Executive Summary
Aquarius explores differences in chassis performance in laboratory conditions as compared to real-world aquatic environments. Through wet lab and dry lab experimentation, we aimed to clarify the effects of environmental complexities—such as microbial competition, water flow, and nutrient variability—on the survival, behavior, and functioning of engineered microorganisms.
We focused on three aquatic case studies, each designed to test and compare chassis effectiveness in controlled laboratory conditions versus simulated natural environments:
I. Marine Corrosion Prevention with Bacillus subtilis
We engineered biofilm-enhancing variants of Bacillus subtilis and compared their growth, motility, biofilm formation, and corrosion protection in lab media versus seawater microcosms. Experiments included strain construction, growth curve analysis, motility assays, biofilm morphology characterization, and corrosion testing on steel surfaces.
II. Freshwater Harmful Algal Bloom (HAB) Remediation with Cyanophage and Acinetobacter baylyi
We attempted to isolate (and ultimately engineer) novel M. aerguinosa cyanophages, viruses that infect the cyanobacterium M. aeruginosa, through multiple large scale assay-based experiments. We also tested the persistence and algicidal activity of Acinetobacter baylyi, a chassis with natural algicidal capabilities, in static laboratory cocultures and flowing lakewater microcosms. We assessed A. baylyi’s survival and gene expression via colony counts, 16S amplicon sequencing, and transcriptomic analysis.
III. Household Pipe Biofilm Removal Using Phage Therapy
We evaluated phage efficacy in lysing Mycobacterium smegmatis biofilms under lab conditions and in PVC pipe microcosms simulating household plumbing. Experiments included phage titer estimation, survivability testing, biofilm imaging, and gene expression analysis of biofilm-forming bacteria.
Bioinformatics & RNA-Seq Meta-Analysis
To complement experimental data, we conducted a meta-analysis of transcriptomic and metatranscriptomic datasets comparing lab and environmental conditions. Using differential expression analysis and functional enrichment tools, we identified key genes and pathways that inform design principles for deploying synthetic biology in aquatic systems.
Aquatic Case Study I:
Corrosion Prevention Using Engineered Bacillus subtilis in Marine Environments
Overarching Goal: To compare ability of engineered B. subtilis to prevent corrosion in lab conditions versus simulated real-world marine seawater conditions.
Overall Rationale:
Domesticated bacterial strains commonly used as synthetic biology chassis are optimized for laboratory performance but often lose key traits during the domestication process. To ensure their effectiveness for real-world implementations, domesticated strains must be compatible with the pathways they are engineered to regulate. For instance, while natural Bacillus subtilis strains can form robust biofilms, competent lab strains like B. subtilis 168 have lost their ability to secrete extracellular polymeric substances (EPS), which are critical for biofilm development (McLoon et al., 2011).
Microbial-Induced Corrosion Inhibition (MICI) presents a promising, sustainable, and self-healing approach to reducing corrosion. However, current MICI strategies lack systematic experimental validation, and there is limited research comparing biofilm behavior in laboratory versus natural seawater environments—raising questions about their real-world feasibility.
We hypothesized that natural B. subtilis biofilms may not be sufficiently robust to prevent corrosion in realistic conditions. We aimed to engineer B. subtilis strains to have enhanced biofilm-forming capabilities by upregulating EPS-related genes. Using the competent but biofilm-deficient lab strain B. subtilis 168, we tested multiple regulatory pathways to identify those most effective at promoting robust biofilm formation and corrosion resistance in real-world environments.
Experiment 1:
Construction of Engineered B. subtilis Strains
Rationale: Gao et al. (2022) suggested that Bacillus subtilis strains that produce thicker biofilms will more effectively prevent corrosion than those with weaker biofilm-forming capabilities. We selected 4 genes from distinct regulatory pathways in B. subtilis (strain 168) for targeted editing to create thicker and denser biofilms for corrosion prevention on steel surfaces. We used an integration vector to ensure stable genomic incorporation, minimize risk of construct loss during cell passage, and prevent horizontal transfer of antibiotic resistance genes typically required for plasmid-based engineering. We also employed inducible promoters to maintain controlled gene expression.
