Synthetic Gene Circuits Enable Systems-Level Biosensor Trigger Discovery at the Host-Microbe Interface

The gut is a largely obscure and inaccessible environment. The use of live, engineered probiotics to detect and respond to disease signals in vivo represents a new frontier in the management of gut diseases. Engineered probiotics have also shown promise as a novel mechanism for drug delivery. However, the design and construction of effective strains that respond to the in vivo environment is hindered by our limited understanding of bacterial behavior in the gut. Our work expands the pool of environmentally responsive synthetic circuits for the healthy and diseased gut, providing insight into host-microbe interactions and enabling future development of increasingly complex biosensors. This method also provides a framework for rapid prototyping of engineered systems and for application across bacterial strains and disease models, representing a practical step toward the construction of clinically useful synthetic tools.

are a promising chassis for the deployment of synthetic biological circuits. Such circuits can enable the construction of diagnostic strains that record biological and environmental factors in the intestine, which can be analyzed after transit through the gut to provide a noninvasive survey of an obscure environment. Also compelling is the potential for in vivo production of biological therapies. A range of studies have engineered bacteria to serve as potential treatments for inflammation, metabolic diseases, cancer, and infection by producing therapeutic proteins in vivo or stimulating immune responses (1,2).
The majority of studies developing engineered gut bacteria have focused on the expression of therapeutic proteins, either constitutively or by induction with exogenously supplied compounds. However, specific environment-and disease-responsive functions-which could minimize both the metabolic burden of engineered systems on the bacteria and off-target effects on the patient-offer exciting prospects for clinical applications. To this end, recent in vivo approaches have developed sensors responding to inflammation (3,4), intestinal bleeding (5), and pathogen quorum-sensing systems (6,7). However, the construction of disease-responsive circuits in bacteria has been hindered by the limited number of characterized bacterial systems that can be reliably used as sensors.
Mining the genomes of native gut bacteria is a promising approach for discovering new sensors that respond under conditions of interest, such as in the healthy or diseased gut. To date, these efforts have largely relied on transcriptome sequencing and proteomics of fecal samples. However, to obtain an instantaneous snapshot of bacterial behavior inside the gut using these techniques, invasive sampling is required (i.e., colonoscopy and biopsy). Furthermore, transient or low-abundance signals may not be detected, and any responsive genetic elements identified with these techniques may not function predictably when used in synthetic circuits. Approaches such as in vivo expression technology (IVET) and recombinase-based IVET (RIVET) have been used to track in vivo-expressed genes noninvasively, primarily focusing on pathogenicity mechanisms in infectious strains (8)(9)(10)(11). However, these approaches detect only constitutive expression (for IVET) and may have high false-positive rates. Nevertheless, these technologies show the potential for systems-level approaches to interrogate the behavior of the microbiota.
We have previously developed an approach for noninvasive measurement of bacterial responses in the gut, based on a robust synthetic memory circuit, which records environmental stimuli via a transcriptional trigger (3,12). When activated, the trigger turns on a memory switch, which can retain the memory-on state for over a week in the gut (12). After the bacteria pass through the host, their memory state can be determined via reporter gene expression, enabling noninvasive readout of transient signals within the gut. The circuit can maintain functional and genetic stability during 6 months' colonization of the mouse gut, demonstrating its suitability for longitudinal studies and its potential to support the development of stable, engineered biosensor strains for in vivo deployment (3).
Here, we adapt this memory circuit for parallel, high-throughput screening of hundreds of potential triggers. We apply this method to identify new biosensor triggers responding specifically to the in vivo gut environment. Through comparison between healthy mice and those suffering from inflammation, we also identify triggers that respond differentially during disease. Together, these results provide a platform for in vivo noninvasive biosensor trigger discovery and longitudinal testing.
(This article was submitted to an online preprint archive [13].)

