Effects of intraspecies and interspecies competition on genetic device construction and performance
Samantha Thompson, A. Robert Williams, Veronica Dill, Deven Marshall, Emily Sawyer, Mason Alexander, Lilah Rahn-Lee, Joseph De-Chung Shih
TL;DR
Genetic devices in microbes often fail in complex environments due to competition, but using regulated promoters can help maintain their function.
Contribution
The study reveals how intraspecies and interspecies competition affect genetic device performance and proposes regulated promoters as a solution.
Findings
High-strength constitutive promoters in E. coli lead to unstable expression and inactivating mutations in monoculture.
Co-culture with P. aeruginosa increases inactivating mutations in E. coli with medium- and low-strength promoters.
Regulated promoters enable stable expression in complex microbial environments like wastewater.
Abstract
One exciting class of future genetic devices could be those deployed in microbes that join complex microbial environments in the wild. We sought to determine whether genetic parts designed for monoculture are predictable when used in co-culture by testing constitutive Anderson promoters driving the expression of chromoproteins from a plasmid. In Escherichia coli monoculture, a high copy number origin of replication causes stochastic expression regardless of promoter strength, and high constitutive Anderson promoter strength leads to selection for inactivating mutations, resulting in inconsistent chromoprotein expression. Medium- and low-strength constitutive Anderson promoters function more predictably in E. coli monoculture but experience an increase in inactivating mutations when grown in co-culture over many generations with Pseudomonas aeruginosa. Expression from regulated promoters…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig 1
Fig 2
Fig 3
Fig 4
Fig 5
Fig 6| Primer name | Sequence (5′ → 3′) |
|---|---|
| J23100_AflII_F | tatt |
| J23100_XbaI_R | tata |
| J23108_AflII_F | tttt |
| J23108_XbaI_R | ggtt |
| J23100_SDM_F | |
| J23100_SDM_R | |
| J23108_SDM_F | |
| J23108_SDM_R | |
| dps_promoterF_XbaI | ttcc |
| dps_promoterR_NdeI | ttcc |
| grxA_promoterF_XbaI | ttcc |
| grxA_promoterR_NdeI | ttcc |
| CAT-F |
|
| CAT-R | GAGGCATTTCAGTCAGTTGC |
| CAT-Rinv | GCAACTGACTGAAATGCCTC |
| pBRforEco | AATAGGCGTATCACGAGGC |
- —National Science Foundationhttp://dx.doi.org/10.13039/501100008982
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGene Regulatory Network Analysis · Evolution and Genetic Dynamics · Evolutionary Game Theory and Cooperation
INTRODUCTION
Starting with the expression of human insulin in E. coli in the 1970s (1), there has been an evolution in the design of genetic devices. The first generation of biotechnology utilized biological parts, such as promoters, terminators, and coding sequences (CDSs), that were designed to maximize biotherapeutic or industrial yield in controlled monoculture environments such as bioreactors (2). To expand beyond this model for biotechnology into more complex environments, such as natural ecosystems and the human body, the next generation of genetic devices that are designed to function in microbial communities will need to be tested in communities consisting of multiple microbial species.
One example of a field that could use such answers for next-generation genetic devices is bioremediation. Microorganisms have been discovered in many different environments with natural bioremediation properties, such as the ones that were enriched during the 2010 Deepwater Horizon oil spill that were shown to degrade oil (3), microorganisms that degrade plastics found in the ocean (4), and others that sequester heavy metals in the soil (5). These naturally occurring microorganisms found in the above studies have some bioremediation capability but are slow in their natural rates of bioremediation and so possess only limited natural bioremediation efficacy. To enable bioremediation at a scale and rate that would be environmentally useful, there are ongoing efforts to either genetically modify these organisms to increase their bioremediation rates or express combinations of enzymes important in bioremediation from naturally occurring but genetically intractable organisms in more genetically tractable organisms (6, 7). However the organisms are made to enable scalable bioremediation, it is very likely that these organisms will have to exist in natural environments as a member of complex microbial communities to perform their designed functions.
We need to understand how the microbial community is affecting the genetic device and vice versa. Several questions will need to be addressed to achieve this goal. First, can a genetically modified organism enter an existing community and establish itself as a member? Genetically modified organism establishment has been demonstrated using nutrient selection, such as in the case of Bacteroides engineered to consume porphyrin fed as a prebiotic (8), but using such a specific nutrient limits potential utility. The rules that govern how genetically modified organisms establish themselves in a microbial community without such specific nutritional limits are being explored with in vitro and in vivo studies (6, 9, 10). Second, does the introduction of the genetic device affect microbial community structure or function? Studies in soil microbe communities have shown that genetic device introduction can affect microbial community structure through changes in species abundance (11, 12), while functional changes in microbial communities caused by genetic device introduction have been modeled (13). Finally, does the community impact genetic device function? We address this question in our study by putting a simple genetic device into microbial communities.
Therefore, we combined constitutive Anderson promoters, publicly available promoters with known performance capabilities (14), with commercially available chromoproteins from ATUM (15) in Escherichia coli and grew that E. coli with Pseudomonas aeruginosa in a simple co-culture community (10). Constitutive Anderson promoters and genes encoding chromoproteins are examples of genetic parts that could be particularly important for next-generation genetic devices to operate in the environment, as constitutive promoters could be used to keep a genetic device on regardless of environmental conditions, and chromoproteins could be useful as environmental reporters because they are detectable by the naked eye without the need for a diagnostic device. These promoters and chromoprotein reporters have been individually optimized for expression in E. coli monoculture (14, 15), so assembling them as a genetic device in E. coli allowed us to test these parts in a microbial community. We determined that high-strength constitutive Anderson promoters are poorly optimized devices in monoculture when expressed from high copy number plasmids due to intraspecies selection for inactivating mutations, and all constitutive Anderson promoters were poorly optimized in co-culture due to selection for inactivating mutations stemming from both intraspecies and interspecies competition. We suggest alternative methods for reliable expression of genetic devices undergoing both intraspecies and interspecies competition using regulated promoters and ribosome-binding sites (RBS) that can function in co-culture.
