Introduction
We decided to utilise B. subtilis as the chassis for our protective biofilm, due to the ability of B. subtilis to naturally form protective biofilms on plant roots, in addition to it being classified as a generally recognized as safe organism with no known pathogenic strains. Whilst there is some literature suggesting that wild type B. subtilis biofilms can be effective at reducing pathogen incidence and mitigating microbially induced corrosion, a field study testing Bacillus and Pediococcus protective biofilms in broiler chicken farms showed that they seem to lack long-term efficacy, possibly due to the identified limited surface coverage (Guéneau et al., 2022). Additionally, B. subtilis biofilms normally disperse and form spores during prolonged nutrient stress. Therefore, we aimed to engineer a B. subtilis strain that can form more effective protective biofilms, by designing and integrating constructs into the genome to achieve the following:
- Non-Sporulating Biofilm
- Increased Biofilm Formation Rate
- Increased Biomass of Mature Biofilm
- Inducible Biofilm Formation and Orthogonal Nutrient Utilisation
- Incorporated Pathogen Biosensor
- Incorporated Biocontainment Mechanisms
1. Non-sporulating Biofilm
Reason:
From our meeting with Dr Diana Fusco, we learned that a major limitation of B. subtilis as a protective biofilm is that B. subtilis biofilms disperse and sporulate under prolonged nutrient stress; hence, wild type B. subtilis may not provide effective long-term surface coverage.
Research:
As sporulation and biofilm formation share significant upstream regulatory pathways, it was important that we targeted a knock-out candidate sufficiently downstream in the sporulation pathway to not affect biofilm formation; however, targeting a component too far downstream may result in cells that are committed to the spore state, but are unable to proceed and hence simply die. From Gray (2016), we identified spoIIE – a phosphatase required for correct asymmetric division during sporulation – as a promising knockout candidate to meet these criteria.
Design
To knock out spoIIE, we utilised homologous recombination in B. subtilis by designing a construct containing ~1,000 bp homology regions flanking either side of the spoIIE gene, to be integrated into the genome of B. subtilis.
Build
We assembled both spoIIE homology regions into the pMiniMAD backbone, utilising Golden Gate assembly. The construct was then transformed into E. coli and miniprepped before transformation into B. subtilis by homologous recombination.
Test
The ΔspoIIE strain showed almost complete loss of spore formation, exactly as hoped. ΔspoIIE also formed a mature biofilm significantly faster than wild type B. subtilis and appeared to have slightly greater total biofilm biomass overall; hence, there was no notable negative effect on biofilm formation, and instead a significant benefit to biofilm formation was identified.
Learn
From our meeting with Dr Briandet, he informed us that commercial B. subtilis probiotics are commonly distributed as spores, due to their increased stability; hence, whilst knocking out sporulation may increase long-term stability and increase biofilm formation, this may limit the commercial viability of B.Max.
Improve:
Based on Dr Briandet’s advice, we designed a construct for genome integration of spoIIE under an inducible promoter; in theory, this could be integrated into a ΔspoIIE strain to allow for inducible sporulation in the presence of the inducer during culturing to allow for spore harvesting for commercial distribution, but during application, in the absence of the inducer, the biofilm would be non-sporulating.
Additionally, as knocking out spoIIE from the sporulation genetic pathway resulted in significantly increased biofilm formation rate, we designed constructs to modify genes within the biofilm genetic pathway to hopefully achieve a similar outcome.
2. Increased Biofilm Formation Rate
Reason:
Based on field studies of protective Bacillus and Pediococcus biofilms in broiler chicken farms, these protective biofilms seemed to lack long-term efficacy, possibly due to the identified limited surface coverage (Guéneau et al. 2022). Hence, increased biofilm formation rate may help increase surface coverage, providing a more effective protective biofilm than non-engineered strains. Additionally, increased biofilm formation rate should allow for the protective biofilm to form quicker, providing less time for pathogens or corrosive microbes to take up the empty space following cleaning.
Research:
We created a gene regulatory model for biofilm formation in B. subtilis, based on the data in published literature. From this model, we identified that increasing the concentration of sinI, abbA, or remA may result in increased rate of biofilm formation.
Design
We designed constructs for expression of the regulatory genes sinI, abbA, and remA under the control of a xylose inducible promoter (BBa_K733002), with each construct construct flanked by homology regions for subsequent integration into the genome of B. subtilis.The inducible promoter should allow for testing the effect of different levels of gene expression by varying the inducer concentration.
Build
Each construct was assembled in an 8-part Golden Gate assembly. Correct size bands were identified for each construct by colony PCR following transformation into E. coli, however only sinI was successfully subsequently integrated into the genome of B. subtilis 3610 in the time we had available.
