Colonic goblet cell-associated antigen passages mediate physiologic and beneficial translocation of live gut bacteria in preweaning mice

Abstract

Gut-resident microbes have time-limited effects in distant tissues during early life. However, the reasons behind this phenomenon are largely unknown. Here, using bacterial culture techniques, we show that a subset of live gut-resident bacteria translocate and disseminate to extraintestinal tissues (mesenteric lymph nodes and spleen) in preweaning (day of life; DOL 17), but not adult (DOL 35), mice. Translocation and dissemination in preweaning mice appeared physiologic as it did not induce an inflammatory response and required host goblet cells, the formation of goblet cell-associated antigen passages, sphingosine-1-phosphate receptor dependent leukocyte trafficking, and phagocytic cells. One translocating strain, Lactobacillus animalis^WU, exhibited antimicrobial activity against the late onset sepsis pathogen E. coli ST69 in vitro, and its translocation was associated with protection from systemic sepsis in vivo. While limited in context, these findings challenge the idea that translocation of gut microbiota is pathological and demonstrate physiologic and beneficial translocation during early life.

Introduction

The mammalian gut microbiota harbors vast numbers of diverse microorganisms. The contribution of gut-resident bacteria to host health is well appreciated including supporting gut immune development, protecting from enteric infection and facilitating gut nutrient uptake^1–9. Some of the reported gut bacteria induced host effects extend to distant tissues, such as the spleen ^10,11, thymus ^12,13, brain ¹⁴ and mammary glands ¹⁵. Gut-resident bacteria induce these effects on the host via multiple mechanisms such as production of exopolysaccharides ^3,16, bacterial metabolic intermediates ^5,17, antimicrobial compounds¹⁸, short-chain fatty acids^19,20, and by modification of host bile acids ²¹. Current concepts of how gut-resident bacteria mediate effects at distant sites include local activities of bacterial products that affect circulating cellular populations and/or bacterial products produced in the gut lumen diffusing and acting beyond the gut. While these models are sufficient to explain many of gut-resident bacteria’s effects at distant host sites, they appear less adequate to explain distant effects of gut-resident bacteria mediated by products produced in low concentrations and/or distant events mediated by unstable bacterial derived metabolites ¹².

Moreover, some gut-resident bacteria driven host effects are time-limited, only occurring during a period in early-life, and disruption of these critical time-limited events leads to long-term outcomes for the host ^22–24. We have previously reported that microbial antigen exposure during a specific preweaning period is critical for the induction of immune tolerance to select gut microbes ²⁵. Further, DNA of gut-resident bacteria has been found in the spleen and thymus in early life in healthy mice, however, the implications of its presence and how it reaches these distant sites is not understood ¹³. The gut microbiota changes dramatically in early life, and it is possible that the transitory presence of select gut microbes explains this restricted interval of benefit conferred by gut-resident microbes. However the relationship between the presence and/or abundance of these gut resident microbes and their extraintestinal effects is not straightforward as, at least for some microbe dependent events, the specific taxa are more abundant later in life when these events do not occur²⁵. Together these observations could suggest another dimension to our interactions with our gut microbes that explains why some gut-resident microbe dependent events are limited to early life and occur at extraintestinal sites despite being driven by scarce and/or unstable microbial products.

The healthy gut has barriers preventing translocation and systemic dissemination of live gut-resident bacteria ^26–30, an event overwhelmingly viewed as pathologic and detrimental to the host and the microbe. Accordingly, translocation and dissemination of live bacteria as a potential mechanism to promote beneficial gut microbiota mediated events is underexplored. However, in early life the mammalian gut is less developed and viewed as more permeable. The preweaning gut epithelia have differences from the adult including the differential expression of microbial pattern recognition receptors^31,32, the presence of vacuolated fetal enterocytes in the small intestine^33,34, and the presence of physiologically formed colonic goblet cell associated antigen passages (GAPs) ²⁵. We questioned whether gut-resident bacteria could take advantage of differences in unique early life gut epithelial biology or other unique features to translocate to extraintestinal tissues and mediate some of these beneficial gut microbial events in early life.

Results

Select gut bacteria translocate and disseminate in early life

We evaluated day of life 17 (DOL17; preweaning) mice, a time of life in which there is altered intestinal epithelial cell TLRs expression ²⁵, the presence of vacuolated enterocytes in rodents³⁴, and the presence of physiologically formed colonic GAPs²⁵, for translocation and dissemination of gut-resident bacteria. We tested multiple bacterial culture media/conditions and found that culture with brain heart infusion (BHI) media recovered all live bacterial taxa, identified by full length 16s rRNA sequencing of isolates, recovered from the mesenteric lymph nodes (MLNs) of DOL17 SPF housed C57BL/6 mice (Supplementary Table 1). Interestingly we did not observe live bacteria in MLNs of SPF housed C57BL/6 DOL35 (adult) mice in any culture condition. Accordingly, we used this approach to evaluate translocation in greater depth. To mitigate litter effects and evaluate effects of sex, we evaluated preweaning (DOL17) mice and adult (DOL35) mice from the same litter and analyzed by sex (schematic Fig 1A). Again, we observed that live bacteria were recovered significantly more often from the extraintestinal tissues (MLNs and spleen) of preweaning mice when compared to adult mice and that this phenomenon was not related to litter or sex, but highly dependent upon age (Figure 1B, Supplementary Table 2). Translocation of bacteria in preweaning mice was seen in a binomial distribution, with hundreds-thousands of organisms detected in a tissue or not occurring at all within the same litter (Figure 1B, Extended Data Figure 1A and related Supplementary Table 2). Mice with translocation appeared healthy and were indistinguishable from their littermates in which translocation was not detected. Using full-length 16s rRNA sequencing we found a limited number of bacterial taxa, dominated by Lactobacillus species, in the MLNs of preweaning mice, which represented a minor subset of the taxa present within the gut lumen (Figure 1C and Extended data Figure 1B). Thus, live bacteria spontaneously translocated in preweaning mice under physiological conditions and this stochastic event depended upon age but not sex or litter.