MATERIALS AND METHODS
Strains used in this study
E. coli strains DH5alpha and MG1655 were used where noted, and P. aeruginosa PAO1 was used for all co-culture assays (16). All bacteria were grown in LB Miller broth (Fisher Scientific, DF0446-70-5). E. coli DH5 alpha was purchased from New England Biolabs, E. coli MG1655 was purchased from Addgene.org, and P. aeruginosa PAO1 was a gift from Dr. Stephen Lory. For E. coli with pUC19 plasmids, 100 μg/mL ampicillin (Sigma, A9518) was used for selection, while for E. coli with pSB1C3 plasmids, 25 μg/mL chloramphenicol (Sigma, C0378) was used for selection. For overnight growth, cells were grown in LB Miller broth shaking in 14 mL polystyrene snap-cap tubes (Corning, 352051) at 200–270 revolutions per minute and 37°C. For long-term assays, cells were grown statically at 37°C in 5 mL of LB Miller broth in 14 mL polystyrene snap-cap tubes with 4–5 sterile glass beads (10). Co-cultures of E. coli carrying pSB1C3 plasmids with P. aeruginosa were grown in LB media with chloramphenicol. To obtain roughly even starting co-culture cell concentrations, a 10:1 optical density at 600 nm (OD_600_) ratio of E. coli to P. aeruginosa was used for inoculation. Long-term co-cultures and colony counts used methods from Ellis et al. (10)
Device design and construction
IPTG-inducible ATUM chromoproteins on the pUC19 plasmid with ampicillin resistance were purchased from ATUM (15) and chemically transformed into E. coli DH5alpha cells. J23100 and J23108 Anderson promoters were ordered as primer dimers (Table 1) with AflII and XbaI restriction sites at the appropriate ends, PCR annealed using touchdown PCR from with denaturation at 95°C, annealing from 65°C to 55°C for 10 cycles followed by annealing at 55°C for another 25 cycles, and extension at 72°C, and inserted into the vector containing Tinsel Purple (tsP) and Cupid Pink (cuP) chromoproteins at AflII and XbaI restriction sites that cut out the original IPTG-inducible promoter using standard molecular biology techniques.
tsPurple, a plasmid expressing tsP under the J23110 promoter, was a gift from Anthony Forster (Addgene plasmid # 117848) (17). Site-directed mutagenesis was performed using the Phusion High Fidelity PCR Kit (New England Biolabs) and SDM primers (Table 1) to mutate the J23110 promoter into the J23108 and J23100 promoters. DpnI (New England Biolabs) was used to digest the plasmid, and chemical transformation into E. coli DH5alpha cells was performed to produce plasmids JDS83 and JDS89, respectively. Successful mutagenesis was confirmed by Sanger sequencing using the pRBforEco primer (Table 1) (18). tsPurple and JDS83 were chemically transformed into E. coli MG1655 for co-culture with P. aeruginosa.
To replace the constitutive Anderson promoters and strong RBS with promoters and RBS from the dps and grx (19) operons, dps_promoter and grxA_promoter primers (Table 1) were used to PCR amplify ~400 bp of sequence upstream of those genes. The primers themselves contained XbaI and NdeI restriction sites that were used to insert the PCR products into the tsPurple plasmid at XbaI and NdeI restriction sites, cutting out the J23110 promoter and strong RBS using standard molecular biology techniques. The ligated product was chemically transformed into E. coli DH5alpha cells to produce plasmids JDS108 and JDS109 for the promoters and RBS from the dps and grx (19) operons, respectively. These plasmids were extracted using the ZymoPURE plasmid miniprep kit (Zymo Research) and chemically transformed into E. coli MG1655 for co-culture with P. aeruginosa.
Diagnostic PCR and sequencing
Diagnostic PCR to detect the presence of the chloramphenicol resistance (cmR) gene was performed with primers CAT-F and CAT-R. Diagnostic PCR to detect the presence of the plasmid and to determine whether deletions were present in the genetic device was performed with primers pBRforEco (18) and CAT-R, and Sanger sequencing confirmation was conducted using primers pBRforEco and CAT-Rinv (Table 1). A 1 kb plus DNA ladder (N3200S) was purchased from New England Biolabs.
Native cell lysis
Liquid cultures of each overnight culture were centrifuged at 5,000 × g for 1 min, and the supernatant was removed. Each cell pellet was resuspended in one volume of native cell lysis buffer consisting of 25 mM Tris-HCl pH 8, 50 mM sodium chloride, 2 mM ethylenediaminetetraacetic acid, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1 mg/mL of lysozyme from chicken egg white (Millipore), and incubated for 90–180 min at 37°C. The reaction was then centrifuged at 17,000 × g for 5 min, and the supernatant was removed and placed into cuvettes for spectrophotometry.
Spectrophotometry and data analysis
Visible wavelength spectrophotometry from 400 nm to 800 nm was conducted on a Varian Cary 50 UV-Vis spectrophotometer (Agilent) with a resolution of 1 nm using LB media as a blank for whole-cell measurements or native cell lysis buffer without lysozyme as a blank for native cell lysis measurements. For whole-cell measurements, spectra were routinely above 1 OD in absorbance for the entire visible spectrum, so spectrometry data were aligned such that 800 nm = 1 OD for comparison purposes. For native cell lysis, spectrophotometry data were aligned such that each spectrum’s minimum value = 0 for comparison purposes. For chromoprotein peak calculations, a spectrum of E. coli MG1655 without chromoprotein expression was background-subtracted. Statistical analysis of chromoprotein peak intensities was conducted using the emmeans package in R with the Tukey adjustment (20).
Mutant analysis
E. coli colonies containing plasmids with the tsP gene with no visible chromoprotein expression on LB agar plates with chloramphenicol selection were picked and grown overnight in LB media with chloramphenicol. An additional 20 μg/mL cefsoludin was added to the agar plates and liquid cultures if the E. coli were from co-cultures with P. aeruginosa to prevent P. aeruginosa contamination. Colonies were counted as not expressing tsP chromoprotein if there was no visible color after overnight incubation at 37°C and two additional days at 4°C. Plasmid DNA was isolated using the ZymoPure Plasmid Miniprep Kit (Zymo Research), and Sanger sequenced using either the pBRforEco sequencing primer (18) or CAT-Rinv (Table 1).
Device testing in a wastewater microbial community
Wastewater was collected from the Liberty Wastewater Treatment Facility in Liberty, Missouri, from the effluent prior to any treatment. LB media was inoculated with wastewater, then increasing concentrations of chloramphenicol were added, and OD_600_ and percentage of colonies expressing tsP were measured to determine the appropriate concentration for the long-term assay. Every 2 days, cell pellets were collected, colony PCR was conducted using pBRforEco and CAT-R primers to detect the presence of the plasmid, and on days 6 and 12, cell pellets underwent native cell lysis and spectrophotometry. Three biological replicate cell pellets, each of wastewater grown overnight in LB media, grown 12 days in LB media, grown overnight in LB with 10 μg/mL chloramphenicol, and grown 12 days in LB with 10 μg/mL chloramphenicol, co-cultured overnight with E. coli MG1655 containing the JDS108 plasmid, and co-cultured for 12 days with E. coli MG1655 containing the JDS108 plasmid, were collected and sent to Genewiz from Azenta Life Sciences for 16S rRNA metagenomic sequencing (16S EZ service). Amplicon sequence variants (ASVs) were inferred using dada2 (21), and taxa were assigned to ASVs using the Silva training set version 138.2 (22). ASV abundances were graphed with phyloseq (23).