Test
Biofilm biomass safranin staining assays confirmed that increased production of SinI results in significantly increased rate of biofilm formation. There was no significant difference between the induced and uninduced condition (with and without 0.5% xylose), suggesting that the BBa_K733002 xylose inducible promoter is significantly leaky.
Learn
Due to difficult troubleshooting an 8-part Golden Gate assembly, in the future we would design this construct with fewer parts. With more time, we would also like to test the effect of increased expression of abbA and remA.
3. Increased Biomass of Mature Biofilm
Research:
From Bassler et al.(2015) and McKenzie et al. (2022), we identified that the density and thickness of the biofilm extracellular matrix appears to play a role in preventing pathogen invasion. Hence, increasing total biomass (and increasing density and thickness) may result in B.Max having increased resistance to invasion.
From our gene regulatory network model of biofilm formation, it seemed that increasing expression of the signalling genes may also result in increased total expression of the biofilm associated genes, hence resulting in increased total biomass. Additionally, we considered increasing the expression of the major polysaccharide and protein components of the biofilm matrix, the eps and tapA-sipW-tasA operons, in addition to ypqP gene from B. subtilis NDmed (involved in strong biofilm formation, disrupted in common lab strains of B. subtilis, such as 3610, likely involved in synthesis of extracellular polysaccharides) (Sanchez-Vizuete et al. 2015).
Design
Designed constructs consist of eps operon, tapA-sipW-tasA operon, and ypqP gene under the control of a xylose inducible promoter (BBa_K733002), with the whole construct flanked by homology regions for subsequent integration into the genome of B. subtilis. The inducible promoter should allow for testing the effect of different levels of gene expression by varying the inducer concentration.
Build
Each construct was assembled in an 8-part Golden Gate assembly. Correct size bands were identified for the tapA-sipW-tasA operon and ypqP gene by colony PCR following transformation into E. coli, however only the tapA-sipW-tasA operon was successfully subsequently integrated into the genome of B. subtilis 3610 in the time we had available.
Test
Biofilm biomass safranin staining assays confirmed that increased expression of the tapA-sipW-tasA operon results in significantly increased biofilm formation, in addition to increased expression of the regulatory gene sinI and the downstream gene mstX (discussed in the following section).
Learn
Due to difficulty troubleshooting an 8-part Golden Gate assembly, in the future we would design this construct with fewer parts. With more time, we would also like to test the effect of increased expression of the eps operon and ypqP.
4. Inducible Biofilm Formation and Orthogonal Nutrient Utilisation
Reason:
As we want to increase total biomass of the protective biofilm, we considered that nutrient usage/availability may be a limiting factor in biofilm formation. Published literature suggested that B. subtilis biofilms primarily form under low nutrient conditions and hence biofilm regulatory pathways may not be sufficiently active under high nutrient conditions. Inducible biofilm formation may allow for biofilm formation under higher nutrient conditions, and hence possibly increased total biofilm biomass.
Additionally, given that surface coverage appears to be a major limitation of protective B. subtilis biofilms, it is possible that inducing biofilm formation prior to application may result in significantly increased adhesion, due to production of the sticky extracellular matrix components, resulting in improved surface coverage.
Research:
Choe et al. (2013) demonstrated that overexpression of the mstX gene resulted in activation of KinC (which normally can act as a sensor for low nutrient conditions), resulting in the induction of the biofilm formation pathways.
Design
The designed construct consists of the mstX gene under the control of a xylose inducible promoter (BBa_K733002), with the whole construct flanked by homology regions for subsequent integration into the genome of B. subtilis.
Build
The mstX construct was assembled in an 8-part Golden Gate assembly. Correct size bands were identified by colony PCR following transformation into E. coli, and correct size bands were identified by colony PCR in B. subtilis, hence integration into the B. subtilis genome was successful.
Test
Biofilm biomass safranin staining assays identified that increased expression of mstX appears to result in significantly increased rate of biofilm formation.
Inducible expression of mstX also resulted in significantly increased biofilm formation under high nutrient conditions (left = biofilm formation of mstX overexpressing B. subtilis 3610 (in pink) and wild type B. subtilis 3610 on standard MSgg media, right = biofilm formation of mstX overexpressing B. subtilis 3610 (in pink) and wild type B. subtilis 3610 on 4x nutrients containing MSgg media)
Learn
Whilst overexpression of mstX seems to increase rate of biofilm formation, we considered that even if nutrient utilisation was normally limiting, providing external nutrients to an environment such as farm surfaces would also provide these nutrients to pathogens or corrosive microbes. Hence, ideally we would be able to apply an external nutrient source that our protective biofilm could digest/utilise, but pathogens and corrosive microbes could not.
