Kinetic and mechanistic diversity of intestinal immune homeostasis characterized by rapid removal of gut bacteria

ABSTRACT

Symbiotic microbiota critically contribute to host immune homeostasis in effector cell-specific manner. For exclusion of microbial component, germ-free animals have been the gold standard method. However, total removal of the entire gut microbiota of an animal from birth significantly skews physiological development. On the other hand, removal of gut microbiota from conventional mice using oral antibiotics has its own limitations, especially lack of consistency and the requirement for long-term treatment period. Here, we introduce an improved regimen to quickly remove gut microbiota and to maintain sterility, that is well received by animals without refusal. Rapid and consistent exclusion of resident bacteria in the gut lumen revealed kinetic differences among colonic lymphocyte subsets, which cannot be observed with typical germ-free animal models. Furthermore, the proposed method distinguished the mechanism of microbiota contribution as a direct stimulus to capable effector cells and a homeostatic cue to maintain such cell types.

KEYWORDS:

• Germ-free mice
• antibiotic treatment
• mucosal immunology
• gut microbiota

Introduction

Symbiotic bacteria are critical in shaping host immunity. Germ-free animals, from which live bacteria are totally absent from birth, show dramatic differences in both their innate and adaptive immune systems, as well as their responses to infection and inflammatory stimuli¹. Although such animals have been the gold standard for investigating host–microbiota interactions, total exclusion of trillions of bacteria from birth also causes significant developmental and physiological discrepancy², which may hinder the dissection of immunological phenotypes by comparison with conventionally colonized animals. Conventionalization of germ-free animals can normalize some, but not all, host responses, because many phenotypes such as developmental or immunological are age-specific³. To delineate and normalize time-dependent impacts, it is necessary to eliminate symbiotic microbiota for only a transient period. In addition, for some in vivo studies, it is often impractical to re-derive and maintain multiple (usually genetically manipulated) mouse strains on a germ-free background; therefore, alternative strategies would be necessary to study the role of gut bacteria in those mice.

As a practical alternative to germ-free animals, deliberate removal of oro-gastrointestinal bacteria, which comprise the vast majority of commensal microbiota by cell mass, has been widely used⁴. Most commonly, animals are treated with permutations of orally deliverable antibiotics (for example, ampicillin, neomycin, metronidazole, and vancomycin⁵ dissolved in their drinking water). Although the procedure is simple and takes little labor, ad libitum administration yields dramatic differences in compliance among individual subjects. Several antibiotics (especially metronidazole, which is critical for removing several anaerobic species) have an unpleasant taste, and many animals refuse to drink water containing them, which can cause severe dehydration and death⁶. Not only does this cause serious health issues for the animals that are hard to mitigate, although oral or subcutaneous supplementation of fluid can let animals minimize drinking foul-tasting water and still survive, but it also results in unintentional selection of subject animals by behavioral adaptation, which could skew in vivo investigations in unpredictable ways. In addition, removal of gut bacteria is usually confirmed after a prolonged treatment period, usually 14 to 28 d. This delay renders time-dependent host responses that occur immediately after removal of bacteria unobservable. To address this issue, several protocols have been developed by combining drinking water-dissolved and orally gavaged antibiotics^6,7. Nonetheless, these approaches require frequent manipulations, hence not easily scalable to larger and longer experiments. Finally, on some occasions, resistant species are selected from non-continuous (given as bolus dose) antibiotic administration and ‘re-contaminate’ animals during later experimental stages.

To address these compliance and health issues, as well as to develop a consistently successful method of rapidly removing gut microbiota, we used metronidazole benzoate, which is a chemical derivative form of metronidazole and has been clinically used for pet animals with good compliance and gentamicin to replace neomycin in some cases, which is more efficient in preventing the development of resistant bacteria. This combination of vancomycin (V), gentamicin/neomycin (G/N), metronidazole benzoate (Mb), and ampicillin (A) (VGMbA/VNMbA) was tested in vivo for both the efficiency of removal as well as host compliance and survival. VGMbA treatment was further assessed to dissect different time courses of gut immune cell responses, as well as to delineate host responses in which commensal bacteria contribute as direct effectors of the immune response or as maintenance signals of gut immune homeostasis.

