Results

Introduction

The PhytoBlock project aimed to develop a sustainable biocontrol solution against Phytophthora species that cause devastating black pod disease in cocoa plants. Through systematic engineering and testing, we have achieved significant milestones in creating a functional microbial biocontrol agent based on Bacillus subtilis. Our goals is to develop a sustainable biocontrol solution against Phytophthora species that cause the devastating black pod disease in cocoa plants. Through multiple cycles of engineering and testing, we have achieved significant milestones in creating a microbial biocontrol agent based on Bacillus subtilis.
Our results demonstrate successful implementation of our dual-input biosensing system, effective production and secretion of antimicrobial peptides, and validation of their activity against target pathogens. This section presents comprehensive findings from our laboratory investigations, including molecular characterization, functional validation, and preliminary efficacy studies.

Finding suitable AMPs to inhibit Phytophthora growth

Upon reviewing the antimicrobial peptides (AMP) database, as described in the Model page, we chose the following AMPs as leads:

• Two vicilin-like peptides from the macadamia tree (Macadamia integrifolia) (MiAMP2-2 and MiAMP2c-2). These AMPs have been reported to suppress hyphal growth in oomycetes like P. cryptogea [1]. Therefore, these two candidates were selected for wet lab validation.
• Two European spindle (Euonymus europaeus) proteins. Ee-CBP is found on leaves, while a closely related isoform is produced in the bark. These lectin-like peptides have been shown to bind cell wall carbohydrates in P. cryptogea, thus disrupting growth of the organism. Their activity has been reported independently for different isolates, providing support that extends beyond a single strain and justifying inclusion for in vitro testing [2][3].
• Two short antifungal peptides (Antifungal peptide 1 and Antifungal peptide 2) originating from the rubber tree (Eucommia ulmoides), with documented effects on P. infestans. These peptides are small, cationic and have traits associated with membrane permeabilization and rapid killing of pathogens. They have confirmed activity against Phytophthora [4], and since P. infestans and P. palmivora share similar membrane lipid compositions and cell wall polysaccharides, these AMPs have been shortlisted for further testing.
• Lucimycin (LSer-AFP), derived from the green bottle fly (Lucilia sericata), complements our predominantly plant-based list with a defense molecule derived from an invertebrate. Although it also exhibits some antibacterial activity, its antifungal potency and resistance to proteolytic degradation are particularly relevant, with reported activity against Phytophthora parasitica [5]. Its mode of action likely involves membrane or cell-envelope disruption, which supports our rationale for expecting activity against Phytophthora palmivora.
• EP-20, derived from Xenorhabdus budapestensis, was included because antimicrobial peptides from entomopathogenic bacteria often display broad antifungal activity. EP-20 has been reported to inhibit Phytophthora capsici [6]. Its short length, cationic charge, and proven efficacy against oomycetes make it a suitable candidate for expression and screening.
• Cecropin from the tsetse fly (Glossina morsitans). Although cecropins are best known for their antibacterial activity, this peptide has also been shown to inhibit Phytophthora infestans [7].

Taken together, the nine selected AMPs represent a consistent trend in our database search: each peptide is documented in public repositories with primary literature support, demonstrates at least one reported activity against a Phytophthora species, and possesses a plausible mechanism of action that could extend to Phytophthora palmivora. Preference was given to AMPs with either inhibitory effects against Phytophthora, mechanisms involving interaction with cell wall components or membrane disruption.
In the future, we aim to gain a deeper understanding of the physical properties of these peptides, enhance efficacy and improve stability through protein engineering.

Phylogenetic analyses of the Phytophthora genus allow us to refine predictions of AMP activity