RESULTS
ATUM chromoprotein expression is quantifiable
To determine whether chromoprotein expression was robustly quantifiable, we first measured the visible absorbance spectra of 11 IPTG-induced chromoproteins expressed from the lac operon promoter on the pUC19 plasmid obtained from ATUM (15) in E. coli and compared it with the visible absorbance spectrum of E. coli MG1655 without the plasmid. While all E. coli expressing chromoproteins formed pellets with qualitatively visible chromoprotein color as advertised by ATUM, their quantifiable absorbance peaks were highly variable. Of the purple chromoproteins, tsP, Vixen Purple, and Prancer Purple had obvious absorbance peaks in the expected 580–590 nm range that had 0.27–0.54 greater absorbance than the E. coli MG1655 background (Fig. 1A), while cuP had the largest absorbance peak of the non-purple chromoproteins in the 530–535 nm range that was 0.28–0.32 greater absorbance than the E. coli MG1655 background (Fig. 1B). The large purple peaks found in tsP, Vixen Purple, and Prancer Purple also had secondary peaks or shoulders at shorter wavelengths. Small peaks that had 0.07–0.16 greater absorbance than the E. coli MG1655 background were also noted for the rest of the purples and for Blitzen Blue, Dreidel Teal, and Serpha Pink, while Scrooge Orange had two small peaks at 505–515 nm and 545–555 nm. We conclude that chromoprotein expression is quantifiable in cells, as all the chromoproteins produced quantifiable primary peaks, but due to the variability of secondary peaks and shoulders, we restrict our analysis to the primary peaks.
Chromoprotein absorbance spectroscopy. Absorbance spectroscopy of 11 IPTG-inducible chromoproteins grown in E. coli purchased from ATUM (15). Native cell lysis of cells expressing chromoproteins to preserve protein structure, followed by pelleting and removal of cellular debris, reduces background for measuring chromoprotein expression in monoculture. (A) Purple chromoproteins in whole-cell spectroscopy. (B) Other chromoproteins in whole-cell spectroscopy. (C) Purple chromoproteins after native cell lysis. (D) Other chromoproteins after native cell lysis. The black line represents control E. coli cells not carrying chromoprotein plasmids.
To reduce the high background we observed in whole-cell spectroscopy measurements, we performed native cell lysis to remove insoluble materials that scatter light, such as the lipid membrane and cell wall polymers, and measured the clear cell lysates after insoluble material was pelleted in the visible wavelength spectrum. Native cell lysis did reduce background, but it was a mixed success at keeping chromoproteins intact and functional as determined by absorbance peaks of the lysates. Of the purple chromoproteins, tsP, Prancer Purple, and Donner Magenta benefited from native cell lysis, with peaks that had 0.05–0.1 greater absorbance than the E. coli MG1655 background, Vixen Purple’s absorbance spectrum had a reduced peak after native cell lysis than in whole-cell measurements, while the already small peaks for Maccabee Purple and Virginia Violet were eliminated (Fig. 1C). In the non-purple chromoproteins, cuP and Scrooge Orange had peaks with 0.08–0.11 greater absorbance than the E. coli MG1655 background, Dreidel Teal and Serpha Pink had peaks with only 0.01–0.02 greater absorbance than the E. coli MG1655 background, and the peak for Blitzen Blue was eliminated (Fig. 1D). Although native cell lysis did reduce the peak intensities of all the chromoproteins to varying degrees, it also reduced the background so significantly that it allowed for more precise measurements of the highest peak intensities, validating this method for use in chromoprotein peak measurements.
Origin of replication and promoter strength affect the successful performance of the constitutive Anderson promoter-driven chromoprotein genetic devices
Based on the highest peak intensities in visible wavelength absorbance spectra of native cell lysates, tsP, and cuP were chosen as the chromoproteins to be paired with constitutive Anderson promoters. We replaced the IPTG-inducible lac operon promoter on the pUC19 plasmid with constitutive Anderson promoters J23100 and J23108 (relative measured strengths = 1 and 0.51, respectively) (14), while retaining the strong T7 phage gene 10 RBS (24) (Fig. 2A). We were successful in producing those constructs as confirmed by chromoprotein expression during cloning and subsequent Sanger sequencing but saw highly variable chromoprotein expression across clonal isolates in both liquid cultures grown overnight under chloramphenicol selection and on plates under chloramphenicol selection in all our constructs regardless of predicted constitutive Anderson promoter strength. This was true for constructs containing both tsP (Fig. 2B) and cuP (Fig. S1A) chromoproteins, demonstrating that it is difficult to have a reliable construct with predictable chromoprotein expression from the pUC19 plasmid.
Chromoprotein expression is highly variable under high copy number origin of replication or high-strength promoter. (A) Device design. (B) Spread plate of high-strength promoter driving tsP chromoprotein expression on the pUC19 plasmid. (C) Lower promoter strength stabilizes expression. Absorbance at 580–590 nm for tsP expressed from constitutive Anderson promoters. N = 10 for high, and 16 for medium and low. (D) Point mutations in the high-strength constitutive Anderson promoter (blue) and Shine-Dalgarno consensus RBS (green) in mutants with no tsP expression. Top line, original sequence; lower lines, sequences of low- and no-expression isolates.
We observed an inconsistency in chromoprotein expression in supposedly clonal strains. We hypothesized that inactivating mutations were being selected for due to the very high copy number origin of replication on the pUC19 plasmid, incurring a high resource cost (25). This hypothesis was supported in a study by Liljerumn et al. (17), who produced stable tsP chromoprotein expression under the control of the J23110 Anderson promoter and a strong consensus prokaryotic RBS (26), but from the pSB1C3 plasmid, which is also considered a high copy number plasmid, but produces only 100–300 copies per cell (27) compared to the 500–700 copies per cell expected from the pUC19 plasmid (25). We obtained this Liljerumn et al. plasmid, named tsPurple, and used site-directed mutagenesis to mutate the J23110 promoter (relative measured strength = 0.33, low strength) into the J23100 and J23108 promoters (relative measured strengths = 1 and 0.51, high and medium strength, respectively) (14), then spread plated each of these constructs after overnight liquid culture, with both plates and liquid culture under chloramphenicol selection (Fig. S1B). From these spread plates, we picked and grew up 10–16 single colonies in liquid culture under chloramphenicol selection, performed native cell lysis, measured the absorbance spectra, and quantified the maximum peak absorbance values after subtracting the E. coli MG1655 absorbance spectrum background (Fig. 2C). The high-strength promoter had the most variability of tsP chromoprotein expression among the individual colonies, with some clonal isolates not producing any chromoprotein expression after native cell lysis. The high-strength promoter was not significantly different from the medium-strength promoter. It did have significantly greater peaks compared to the low-strength promoter (P = 0.0058), but we determined that it was too variable to be used in future assays. Conversely, both the medium-strength and low-strength promoters combined with tsP expressed on the pSB1C3 plasmid in E. coli produced clonal isolates with predictable tsP chromoprotein peaks in the 580–590 nm range as quantified by absorbance spectra of native cell lysates. While the medium-strength promoter produced more variable peak intensities than the low-strength promoter, it had reliably greater peaks than the low-strength promoter (P = 0.0139), leading us to conclude that these promoters driving tsP expression were reliable enough to be used in future assays.