Results

Metronidazole benzoate-containing antibiotic cocktail treatment does not affect mouse health and rapidly and efficiently removes gut bacteria

To assess whether a modified antibiotic mixture treatment removes gut bacteria without affecting health, we treated the mixture to specific pathogen-free (SPF) C57BL/6 (B6) mice (from Taconic Biosciences, Tac B6). The average daily VGMbA water consumption during the treatment was comparable to that of the control (Extended Figure S1A). VGMbA cocktail treatment led to a slight loss of body weight 1 and 2 d after treatment but it was recovered quickly (Figure 1(a)). In parallel, VNMbA cocktail was used for pre-conditioning for large-scale animal study (Extended Table 1), showing 7 d of antibiotic cocktail treatment containing metronidazole benzoate in drinking water did not cause weight loss or show sign of dehydration.

Figure 1. VGMbA treatment does not affect mice health and rapidly and efficiently removes gut bacteria.

Note: Antibiotic cocktails in drinking water were given to C57BL/6 (B6) mice from Taconic (Tac). a) Percent of body weight changes after antibiotic treatment. Vancomycin (V), Gentamicin (G), Metronidazole benzoate (Mb), and Ampicillin (A). b) Cultivable total bacterial CFU from feces were enumerated. (§: estimated counts from lawn plates) c) The levels of short chain fatty acids from feces were measured by LC/MS/MS. N = 5 for each group.

We observed that VGMbA treatment very rapidly cleared gut bacteria by counting cultivable bacteria in feces (Figure 1b). Although treatment without Mb cleared bacteria in a fashion similar to VGMbA, Mb was indispensable to prevent outgrowth of resistant bacteria after 7 d, as confirmed by live bacterial CFU (Figure 1b), bacterial DNA from feces quantified with qPCR (Extended Figure S1B), and the host IgA response (Extended Figure S1C). Rapid clearance of gut bacteria was confirmed in SPF animals from different commercial vendors (Jackson Laboratory C57BL/6: JAX B6, Extended Figure S2A) without fungal overgrowth in the gut (Extended Figure S2B). Along with the removal of gut microbiota, short chain fatty acid (SCFA) level in the gut lumen was also used as a surrogate readout of bacterial metabolic activity. Acetate, propionate, and butyrate levels in the gut lumen rapidly fell, coinciding with the time course of bacterial burden decrease (Figure 1c). These results collectively imply that VGMbA in drinking water can be used as a universal method to rapidly achieve quasi germ-free status in commercially available SPF mice.

Rapid removal of microbiota distinguishes different kinetics in maintenance of colonic T cell subsets.

Next, we investigated how the loss of gut microbiota affects the homeostasis of colonic T cells in a time-dependent manner. VGMbA treatment could suppress fecal microbiota by the factor of 10⁶ up to 28 d (Extended Figure S3A). During the time course, we analyzed the frequencies of colonic Foxp3+ regulatory T cells (Tregs), RORγt+ Tregs, Th17 cells, and NKT cells (Figure 2a–d and Extended Figure S3b–f)^8,9. Of note, Foxp3+ Tregs and Th17 cells responded very quickly to gut bacterial loss and began to decrease in number within 7 d. In contrast, RORγt+ Tregs were maintained longer and started to decrease at 14 d. On the other hand, colonic NKT cell levels remained unchanged, confirming previous findings that bacterial signals affect NKT cell proliferation only in early life^10,11. Through rapid and efficient bacterial removal, kinetic differences in homeostatic maintenance of individual T cell subsets in response to bacterial signals were revealed.

Figure 2. Rapid removal of microbiota shows different kinetics of colonic immune cells.

Note: a-d) B6 mice from Jackson (JAX) were treated with VGMbA for the indicated time periods and colonic lamina propria cells from SPF, VGMbA-treated SPF and GF mice were prepared to analyze colonic immune cells. The frequencies of colonic a) Foxp3+ Tregs, b) RORγt+ Tregs, c) Th17 cells and d) NKT cells were indicated. Each dot is an individual mouse. Data are representative of three independent experiments.