Our Phytophthora genome database covers 174 different species. Most of them have a hemi-biotrophic lifestyle and they are found all across the world. Common hosts are cocoa trees, tomato plants, soybean, and citrus. The combination of conserved lifestyle and biology makes envelope-targeting AMPs especially attractive. As a result, mechanisms that bind β-glucans or destabilize membranes are likely to maintain efficacy across the genus.
We have taken particular interest in three Phytophthora species: P. infestans, P. cryptogea, and P. palmivora. While our main focus is on P. palmivora affecting cocoa trees, P. infestans and P. cryptogea are naturally present in Belgium. Consequently, these local species can serve as suitable models for evaluating the effectiveness of our antimicrobial peptides (AMPs) in case it is impossible to obtain Phytophthora palmivora.
The P. infestans T30-4 genome (228.5 Mb, 51% GC) is large and repeat-rich, making it particularly susceptible to AMPs that target conserved membrane and cell wall components [8]. This supports our selection of cationic, amphipathic, and lectin-like peptides effective against P. infestans, P. cryptogea, and related species.
P. cryptogea CBS 418.71 (63.8 Mb, 52.5% GC) is a hemibiotrophic species with a broad host range [9]. Several of our candidate AMPs, including vicilin-derived peptides such as MiAMP2 and carbohydrate-binding proteins like Ee-CBP, have demonstrated inhibitory effects on this species, supporting its suitability as a potential model for future testing of our system.
P. palmivora sbr112.9 (107.8 Mb, 49% GC) is a hemibiotrophic species that primarily infects Theobroma cacao and is distributed across major cocoa-growing regions [10]. Its β-glucan– and cellulose-rich cell wall aligns well with the mechanisms of our selected AMPs, including β-glucan–binding lectin-like peptides and short, cationic amphipathic helices. Together with P. infestans and P. cryptogea, these reference genomes represent key traits relevant to our research, increasing the likelihood that our chosen AMPs will demonstrate effective inhibition against phylogenetically or functionally similar strains.

Using MUMmer4 [11] genome alignment system, we conducted a comparative homology study with the previously listed reference genomes. The whole-genome alignment results indicated that P. infestans, P. cryptogea and P. palmivora are closely related at the genomic level.
The MUMmerplot dotplot illustrates the genomic alignment between two species, highlighting regions of sequence similarity and structural variation. The prominent blue diagonal line from the bottom left to the top right indicates a strong level of synteny, suggesting that large portions of the two genomes are conserved in both sequence and gene order. This reflects a close evolutionary relationship or limited genomic rearrangement between the species.
The presence of scattered purple and blue dots off the main diagonal signifies local sequence similarities that are either rearranged, inverted or fragmented due to assembly differences. Therefore, the plots support a high degree of homology between the species, with some evidence of genomic inversions and structural variations.
The conserved synteny and the overlap in predicted gene content all point to these three species sitting near one another within Phytophthora, differing primarily at the margins of their genomes rather than in their backbone.

Nine sequence-optimized basic parts expand the SubtiToolKit for production of AMPs in B. subtilis

Thanks to the kind support of Prof. Joaquin Caro-Astorga, we obtained access to the SubtiToolKit (STK). The use of STK enables efficient assembly of the desired constructs by using Golden Gate Assembly. This technique utilizes specific Type IIS enzymes to cut and assemble in a modular way. Acquired STK strains contain plasmids encoding various promoters, RBSs, reporter proteins, secretion tags and terminator sequences optimized for use in B. subtilis, which allowed us to streamline our cloning approach. Each STK part is equipped with specific adapters, whose sequence depends on the type of the part (Figure 2). Variation is possible for the CDS part, which can be split into Cx, Cy and Cz regions, allowing assembly of proteins with different N- and C-terminal tags.

We aimed to expand the STK by incorporating novel parts – encoding selected AMPs, a reporter PhoA protein and two fusion tags, aimed to facilitate protein purification – into STK syntax and testing their performance in B. subtilis. All basic parts were codon-optimized for efficient expression in B. subtilis using GenSmart™ Codon Optimization tool. To ensure compatibility with STK and compliance with iGEM registry guidelines, we avoided BsaI, SapI, BbsI and BsmBI restriction sites in the design.
The parts were synthesized as gBlocks by Twist or IDT. To ensure efficient Golden Gate Assembly (GGA) of the final constructs, the parts were pre-cloned into level 0 plasmids (comparable to basic parts) and stored in E. coli DH10B. Reaction conditions were optimized (see the Engineering Cycle 1 for more information). The final GGA cloning protocol is described on the Experiments page. After transformation, non-fluorescent colonies were verified by colony PCR, and plasmids from positive clones were isolated and sent for sequencing.
By the end of the project, we successfully cloned the following basic parts into level 0 plasmids:
Cy:

