All animals were bred, raised, and housed in the facilities of the “Forschungseinrichtungen für Experimentelle Medizin” (Charité – University Medicine Berlin, Germany) under specific-pathogen-free (SPF) conditions. Female age-matched C57BL/6j wild-type mice were used.

In order to virtually deplete the intestinal microbiota, 8–10 weeks old mice were transferred to sterile cages and subjected to quintuple antibiotic treatment for 8 weeks as previously described (24). Three days prior to peroral FMT, the antibiotic cocktail was withdrawn and replaced by sterile drinking water. Successful eradication of the cultivable intestinal microbiota was confirmed as described previously (24). Fresh murine fecal samples were collected from 10 age- and sex-matched SPF control mice, pooled, dissolved in 10 ml sterile phosphate-buffered saline (PBS; Gibco, Life Technologies, Paisley, UK) and the supernatant perorally applied by gavage (in 0.3 ml PBS) in order to reconstitute secondary abiotic (i.e., gnotobiotic) mice with a complex intestinal microbiota.

Mice were sacrificed by isoflurane treatment (Abbott, Greifswald, Germany) at day 7 or day 28 post-FMT. Luminal large intestinal samples as well as ex vivo biopsies from spleen, MLN, ileum, and colon were taken under sterile conditions. Ileal and colonic ex vivo biopsies were collected from each mouse in parallel for immunological, microbiological, and immunohistochemical analysis. For immunohistochemical stainings, ileum and colon samples were immediately fixed in 5% formalin and embedded in paraffin, and sections (5 µm) were stained with distinct antibodies as described below.

In situ immunohistochemical analysis of ileal and colonic paraffin sections was performed as previously described (25–28). Primary antibodies against cleaved caspase-3 (Asp175, Cell Signaling, Beverly, MA, USA, 1:200), Ki67 (TEC3, Dako, Glostrup, Denmark, 1:100), CD3 (#N1580, Dako, 1:10), Foxp3 (FJK-16s, eBioscience, San Diego, CA, USA, 1:100), B220 (eBioscience, 1:200), and F4/80 (# 14-4801, clone BM8, eBioscience, 1:50) were used. For each animal, the average number of positively stained cells within at least six high power fields (HPF, 400× magnification) was determined microscopically by an independent blinded investigator.

Single cell suspensions were generated from spleens and MLN, and erythrocytes were removed from splenic samples by 1.66% ammonium chloride. All samples were resuspended in defined volumes of PBS/0.5% bovine serum albumin (BSA) and subjected to further processing (23).

Segments of the murine gut were removed and freed from fat, connective tissue, and PP, cut longitudinally, and cleared from luminal content and mucus with ice-cold PBS. The isolation of lamina propria lymphocytes (LPL) followed a standard protocol with minor modifications (29). Briefly, the intestines were cut into 0.5 cm pieces and incubated twice with 25 ml Hanks’ balanced salt solution (HBSS; Gibco) containing 1 mM dithioerythritol (Carl Roth) for 20 min at 37°C and 220 rpm. Afterward, the intestines were introduced to HBSS containing 1.3 mM ethylenediaminetetraacetic acid (Life Technologies, Eugene, OR, USA). Subsequently the cells were placed in digestion solution, containing 0.5 mg/ml collagenase A (Roche, Mannheim, Germany), 0.5 mg/ml DNAse I (Roche), 10% fetal calf serum (FCS), and 1 mM of each CaCl₂ and MgCl₂ (both Carl Roth). Digestion was performed through incubation for 45 min at 37°C and 220 rpm. After the incubation, the digested tissues were washed with RPMI containing 5% FCS, resuspended in 5 ml 44% Percoll (GE Healthcare, Uppsala Sweden), and overlaid on 5 ml 67% Percoll in a 15 ml Falcon tube. Percoll gradient separation was performed by centrifugation at 600 g for 20 min at room temperature. LPL were collected from the interphase, washed once, and suspended in PBS/0.5% BSA.

Surface staining was performed using following antibodies: FITC-anti-CD4 (Clone RM4-5; 1:200), PerCP-anti-CD8 (Clone 53-6.7; 1:100), PacBlue-anti-B220 (Clone RA3-6B2, 1:200), APC-Cy7-anti-CD25 (Clone PC61, 1:200), PE-anti-CD44 (Clone IM7, 1:200), and APC-anti-CD86 (Clone B7-2, 1:200) (all from BD Biosciences, San Jose, CA, USA).