Figure 1: Live gut-resident bacteria spontaneously translocate to MLN and Spleen in preweaning mice:

(A) Schematic representation of the experimental set up used in the identification of translocating bacteria in mice. (B) CFU/organ (colony forming units per organ) of bacteria recovered from brain heart infusion (BHI) plates of spleen and MLN homogenates from DOL17 and DOL35 littermates (n=27 for DOL17 MLN, n=26 for DOL35 MLN, n=15 for DOL17 spleen and n=14 for DOL35 spleen) (C) Specific bacterial taxa identified from six DOL17 mice (M1 – M6) MLNs by full length 16s rRNA Sanger sequencing of isolates (D) Schematic representation of the experimental set up used to track L. animalis^WU colonization in intestinal tissues and translocation to MLN and spleen in DOL17 and DOL35 mice. (E) CFU/organ of L. animalis^WU recovered from intestinal and extraintestinal tissues of L. animalis^WU fed DOL17 (N=15) and DOL35 (N=15) mice. For B and E litters are color-coded, circle denotes females, and triangle denotes males to demonstrate that translocation was not litter or sex dependent but was age dependent. Graphs are presented as the mean +/− SEM. Statistical comparisons were performed using a one-sided cumulative binomial distribution probability for MLN and Spleen in panels 1B and 1E (see supplementary tables 2 and 3 for details) and two-tailed Students t-test for intestinal tissues, P values are as denoted.

The gut microbial community differs dramatically between preweaning and postweaning mice and therefore we questioned if the lack of translocation in postweaning mice was due to the lack of specific bacterial strains with the property to translocate. We generated an antibiotic resistant translocating Lactobacillus animalis strain isolated from the MLN of DOL17 mice (hereafter referred to as L. animalis^WU) and gavaged mice with this isolate (Figure 1D). L. animalis^WU colonized the intestinal tract of DOL17 and DOL35 mice equivalently and was found in both the luminal contents and mucosal scraping, suggesting it can reside in the lumen and close to the epithelium (Extended Figure 1C), but translocated significantly more often in DOL17 mice (Figure 1E, Supplementary Table 3). All preweaning mice colonized with L. animalis^WU appeared healthy irrespective of translocation. Translocation of L. animalis^WU in preweaning mice again appeared to be stochastic with a binomial distribution despite high levels of L. animalis^WU colonization of the intestinal tract (Figure 1E, Supplementary Table 3), indicating that factors other than the presence/abundance of bacteria with the ability to translocate were driving this process. Therefore, we sought to understand host dependent mechanisms facilitating translocation of L. animalis^WU in early life.

Early life translocation is facilitated by the host

Intestinal goblet cells (GCs) can form GC associated antigen passages (GAPs) to deliver luminal substances to the lamina propria immune system³⁵. In adult mice, GAPs are rare in the colon in the healthy state due to GC intrinsic Myd88 dependent sensing of the gut microbiota, suppressing colonic GCs’ ability to form GAPs ²⁷. However, when colonic GAPs form pathologically in adult mice they facilitate the translocation of gut-resident bacteria and induce inflammatory responses³⁶. In contrast colonic GAPs physiologically form for a defined preweaning period in early life which overlaps with the timing of translocation²⁵. We therefore evaluated if translocation was dependent upon GCs and GAPs. Gut bacteria could be found within colonic GCs of unmanipulated DOL17 mice using an eubacterial fluorescence in situ hybridization (FISH) probe (Figure 2A, Supplementary video 1) consistent with GCs/GAPs facilitating gut-resident bacterial translocation. Mouse atonal homologue 1 (Math1) is a transcription factor required for intestinal GC development. Deletion of GCs in preweaning mice through inducible deletion of Math1 in epithelial cells (Math1^f/f Vil-Cre-ER^T2 mice) did not impair the ability of L. animalis^WU to colonize the intestinal tract but significantly impaired the translocation and dissemination of L. animalis^WU (Figure 2B). Notably deletion of GCs increases intestinal leak³⁷ while paradoxically inhibiting translocation, consistent with translocation being a controlled rather than sporadic event. Prior to DOL10, colonic GAPs are inhibited by high luminal levels of epidermal growth factor (EGF) originating from the breastmilk, and gavage with exogenous EGF after DOL10 inhibits colonic GAPs³⁸. Gavaging preweaning mice with EGF to inhibit GAPs (Extended Figure 1D) did not increase gut barrier function (Extended Figure 1E), and trended toward increasing leak, or impair the ability of L. animalis^WU to colonize the intestinal tract, but significantly impaired the ability of L. animalis^WU to translocate to the MLN and spleen (Figure 2D), consistent with the role of GAPs facilitating translocation in preweaning mice.