• MiAMP2-2 – BBa_25DO0ICK
• MiAMP2c-2 – BBa_258YJPQ3
• Ee-CBP-Leaves – BBa_25OQ2MUK
• Antifungal peptide 1 - BBa_25LAH38O
• Antifungal peptide 2 - BBa_25CSAZHD
• Lucimycin – BBa_25N58AWT
• EP-20 – BBa_25DO0ICK
• Cecropin – BBa_25CH9T79

Cx:

• HIS-Thioredoxin-TEV – BBa_25FN8UNC
• HIS-SUMO – BBa_25FN8UNC

Cx/Cz:

• PhoA from E. coli – reporter protein

Thus, we expanded the STK with parts codon-optimized for production in B. subtilis, enabling production of seven different AMPs, SUMO and thioredoxin fusion tags, and a reporter PhoA gene.

Codon-optimized alkaline phosphatase (PhoA) secretion reporter in B. subtilis

To visualize the effects of xylose induction and provide a platform for testing different secretion tags, we designed a reporter system employing alkaline phosphatase (PhoA) from E. coli as a reporter gene. The amino acid sequence was downloaded from the GenPept database (accession number NP_414917.2) and optimized as described above. To lower the complexity of the GGA reaction, the designed overhangs allowed assembly with the secretion tag and terminator parts, omitting the Cz linker part. We assembled and tested six constructs differing in the type of Cx tag and RBS. All designs included the xylose promoter part (STK161_pSTK-0A-XylR-Pxyl) and a strong terminator (STK080_pSTK-0D-bkdB Cold double terminator). The constructs contain either Apr or SacB secretion tag or a Cx linker. Additionally, we tested two RBS sequences.
PhoA has been used previously as a reporter of secretion in B. subtilis [12], but, to our knowledge, a detailed secretion assay protocol for this chassis has not been described by any previous iGEM team. Our goal was to design a protocol that can be easily reproduced and adapted by future teams.
In our pilot assay, we employed the following controls:

• Purified bovine alkaline phosphatase from Promega – positive control
• Pellet and supernatant samples from B. subtilis 168 WT – negative control

Moreover, we baseline-corrected our measurements with the growth media (LB).
Out of all engineered strains, only B. subtilis transformed with pPB026 showed clear accumulation of product in the supernatant sample. In that construct, PhoA was fused with the Apr tag. This secretion tag is documented to permit secretion of PhoA in B. subtilis [11]. The efficiency of secretion is notably high, with the signal from the supernatant over 10-fold higher than that of the cell lysate sample (Figure 3).

To assess the reproducibility of the assay, we plotted the increase of OD410 over time for the strain transformed with pPB026, measured in three assay runs (Figure 4). The results were highly similar across the first two repeats. We assume that slight modifications to the protocol in the third repeat (described in the Notebook) caused the higher variability observed.

Future prospects:

• Further optimization of the assay – the goal would be to observe the whole saturation curve, allowing to incorporate specific models and measurements of the kinetic properties of secretion tag-PhoA fusions. Moreover, we need a way to relate the enzymatic activity to the absolute amount of the enzyme in the medium.
• The current protocol involves long incubation times. Perhaps the modification of the protocol (induction with higher amount of xylose, addition of higher concentration of the substrate) would increase the sensitivity of the assay, allowing to shorten the incubation time post-induction.
• The first paper describing the use of PhoA as a secretion reporter in B. subtilis mentions the prospect of creating tri-partite fusions to analyze the secretion tag preference of the protein of choice by measuring the activity of PhoA fused to that protein [11]. The main end goal of this assay is to evaluate this approach to measure the secretion of AMPs. First, we need to adapt a protocol allowing to confirm the presence of the AMP in the supernatant sample upon secretion via a particular secretion tag (SDS-PAGE electrophoresis, HPLC), then compare the results of such assay to the phosphatase activity of the secretion of the fusion of such AMP with the same secretion tag and PhoA.

Our plate-based dual culture assay setup allows standardization of microbial growth measurements

We adapted and described one of the classic dual-culture plate assay formats, wherein the mycelium plug with Phytophthora is placed in the centre of the plate, surrounded by three bacterial colonies. Secretion of anti-oomycete compounds by tested bacteria inhibits the growth of Phytophthora mycelia near the bacteria, leading to the formation of “halos”.

