Engineering

Fig.1.0 Overview

Our project tackles biofouling, a phenomenon that significantly reduces ship speed and increases fuel consumption, through the production of zosteric acid.Through a multi-cycle Design–Build–Test–Learn (DBTL) framework, we systematically constructed and optimized an E. coli biosynthetic system for Zosteric Acid (ZA) production. In Cycle 1, we validated pathway feasibility by deleting cysH to enhance PAPS availability and assembling the full biosynthetic route. Although ZA was successfully detected, conversion efficiency remained low due to limited SULT1A1 activity and sulfur donor deficiency. Cycle 2 addressed the enzymatic bottleneck: rational mutagenesis identified the M12 variant with a 2.5-fold higher catalytic efficiency, and fusion-protein engineering demonstrated that flexible linkers such as (GGGGS)₂ and correct domain orientation could quadruple ZA yield. Cycle 3 overcame the sulfate-supply limitation by overexpressing the endogenous cysDNCQ enzymes, which established a PAPS-recycling loop and increased conversion efficiency nearly fourfold over control. Integrating these findings, Cycle 4 combined the key optimized elements into a next-generation "super strain" (25AIS-S128) through the co-expression of a SULT1A1(M12)-2GS-TAL fusion protein and the cysDNCQ precursor supply module.Finally, antifouling performance tests of ZA-enriched coatings confirmed substantial inhibition of marine biofilm formation across multiple assay systems, validating the industrial relevance of the engineered strain.

Design & Build

Part 1. cysH Deletion

Fig.1.1 Illustrations of design and build of cycle 1. (A) Schematic diagram of knockout of cysH using CRISPR Cas9 method to construct 25AIS-S16; (B) PCR verification of cysH knockout. Lane C represents the negative control (BL21(DE3)).1, 2, 3, 4 represents single colony samples on the plate; (C) Plasmid curing pEcCas and pEcRNA in strain BL21(DE3)△cysH; (D). Sequence Alignment Map of the cysH knockout region.

Based on Zhang et al. (2023)^[1], to reduce PAPS consumption, we knocked out cysH in E. coli BL21(DE3) using CRISPR-Cas9. A specific gRNA (GTTCCCATTTACGGGTTGTA TGG) was designed following Li et al. (2021) ^[2] and cloned into pEcRNA plasmid. After transformation pEcCas & pEcRNA and gene editing, the engineered strain 25AIS-S16 (BL21(DE3) ΔcysH) was obtained. Successful knockout was confirmed by PCR and sequencing (Fig. 1.1 B,D), and subsequent plasmid curing was verified (Fig. 1.1 C).

Part 2. Pathway Construction and Protein Expression

Fig.1.2(A) Gene circuits of 25AIS-S67, 25AIS-S65, 25AIS-S77, 25AIS-79 and 25AIS-24, which were independently transformed into E. coli BL21(DE3) for expression. (B) SDS-PAGE analysis of recombinant protein expression in E. coli BL21(DE3). Samples include BL21(DE3), 25AIS-S16, 25AIS-S67, 25AIS-S65, 25AIS-S77, 25AIS-S79, and 25AIS-S24 strains before (-) and after (+) IPTG induction. Target proteins are indicated by arrows: TAL and KlATPSL(~55 kDa), SULT1A1 (~30 kDa) and PcAPSK (~23 kDa) .

According to the metabolic pathway shown above(Fig. 1.0), we conducted strain 25AIS-S24 by co-tranformed two plasmid (pETdute-rSULT1A1-TAL and pRSFduet-KlATPSL-PcAPSK) with different ori, into our engineering strain 25AIS-S16. All plasmids were directly synthesized by GenScript.

To evaluate the protein expression of each construct, we performed induction under specific conditions (At an OD₆₀₀ of 0.6–0.8, 0.4mM IPTG was added for induction, and the cultures were shaken at 37 °C for 2 hours.) and observed the protein expression by SDS-Page.