RESULTS
Bacterial memory as a biosensor trigger screening tool. To enable screening of new potential biosensor triggers in parallel, we modified our previously developed Escherichia coli memory circuit, which is based on the phage lysis-lysogeny switch (see Fig. S1A in the supplemental material) (12). This modified circuit is referred to as the high-throughput memory system (HTMS) (Fig. 1a). Both the original memory circuit and the HTMS consist of a trigger-based on a transcriptional promoter activated in the presence of a certain stimulus-and a bistable memory switch. The memory-on and memory-off states of the switch correspond to the mutually repressive proteins Cro and CI, which are under the control of the P R and P RM promoters, respectively. In addition, a ␤-galactosidase (LacZ) reporter is produced in the memory-on state.
One key modification in the HTMS is the triggering of memory using a dominantnegative mutant of the cI gene (cI DN ), instead of the cro gene used in the original circuit's trigger. In the original circuit, the continued presence of a stimulus can lead to the production of high levels of Cro from the trigger, which can repress both the P R and P RM promoters and prevent switching to the memory-on state while the stimulus is present. In contrast, the trigger used in the HTMS produces CI DN monomers upon induction, which have an N55K mutation in their DNA-binding region (14). These CI DN monomers dimerize with wild-type (WT) CI monomers expressed in the memory-off state to create heterodimers that are deficient in DNA binding. This leads to derepression of P R and transition to the memory-on state, even during continued induction of the trigger. As with the CI used in the memory switch, CI DN carries a mutation to prevent RecA-mediated cleavage (ind-) (15).
Use of CI DN in the trigger ensures that there is no delay of switching to the memory-on state in the case of high, or constant, activation of the trigger promoter. To test this, a P tet trigger driving cI DN or cro was integrated into E. coli K-12 MG1655 and NGF-1 strains containing a memory switch. E. coli NGF-1 is a strain originally isolated from the murine gut, which has proven to be an efficient and persistent colonizer and a reliable platform for the deployment of engineered circuits (3,12,16). When grown in the presence of a high concentration (100 ng/ml) of anhydrotetracycline (aTc) inducer, cI DN   Comparison of memory switch induction with cro and cI DN triggers illustrates differences in induction dynamics. Control and memory strains with P tet triggers (PAS132, PAS133, PAS807, and PAS808) were grown in liquid media and then spotted on indicator plates, each with or without aTc induction (100 ng/ml). (c) Selection of memory-on HTMS with spectinomycin. Memory-off, memory-on, and 50-50 mixed cultures of PAS810 were plated on indicator plates with or without spectinomycin (inset photo). Graph shows comparison of CFU counts between ϩspectinomycin and -spectinomycin plates. Error bars represent the SE of eight biological replicates (for 0 and 100%) and five biological replicates (for 50%).
Synthetic Gene Circuits Enable Gut Biosensor Discovery triggered strains switched only after a subsequent period of growth in the absence of aTc (Fig. 1b). The original memory circuit expresses a lacZ reporter gene for screening on indicator plates (12). To analyze pooled libraries containing many strains with varied trigger promoters, the HTMS also expresses a spectinomycin-selectable resistance gene (aadA) in the memory-on state.
This antibiotic-selectable memory maintains response characteristics similar to the original memory switch. To test this, a P tet trigger driving cI DN was integrated into strains containing lacZ (original) or aadAϩlacZ (HTMS) memory switches, creating PAS809 and PAS810, respectively (see Table 1 for strain list). Strains were induced by aTc (0 to 100 ng/ml) and the response quantified by plating cultures on indicator plates containing X-Gal (5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside), which turns blue in the presence of LacZ, indicating a memory-on state (Fig. S1B). Both strains responded similarly to aTc (original memory 50% effective concentration [EC 50 ], 4.1 to 4.6 ng/ml, 95% confidence interval [CI]; HTMS EC 50 , 4.0 to 4.3 ng/ml, 95% CI), confirming the circuit's modularity to additional reporters in the memory-on state.