To standardize the conditions of the assay, we measured the growth of both tested Phytophthora strains and B. subtilis 168 on five different solid media: carrot agar (CA), pea sucrose agar (PSA), Lysogeny broth agar (LB), V8-juice agar (V8) and potato dextrose agar (PDA). This was tested at two different temperatures, 27°C and 30°C. In each test run, we included 4 replicates of each microorganism growing on a particular type of media.
Cost analysis suggests that our biocontrol agent could be economically viable for smallholder farmers, particularly when considering reduced environmental impact and potential for resistance management compared to chemical alternatives.
In general, we observed better growth of P. capsici than P. palmivora on all tested media except PSA (Figure 5). The overall effect of temperature on incubation was not significant (p = 0.2102 in 3-way ANOVA), although it appeared to impact P. palmivora growth on PSA and PDA (Figure 5).

Analysis of the growth data indicated that pea sucrose agar (PSA) at 27 °C supported the most consistent and robust growth for both organisms (Figure 6).

Future prospects:

• Gathering of the measurements of the B. subtilis media preference needs optimization (Fig. 7). Despite plating the same amount of bacterial cultures, diluted to a particular optical density, plate-to-plate variation between the technical replicates in the same assay is high. Perhaps the assessment of the growth curves of bacteria growing in the liquid media would be a useful indicator of the media preference.
• Ideally, we would repeat the growth assay at least twice per temperature variant to increase our confidence in the collected results.

After establishing the initial conditions of the assay, we conducted a test dual culture experiment, where Phytophthora strains were cultured with B. subtilis strains expressing GFP upon induction with xylose. This served as a validation step for the final assay.
During this stage, we tested two parameters:

1. Whether different xylose concentration improves protein expression without negatively impacting Phytophthora growth.
2. The optimal volume of bacterial suspension required per assay plate.

We observed that GFP fluorescence was much stronger at 2% xylose compared to 1% xylose, confirming that induction efficiency improved, but at the cost of negatively impacting the growth of P. palmivora (Figure 8). Due to this effect we decided to use 1% xylose in our final competition assay experiment.

Future prospects:

• Further evaluation of the xylose content in the assay media would allow us to pinpoint the highest concentration of xylose that has an effect on the induction of the Pxyl promoter without inhibiting P. palmivora growth.

Constructs allowing for the secretion of two AMPs can be easily transformed to B. subtilis

The main objective of the project was engineering and evaluation of the activity of AMP-secreting B. subtilis strains. In our assemblies, we used parts from the SubtiToolKit alongside the basic parts we incorporated into the STK syntax. We followed the experimental procedures described in the STK paper with slight modifications (see Experiments page for detailed protocols).

The initial design was straightforward: expression of the AMP–secretion tag fusion was regulated by the xylose promoter. To mitigate potential toxicity effects in E. coli, we included the “E. coli Expression Blocking Device” RBS part from STK in a variant of the assembly. Additionally, we tested two secretion tags (Apr and SacB) and a Cx linker part (non-secreting) to account for different secretion-tag preferences of the tested AMPs.
Assembly of the final constructs allowing for the production and secretion of AMPs proved to be more difficult than anticipated. The troubleshooting process is described extensively in the Engineering Cycle 4 . In the end, we successfully assembled plasmids allowing for the xylose-dependent production of the Vicilin-like antimicrobial peptide 2-2 and cecropin fused to the SacB secretion tag or Cx linker, as confirmed by sequencing data. Those constructs were then transformed to B. subtilis 168 strain. Engineered bacteria were then evaluated for their ability to inhibit the growth of Phytophthora.
We did not see formation of mycelia-devoid halos around the B. subtilis colonies, which would signify inhibitory activity of the engineered strains. Interestingly, we observed two different mycelial growth behaviors depending on incubation with specific B. subtilis strains. P. palmivora appeared to exhibit faster growth (compared to the culture cultivated alone) when cultured with B. subtilis strains expressing tested AMPs with a SacB linker. In contrast, P. capsici seemed to grow equally fast when incubated with strains carrying SacB–AMP fusions or grown alone, but its growth when cultured with B. subtilis expressing GFP was slower.
The time constraints did not allow us to repeat the experiment. We believe that the observed results might be related to a different amount of plated mycelia, which could allow for faster Phytophthora growth independent from the effect of the treatment.
We observed that the morphology of B. subtilis colonies differed from plate to plate (Figure 9). Strains transformed with a secretion tag and MiAMP2-2 (Figure 9C) or cecropin (Figure 9D) appeared “granular,” as if bacterial growth was affected, perhaps by the produced AMPs. This would signify a potential toxic effect of the selected peptides on engineered bacterial cultures and the need to screen for new AMPs or explore fusion tags protecting the production host.

