Streptomycin pretreatment is often used to sensitize mice to S. Typhimurium orogastric infection. This antibiotic reduces the microbiota and is thought to heighten the pathogen burden and consistency of intestinal colonization (26). We orogastrically administered 10⁸ colony-forming units (CFU) of a barcoded S. Typhimurium library into 8-week-old untreated and streptomycin-treated C57BL/6J mice to investigate how streptomycin pretreatment modifies S. Typhimurium population dynamics. We monitored common markers of disease and colitis during infection (S1 Fig). Consistent with previous reports (26–28), the streptomycin-treated animals lost weight more rapidly than animals that were not streptomycin-treated and were sacrificed on day 4 post-inoculation (dpi), after they had lost >20% of their body weight (Figure 1A). Fecal shedding of S. Typhimurium gradually increased in untreated mice. In contrast, animals pretreated with streptomycin rapidly and uniformly began shedding S. Typhimurium, with fecal burdens higher than the inoculum by 24 hours post-inoculation and sustained throughout the experiment (Figure 1B).

We measured S. Typhimurium burden in 12 organs/fluids at 4 (+SM) or 5 (untreated/-SM) days post-inoculation, while they were at the peak of disease. Organs/tissues were classified into three groups based on the pathogen burden and differences between the streptomycin-treated and untreated animals (Figure 1Ci-iii). The gastrointestinal (GI) tract organs, including the proximal and distal small intestine (SI), cecum, colon, and Peyer’s patches (PP), are presumed to be the initial site of infection after orogastric inoculation and generally had 100-10,000-fold higher burden in streptomycin-treated animals (Figure 1Ci); also, there was much more variation in pathogen burdens within the GI tracts of untreated animals. In contrast, in the extraintestinal immune-rich organs, such as the mesenteric lymph node (MLN), spleen, and liver, the S. Typhimurium burdens and variation in burden were similar in the two groups (Figure 1Cii). In the remaining organs and fluids, including the bile, pancreas, perigonadal adipose tissue (‘adipose’), and peritoneal wash (‘wash’), bacterial burdens were generally lower than in other tissues, except for bile, and the burdens in treated and untreated animals were similar. Together, these observations reveal that although streptomycin treatment markedly elevates the S. Typhimurium burden in the GI tract, it does not appear to significantly alter the pathogen burden at sites beyond the intestine.

Counting the number of S. Typhimurium cells within an organ does not directly measure the bottleneck the bacterial population encountered in establishing its niche because bacterial replication obscures the effects of bottlenecks (i.e., the observed population is the net outcome of a bottleneck followed by replication and migration from other sites). To directly measure the bottleneck impeding intestinal colonization, the experiments described above were carried out with genomically barcoded S. Typhimurium. We introduced neutral tags into our S. Typhimurium population by integrating short ∼25 base pair sequences (barcodes) into the genome on a Tn7 transposon, resulting in a library of ∼55,000 strains that are isogenic except for their barcodes (16, 17). These barcodes can be detected by amplicon sequencing and changes in the abundance and frequencies of barcodes in the library allow us to determine the number of unique cells from the inoculum that gave rise to the observed population, referred to as the “founding population” or “founders”. The number of founders is estimated using the STAMPR analytical pipeline by a metric called Ns, which employs a multinomial resampling strategy to determine the sampling depth required to observe a specific number of barcodes given the distribution of barcodes in the inoculum (16). The size of the founding population in an organ and its barcode diversity reflects the bottleneck experienced by the inoculum, with tighter bottlenecks resulting in fewer unique barcodes than wider bottlenecks (Figure 2A). Before performing animal experiments, we determined the resolution of STAMPR analysis with our library by creating experimental bottlenecks in a culture of the S. Typhimurium library by plating serial dilutions and found that our calculations accurately predicted the number of colony- forming units (which can be considered the true size of the founding population) up to ∼700,000 CFU (Figure S2).

There was considerable variation in S. Typhimurium founding population (Ns) sizes in untreated animals (Figure 2B), reflecting inter-animal variation in the bottleneck to intestinal colonization, which likely accounts for the variable timing and quantity of fecal shedding observed in this model (Figure 1B). In most parts of the intestine of untreated mice, there were only ∼10² unique founders (Figure 2B), suggesting that at this dose (10⁸ CFU) there is a severe intestinal colonization bottleneck that reduces the inoculum population by about ∼10⁶-fold. Streptomycin treatment widened this bottleneck considerably, increasing the founding population by 10-100- fold in all regions of the intestine (Figure 2B). This increased founding population following streptomycin treatment demonstrates that streptomycin clearance of the microbiota and its downstream consequences, such as inflammation (26), removes some of the barriers to S. Typhimurium establishing a replicative niche in the intestine.

