains (Fig. 82).

Several Chaiyaphumine variants carrying two exchanged amino acids exhibited greater antimicrobial activity than the initial hit. In particular, variants 3, 6, 7, and 9 produced clear zones of inhibition against Staphylococcus aureus (MRSA) and Enterococcus faecalis (Fig. 83), with 7 and 9 showing the strongest bioactive inhibition. Independent replicate assays reproduced these outcomes, underscoring the robustness of the modification approach. Notably, all four active derivatives were altered at the initiator position, whereas modifications in elongator modules yielded no detectable activity, suggesting that the XUT^I‑2 site is especially favorable for enhancing bioactivity.

Even though, the results were reproducible on S. aureus and E. faecalis for the double derivatized Chaiyaphumines, the observation that the 'hit' compound caused an inhibitory zone on B. subtilis was not reproducible. Therefore, these results need further validation with replicates. The plate images from the bioactivity screening of the derivatives of the hit compound are provided here.

🔍

Interestingly, among the ESKAPE(E) pathogens, the double-derivatized NRPs exhibited activity only against S. aureus and E. faecalis, the two Gram-positive members of the group. In contrast, Gram-negative bacteria such as Klebsiella pneumoniae possess an additional outer membrane that restricts compound uptake and diminishes antibiotic efficacy. To overcome this barrier, we developed a complementary drug-delivery strategy within our project (Drug Delivery).

Furthermore, these results serve as a proof of concept that derivatized NRPS fragments can be recombined not only with their native clusters but also with one another.

Outlook

• Determining the MIC

The next step was to determine the minimum inhibitory concentration (MIC) of the most promising extracts or purified compounds. While previous assays provided qualitative results, MIC testing enables quantitative potency comparison with known antibiotics. To support this, we established a peptide concentration assay, learn more on our measurement page.

• Screening for Stress-Inducing NRPs Using a Pro moter-Based Approach

To further assess the antimicrobial potential of our NRPs, the 2018 Leiden iGEM team’s strategy could be applied. They used stress-activated promoters to detect compounds inducing cellular stress, including membrane disruption or proteotoxicity, allowing identification of NRPs with non-classical modes of action. By exposing bacteria to our extracts and monitoring stress-responsive gene activation, peptides triggering stress responses can be detected, supporting the discovery of novel treatments against antibiotic-resistant pathogens. For details, see iGEM Team Leiden 2018.

• Determination of Mode of Action

With more data, structural or experimental patterns linked to improved antimicrobial activity may be identified, guiding peptide design and refining culture or extraction conditions.

• Cytotoxicity Assay

Our NRPs may not only target microbes but could also be toxic to human cells, making cytotoxicity an important consideration. To assess this, we planned to employ the xCELLigence RTCA (Real-Time Cell Analysis) assay, which continuously monitors the electrical impedance of cells grown on microelectrode plates. Healthy, adherent cells maintain high impedance, whereas cytotoxic effects reduce it, providing a dynamic, label-free measurement of cell viability in real time.

• Finding Patterns for Improved Bioactivity

With more data, structural or experimental patterns linked to improved antimicrobial activity may be identified, guiding peptide design and refining culture or extraction conditions.

• Expanding the Tested Organisms

Our current testing panel included a broad but limited range of organisms. Expanding to additional clinically relevant species could further evaluate peptide bioactivity supporting the development of rare, narrow-spectrum antibiotics.

• Automation of Testing Methods

Our activity assays involved extensive manual handling. Partial automation could improve reproducibility, reduce errors, and enable higher-throughput testing.

Drug Delivery

Key Points

• Integration of an azide handle for click chemistry into a non-ribosomal peptide by NRPS Engineering.

• Functionalization of an NRP by conjugation to a siderophore and a fluorescent tag via click chemistry.

Graphical Abstract

Introduction

Our project focuses on producing peptides with potential antibiotic activity. A major obstacle for many antimicrobial compounds, particularly against gram-negative bacteria, is the bacterial cell envelope, which often prevents molecules from reaching their intracellular targets.