Procedure: We selected 4 genetic targets known to influence biofilm properties: SinI to promote biofilm thickness by alleviating SinR-mediated repression, BslA to enhance surface hydrophobicity through the formation of a protective protein layer, and the TapA-SipW-TasA operon and TasA to strengthen biofilm architecture via amyloid fiber production. We constructed a vector that overexpresses each gene to assess their respective contributions to biofilm morphology and corrosion inhibition.
Summary Results: We successfully transformed B. subtilis 168 with the integration vector using homemade chemically competent cells. We confirmed the transformation’s success with Starch-Iodine staining. Attempts to transform B. subtilis 3610 were unsuccessful due to the strain’s lack of natural competence.
Experiment 2:
Establishing growth curves to compare growth dynamics of B. subtilis 168, 3610, and the 4 engineered 168 variants under laboratory conditions
Rationale: Studies suggest that upregulating certain biofilm-formation genes adversely affects bacterial growth rates (Bai et al., 1993). To confirm that our engineered constructs do not impose substantial burden on the chassis, we conducted multiple growth curve measurements with the engineered strains and under varying concentrations of xylose (the inducer) in LB media.
Procedure: Strains were cultured in LB medium with varying xylose induction levels in 96-well plates. A 24-hour growth curve and a 36-hour growth curve were conducted using a microplate reader for downstream comparison of growth rates between the lab strain, natural strain, and engineered variants under different inducer concentrations.
Summary Results: Both the 24- and 36- hour growth curves showed that the xylose inducer concentration had no substantial effect on bacterial growth in normal LB media (which does not promote biofilm growth). The 4 modified strains and the unmodified lab strain of Bacillus subtilis 168 did not differ substantially in growth rate, while the growth rate of the natural strain B. subtilis 3610 was substantially greater than all other strains.
For detailed results, click here!Experiment 3:
Evaluating the motility of B. subtilis 168, 3610, and the 4 engineered 168 variants
Rationale: Studies found that the laboratory strain Bacillus subtilis 168 lacks extracellular polymeric substances (EPS), a key biofilm component, and therefore does not exhibit EPS-dependent sliding motility (Dergham et al., 2021). We aimed to determine whether specific biofilm-enhancing genetic modifications restore or alter motility behavior in Bacillus subtilis 168.
Procedure: Motility was assessed using semi-solid agar plates (0.3%, 0.7%, and 1.5% agar) all supplemented with a 0.2% concentration of xylose inducer. Movement on 0.3% agar reflects flagellum-dependent swarming, while diffusion on 0.7% agar indicates EPS-dependent sliding. Spreading diameters were measured after 12 hours. The bacterial strain is considered to be mobile when the diameter of the spread region is at least 20% greater than the diameter of the colony of the same strain developed in the 1.5% agar medium that does not allow any movement (Bartolini & Grau, 2019)
Summary Results: Among all tested strains, only Bacillus subtilis 3610 exhibited both swarming and sliding motility, highlighting its robust native biofilm-associated movement. Among the engineered variants, B. subtilis 168 and the SinI-modified strain demonstrated strong swarming capability, while the BslA+ and TapA-SipW-TasA+ variants showed moderate swarming. In contrast, the TasA+ variant exhibited neither swarming nor sliding motility, which may be attributed to potential toxicity arising from TasA overexpression.
For detailed results, click here!Experiment 4:
Determining effect of our genetic modifications on biofilm robustness and structure through characterization of biofilm and colony morphology of laboratory and engineered strains of Bacillus subtilis
Rationale: Brück et al. (2019) demonstrated that Bacillus subtilis 168 loses its ability to produce extracellular polymeric substances (EPS) during domestication, resulting in diminished biofilm formation. Therefore, we aimed to test the effect of our genetic modifications on biofilm morphology and robustness.
Procedure: 3 μL of overnight LB culture inoculation from each strain was spotted onto various agar media, including: LB, LB supplemented with 0.03% xylose, MSgg (biofilm-inducing minimal medium), MSgg with 0.03% xylose spread on the surface, MSgg with 0.03% xylose mixed into the agar prior to pouring, and MSgg with 0.2% xylose. Plates were incubated at 37 °C for 48 hours. Colony and biofilm morphology were documented using phone photography and Hyrox microscopy for higher-resolution imaging.