Streptomycin treatment increased both the total S. Typhimurium burden (Figure 1Ci) and the number of founders (Figure 2B) in the intestine. Replication of founders generally accounts for the majority of the burden in most tissues. The average net expansion of the population can be measured by dividing bacterial burden by the founding population (CFU/Ns). This calculation revealed that there is generally more net expansion in streptomycin-treated versus untreated animals (Figure 2C), i.e., the increase in CFU following streptomycin treatment is caused by both increased number of founders and increased net replication. These observations suggest that streptomycin treatment not only widens the bottleneck impeding S. Typhimurium colonization (increased founders) but also creates a more permissive niche for pathogen population expansion.

The STAMPR analytical pipeline enables comparison of barcoded bacterial populations sampled from different sites within the same animal. The similarity in the frequency of barcodes between organs is quantified with a metric of genetic distance (Figure 2A) (17). Organs containing barcodes with indistinguishable proportions have a genetic distance (GD) of zero, while those that do not share barcodes have a genetic distance of nearly one (see methods). We compared the similarity of barcode frequencies in samples from different intestinal regions. Except for the cecum and colon, in untreated animals the S. Typhimurium populations in different regions of the intestine were dissimilar (Avg. GD ranged from 0.369 to 0.729, 2D left), suggesting that S.

Typhimurium replicates in a compartmentalized manner with little sharing between populations throughout the intestinal tract (i.e., barcodes found in one region are not in other regions, Figure S3). This was surprising as the luminal contents, an environment presumably conducive to bacterial movement, were not removed from these samples. Further supporting this compartmentalized population hypothesis is the lack of barcode overlap between individual Peyer’s patches taken from the proximal, medial, and distal small intestine of an individual mouse (Figure S3). Streptomycin treatment increased the similarity in S. Typhimurium populations derived from different parts of the intestine (Avg. GD ranged from 0.157 to 0.407, Figure 2D right), suggesting that reduction of the microbiota through streptomycin pretreatment increases the mixing of the populations between the intestinal compartments.

S. Typhimurium is primarily transmitted through the feces. We attempted to identify the origin from within the intestinal tract of fecal S. Typhimurium through genetic distance comparisons of fecal samples to different regions of the intestine. Unexpectedly, in approximately half of the untreated animals, the fecal barcodes were highly dissimilar from the barcodes present in all regions of the intestine (GD >0.75; Figure 2E), indicating that the majority of the S. Typhimurium population at intestinal sites are not being shed into the feces. In contrast, mice treated with streptomycin had similar populations in intestinal and fecal samples (Avg. GD = 0.299 ± 0.06; Figure 2D), reflecting an increase in sharing between these populations. Thus, STAMPR-based analysis of the high-density barcoded S. Typhimurium library enabled quantitative assessments of intestinal bottlenecks and revealed that streptomycin treatment markedly alters the compartmentalization of S. Typhimurium replication in the intestine and the source of the pathogen shed in the feces.

In mice, S. Typhimurium routinely spreads out of the intestine to colonize extraintestinal organs. In the intestine, streptomycin treatment increased S. Typhimurium burdens by 100-10,000-fold relative to untreated animals. In contrast, streptomycin treatment did not markedly increase the burden of S. Typhimurium in systemic organs (Figure 1Cii-iii). Nevertheless, barcode analysis revealed that streptomycin treatment increased the size of the founding population in extraintestinal organs by ∼10-fold (Figure 3A), suggesting that the net bottleneck to dissemination is relaxed by streptomycin treatment, likely contributing to the more rapid disease progression observed in streptomycin-treated animals (Figure 1A). However, since the founding population is calculated based on the abundance of barcodes in an organ sample at sacrifice, it is difficult to discern if streptomycin treatment modifies the net bottleneck to dissemination by relaxing the individual bottlenecks to reaching the intestine, colonizing the intestine, disseminating from the intestine, and/or establishing a niche in extraintestinal organs.