The HTMS allows faithful selection of memory-on colonies with spectinomycin treatment. Plating of fully memory-off, fully memory-on, and 50-50 mixed cultures of PAS810 on indicator plates with and without spectinomycin further demonstrated that all spectinomycin-selected colonies were also LacZ positive (Fig. 1c). Spectinomycin did not yield false-positive results by inducing memory switching (fully memory-off, 0% Ϯ 0% standard error [SE], n ϭ 8), nor excessive false-negative results through inhibition of memory-on bacterial growth (fully memory-on, 93.0% Ϯ 4.2% SE, n ϭ 8; 50-50 mix, 49.7% Ϯ 1.5% SE, n ϭ 5) (Fig. 1c). Together, these results demonstrate the ability of the HTMS to measure biosensor trigger response and allow selection for downstream pooled analyses.
Biosensor trigger library construction. To build biosensor trigger libraries for genomic integration, we adapted a Tn7 transposon genome insertion plasmid (17) for rapid Golden Gate assembly (18) of bacterial promoters upstream of the cI DN trigger and insertion into the genome of memory bacteria ( Fig. 2a and S2). The modularity of this cloning strategy allows for adjustment of trigger sensitivity through incorporation of ribosomal binding site (RBS) variants, which vary the translation rate of mRNA transcripts (Fig. 2a). To test this concept, triggers consisting of a P tet promoter combined with nine synthetic RBS sequences-previously demonstrated to vary widely in their translation strength (19) (Fig. 2a)-were constructed and inserted into the genome of HTMS bacteria, and the HTMS response to varying concentrations of aTc (0 to 100 ng/ml) was characterized (Fig. 2b). The RBS variants differed in their extent of memory induction at 0.1 to 10 ng/ml aTc (EC50 ranging from 0.5 to 4.1 ng/ml for responsive strains), illustrating our ability to tune trigger sensitivity. We explored two approaches for generating biosensor trigger libraries: (i) a comprehensive collection of trigger promoters that would enable screening a wide range of a bacterium's transcriptional responses (MG1655 library) and (ii) a curated collection of promoters with sensitivity variants aimed at detecting inflammation (Nissle 1917 library). Both libraries were assembled into an HTMS-containing E. coli NGF-1 strain (PAS811). The comprehensive library was sourced from a previously published collection of 1,600 unique promoters selected from across the genome of E. coli K-12 MG1655 (20). Promoters and their wild-type RBSs were amplified by PCR from this collection, assembled into our transposon plasmid, and integrated as triggers into the genome of PAS811, creating a genome-wide library. However, because our method specifically detects off-to-on sensor transitions, the genome-wide library was further subsampled by pooling 500 colonies that were LacZ negative under routine in vitro culture to produce a comprehensive, "off-in vitro" library. This subsampling maximized sequencing depth for HTMS trigger candidates. Sequencing confirmed the presence of 155 unique strains in this final comprehensive library.
Our second library was constructed with a subset of promoters sourced from the human probiotic E. coli Nissle 1917, which are involved in anaerobic respiration of sulfur or nitrogen oxides or nitrate, produced by the gut epithelium during inflammation (21,22). For each promoter, triggers with the wild-type RBS, as well as with five different synthetic RBSs (MCD5, MCD10, MCD15, MCD17, and MCD23) (19), were included to tune sensitivity. Throughout this study, strains from this library are referred to by an abbreviation consisting of the first gene of the operon from which their trigger is derived and the number of the synthetic RBS used. For instance, "ynfE15" denotes the trigger consisting of the ynfEFGH promoter with MCD15. Sequencing confirmed that the assembled library contained 61 unique strains of 66 total designed constructs.
