Maternal gut microbiome-induced IgG regulates neonatal gut microbiome and immunity

Maternal IgG antibodies are transferred to maternal milk through FcRn and coat bacteria in the neonatal intestine

We previously reported the presence of gut microbiome-specific IgG antibodies that circulate systemically in adult mice and confer critical protection against systemic infection by pathogens that share conserved IgG surface antigens. These IgG antibodies, however, were scarce in the sera of mice 4 weeks after birth, as well as in the intestinal lumen of adult mice at steady state (14). To appreciate the function of these antibodies in the neonatal gut, we first asked whether neonates acquire these protective IgG antibodies from the mother. We detected robust levels of IgG in WT maternal milk that recognized proteins of gram-negative gut commensal or pathogenic bacteria, E. coli and C. rodentium, respectively (Figure 1A). During fetal development, maternal antibodies are transported through the placenta to the fetus during the third trimester via the neonatal Fc receptor FcRn (Fcgrt) (21). FcRn is expressed at high levels in epithelial cells in human mammary glands to facilitate the transfer of serum IgG to maternal milk (22), as well as in neonatal intestinal enterocytes to facilitate the uptake of maternal milk and transcytosis to circulation (23). We found that the transfer of gut microbiome-specific IgG antibodies to maternal milk is dependent on FcRn as well, as gut microbiome-specific IgG antibodies were almost undetectable in Fcgrt^−/− maternal milk (Figure 1A).

Binding to FcRn prolongs the half-life of serum IgG. Consistently, we detected a 20–30% reduction in serum IgG1 and IgG2a in adult FcRn-deficient (Fcgrt^−/−) mice (Figure S1A). The amounts of serum IgG against fecal bacteria in adult Fcgrt^−/− mice were only ~40% of that detected in the sera of WT adult mice (Figure S1B–C). This demonstrates that the scarcity of gut microbiome-specific IgG in Fcgrt^−/− maternal milk is attributable to two factors: lower levels of serum IgG against gut bacteria in the dam and lacking FcRn to transfer these IgG antibodies from serum to maternal milk.

We further demonstrated that FcRn mediates the transfer of gut microbiome-reactive maternal IgG to the neonatal intestine, which directly targets gut bacteria in neonatal mice (Figure 1B). The same trend was observed for IgA-coating of gut bacteria between neonatal and adult mice (Figure 1C). This may reflect increased abundance of gut microbiome-specific IgG and IgA in the neonatal intestine due to direct supply of maternal milk to the neonatal intestine; whereas in adults, IgG in the intestinal lumen is limited at steady state and increases only in response to gut injury or infection, as implicated in our previous study (14). Furthermore, IgA-coated fecal bacteria from neonates and adults were almost all positive for IgG as well (Figure 1D), suggesting originally within the dam the same population of gut commensal bacteria elicited both IgA and IgG antibody response.

Maternal IgG antibodies opsonize and inhibit colonization and mucosa-attachment of C. rodentium in the neonatal gut