We suspect that the high-strength promoter was producing selection against tsP chromoprotein production due to high metabolic resource usage, much like the high copy number origin on pUC19 did, and therefore mutations occurring in the promoter or CDS that disabled or reduced chromoprotein production would provide a fitness advantage that was selected for over time. We grew five independent inoculations of E. coli with the high-strength constitutive Anderson promoter driving tsP chromoprotein expression in monoculture over 12 overnight passages in static conditions, plated dilutions of the final culture, screened for E. coli colonies with no visible tsP chromoprotein despite chloramphenicol selection, and Sanger sequenced the promoter, RBS, and tsP CDS of 24 of these isolated colonies from across the five independent inoculations (Fig. 2D). All the isolates had at least one point mutation in the promoter, with a few that had many different point mutations in the promoter and RBS, supporting our hypothesis that mutations that reduced or disabled chromoprotein production conferred a growth advantage that was selected for over time. The most common point mutation that all the isolates had was the G to T transversion in the third nucleotide of the promoter. This transversion is found in 8 out of 18 Anderson promoters that have reduced measured promoter strength compared to the strongest constitutive Anderson promoter (14), suggesting that this position influences promoter strength. Some isolates had mutations in the middle region, which is constant among most constitutive Anderson promoters, and this would likely lead to a loss of promoter function due to the strong sequence conservation found in that region of constitutive Anderson promoters (14).
Growth of constitutive Anderson promoter-driven genetic devices in co-culture with P. aeruginosa eliminates chromoprotein expression in E. coli by producing selection against functional devices
Since tsP expression from the pSB1C3 plasmid driven by the medium- and low-strength promoters had predictable expression in E. coli MG1655 monoculture, we decided to co-culture E. coli MG1655 containing each device with P. aeruginosa PAO1, as it is well established that E. coli MG1655 and P. aeruginosa PAO1 can exist in stable co-culture in static conditions (6, 28–30). We grew this co-culture under chloramphenicol selection to keep the plasmid in E. coli, as P. aeruginosa PAO1 is naturally chloramphenicol-resistant (31). We found that hosting neither genetic device changed the ability of E. coli to participate in co-culture with P. aeruginosa (Fig. 3A and B), as colony-forming units (CFU) per mL (CFU/mL) were comparable with prior co-culture results seen by Ellis et al. (10) We could not detect any tsP expression in liquid culture, either by eye or by native cell lysis and spectrophotometry at any point in our co-culture beyond the original day of inoculation (Fig. 3C). Monocultures of E. coli with the genetic devices being grown statically and passaged for the same amount of time as the co-cultures reliably expressed tsP (Fig. 3C). One possible explanation for not detecting tsP in co-culture is that the presence of cell lysate material from P. aeruginosa could interfere with the absorbance of tsP, while another explanation is that tsP is not being expressed in co-culture. To determine which of these possibilities is true, we mixed statically grown monoculture populations of E. coli and P. aeruginosa at similar ratios as seen in long-term co-culture and immediately conducted native cell lysis on them (Fig. 3D). While the peaks in the lysate of mixed monocultures were lower and the background higher than in E. coli monocultures, the peaks were visible in these controls, suggesting that interference by P. aeruginosa cell lysate debris alone cannot account for the complete lack of tsP peaks in long-term co-cultures. This result suggests that the co-culture prevents some or all tsP expression either by reducing transcription and/or translation of tsP or by selecting for mutations that inactivate the functional device.
*Genetic devices in E. coli co-cultured with P. aeruginosa do not alter community stability but result in no tsP chromoprotein expression. (A) Log 10 of CFU/mL for E. coli expressing tsP from constitutive Anderson promoters (purple) co-cultured with P. aeruginosa (green). N = 6. (B) Colony counts for E. coli (black) co-cultured with P. aeruginosa (green). N = 4. Thin lines show individual experiments, and thick lines show the average. (C) No detectable tsP expression from E. coli grown in co-culture with P. aeruginosa. Absorbance of medium-strength (purple) or low-strength (pink) promoters driving tsP expression in monoculture (solid) or co-culture with P. aeruginosa (dashed). E. coli (black) was included as a negative control. (D) P. aeruginosa cell debris presence in lysate cannot completely account for the loss of tsP detection in co-cultures. E. coli was grown in monoculture and then mixed (dotted) or not (solid) with P. aeruginosa directly before lysis and measurement. Colors as in C. Proportion of E. coli colonies carrying functional devices with medium- (E) and low- (F) strength promoters after co-culture with P. aeruginosa is lower than in monoculture over time. N = 4. P < 0.05 comparing monocultures to co-cultures at days 6 and 12, Student’s T-test. (G) Qualitative summary of mutations found in medium- and low-strength promoters and Shine-Dalgarno consensus RBS in 32 mutants with no tsP expression. Sections indicated by black brackets are deletions. The mutations are as follows: (I) point mutations in the promoter, (II) 79 bp frameshift deletion in tsP (2–80) resulting in an early STOP after 46 amino acids, (III) 78 bp in-frame deletion in tsP removing 26 amino acids, (IV) deletion of the RBS and tsP CDS, (V) deletion of the promoter, RBS, and first 400 bp of tsP, and (VI) deletion of part of the promoter, the RBS, and tsP CDS.
To look for mutations that inactivate the functional device, we tested whether the device was functional by plating liquid co-cultures on plates with cefsulodin selection, which excludes P. aeruginosa, and chloramphenicol selection, which retains the device. Since a colony on a plate is a physiological condition in which functional devices would produce tsP, any white colonies on these plates would be due to mutants that formed and were selected for during co-culture. Over time, we observed a decrease in E. coli colonies that showed expression of tsP after co-culture with P. aeruginosa compared to E. coli grown in monoculture despite antibiotic selection on plates for the plasmid (Fig. 3E and F). As a control, we removed chloramphenicol selection on plates, resulting in a more rapid decline of tsP expression in the E. coli colonies than on plates with chloramphenicol selection for both monocultures and co-cultures (Table S1). This control argues against plasmid loss as the cause of most of the lost tsP expression in E. coli co-cultured with P. aeruginosa. Much like the high-strength promoter is producing selection against tsP expression in E. coli grown in monoculture (Fig. 2C), we conclude that E. coli in co-culture with P. aeruginosa is producing increased selection against tsP expression compared to E. coli monoculture in the medium- and low-strength promoters, and therefore selecting for inactivating mutations occurring in the promoter or CDS over time.