If the production of the AMPs affects the growth of B. subtilis, we hypothesize that perhaps the observed effect on the growth speed of Phytophthora may be due to the release of other compounds by B. subtilis under stress to the medium.
Future prospects:

• To confirm whether the observed differences are biologically significant, we aim to repeat the assay and report better on the amount of mycelia used in the assay.
• Moreover, we need to optimize the method of data collection. ReShape Biotech Imaging Device cannot properly track the growth of Phytophthora over time in presence of three colonies belonging to different species. We measured the changes in the area by manually analyzing the pictures of the plates with Fiji (ImageJ) software.
• We hypothesize that the observed effect – if true – is not directly related to the ability of tested AMPs to affect the growth of Phytophthora. We want to transform our constructs to another B. subtilis strain lacking proteases/genome minimized to ensure the stability of the AMPs. Additional protocols involving visualization of AMPs on SDS-Page gels or HPLC, without relying on the functional assays to confirm the peptide production, are also in plans.
• We aimed to keep our cloning strategy as versatile as possible – thus, the assembly method enables the substitution of the secretion tag with a SUMO fusion tag. This will allow us to purify the peptides and measure their effect on the Phytophthora strain directly to assess their specificity.

Versatility of our system: from biocontrol strain engineering to peptide purification

Anticipating problems with assembly of constructs containing AMP–secretion tag fusions, we included 6xHis-SUMO and 6xHis-thioredoxin Cx parts in our basic parts repertoire. This allowed us to substitute the secretion tag for a fusion tag in the GGA reaction when needed.

Engineering bacteria to produce peptide–SUMO fusions has proven a successful strategy to obtain significant yields of numerous AMPs [13–17]. This approach is usually used to purify AMPs from cell lysates or culture media (as opposed to using the engineered strain directly as a biocontrol agent), as attachment of SUMO may interfere with inhibitory effects. Conversely, fusion tags can lower peptide toxicity to the production host. As we struggled to engineer B. subtilis strains secreting AMPs, assemblies enabling purification of the peptides would allow us to assess their activity on the tested Phytophthora spp.
We assembled the construct bearing the sequence of 6xHis-SUMO–cecropin and transformed it to B. subtilis 168. Interestingly, we did not observe growth of transformants in LB liquid medium. Production of SUMO fused with a cecropin AD variant in B. subtilis has been reported in the literature [17, 18]. Thus, we believe the lack of growth is most likely the result of technical issues or contamination of the batch of competent B. subtilis cells.
Future prospects:

• Persisting issues regarding cloning and evaluation strategies reinforces the need to quality-check all used sources (chemocompetent E. coli and B. subtilis cells, reagents, STK strains, media) to exclude contaminations
• We would like to test the effect of the growth conditions (rich/minimal medium, growth temperature, shaking conditions) on the performance of the strain
• Lastly, we could test other fusion tags like 6xHis-thioredoxin-TEV fusion for their ability to increase the yield of the produced AMP. Several carrier proteins have been used to increase the yield of producing peptides in bacteria, each characterized by its own advantages and disadvantages [19].

Conclusion

We successfully cloned and assembled multiple AMP constructs in Bacillus subtilis and developed benchmark assays to identify optimal dual-culture assay conditions. Although functional assays revealed noticeable morphological changes in B. subtilis expressing AMPs, no inhibition zones or significant effects on Phytophthora growth were observed. These findings suggest that our AMP library may need to be expanded and refined to include more effective candidates. Overall, the results provide a solid foundation for future work, emphasizing the need for further optimization of construct expression and secretion to achieve consistent antifungal activity.