Similar clearance of C. rodentium was observed in adult Fcrgt^−/− and WT (Fcrgt^+/+) mice (Figure 2A). To investigate whether maternal IgG antibodies in the neonatal gut confer protection against enteric pathogens in neonates, we orally infected 18-day old WT or Fcgrt^−/− mice with gram-negative C. rodentium. While pathogen loads detected in feces from WT infected neonates gradually increased but started to drop around day 7 following infection, the bacterial loads in Fcgrt^−/− neonates were detected at higher levels as early as day 2–3 after infection and remained significantly higher beyond day 7 (Figure 2B). This suggests a possible inhibitory effect of maternal IgG on the colonization of C. rodentium in the neonatal gut. In contrast to almost complete protection against infection in WT and IgA-deficient neonates, we observed 100% mortality in Fcgrt^−/− neonates, akin to that found in B cell-deficient J_H^−/− mice (Figure 2C). Of note, stool bacteria of Fcgrt^−/− neonates transplanted into germ-free (GF) neonates showed similar colonization resistance against C. rodentium as stool bacteria from WT neonates (Figure 2D), indicating the increased mortality in Fcgrt^−/− neonates (Figure 2C) was likely due to lack of maternal IgG rather than reduced colonization resistance against C. rodentium by the gut bacteria in Fcgrt^−/− neonates. To understand the mechanisms underlying maternal IgG-mediated protection, we orally infected WT, B cell-deficient J_H^−/− (24), or Fcgrt^−/− neonates with a GFP-expressing strain of C. rodentium. On day 6 post infection, ~10% luminal GFP-C. rodentium with surface IgG coating was observed in WT infected mice, in contrast to almost complete absence of IgG-coated luminal GFP-C. rodentium in Fcgrt^−/− mice (Figure 2E). Previous studies showed that virulent C. rodentium is able to adhere to the intestinal mucosa, and inability to inhibit attachment of virulent C. rodentium to the intestinal mucosa is associated with increased penetration through the intestinal epithelium and systemic dissemination of C. rodentium (25). To assess whether IgG-coating would affect the adherence of virulent C. rodentium to the mucosa, we used a strain of C. rodentium that expressed luminescence under the Ler promoter (C. rodentium-ler-lux), which allowed detection of C. rodentium expressing Ler-dependent virulence factors in vivo (26). After removing luminal contents prior to imaging, we found increased luminescence-expressing mucosa-associated C. rodentium on the ceca and colons of infected Fcgrt^−/− neonates compared to WT neonates 6 days after infection, which persisted even on day 12 following infection (Figure 2F). In addition, the colon length of infected Fcgrt^−/− neonates was noticeably shorter than that of WT mice, thus consistently indicating increased intestinal inflammation in infected Fcgrt^−/− neonates (Figure 2F). These observations indicated increased adherence of virulent C. rodentium to the mucosa in the absence of IgG. Consistently, increased numbers of translocated C. rodentium were detected in the blood, spleens and livers of Fcgrt^−/− neonates (Figure 2G), likely leading to the ultimate demise of these mice (Figure 2C).

Since maternal IgG antibodies are transferred placentally during pregnancy and via maternal milk after birth, to delineate the contributions of prenatal and postnatal transfer of maternal IgG antibodies to neonatal immune defense against enteric infection, WT newborns were cross-fostered with a Fcgrt^−/− dam immediately after birth. Upon oral infection with C. rodentium, the fostered neonates exhibited increased mortality (>70%) compared to 100% survival in littermates that were nursed by biological WT dams (Figure 2H). This result suggests postnatal maternal IgG transfer through maternal milk accounts for most of the immune protection by maternal IgG against enteric gram-negative pathogens in neonates.

Colonization of IgG-inducing gut bacteria, or maternal immunization with OMP-A, confers protection against enteric infection via induction of IgG

To further elucidate the role of the gut microbiome in the induction of protective IgG against C. rodentium, we isolated luminal bacteria with or without IgG coating from 3-week-old WT mice, which were inoculated by oral gavage into 3-week-old WT germ-free (GF) mice (Figure S2A–C). Two weeks later, serum IgG antibodies against fecal bacteria were induced at higher levels in the ex-GF mice that were transplanted with IgG-coated bacteria (IgG⁺ exGF) compared to that in mice transplanted with bacteria not coated with IgG (IgG- exGF) (Figure 3A). Of note, there was no difference in the amount of serum IgA against fecal bacteria between the two groups of mice. Despite differential abundance of some bacterial taxa in IgG⁺ exGF mice, there was no apparent gut inflammation in either group prior to C. rodentium infection (Figure S2B–C). Importantly, following oral infection with C. rodentium, we detected lower numbers of translocated C. rodentium in the spleens and livers of IgG⁺ exGF mice (Figure 3B). Taken together, our results further demonstrated protection against C. rodentium by IgG-inducing gut bacteria.