Improve:
Shepherd et al. (2018) identified a ~40-60 kb region within the genome of Bacteroides plebeius/Phocaeicola plebeius that provides the ability to digest the polysachharide, porphyran, found in seaweed. We designed constructs to integrate the minimum required ~40 kb of this construct into the genome of B. subtilis by homologous recombination of 6 ~7 kb fragments, utilising 2 different resistance genes at the 3’ end that would be deleted from the genome by homologous recombination of the subsequent fragment, due to the 5’ homology region being homologous to the region upstream of this resistance gene.
5. Pathogen biosensor
Reason:
Although B.Max is designed to minimise pathogen invasion, it inevitably would not be completely effective. An incorporated pathogen biosensor for major foodborne pathogens may allow for targeted cleaning/use of biocides if necessary.
Research:
When scavenging across the literature on the molecular signals that enteric pathogens produce, we came across autoinducer-3 (AI-3). AI-3 is used by enteric pathogens to sense and signal that they are in the intestinal environment, and is sensed by the two-component system (TCS) QseBC. Hence we chose this system as the base of the biosensor due to AI-3’s heavy involvement in the virulence of enteric pathogens. In the TCS, upon binding of AI-3 by QseC, QseC phosphorylates QseB. QseB-P will then bind onto a promoter and activate the transcription of the genes downstream, for example a GFP so that the system acts as a biosensor for AI-3.
Design
Our final construct for the AI-3 biosensor was designed as below (BBa_4769605):
The QseB and QseC genes are controlled under pVeg, a constitutive promoter, so that B. subtilis would be able to receive AI-3. A superfolder GFP gene is downstream of the QseB-activated promoter. By not placing a terminator after the sfGFP, QseBC would also be expressed after AI-3 reception, hence increasing B. subtilis’ sensitivity to AI-3 and increase the fluorescence produced. The gene is inserted into B. subtilis using pMiniMAD with the homology regions of AmyE.
We have decided to test our idea on E. coli DH5α as a proof of concept. To do so, we have built a construct consisting of the QseB-activated promoter (BBa_K4769600) which we extracted from the E. coli genome, a T7 g10 RBS (BBa_K4769607), a superfolder GFP which we amplified from pJUMP28-1A(sfGFP) (BBa_J428353), and a conventional terminator (BBa_J428091). We have selected AE_lacZ pDEST (BBa_J435320), a high copy number expression plasmid, as the backbone. This construct assumes QseB and QseC are expressed from the E. coli genome.
Build
We have assembled our constructs using Golden Gate Assembly and transformed them into E. coli DH5α and B. subtilis respectively.
Test
In the DH5α strain containing the proof of concept plasmid, the transformed bacteria doesn’t fluoresce, in contrast to what is expected as QseBC/AI-3 is a quorum sensing system and the AI-3 concentration should increase as cell density increases.
Learn
We went back to literature and found evidence showing that AI-3 expression is linked to AI-2 quorum sensing, and AI-2 is not produced in the DH5α strain, which may contribute to the fact that our transformant doesn’t fluoresce. A method to overcome this is to directly add AI-3, or its analog adrenaline, to the media and measure the fluorescence. However, adrenaline cannot be obtained in our lab due to the risk involved, hence we decided to switch our proof of concept to proving that QseC can be properly incorporated into the plasma membrane of B. subtilis, as QseC is only present in gram-negative bacteria and is usually inserted into the inner membrane.
Moreover, we identified another potential reason for the construct to fail, where upon inspecting the genome of DH5α, an araC family transcriptional repressor is present in front of the QseBC genes. As our construct assumes the genomic expression of the two proteins, the repressor could be problematic.
We also ideated a protocol which doesn’t require adrenaline as a replacement for AI-3. To do so, we would culture an AI-3 producing bacteria, then centrifuge it to obtain the AI-3 containing supernatant, which can be added to our transformant that is cleaned with 10% glycerol.
Improve:
We have hence designed a fusion protein between QseC and sfGFP (QseC-sfGFP), using the linker sequence GSAGSAAGSGEF (Waldo et al., 1999) to connect both proteins. An Alphafold prediction of the structure shows a stable sfGFP can be formed.
To overcome the araC family repressor problem, we have also designed a construct for E. coli that includes the QseBC coding genes. The construct is assembled in a similar fashion as to the B. subtilis biosensor, but with Pveg replaced by J23100. (BBa_J23100), a standard constitutive promoter, and the B. subtilis RBS replaced with T7 g10-L RBS for E. coli.
Test
Unfortunately we didn’t have the time to transform the QseC-sfGFP construct into B. subtilis. However, if given the time, a successful localisation would have fluorescence around the periphery of the cell.
6. Biocontainment
Reason:
Whilst B. subtilis is generally recognised as safe and is already used as a probiotic for humans, plants, and animals, it is important that our engineered strain does not end up where it was not designed to be and possibly cause unintended consequences. Additionally, although our engineered strain is designed to minimise the probability of invasion, over time there is the possibility of these beneficial biofilms being invaded by pathogenic or corrosive microorganisms, which may then be difficult to remove during cleaning. Hence, we recognised the need for biocontainment to prevent unintended spread of B.Max, in addition to the need for inducible biofilm disruption (self-destruction) to allow for effective cleaning of any invading microbes.