Several constructs displayed bands at the expected molecular weights, although with varying intensities (Fig.1.2 B). The expression levels of TAL (~55kDa) and PcAPSK (~23kDa) remained stable during co-expression. KlATPSL(~56.5kDa) expression was consistently poor, evidenced by a weak band on the SDS-PAGE gel. This low expression level was observed in both strain S77 (expressing KlATPSL alone) and S79 (co-expressing it with other proteins), suggesting an inherent issue with its expression or stability. SULT1A1 (~30kDa) showed strong expression in strains S65 and S67, yielding a dark band on the SDS-PAGE gel. Conversely, its expression was markedly reduced in strain S24, where it was part of a four-enzyme co-expression system. The diminished band intensity suggests potential interference or cellular stress resulting from the simultaneous expression of multiple heterologous proteins.

Overall, these results provide preliminary evidence that some constructs can be expressed in E. coli, but the expression levels remain suboptimal, highlighting the need for further optimization of induction conditions or improvement of the expression system.

Test & Learn

Test

To estimate the amount of ZA produced, fermentation was carried out in M9 medium, supplemented with 2 mM L-Tyr, 5 g/L (NH₄)₂SO₄, 10g/L casein hydrolysate, and the appropriate antibiotic. We further analyzed the fermentation broth using HPLC, showcasing the production situation of ZA.

Fig.1.3 Illustration of ZA production (A). The results of HPLC. The top one shows the HPLC chromatogram of zosteric acid (ZA) standard, showing a major peak at around 7.91 min. The middle graph presents the chromatographic profile of p-coumaric acid (pHCA) standard, showing its principal peak around 8.84 min. The bottom graph shows the chromatographic profile of bacterial solution 25AIS-24, with the main peak detected around 7.89 min (ZA) and 8.83min (pHCA); (B). Standard curve of Zosteric acid (ZA) ;(C). Standard curve of p-coumaric acid (pHCA);(D). Comparison of ZA and pHCA concentration in 25AIS-24

HPLC analysis was performed to identify the substrate and product in our fermentation broth (Fig.1.3 A). First, we analyzed chemical standards to determine their retention times. The zosteric acid (ZA) standard eluted as a sharp peak at approximately 7.9 minutes (top panel). The p-coumaric acid (pHCA) standard eluted later, at approximately 8.8 minutes (middle panel). Next, the fermentation sample from strain 25AIS-S24 was analyzed (bottom panel). The chromatogram shows two main peaks. A large peak is present at 8.8 minutes, which matches the retention time of the pHCA standard, indicating a significant amount of unconsumed substrate. Crucially, a new, smaller peak appeared at 7.9 minutes, perfectly matching the retention time of the zosteric acid standard. This confirms that strain 25AIS-S24 successfully converted a portion of the pHCA substrate into the desired product, zosteric acid.

We quantified product and substrate concentrations using HPLC standard curves. The standard curves for both zosteric acid (ZA) and p-coumaric acid (pHCA) were highly linear (R² > 0.99) (Fig.1.3 B-C). Based on these calibrations, the final titer of ZA produced by strain 25AIS-24 was determined to be 0.2 mM (Fig.1.3 D).

According to results gained from HPLC, the amount of pHCA produced is greater than the production of ZA, showcasing a low conversion rate from pHCA to ZA.

Learn

SULT1A1 is a crucial part of ZA production, yet our experimental results, including SDS-Page analysis, revealed that its expression was suboptimal. Consequently, HPLC analysis also showed that the conversion rate of pHCA to ZA was low and the product ratio was not as expected. We analysed the results and found that low efficiency of SULT and sulfate donor deficiency might be the possible reasons for the low production of ZA.