Furthermore, to test whether maternal immunization to increase gut microbiome-specific IgG would enhance neonatal immunity, potential gut microbiome-derived IgG antigens from the outer membrane of E. coli K12 was immunoprecipitated using purified IgG from naïve WT adult mouse sera (Figure S2D). Immunoprecipitated proteins were analyzed by mass spectrometry. One of the top proteins identified was outer membrane protein (OMP) (Figure S2E), which is present in all gram-negative Enterobacteriaceae (27). To verify the presence of anti-OMP IgG in mice, we used cell lysates from E. coli strains that were deficient in one of the known OMPs (OMPs C, A, F and X), and confirmed the presence of anti-OMP-A IgG in sera from naïve adult WT mice (Figure 3C). Furthermore, we immunized female mice with purified OMP-A prior to mating and re-boosted with OMP-A administration two weeks later, which resulted in increased levels of anti-OMP-A serum IgG in immunized females (Figure S2F). The offspring from the immunized or non-immunized dams were orally infected with a high dose of C. rodentium (5×10⁸), which caused 100% mortality in the offspring from non-immunized dams, in contrast to ~20% mortality and lower bacterial loads in the blood in the offspring from immunized dams (Figure 3D–E). Importantly, the protection conferred by maternal immunization was absent in Fcgrt^−/− mice (Figure 3F) and independent of IgA (Figure 3G), thus suggesting the protection was mediated through induction of IgG antibodies. Collectively, our data suggested the importance of gut microbiome as a source of IgG antigens to elicit homeostatic IgG that can be transferred via maternal milk to confer critical protection against enteric pathogens in neonates.

IgG-deficient neonatal mice are more susceptible to enteric C. rodentium infection

To further investigate how lacking maternal IgG may affect gut immune response and susceptibility to enteric infection in neonatal mice, we generated a mouse model that is deficient in all IgG isotypes (IgG1, 2a, 2b and 3) by deleting >80 kb of DNA fragments containing the mouse IgG-encoding loci using a CRISPR-Cas 9 gene editing system (Figure S3A–C). The founders were on B6 background and were further backcrossed to the B6 background for more than 10 generations. Deficiency of all IgG isotypes was confirmed in the IgG KO mice, while serum IgA, IgE, and IgM were intact in these mice compared to WT littermates (Figure 4A and S3C). Consistent with findings in Fcgrt^−/− neonates, P7 and P14 IgG KO neonates showed increased systemic translocation of C. rodentium (Figure 4B–C), increased gut barrier permeability and intestinal inflammation (Figure S3D–E) and mortality following C. rodentium infection (Figure 4D).

IgG deficiency in neonates is associated with alterations in the gut microbiome and increased IL-17A-producing cells in the intestine

To investigate whether lacking maternal IgG would affect the composition of gut microbiome, we analyzed the gut microbiome in IgG KO and WT littermates that were separated right after birth and nursed by IgG KO and heterozygous foster dams (both dams were littermates that were co-housed until used for cross-fostering), respectively. We found differential abundance of several Operational Taxonomic Units (OTUs) within the bacterial populations in both the small intestines and colons of 2-weeks-old IgG KO and WT neonates (Figure 5A). Specifically, in the colon, two OTUs of Porphyromonadaceae in particular were more abundant in WT than IgG KO neonates. Additionally, while an OTU of gram-positive Lactobacillus was more abundant in WT SI, an OTU of gram-negative Enterobacteriaceae and an OTU of Prevotellaceae were more abundant in IgG KO SI (Figure 5A).