For intracellular staining, cells were restimulated for 5 h with 10 ng/ml phorbol myristate acetate and 1 µg/ml ionomycin, in a tissue culture incubator at 37°C (both Sigma-Aldrich). Ten micrograms per milliliter brefeldin A (Sigma-Aldrich) were added to the cell suspensions after 1 h of polyclonal restimulation. Then cells were treated with LIVE/DEAD Fixable Aqua Dead Cell Stain kit (Life Technologies) and hereafter fixed with 2% paraformaldehyde (Sigma-Aldrich) for 20 min at room temperature. Cells were stained in 0.5% saponin (Sigma-Aldrich) using following antibodies: PacBlue-anti-CD4 (Clone RM4-5; 1:400), PE-Cy7-anti-IFN-γ (Clone XMG 1.2; 1:400) (both from BD Biosciences), FITC-anti-IL17A (Clone TC11-18H10.1; 1:200, BioLegend, San Diego, CA, USA), PE-anti-IL10 (Clone JESS-16E3; 1:100), and APC-anti-IL22 (Clone IL22JOP; 1:100) (both from eBioscience). All data were acquired on a MACSQuant analyzer (Miltenyi Biotec, Bergisch Gladbach, Germany) and were analyzed with FlowJo Software v10.1 (Tree star, Ashland, OR, USA).

Expression levels of pro- and anti-inflammatory cytokines including IFN-γ, IL-22, IL-17A, and IL-10 mRNA were determined in snap frozen ileal and colonic ex vivo biopsies using Light Cycler Data Analysis Software (Roche) as stated elsewhere (30). The mRNA of the housekeeping gene for hypoxanthine-phosphoribosyltransferase was used as reference; the mRNA expression levels of the individual genes were normalized to the lowest measured value and expressed as fold expression (arbitrary units) (31).

For molecular analysis of the intestinal microbiota, DNA was extracted from fecal samples as described previously (24). Briefly, DNA extracts and plasmids were quantified using Quant-iT PicoGreen reagent (Invitrogen, Paisley, UK) and adjusted to 1 ng/µl. Then, abundance of the main bacterial groups of murine intestinal microbiota was assessed by quantitative real time-PCR with group-specific 16S rRNA gene primers (Tib MolBiol, Berlin, Germany) as described previously (5, 32, 33). The number of 16S rRNA gene copies per microgram DNA of each sample was determined, and frequencies of respective bacterial groups calculated proportionally to the eubacterial (V3) amplicon.

Medians, means, SDs, and significance levels were determined using Mann–Whitney U test or one-way analysis of variance with Tukey’s post hoc test for multiple comparisons (GraphPad Prism Software v6, La Jolla, CA, USA) as indicated. Two-sided p values ≤ 0.05 were considered significant. Data shown were pooled from two independent experiments (n = 10–15 per group).

To confirm successful depletion of the intestinal microbiota, we applied cultural analyses of fecal samples derived from ABx mice. In fact, all fecal samples were culture negative for aerobic, microaerobic, and obligate anaerobic species as assessed by direct plating and enrichment procedures (not shown). To additionally assess abundance of fastidious and uncultivable intestinal bacteria, we next determined the main bacterial groups abundant in the murine intestinal tract by quantitative 16S rRNA-based PCR analysis of fecal samples derived from conventionally colonized and ABx mice as compared to respective bacterial groups abundant in autoclaved food pellets. In ABx mice, bacterial 16S rRNA gene numbers were decreased by up to 10 orders of magnitude as compared to conventional SPF controls (p < 0.001; Figure 1). Remarkably, mean 16S rRNA gene numbers in fecal samples derived from ABx mice and autoclaved food pellets were comparable, indicating a successful and biologically relevant depletion of the intestinal microbiota following broad-spectrum antibiotic treatment. Notably, one cannot differentiate whether detected 16S rRNA gene numbers in ABx mice were derived from avital (“dead”) or viable fastidious/uncultivable bacterial cells. We next determined the efficiency of intestinal microbiota reconstitution upon FMT of ABx mice. As assessed by molecular methods, 16S rRNA gene numbers of the main bacterial intestinal microbiota groups were comparable in conventional mice and ABx mice at days 7 and 28 post-FMT (Figure S1 in Supplementary Material) indicating a successful reconstitution of the intestinal microbiota by FMT.