Figure 2: Early-life L. animalis^WU translocation requires goblet cells, GAPs, and host cells that express S1PR and are phagocytic :

(A) Representative images from the proximal colon of DOL17 mice with gut-resident bacteria (eubacterial FISH probe; red) identified within a goblet cell (UEA1+ green) and DAPI (in blue) visualized in XY, YZ and XZ plane. (B) CFU/organ of L. animalis^WU recovered from intestinal and extraintestinal tissues of Math1^f/f (n=9) and goblet cell deficient Math1^f/fvil-Cre-ER^T2 (n=13) DOL17 mice fed with L. animalis^WU. (C) CFU/organ of L. animalis^WU recovered from the intestine and extraintestinal tissues of nontreated (n=14), EGF treated (n=14) and pan-S1PR modulator (FTY720) treated (n=6) DOL17 mice fed with L. animalis^WU. (D) Flow cytometry plots and (E) graph showing depletion of F4/80+ MHCII+ CD45+ splenic cells in Clodrosome treated preweaning mice (n=5) when compared to nontreated mice (n=4). (F) CFU/organ of L. animalis^WU recovered intestinal and extraintestinal tissues of PBS (n=4) and Clodrosome (n=4) treated preweaning mice. (G) CFU of in vitro cultured L. animalis^WU recovered after treatment of 10³ L. animalis^WU with gentamicin at the denoted concentrations for 2 hours. CFU/organ of L. animalis^WU recovered from (H) MLNs (n=3) and (I) Spleens (n=4) of preweaning mice in presence or absence of gentamicin treatment before cell lysis. Image in panel A is representative of colon sections from unmanipulated DOL17 mice from 3 independent litters in which 13 of 125 goblet cells (UEA1+) and 1 of 553 (UEA1-) colonocytes imaged in 3 dimensions (z-stacks) contained bacteria (eubacterial FISH probe +). Bar graphs presented as mean+/− SEM. Data points in panels H and I represent paired observations of CFUs within one half of the same MLN or spleen. Two- tailed Student’s t-test was used in E, H and I and intestinal tissues in B and F. A one-sided cumulative binomial distribution probability test was used for extraintestinal tissues in B, C, and F. A one-way ANOVA with at Dunnett’s post test was used for intestinal tissues in C and G.

Sphingosine-1-phosphate receptor (S1PR) is a cell surface receptor required for antigen presenting cells (APCs) acquiring luminal antigens from the majority of colonic GAPs to traffic to the draining lymph nodes ³⁹. Some gut bacteria can survive within lamina propria dendritic cells ^40,41, and S1PR is required for trafficking of some pathogenic bacteria to lymph nodes⁴². Therefore we evaluated if S1PR expressing cells facilitated the dissemination of L. animalis^WU to extraintestinal tissues in early life by treating preweaning mice with the S1PR modulator FTY720, which inhibits leukocyte trafficking (Extended Data Figure 1D). FTY720 treatment significantly reduced S1PR⁺ cells in spleen but not in the colon lamina propria, consistent with S1PR modulation affecting cell trafficking in preweaning mice (Extended Data Figure 2). FTY720 treatment did not affect L. animalis^WU colonization of the gut, but impaired its ability to translocate and disseminate (Figure 2C), consistent with S1PR expressing cells carrying L. animalis^WU to distant sites. We employed genetic and pharmacologic approaches to identify specific host cell subsets facilitating L. animalis^WU translocation. CSFR1 blockade inhibits the development of CSFR1 dependent populations which are largely myeloid derived phagocytes^43,44. CSF1R blockade depleted CD45+MHCII+CD11c-populations in the preweaning colon that were largely CX₃CR1-F4/80- or CX₃CR1+F4/80+ but did not affect CD45+MHCII+CD11c+ cells (Extended Data Figure 3A-E). CSF1R blockade reduced L. animalis^WU colonization in the small intestine but did not affect translocation and dissemination (Extended Data Figure 3F). We observed that despite lower colonization of L. animalis^WU in the colon, the lack of cDC1s in preweaning mice (Irf8 delta +32 enhancer)⁴⁵ did not impact dissemination of L. animalis^WU (Extended Figure 3G). Deletion of CX₃CR1 has been observed to inhibit dissemination of gut bacteria in adult mice^46,47, however surprisingly CX₃CR1 deficiency did not affect the translocation of L. animalis^WU in preweaning mice, despite lower levels of colonization in the colon (Extended Figure 3G). Clodronate encapsulated capsule liposomes (Clodrosome) effectively depletes phagocytic cell populations^48,49 and accordingly Clodrosome treatment effectively depleted CD45+ MHCII+ F4/80+ cellular populations in the spleen of preweaning mice (Figure 2D-E) and impaired L. animalis^WU translocation to spleen with a trend toward impairment in the MLN (Figure 2F). We assessed if L. animalis^WU could be found inside host cells by treating MLNs and splenic cellular populations from L. animalis^WU fed preweaning mice with gentamicin, which does not penetrate eukaryotic cell membranes and thus selectively kills extracellular but not intracellular bacteria. Gentamicin treatment killed cultured L. animalis^WU (Figure 2G), but L. animalis^WU was still recoverable from splenic (Figure 2H) and MLN (Figure 2I) cellular populations, suggesting that L. animalis^WU was able to reside to some extent intracellularly. Together these observations suggest that translocation and dissemination of gut-resident commensal bacteria in early life is a controlled process that is in part facilitated by the host.