Additionally, to determine whether IgG deficiency may change gut immune cell responses in the neonatal gut, we analyzed lamina propria cells in 2-week-old IgG KO and WT neonates. We found increased numbers of IL-17A-producing cells within the CD3⁺CD4⁻CD45⁺ lymphocyte populations in both the small intestines and colons of IgG KO neonates (Figure 5B–C, S4A–B). Consistent with the enhanced IL-17A response, we found increased neutrophils in the colons of IgG KO neonatal colons (Figure 5D, S4C). Of note, almost all of these IL-17A-producing cells expressed a γδ TCR, suggesting they were γδ T cells (Figure 5E). In addition, these IL-17A-producing γδ T cells were diminished in WT germ-free (GF) neonates compared to specific-pathogen-free (SPF) WT neonates (Figure S4D), suggesting the gut microbiome as a driver of the development of γδ T cells in the neonatal gut. To determine whether increased IL-17A production in the IgG KO neonatal gut was driven by altered gut microbiome in these mice, we re-derived IgG KO mice germ-free (Figure 5F). Similar numbers of IL-17A-producing CD3⁺ T cells were observed in WT and IgG KO GF mice (Figure 5G). Additionally, 2 weeks following oral transplantation of stool from P7 SPF WT or IgG KO stool into P7 WT GF neonates, we found increased IL-17A⁺ γδ T cells and IL-17A⁺ CD4⁺ cells in the intestines of the recipient mice with stool from IgG KO donors (Figure 5H, S4E). Furthermore, cross-fostering by a WT dam immediately after birth led to reduced numbers of IL-17A⁺ γδ T cells and IL-17A⁺ CD4⁺ cells in IgG KO P14 neonates (Figure 5I, S4F). Collectively, our findings suggest that postnatal maternal IgG is critical to dampen intestinal IL17-producing γδ T cells in neonates via regulation of the gut microbiome.

Altered gut microbiome increases the susceptibility to DSS-induced colitis in IgG KO adult mice

To investigate whether the role of gut microbiome-specific IgG in shaping gut microbiome and immune cells is restricted to the early developmental stage, IgG KO and WT adult mice (at least 8 weeks of age) were first co-housed for two weeks before separation and induction of colitis by administration of 2% dextran sulfate sodium (DSS) in drinking water for 7 days. Reduction in colon lengths, an indicator of disease severity, was more pronounced in DSS-treated IgG KO mice (Figure 6A), with increased numbers of IL-17A-producing colonic lamina propria T cells and macrophages (Figures 6B, S5A). Despite homogenization of the gut microbiome between IgG KO and WT mice after two weeks of co-housing, the gut microbiome appeared to diverge during the course of DSS treatment during which these mice were separated (Figures 6C–E). For example, an OTU corresponding to Akkermansia was significantly reduced in IgG KO DSS-treated mice, which had enriched Erysipelotrichaceae (Figure 6D). This suggests IgG antibodies are important to control the gut microbiome in adults as well, which might be even more critical in the setting of gut inflammation where opportunistic gut bacteria over-bloom and further exacerbate gut injury. To further interrogate the role of the gut microbiome in the altered immune response to DSS in IgG KO mice, IgG KO and WT adult mice were first co-housed for two weeks before DSS treatment and remained co-housed throughout the DSS treatment (Figures 6F–G). The differences in the composition of the gut microbiome, colon shortening and colonic lamina propria immune cells, including IL-17A⁺ T cells, were no longer observed between the two genotypes (Figures 6F–G, S5B–E), thus suggesting the gut microbiome was a driver in the increased gut inflammation in IgG KO DSS-treated mice when housed separately from WT mice (Figures 6A–E). Of note, in naïve WT mice, we found that co-housing with IgG KO mice led to increased numbers of IgG-coated fecal bacteria, implicating the transfer of IgG-targeting bacteria that over-bloom in IgG KO mice to the WT co-housed mice. Consistently WT mice co-housed with IgG KO mice during the course of DSS treatment had increased IL-17A⁺ intestinal cells to the levels observed in the co-housed IgG KO mice (Figure 6G); this suggests the transferred IgG-coated bacteria (Figure S5F) in the WT co-housed mice were likely a driver of the IL-17A response. Together, these findings demonstrate that the critical role for IgG to control the gut microbiome and immune cells extends beyond the early developmental period.

Study Design

The aim of this study was to investigate the role of maternal gut microbiome-reactive IgG antibodies in the regulation of neonatal immunity against enteric bacterial pathogens, and the development of the gut microbiome and immune cells in the neonatal offspring. We employed flow cytometry to identify IgG-coated gut commensal and pathogenic bacteria, a mouse model of enteric infection with C. rodentium, gnotobiotic mice to induce gut microbiota-specific IgG, and lastly, a genetic model of complete IgG deficiency to analyze the development of neonatal gut microbiome and immune cells in the absence of IgG.