To gain further insight into when S. Typhimurium disseminates to extraintestinal sites, we compared the similarity of barcodes found in S. Typhimurium isolated from the liver to those in the intestine. The similarity between the initial site of infection (in our case the intestine) and secondary sites differentiates whether the secondary population arose soon after inoculation, before the population experienced the profound intestinal bottleneck (‘early’ in Figure 3B), or at a later point, after the initial bottlenecks and subsequent replication (‘late’ in Figure 3B) (18).

We expect that early spread would yield dissimilar barcode frequencies between intestinal and extraintestinal sites, whereas, with late spread, there would be greater barcode similarity in the two populations. Genetic distance comparison of liver samples to other sites revealed that, regardless of streptomycin treatment, there was very little sharing of barcodes between the intestine and extraintestinal sites (Avg. GD >0.75, Figure 3C). Thus, these analyses strongly support the idea that S. Typhimurium disseminates to extraintestinal organs relatively early following inoculation before it establishes a replicative niche in the intestine.

It is possible that orogastric gavage can cause trauma to the esophagus that may result in the introduction of bacteria to the bloodstream during inoculation (29), potentially giving rise to the independent populations observed in the intestine and extraintestinal sites (‘direct’ in Fig 3B).

To exclude the possibility that we unintentionally introduced the pathogen into circulation through orogastric gavage, streptomycin-treated mice were administered water containing 10⁸ CFU of the S. Typhimurium barcoded library in their oral cavity and allowed to drink (+SM, drinking). These animals showed highly similar weight loss and fecal shedding patterns as observed in streptomycin-treated mice orogastrically gavaged with the same dose of S. Typhimurium (Figure S5). Furthermore, the bacterial burdens in the intestine and extraintestinal organs were indistinguishable regardless of the method of oral inoculation (Figure 3D).

Drinking and orogastric gavage also yielded very similar founding population sizes in all organs (Figure 3E). The similarity in number of founders suggests that the two inoculums experienced indistinguishable bottlenecks, an unlikely event if orogastric gavage led to routine bloodstream inoculation. Moreover, changing oral administration from orogastric gavage to drinking did not impact the lack of similarity between S. Typhimurium barcodes present in liver samples and other organs (Figure 3F). Together, these observations provide strong support for the idea that that orogastric gavage did not routinely lead to tissue damage that enables S. Typhimurium entry into the bloodstream. Instead, these data buttress the idea that S. Typhimurium disseminates from the intestine to extraintestinal organs early in the infection, sometime after oral inoculation and prior to significant intestinal replication.

To confirm our hypothesis that S. Typhimurium disseminates from the intestine to extraintestinal organs early in the infection, we sacrificed untreated mice 5 hours after orogastric inoculation with ∼10⁸ S. Typhimurium. Even at this early timepoint, ∼100 S. Typhimurium CFU were found in liver samples (Figure 3G), indicating that the pathogen transits out of the intestine to the liver within 5 hours. At this point, there were also ∼100 founders present in the liver samples, suggesting that little, if any, replication had occurred (∼1 CFU per founder). Notably, the number of liver founders at 5 hours was similar to the number of founders observed at 5 days (120 hours, Figure 3H), consistent with the idea that dissemination of S. Typhimurium from the intestine to the liver establishes the pathogen population that subsequently undergoes considerable replication. Furthermore, the barcodes present in liver samples were not present in any other samples. Together, these data suggest the initial rapid dissemination from the intestine seeds the majority of clones in the liver that are observed at later time points. In contrast to the liver, there were more founders present in samples from the intestine (particularly in the colon) at 5 hours versus 120 hours (Figure 3H), demonstrating that the forces restricting the S. Typhimurium population in the intestine act over a period of > 5 hours.

The net bottleneck to S. Typhimurium spreading to extraintestinal organs following oral inoculation is a combination of individual bottlenecks, including barriers the pathogen encounters within the GI tract, escaping from the intestine, traveling to extraintestinal organs, and establishing a niche at extraintestinal sites. The proportional contribution of each of these bottlenecks is unknown. It is possible that they play equal roles or that one constitutes the majority of the net bottleneck. To further probe how the size of the founding population in extraintestinal organs is impacted by the bottlenecks S. Typhimurium experiences within or while exiting the intestine, we bypassed the intestinal bottleneck by administering the barcoded S. Typhimurium library intraperitoneally (IP) and intravenously (IV) (Figure 4A), routes that have been used to model the pathogen’s interactions with the immune system in extraintestinal infections (2, 14). Lower doses of 10⁴ (IP) and 10³ (IV) CFU were used to avoid rapid death (9). In addition, infections via orogastric gavage were performed at a dose of 10⁴ CFU both with and without streptomycin pretreatment to facilitate direct comparison between inoculation routes (Figure 4A).