Bacterial signals can function as direct stimuli, as well as homeostatic cues for host immunity.

In addition to diverse peripheral T cells, immunoglobulin A (IgA) plays an important role in intestinal homeostasis by preventing intestinal bacteria from crossing the epithelium¹². Gut luminal IgA is much lower in germ-free mice so is IgA-producing plasma cell number in lamina propria¹³. Of interest, VGMbA treatment immediately (as early as 1 d after treatment) decreased gut luminal IgA (Extended Figure S1C). In comparison, IgA-producing plasma cell (IgA-PC) levels changed much more slowly, decreasing only after 14 d or later (Figure 3a and Extended Figure S4). Such discrepancies imply the dual roles of microbiota, not only as a direct stimulus for capable PCs to secrete IgA, but also as a component necessary for the development and maintenance of PC¹⁴. These two different roles of gut microbiota could be further delineated by using hosts given short- (3 d) or long-term (14 d) VGMbA (Figure 3b). Loss of gut luminal IgA resulting from short-term removal of gut microbiota can be rapidly recovered by reintroduction of microbiota (Figure 3c); in contrast, loss of tissue-resident IgA-PCs due to long-term removal of microbiota causes delays in recovering luminal IgA production (Figure 3d), which is comparable to the time-course of germ-free animals associated with bacteria for the first time in life¹⁵.

Figure 3. Long-term antibiotic treatment delays IgA responses against re-introduction of gut bacteria.

Note: a) Mice were treated with VGMbA for the indicated time periods, and frequency of IgA-producing plasma cells from colonic lamina propria was analyzed. b) Experimental scheme of VGMbA treatment and fecal transfer. c-d) Fecal IgA levels after re-introduction of gut bacteria after 3 d (c) or 14 d (d) VGMbA treatment. Each dot is an individual mouse. Data are representative of at least two independent experiments.

Discussion

Even before the modern concept of symbiotic microbiota-mediated host development was established, commensal microbes are considered as a critical factor for host survival¹⁵. Although Louis Pasteur’s argument may have not been entirely correct, physiological and immunological development of germ-free mammals differs significantly from that in animals reared in the presence of conventional microbiota. By removing gut microbiota in a rapid and consistent manner, we were able to characterize the time course of host immune responses to the removal of gut microbiota. In two extremes, some immune phenotypes, such as IgA secretion, immediately follow the decrease in microbial burden, whereas others, such as colonic NKT cells, are not affected. Many conventional T cell subsets fall in between, showing different kinetics in homeostatic control.

With regard to delineating the kinetics of host immune responses, our approach could distinguish the impact of microbiota as a direct stimulus on effector cells from the homeostatic cue for immune homeostasis^16,17. As seen in IgA-producing PCs, short-term removal of microbes preserves the host effector cell population, which can immediately respond to the re-introduction of microbes by secreting IgA. In contrast, the prolonged absence of microbiota (14 d) turns off the host homeostatic control by losing the ability to maintain effector cells. Therefore, the latter case causes delayed IgA responses, since the host first needs to replenish effector cells in the niche.

Limitation of study

Proposed methods, or antibiotic treatment to deplete gut microbiota in general, have two major limitations. Firstly, antibiotic treatment can only confirm that the colonic contents have less than certain level of culturable bacteria in it. Trace amounts of resistant bacteria or non-bacterial microbes still may exist, hence additional quantification of bacterial and fungal DNAs might be necessary as shown. Secondly, antibiotic treatment itself can affect host physiology in microbiota-independent manner. The method presented here, which preserves normal development in the host-microbiota context and rapidly removes bacteria at the desired animal age, can reduce antibiotic impact to host. The results also suggest that antibiotic treatment is not just as a ‘poor scientists’ alternative’ approach to germ-free animals but can be a reliable and complementary approach, capable of contributing to investigations on temporal dissection of the host–microbiota interactions.