Early life L. animalis^WU translocation does not trigger inflammation

Translocation of live gut-resident bacteria to systemic tissues is generally perceived as an unwanted event that occurs under pathologic conditions resulting in breach of the gut barrier and is linked with systemic pathologies detrimental to the host ^36,50–52. Supporting this concept, we have previously reported that pathologic opening of colonic GAPs in adult mice results in translocation and dissemination of live gut commensal bacteria triggering inflammation³⁶. Further, early-life bacterial translocation of gut resident pathogens is linked with late onset neonatal sepsis^38,53,54. Since L. animalis^WU was recoverable from extraintestinal tissues of healthy, conventionally raised laboratory mice and its translocation occurred via mechanisms in part dependent upon the host, we speculated that, unlike the so far studied pathological translocations, the translocation of L. animalis^WU might not be harmful but rather potentially beneficial to the preweaning mice.

Given that host cells harbor and carry L animalis^WU to distant tissues, we assessed the effects of translocation on the transcriptome of MLN APCs (CD45⁺CD19⁻B220⁻ MHCII⁺) from preweaning mice fed L. animalis^WU (confirmed translocation positive) or fed L. animalis^WU and EGF (confirmed translocation negative) (Figure 3A and Supplementary Table 4). Surprisingly, despite the presence of live bacteria in the translocation positive group and lack of live bacteria in the translocation negative group, only 69 differentially expressed transcripts were detected (Figure 3B, Supplementary Table 5: determined by DESeq2 analysis, significance P<0.05, FDR <0.05, fold change >2) which were largely not associated with immune signaling/inflammatory pathways expected in tissues/cells harboring live bacteria. Genes associated with regulation of intracellular signaling cascades (Abl1, Ccdc125, Hmgcr, Nucb2, Pdpk1, Prxl2c, Sos2 and Usp7) were upregulated and genes associated with cell junction organization (Hip1r, Myo1c, Ptpra, Rhob, Smad7, Tjp1 and Zfp703) were downregulated in the translocation positive group. (Figure 3B, Supplementary Table 5). Pathway analysis revealed 22 pathways that were enriched (P<0.05), none of which were related to infection or met the criteria of FDR<0.05 (Supplementary Table 6). Furthermore, blood neutrophils (Figure 3C), serum cytokines (IL6, TNFα, IL10; Figure 3D-F), and chemokines (CCL2, CXCL9 and CXCL10 Figure 3G-I) were not different between age-matched control preweaning mice, mice fed L. animalis^WU or mice injected with L. animalis^WU intraperitoneally (i.p.). IFNα was the only upregulated cytokine seen in mice fed with L. animalis^WU orally (Figure 3J). We assessed if L. animalis^WU translocation induced the production of Th1, Th2, or Th17 related cytokines by CD4⁺ T cells in MLNs and spleen (Extended data Figure 4). We did not observe T cells expressing multiple cytokines related to a T helper subset. However, we observed significantly increased expression of TNFα and IL17 secretion by splenic CD4⁺ T cells in L. animalis^WU colonized mice that was reversed by EGF treatment; this was not seen in MLN CD4+ T cells (Extended Data Figure 4 D and O compare with B and M). Interestingly inhibition of translocation by EGF increased production of IL17 by MLN CD4+ T cells (Extended Data Figure 4M). This suggests that in early life translocation is detected, or affects, to some extent, the adaptive immune system, however, the responses may vary by organ and not be as robust as expected in adults. The lack of systemic inflammatory response was not universal to all bacteria, as preweaning mice gavaged with E. coli ST69, a pathogen isolated from the blood of a child with late onset sepsis (LOS), displayed neutrophilia (Extended Data Figure 5).

Figure 3: L. animalis^WU translocation does not trigger a substantial systemic inflammatory response in preweaning mice:

(A) Schematic representation of experimental setup used in investigating transcriptomic changes associated with L. animalis^WU translocation in MLN of DOL17 mice. (B) Heatmap of 69 differentially expressed genes in sorted cellular populations defined as in (A) and confirmed by CFUs in the liver. (C) Peripheral blood neutrophils, and serum (D) IL6, (E) TNFα, (F) IL10, (G) CCL2, (H) CXCL9, (I) CXCL10 and (J) IFNα measured in DOL17 mice that are non-treated (n=6) or treated with L. animalis^WU fed orally (L. ani oral) (n=6) or administered by i.p. injection (L. ani i.p.) (n=6). Graphs presented as mean +/− SEM. Statistical analyses performed using DEseq2 with FDR<0.05 and >2 fold change for B in Partek^® FLOW^® and with one-way ANOVA with a Dunnett’s post test for C-J.

