Male chimera mice were generated and validated as previously described (Ahmari et al., 2016). Control chimera mice (Control) were generated by reconstitution of WT C57BL/6J whole BM cells into sublethally irradiated age- and sex-matched C57BL/6J mice by comparable irradiation and reconstitution protocols. Following irradiation, all chimera mice were treated with 0.4% Baytril antibiotic (enrofloxacin; Bayer) in drinking water, and were given moist chow and Nutrical supplement daily during the first 3 weeks. Following this, the mice were allowed to recover for 9 weeks before fecal collections. All experimental procedures were completed in accordance with the approved protocols by the University of Florida Institute for Animal Care and Use Committee, and complied with the standards stated in the National Institutes of Health's Guide for the Care and Use of Laboratory Animals. All mice were housed in a temperature-controlled room (22°–23°C) on a 12:12 h light-dark cycle, in specific pathogen-free cages, and had access to standard mouse chow and water ad libitum.

Fecal samples were collected from C57 chimera and b1/2 ARs KO chimera mice 3 months post-irradiation and BM reconstitution, to allow for full recovery of the mice, as previously described (Ahmari et al., 2016). The analyses of microbial composition were performed, as previously described (Yang et al., 2015). Briefly, fecal DNA (N = 6/group) was extracted by using ZR fecal DNA MiniPrep (Zymo Research, Irvine, CA). Bacterial 16S rRNA containing V4-V5 regions were obtained using Illumina Miseq compatible primers. PCR amplicons were purified (Qiagen, Madison, WI) following separation on an agarose gel, and quantified by the Qubit Fluorometer (Invitrogen, Grand Island, NY). Equal amounts of amplicon for each sample were pooled and qualified with Kapa SYBR fast qPCR kit (Kapa Biosystems, Inc., Woburn, MA). Pooled samples were run on the Illumina Miseq platform using Miseq v2 reagent kit (Illumina, Inc., San Diego, CA). Primers used for amplification are listed in Table 1.

The bioinformatics analyses were performed as described previously (Lightfoot et al., 2014; Yang et al., 2015). Briefly, sequenced reads were demultiplexed according to a combination of forward and reverse indices. Subsequently, obtained reads were filtered based on additional requirements: exact match to sequencing primers and quality score of 30 or more. Filtered reads were joined using FLASh (Fast Length Adjustment of Short reads) software, applied against Silva non-redundant 16S rRNA references database (v108), assigned taxonomic classifications at a 97% clustering threshold, and summarized in a refOTU table. Alpha, beta diversity, and principal coordinate analyses (PCoA) measures were created using custom R scripts and scripts from the QIIME package. Linear discriminant analysis (LDA) along with effect size measurements (LEfSe) in Galaxy was used to distinguish differentially significant features at each taxonomic level. Bacterial genus were identified based on 16S sequencing data and grouped based on the primary metabolite according to Bergey's manual as before described (Antharam et al., 2013; Yang et al., 2015).

Fecal pellets (N = 4/group) were collected and SCFAs were extracted, as previously described (De Baere et al., 2013) with some modifications. Briefly, 200 μL of HCl was added to the fecal homogenates to preserve the volatile SCFAs, and vortexed vigorously to evenly suspend the fecal mass. Five milliliters of methylene chloride was used to extract the SCFAs with gentle rotation at room temperature for 20 min. After centrifugation at 3,500 rpm for 5 min, the supernatant was discarded. Five hundred microliters of 1N NaOH was added to the organic pellet, followed by an additional 20 min rotation at room temperature. After spinning at 3500 rpm for 5 min, the top aqueous phase was collected and mixed with 100 μL of HCl before filtered for injection. A 50 μL aliquot was used for each injection. A standard linear regression curve was generated for each of the acetate, lactate, propionate, and butyrate solutions separately, for calibration purposes. All of the standard solutions were extracted according to the method described above. Relative standard deviation (< 3.0%) was achieved by sequential injections of 10 mM standard mix (of all three SCFAs) into the mobile phase. The chromatographic separation was performed at room temperature using Perkin Elmer Series 200 HPLC System (Perkin Elmer Instruments, Norwalk, CT), equipped with an autosampler, quaternary pump, and 200 series UV/VIS detector. The analytical column used was Hypersil Gold aQ 150 × 4.6 mm 3 μm (Thermo Fisher Scientific, Walthan, MA). Samples were run in a combination of two solutions: 90% of mobile phase A 0.02 M phosphate buffer pH 2.2 and 10% of mobile phase B acetonitrile. Flow rate was set as 0.8 ml/min for 25 min. Wavelength used for detection was 210 nm.

Extraction of RNA from the distal colon was performed using 1 mL TRIzol® Reagent (Life Technologies, Carlsbad, CA), as per manufacturer's protocol. Final nucleic acid pellets were resuspended in 30 μL RNAse-DNAse free water. Total RNA integrity for all samples used in microarray and real-time PCR analyses was determined using the RNA 6000 Nano Assay Kit with the 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). The mean RNA integrity value for all samples was 8.16 (SD ± 0.76).