To address this, we employed the “Trojan Horse” strategy which disguises our potential peptide antibiotics by conjugating them to siderophores. The siderophore binds iron and is recognized by specific bacterial transporters for nutrient uptake, enabling the attached peptide to enter the cell (Fig. 84). This approach leverages a natural bacterial survival mechanism to enhance drug delivery. A central challenge in our project was introducing a functional “handle” into nonribosomal peptides (NRPs) that allows conjugation to siderophores via click chemistry, forming the foundation of our targeted delivery strategy.

🔍

To achieve this, we systematically explored different engineering strategies - from feeding small-molecule analogs, to constructing exchange units, to precise A-domain swaps - each representing a new iteration in the design-build-test-learn (DBTL) cycle. Alongside, we also developed a chemical method to synthesize siderophores carrying reactive groups for conjugation. The following results outline our experimental journey, including both setbacks and successful breakthroughs, and illustrate how rational engineering and chemistry can converge to expand the NRPieceS library.

Results

Exploring Siderophore Conjugation via Starter Domain Promiscuity

Since we needed a way to conjugate the NRP to a siderophore, we considered exploiting the natural promiscuity of certain NRPS domains. We began by investigating the native clusters in the NRPieceS library and noticed that both Chaiyaphumine and Xentrivalpeptide possess a starter condensation (C) domain that naturally incorporates acetic acid (C2) as well as phenylacetic acid (PAA) (Fig. 85). This suggested that the domain might exhibit substrate flexibility, potentially allowing direct attachment of a siderophore to the nonribosomal peptide.

🔍

To explore this, we defined three key requirements for a suitable siderophore candidate: it should contain an iron-chelating group, possess a carboxyl group compatible with condensation, and structurally resemble acetic or phenylacetic acid to maximize the likelihood of successful incorporation.

All these criteria are met by DOPAC (3,4-dihydroxyphenylacetic acid), which contains both a catechol group for iron chelation and a carboxyl group for condensation (Fig. 86). Its close structural similarity to phenylacetic acid is particularly advantageous, as it increases the probability that the C-domain will recognize and incorporate DOPAC in place of its natural substrate.

🔍

We supplemented the production medium with DOPAC and analyzed the extracts regarding the incorporation of the siderophore by the initial condensation domain (Fig. 87). However, LC-MS analysis revealed that the starter C domain did not accept the siderophore as a substitute substrate and only the native chaiyaphumine A and D were produced. The experiment was also conducted with Xentrivalpeptide and yielded the same result.

🔍

This result indicates that the domain exhibits more restricted substrate specificity than anticipated, highlighting the need for either a modified conjugation strategy or chemical tailoring of the siderophore to overcome this limitation.

Feeding a Handle for Click Chemistry

As direct siderophore feeding proved unsuccessful, we explored an alternative conjugation strategy. Instead of using the starter condensation domain for direct siderophore incorporation, we repurposed it as an entry point for introducing a chemical handle that could later enable siderophore attachment via click chemistry (Fig. 88A). Since this domain naturally accepts phenylacetic acid, we hypothesized that it might also tolerate structurally related phenylalanine derivatives, such as 4-azido-L-phenylalanine (AzF), which provides an azide group suitable for click-based conjugation (Fig. 88B & C).

🔍

We supplemented the production medium with 4-azido-L-phenylalanine (AzF) and analysed the extracts by LC-MS (Fig. 89). The results show that the starter C domain did not accept AzF as a substitute substrate, as only the native chaiyaphumine variants were detected. The same outcome was observed for Xentrivalpeptide, indicating that neither system tolerated incorporation of the modified precursor.

🔍

These results demonstrate that the starter C domain, while naturally able to process small substrates like AA or PAA, is not permissive toward bulkier or modified analogues. This implies conjugation through the native starter C domain is not a viable strategy and motivating a shift toward alternative approaches for introducing handles in NRPS.

Engineering NRPS for the incorporation of a handle for Click Chemistry

As described in the NRPS Engineering, swapping NRPS modules or units allows redirection of the amino acid incorporated into the peptide. Having established that feeding the starter condensation domain was not a viable strategy for incorporation AzF handle, we turned to promiscuous adenylation domains, which are known to accept chemically modified substrates such as AzF^[18].