Pipeline for biosensor trigger library screening. To screen for biosensor trigger response, HTMS libraries are exposed to a condition of interest (Fig. 3a), and put through a processing, sequencing and analysis pipeline (Fig. 3b). After exposure, HTMS bacteria are recovered and cultured. The initial culture is split into two and backdiluted, and one of the two new cultures is subjected to spectinomycin selection. After selection, the trigger regions of both cultures are sequenced and analyzed to produce an odds ratio for each trigger promoter in the library, corresponding to that trigger's memory state. To calculate odds ratios, the results are normalized to a positive normalization strain (PAS812) (Fig. 3b). PAS812 is an HTMS strain containing a trigger promoter from the E. coli fabR gene, which was observed to be consistently memory-on under in vitro culture conditions.  Library screening faithfully reports biosensor trigger response. To test our library screening pipeline, the Nissle 1917 library was cultured aerobically in liquid media and analyzed to obtain odds ratios as described above. Concurrently, individual strains from this library were grown on indicator plates to assess each strain's in vitro memory state directly. Both tests showed strong agreement, with strains that were LacZ  moaA17  cydA10  moaA23  cydA23  cydA17  moaA10  moaA15  moaA5  cydA15  cydA15  fdnGWT  cydAWT  narZ23  fdnG10  narK10  yeaRWT  narK15  fdnG5  ynfE15  hycAWT  ynfE17  narK5  yedY17  yedY23  moaAWT  fdnG15  hycA5  hycA10  hycA15  hycA17  hycA23  narKWT  narK23  yeaR5  yeaR10  yeaR15  yeaR17  yeaR23  yedY10  yedY15  ynfEWT  ynfE5  positive also displaying higher odds ratios (Fig. 3c). Receiver operating characteristic analysis confirmed efficient distinction between memory-on and memory-off states, with an odds ratio of approximately 0.02 delimiting the boundary (Fig. 3d). This confirmed our sequencing method as a reliable indicator of biosensor memory state.
Differential biosensor trigger response in the healthy mouse gut. To screen for biosensor trigger response to growth within the murine gut, the MG1655 library was administered to specific-pathogen-free (SPF) mice by oral gavage (ϳ10 7 bacteria/ mouse), and fecal samples were collected over 1 (n ϭ 2) or 7 (n ϭ 3) days. High library diversity was maintained in both experiments (92 and 82% of strains identified in gavage samples present at the experiment endpoint, respectively; Data Sets S1 and S2). Analysis of HTMS strains recovered from gavage suspension and fecal samples identified 23 unique strains that responded specifically to growth within the gut (gavage: odds ratio Ͻ 0.02; fecal samples: Ն1 time point with an odds ratio Ն 0.02, and P Ͻ 0.05) ( Fig. 4a and b; Data Sets S1 and S2). Five strains (containing ydiL, ydjL, gatY, gcvA, and ubiG triggers) were detected in the memory-on state in at least four of five mice. The two most consistent responders (ydjL and ydiL triggers) were selected for follow-up testing.
To validate the response of the ydiL and ydjL triggers during gut transit, memory bacteria containing these triggers (ydiL, PAS813; ydjL, PAS814) or a promoterless cI DN gene (negative control, PAS815) were administered to SPF mice as monocultures. Fecal samples were collected and analyzed over the subsequent 5 days. Culture on indicator plates demonstrated an absence of memory activation in all three strains prior to gavage. However, when recovered from fecal samples, PAS813 and PAS814 colonies were consistently memory-on, confirming activation during gut transit (at day 2, PAS813: 51% Ϯ 8% SE; PAS814: 30% Ϯ 4% SE; negative control: 0% Ϯ 0% SE; n ϭ 3) (Fig. 4c).
The Nissle 1917 library was also screened to discover promoters responding in the healthy mouse gut. Testing of the Nissle 1917 library over 5 days following gavage (ϳ10 7 bacteria/mouse) identified 11 strains that specifically responded to in vivo growth (Data Set S3). Ten of these, derived from three unique promoters (ynfEFGH, torCAD, and yeaR-yoaG operons) registered a memory-on state in the majority of time points and all mice tested (n ϭ 4) (Fig. 4d and e). Promoter response was similar during parallel analysis in the inflamed mouse gut (n ϭ 4; see below and Fig. 5 for experimental details), further validating these results ( Fig. S3 and Data Set S3).