Without streptomycin treatment at this lower dose, orogastric gavage (Oral -SM) did not lead to extraintestinal infection. Furthermore, there was no detectible S. Typhimurium (and consequently zero founders) in all the organs sampled (Figure 4B-C), consistent with a bottleneck that completely eliminated the 10⁴ cells in the inoculum either before S. Typhimurium escaped from the intestine or after spread from the intestine but prior to establishing a niche at extraintestinal sites. Moreover, these untreated mice did not exhibit any signs of disease, including diarrhea, lethargy, or weight loss (Figure S6). Thus, the net bottleneck to extraintestinal dissemination following orogastric inoculation results in less than 1 founder after inoculation of 10⁴ CFU.

The pathogen burden and number of founders in immune-rich organs were highly similar following IV and IP inoculation (Figure 4B-C). Furthermore, the numbers of founders in these organs were similar to the size of the inoculum, suggesting that S. Typhimurium does not experience a significant bottleneck when colonizing immune-rich organs following IV and IP inoculation. In streptomycin-treated animals given 10⁴ CFU by orogastric gavage, the pathogen burden and number of founders observed in immune-rich organs were ∼10-fold lower than after IP and IV inoculation. These observations suggest that the net bottleneck that S. Typhimurium experiences while spreading from oral inoculation to extraintestinal immune-rich organs is tighter than the net bottleneck following either IV or IP inoculation. This additional ∼10-fold constriction of the bottleneck is likely contributed by the extra steps required for an oral versus IP or IV inoculum to reach systemic organs, including escaping from the intestine.

Remarkably, even though most animals had a high S. Typhimurium burden in bile extracted from the gallbladder (Figure 4D), the number of founders was generally very low. Of animals with pathogen in the bile, more than half after all inoculation routes contained only a single founder (Figure 4E). The high frequency of clonality in the bile indicates the presence of a very tight bottleneck to colonization of the gallbladder. The factors that impede S. Typhimurium colonization of the gallbladder and bile are not clear, but once a single or a few bacterial cells take hold, they can replicate to yield a very high burden, with on average ∼3x10⁶ (IP) and ∼9x10⁴ (IV) CFU. Collectively, these observations support the hypothesis that the most restrictive bottleneck to S. Typhimurium dissemination occurs within or while the pathogen exits the intestine. However, spread to other organs, such as the gallbladder, are impeded by additional bottlenecks.

After IV and IP injection of S. Typhimurium there were similar burdens (Figure 4B) and founding population sizes (Figure 4C) in immune-rich organs. However, these routes yielded divergent outcomes at other sites. In the pancreas, perigonadal adipose tissue, and peritoneal washes, S. Typhimurium burdens were 100-1,000-fold greater following IP versus IV inoculation (Figure 4D). Furthermore, at these sites, there was a trend toward more founders after IP inoculation (Figure 4E), although this trend may be explained by the 10-fold higher IP dose. Along with the increased pathogen burden, white lesions were also observed in the adipose tissue of many IP- inoculated and some IV-inoculated mice (Figure S7). We speculate that the increased S. Typhimurium burden may originate from replication in the lesions. Exclusively after IP inoculation, there was a considerably higher pathogen burden in these three extraintestinal samples than the size of the inoculum or founding population, suggesting an increase in localized replication in the peritoneal cavity, the site of inoculation, and adjacent tissue.

Barcode comparisons revealed that there was a consistently high level of sharing between the spleen and liver after IP and IV inoculation of S. Typhimurium (Avg. GD <0.25, Figure 4F), indicating the spleen and liver are within the same compartment. Previous studies suggest sharing between the liver and spleen is initially minimal but increases over time (13, 14), consistent with early dissemination followed by sharing within the compartment as the infection progresses. After IP inoculation, however, the other extraintestinal sites were dissimilar to the liver/spleen compartment and each other, suggesting they represent separate compartments (GD >0.5, Figure 4F, top). In contrast, following IV inoculation there was greater similarity (Avg. GD <0.5, Figure 4F, bottom) between the S. Typhimurium populations in the liver/spleen compartment and those from the pancreas and adipose tissue (Figure 4F, more dark blue squares in the IV versus IP). Thus, these analyses of barcoded bacteria unveil how the route of inoculation can result in divergent population dynamics.