Given that neither antibiotic treatment nor FMT affected mice clinically and resulted in macroscopic sequelae such as wasting, diarrhea, or occurrence of blood in fecal samples (not shown), we assessed potential microscopic changes in intestinal ex vivo biopsies derived from ABx and FMT mice. To address this, we determined numbers of apoptotic cells in small and large intestinal paraffin sections following staining against caspase-3, given that apoptosis is an established parameter used for histopathological evaluation and grading of intestinal inflammation (5). Neither small nor large intestinal epithelial apoptotic cell numbers differed in conventionally colonized, ABx, and reconstituted mice at days 7 and 28 post-FMT (Figure 2A). Interestingly, quantification of Ki67 expression cells as a sensitive measure for cell proliferation and regeneration (34) revealed reduced numbers of proliferating cells in both ileal and colonic epithelia of ABx mice (p < 0.001; Figure 2B). As early as 7 days post-FMT, however, Ki67⁺ cell numbers reached basal counts again (p < 0.001 vs ABx; Figure 2B). Hence, neither broad-spectrum antibiotic treatment nor FMT results in increased intestinal apoptosis, whereas reconstitution with complex intestinal microbiota is essential for restoring cell proliferative and regenerative measures during physiological tissue turnover within the intestinal tract.

To examine the impact of the intestinal microbiota on abundances of distinct immune cell populations in the small and large intestines, we microscopically quantitated respective immune cell subsets in small intestinal and colonic paraffin sections applying in situ immunohistochemistry. In microbiota-depleted mice, significantly reduced numbers of CD3⁺ T lymphocytes (p < 0.001; Figures 3A,E), B220⁺ B lymphocytes (p < 0.001; Figures 3B,F), Foxp3⁺ regulatory T cells (Treg, p < 0.001; Figures 3C,G) as well as of F4/80⁺ monocytes and macrophages (p < 0.01–0.001; Figures 3D,H) in both ileum and colon as compared to conventional mice could be observed. Following FMT colonic, but not ileal numbers of respective immune cell populations increased back to counts observed in control mice as early as 7 days post-FMT (p < 0.01–0.001 vs ABx, Figures 3E,H). In small intestines, however, T and B lymphocytes as well as Treg numbers were lower at day 7 post-FMT as compared to SPF mice but reached comparable or even higher counts thereafter (Figures 3A,C). Hence, our data underline the essential association of the complex commensal microbiota and the repertoire of innate and adaptive immune cell populations in both the small and large intestines.

To further elaborate the role of the intestinal microbiota on adaptive immunity in mucosal, peripheral, and systemic compartments, we isolated lymphocytes of the small and large intestinal LP, MLN, and spleen and analyzed defined immune cell populations by flow-cytometric analysis. Gating strategies are depicted in Figures S2A–F in Supplementary Material.

Broad-spectrum antibiotic treatment induced a significant reduction of both relative abundances and absolute numbers of CD4⁺ helper T lymphocytes in the small and large intestinal LP that could be restored at day 7 post-FMT, whereas colonic CD4⁺ cell concentrations further declined thereafter (p < 0.01–0.001; Figures 4A–D). Interestingly, within the MLN, but not other compartments, decreased percentages of CD4⁺ cells could be observed at day 7 post-FMT (p < 0.05; Figures 4E,F). In the spleen, frequencies of CD4⁺ cells were not affected by microbiota depletion (n.s.; Figure 4G). Notably, increased splenic CD4⁺ cell numbers were determined in ABx mice that slightly declined until day 7 post-FMT but increased to even supra-basal levels thereafter (p < 0.01; Figure 4H). These results point toward a potential “systemic accumulation” of these cells in the splenic compartment in absence of the intestinal microbiota.

In the small intestinal, LP decreased frequencies and cell numbers of CD8⁺ cytotoxic T cells were observed following microbiota depletion (p < 0.01–0.001; Figures 5A,B) that could be completely restored upon FMT. In the colonic LP, CD8⁺ cell frequencies were reduced upon antibiotic treatment and reestablished rather late following microbiota reconstitution (i.e., until day 28 post-FMT) (p < 0.001, d28 vs ABx; Figure 5C). Notably, colonic CD8⁺ cell numbers were profoundly affected by antibiotic treatment of mice, irrespective whether recolonized or not as indicated by lower counts as compared to SPF controls (p < 0.05–0.001; Figure 5D). In line with CD4⁺ cell numbers, CD8⁺ cells within MLN were not affected by antibiotic treatment and/or subsequent bacterial recolonization (n.s.; Figures 5E,F).