Early life translocation protects against systemic infection

Gut commensal bacteria are associated with immunomodulatory capabilities ⁵⁵, particularly with the induction of RORγt expressing colonic T regulatory cells (Tregs). Induction of colonic RORγt+ Tregs in early life has been attributed to short chain fatty acid (SCFA) producing bacterial taxa ⁵⁶, which largely do not include Lactobacilli. However L. rhamnosus has been observed to induce colonic RORγt Tregs to some degree ⁵⁶. Consistent with the transcriptomic analysis, gavage of L. animalis^WU from DOL10–20 did not increase the frequency of Tregs or RORγt+ Tregs in the spleen or MLN (Extended Data Figure 6A-D), however inhibition of GAPs and translocation with EGF reduced the population of RORγt+ Tregs in the MLN, consistent with prior observations of colonic GAP manipulation on colonic RORγt+ Treg populations in preweaning mice⁵⁷. This suggests that Treg induction is not the major beneficial effect L. animalis^WU translocation is providing to the host at this time in life.

Since our observations suggested that L. animalis^WU translocation was physiologic, yet not associated with overt immunological effects on the host, we sought to assess bacteria intrinsic properties that could be beneficial to the host. Sequencing and alignment of the L. animalis^WU genome using Prokka identified coding sequences (CDS) for a putative gramicidine-tyrocidine antibiotic synthesizing gene cluster (tycB, tycC, grsA, pikAV and sfp), resembling the organization of an operon(Figure 4A). Further analysis with antiSMASH⁵⁸ revealed five additional regions putatively coding for genes synthesizing antimicrobial products (Supplementary table 7, Supplementary files 1-6). These gene clusters were not observed in publicly available databases for Lactobacilli, including L. animalis, and therefore appeared to be a unique feature of this strain. We questioned if L. animalis^WU translocation might confer antimicrobial, or other potential benefits, in extraintestinal tissues which could limit systemic bacterial infections and/or septic responses in early life when the immune system is not fully developed. To test this concept, we utilized E. coli ST69⁵⁹, as we have demonstrated it causes disease in a preweaning mouse model of LOS³⁸. L. animalis^WU culture supernatants exhibited antibacterial activity against E. coli ST69 in vitro, which was reduced by treatment with proteinase K, suggesting the presence of L. animalis^WU protein products conferring antimicrobial activity (Figure 4B and 4C). Culture independent approaches that exclude free DNA within the tissue^60,61 did not demonstrate quantitative changes in the bacterial taxa in the MLNs of mice receiving augmented L. animalis colonization (Extended Figure 7), likely reflecting selection/evolution of the early life gut microbial community to co-exist with taxa possessing these properties.

Figure 4: L. animalis^WU translocation in early-life helps protect against systemic E. coli ST69 infection:

(A) Putative gramicidin-tyrocidine antibiotic coding sequences identified in L. animalis^WU by Prokka annotation and five other gene regions predicted by antiSMASH to encode secondary metabolites with potential antimicrobial activity (red and supplementary table 7). (B) Representative images of antibacterial activity of L. animalis^WU by agar diffusion in E. coli ST69 LB plates incubated with MRS broth (control), supernatant from overnight L. animalis^WU culture, L. animalis^WU culture supernatant treated with proteinase K (experiment was performed three times) (C). E. coli ST69 recovery from 6 hour cultured E. coli ST69 (n=3 per treatment group) grown in LB broth with no additional treatments (growth control), treated with MRS broth (vehicle control), L. animalis^WU culture supernatant or, L. animalis^WU culture supernatant treated with proteinase K, plated on LB agar. (D) Schematic representation of experimental setup used to assess the protective role of bacterial translocation in preweaning mice against systemic E. coli ST69 infection. Comparison of survival in (E) nontreated (n=9) and VNAM antibiotic treated (n=7) mice and (F) L. animalis^WU (n=5) and EGF + L. animalis^WU treated (n=5) mice that were infected with E. coli ST69 by i.p. injection. (G) Body weight, (H) percentage of segmented neutrophils in the peripheral blood (I) representative images of hepatic abscess (marked by black arrows in mice in both groups) and (J) quantification of hepatic abscesses per mm² in L. animalis^WU (n=4) and EGF + L. animalis^WU treated (n=4) mice infected 2.5 days post infection with E. coli ST69 by i.p injection. Graph in 4C represents mean+/−SD. Graphs in 4G, H, and J represent mean +/− SEM. Statistical analyses were performed by one-way ANOVA with a Dunnett’s post test for C and Log Rank (Mantel-Cox) test for E and F in GraphPad Prism, and a two-tailed Student’s t test for G, H and J; P values are as denoted.