Therefore, we began to re-examine these problems in search of an optimal solution. To improve SULT1A1 expression, we applied modeling to identify several potential mutant sequences (cycle 2). Introducing these mutations is highly likely to enhance SULT1A1 expression, thereby increasing ZA yield and ultimately achieving our goal of efficient ZA biosynthesis. We also constructed strains overexpressing cysDNCQ to increase the supply of SO₄²⁻ (cycle 3)

In summary, this round of experiments provided preliminary evidence for the feasibility of certain constructs, but the overall product yield remained unsatisfactory. These findings highlight the need for optimization in the next cycle, including adjusting induction conditions, improving the expression system, and enhancing the efficiency of the metabolic pathway.

All fermentation and induction procedures in this cycle were performed under the same conditions as described in Cycle 1, including culture medium composition, IPTG induction parameters, and incubation duration.

Based on the findings of Cycle 1, low expression and activity of SULT1A1 were identified as major bottlenecks for ZA biosynthesis. To address this challenge, we designed two engineering routes based on our modeling strategies.

Cycle 2.1: Single enzyme modification

Design & Build

Using computational modeling tools such as ProtSSN and Rosetta ΔΔG stability analysis, we identified several key residues potentially influencing substrate affinity, catalytic activity, and enzyme stability. Mutant libraries were designed by introducing targeted point mutations at residues including Y42, Y236, P250, and T256, which were prioritized based on predicted ΔΔG changes in protein stability.

Fig.2.1 The mutation list of SULT1A1

To implement this design, we constructed twelve engineered strains carrying different SULT1A1 mutants construction (Fig. 2.1 and 2.2).

Fig.2.2 Gene circuit of SULT1A1(M1-M12) and TAL1

Protein expression of 12 SULT-mutant constructs was analyzed via SDS-PAGE (Fig.2.3).

Fig.2.3 SDS-PAGE analysis of TAL and SULT1A1 (WT and mutants M1–M12) before(-) and after(+) IPTG induction. Arrows indicate the expected protein bands at the corresponding molecular weights, approximately 56.6 kDa for TAL and 33.9 kDa for SULT1A1.

Distinct bands appeared after induction at approximately 56.6 kDa and 33.9 kDa, corresponding to TAL and SULT1A1, respectively. The presence of clear bands in both WT and mutant lanes confirms successful expression of all constructs in E. coli BL21(DE3). Variations in band intensity among mutants indicate differences in expression efficiency or protein stability, with M12 showing relatively higher expression levels.

Test

Fig. 2.4 Comparison of Zosteric acid Production in 12 SULT-Variants.(A) Concentrations of ZA (red bars) and pHCA (blue bars) in different mutants (M1–M12) compared with strain S24 (WT);(B) ZA conversion efficiency of mutants M1–M12 compared with strain S24 (WT). Data are shown as mean ± SEM from n=3 parallel replicates. Statistical analysis was performed using ordinary one-way ANOVA with Dunnett’s multiple comparisons test (WT vs each mutant). Significance is indicated as p<0.05 (*), <0.01 (**), <0.001 (***), <0.0001 (****), and not significant (ns).

To directly assess the catalytic efficiency of engineered strains, we analyzed fermentation broth using HPLC. Each engineered strain (M1–M12) was tested for the production of ZA and the residual amount of pHCA. The results revealed clear differences among the mutants. Notably, the M12 variant (0.663mM) achieved a 3.58-fold enhancement in ZA production over the wild-type (0.184mM) (Fig.2.4 A). HPLC quantification revealed distinct differences in conversion efficiency among the twelve SULT1A1 variants (Fig.2.4 B). Most mutants (M1–M7, M9–M11) exhibited markedly reduced activity, with conversion rates dropping by 80–100% relative to the wild type (7.15%). These results suggest that most combinations of mutations introduced structural destabilization or impaired substrate binding, leading to nearly complete loss of catalytic function. M8 retained partial activity with a conversion rate of 5.57%, approximately 22% lower than WT, indicating that its mutations mildly perturbed—but did not abolish—enzymatic catalysis. In sharp contrast, M12 demonstrated a significant improvement, achieving a conversion rate of 18.04%, corresponding to a 2.5-fold increase over the wild type.