Relative abundances as well as absolute numbers of splenic CD8⁺ cells increased following antibiotic treatment (p < 0.05–0.001; Figures 5G,H), decreased as early as 7 days post-FMT but increased again 28 days after FMT to higher levels than in SPF mice.

We next addressed whether also B lymphocytes were affected following antibiotic treatment and restoration of the intestinal microbiota. In fact, also absolute numbers of B220⁺ cells decreased in small intestines, colon, and MLN upon microbiota depletion, but conversely increased in the spleen. FMT could sufficiently restore small intestinal, but not colonic B220⁺ cell counts as early as 7 days thereafter (Figures S3B,D in Supplementary Material). Again, microbiota depletion-induced elevation of splenic B220⁺ cells was reversed until day 7 post-FMT (p < 0.05; Figure S3H in Supplementary Material). Interestingly, proportions of B220⁺ B lymphocytes within small intestine, colon, and spleen were rather unaffected by the colonization status of mice (Figures S3A,C,G in Supplementary Material), whereas an increased proportion of B cells could be observed in MLN at day 7 post-FMT as compared to the other groups (p < 0.001; Figure S3E in Supplementary Material).

Taken together, these data point out that a virtual eradication of the intestinal microbiota by broad-spectrum antibiotic treatment affects both absolute numbers and relative abundances of distinct adaptive immune cell population not only locally (i.e., in the intestinal tract) but also has far-reaching consequences on systemic immune functions.

We next surveyed the impact of broad-spectrum antibiotic therapy and FMT on defined T cell subsets and on the activation status of distinct cell populations. To address this, the surface marker CD44 that is expressed upon antigen contact (35) on CD4⁺ and CD8⁺ cells was analyzed. Upon broad-spectrum antibiotic treatment, both CD44 positive CD4⁺ and CD8⁺ memory/effector cells were less abundant in all surveyed lymphoid compartments (p < 0.01–0.001; Figure 6). This effect could, however, be reversed upon FMT as indicated by higher abundances of CD44 positive CD4⁺ and CD8⁺ cells at days 7 and 28 post-FMT as compared to ABx mice (p < 0.05–0.001; Figure 6). At day 7 post-FMT, CD8⁺CD44⁺ in the MLN were even more abundant than in SPF mice but reached basal levels thereafter (p < 0.05; Figure 6F).

As already observed for the other immune cell subsets, also CD4⁺CD25⁺ Treg were strongly diminished in small and large intestines, MLN, and spleen of ABx mice (p < 0.01–0.001; Figures S4A,C,E,G in Supplementary Material), but this reduction could be reversed as early as 7 days following FMT (p < 0.001; Figures S4A,C,E,G in Supplementary Material). Interestingly, in all analyzed immunological compartments, Treg numbers were even higher at day 7 post-FMT as compared to naive mice but declined back to basal levels thereafter (p < 0.01–0.001; Figures S4A,C,E,G in Supplementary Material).

As for Treg, a strong downregulation of CD86 expression, a costimulatory molecule marking activated DC (36), could be observed in the small intestine, colon, MLN, and spleen upon broad-spectrum antibiotic treatment (p < 0.001; Figures S4B,D,F,H in Supplementary Material). Within 7 days following FMT, however, CD86⁺ cells reached basal levels again (Figures S4B,D,F,H in Supplementary Material). As already described for Treg, activated DC were even more abundant in spleens at day 7 post-FMT as compared to naive mice (p < 0.001; Figures S4H in Supplementary Material) but declined to basal levels thereafter.

Hence, memory/effector T cells, Treg, and activated DC are highly microbiota dependent, are disturbed following broad-spectrum antibiotic treatment, but can be restored upon FMT.