Summary Results: Biofilm formation varied significantly across strains and conditions. Bacillus subtilis 3610 (natural strain) was the only strain to produce a distinct biofilm, while the engineered and laboratory strains showed minimal or no biofilm characteristics. Media choice played a critical role: LB supported basic growth but was unsuitable for biofilm development, whereas MSgg enabled biofilm formation. However, minimal differences in xylose concentration had little effect on biofilm enhancement, while varying methods of inducer distribution had an effect on colony morphology, suggesting that more precise control of induction conditions is necessary for consistent biofilm formation.
For detailed results, click here!Experiment 5:
Evaluating corrosion protection abilities of each strain on steel surfaces in a simulated marine environment
Rationale: The effectiveness of corrosion prevention mechanisms likely differs between sterile laboratory conditions and natural environments due to microbial community dynamics. In seawater microcosms, native bacteria may infiltrate Bacillus subtilis biofilms, potentially disrupting their uniformity and compromising protective function. This experiment evaluates whether engineered enhancements to biofilm structure improve corrosion resistance, and how each strain—lab strain 168, natural strain 3610, and engineered variants Sini+ and BslA+ —performs in a simulated marine setting that reflects real-world conditions. SinI+ and BslA+ were selected from the 4 engineered strains based on their demonstrated motility and potential for biofilm formation.
Procedure: We used recycled 1000 mL pipette tip boxes as microcosm water tanks, each containing 200 mL of artificial seawater (35 g Instant Ocean salt / L). Seven microcosms were prepared: one control and six inoculated with Bacillus subtilis strains (3610, 168, SinI+, BslA+, TasA+, and TapA–SipW–TasA+), each induced with 0.2% xylose and inoculated at 0.5% (v/v) of the seawater(1ml). Three additional high-biomass microcosms with 50 mL of overnight bacterial culture (with 0.2% xylose) to 150 mL of seawater(15% (v/v)) were prepared for RNA extraction using B. subtilis 168 SinI+, BslA+, and 3610. A36 steel plates were thoroughly cleaned, painted with bacterial strains, and incubated in seawater microcosms. After 3 to 7 days of incubation, corrosion levels were visually assessed using phone pictures and via scanning electron microscopy (SEM).
Summary Results: The experiment revealed notable differences in corrosion protection among natural, lab, and engineered Bacillus subtilis strains under realistic conditions. The uninoculated seawater control showed heavy oxidation, while samples treated with B. subtilis 168 exhibited moderate corrosion. Novel surface features—including irregular deposits and parallel, unidirectional metallic crystals—were observed across various samples, suggesting that microbial activity and strain-specific interactions can influence surface mineralization. This novel structure, which has not been observed in the literature, may be attributed to species interactions in biofilm formation.
For detailed results, click here!Experiment 6:
Analysis of gene expression differences between lab and microcosm environments in strains with strong corrosion protection
Rationale: We aimed to use RNA-seq to uncover transcriptional changes that explain the observed differences in biofilm formation and corrosion inhibition capabilities of engineered and non-engineered B. subtilis strains. This would provide insight into the genetic and regulatory mechanisms underlying biofilm production in simulated natural environments, allowing future synthetic biologists to rationally design Bacillus strains with optimized biofilm and surface-protective properties. We also use RNA-seq and 16S analyses to enable direct comparison between strain-specific transcriptional profiles and community-level metagenomic data, to showcase the ability of B. subtilis to survive in a community dynamic environment.
Procedure: Three strains: SinI overexpressing B. subtilis 168, BslA overexpressing B. subtilis 168, and the natural strain B. subtilis 3610, were selected for transcriptomic analysis via RNA sequencing based on the morphology test assays, motility test assays and their suggested corrosion inhibition ability as shown in the SEM image. RNA was extracted from cultures grown under sterile laboratory conditions and from samples retrieved from seawater microcosms to compare gene expression profiles across environments. Additionally, 16S DNA sequencing was performed on microcosm samples inoculated with varying concentrations (0.5% and 25%) of each strain. This analysis aimed to identify the microbial community composition within the microcosms and assess whether population structure influences the corrosion protection efficacy observed in Experiment 4.