To test if the early-life endogenous gut microbiota, which contains L. animalis^WU, is protective against systemic E. coli ST69 infection in vivo, we gave control and oral antibiotic (vancomycin, neomycin, ampicillin, metronidazole) treated preweaning mice systemic E. coli ST69 (Figure 4D). In contrast to the previously described LOS model³⁸, E. coli ST69 was given interperitoneally (i.p.) to avoid effects of early life gut microbes on the growth of E. coli ST69 in the gut lumen. Mortality of antibiotic treated mice was delayed by one day, which might reflect presence of residual systemic antibiotics with efficacy to ST69 (Figure 4 and Extended Data Figure 8). However, antibiotic treatment paradoxically made preweaning mice more susceptible, rapidly succumbing to systemic infection despite these antibiotics having activity against ST69 (Figure 4E). Notably L. animalis^WU displayed sensitivity to these antibiotics as well (Extended Data Figure 8). These observations are consistent with the endogenous gut microbiota, containing L. animalis^WU, being protective against systemic E. coli ST69 induced mortality in preweaning mice. To evaluate if the protective effect was further enhanced by supplementing with L. animalis^WU and if the protection was reversed by inhibiting translocation, mice were colonized with L. animalis^WU with or without EGF treatment to inhibit translocation. Inhibition translocation by EGF treatment increased mortality when compared to mice treated with L. animalis^WU alone (Figure 4F). Mice succumbed to E. coli ST69 with little forewarning, such that attempts to quantitate live pathogen burden in the systemic tissues while the mice were alive were unsuccessful. Therefore, we evaluated mice by necropsy 2.5 days following E. coli ST69 infection, a time in which all mice given L. animalis^WU with or without EGF were alive. Mice given E. coli ST69 i.p., L. animalis^WU, and EGF had significantly reduced body weight (Figure 4G). While mice in both groups displayed neutrophilia, mice treated with EGF had a significantly increased percentage of peripheral blood segmented neutrophils consistent with a more severe infection (Figure 4H and Supplementary Table 8). There were no significant gross findings in mice in either group. Histologic examination revealed no significant abnormalities in the brain, heart, large intestine, pancreas, and kidneys, mild enteritis of the small intestine, and mild to moderate extramedullary hematopoiesis in the spleen of mice in both groups. Moderate neutrophilia was seen in the lungs of two mice not receiving EGF. However, mice given E. coli ST69 i.p., L. animalis^WU and EGF had significantly increased number of hepatic abscesses (Figure 4I and J), consistent with worse control of systemic E. coli ST69 infection, or the response to the infection, when translocation was inhibited. In further support of L. animalis^WU providing protection, mice receiving augmented L. animalis^WU colonization showed enhanced survival during E. coli ST69 infection when compared with unmanipulated mice (Compare black line Figure 4E with blue line Figure 4F, P=0.0190 (Log-rank Mantel-Cox test)). Therefore, the endogenous preweaning microbiota, and specifically L. animalis^WU, enhanced survival to systemic E. coli ST69 infection and this effect was reversed by inhibiting physiologic translocation of resident gut microbes during this preweaning period, however the definitive mechanism(s) behind this benefit (antimicrobial activity vs other effects) remain to be explored.

Discussion

There is a growing appreciation that the gut microbiota is not merely a passive resident but a large contributor to host physiology. Beyond gut centric effects, the gut microbiota has been implicated in disease, and by extension health, in distant organs systems including the endocrine, dermatologic, neurologic, pulmonary, renal, and reproductive systems. Current concepts are that gut residing microbes mediate these extraintestinal events by imprinting local cellular populations which migrate to extraintestinal sites and/or microbial products produced within the gut diffuse to distant sites⁶². While these concepts are sufficient to explain many, if not most, of the distant effects attributed to the gut microbiota, some observations extraintestinal events mediated by rare and/or unstable gut microbial products¹² seem at odds with this model.

Here we identified that select live gut resident bacteria have the capacity to translocate and disseminate in early life. This phenomenon was not dependent upon litter or sex but was dependent upon age. Translocation in early life appeared stochastic with a binary distribution, either occurring with hundreds to thousands of live bacteria translocating or no live bacteria translocating despite equivalent levels of gut colonization by translocating bacteria. We do not understand the basis for this pattern of translocation. Our impression, based on the timing of colonic GAP formation in early life and observations in mice within or outside of this period of colonic GAP formation, is that translocation occurs during a ~10 day preweaning window and we may have missed translocation in some DOL17 mice. Other possibilities such as alterations in the developing microbiota, alterations in maternal or other environmental factors are also potential explanations. However, translocation varied within litters, which have similar exposure to maternal and environmental factors and similar gut microbiotas. Further we did not observe differences in the gut microbiota from mice with and without translocation within the same litter, albeit analysis used 16S rRNA variable region sequencing, which may not have granularity to decipher subtle differences. Consistently we observed is that only a subset of the gut resident bacteria translocate in early life, which was not related to relative abundance with the gut. Lactobacilli were the most common translocating taxa in these studies, yet Lactobacilli made up ~5% of the taxa identified in the preweaning gut microbiota. We do not favor that this is solely a result of failure to culture some taxa, as taxa, which we observed not to translocate, should be culturable by our methods. Based upon observations here and other work we believe translocation is a property of non-adherent bacteria that can reside in the mucus layer. Other factors that account for the capacity to translocate is unclear but the subject of ongoing work. Likewise, determination of the preferential tissue niche(s) and function(s) of other physiologically translocating species and their effects on the host will require further study.