Collectively, these findings indicate that while most mutations adversely affected enzyme performance, M12 successfully enhanced catalytic efficiency and metabolic flux, validating its design as the most functionally improved variant among all tested mutants.

Learn

SULT expression level is a major limiting factor for ZA yield. SDS-PAGE analysis demonstrated that TAL was consistently expressed at stable levels across all constructs, indicating that TAL was not the limiting factor in ZA production. In contrast, SULT1A1 variants showed markedly different expression profiles. While WT SULT1A1 was expressed at relatively low levels, some mutants, particularly M9 and M12, exhibited significantly stronger expression bands around 35 kDa. Several other variants (e.g., M4, M5, M6) showed weak or unstable expression, suggesting that the introduced mutations negatively affected protein folding or stability.

Consistent with the expression data, HPLC quantification revealed distinct metabolic outcomes. Most mutants produced negligible amounts of ZA, with residual pHCA levels close to the substrate concentration, confirming that destabilizing mutations impaired enzymatic activity. In contrast, M8 displayed a modest improvement in ZA accumulation relative to WT, while M12 showed a substantial increase in ZA concentration. These results indicate that rationally guided mutations based on ΔΔG analysis can indeed enhance both the expression and catalytic efficiency of SULT1A1.

Taken together, this round of experiments demonstrated that enzyme optimization at the single-mutation level is a viable strategy to overcome the limitations of the WT enzyme. While many mutations reduced activity, M12 emerged as promising variants, providing both higher protein stability and improved catalytic performance.

Cycle 2.2: Fusion protein design

Design & Build

To improve metabolic efficiency and reduce diffusion loss of the intermediate pHCA, we designed fusion constructs between TAL and SULT1A1. Different linker strategies were tested (Fig. 2.5): flexible linkers (GGGGS)2 to allow conformational freedom, and rigid linkers (EAAAK)2 to enforce close proximity. In addition, we incorporated the SpyTag–SpyCatcher system to evaluate a modular co-localization approach, enabling TAL and SULT1A1 to be brought into close spatial proximity without direct fusion. These designs aimed to facilitate substrate channeling and enhance overall catalytic throughput.

Fig.2.5 The designs of fusion protein

Fig.2.6 The result of fusion protein plasmids construction. (A) Sequencing results;(B) Agarose gel electrophoresis results.

Fig.2.7 SDS-PAGE analysis of different fusion protein constructs before(-) and after(+) IPTG induction. Lanes show SULT–2GS–TAL (S113), SULT–2EA–TAL (S114), TAL–2GS–SULT (S115), TAL–2EA–SULT (S116), and SpyTag/SpyCatcher (S125) constructs. Arrows indicate bands corresponding to the expected molecular weights of the fusion proteins. Specifically, the TAL–SULT1A1 fusion proteins containing the 2EA or 2GS linkers exhibited an approximate molecular weight of ~91.2 kDa, while the individual components SULT1A1–GS–Catcher and TAL–GS–Tag were detected at ~47.8 kDa and ~59.1 kDa, respectively.

The sequencing results (Fig.2.6 A) and agarose gel electrophoresis (Fig.2.6 B) showed that these fusion proteins were successfully constructed. According to the sequencing results, the fusion protein constructs were successfully expressed in E. coli BL21(DE3), as confirmed by SDS-PAGE analysis (Fig.2.7). Distinct bands were observed at the expected molecular weights.

Test

Fermentation experiments were performed to evaluate whether fusion proteins could improve substrate channeling between TAL and SULT1A1. The fermentation broth was analyzed by HPLC to measure ZA production and residual pHCA

Fig.2.8 Comparison of Zosteric acid Production in different fusion stradegy. Concentrations of ZA and pHCA (A) and ZA conversion efficiency(B) in different fusion protein constructs (SULT–2GS–TAL, SULT–2EA–TAL, TAL–2GS–SULT, TAL–2EA–SULT, and SpyTag/SpyCatcher) compared with strain S24 (WT). Data were shown as mean ± SEM from n=3 parallel replicate experiments. Statistical analysis was performed using two-way ANOVA with Šidák’s multiple comparisons test (WT vs each construct). Significance is indicated as p<0.05 (*), <0.01 (**), <0.001 (***), <0.0001 (****), and not significant (ns).