Materials and methods

Mice

All animal experiments were performed in accordance with the guidelines of Harvard Medical School and Brigham Women’s Hospital. Specific pathogen-free (SPF) 4–6 weeks old C57BL/6 mice were obtained from Taconic Biosciences and Jackson Laboratory. Germ-free (GF) C57BL/6 mice were bred and housed in inflatable plastic isolators. Mice were treated with 0.5 g/L vancomycin, 0.5 g/L gentamicin or neomycin, 0.25 g/L metronidazole benzoate, and 1 g/L ampicillin dissolved in drinking water for indicated time periods. Due to the insolubility of metronidazole benzoate in water, 10% of metronidazole benzoate solution in DMSO was prepared and added to the VGA or VNA-predissolved solution by dropper with vigorous stirring. Drinking water bottle was changed twice per week, as well as water intake (water levels dropping every time we changed the water) and body condition was inspected. During treatment, both antibiotic-treated groups and control groups were housed in autoclaved cages. Autoclaved diet (LabDiet 5K67) and water were given ad libitum.

Bacterial CFU enumeration

Fecal pellets collected from SPF and antibiotic-treated mice were resuspended in sterile PBS, and the suspension was further diluted in PBS as needed and used for plating. The suspension was plated onto Trypticase Soy Agar (TSA) plates with 5% sheep blood (BD Biosciences and Thermo Scientific) for aerobic bacteria culture and incubated under aerobic conditions at 37°C. Brucella agar plates with 5% sheep blood, Hemin, and Vitamin K1 (BD Biosciences and Thermo Scientific) were used for anaerobic bacteria culture and were incubated under anaerobic conditions at 37°C. Colonies were counted, and bacterial CFU were enumerated.

Bacterial and fungal DNA quantification

Bacterial DNA and fungal DNA from fecal pellets were extracted per the manufacturer’s instructions using DNA QIAamp PowerFecal Pro DNA Kit (Qiagen) and Quick-DNA Fungal/Bacterial Miniprep Kit (Zymo Research), respectively. To quantify bacterial DNA and fungal DNA, 16s rRNA gene and 18s rDNA gene were amplified by qPCR with a set of primers as follows: bacterial 16s rRNA (Fwd-515f 5’ GTGYCAGCMGCCGCGGTAA 3’ and Rev-806 r 5’ GGACTACNVGGGTWTCTAAT 3’)¹⁸ and fungal 18s rDNA (Fwd-UF1 5’ CGAATCGCATGGCCTTG 3’ and Rev-EU1 5’ TTCTCAGGCTCCCTCTCC 3’)¹⁹. qPCR was performed on the CFX-96 Real Time System (Bio-Rad) using SsoAdvanced Universal SYBR Green Supermix and iTaq™ Universal SYBR® Green One-Step Kit (Bio-Rad).

Colonic lamina propria cell isolation

The large intestines from SPF, GF, and antibiotic-treated SPF mice were collected, and the fat tissues were removed. The intestines were opened longitudinally, fecal contents were removed, and the intestines were cut into 1-inch pieces and shaken in HBSS containing 2 mM EDTA for 50 min at 37°C. After the removal of epithelial cells, the intestines were washed in HBSS and incubated with RPMI 1640 containing 10% FBS, 1.5 mM HEPES, 1% Penicillin/Streptomycin, collagenase type VIII (1 mg/ml), and DNase I (0.1 mg/ml) (Sigma-Aldrich) for 45 min at 37°C under constant shaking. The digested tissues were mixed with FACS buffer (PBS with 2% FBS and 1 mM EDTA), filtered twice (mesh sizes 70 and 40 μm), and used for flow cytometry.

Flow cytometry

Isolated cells from large intestinal lamina propria were used for flow cytometry analysis. For intracellular staining, surface-stained cells were fixed and permeabilized with eBioscience transcription factor staining buffer set. Cells were analyzed on an LSR II and Fortessa (BD Biosciences). Data were analyzed with FlowJo software (BD Biosciences).