Building on the modular design of the NRPieceS library, we created a donor vector carrying an AzF-specific exchange unit from Photorhabdus luminescens TT01 for integration at three positions within the native Chaiyaphumine cluster (Fig. 90).

🔍

For incorporation of this unit into the initiator, LC/MS analysis showed no detectable AzF (Fig. 91) Notably, even in the absence of AzF feeding, phenylalanine—despite being present in the reaction mixture—was incorporated at a substantially lower level than expected. We therefore hypothesized that AzF may have been inserted into the growing peptide chain, but the NRPS complex was unable to release the intermediate. As a result, the process stalled, preventing further peptide production with either AzF or phenylalanine. Since similar domains have previously been reported to accept AzF, the most plausible explanation seems to be that downstream processing interferes with elongation, highlighting the need for refined integration strategies.

🔍

When incorporated into the elongator, the system yielded a non-functional NRPS, as neither phenylalanine nor AzF was incorporated (Fig. 92).

🔍

When incorporated into the terminator in place of proline, LC–MS analysis revealed a low-intensity peak consistent with AzF incorporation (Fig. 93). At the same time, the phenylalanine peak was greatly reduced, and several additional peaks with similar masses appeared. These overlapping signals make it difficult to clearly identify the intended product and suggest that abnormal interactions or faulty processing may disrupt elongation. The NRPS might also stall, producing truncated peptides. Such incomplete products can have masses similar to the target peptide, further complicating analysis and indicating that the product mixture is not suitable for further use.

🔍

Introduction of the AzF exchange unit in the last position

To address the issue of failed elongation observed in earlier designs, we placed the AzF exchange unit at the final position of the Chaiyaphumine cluster. By exchanging the last elongation unit via Gibson Assembly, the construct contained only the terminal thioesterase (TE) domain downstream, which minimized processing steps and was expected to reduce steric interference during peptide synthesis. This design aimed to maximize the chance of successful AzF incorporation by ensuring that the chemical handle was introduced immediately before peptide release.

🔍

When testing the new construct, it became clear that phenylalanine was incorporated in place of AzF, indicating that even at the terminal position the exchange unit was not permissive for the modified substrate. Despite rational redesign and targeted placement, incorporation still failed (Fig. 95), highlighting that whole-unit exchanges may not preserve essential condensation interfaces and that alternative engineering strategies will be necessary moving forward.

🔍

Exchanging the terminal A domain

Adenylation (A) domains are the key determinants of substrate specificity in NRPS systems, making them a logical target for engineering. Unlike whole-module exchanges, which often disrupt condensation interfaces, direct A-domain swapping can redirect substrate incorporation while preserving functional context. Guided by this principle, we aimed to incorporate 4-azido-L-phenylalanine (AzF) into Chaiyaphumine by replacing its terminal A domain.

🔍

This strategy was motivated by two factors: (i) placing AzF at the peptide terminus minimizes elongation challenges, and (ii) literature precedent shows that even a single mutation in phenylalanine-specific A domains can enable activation of non-natural aromatic amino acids functionalized with azides or alkynes.

In particular, Kries et al. reported that the GrsA A-domain can be engineered to accept AzF with improved efficiency^[19]. In discussions of our drug delivery concept, Dr. Hajo Kries further emphasized that A-domain targeting is among the most robust routes toward substrate engineering.

Guided by these insights, we constructed two derivatives of Chaiyaphumine: one in which the terminal A domain was replaced with the naturally promiscuous GxpS A-domain (BBa_255Z1HPP), and another with the engineered GrsA variant (BBa_25EW3Z6A). Domain swaps were carried out using XUT^I and XU recombination sites (Fig. 97), which flank the native A domain and allow precise excision and insertion of engineered substitutions (A domain exchange).

🔍

When expressed in the presence of AzF and analyzed by LC–MS, both constructs successfully incorporated the chemical handle, with higher incorporation efficiency observed for the GxpS domain than for the engineered GrsA. (Figs. 98 & 99).