Together, these results demonstrate the ability for HTMS analysis to rapidly identify biosensor triggers in vivo and the power of varying trigger sensitivity to tune the strength of biosensor strain response.
Identification of disease-specific biosensor triggers. To look for sensors responding differentially to disease, we compared the response of the Nissle 1917 library in healthy mice (n ϭ 4; as previously displayed in Fig. 4d and e   Synthetic Gene Circuits Enable Gut Biosensor Discovery murine intestinal inflammation model ( Fig. 5a and b, Fig. S4, and Data Set S3). After library gavage, SPF mice were provided water containing 4% (wt/vol) dextran sulfate sodium (DSS) ad libitum for 5 days, and HTMS analysis was performed on fecal samples throughout. Weight loss (Fig. S4A), colon length reduction at endpoint (Fig. S4B), and increased E. coli CFU counts (Fig. S4C) were all consistent with increasing inflammation throughout the experiment. Six strains registered memory-on at more time points in the DSS-treated group than in the control group ( Fig. 5a and b). In particular, the ynfE17 trigger strain (PAS819) responded specifically in DSS-treated mice (control: no response; DSS-treated: 93% of time points with an odds ratio Ն 0.02 and P Ͻ 0.05) (Fig. 5a and  b; Data Set S3).
To validate ynfE17 (PAS819) response to inflammation, the strain was administered to SPF mice as a monoculture, after which a subset of the mice was provided water containing 4% DSS. Fecal samples were cultured for memory bacteria on indicator plates for 7 days after gavage. As above, body weights (Fig. S4D), post-dissection colon lengths (Fig. S4E) and CFU counts of HTMS bacteria (Fig. S4F) reflected increased inflammation in DSS-treated mice. Confirming the screen results, ynfE17 showed increased response in DSS-treated mice compared to untreated controls, with the greatest difference between groups at days 6 and 7 (at day 6, ϩDSS: 24% Ϯ 9% SE, n ϭ 4; control: 5% Ϯ 2% SE, n ϭ 8) (Fig. 5c). The strong response of PAS819 in the absence of DSS in one of the control group mice indicates that in vivo conditions other than DSS treatment can activate the ynfE17 trigger. In vitro anaerobic growth both in rich media and in cecal contents did not induce ynfE17 (rich media: 0 Ϯ 0% SE, n ϭ 7; cecal contents: 0 Ϯ 0% SE, n ϭ 3), in contrast to ynfE15 (Fig. 4g), suggesting a lower nitrate threshold for ynfE17 activation and that individual bacteria experience low nitrate conditions within the inflamed mouse gut.

DISCUSSION
Here, we have expanded the use of a robust genetic memory circuit to assess the in vivo responses of multiple bacterial sensors in parallel. The original memory circuit (12) was altered to allow off-to-on transitions in the presence of constant induction and to enable selection of memory-on strains from pooled libraries using spectinomycin. We developed a screening, sequencing, and analysis pipeline to efficiently identify in vivo-responding trigger-RBS combinations. Tests conducted with both comprehensive and curated libraries containing hundreds of sensors demonstrated that our method is an effective, noninvasive way to identify new biosensor triggers responding in the gut. We identified and validated biosensor triggers responding to growth in the healthy mouse gut and preferentially in inflamed conditions. Together, these results demonstrate the power of tuning trigger sensitivity to physiological responses and for the HTMS to assess unique features of the mammalian gut environment in vivo.
