In our first cycle, we aimed to generate functional parts from the pysp1 template that could be used to construct synthetic pyriform silk genes. To achieve this, we first assessed the structural composition of PySp1 using multiple AlphaFold v3 predictions (Abramson et al., 2024). After consulting Dr. Florian Hof and Prof. Dr. Lukas Stelzl on AlphaFold predictions and disordered proteins, we concluded that the modularity of PySp1 corresponds to its domains and DNA sequences.

Based on this, we divided pysp1 into three modular parts, NTE (N-terminal element), RPE (repeat element), and CTE (C-terminal element), that could later be assembled into synthetic pyriform silk genes. The sequences were codon-harmonised for B. subtilis W168 to reduce tRNA depletion and preserve co-translational folding pauses, thereby increasing the likelihood of producing a functional PySp1 protein.

A key challenge in assembling highly repetitive DNA sequences is that PCR-based methods often fail due to mispriming and polymerase slippage. To avoid these issues, we deliberately designed our cloning strategy to be PCR-free. Hence, we developed Pyricloning, a modular cloning strategy combining elements of RFC25 and the RFC1000 MoClo framework to generate higher-order repeat oligomers without compromising flexibility in downstream assembly. After discussing our approach with Dr. Farley Kwok van Giezen and considering alternative methods for assembling repetitive sequences, we added specific restriction enzyme recognition sites to our parts.

We implemented Pyricloning to iteratively construct repeat oligomers from dimers up to octamers. These oligomers were then used to assemble CDS constructs with varying repeat numbers.

Restriction digests, agarose gel electrophoresis, and sequencing validated our assembly strategy, with distinct bands corresponding to different repeat oligomers (Figure 2) and CDS constructs (Figure 3). Additionally, we computationally confirmed that codon harmonisation produced a distribution closer to the native profile by comparing codon usage in our sequences with that of B. subtilis W168, and contrasting it with the wild-type A. ventricosus codon usage.

Pyricloning enabled robust, PCR-free construction of repeat oligomers up to octamers (and, in theory, beyond), which could be seamlessly integrated into complete CDS constructs. However, we observed occasional recombination events during repeat generation, likely due to using a recA⁺ E. coli strain rather than issues with the assembly strategy itself. These artifacts did not appear in later assemblies, but the experience prompted us to carefully validate construct sizes after each cloning step.

Codon harmonisation produced a distribution closer to the native sequence, which should reduce the risk of tRNA depletion and misfolding during translation. While we cannot directly quantify its effects, literature suggests that harmonisation increases the likelihood of successful protein expression, thus as an advantage we expect to benefit in future production attempts.

In recombinant spidroin production, the precise control of transcription and translation is crucial for success and overcoming common challenges such as low yields and the accumulation of insoluble protein, both frequently linked to limitations in transcription and translation (Connor, Zha & Koffas, 2024). To strengthen our approach, we consulted Dr. Georg Fritz regarding the MoClo strategy. He provided valuable guidance on how to efficiently utilize this methodology, offering feedback and resources such as detailed guides and plasmid materials to optimize our workflow.

Therefore, prior to assembling complete transcription units with our CDS constructs, we evaluated transcriptional and translational regulatory elements to identify combinations that balance protein yield with cellular stress. We selected candidate promoters (PxylA, PbceA, PliaI, PhpaII) and ribosome binding sites (st4, st7, st11, wk8) that are well-characterized in B. subtilis. As a reporter, we used sfgfp to quantify expression as the protein matures rapidly with a folding time of 6 min in E. coli and is highly stable, thereby producing consistent fluorescence signals (Pédelacq et al., 2006; Shields et al., 2019).

We assembled various reporter transcription units (promoter–RBS–sfgfp–terminator) using the MoClo RFC1000 framework. To ensure reliable comparisons, the constructs were genomically integrated in a single-copy format.

Reporter strains were analysed using plate-reader fluorescence measurements and single-cell fluorescence microscopy. Both static end-point and kinetic measurements were performed, with time courses extending up to 60 min post-induction.

We observed a broad dynamic range of expression strengths across different promoter–RBS combinations, providing the flexibility to fine-tune constructs for future production attempts. PliaI emerged as the strongest inducible promoter, while PxylA displayed the weakest activation. Among the RBSs, st11 supported rapid and robust translation, whereas st7 exhibited slightly slower kinetics, which may reduce cellular stress during expression of repetitive proteins. These results formed the basis for promoter and RBS choices in subsequent spidroin transcription units.