Summary Results: Across all samples, a substantial presence of non-Bacillus subtilis bacteria was detected, indicating the establishment of mixed microbial communities. The wild-type B. subtilis 3610 showed limited persistence within these communities, whereas both engineered strains exhibited moderate survival. In the microcosms inoculated with 25% bacterial culture, Ochrobactrum emerged as the dominant contaminant and the most abundant genus overall. The 0.5% inoculum microcosms consistently contained substantial populations of Acinetobacter, suggesting that the amount of Bacillus added influences microbial community composition and competitive interactions. RNA-seq showed that all three B. subtilis strains suppressed sporulation genes but activated metabolic ones in marine microcosms. The BslA+ strain also upregulated flagellar assembly genes. The data confirm distinct gene-expression behaviors between microcosm and laboratory conditions.
For detailed results, click here!Aquatic Case Study II:
Freshwater Algal Bloom Remediation
Overarching Goal: To compare behavior and ability of cyanophages and/or Acinetobacter baylyi to remediate HABs in lab conditions versus simulated real-world conditions.
Experiment 7:
Microcystis Cyanophage Isolation for Harmful Algal Bloom Remediation
Experimental Goal: To isolate (and potentially engineer) novel Microcystis cyanophages for harmful algal bloom (HAB) remediation as a case study to compare phage behavior in laboratory conditions versus simulated real-world environments.
Rationale: Cyanophage, viruses that infect cyanobacteria, may represent a promising tool for HAB remediation (Grasso et al., 2022; Bhatt et al., 2023; Aranda et al., 2023), however 1) few M. aeruginosa cyanophages have been isolated (Guo et al., 2024) and 2) cyanophage deployment in real lakewater environments may be challenging given dynamic factors such as water flow and unique features of HAB structure and ecology— e.g., host resistance evolution and phage lysogeny during blooms (Grasso et al., 2022; Huang et al., 2025; Bhatt et al., 2023; Aranda et al., 2023).
To address the lack of readily available phages infecting M. aeruginosa, we aimed to isolate novel Microcystis cyanophages for future use in HAB remediation and for downstream optimization via engineering. To clarify the effect of water flow on cyanophage persistence and infection of M. aeruginosa, we aimed to test novel cyanophages under simulated lakewater conditions. Our ultimate goal was to identify and implement design principles that would enhance phage viability for HAB mitigation in real-world water environments.
Procedure: The general procedure for isolating Microcystis cyanophages includes 1) collection of water samples, 2) filtration of water samples and enrichment in Microcystis liquid culture, 3) enrichment filtration (0.22 μm) for phage particles, and 4) phage infection assays via two methods:
A) Liquid Assays: M. aeruginosa was cultured with sample filtrate in 35 mm dishes and monitored for bacterial clearing indicative of infection. Cleared assays were filtered (0.22 μm) and filtrate was used to conduct re-infection assays.
B) Plaque Assays: Concentrated M. aeruginosa was mixed with sample filtrate and plated on BG-11 plates. Microcystis lawns were monitored for phage plaque formation. Plaques were resuspended in liquid Microcystis culture and monitored for clearing. A plate containing prominent plaques was flooded and lysate harvested and used for more reinfection assays.
Summary Results: We observed bacterial clearing compared to control assays in 39 of 205 liquid assays and putative cyanophage plaque formation on 2 of 65 agar assays. However, we were unable to consistently propagate any phage, likely due to strong host defense systems. Following the guidance of IHP experts as well as our mathematical modeling, we then transitioned to investigating the use of Acinetobacter baylyi, a commonly-used chassis with algicidal capabilities, for HAB remediation.
For detailed results, click here!Experiment 8:
Comparison of Acinetobacter baylyi Behavior in Laboratory vs Simulated Real-World Microcosms
Experimental Goal: To test habitat remediation with Acinetobacter baylyi ADP1-ISx as a case study comparing bacterial behavior (persistence and gene expression) in laboratory conditions versus simulated real-world environments in order to inform the development of design principles for synthetic biology applications in aquatic systems.