Translocation and dissemination of gut resident bacteria is almost universally viewed as an unwanted event with detrimental outcomes for the host. Here we observed that translocation of live gut resident commensal bacteria did not adversely affect the host’s health. Moreover, this process was in part host dependent as it required host goblet cells, the formation of GAPs, S1PR function, and phagocytic cells for translocation and dissemination. In contrast to what has been described in adult mice^46,47, we did not find that dissemination was dependent upon CX₃CR1, nor was it dependent upon cDC1 DCs or newly developing CSFR1 dependent cells. This likely reflects unique cell populations inhabiting the gut in early life with differing biology from those in adults^63,64. Further systemic dissemination of live translocating bacteria in early life did not elicit a significant inflammatory response, which is in part due to properties of the bacteria and in part due to age specific features of the host. Together these observations suggest this is a physiologic event, and to be evolutionarily conserved, has some benefit to the host at this time in life. Given the role of gut commensal bacteria in shaping the immune system, we were surprised that we did not observe a role for translocation of L. animalis^WU in immune development, although admittedly our investigations were not exhaustive. Whole genome sequencing of L. animalis^WU revealed the presence of a clusters of genes putatively producing antimicrobial products, suggesting a bacteria intrinsic purpose for translocation. The regions were not found in publicly available databases of other Lactobacillus species and are therefore inferred to potentially be unique to this strain. L. animalis^WU products demonstrated antimicrobial activity in vitro, and in vivo studies demonstrated a benefit of L. animalis^WU in a systemic sepsis model, which could be reversed by limiting the translocation of L. animalis^WU. Admittedly the studies presented here do not definitively prove this is the mechanism providing protection in this model in vivo. L. animalis^WU might have other yet to be discovered properties, potentially including immunomodulation, that benefit the preweaning host during systemic infection.

Lactobacilli spp. have demonstrated antibacterial activity in vitro⁶⁵ and a Lactobacillus species, L. salivarius UCC118, produces the antibiotic bacteriocin, Abp118, which acts in the gut lumen to protect from infection by the enteric pathogen L. monocytogenes⁶⁶. In addition, L. murinus, has also been found to be protective in a model of LOS by acting within the gut lumen to limit pathobiont colonization⁶⁷. Thus, although not investigated here, it is likely that L. animalis^WU would have role(s) in out competing other microbes and/or conferring antimicrobial activity within the gut lumen which, could likewise confer benefits to the early life host to decrease enteric infections, limit the dissemination of other gut microbes, and facilitate the assemblage of the healthy gut microbial community.

Here we describe a surprising, and by current concepts almost heretical, role for translocation and dissemination of live gut bacteria. Although surprising from today’s perspective of live bacterial translocation and septic outcomes, this phenomenon may be less surprising from an evolutionary perspective that suggests endosymbiotic intimate relationships between prokaryotes and their eukaryotic hosts resulted in the development of organelles like mitochondria and chloroplasts⁶⁸. Importantly our studies were performed in a limited context regarding host species/strain, housing conditions, and gut microbial community, and thus further investigation will be required to define how far these observations extend in other settings. Logically one speculates if this phenomenon is also true in humans. While a study of random culture of blood and tissue of healthy human infants and monitoring outcomes in the absence of antibiotic therapy, if culture positive, is not possible due to ethical concerns by today’s standards, a study six decades ago evaluated random blood cultures from 131 three day old asymptomatic children found ~13% were culture positive without evidence of sepsis or adverse outcomes⁶⁹. The authors conclude these positive cultures could result from contamination or transient bacteremia, which were indistinguishable, but might be consistent with physiological translocation in humans in early life.

In summary, the gut microbiota is increasingly appreciated as a major driver of host development and physiology. Current concepts are that benefits of the gut microbes conferred to the host are restricted to microbes residing in the gut lumen and once the gut barrier is breached, microbes have universally pathophysiologic effects on the host. However, our observations suggest a new level of symbiosis with our gut microbes in early life where translocation and dissemination of live microbes is not an internecine event resulting in mutual destruction, but an unappreciated and intimate level of symbiosis.

Methods

See supplementary information for detailed methods.

Mice

Animal procedures and protocols were performed in accordance with ethical and other regulations per the IACUC at Washington University School of Medicine. All mice were bred in house for experiments. Mice were housed in a specific-pathogen-free facility and fed routine chow diet. Mice of both sexes were used in this study and where possible littermates were used as controls.

Extended Data

Extended Data Figure 1: Bionomial distribution of translocation, characterization of colon bacterial taxa at family level in preweaning and adult mice, luminal vs mucosal residence of L. animalis^WU, and changes in gut barrier with EGF:

A) Pattern of number of CFUs recovered from the MLN and spleen of DOL17 mice (relates to supplementary table 2. B) Relative frequency of bacterial taxa at family level compared by 16S v4 sequencing of specific pathogen free laboratory mice at DOL17 (n=3) and DOL35 (n=4). Lactobacillaceae family (denoted by red box in the legend and arrows on the graph) was one of the most common translocating taxa, is present at both ages. C) Culture of luminal vs mucosal scrapings from L. animalis^WU fed DOL17 mice (n=4) reveals that L. animalis^WU can be present in both compartments. D) Schematic of EGF treatment to inhibit GAPs and FTY720 treatment to inhibit trafficking. E) Assessment of gut barrier function in L. animalis^WU colonized mice in the presence and absence of EGF (n=3 per treatment group) using the 4kD FITC dextran leak assay. Graphs represent the mean +/− SEM. P value calculated using two-tailed Student’s t-test.

Extended Data Figure 2: S1PR modulation alters immune cell trafficking to spleen in preweaning mice:

Percentage of immune cells expressing S1PR (%S1PR^GFP+CD45⁺) assessed in colon and spleen of S1PR-GFP reporter mice (B6.129P2-S1pr1^tm1Hrose/J) treated at DOL17 with pan-S1PR inhibitor (FTY720) (n=3) or not treated (control) (n=3). Statistical analyses were performed by two-tailed Student’s t test in GraphPad Prism. Data represented as mean with individual values. P values are as denoted.