Orogastric gavage of 10⁴ S. Typhimurium in streptomycin-treated animals resulted in high-level shedding (10¹⁰ CFU/g) by 1 day after inoculation, likely reflecting a rapid blooming of the S. Typhimurium population in the intestine. In marked contrast, S. Typhimurium shedding was not observed until 4 days after IV or IP pathogen inoculation; at this point, fecal shedding was detected in 8/16 and 11/16 animals inoculated IP or IV, respectively (Figure 5A). In addition, the S. Typhimurium burden in intestinal samples was generally higher in streptomycin-treated animals inoculated by the orogastric route versus untreated animals receiving an IV or IP inoculum (Figure 5B). The presence of S. Typhimurium in the intestine and fecal shedding after systemic inoculation indicates that there are route(s) for S. Typhimurium to spread back into the intestine from extraintestinal organs. However, the absence of shedding in some animals and the marked delay in shedding after IP and IV relative to orogastric inoculation suggests that the S. Typhimurium population encounters substantial bottleneck(s) on the route(s) from extraintestinal sites back to the intestine.

Notably, the number of founders in intestinal samples from animals inoculated IP or IV had a bimodal distribution (Figure 5C). In some animals, IV and IP inoculation resulted in very few or even only a single founder in intestinal samples (‘clonal’ group, n=9), indicating that the population that re-seeded the intestine had been affected by a severely restrictive bottleneck (Figure 5C). The remaining animals in the IV and IP groups with colon colonization (n=22) had a much larger number of founders, approaching the number observed in the immune-rich samples (Figure 4C), suggestive of a highly permissive bottleneck. Together, these data strongly suggest that there are at least two routes by which extraintestinal S. Typhimurium can return to the intestine; one route has little, if any, bottleneck and gives rise to a diverse population in the intestine and the other route includes a very tight bottleneck and often results in a clonal intestinal population.

Comparisons of the genetic distances showed that the IV and IP groups with clonal intestinal populations have the same clone (barcode) throughout their respective GI tracts (clonal colons GD=0, Figure 5D). In contrast, colons with diverse populations had less similarity to other regions of the intestine. Comparisons of the barcode similarity of clonal colon samples to extraintestinal samples provided insight into the likely extraintestinal source of the intestinal clone. 7 of the 9 animals with clonal populations in the colon had bile samples with genetic distances of zero, indicating the presence of the same clone and strongly suggesting that the bile clone seeded the intestine through the common bile duct. However, even though most bile samples containing S. Typhimurium from IP/IV inoculated mice were clonal (14/26) or pauciclonal (5/26 with Ns=2-10), most intestinal samples were not clonal, indicating that bile is often not the dominant extraintestinal site that seeds the intestine. Furthermore, clonal bile did not necessarily give rise to a dominant intestinal clone (Figure 6A, animals 1 and 2).

Interestingly, the bottleneck that characterizes the more permissive route to seeding the intestine from extraintestinal sites was even more permissive than the bottleneck that the pathogen experiences following orogastric inoculation (i.e., Ns values for the permissive IP and IV groups were generally greater than those for the orogastric group, Figure 5C). We speculate that the re-seeding route leading to the diverse S. Typhimurium population in the intestine represents a reversal of known mechanisms the pathogen employs to escape from the intestine (30).

Unexpectedly, we observed that animals with clonal colon populations often had clonal bile populations as well as visible biliary pathology, such as darkening, cloudiness, or hardening (Figure 6A-B). We noticed that animals with diverse colon populations often had gallbladders and bile lacking any visible pathology, regardless of the bile’s clonality. This led us to further analyze the association of gallbladder/bile pathology with the bile-driven intestinal re-seeding pathway. Overall, there was a small chance for any gallbladder/bile phenotype in all the infection schemes we tested, but there was a lower rate of bile colonization in untreated orogastrically gavaged animals (n=2) (Figure S8).