🔍

🔍

After several unsuccessful attempts in earlier iterations, this approach successfully produced a Chaiyaphumine variant containing AzF. These results show that targeted A-domain exchange is an effective and reliable method for introducing a click chemistry handle into NRPs, providing a foundation for conjugating the peptide with a siderophore to facilitate efficient drug delivery.

Chemical Synthesis of a Siderophore

Having successfully introduced the handle into the NRP, the next step was to create a suitable conjugation partner. For this, we chose to chemically synthesize a catechol derivative bearing an alkyne group.

To conjugate our peptide to a siderophore via click chemistry, a siderophore with a free terminal alkyne group was needed. We decided to perform an amide coupling of protocatechuic acid (1) and propargyl amine (2) to synthesize the amide (3), which contains a catechol and an alkyne moiety (Fig. 100).

🔍

🔍

We chose this approach because amide couplings (especially with primary amines) are usually very reliable. Nonetheless, the reaction failed twice when performed at room temperature in the standard solvents dimethylformamide (DMF) and dichloromethane (DCM). Only after these attempts did we find a patent in which this synthesis was described with a slightly more unusual method - the authors used acetonitrile as the solvent, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) as coupling reagent and heated the reaction mixture to 80 °C under reflux conditions^[20].

We adjusted this protocol by exchanging EDC with the comparable N,N’-dicyclohexylcarbodiimide (DCC) and using the cheaper triethylamine instead of a second equivalent of propargyl amine as base. The reaction scheme is shown in figure 100. After column chromatography (Methanol/DCM) and solvent evaporation under reduced pressure, product 3 (225 mg, 1.18 mmol, 73%) was obtained as a yellow solid (Fig. 101).

The yellowish color of our product suggests that it was not completely pure - but color can often be misleading in organic chemistry, as very small amounts of highly colored impurities can result in very visible color changes. Indeed, a 1H-NMR spectrum of the product showed only methanol as a relevant impurity (Fig. 102).

🔍

For more information about this synthesis, including a more detailed analysis of all 1H-NMR peaks, please refer to our protocols page.

Click Chemistry Reaction

🔍

Over the past 25 years, click chemistry has transformed chemical and biological research, including synthetic biology projects within iGEM. Recent teams, such as HunanU 2024, have incorporated azidophenylalanine via an amber suppressor system to enable precise protein linkage. However, no iGEM team has yet combined click chemistry with NRPS systems.

In our iGEM project, click chemistry served as a key tool for site-specific functionalization of the engineered compound Chaiyaphumine within our biosynthetic pipeline. This enabled its conjugation to two molecular tags: a chemically synthesized siderophore for potential drug delivery and a fluorescent FAM-alkyne, a carboxyfluorescein derivative, tag for visualization of i.e. cellular uptake in biological assays (Fig. 103).

To achieve reliable and high-yielding click reactions, we implemented an improved variant of the copper-catalyzed azide–alkyne cycloaddition (CuAAC). This optimized protocol involves the in situ generation of catalytically active Cu(I) via reduction of copper(II) sulfate (CuSO₄) with sodium ascorbate, in the presence of the Cu(I)-stabilizing ligand THPTA (tris(3-hydroxypropyltriazolylmethyl)amine). This method effectively prevents the oxidation of Cu(I) by atmospheric oxygen, leading to more consistent and reproducible reaction conditions^[21][22].

We scaled up production of the AzF–Chaiyaphumine variant and performed the click reaction using crude extracts following the established protocol. LC–MS analysis revealed strong signals corresponding to the expected conjugates, confirming successful reactions and demonstrating that both the siderophore and fluorophore were successfully clicked to Chaiyaphumine (Fig. 104).

🔍

Through this work, we provide a robust and accessible CuAAC protocol tailored for the iGEM community, especially for teams interested in the chemical functionalization of biomolecules within synthetic biology and biomedical applications.

Outlook

• Bioactivity Testing of Siderophore-NRP Conjugate

Although we achieved siderophore–NRP conjugation and confirmed its feasibility, time constraints prevented testing for improved uptake. Future work should assess the biological activity of these conjugates, focusing on compounds with known activity to determine whether conjugation enhances uptake or extends efficacy to Gram-negative bacteria.