HPLC quantification revealed that fusion orientation and linker type had a pronounced impact on ZA production. Impressively, SULT-2GS-TAL achieved 3.08-fold enhancement in ZA production over wild-type (Fig.2.8A). SULT-2GS-TAL and TAL-2GS-SULT (flexible linkers, GGGGS) achieved the most significant improvements, with conversion efficiencies reaching about 3.50-fold WT and reduced residual pHCA (Fig.2.8 B), confirming that flexible linkers effectively facilitated substrate channeling. By contrast, rigid linkers (EAAAK) showed orientation-dependent outcomes: SULT-2EA-TAL produced little ZA and remained close to WT, while TAL-2EA-SULT achieved high conversion efficiency but accumulated very low ZA, suggesting that product instability or steric hindrance hindered effective biosynthesis. The SpyTag/SpyCatcher system failed to improve productivity and even reduced performance relative to WT, likely due to inefficient complex formation or unfavorable spatial geometry. These results demonstrate that both linker flexibility and fusion orientation are critical determinants of fusion enzyme performance.

Learn

From this cycle, we learned that rational fusion design can substantially enhance the catalytic throughput of the TAL–SULT1A1 module, but only under optimal configurations.

Notably, the fusion orientation ——that is, whether TAL or SULT1A1 occupies the N-terminus——strongly influenced catalytic performance. Constructs following the TAL–SULT arrangement consistently yielded lower ZA concentrations than those following the SULT–TAL configuration, suggesting that substrate channeling is less efficient when TAL precedes SULT in the fusion order. The reduced activity may be due to the TAL C-terminal fusion, which potentially obstructs the assembly or function of the TAL tetramer. As a homotetrameric enzyme, TAL requires free access to its C-terminal interface to assemble into a functional tetramer. When placed at the N-terminal position, this assembly interface becomes sterically hindered by the downstream SULT1A1 domain, disrupting tetramer formation and significantly impairing catalytic efficiency.Consequently, optimal fusion performance was achieved only when SULT1A1 preceded TAL, preserving TAL’s native quaternary structure and allowing effective substrate channeling through flexible linkers.

Besides fusion orientation, the linker type also critically affects substrate channeling efficiency. Flexible linkers such as (GGGGS)₂ provided sufficient conformational freedom between the two domains, enabling adaptive domain motion and efficient substrate handover, which maximized ZA yield. In contrast, rigid linkers such as (EAAAK)₂ imposed strong geometric constraints that limited inter-domain flexibility and often misaligned the catalytic centers, leading to lower product accumulation even when the overall orientation appeared favorable.

The SpyTag/SpyCatcher co-localization strategy, while conceptually promising for modular enzyme assembly, failed to enhance performance as anticipated. SDS-PAGE analysis indicated extremely low expression of the SULT1A1-GS-Catcher construct, implying that the SpyCatcher tag might impede proper folding or translation. Furthermore, the inconsistent expression ratio between SpyCatcher and SpyTag likely prevented effective isopeptide bond formation, resulting in no detectable functional fusion complex between SULT1A1 and TAL.

This outcome highlights that simple fusion expression is insufficient to enhance catalysis; effective improvement requires both precise spatial compatibility and dynamic flexibility, which are better achieved through direct covalent linkers (GGGGS/EAAAAK) and controlled domain orientation rather than non-direct covalent modular attachment (spyTag-spyCatcher).

Design & Build

Due to the insufficient supply of SO₄²⁻, we constructed strain 25AIS-92 to overexpress cysDNCQ. This aims to increase the rate of ZA production.