Sanger sequencing of the promoter, RBS, and tsP CDS of 22 E. coli colonies with devices with medium- or low-strength constitutive Anderson promoters with no tsP expression after co-culture with P. aeruginosa for 12 days, and in 10 E. coli colonies grown in monoculture for 12 days, all showed either point and/or deletion mutations (Fig. 3G; Table S2). The point mutations were in the conserved middle of the promoters, similar to prior mutations seen in the high-strength constitutive Anderson promoters (Fig. 2D), and the deletions removed some or all of the promoter region and some or all of the tsP CDS. Two deletions were frameshift deletions within the tsP CDS. One deletion is predicted to produce a truncated 46 amino acid peptide, while the other deletion is an in-frame deletion, but an Alphafold 3 prediction of the mutant (32) produces an in-frame deletion that removes the majority of the first strand and all of the second strand of the beta barrel, causing incomplete beta barrel formation and suggesting a loss of function due to loss of beta barrel integrity.
A genetic device utilizing a regulated promoter results in detectable expression in co-culture
While constitutive Anderson promoters may only be suitable for E. coli monoculture, it may be desirable in the future for genetic devices that interact with microbial communities to have high expression either only in communities or in both communities and in monoculture. In long-term co-culture of E. coli MG1655 with P. aeruginosa PAO1 (10), we used RNA sequencing to discover genes in E. coli that were more highly expressed in co-culture with P. aeruginosa than in monoculture. We selected dps and grxA as genes with high base expression in monoculture and higher expression in co-culture than monoculture, with dps having roughly an order of magnitude more expression than grxA in both scenarios (Fig. 4A). We asked whether genetic devices could be optimized for co-culture by utilizing promoters and RBS from these genes. We replaced the constitutive Anderson promoters and the consensus strong RBS with promoters and RBS from the dps and grx operons in E. coli MG1655 (19) while keeping the tsP chromoprotein and origin of replication on the pSB1C3 plasmid intact (Fig. 4B), creating P_dps_-tsP and P_grxA_-tsP devices, respectively. We found that neither genetic device in E. coli significantly changed the proportion of E. coli in co-culture with P. aeruginosa. There was no statistical difference in CFU/mL with E. coli MG1655 co-cultured with P. aeruginosa and prior co-culture results (10) (Fig. S2). Monoculture of E. coli with both devices expressed tsP (Fig. 4C), with P_dps_-tsP producing much greater peaks than P_grxA_-tsP, in line with their baseline E. coli expression (Fig. 4A). In long-term co-culture with P. aeruginosa, E. coli with the P_dps_-tsP device produced small but detectable amounts of tsP, while E. coli with the P_grxA_-tsP device failed to produce any detectable tsP in co-culture with P. aeruginosa using the cell lysis assay (Fig. 4C). Monocultures of statically grown populations of E. coli with the P_dps_-tsP device mixed with P. aeruginosa at similar ratios as seen in long-term co-culture with native cell lysis produced reduced but still visible peaks (Fig. 4D), similar to what was seen in long-term co-culture (Fig. 4C), suggesting that the P_dps_-tsP device functioned, expressing tsP as well in co-culture with P. aeruginosa as in monoculture. Both regulated promoter devices we used experienced selection against functional devices across our 12-day assay (Fig. 4E and F). E. coli with P_dps_-tsP had an increased proportion of colonies expressing tsP when co-cultured with P. aeruginosa from day 6 to day 12 (Fig. 4E), suggesting either sampling bias or drift. Monocultures of E. coli with either device initially had high percentages of colonies expressing tsP, but by day 12, these percentages were similar to those of colonies in co-culture with P. aeruginosa, suggesting comparable expression levels and selection for inactivating mutations in both co-culture and monoculture.
Co-culture optimized promoters improve tsP expression. (A) Differential gene expression of E. coli after co-culture with P. aeruginosa. Black dots are genes significantly upregulated or downregulated in co-culture compared to monoculture. Base mean represents the average monoculture expression. Green diamond = dps, Blue triangle = grxA. (B) Device design. (C) Pdps-tsP produces measurable chromoprotein expression in co-culture. Absorbance of Pdps (green) or PgrxA (teal) promoters driving tsP expression in monoculture (solid) or co-culture with P. aeruginosa (dashed). E. coli (black) was included as a negative control. (D) Pdps-tsP expression from E. coli is detectable when mixed with P. aeruginosa. E. coli was grown in monoculture and then mixed (dotted) or not (solid) with P. aeruginosa directly before lysis and measurement. Colors as in C. Proportion of E. coli colonies carrying functional devices with Pdps-tsP (E) or PgrxA-tsP (F) in co-culture with P. aeruginosa. N = 4.
At the end of our long-term assay with the regulated promoter devices, we isolated plasmid DNA from 61 E. coli colonies with no visible tsP expression, 24 from the P_dps_-tsP device, and 21 from the P_grxA_-tsP device grown in co-culture with P. aeruginosa for 12 days and 8 E. coli colonies grown in monoculture for 12 days for each regulated promoter device, and conducted diagnostic PCRs to detect the presence of various parts of the plasmid. PCR with primers internal to the cmR gene on the plasmid produced mostly positive hits, with 16/17 diagnostic reactions, 9 for P_dps_-tsP, and 8 for P_grxA_-tsP, producing the expected 511 bp band (Fig. 5A), and the one not producing the expected 511 bp band produced a slightly bigger band, indicating retention of the cmR gene in most colonies. Sanger sequencing of colonies from regulated devices that had no visible tsP expression when co-cultured with P. aeruginosa using the pBRforEco primer revealed some of the colonies had mutations and some did not, with 6/9 P_dps_-tsP and 1/14 P_grxA_-tsP colonies producing chromosomal integration events excluding the tsP CDS (Fig. 5C, mutation II, Table S3). Almost half of the Sanger sequencing reactions using pBRforEco produced no result, suggesting a lack of primer binding. Diagnostic PCR with pBRforEco and CAT-R of devices grown in co-culture with P. aeruginosa for 12 days revealed that 22/45 colonies, 15 from P_dps_-tsP and 7 from P_grxA_-tsP, did not produce any band at all (Fig. 5B). Since the cmR gene was confirmed to be present (Fig. 5A), we suspected plasmids from these colonies were missing the pBRforEco primer-binding site, so we sequenced with CAT-Rinv, an inverted sequence of CAT-R primer found in the cmR gene which is predicted to be 227 bp upstream from pBRforEco. Sanger sequencing with CAT-Rinv resulted in the discovery of additional chromosomal integration events and deletions that resulted in loss of the pBRforEco primer-binding site (Fig. 5C, mutations I, IV, and V, Table S3). Overall, 21/24 isolated E. coli colonies with P_dps_-tsP co-cultured with P. aeruginosa with no tsP expression had deletion mutations or random chromosomal integration events, and 8/21 isolated E. coli colonies with P_grxA_-tsP co-cultured with P. aeruginosa had deletion mutations or random chromosomal integration events. Meanwhile, 3/8 isolated E. coli colonies with P_dps_-tsP grown in monoculture produced the same mutation (Fig. 5C, mutation III, Table S3), a deletion of most of the promoter and RBS and 78 bp of the tsP CDS with a small 15 bp insertion, while none of the E. coli colonies with P_grxA_-tsP grown in monoculture displayed a mutation. Since all the colonies we isolated had no visible tsP expression, this suggests that there are other mutations responsible for the phenotype, either elsewhere on the plasmid or in the E. coli MG1655 genome, that we could not detect.