Mice

Wild-type and Fcgrt^−/− mice (B6.129X1-Fcgrt^tm1Dcr/DcrJ; strain 003982) on a C57BL/6 background were originally purchased from the Jackson Laboratory. IgA-deficient mice were originally from Dr. Bernard Arulanandam (40). J_H^−/− mice were kindly provided by Marilia Cascalho (Univ. of Michigan) (41). All animal experiments were approved by the Institutional Animal Care and Use Committee at Weill Cornell Medicine.

Isolation of mouse intestinal lamina propria cells

Mouse intestines were removed, cleaned from remaining fat tissue, and washed in ice-cold phosphate-buffered saline (PBS) (Corning). Intestines were opened longitudinally, washed in ice-cold PBS, and cut into small pieces (~2mm). Dissociation of epithelial cells was performed by incubation on a shaker in Hanks’ balanced salt solution (Sigma-Aldrich) containing 5 mM EDTA (Thermo Fisher Scientific), 1 mM dithiothreitol (DTT) (Sigma-Aldrich), and 2% heat-inactivated fetal bovine serum (FBS) for 15 min each at 37°C in a glass beaker with stirring. The dissociated cells were filtered through a 70um cell strainer and washed one time prior to enzymatic digestion in digestion buffer containing dispase (0.4 U/ml; Thermo Fisher Scientific), collagenase III (1 mg/ml; Worthington), and deoxyribonuclease (DNase) I (20 μg/ml; Sigma-Aldrich) in 10% FBS/Dulbecco’s modified Eagle’s medium (DMEM) (Corning) for 30 min at 37°C. Leukocytes were further enriched by a 40/75% Percoll gradient centrifugation (GE Healthcare) at 700g for 20 min.

Milk collection from mice

Milk collection from mice was performed as previously described (19). Briefly, nursing dams (P7-P16) were separated from their pups for 4–6 h to allow accumulation of milk, followed by anesthetization using 2% isoflurane (VetOne), and 2–8 units per mouse of oxytocin (Sigma) was administered intraperitoneally to induce milk flow. Samples were collected using an electric human breast pump that was modified to accommodate mice and small liquid volumes. All milk was stored at −20°C until use. For western blotting, mouse milk was diluted 1:100. For ELISA, mouse milk was diluted 1:10.

Measurement of Mucosa-Associated C. rodentium

For detection of mucosa-attached C. rodentium, WT or Fcgrt^−/− neonatal mice were infected with ∼0.5×10⁸ CFU of a LER-driven bioluminescence-expressing strain of C. rodentium (C. rodentium ler-lux). Both the cecal and colonic tissues were harvested 6 or 12 days after infection. To remove non-adherent C. rodentium, luminal contents in the cecum and colon were flushed with PBS and further washed by with sterile PBS twice. Washed tissues were immediately placed into the light-tight chamber of a CCD in vivo imaging system (IVIS 200, PerkinElmer) and bioluminescence in the tissue was captured using the software Living Image (PerkinElmer), as previously described (25). Both control and treatment groups were analyzed at the same time and using the same software settings.

ELISA for bacteria-specific antibodies

For measurement of commensal bacteria or C. rodentium-specific IgG, IgM, and IgA, fecal bacteria were isolated from naive WT SPF mice (8–10 weeks old) used to coat ELISA plates as described in our previous studies. Fecal bacteria from ∼200 mg fecal pellets were isolated by homogenization in sterile PBS, filtered through a 40 μm cell strainer, and separated from debris/mouse cells by removing the pellet after centrifugation at 1,000 rpm for 5 min. Isolated fecal bacteria were washed twice, heat-killed at 85°C for 1 hr, and resuspended in 10 ml coating buffer, and100 μL was added to each well of a 96-well ELISA plate for overnight coating at 4°C. Mouse sera were diluted at 1:20 and 1:100 and incubated overnight at 4°C for detection of IgG, IgM, and IgA. Anti-mouse IgG and HRP substrate were purchased from Bethyl Laboratories. Anti-mouse IgA and IgM were from Southern Biotech.