Fig.3.1 Construction in cycle3. (A) Gene circuits of 25AIS-S92 and 25AIS-S24; (B) Sequence Alignment Map of the cysDNCQ; (C) SDS-PAGE analysis of recombinant protein expression. Samples include 25AIS-S92, and 25AIS-S24 strains before (-) and after (+) IPTG induction. Target proteins are indicated by arrows: TAL and KlATPSL are expected at approximately 55 kDa, while SULT1A1 and PcAPSK are anticipated at around 30 kDa. CysD, cysN, cysC, cysQ are expected to at approximately 35.2kDa, 55kDa, 22.3kDa and 27.2kDa respectively.

To validate the function of the enzymes in the PAPS recycling pathway, we constructed strain 25AIS-S92 and used 25AIS-S24, as a control for comparison (Fig.3.1 A). The construct was sequence-verified and showed a 100% match with the intended design of cysDNCQ (Fig.3.1 B).

To evaluate the protein expression of S92 and S24 (Fig.3.1 C), we performed induction under specific conditions (same condition suggested in cycle 1) and analyzed the results using SDS-Page. Both S92 and S24 exhibited bands at the expected molecular weights after induction, although with differences in band intensity. For S92 after induction, clear bands appeared at ~55 kDa, ~30 kDa, and ~23 kDa, corresponding to TAL and cysN(~55 kDa), SULT1A1(~33.9kDa), cysD(~35.2 kDa) and cysQ(~27.2kDa) respectively, though the intensities of these bands were relatively weak. CysC(~22.3kDa), however, was not successfully shown on the SDS Page. The selected gel concentration (10%) we used was not suitable for resolving proteins below a molecular weight of 23 kDa, thus explaining the absence of corresponding bands. For S24, bands were relatively strong at approximately 55kDa, showcasing the a good expression of KlATPSL and TAL. Bands also appear at approximately 30kDa, indicates the expression of SULT1A1 and PcAPSK. In contrast, the lanes of both S92 and S93 before induction showed no distinct target bands, confirming that protein expression was successfully triggered only after induction.

Strains were constructed successfully.

Test

To estimate the amount of ZA produced, fermentation was carried out in M9 medium, supplemented with 2 mM L-Tyr, 5 g/L (NH₄)₂SO₄, 10g/L casein hydrolysate and the appropriate antibiotic. IPTG was then added. We further analyzed the fermentation broth using HPLC, showcasing the production situation of ZA.

Fig.3.2 Comparison of Zosteric acid Production in different fusion stradegy. (A) HPLC analysis of comparing Zosteric Acid (ZA) production in strains 25AIS-S92 and 25AIS-S24 (as a control). The top chromatogram for strain S92 shows a prominent peak for zosteric acid at ~7.9 minutes and a peak for the remaining p-coumaric acid substrate at ~8.8 minutes. The bottom chromatogram for strain S24 shows a smaller zosteric acid peak and a larger p-coumaric acid peak compared to S92, indicating lower conversion efficiency; (B) Zosteric Acid Conversion Efficiency; (C) Final concentrations of ZA (red bars) and residual pHCA (blue bars). Data are shown as mean ± SEM from n=3 parallel replicate experiments. Statistical analysis was performed using two-way ANOVA with Šidák’s multiple comparisons test (WT vs each construct). Significance is indicated as p<0.05 (*), <0.01 (**), <0.001 (***), <0.0001 (****), and not significant (ns).

This panel (Fig.3.2 A) displays the HPLC chromatograms comparing the engineered strain 25AIS-S92 (top) with the control strain, 25AIS-S24 (bottom). Both chromatograms show a large peak at a retention time of ~8.8 minutes, corresponding to the substrate, p-coumaric acid. However, a distinct peak at ~7.9 minutes, corresponding to the product zosteric acid (ZA), is substantially larger in the S92 sample compared to the very small product peak in the control. This provides clear qualitative evidence that the engineered pathway in S92 is highly active, while the S24 has negligible ability to produce ZA. This result further demonstrates that overexpressing the endogenous cysDNCQ effectively enhances zosteric acid production.