• Testing the Cellular Uptake

To validate the concept, a reliable method is needed to confirm that the siderophore–NRP conjugate is actively taken up by bacterial cells. This could be achieved using fluorescent or isotopically labeled NRPs for visualization or quantification. Notably, some siderophores such as P. aeruginosa pyoverdine are naturally fluorescent^[23], offering a simple way to monitor uptake without additional labeling - e.g., by linking pyoverdine to a peptide antibiotic.

• Expanding the Conjugation Partners

The ability to link virtually any molecule to an NRPS-derived peptide via click chemistry unlocks broad applications. Beyond siderophores, the principle can be expanded to heme, sugars, nucleobases, vitamins, lipids, or fatty acids. The approach also shows promise for targeted drug delivery against cancer, fungal pathogens, parasites, and agricultural pests, enabling tailored uses across medicine and biotechnology.

• Producing the Siderophores by NRPS

Since siderophores can also be synthesized by NRPSs, it may be possible to replace chemical synthesis of the conjugation partner with a biotechnological approach. Possibly NRPS-linked siderophores could be produced by engineering and recombining naturally occurring biosynthetic clusters.

Conclusion

Our wet lab work set out to address one of synthetic biology’s long-standing challenges: the engineering of nonribosomal peptide synthetase (NRPS) systems to access new chemical diversity and unlock novel bioactive compounds. Through a stepwise yet interconnected design-build-test-learn cycle, we developed a comprehensive modular framework that integrates DNA assembly, protein engineering, high-throughput screening, and chemical conjugation into a cohesive antibiotic discovery platform.

Beginning with Cluster Selection, we successfully expressed three cyclic peptide-producing NRPS clusters from Xenorhabdus in E. coli, establishing a robust heterologous expression system based on split-intein reassembly after expression of orthogonal plasmid backbones in E.coli. This achievement overcame the traditional bottlenecks of NRPS size and complexity, validating E. coli as a tractable host for otherwise inaccessible gene clusters.

Through Designing and Building our Parts, we created the NRPieceS system-a modular NRPS engineering toolkit composed of 9 acceptor and 35 donor modules, enabling combinatorial reassembly of biosynthetic units. By cloning 105 hybrid constructs, we proved that complex NRPSs can be functionally recombined, paving the way for large-scale diversification of peptide structures. The scalability of this platform expands the accessible peptide chemical space to millions of potential combinations, offering unprecedented opportunities for antibiotic innovation.

In Characterizing our Parts, we demonstrated that this modularity translates into functional peptide biosynthesis. 95 new-to-nature peptides were successfully produced and analyzed, confirming that our three-plasmid intein- Golden Gate based expression system supports efficient combinatorial screening. The insights gained: such as position-dependent module compatibility and phylogenetic approach form the basis of predictive NRPS design principles.

The Screening stage translated molecular diversity into biological function. Our high-throughput assays against multiple bacterial strains identified four active compounds, including derivatives with inhibitory effects on methicillin-resistant Staphylococcus aureus (MRSA) and Enterococcus faecalis. The iterative refinement of our assay design, from antibiotic-free production cultures to improved bioassay formats, exemplifies the adaptive optimization required in translational biosynthetic research.

Finally, through the Drug Delivery module, we extended NRPS engineering toward functionalization and targeting. By incorporating an azide handle for click chemistry and conjugating peptides to siderophores, we demonstrated the feasibility of tailoring NRPs for specific delivery pathways. This strategy achieves targeted delivery and evasion of bacterial resistance mechanisms.

Taken together, this work establishes an integrated, modular, and scalable platform for nonribosomal peptide engineering, bridging the gap between natural product biosynthesis and rational antibiotic design. It not only demonstrates that complex NRPS systems can be reassembled and engineered in E. coli, but also that their products can be systematically characterized, screened, and functionally modified. Our results collectively advance the vision of a universal NRPS chassis for discovery and development of next-generation antimicrobial agents.