Fecal IgA measurement

Fecal pellets were suspended in PBS and homogenized, and large particles were removed by centrifugation at 13,000 rpm, 4°C for 20 min. The supernatant was used for IgA ELISA (Invitrogen), which was conducted per the manufacturer’s instructions. Colorimetric reaction was measured by optical density at 450 nm by SpectraMax (Molecular Devices).

SCFA derivatization

The levels of three SCFAs (acetic acid, propionic acid, and butyric acid) were quantified using the chemical derivatization method proposed in an earlier previous study²⁰. Briefly, fecal pellets were resuspended in 75% acetonitrile containing deuterium-labeled internal standards (d₄-acetic acid, d₅-propionic acid, d₇-butyric acid). Samples were sonicated for 10 min and centrifuged at 10,000 g for 5 min. Forty microliters of the supernatant was used for derivatization. Before the derivatization, 20 μL of water was added to the supernatant to make 50% acetonitrile solution. Then, 20 μL of 200 mM 3-nitrophenylhydrazine (3NPH) hydrochloride and 20 μL of a mixed 120 mM of N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride-6% pyridine solution were added to the samples and the mixtures were incubated at 40°C for 30 min. The mixtures were incubated on ice for 1 min and diluted two-fold with the addition of 100 μL of water. Three authentic standards (acetic acid, propionic acid, and butyric acid) were also derivatized and used as calibration standards.

LC-MS and MS/MS conditions

The derivatized SCFAs were analyzed with a Thermo Scientific Vanquish UPLC and a Q Exactive Orbitrap. Chromatographic separation was performed with an Agilent Zorbax C18 column (4.6 mm × 75 mm,1.8 µm) at a flow rate of 0.6 mL/min at 40°C. The gradient profile was as follows: 25% acetonitrile/0.05% formic acid isocratic for 3 min, a linear gradient to 97.5% acetonitrile/0.05% formic acid over 5 min and hold for 6 min, back to 25% acetonitrile/0.05% formic acid over 0.1 min and hold for 5.9 min. The heated electrospray ionization parameters were as follows: sheath gas, 60 AU; auxiliary gas, 15 AU; spray voltage, 3 kV; capillary temperature, 320°C; aux gas heater temperature, 400°C. Negative ion mode MS1 full scans (R = 70,000 at 200 m/z) were acquired in the m/z range from 100 to 500. The parallel reaction monitoring (PRM) mode (R = 17,500 at 200 m/z) was used for 3NPH derivatized SCFAs and internal standards with a normalized collision energy of 35.

SCFA quantification

SCFAs were quantified using the data acquired in PRM mode. The extracted ion chromatograms (XICs) of the paired precursor (3NPH derivatized SCFAs) and fragment ions (137 m/z) were obtained using Qual Browser (Thermo Scientific Xcalibur 4.0). Peak areas of XICs were normalized by internal standard recovery and sample weight. The concentrations of SCFAs were calculated based on the standard calibration curve.

Statistical analysis

All statistical analyses were carried out with Prism software (version 9, GraphPad). Line graphs are shown as mean ± SEM. Bar graphs represent mean ± SEM, and each dot represents an individual datapoint. To determine P values for two groups as specified in each figure legend, two-tailed, unpaired Student’s t-test was used. One-way ordinary ANOVA was performed for comparisons of more than two groups with multiple comparisons. For in vivo results presented in Figure 3c,d, one-way ANOVA for the paired groups with multiple comparison was used. Statistical significance is defined in the figures as follows: *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

Acknowledgments

We thank P. Guttry for manuscript preparation, W. Zheng and T. Yanostang for technical assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19490976.2023.2201154

Additional information

Funding

The work was supported by the National Center for Complementary and Integrative Health [R01-AT010268, S.F.O.]; National Institute of Allergy and Infectious Diseases [R01-AI165987, S.F.O.]; National Research Foundation of Korea [2021R1A6A3A14039202, D-J.J.]; National Research Foundation of Korea [2021R1A6A3A14044113, J-S.Y.].