Extended Figure 3: CSF1R blockade, loss of cDC1 cells, and CX₃CR1 deletion does not impair L. animalis^WU translocation in preweaning mice:

(A) Flow cytometry plots and (B-E) graphs demonstrate that CSFR1 blockade in preweaning mice (n=3) reduces the CD45+ MHCII+ CD11c- colonic LP cellular population which can also express CX₃CR1 and F4/80. (F) CFU/organ of L. animalis^WU recovered from intestinal and extraintestinal tissues of L. animalis^WU fed preweaning mice that were nontreated controls (n=4) or treated with anti-CSF1R (n=3). (G) CFU/organ of L. animalis^WU recovered from intestinal and extraintestinal tissues of L. animalis^WU fed wildtype (n=6), cDC1 deficient (Irf8 delta 32) mice (n=6), or CX₃CR1 deficient preweaning mice (n=5). Graphs represent mean +/− SEM. Statistical analyses were performed by two- tailed Student’s t test for B - F, one-way ANOVA for intestinal tissues in G, and one-sided cumulative binomial distribution probability test for extraintestinal tissues in F and G.

Extended Figure 4: Induction of Th1, Th2 and Th17 cytokines by MLN and splenic CD4⁺ T cells from preweaning mice given L. animalis^WU with and without EGF :

(A) Flow cytometry gating strategy for identifying Th1 cytokines (TNFα and IFNɣ) secreted by CD45⁺CD3⁺CD4⁺ T cells in MLNs and spleens. Frequency of TNFα+ and IFNɣ+ CD4+ T cells in (B-C) MLNs and (D-E) spleens of nontreated (control) (n=9), L. animalis^WU fed (n=5) and L. animalis^WU+EGF fed mice (n=6). (F) Flow cytometry gating strategy for identifying Th2 cytokines (IL4 and IL13) secreted by CD45⁺CD3⁺CD4⁺ T cells in MLNs and spleens. Frequency of IL13+ and IL4+ CD4+ T cells in (G-H) MLNs and (I-J) spleens of nontreated (control) (n=9), L. animalis^WU fed (n=5) and L. animalis^WU+EGF fed mice (n=6). (K) Flow cytometry gating strategy for identifying Th17 cytokines (IL17 and IL22) secreted by CD45⁺CD3⁺CD4⁺ T cells in MLNs and spleens. Frequency of IL17+ and IL22+ CD4+ T cells in (L-M) MLNs and (N-O) spleens of nontreated (control) (n=9), L. animalis^WU fed (n=5) and L. animalis^WU+EGF fed mice (n=6). Statistical analyses were performed by one-way ANOVA with Dunnett’s post test. Graphs represent mean+/− SEM. P values are as denoted.

Extended Data Figure 5: Peripheral blood neutrophils increase in preweaning mice infected with E. coli ST69:

Neutrophil (CD45⁺Ly6G/C⁺ cells) numbers in non-infected mice (n=3) or mice infected with E. coli ST69 (n=5) gavaged orally at DOL17. Statistical analyses were performed by two-tailed Student’s t test in GraphPad Prism. Graph represents mean +/− SEM. P values are as denoted.

Extended Data Figure 6: L. animalis^WU does not induce regulatory T cell subsets in preweaning mice:

A) Graphs and B) representative flow plots of Foxp3+ and RORγt+ Foxp3+ regulatory T cell populations in the MLN and spleen of unmanipulated mice (Control) (n=4 for MLNs and 3 for spleens), mice fed L. animalis^WU from DOL10–20 (L. animalis^WU) (n=4 for MLNs and 3 for spleens), and mice fed L. animalis^WU given EGF gavage from DOL10–20 (L. animalis^WU +EGF) (n=4 for MLNs and spleen). Graphs represent mean +/− SEM. Statistical analyses were performed by one way ANOVA with a Dunnett’s post test.

Extended Figure 7: Characterization of MLN and splenic bacterial taxa at family level in preweaning mice with and without L. animalis^WU feeding.

Cellular populations from the A) MLN and B) spleen were isolated and treated with propidium monoazide and photoactivation, DNA was isolated and bacterial taxa were characterized by 16s rRNA v4 sequencing. The number of different taxa identified were not dramatically altered by L. animalis^WU feeding.

Extended Data Figure 8: E. coli ST69 and L. animalis^WU are sensitive to vancomycin, neomycin, ampicillin and metronidazole:

Colony forming units (CFU) of (A-D) E. coli ST69 (n=3) and (E-H) L. animalis^WU (n=3) after plating bacteria treated with (A, E) Vancomycin (B, F) Neomycin (C, G) Ampicillin and (D, H) Metronidazole at the specified concentrations for 4 hours. Statistical analyses were performed by one way ANOVA with a Dunnett’s post test. Data represented as mean +/− SEM P values are as denoted.

Data availability

The stool sequencing data and L. animalis^WU whole genome sequencing data are archived on the Sequence Read Archive (BioProject: PRJNA1067122 and PRJNA1066880). The RNA-seq data is archived on Gene Expression Omnibus (GSE278303).