Overall Rationale: Bacteria with natural algicidal capabilities offer an alternative HAB mitigation option that may be easier to engineer and deploy than phage. Lab-based co-culture experiments have suggested that Acinetobacter species can almost completely eliminate an M. aeruginosa culture within as few as 5 days (Li et al., 2016, Yi et al., 2015). While algicidal bacteria are highly effective in a lab setting (Kong et al., 2022), our meta-analysis indicated that bacterial species, including engineered strains, are unlikely to function optimally in real-world lakewater environments, where they are subjected to stress due to mechanical flow and the presence of diverse, competing microbial communities.
To more fully understand the effects of in vivo, water-associated complexities on the functioning of engineered, lab strain algicidal bacteria and assess the feasibility of real-world deployment, we measured the persistence and gene expression of A. baylyi ADP1-ISx in flowing lakewater HAB microcosms as compared to static control co-cultures.
We inoculated three lakewater microcosms and three control flasks with equivalent ratios of approximately log phase A. baylyi and M. aeruginosa. Microcosms contained water from a local lake and were equipped with water pumps to simulate turbulent lakewater flow. Control flasks were static and contained BG-11 media. For details on the microcosm setup, see our Results section.
Experiment 8a:
Estimating A. baylyi survival in microcosms and control cocultures via colony counts
Rationale: We hypothesized that realistic in vivo lakewater conditions (including water flow and microbial communities) would reduce A. baylyi survival and could limit real-world deployment potential. To estimate the extent to which realistic conditions alter A. baylyi survival over time, we grew spread plates and quantified A. baylyi colony formation (from microcosms and control cultures) at each experimental timepoint.
Procedure: At each timepoint, we spread 100 μL of fluid from each microcosm and control culture on LB agar plates. After approximately 48 hours of incubation at 37°C, we stored plates at 4°C for downstream A. baylyi population estimation via colony counts. Colony counts were generated by Google Gemini using AI image analysis. We validated Gemini’s counts visually and adjusted values to match our observations.
Summary Results: Cocultures consistently yielded uniform A. baylyi lawns, while microcosms yielded decreasing A. baylyi colony counts over the course of the experiment.
For detailed results, click here!Experiment 8b:
Quantifying species abundance in microcosms and control cocultures via 16S amplicon sequencing
Rationale: Real-world aquatic environments contain dense microbial communities that we hypothesized would adversely affect chassis survival and functioning during in vivo deployment. To quantify the extent to which native lakewater species outcompete or interfere with A. baylyi survival under realistic conditions, we used 16S data to approximate relative species abundances in the microcosms and control cocultures over the course of the experiment.
Procedure: We extracted total genomic DNA via phenol chloroform isoamyl alcohol (PCI) extraction from frozen sample pellets collected on day 1 and day 7 for each microcosm and control culture. We amplified 16S fragments via PCR and purified them using the NEB PCR cleanup kit. Purified samples were sequenced by Plasmidsaurus.
Summary Results: 16S percent abundance data indicated that Acinetobacter baylyi was outcompeted by native lakewater organisms within a week after initial microcosm inoculation. The average microcosm A. baylyi population declined from an initial 73.58% abundance (on day 1) to 7.83% abundance (day 7). Control abundances on day 7 were approximately comparable to those measured on day 1 (on average at 77.41% on day 7 as compared to 81.33% on day 1).
For detailed results, click here!Experiment 8c:
Quantifying differential gene expression of A. baylyi in microcosms and control cocultures via RNA sequencing
Rationale: We hypothesized that aquatic environments may affect specific aspects of chassis behavior and functioning in addition to altering survival. To quantify the effect of simulated real-world deployment conditions on A. baylyi functioning and HAB remediation potential, we extracted RNA samples for downstream differential gene expression analysis comparing simulated in vivo microcosm and in vitro coculture conditions.
Procedure: We extracted total RNA via the NEB Monarch Spin RNA Isolation Kit from frozen sample pellets collected on day 1, day 5, and day 7 for each microcosm and control culture. We conducted preliminary analysis of RNAseq data using DEseq2 on the Galaxy platform.