The bar chart (Fig.3.2 B) quantifies and compares the conversion efficiency between S92 and the control S24, exhibited a baseline conversion efficiency of 7.15%. In stark contrast, the engineered strain S92 demonstrated a nearly four-fold improvement, achieving a conversion efficiency of 27.27%. This result quantitatively proves that the genetic modifications in S92 are highly effective and directly responsible for the significant increase in ZA production over the wild-type background.

The bar chart (Fig.3.2 C) details the final concentrations of the product (zosteric acid, red bars) and residual substrate (p-coumaric acid, blue bars). The control strain S24 produced a minimal amount of ZA, reaching a concentration of only 0.17 mM, leaving the vast majority of the substrate unconsumed (~2.5 mM).The engineered strain S92, however, achieved a significantly higher ZA titer of 0.77 mM, representing a 4.5-fold increase over the control strain S24. This was accompanied by a corresponding decrease in the final substrate concentration to ~2.2 mM, reflecting the effective conversion of p-coumaric acid into the target product by the engineered pathway.

To sum up, these results provide definitive evidence that our strategy of overexpressing the endogenous cysDNCQ enzymes is highly effective for boosting the biosynthesis of zosteric acid.

Learn

The importance of our initial Cycle 1 decision to knock out cysH is starkly highlighted when comparing our work to the 2022 UNILausanne iGEM team. Their project used a similar architecture (overexpress cysDNCQ) but failed to produce any Zosteric Acid (ZA). The critical difference was their retention of the native cysH gene. Our success, therefore, provides compelling evidence that the cysH deletion is a prerequisite for enabling this entire biosynthetic pathway, a finding that clarifies a key bottleneck for future projects.

Based on these learnings, and acknowledging that SULT1A1 expression remains a bottleneck, our next step is to construct a 'super strain' by fusing our optimized SULT enzyme (from Cycle 2) with the highly effective cysDNCQ module.

Fig.4.1 The gene circuit of the first-generation engineered strain 25AIS-S24 and "Super strain" 25AIS-S128

Through a combination of computational modeling and experimental validation, we confirmed that our designs for a mutated enzyme and a fusion protein significantly enhanced the stability and catalytic performance of SULT1A1. Concurrently, by replacing the heterologous sulfur donor pathway with the endogenous cysDNCQ enzymes, we improved the supply of PAPS, which in turn boosted the production of Zosteric Acid (ZA).

Based on these successful outcomes, we have designed our next-generation "super strain," 25AIS-S128 (Fig.4.1), which integrates all of the aforementioned optimizations. Due to the time constraints of the iGEM competition, this super strain is still under development. We look forward to presenting our latest progress and final results on the stage at the Grand Jamboree.

To systematically assess antifouling performance, we designed three complementary assays, each representing a distinct level of environmental realism:

1. 96-well Plate Assay: Standard and high-throughput quantification of biofilm biomass under uniform conditions.

2. Tube Assay: Simulates the liquid–air interface, where biofilms typically form in static aquatic systems.

3. Slide Biofilm Assay: Mimics real coating conditions using galvanized metal slides to simulate ship hull surfaces.

These mutilevel approaches allow us to evaluate Fer’s antifouling efficiency from laboratory-standard conditions to ocean application-oriented environments.

Two coating formulations were tested: a control consisting of pure silicone elastomer (SYLGARD 184, referred to as the 184 coating) and an experimental coating (Fer) incorporating a ZA‐enriched fermentation product. All assays were conducted using the marine bacterium Vibrio natriegens ATCC 14048, a fast growing and garam negative model organism relevant to marine biofilm research. Bacterial cultures were grown in LB medium supplemented with v2 nutrients and inoculated at an initial OD₆₀₀ of approximately 0.1 to ensure consistent and reproducible biofilm formation.