One advantage of our method is its ability to discover sensors that could not be rationally designed based on existing knowledge, presenting an opportunity to apply the rapidly increasing but largely uncharacterized genetic diversity identified through microbiome sequencing. For instance, the two validated healthy-gut sensors from our E. coli MG1655 library (PAS813 and PAS814) are derived from operons with largely uncharacterized function and regulation. PAS814 is triggered by the promoter of the ydjLKJIHG operon, which putatively includes a kinase, a transporter protein, two dehydrogenases, an aldolase and an aldo-keto reductase. Only the activity of the aldo-keto reductase, YdjG, has been confirmed through reduction of methylglyoxal (24,25). Interestingly, a previous analysis of E. coli protein expression in germfree mice showed that YdjG was expressed 3.5-fold higher in the cecum than in vitro (26). Another gene which has been studied in this operon, ydjK, may play a role in osmotolerance, showing a 50% increased growth rate in high-salt media when overexpressed in E. coli (27). It is not known whether methylglyoxal or osmotic stress can directly trigger transcription of the ydjLKJIHG operon. However, methylglyoxal occurs in many foods and is also produced by intestinal bacteria (28); it can also inhibit bacterial growth, suggesting a possible motivation for expression of ydjLKJIHG in the gut.
Promoters derived from three unique Nissle 1917 operons (ynfEFGH, torCAD, and yeaR-yoaG) showed memory response in the healthy mouse gut (Fig. 4d and e). The ynfEFGH operon encodes a DMSO reductase which has also been shown to reduce selenate (29,30). It is activated by FNR under anaerobic conditions and repressed by phosphorylated NarL in the presence of nitrate (23), which was further confirmed by our in vitro tests with PAS816 (Fig. 4g).
Tuning of trigger sensitivity (e.g., by RBS modulation) is important for generating responses to physiological conditions of interest and for successful application in synthetic engineered circuits. As we observed, RBS tuning can be used to increase the response of the ynfE promoter to as high as 100% in healthy mice (PAS816; Fig. 4f), and to adjust the response to distinguish the inflamed gut state (PAS819; Fig. 5c). Importantly, the sensors we identify can be used directly in downstream applications with the memory circuit. This provides an engineering advantage over any responsive genetic elements identified through analysis in their native context, for which incorporation into synthetic circuits would routinely require additional optimization.
Inflammation leads to an increase in nitric oxide produced by the host, which generates nitrate in the intestine (31). However, because the ynfE promoter is activated by a decrease in nitrate, our results suggest that DSS-induced inflammation may lead to lower levels of free nitrate available to E. coli NGF-1, possibly due to increased local competition for nitrate via respiration by NGF-1 and other Enterobacteriaceae. This idea is supported by our observation of increasingly higher NGF-1 bacterial loads in fecal samples of DSS-treated mice ( Fig. S4C and F), suggesting a bloom of E. coli-and potentially other Enterobacteriaceae capable of nitrate respiration-correlated with increasing duration of DSS treatment. This is consistent with previous descriptions of E. coli experiencing a growth advantage due to anaerobic respiration of host-derived nitrate (31). Thus, we hypothesize that PAS819 responds in DSS-treated mice specifically through sensing inflammation-induced changes in its own microenvironment.
The HTMS enables both the recording of transient signals and the amplification of low-abundance signals through antibiotic selection. These features serve as a useful complement to other techniques, such as meta-transcriptomic or -proteomic studies which capture an instantaneous snapshot of total RNA or protein content. Screening of broad libraries increases the chances of discovery of new, uncharacterized sensors. Combining comprehensive libraries with RBS tuning would further increase the chances of identifying triggers specific to conditions of interest. Our use of the E. coli NGF-1 strain as a chassis allows reliable colonization of the mouse gut without requiring antibiotic maintenance, leading to retention of high bacterial loads and high library complexity in fecal samples even after long periods in the gut.