Summary Results: Preliminary RNA sequencing analysis indicated that a number of A. baylyi genes displayed significant (p-adjusted ≤ 0.05) differential gene expression across microcosm and control coculture conditions approximately 24 hours post-inoculation. The existence of differentially expressed genes across the two conditions provides evidence that A. baylyi behaves differently in lakewater conditions than in a traditional lab setting and may suggest that real-world conditions could limit the chassis’s HAB remediation potential.
We are currently in the process of analyzing and interpreting RNAseq data to quantify A. baylyi differential gene expression between microcosms and control cocultures at all three timepoints.
For detailed results, click here!Aquatic Case Study III:
Removal of Biofilms in Household Pipes
Overarching Goal: To compare ability of phages to lyse biofilms in lab conditions versus real world conditions of household pipes.
Experiment 9:
Determining efficiency of specific mycobacteriophages to lyse mycobacteria under laboratory conditions
Rationale: Ultimately, we wish to test the ability of phage to treat biofilms in pipes, but to do so we need to ensure that the phage lyse bacteria effectively under controlled lab conditions. Therefore, for our first experiment we tested the ability of phage to successfully lyse mycobacteria under controlled lab conditions. This would allow us to generate comparative results between the efficiency of phage in lysing mycobacteria in the form of a lawn on a petri dish and in the form of a biofilm on a pipe.
Procedure: We conducted several experiments to estimate the titer of phage and to simultaneously increase titer in preparation for further experiments outside of standard lab conditions. Phage CrimD, Phage Neighly, and Phage Raid were filtered and diluted with a phage buffer solution. Dilutions (10^-2 to 10^-10) were combined with Mycobacterium smegmatis and plated with top agar, and incubated overnight. Dilutions were determined based on plaque assays. Titers were estimated by counting plaque forming units (PFU). Plates with bacterial lawns with webbed lysis phage patterns were flooded with a phage buffer, and filtered to generate high titer phage lysate.
Summary Results: Phage titers ranged from approximately 10^10 to 10^13 PFU per milliliter of lysate. Phage CrimD and Phage Neighly formed webbed lysis at 10^10 dilutions and Phage Raid at 10^8 dilutions, successfully killing most of the mycobacterial lawn.
For detailed results, click here!Experiment 10:
Determining efficiency of phage to lyse of biofilm-forming mycobacteria on PVC pipes simulating household plumbing conditions
Experimental Goal: To evaluate whether phages can effectively lyse mycobacteria under conditions that mimic household water systems—beyond controlled laboratory settings—and to assess their potential as a viable strategy for biofilm removal in pipes.
Rationale: In order to compare with laboratory conditions, we evaluated the effectiveness of phage to lyse biofilms in more dynamic, complex environments simulating household environments compared to controlled lab conditions and the ability of phage to penetrate a biofilm. We constructed microcosms mimicking household water systems where mycobacterial biofilms are often found and administered phage as treatment of these biofilms and compared these results with phage lysing efficiency under lab conditions. We did so by the following types of measurements:
Figure: Experimental design for the microcosm that was used for experiments 10a, 10b, and 10c.
Experiment 10a:
Determining titer of viable bacteria in the microcosm effluent before and after phage infection
Rationale: Bacterial population estimates are difficult to evaluate in the biofilm due to the complex structure of the extracellular matrix and the adhesion of the bacteria to the biofilm. Therefore, we sampled from the microcosm effluent to determine the bacterial titers in the effluent.
Procedure: Biofilms with Mycobacterium smegmatis were grown out on PVC pipes in a nonsterile environment, and placed into 3D adaptors to simulate a showerhead. The pipes were flushed once a day with tap water, and bacterial titers from the effluent were taken after each flushing. The M. smegmatis was transformed with a red fluorescent protein (RFP) before biofilm growth with a plasmid that displayed hygromycin resistance, so bacterial titers were evaluated on hygromycin resistant plates. After several days, phage was added to the tap water and bacterial titers were taken every day after phage treatment.