Flow cytometry on lamina propria cells

Lamina propria cells from mouse intestines were isolated as previously described, followed by stimulation for 4 hours ex vivo with phorbol 12-myristate 13-acetate (PMA; 100 ng/ml; Sigma-Aldrich) and ionomycin (1 μg/ml; Sigma-Aldrich) in the presence of brefeldin A for the last hours (10 μg/ml; Sigma-Aldrich) in a 37°C incubator (5% CO₂). Cells were washed with cold PBS and surface-stained before fixation and permeabilization using the Cytofix/Cytoperm Fixation/Permeabilization solution and Perm/Wash buffer according to the manufacturer’s protocol (BD Biosciences). Stained bacteria/mouse lamina propria cells were then washed with FACS buffer once, and analyzed by BD Aurora (BD Bioscience). Data were analyzed by Flow Jo (Tree Star).

Maternal immunization and infection

50 μg of recombinant outer membrane protein A (OMP-A; Creative BioMart) was mixed 1:1 (v/v) with Imject Alum (Thermo Fisher Scientific) and IP injected into female mice (WT, Fcgrt^−/−, IgG-deficient, and IgA-deficient) 2 days prior to mating. A second protein/alum injection was administered 2 weeks later (approximately one week prior to giving birth). Pups from naïve or immunized females were orally infected with 5 × 10⁸ C. rodentium and the survival of the pups was monitored. Some infected pups were euthanized at 6 days following infection to enumerate translocated C. rodentium in the spleens and livers.

Generation of IgG-deficient mice

IgG-deficient mice were generated using the CRISPR/Cas9 system. Cas9 mRNA and sgRNA were designed as described previously (42). Briefly, Cas9 mRNA and sgRNAs were microinjected into fertilized embryos of C57BL/6J mice to allow Cas9-mediated cleavage of DNA sequences about ~1kb upstream of Ighg3 and ~1kb downstream of Ighg2b (performed by the University of Michigan Transgenic Animal Model Core). Deletion of a ~80kb DNA fragment containing all IgG-encoding loci (Ighg1, Ighg2a, Ighg2b and Ighg3) was confirmed by Sanger sequencing of PCR amplicons (935 bp) of potential founders using genotyping primers The genotyping primers and the full predicted amplicon sequence are provided in the Supplementary Methods. The embryos were transferred into pseudopregnant C57BL/6 females. Confirmed founders from the offspring were additionally mated with C57BL/6 WT mice and subsequent crossing of heterozygous offspring for >10 generations. ELISA for serum IgG was performed to confirm loss of all IgG isotypes in IgG-deficient mice.

Fecal transplantation of WT and IgG KO bacteria into GF neonates

P7 WT or IgG KO mice were sacrificed and the luminal contents from the small intestine and colon were collected and homogenized in PBS (~10 μL/mg of luminal contents). P7 GF WT were given the 50 μL of the homogenate by oral gavage. After 2 weeks, the mice were sacrificed and the colons were collected for further analysis of lamina propria immune cells.

Cross-fostering of IgG KO neonates with WT dams

Postnatal day 1 (P1) IgG KO mice were given to either a nursing IgG KO dam or nursing WT dam (see schematic). After 2 weeks, P14 mice were sacrificed and their colons were collected for further analysis of lamina propria immune cells.

FITC-dextran permeability after C. rodentium infection

For in vivo infection, Chloramphenicol-resistant (Chl^R) C. rodentium was grown overnight in Luria-Bertani (LB) broth supplemented with Chl (10 μg/ml) with shaking at 37°C. P14 WT or IgG KO mice were infected by oral gavage with 100 μL of PBS containing 5 × 10⁷ CFU of C. rodentium. Three days post infection, the mice were fasted for 4 hours. Mice were then given FITC-dextran (500 mg/kg body weight dissolved in PBS, Sigma Aldrich) via oral gavage. After 1 hour, blood was collected and fluorescence intensity in the serum was determined using a spectrophotometer (excitation 485 nm; emission 525 nm).