For clinical applications, an expanded arsenal of characterized sensors presents opportunities to construct more complex disease-responsive circuits. For instance, the combination of multiple redundant sensors would increase response accuracy and specificity under variable in vivo conditions, while complementary sensors may allow "fingerprinting" of different disease states. An exciting possibility is the use of more complex logic and signal processing within a single engineered strain, which may sense multiple inputs and produce anti-inflammatory, antimicrobial, or other therapeutic proteins only when a precise set of conditions is satisfied (2). Sensors responding differentially based on localization within the intestine may create opportunities for more targeted drug delivery or for the construction of new safety and containment mechanisms-another important consideration in the deployment of engineered organisms.
The potential to engineer synthetic circuits into commensal gut bacteria is a promising new approach to the management of intestinal disease. Synthetic biology is just beginning to tap into the evolutionary breadth of capabilities found in natural systems, and our method represents a practical means for expanding the toolkit of useful sensors for in vivo application.

MATERIALS AND METHODS
Media and culture conditions. Unless otherwise mentioned, bacterial cultures were grown at 37°C in LB broth or agar (10 g/liter NaCl, 5 g/liter yeast extract, 10 g/liter tryptone). Mixed liquid cultures (i.e., libraries) were grown in LBPS, which contains Peptone Special (Sigma) instead of tryptone. To quantify memory response on indicator plates, agar was supplemented with 60 g/ml X-Gal.
High-throughput memory system construction. The spectinomycin resistance gene, aadA, was added downstream of the P L promoter in the original memory switch (12) by overlap extension PCR and genomically integrated by Red recombineering (32) into E. coli TB10 (33) between mhpR and lacZ, driving endogenous lacZ as a memory-on reporter. From TB10, transfer into streptomycin-resistant E. coli NGF-1 was done by P1vir transduction.
Biosensor strain and library construction. All triggers were cloned into pDR07 (Fig. S2), a Tn7 transposon insertion plasmid derived from pGRG36 (17). BsaI sites directly upstream of cI DN allow modular insertion of promoter-RBS sequences via Golden Gate assembly (18) (Fig. 2a). Assembled trigger plasmids were electroporated into PAS811. After recovery (90 min, 30°C in SOC medium) transformants were selected overnight (30°C in LB-ampicillin [100 g/ml]). Cultures were then back-diluted 1:100 into LB-chloramphenicol (25 g/ml) plus 0.1% arabinose to induce transposase genes. After Ͼ6 h at 30°C, temperature-sensitive pDR07 plasmids were cured from integrants by 1:100 back-dilution into LBchloramphenicol and Ͼ6 h growth at 42°C. This cure step was repeated a second time. Plasmid loss was confirmed by restreaking on LB-ampicillin agar.
For individual strains, post-cure cultures were plated on LB-chloramphenicol agar and attTn7 integrations confirmed by PCR and Sanger sequencing. For pooled libraries, library composition was confirmed by Illumina MiSeq sequencing of pooled PCRs of trigger regions.
Assessment of memory state by LacZ assay. Cultures or fecal supernatants containing memory bacteria were plated on agar plates containing streptomycin (200 g/ml), chloramphenicol (34 g/ml), and X-Gal (60 g/ml). The percentages of memory-on colonies were assessed by counting blue (on) and white (off) colonies.
In vitro induction. Overnight liquid cultures were back-diluted 1:100 into fresh media containing inducer, followed by 4 h growth and plating on X-Gal agar. For induction in cecal contents, contents of ceca from three female SPF C57BL/6J mice and suspended at 10% (wt/vol) in phosphate-buffered saline (PBS). Suspensions were vortexed 90 s and centrifuged for 3 min at 4,300 relative centrifugal force (rcf). The supernatant was recovered, supplemented with 200 g/ml streptomycin and used for growth of HTMS bacteria.