Intraspecies and interspecies competition selects for different types of mutations. (A) The cmR gene is retained during co-culture. Co-cultured E. coli isolates with PgrxA-tsP (1–8) or Pdps-tsP (9–17), + = positive control, L = 1kb ladder. (B) Many co-cultured isolates have deletions in the device. PCR assay for the presence of the pBRforEco-binding site. Co-cultured E. coli isolates with PgrxA-tsP (1–4) or Pdps-tsP (5–8), + = positive control, L = 1kb ladder. (C) Summary of mutations from 61 co-culture and monoculture mutants with no tsP expression. Black brackets represent deletions. Green brackets represent plasmid regions that were integrated into the E. coli chromosome. The mutations are as follows: (I) Chromosomal integration of the cmR gene, (II) chromosomal integration of the cmR gene and most of the promoter and RBS region (III) replacement of the sequence encompassing the promoter and beginning of tsP with a 15 bp insertion (TGGTGGCAAGTTATG), (IV) deletion between the pBRforEco primer and 329 bp of tsP, and (V) deletion between 40 bp downstream of cmR and 394 bp of tsP.
A regulated promoter and RBS device is functional in a natural microbial community
Since E. coli with P_dps_-tsP functioned in co-culture with P. aeruginosa, we wanted to assess how it would perform in a natural community over time. To do this, we obtained untreated wastewater effluent from a local wastewater treatment facility and tested the device in the presence of diverse and abundant wastewater bacteria to determine whether tsP would still be expressed and whether it would impact the wastewater microbial community.
Our P_dps_-tsP device is on a plasmid with the cmR gene, so it is important that selection for the device is maintained with chloramphenicol. Therefore, we had to determine what concentration of chloramphenicol would have minimal effect on the natural community members of the wastewater while still maintaining selection for the device, which we experimentally determined to be 10 μg/mL (Fig. 6A). We performed a modified long-term assay under 10 μg/mL instead of 25 μg/mL chloramphenicol selection for 12 days. Triplicate cell pellets of the wastewater co-cultured with E. coli with P_dps_-tsP were purple to the naked eye for all 12 days, with a visible decrease in tsP expression by day 10, demonstrating expression of the device (Fig. 6B). Plasmid extraction and diagnostic PCR using pBRforEco and CAT-R primers targeting the plasmid confirmed the presence of plasmid (Fig. S3), and spectrophotometry of the cell lysate on days 6 and 12 showed obvious tsP peaks (Fig. 6C), leading us to conclude that our P_dps_-tsP device remained present and functional in the natural wastewater community under 10 μg/mL chloramphenicol selection.
Expression of a regulated promoter device in a wastewater community. (A) Determination of chloramphenicol concentration. OD600 of wastewater microbes after overnight growth in chloramphenicol (black line). N = 6. Percent of purple isolates from E. coli with Pdps-tsP grown overnight in chloramphenicol (gray bars). N = 3. (B) Visible tsP expression in the wastewater community cell pellets. Each row is a biological replicate. (C) tsP expression of E. coli with Pdps-tsP cultured for 6 days (solid green) and 12 days (dashed green) with wastewater community. E. coli monoculture (black) and wastewater community (gray) were included as negative controls. (D) Normalized abundance of the genera of the most common 400 taxa identified by 16S rRNA sequencing from wastewater inoculum grown in LB (control), LB with 10 μg/mL chloramphenicol (selection), or LB with 10 μg/mL chloramphenicol and co-inoculated with E. coli carrying Pdps-tsP (device).
We performed 16S rRNA metagenomic sequencing to determine what members were in the wastewater community and if our P_dps_-tsP device with chloramphenicol selection altered the community (Fig. 6D). Chloramphenicol selection did enrich for some of the most common genera originally found in the wastewater after 1 day of growth compared to growth without chloramphenicol, but it merely accelerated enrichment that long-term growth in rich LB media promoted, as there was little difference in the abundance of the top three genera after 12 days for growth in LB with or without chloramphenicol selection or between these and 1 day of chloramphenicol selection. Adding E. coli with P_dps_-tsP increased the proportion of Escherichia-Shigella in both day 1 and day 12, as would be expected if E. coli with P_dps_-tsP were maintained as a significant portion of the microbial community, but did not significantly alter the frequency of the other top two genera in the microbial community grown with 10 μg/mL chloramphenicol in either day 1 or day 12, leading us to conclude that our functional device had minimal impact on wastewater community composition.
DISCUSSION
Our goal is to determine whether robust genetic parts that are designed for predictable use in monoculture are as robust as claimed when put into complex microbial communities. This is particularly relevant for future genetic devices designed to function in complex microbial environments. Based on our findings, we conclude that current genetic devices optimized for monocultures are not robust when asked to perform the same functions in complex microbial communities. Conversely, at no point in our experiments did it seem that our genetic devices had any significant effect on community composition within a complex microbial community, either in E. coli co-cultured with P. aeruginosa (Fig. 3A) or in the wastewater microbial community (Fig. 6D). We cannot rule out changes in genetic expression and community function in organisms when a functional genetic device is present that could be detectable using transcriptional profiling, but complex microbial communities seem to be structurally robust in a way that genetic devices themselves are not.