Summary Results: Overall, populations of hygromycin resistant bacteria did not decrease after phage treatment, implying that phage were unable to infect and lyse mycobacteria in the biofilm along the pipe efficiently enough to make a significant impact.
For detailed results, click here!Experiment 10b:
Determining phage survivability in microcosm conditions
Rationale: Real world environments present challenges to the survivability of bacteriophage, since they are largely susceptible to degradation, with the presence of proteases in drinking water. We aimed to determine if there would be degradation of phage in tap water by testing its survivability in microcosm effluent after 24 hours.
Procedure: To determine if phage can not only lyse mycobacteria but survive in real world conditions without complete degradation, phage titer was compared before and after microcosm flushing. Phage of known titer was diluted with standard tap water and flushed through the microcosm. The effluent was collected and let stand for 24 hours, and the effluent was filtered and spotted on a mycobacterial lawn formed with top agar.
Summary Results: After 24 hours, there was a decrease in PFU, signifying that there is degradation of phage under microcosm conditions.
For detailed results, click here!Experiment 10c:
Testing phage presence in microcosm effluent at the final time point
Rationale: When phage was flushed through the biofilm, we predicted through literature and mathematical modeling that the phage would adhere to the biofilm and continuously replicate while lysing the bacteria. Therefore, phage presence should have persisted throughout the time after inoculation since we still saw a persistence of bacterial presence, although decreased.
Procedure: Microcosm effluent was collected and filtered. It was combined with M. smegmatis, plated with top agar, and incubated overnight. Plaques were quantified.
Summary Results: There was no phage presence in the effluent at the final time point, meaning that phage did not maintain presence in the biofilm a week after treatment.
For detailed results, click here!Experiment 10d:
Determining the structure of the biofilm in the pipe following phage treatment
Rationale: Mycobacterial biofilms are largely made up of sugars and polysaccharides, which comprise the extracellular matrix (Esteban et al., 2018). In order to understand how the structure of the biofilm was affected by phage treatment, we placed it under a microscope in order to assess the differences in biofilm structure and visual composition with and without phage treatment.
Procedure: In order to evaluate the effect of phage on the biofilm structure, we constructed separate microcosms, where two were treated with phage and two were not. 36 hours after flushing either with or without phage, a slice of the PVC pipe was viewed under the Hirox RV 2000 microscope to generate a 3-dimensional image of the biofilm. Because the mycobacteria was transformed with a RFP, mycobacteria was identified by the expression of red fluorescence under the microscope camera. The average height of the biofilm as well as surface area covered by biofilm on the PVC pipe were compared between pipes with and without phage treatment.
Summary Results: There was not significant variation in maximum height of the biofilm, but the average height of the biofilm was lesser in pipes exposed to phage treatment. There was also a formation of plaque-like clearings on the PVC pipe, with a significant decrease in surface area covered by biofilm.
For detailed results, click here!Experiment 10e:
Determining gene expression differences of mycobacteria in laboratory versus microcosm conditions
Rationale: Mycobacterial gene activation in biofilm formation can largely inhibit phage infection, with the extracellular matrix creating difficulty for phage accessibility to the bacteria. Some phage have adapted to this by inhibiting genes that regulate biofilm maturation such as GroEL1 (Ojha et al., 2005). For more informed phage selection for biofilm treatment, RNA from biofilm forming mycobacteria was extracted and sequenced. Biofilms were grown in a nonsterile environment in the presence of PVC material and tap water in order to create conditions more similar to those inside of a household pipe.
Procedure: Mycobacterium smegmatis culture was diluted with tap water and placed in a nonsterile tray containing PVC pipe fragments and incubated for 7 days. RNA was extracted using the NEB Monarch Spin RNA Isolation Kit.
Summary Results: We are in the process of analyzing RNA sequencing data to determine the genes expressed in mycobacteria while it is forming biofilms in a nonsterile environment.
For detailed results, click here!Experiment 10f:
Determining bacterial community structure of microcosm biofilm
Rationale: Findings in literature suggest that mycobacteria most efficiently forms biofilms in the presence of other species and adheres most efficiently to biofilms produced by other bacteria or yeast (Gomez-Smith et al., 2015). Therefore, the biofilms were grown on PVC pipes in a nonsterile setting to allow for more effective adherence to the PVC material. To determine the bacterial species present in the biofilm, we performed 16s rRNA metagenomic sequencing.