We further addressed whether microbiota depletion and subsequent reconstitution by FMT affected pro- and anti-inflammatory cytokine expression of CD4⁺ cells in intestinal and systemic lymphoid compartments by performing intracellular cytokine stainings. Gating strategies are depicted in Figures S2G–I in Supplementary Material. Decreased IFN-γ-expressing CD4⁺ cells were detected in small and large intestines of ABx mice but increased back to naive levels until day 7 following microbiota reconstitution (p < 0.001; Figures 7A,B). Of note, at day 7, but not day 28 following FMT, IFN-γ-expressing CD4⁺ cells were more prominent in MLN as compared to naive controls (p < 0.05; Figure 7C). Splenic CD4⁺IFN-γ⁺ cells, however, were virtually unaffected by antibiotic treatment and bacterial reconstitution (n.s.; Figure 7D). Notably, CD4⁺ cells expressing IL-17 and IL-22 were downregulated in small and large intestines, in MLN as well as in the spleen upon broad-spectrum antibiotic treatment but could be reversed as early as 7 days following FMT (p < 0.01–0.001; Figure 8). In addition, CD4⁺IL17⁺ cells in the MLN and spleen as well as small and large intestinal CD4⁺ IL-22⁺ cells were even higher at day 7 following FMT as compared to naive animals (p < 0.05–0.001; Figures 8B,D,E,G) but declined back to basal levels thereafter (p < 0.01–0.001). Furthermore, a strong reduction of CD4⁺ lymphocytes producing the anti-inflammatory cytokine IL-10 could be determined in all immunological sites following antibiotic therapy (p < 0.001; Figure 9). FMT could completely restore this cell population in all examined lymphoid compartments. Importantly, higher levels of IL-10 productions were observed in the small intestinal LP and MLN of mice at day 7 post-FMT than in their naive counterparts (p < 0.01–0.001; Figures 9A,C).

These findings were further supported by results obtained from mRNA analysis of respective cytokines measured in ileal and colonic ex vivo biopsies (Figure S5 in Supplementary Material). IFN-γ expression was downregulated upon antibiotic treatment in the small intestine only, an effect that was fully reversed until day 28 following FMT (Figure S5A in Supplementary Material). Moreover, antibiotic treatment resulted in strong suppression of both IL-17 and IL-22 mRNA expression. In the colon, expression of respective cytokines fully recovered as early as 7 days post-microbiota reconstitution, whereas small intestinal IL-17 mRNA levels measured at day 28 following FMT were approximately 300 times higher than in ABx mice, but still significantly lower as compared to naive SPF control mice (Figures S5B,C in Supplementary Material). In recolonized mice, a trend toward higher small intestinal expression of IL-22 mRNA could be observed (n.s.; Figures S5B,C in Supplementary Material). Furthermore ABx mice also displayed lower levels of IL-10 mRNA expression in both small and large intestines than their conventional counterparts (Figure S5D in Supplementary Material). At day 7 post-FMT, a strong IL-10 response could be only observed in the colon, whereas 28 days after microbiota reassociation, IL-10 expression reached naive levels in both compartments (Figure S5D in Supplementary Material). Hence, long-term broad-spectrum antibiotic treatment leads to suppression of both pro- and anti-inflammatory cytokines. These effects can, however, almost fully be reversed by recolonization with complex murine intestinal microbiota.

In the present study, we have focused on the effects of a complex murine microbiota on the immune system following antibiotics induced impairments not only of the intestinal ecosystem but also of peripheral as well as systemic immune functions. Whether the here displayed beneficial restoring effects exerted by reintroduced microbial antigens are due to the large bacterial loads, complexity and/or diversity of the introduced complex microbiota, or whether distinct species in concert with each other play a more important role in an orchestrated fashion, with the host immune system as the conductor, should be unraveled in more detail, but appears literally rather as a search for the needle in the hay stack. Nevertheless, it remains an outstanding and challenging issue to characterize the effects of single species and their products on the balance between pro-inflammatory and regulatory immune responses.

We are all aware of the fact that a rational and responsible antibiotic treatment is unavoidable under specific clinical conditions, but it is crucial to keep the effects of this therapy on the immune system in mind. These effects might be due to potential immune-modulating properties of the antimicrobial compound itself and/or due to microbiota-modulating (-depleting) sequelae of therapy or prophylaxis. Further knowledge of the orchestrated microbiota–host interplay could offer valuable contributions to the development of novel therapeutic approaches including strategies to enhance immunity and manipulating microbiota composition toward more beneficial (i.e., probiotic) species.