OD₅₉₀ Quantification of Biofilm Biomass

Fig.5.1 Crystal Violet detection. Biofilm formed by Vibrio natriegens ATCC 14048 on different coatings in 96-well plate, tube and slide Assays. 184(blue) represent the coating without ZA, Fer(orange) represent the coating with ZA-enriched fermentation product. The bar chart is shown above, with the actual photographs of the experimental results displayed beneath it. Data are shown as mean ± SEM from three parallel replicates. Statistical analysis was performed using mutiple t-test. Significance indicates as p < 0.05(*), p<0.01(**). Created by Graphpad Prism 8.0.2.

Based on the crystal violet assay for biofilm formation, the OD₅₉₀ value—which is proportional to biofilm biomass—was consistently lower in the Fer group than in the 184 group across all three substrates (96-well plate, tube, and slide), corresponding to inhibition rates of 48.63%, 14.82%, and 26.75%, respectively. This quantitative result corroborated the initial qualitative observation that the 184 group exhibited visibly darker purple staining than the Fer group after staining(Fig.5.1). Together, these consistent findings indicate that the Fer coating possesses significant anti-biofilm activity under various substrate conditions.

ImageJ Surface Coverage Analysis

Fig.5.2 Biofilm surface coverage anlysis of slide by ImageJ. (A) Actual phaotographs and corresponding ImageJ-based denoising analysis; (B)Comparison of biofilm coverage area with different coatings on slide. Data are shown as mean ± SEM from three parallel replicates. Statistical analysis was performed using t-test. Significance indicates as p < 0.05(*). 184(blue) represent the coating without ZA, Fer(orange) represent the coating with ZA-enriched fermentation product.

We performed ImageJ analysis on bright galvanized slides (Fig.5.2A). Using a standardized enhancement protocol (contrast uniformly increased by 0.3%; threshold range 0–100), the average biofilm coverage area was quantified (Fig.5.2B). The results showed coverage of 48.27% for the 184 group and 29.92% for the Fer group, representing a significant reduction of 37.99% (p<0.05) in biofilm attachment area for the Fer coating compared to the 184 control.

The segmentation was performed using the Otsu threshold method. Slight deviations (e.g., sample 184-2) were attributed to lighting artifacts rather than true biomass variation.

Overall, ImageJ analysis confirmed the OD₅₉₀ trend, demonstrating lower biofilm biomass on Fer coatings.

This project established the first experimentally verified E. coli platform capable of producing Zosteric Acid efficiently through systematic metabolic and protein engineering. Each cycle contributed a key advancement—from confirming pathway feasibility, to enhancing enzyme performance and sulfur-donor flux, and finally integrating all improvements into a unified chassis strain. Quantitative HPLC results demonstrated progressive yield increases, culminating in a over 3-fold improvement relative to the initial construct. Functional assays further proved that the resulting ZA-based coating significantly suppresses marine biofilm formation, achieving up to 48.6 % inhibition under laboratory conditions. Collectively, these outcomes not only demonstrate the feasibility of sustainable, biobased antifouling solutions but also provide a transferable synthetic-biology framework for future marine-biotechnology applications.

1. Peichao Zhang, Jing Gao, Haiyang Zhang, Yongzhen Wang, Zhen Liu, Sang Yup Lee, Xiangzhao Mao, Metabolic engineering of Escherichia coli for the production of an antifouling agent zosteric acid, Metabolic Engineering, Volume 76, 2023, Pages 247-259, ISSN 1096-7176, https://doi.org/10.1016/j.ymben.2023.02.007

2. Li Q, Sun B, Chen J, Zhang Y, Jiang Y, Yang S. A modified pCas/pTargetF system for CRISPR-Cas9-assisted genome editing in Escherichia coli. Acta Biochim Biophys Sin (Shanghai). 2021 Apr 15;53(5):620-627. doi: 10.1093/abbs/gmab036. PMID: 33764372.