In vivo induction of strains and libraries. The Harvard Medical School Animal Care and Use Committee approved all animal protocols. Experiments were conducted in female 7-to 14-week-old BALB/c mice (Charles River; MG1655 library) or C57BL/6J mice (Jackson; Nissle 1917 library). Before experiments, all mice were confirmed to be free of native streptomycin-and chloramphenicol-resistant flora. Food and water were removed ϳ4 h before each gavage; water was replaced immediately, and food was replaced Ͻ2 h after gavage.
One day prior to bacterial gavage, mice were provided streptomycin (20 mg in PBS) by oral gavage. The next day, overnight cultures of memory strains or libraries were washed once and then diluted 10-fold in PBS and administered by gavage (100 l; ϳ10 7 bacteria/mouse).
Gavage suspension and fecal samples were plated to track bacterial load and, for individual strains, to assess memory state. Libraries were processed according to the postexposure processing protocol below. To plate fecal bacteria, samples were suspended at 100 mg/ml in PBS, vortexed 5 min, and centrifuged 20 min at 4 rcf to obtain fecal supernatant.
For inflammation experiments, water containing 4% DSS (36,000 to 50,000 molecular weight; MP Biomedicals, catalog no. 160110) was substituted 2 h after bacteria administration. Mice were dissected at the end of the experiment to measure colon length.
Postexposure library processing. Fecal supernatant or in vitro culture was diluted 1:100 into LBPS-chloramphenicol (25 g/ml) to achieve ϳ10 6 CFU/ml. Concurrently, an overnight culture of the positive normalization strain, PAS812 was back-diluted 1:100 into LBPS-chloramphenicol. Cultures were grown 4 h or until an optical density at 600 nm (OD 600 ) of ϳ1 was achieved. The PAS812 OD 600 was adjusted to match the library culture then diluted 1:1,000 into the library culture. The resulting mix was back-diluted 1:1,000 into 50 ml of LBPS-chloramphenicol and immediately split into two 25-ml volumes. Spectinomycin (50 g/ml) was added to one culture, and both were grown overnight before centrifugation to collect bacterial pellets, which were stored at -80°C.
Library sequencing and odds ratio calculation. Genomic DNA was extracted from frozen cell pellets using a Qiagen DNeasy Blood & Tissue kit. Using genomic DNA as a template, trigger regions from HTMS libraries were amplified by PCR and sheared with a Covaris M220 ultrasonicator to 200-to 600-bp fragments. Sheared products were prepared using a New England Biolabs NEBNext Ultra II Prep kit and sequenced by Illumina MiSeq.
Raw reads were trimmed using Trimmomatic 0.36 (34) and aligned to a reference file (Data Sets S4 and S5 for MG1655 and Nissle 1917 libraries, respectively) using BWA mem 0.7.8 (35). The number of uniquely mapped reads for each trigger was counted.
The odds ratio is expressed as (T x-spect /PNS spect )/(T x /PNS), where T x and T x-spect are the numbers of mapped reads for a particular trigger in the untreated and spectinomycin-treated cultures, respectively, and PNS and PNS spect are the numbers of mapped reads for the positive normalization strain (PAS812) in the untreated and spectinomycin-treated cultures, respectively. Triggers with Ͻ5 reads in the gavage suspension were discarded, unless they registered Ͼ20 reads at any subsequent time point. For each pair of untreated and spectinomycin-treated cultures (from a single fecal sample), odds ratios were calculated for each trigger with Ն5 reads in the untreated culture. The statistical significance was assessed with a one-tailed Fisher exact test (H 0 , odds ratio ϭ 0.02; H a , odds ratio Ͼ 0.02). The odds ratio calculation compares each trigger only with itself (between spectinomycin-treated and untreated cultures), normalizing any sequencing length bias between triggers. It also normalizes to the positive normalization strain (PAS812) in each sample, negating read depth disparities between samples.
Data availability. Raw sequence data from library screening experiments have been deposited at the NCBI Sequence Read Archive as BioProject ID PRJNA542391. Other data and resources are available from the corresponding authors upon request.