While we acknowledge that the primary objective of ATUM was to create chromoproteins with qualitative visibility, it was surprising to us that some of them were difficult to quantify using whole-cell absorbance spectroscopy in the visible wavelengths, as the chromoprotein peaks were highly variable in size. This quantifiable variability was further amplified by our native cell lysis procedure, which did succeed in dramatically reducing background but at the cost of reducing chromoprotein peak intensities overall. While the largest peaks were more precise following native cell lysis, many of the small peaks were eliminated. Due to the proprietary nature of these chromoproteins, we do not know their protein structure and thus cannot speculate on specific reasons why one chromoprotein may have kept or lost its peak(s), but we can conclude that our attempt to preserve protein structure using our native cell lysis procedure needs refinement. There is likely some protein denaturation, and some modifications to this procedure in the future to reduce protein denaturation while preserving cell lysis could be considered, such as adjusting salt, pH, or buffer concentrations to preserve protein folding or increasing the amount of PMSF protease inhibitor.
We originally created devices with two different chromoproteins and two constitutive Anderson promoters of varying strengths on a plasmid with a high copy number origin of replication. The chromoprotein expression of these devices, as seen on plates, was determined to be highly variable, with no predictable pattern. This variability led us to switch to a plasmid with a lower copy number origin of replication and focus on one chromoprotein, which produced predictable expression of tsP chromoprotein at lower constitutive Anderson promoter strengths. While the RBSs were different on both plasmids, they were both previously known to be strong RBSs (24, 26), so we chose not to alter those when constructing our devices to reduce the complexity of our study. In addition to promoter strength and origin of replication, variation in RBS is another way to vary genetic device expression and could be the focus of future studies. We show that switching from pUC19 to pSB1C3 provides some stability and predictability in genetic device function. We attribute this to the lower cost of expressing a protein from a lower copy number plasmid, though it remains formally possible that the switch from ampicillin to chloramphenicol selection lessened the fitness cost of the chromoprotein device. One way to eliminate the influence of copy number and antibiotic selection on genetic device function would be to integrate genetic devices into the microbe’s genome and avoid plasmids entirely, thus providing stability with a low copy number that would result in less costly genetic device function. However, this would likely come with the trade-off of lower protein expression and, therefore, negatively impact genetic device functionality. As long as initial genetic device construction is conducted on plasmids, plasmid copy number will need to be considered for device validation, and there could be situations when and where it may be suitable to have the genetic device only temporarily in an organism, such as when the microbe naturally exists in the environment and the genetic device is intended for temporary function, like in immediate bioremediation.
Regardless of whether genetic devices are integrated into the genome, high promoter strength can interfere with reliable genetic device function. In our study, once the origin of replication strength was reduced, the constitutive Anderson promoter with the highest promoter strength still demonstrated high variability in chromoprotein expression. Previous work has shown that high-strength constitutive Anderson promoters lose expression upon entering a competitive microbial environment (33) and negatively affect E. coli growth (34), while in a prior study, E. coli with the low-strength constitutive Anderson promoter driving tsP expression from the pSB1C3 plasmid had 75%–80% of the growth rate E. coli compared with a promoterless control plasmid (17). Together, these results lead us to hypothesize that expressing genetic devices reduces growth rates in E. coli, but genetic devices on a high copy number plasmid or with the strongest constitutive Anderson promoter reduce growth rates to the point of inordinately selecting for mutations that inactivate tsP chromoprotein expression. Some of the tsP expression variability from the pUC19 plasmid could also be from selection for similar inactivating mutations in promoters, so it is likely that copy number and promoter strength additively contribute to the cost of a genetic device.
E. coli expressing our constitutive Anderson promoter genetic devices in co-culture with P. aeruginosa failed to express any measurable tsP chromoprotein, despite the equivalent E. coli monoculture producing robust chromoprotein expression, and mixtures of monocultures also producing visible, albeit weaker, chromoprotein peaks. E. coli with P_dps_-tsP had high tsP expression in monoculture and low tsP expression in long-term co-culture with P. aeruginosa, while E. coli with P_grxA_-tsP had low tsP expression in monoculture and no tsP expression in long-term co-culture with P. aeruginosa (Fig. 4C). We propose two explanations for these results. First, there is a dampening effect from P. aeruginosa cell lysate, which we found in E. coli monoculture controls of both constitutive Anderson promoters expressing tsP (Fig. 3D) and regulated devices (Fig. 4D) mixed with P. aeruginosa. It is unknown how much of the lack of tsP chromoprotein expression in co-culture with P. aeruginosa is due to selection for mutants and how much is due to changes in gene expression of functional devices due to the co-culture context. One line of evidence suggesting that expression from chromoprotein devices is reduced during co-culture is that both day 6 and day 12 E. coli monocultures express enough chromoprotein to be easily measured from constitutive Anderson devices, despite the presence of significant inactivating mutations, suggesting that loss of functional devices to mutation cannot explain all the loss of chromoprotein expression in co-culture.
In our co-culture experiments, we tried to recreate the conditions found in wastewater as much as possible in a lab environment, but we acknowledge slight differences due to our need to optimize our E. coli–P. aeruginosa co-culture in the lab. For example, optimal temperatures for wastewater bacterial activity are between 25°C and 35°C (35), while optimal temperature for E. coli and P. aeruginosa in the lab is 37°C. This slight temperature variation should have a negligible impact as it falls within the acceptable temperature range for mesophilic wastewater bacteria of 10°C–45°C (36). Wastewater also has a low oxygen content of around 2 mg/L during treatment (37). We have attempted to recreate such low oxygen conditions by growing our culture statically (10), but we acknowledge that there is variation within the statically grown tubes, such that the surface of the tube is more aerobic than the bottom.
We observed an increase in inactivating mutations in co-cultures of E. coli with medium- and low-strength constitutive Anderson promoters expressing tsP with P. aeruginosa compared to monoculture (Fig. 3E and F), suggesting that co-culture plays a role in selecting for inactivating mutations as it imposes additional interspecies selection pressures. E. coli is in a very different physiological and metabolic state in co-culture with P. aeruginosa compared to monoculture. It undergoes a starvation response and switches away from aerobic respiration (10), factors that could make gene expression more costly and therefore select for inactivating mutations. This suggests that while genetic devices always incur a cost, there is a cost threshold at which selection results in loss of functional devices from the population. In our work here, co-culture had a lower cost threshold, perhaps due to the added interspecies competition, than monoculture did.