Procedure: Biofilm was scraped off the inside of a pipe and suspended in TE buffer. The solution was then bead beat and boiled to release DNA. 16s PCR was performed and the amplicons were sequenced. Raw sequence data was analyzed using Kraken2 and Bracken tools on Galaxy.
Summary Results: The 16s sequencing results suggest that the biofilm largely consisted of acinetobacter, but contained various other species. There is no mycobacteria sequence identified, although there is actinobacterium sequence identified, which may be the mycobacteria.
For detailed results, click here!RNA-Seq Meta-Analysis
Overarching Goal: To identify differentially expressed genes and pathways in chassis organisms by comparing laboratory versus environmental transcriptomic and metatranscriptomic data.
Experiment 11:
To identify and assemble a comprehensive set of synthetic biology chassis genes and pathways that are differentially expressed (both up- and down-regulated) in laboratory versus natural environments, with the ultimate goal of using these data to develop fundamental design principles for synthetic biology in aquatic systems
Rationale: As stated in our Project Description, despite the ever-increasing severity of problems facing aquatic systems and the ineffectiveness and/or harmful side effects of current solutions, synthetic biology solutions—despite their potential—have not been widely deployed. While many synthetic biology solutions have been developed and function under laboratory conditions, there is an overall lack of knowledge on how these chassis and their engineered circuits would perform in realistic environments.
The objective of our project is to establish a rigorous knowledge base for conducting synthetic biology safely, predictably, and effectively in aquatic systems. To this end, we have employed an experimental approach (also presented in this section) in which we rigorously test these differences at all levels. However, these are only three examples. Therefore, it is equally important—if not more so—to conduct a thorough meta-analysis of current literature and to extract as many relevant existing datasets as possible to establish a comprehensive assessment of genes, pathways, and patterns that are differentially expressed (DE) between laboratory and environmental conditions, with the ultimate goal of establishing fundamental design principles for synthetic biology.
Procedure: The first step of this bioinformatic experiment was to establish an exhaustive list of DE genes. For the bioinformatic analysis, we performed a structured survey of the primary literature and public transcriptome repositories (e.g., NCBI SRA) to identify studies profiling the same aquatic chassis under laboratory and more natural or simulated environmental conditions. When DE results were provided by the original authors, we incorporated them directly; otherwise, we reprocessed raw reads through a standardized workflow culminating in DESeq2 for DE analysis. Functional interpretation relied on gene-set and pathway enrichment against Gene Ontology and KEGG catalogues, implemented via DAVID tools. In all cases, we attempted to pair cultured-strain transcriptomes with in situ metatranscriptomes from the same or closely related taxa wherever possible.
Our experiment can be divided into five stages:
- Analysis of transcriptomic data comparing laboratory versus environmental samples
- Analysis of metatranscriptomic data comparing laboratory versus environmental samples
- Functional analysis of DE genes
- AI Analysis of DE Genes:
- As described above, while we devoted significant effort to the bioinformatic analyses, our team also began studying prompt engineering to more effectively and accurately employ AI for our analyses. We were able to significantly enhance our catalog of DE genes with specific prompts for both generally DE genes and those specific to a given genus or species. For our analysis, we employed Microsoft Copilot, Google Gemini, and OpenAI ChatGPT (both version 4 and ChatGPT+).
- Curation and Vetting of AI Results:
- While AI tools are improving at an exponential rate, they still hallucinate and make mistakes. It is therefore absolutely essential to vet the results produced by AI agents. To do so, we not only checked the published literature but also analyzed the outcomes using our own bioinformatics pipelines.
Summary Results: As described in more detail in the Results section, integrating all five approaches allowed us to generate not only specific lists of genes and pathways differentially expressed when comparing laboratory conditions versus natural environmental conditions for the species in which we generated experimental data, but also a more generalizable list of genes and pathways that served as a starting point for developing a set of design principles.