Design - Preventing Unintended Spread
To prevent the unintended spread of B.Max, we considered 3 of the major biocontainment mechanisms – a toxin-based kill switch that results in toxin expression and cell death under certain condtions, engineered auxotrophy that only allows cells to survive in the presence of supplemented nutrients, or conditional expression of essential genes under certain conditions. As B.Max may be broadly applied over large surfaces with varying nutrient compositions, engineered auxotrophy did not appear to be reasonable in this case. Additionally, when considering the long-term evolutionary stability of the biocontainment mechanism (i.e. how easy it is to break it), toxin-based kill switches appear significantly less evolutionarily stable (i.e. easier to break) than conditional essential genes. Hence, we decided to utilise conditional essential genes for this biocontainment mechanism.
As B.Max is intended to survive only in the biofilm-associated state, we decided to design a biocontainment mechanism that allows biofilm-associated cells to survive, but results in the death of planktonic cells. To do so, we considered utilising the biofilm master regulator in B. subtilis, SinR – SinR binds to an operator region upstream of the promoter of genes involved in biofilm formation, repressing expression of these genes during the planktonic state. During the planktonic-to-biofilm transition, SinR is sequestered by SinI, preventing binding of SinR to the operator regions and allowing for expression of biofilm-associated genes. Hence, adding a SinR-binding operator region upstream of the promoter of an essential gene may allow for the desired biocontainment mechanism of survival in biofilm-assocaited state and death in planktonic state.
However, a single SinR-binding operator region upstream of the promoter of a single essential gene also appears evolutionarily unstable (i.e. it could be easily broken by disruption of the SinR-binding operator region). To increase the evolutionary stability of this biocontainment mechanism, we considered adding multiple binding sites for SinR, and binding sites for other transcription factors involved in biofilm formation, upstream of the promoters of multiple essential genes. From this, the probability of all binding sites being disrupted for all of the conditional essential genes would appear to be very low.
As B. subtilis would be planktonic during culturing and application, in this biocontainment system the cells would die prior to being able to form a biofilm. Hence, we considered inserting an additional copy of the essential gene at a different site in the genome, expressed under an inducible promoter. This would effectively form an AND gate for survival, where cells have to either be in the presence of an inducer (which could be provided during culturing) or in the biofilm-associated state.
To identify the optimal essential genes to conditionally express, we analysed the BsubCyc database for every essential gene in B. subtilis to identify those which have annotated promoter regions supported by experimental evidence and contain only a single gene under the control of that promoter (to avoid unintentionally making other genes conditional). From this, we identified gyrA, gltX, and rpsD as promising candidates.
For designing the operator region, we decided to utilise the naturally-occuring operator regions upstream of the eps and tapA-sipW-tasA operons (encoding the major polysaccharide and protein components of the biofilm matrix, respectively). These operator regions are well-documented based on experimental evidence, hence it was possible to design constructs to insert these operator regions upstream of the promoter of the essential genes, with the same distance between the operator elements and transcription start site that occurs in the eps and tapA-sipW-tasA operons.
(Images from BsubCyc)
Build
Each of the three essential genes were assembled as an 8-part assembly by Golden Gate assembly, under the control of a xylose inducible promoter (BBa_K733002), between homology regions for subsequent integration into the genome of B. subtilis. However, due to the xylose inducible promoter being significantly leaky, it was not possible to achieve no expression of the essential genes and hence there was no point proceeding with introducing the operator regions upstream of essential genes.
Design - Inducible Biofilm Disruption
To allow for effective removal of B.Max and any invading microbes during cleaning, we considered the optimal mechanisms for inducible biofilm disruption. Whilst there is data supporting biofilm disruption by application of the DNase, DNase I, and the protease, proteinase K, these are eukaryotic proteins which are unlikely to function optimally in B. subtilis. Instead, we considered other proteins, such as the bacterial DNase, NucB, the bacterial proteases, NprB and Bpr, and the bacterial racemase, YlmE. Expression of some combination of these proteins following application of an inducer may result in significant amounts of biofilm disruption to allow for effective cleaning.
Build
The biofilm-degrading enzymes were each assembled as an 8-part assembly by Golden Gate assembly, under the control of a xylose inducible promoter (BBa_K733002), between homology regions for subsequent integration into the genome of B. subtilis. However, we were unfortunately not able to correctly assemble the constructs in the time we had available.
Learn
Due to difficulty troubleshooting an 8-part Golden Gate assembly, in the future we would design this construct with fewer parts. With more time, we would also like to test the effect of inducible expression of the biofilm-degrading enzymes on degradation of mature biofilms.