Building on the results of the second cycle, we assembled complete transcription units (TUs) with our CDS constructs. To achieve high yields, we chose the strong promoter PliaI. As RBS, we selected st11, aiming for high yields while accounting for stress with a st7 variant with slower kinetics.

The Level-1 transcription units were then assembled into Level-M constructs containing homologous flanks for the amyE locus of B. subtilis W168 and a chloramphenicol resistance cassette for selection.

Because our CDS constructs from Pyricloning were already compatible with the RFC1000 MoClo framework, we were able to directly assemble level-1 TUs with our previously build CDS. Our first TUs contained CDS with monomeric RPEs to assess the production of the simplest constructs. Later we also build CDS with tetrameric and octameric RPEs, which were also used to assemble functional TUs. Successfully crafted level-1 transcription units were then used for Level-M (2) assemblies, which were subsequently integrated into the B. subtilis W168 genome via homologous recombination at the amyE locus.

We validated the Level-1 and Level-M assemblies by restriction digests, agarose gel electrophoresis, and whole-plasmid sequencing. Successful genomic integration was confirmed using a starch hydrolysis assay: colonies with disrupted amyE no longer degraded starch, and indeed, colonies without hydrolysis halos indicated correct integration.

This cycle demonstrated that our cloning strategy was both robust and efficient for assembling synthetic pyriform silk genes into functional transcription units. The constructs proved stable across multiple assembly steps and integration events. These results also gave us confidence that the approach could be quickly adapted to address future challenges, allowing us to proceed with the first production tests using integrated spidroin constructs.

For our first production attempts, we tested two constructs driven by the PliaI promoter. Each carried a CDS with a monomeric RPE, the L3S1P47 terminator, and a hexahistidine tag for purification and Western blot detection. Since our earlier sfgfp reporter tests showed a robust and strong translation initiation for both the st11 and st7 RBS, we chose them to explore the impact of translational fine-tuning on expression of repetitive proteins.

In addition, we designed the cultivation process to probe the impact of growth conditions on spidroin production. Specifically, we compared two different media (LB and 2×YT) to test whether nutrient availability affected expression. We also reduced the cultivation temperature from 37°C to 18°C after induction in one batch to slow metabolism and protein biosynthesis, aiming to alleviate cellular stress associated with spidroin production.

The transcription units were genomically integrated into the amyE locus of B. subtilis W168, and production was tested in both LB and 2×YT medium.

After cultivation, cultures were harvested, and the pellets processed for SDS–PAGE and Western blot analysis. No spidroin was detectable with either method.

Intracellular expression of spidroin constructs proved insufficient regardless of RBS or induction strength. Possible explanations include proteolytic degradation triggered by cytoplasmic stress or expression levels below the detection limits of standard methods. These findings prompted us to adopt a new strategy: increasing gene dosage with a multi-copy approach and using secretion signal peptides to bypass intracellular stress and proteases.

Guided by our consultation with Dr. Jolanda Neef, we implemented her input by converting the pBSMuL1 vector into a MoClo-compatible entry vector containing an mrfp1ΔlacO red/white selection cassette to enhance gene dosage and enable multi-copy expression in Bacillus subtilis W168. To test the functionality, we first cloned sfgfp constructs under two promoters, PhpaII (constitutive) and PxylA (previously described as inducible).

After transforming B. subtilis W168, we evaluated expression and growth using plate-reader assays and fluorescence microscopy.

The system functioned as expected. Nevertheless, against the expectations, PxylA behaved as constitutive in the multi-copy plasmid context, likely because the cognate repressor is encoded only in the chromosome and was insufficient to control the large number of promoter copies. Based on this observation, subsequent designs employed PliaI.

After confirming sfgfp expression, we applied the strategy to pyriform silk constructs. Literature review revealed that secretion signals vary in performance depending on the target protein. Following advice from Prof. Dr. Thomas Wiegert, we incorporated the ssyoaW system, which includes a StrepII–SUMO tag for enhanced folding, alongside sslipA from the pBSMuL1 vector for comparison.

We designed four new constructs: monomeric and tetrameric RPE variants, each with either ssyoaW or sslipA, all driven by PliaI. These constructs were assembled in our new vector and transformed into B. subtilis W168.

Expression was evaluated in LB cultures, with samples analysed via SDS–PAGE and Western blot across multiple cellular fractions and the supernatant.