Analysis of aggregated data from all infection routes and doses showed that increased pathogen burden in bile correlated with cloudy or hardened biliary pathology (Figure 6D), but not darkened bile. Additionally, loss of biliary epithelial cells and marked cellular infiltrates were observed in cloudy and hardened bile samples (Figure 6C). Thus, cloudy and hardened biliary pathology either enables the growth of S. Typhimurium in the bile or, more likely, is the result of S. Typhimurium growth in the bile. Estimates of the fraction of barcodes that contribute to similarity between the bile and colon (FRD, see methods) revealed that a greater fraction of bile barcodes was shared with the colon in animals with cloudy and hardened bile samples compared to bile samples without pathology (Figure 6D). Collectively, these observations show that hardened or cloudy biliary pathology reflects robust S. Typhimurium replication in the gallbladder and is correlated with seeding of the intestine. Indeed, fecal shedding correlates with bile burden in animals with hardened or cloudy biliary pathology, but not with animals whose bile is darkened or lacks pathology (Figure 6F). Thus, STAMPR-based analyses uncovered an unexpected link between gallbladder pathology and bile-related intestinal re-seeding.

8-week-old female and male C57BL/6J mice were purchased from Jackson laboratory. Mice were acclimatized in the biosafety level 2 (BSL2) facility at the Brigham and Women’s Hospital for at least 72 h before use. The facility is temperature (68–75⁰F) and humidity (50%) controlled with 12 h light/dark cycles. All experiments involving mice were performed according to protocols reviewed and approved by the Brigham and Women’s Hospital Institutional Animal Care and Use Committee (protocol 2016N000416) and in compliance with the Guide for the Care and Use of Laboratory Animals.

A streptomycin-resistant strain of Salmonella enterica serovar Typhimurium SL1344 was used to create the barcoded library used here. Unless otherwise noted, S. Typhimurium was grown at 37⁰C in LB broth or solid agar. As needed, media was supplemented with streptomycin (SM, 200 ug/ml) and/or kanamycin (KM, 50 ug/ml).

The barcoded library was created using the pSM1 plasmid donor library, which contains >70,000 barcodes, and helper pJMP1339, which contains the Tn7 transposon and conjugation/replication machinery, as described (16–22). S. Typhimurium was heat shocked for 4 h at 42⁰C. A triparental conjugation was then used to introduce the pSM1 library into S. Typhimurium and transconjugants were selected for as SM- and KM-resistant colonies. Transconjugant colonies were pooled in PBS + 20% glycerol and stored in aliquots at -80⁰C. Illumina sequencing indicated the library contains ∼55,000 unique barcodes with even distribution (Figure S2). The growth of the library in LB supplemented with SM was not significantly different than the parent strain (Figure S2). The resolution limit of STAMPR analysis with our library was ∼700,000 founders; this was demonstrated through creation of a standard curve, comparing the true founding population (CFU following plating of a serially diluted culture) versus the calculated founding population (Figure S2).

The barcoded S. Typhimurium library was prepared by resuspending a frozen aliquot in LB and expanding for 3-5 h at 37⁰C with shaking unless otherwise noted. OD600 was used to estimate the size of bacterial population in the culture before use. Following growth, bacteria were pelleted and resuspended in 1x sterile PBS to the appropriate concentration (orogastric gavage 10⁹ or 10⁵ CFU/ml; oral via drinking 5x10⁹ CFU/ml; IP 10⁵ CFU/ml; IV 10⁴ CFU/ml) and stored at room temperature prior to inoculation.

Streptomycin sulfate (100 μl of 200 mg/ml, United States Pharmacopeia grade) was given by orogastric gavage 24 h before gavage with bacteria. Mice were deprived of food for 2-4 h before orogastric gavage and given light sedation with isoflurane inhalation immediately prior to gavage. 100 μl inocula were gavaged into the stomach using a 1 ml syringe and sterile 18 G, 1.5- inch, 2 mm ball, flexible feeding needles (Braintree Scientific). Before oral inoculation via drinking, mice were deprived of food for 8 h. 20 μl inocula were pipetted into the mouth of scruffed mice, who were then observed to ensure consumption. For IP infections, mice were restrained by a scruff, then a 1 ml syringe with a sterile 27 G, 1-inch needle was inserted into the lower right quadrant of the animal’s abdomen at a -45-degree angle, and 100 μl inoculum was injected. Before IV inoculation, mice were placed into small containers on a heating pad to help promote tail vein dilation and restrained with a Broome-style restrainer (Plas Labs). 100 μl inocula were injected into the lateral tail vein using a sterile 27 G needle.