It is interesting to note that the type of mutations we saw differs by the type of genetic device utilized. E. coli with medium- and low-strength constitutive Anderson promoters experiences both point and deletion mutations regardless of whether they are in monoculture or co-culture (Fig. 3G; Table S2). This is different from the high-strength promoter, where we saw only point mutations (Fig. 2D). We saw that the high-strength constitutive Anderson promoter imposed a large resource burden on E. coli causing loss of tsP expression in monoculture (Fig. 2C; Fig. S1B) and wondered if that burden could explain the selection for point mutations. Surprisingly, we observed no point mutations in the inactivated regulated promoters. Point mutations are more common than other types of mutations, and among other types of mutations, deletions are more common than insertions (38). Most insertions and deletions tend to be 50 bp or less (38), making the deletions and random integrations we see in our inactivated regulated promoters unusual. We speculate that these longer deletions and integrations are more effective in inactivating a genetic device but occur less frequently. Therefore, there are instances, such as utilizing the high-strength constitutive Anderson promoter, where the selection pressure against the functional genetic device is so strong that the need to inactivate the device selects for more common point mutations as the mechanism of inactivation. Conversely, if the selection pressure is not as strong, then point mutations may not produce a big enough phenotypic difference to warrant selection, so over time, the mutations that are selected for are the more effective, longer deletions and integrations that remove large portions of the genetic device.
We observed loss of function mutations arising in both constitutive Anderson devices and regulated promoter devices over time (Fig. 4E and F). However, the P_dps_-tsP device retained functionality in the native lysis assay (Fig. 4C and 6C) and in the wastewater microbial community (Fig. 6B). This suggests that P. aeruginosa lysate masking and loss of function mutations cannot completely explain why the constitutive Anderson devices do not function during co-culture. Ellis et al. (10) have shown that E. coli undergoes the starvation response when in co-culture with P. aeruginosa. Even though constitutive Anderson promoters are conceived of as constitutively expressed, this expression relies on the “housekeeping” σ^70^ sigma factor, which is modulated during the starvation response (39, 40). It is likely that the starvation response increases competition between housekeeping and stress response sigma factors for the RNA polymerase holoenzyme, which could lead to an indirect downregulation of constitutive promoters during stress, such as co-culture with P. aeruginosa. Dps plays a role in protecting DNA during oxidative stress and starvation and is upregulated during the starvation response (41). The dps promoter is bound by σ^70^, but also contains a consensus σ^S^-binding site (GCTATACTTAA) in the −10 position (42, 43). σ^S^ is the stress sigma factor, so it would likely be activated during co-culture with P. aeruginosa when E. coli is starving (10). Therefore, dps and other genes that are activated by Sigma S (44) are good candidates for reliable promoters to drive the expression of genetic devices in the context of co-culture with P. aeruginosa due to activation of the starvation response. Thus, to create a functional device, it is important to consider the RNA transcript data and the regulatory environment of the gene, even when using nominally “constitutive” promoters.
Putting a genetic device in a microbe that provides no survival or growth benefit to that microbe while incurring a resource utilization cost introduces a selective pressure to reduce the expression of that genetic device because metabolic and gene expression resources that could be used for growth are diverted to genetic device function. This is a well-known phenomenon (45–47) further supported by reduced growth rates found in E. coli expressing chromoproteins (17) and by the increase in mutations selected for in co-cultures found during our current study. Selective pressure can be a result of intraspecies competition within one species, as microbes with the functional genetic device are at a competitive disadvantage with the same species microbes with a mutated nonfunctional or reduced-function genetic device. Furthermore, when putting a microbe with a genetic device in an environment with different species, the additional selective pressure of interspecies competition is introduced. This additional interspecies selective pressure has profound consequences, as even devices that seem to function reliably with low selection bias in monoculture may be stretched beyond their functional breaking point in a more complex microbial community. Our results support the need to test next-generation genetic devices in model microbial communities to make sure they can function as designed before any large-scale environmental release.
Conclusions
Based on our study, we reject the hypothesis that genetic devices and parts can have predictable functionality regardless of what other genetic parts they are coupled with or the environmental context they are placed in. Future design and construction of robust genetic devices or parts for predictable functionality need to account for their environmental conditions and genetic context, not just for the organism they are to be expressed. For complex microbial communities, we suggest not using constitutive promoters and using regulated promoters that are optimized for expression in those communities. Constructing genetic devices that will give the microbe with that device a selective survival or growth advantage in a particular environment could also enable robust genetic device function.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Goeddel DV, Kleid DG, Bolivar F, Heyneker HL, Yansura DG, Crea R, Hirose T, Kraszewski A, Itakura K, Riggs AD. 1979. Expression in Escherichia coli of chemically synthesized genes for human insulin. Proc Natl Acad Sci USA 76:106–110. doi:10.1073/pnas.76.1.10685300 PMC 382885 · doi ↗ · pubmed ↗
- 2Steinberg FM, Raso J. 1998. Biotech pharmaceuticals and biotherapy: an overview. J Pharm Pharm Sci 1:48–59.10945918 · pubmed ↗
- 3Gutierrez T, Singleton DR, Berry D, Yang T, Aitken MD, Teske A. 2013. Hydrocarbon-degrading bacteria enriched by the Deepwater Horizon oil spill identified by cultivation and DNA-SIP. ISME J 7:2091–2104. doi:10.1038/ismej.2013.9823788333 PMC 3806270 · doi ↗ · pubmed ↗
- 4Jacquin J, Cheng J, Odobel C, Pandin C, Conan P, Pujo-Pay M, Barbe V, Meistertzheim A-L, Ghiglione J-F. 2019. Microbial ecotoxicology of marine plastic debris: a review on colonization and biodegradation by the “plastisphere”. Front Microbiol 10:865. doi:10.3389/fmicb.2019.0086531073297 PMC 6497127 · doi ↗ · pubmed ↗
- 5Pande V, Pandey SC, Sati D, Bhatt P, Samant M. 2022. Microbial interventions in bioremediation of heavy metal contaminants in agroecosystem. Front Microbiol 13:824084. doi:10.3389/fmicb.2022.82408435602036 PMC 9120775 · doi ↗ · pubmed ↗
- 6Rafeeq H, Afsheen N, Rafique S, Arshad A, Intisar M, Hussain A, Bilal M, Iqbal HMN. 2023. Genetically engineered microorganisms for environmental remediation. Chemosphere 310:136751. doi:10.1016/j.chemosphere.2022.13675136209847 · doi ↗ · pubmed ↗
- 7Sevilla ME, Garcia MD, Perez-Castillo Y, Armijos-Jaramillo V, Casado S, Vizuete K, Debut A, Cerda-Mejía L. 2023. Degradation of PET bottles by an engineered Ideonella sakaiensis PE Tase. Polymers (Basel) 15:1779. doi:10.3390/polym 1507177937050393 PMC 10098701 · doi ↗ · pubmed ↗
- 8Shepherd ES, De Loache WC, Pruss KM, Whitaker WR, Sonnenburg JL. 2018. An exclusive metabolic niche enables strain engraftment in the gut microbiota. Nature 557:434–438. doi:10.1038/s 41586-018-0092-429743671 PMC 6126907 · doi ↗ · pubmed ↗