Despite increased gene dosage and secretion tags, no detectable pyriform silk was observed in either intracellular or extracellular fractions. After consulting Prof. Dr. Tracy Palmer, we hypothesised that overexpression of such a large, complex protein in a multi-copy context overwhelmed the Sec secretion system, triggering protease upregulation and degradation. The high proportion of unfolded regions in spidroins makes them particularly vulnerable to proteolysis.

This outcome highlighted the need for more sensitive detection methods and guided us toward exploring the split NanoLuc assay in the following cycle.

Following valuable feedback from Prof. Dr. Tracy Palmer, we decided to take a step back and return to a single-copy genomic integration approach. This time, we designed constructs carrying secretion signal sequences to facilitate protein export without overwhelming the Sec pathway, which is essential for B. subtilis viability (Ling et al., 2007; Souza et al., 2021).

In parallel, we explored the split NanoLuc assay (Yang et al., 2024; Pereira et al., 2019) for highly sensitive detection and consulted Dr. Margarethe Schwarz (Promega) regarding the Nano-Glo® HiBiT Extracellular Detection System. Based on these discussions, we designed new constructs that included a small C-terminal tag, pep86 (11 amino acids), which complements Promega’s LgBiT fragment to form an active luciferase. Combined with the N-terminal ssyoaW secretion signal provided by Prof. Dr. Thomas Wiegert, this setup offered a more sensitive detection method compared to traditional Western blots.

We generated three new constructs (BBa_255CP6QO, BBa_25VXKHEZ, BBa_25AE0AHE) containing a monomeric, tetrameric, or octameric RPE. Each construct included the PliaI promoter, RBS st11, the N-terminal ssyoaW signal peptide, and the C-terminal pep86 tag, as discussed with Prof. Dr. Tracy Palmer. These were genomically integrated into B. subtilis W168.

We cultivated LB cultures of all three strains and analysed the supernatant and cell pellets. Protein detection was carried out using SDS–PAGE (Figure 16A), Western blot (Figure 16B), and the newly adapted split NanoLuc assay (Figure 17) with Promega’s Nano-Glo® HiBiT Extracellular Detection System.

Using the sensitive split NanoLuc assay, we finally detected small amounts of pyriform silk protein, unlike with SDS–PAGE or Western blots. This confirmed that B. subtilis W168 is capable of producing pyriform silk, but also highlighted the need for significant optimisation in host engineering and downstream processing. Unfortunately, the yields remain far too low for biocementation prototyping.

From further discussions with Prof. Dr. Tracy Palmer and Dr. Florian Hof, our next steps include:

• Integrating the constructs into B. subtilis strains lacking extracellular proteases (e.g., MidiBacillus).
• Adding protease inhibitors post-harvest to reduce degradation.
• Engineering strains with enhanced availability of the tRNAs most required by our synthetic spidroin and increased levels of Sec pathway components.

Initial attempts to integrate the constructs into midi Bacillus failed, as cells underwent rapid lysis after cloning. We also see high potential in medium optimisation, but due to time constraints within the iGEM competition, further investigation and optimisation were not possible.

To evaluate the effect of biopolymers on microbially induced calcite precipitation (MICP)-based biocementation, we tested xanthan in a simple prototype setup that could later be adapted for spider silk once sufficient quantities are available. Based on the valuable feedback from Prof. Dr. Kevin Paine and Mr. Tim Gemünden, we specifically aimed for this setup of precast biocementation blocks, reflecting their advice to begin with small-scale, controlled applications before scaling toward larger structural use. We designed the screen to mimic a non-sterile static environment, comparable to a conventional concrete cast.

Xanthan was obtained by culturing X. campestris and purifying the secreted biopolymer. After lyophilisation, we directly used the product in our prototypes. To keep the setup simple and reproducible, we combined sand, pre-cultured bacteria, defined medium, and xanthan.

The biocementation process was then allowed to proceed for several days, during which the developing bricks were regularly supplied with fresh medium.

Our initial prototypes demonstrated that xanthan had a clear positive effect on both the mechanical properties and calcification of the bricks. This indicated that biopolymers can indeed benefit MICP-based biocementation. For future cycles we would like to change the biocementation towards a more complex but efficient process in flow cycle set up. Due to time constraints and limited resources, we were unable to pursue this in the wet lab before the end of the iGEM competition.