Bacterial counts were enumerated for all inoculums using serial dilutions on LB agar supplemented with SM. Mice were monitored daily for signs of infection including weight loss, fecal shedding, and diarrhea. Mice with weight loss of <5%, indicative of lack of disease, were excluded from further analysis (orogastric gavage -SM after 10⁸ CFU n = 2, IP n = 3).

At the peak of the disease (4- or 5-days post-inoculation), at 5 h post-inoculation, or as needed per the humane endpoint in our animal protocol, mice were sacrificed by isoflurane overdose followed by cervical dislocation. Peritoneal washes were obtained by injecting 5 ml sterile PBS into the peritoneal cavity of the animal with a 27 G needle and clamping the hole shut before gently massaging the abdomen for ∼30 sec before aspirating the wash from the peritoneal cavity. When possible, bile was obtained from the gallbladder with a 31 G short-needle insulin syringe (Sol-M®, VWR) and aspirated into 90 μl sterile PBS. The gallbladder was then removed before harvesting the liver and other organs. If bile was not obtainable, the whole gallbladder was placed into a 1.5 ml tube containing 1 ml sterile PBS and 2x 3.3 mm stainless steel balls and processed. All other organs (perigonadal adipose tissue, cecum, colon, distal third and proximal third of the small intestine, Peyer’s patches, mesenteric lymph nodes, pancreas, liver, and spleen) were removed and individually placed into 1.5 ml tubes containing 1 ml sterile PBS (5 ml tube and 4 ml PBS for liver) and 2x 3.3 mm stainless steel balls and then homogenized using a bead beater (Biospec Products). CFU of all organs was enumerated by serial dilution plating on LB agar containing SM. The remaining sample homogenate was plated on large LB agar plates containing SM for processing before sequencing to enumerate barcode frequency.

The tissue samples were fixed in 4% paraformaldehyde in PBS and dehydrated in 20% sucrose in PBS. The dehydrated tissue was then embedded in OCT compound and sectioned to 10 µm. The sectioned slides were stained with hematoxylin and eosin (Abcam, #ab245880), washed with distilled water, and captured using Olympus Slideview VS200 (Rodent Histopathology Core, Dana-Farber/Harvard Cancer Center).

Samples were prepared for sequencing as described (16–22). Briefly, bacteria from large (150 mm x 15 mm) plates were resuspended in PBS + 20% glycerol and stored at -80⁰C. The frozen suspensions were diluted in water and boiled at 95⁰C for 15 min to obtain genomic DNA. An Eppendorf epMotion 5075 liquid handling robot was used to multiplex primer composition for sample preparation before PCR amplification (primers listed in Table S1). OneTaq HS Quick-Load (New England Biolabs) was used to amplify the genomic barcodes by PCR for 25 cycles. The presence of amplicons was confirmed by agarose gel electrophoresis. Samples were pooled and purified using a Qiagen DNA cleanup kit and DNA concentrations were checked with a Qubit fluorimeter, then sequenced with an Illumina NextSeq 1000/2000. FASTQ files were generated by Illumina’s proprietary pipeline in BaseSpace with DRAGEN BCL Conver v3.10.4. Sequencing reads were then demultiplexed, mapped to the donor barcode library pSM1, and trimmed in R (version 4.3.0) using custom scripts to obtain counts for each barcode (Barcode counts are listed in Dataset S1). The founding population (Ns) was determined through comparison to the frequencies of barcodes in the undiluted library using the STAMPR pipeline (16). Cavalli-Sforza chord distance (2√2/π = 0.9) was used to compare the genetic distance of samples from the same animal. FRD was calculated between sample A and sample B as FRDA-B = is a measure for the number of shared barcodes that contribute to genetic similarity (defined as GD < 0.8) between two samples, described extensively in (44).

STAMPR scripts are available online at [https://github.com/hullahalli/stampr_rtisan] Requests for further information, resources, or reagents will be fulfilled by the Lead Contact, Matthew K. Waldor (mwaldor@research.bwh.harvard.edu).

Statistical analyses were performed using GraphPad Prism version 10.1.2. Information regarding the number of samples and statistical tests are described in the figure legends. Geometric means, geometric standard deviations, and non-parametric tests were used for analyzing bacterial burden and founding population data. Means, standard deviations, and non- parametric tests were used for comparisons of animal weights, genetic distances, and barcode identity. As our lower limit of detection was near 1, we substituted null values with a value between 0.1 and 0.8 for graphical representation.