The SurePrint G3 Mouse GE 8 × 60 K Microarray Kit (Agilent Product Number, G4852A Design ID 028005) was used for expression analyses (Agilent, Santa Clara, CA, USA). Sample sizes for microarray analysis were N = 7 per group. Briefly, the RNeasy Mini Kit was used to purify RNA prior to labeling, as per manufacturer's protocol (Qiagen, Valencia, CA, USA). RNA concentrations were determined using the NanoDrop-2000 spectrophotometer (Thermo Scientific). Microarray hybridizations were performed according to the One-Color Microarray-Based Gene Expression Analysis Low Input Quick Amp Labeling kit (Agilent V6.5, May 2010) and 10 ng total RNA per sample was used for labeling and hybridization as per our previous protocol (Feswick et al., 2014; Ornostay et al., 2016). Microarrays were scanned at 5 μm with the Agilent G2505 C Microarray Scanner, and Agilent Feature Extraction Software (v. 10.1.1.1) was used to extract raw signal intensities from microarray images. Microarray data were evaluated by manual inspection of the quality control parameters. All arrays were deemed high quality. Raw microarray data have been deposited into the NCBI Gene Expression Omnibus (GEO) database (Series GSE71632, GPL13912). Raw intensity data were imported into JMP® Genomics v7.0 (SAS Institute Inc., Cary, NC, USA). Intensity data were normalized using quantile normalization. Control probes were removed prior to identifying differentially expressed genes (DEGs), and the limit of detection was set to an intensity of 2.5 based on the Agilent spike-in controls. Therefore, any probe falling below this value was assigned a normalized intensity of 2.5. DEGs were identified using a one-way analysis of variance (ANOVA) followed by a false discovery rate (FDR) set at 5.0%. Cluster analyses were conducted in JMP Genomics v7.0 using probes that were differentially expressed (unadjusted p < 0.05). Two-way clustering using the Fast ward algorithm was conducted after each row was centered to a mean of zero (0) and variance scaled to one.

Pathway Studio 9.0 (Elsevier) and ResNet 10.0 were utilized for sub-network enrichment analysis (SNEA) of cell processes (Nikitin et al., 2003). The option of “best P-value, highest magnitude fold change” in Pathway Studio was used for duplicated probes. A total number of 32,734 mouse probes were successfully mapped to the program using the official gene name. SNEA was performed to identify transcriptome networks that were enriched in the distal colon. A Kolmogorov–Smirnov test with 500 permutations was conducted to determine whether networks were preferentially regulated compared to the background reference probability distribution. Networks were constructed based on common regulators of expression and regulators of specific cell processes. The enrichment P-value for a gene seed was set at P < 0.05. Additional details on the use of SNEA can be found in a previous publication (Langlois and Martyniuk, 2013).

For distal colon microarray validation, primer sets from target genes were collected from primerbank (Table 2; (Wang et al., 2012)). Transcripts that showed dramatic and significant changes in expression levels were selected for microarray validation. The genes investigated included CFD, Cyp2e1, and Ms4a4b. N = 7–8/group. PCR was performed on CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad, Hercules, California). The PCR conditions were as follows: 95°C for 3 min, followed by 40 cycles of 95°C for 30 s, primer annealing at 60°C for 30 s, and 72°C for 30 s. Dissociation curves were generated, starting at 65°C and ending at 95°C with increments of 0.5°C every 5 s. Gene expression was normalized to GAPDH and was determined using the relative ΔΔCt method based on the method described (Pfaffl, 2001). No reverse transcriptase control was used.

Colon tissue was fixed in 4% paraformaldehyde (PFA) and kept in 30% sucrose solution for 1 day. Tissue was then frozen at −80°C in O.C.T embedding, cross sectioned at 10 μm thickness via cryostat at −20°C, mounted onto glass slides, and routinely stained with hematoxylin and eosin (H&E). Sections from C57 (N = 2) and b1/2-ARs KO (N = 2) chimera mice were qualitatively evaluated for histopathologic lesions. Slides were digitally scanned using an Aperio System to produce representative images captured from the scans.

Colon tissues (N = 3/group) were fixed in 4% PFA, and kept in 30% sucrose solution for 1 day before freezing in O.C.T. Frozen tissues were sectioned at a thickness of 10 μm and mounted onto slides for immunofluorescent staining. Primary mouse anti-CD4 (1:500, #14-0041-85, eBioscience, San Diego, CA), anti-CD94 (1:100, #sc-390776, Santa Cruz Biotechnology, Dallas, Texas) and anti-CD11b (1:200, sc-20050, Santa Cruz Biotechnology, Dallas, Texas) were used, according to the manufacturer's protocol. Secondary antibody conjugated with Alexa Fluor 594 (Abcam, Cambridge, UK) was used for visualizing infiltration of CD4⁺ cells, NK cells, and macrophages, separately, in the intestines of both C57 and b1/2 ARs KO chimera. Three to four images/rat/antibody were used for quantification. Positive cell numbers were divided by total cell numbers, as indicated by DAPI staining.

The rodent gut microbiota is generally dominated by Firmicutes and Bacteroidetes phyla, with Proteobacteria, Actinobacteria, and Tenericutes constituting the remaining phyla (Yang et al., 2013). Shifts in the bacterial proportions and phylotypes are widely used to evaluate gut homeostasis. Our 16S rRNA amplicon sequencing data suggested that Chao richness and Shannon diversity indices did not differ between C57 and b1/2-ARs KO chimeras (Figure 1A), indicating that deficiency in BM b1/2-ARs signaling did not induce changes in the microbiota species richness or biodiversity (Antharam et al., 2013; Yang et al., 2015; Mosca et al., 2016). However, a shift from a dominance of Firmicutes to Bacteroidetes, resulting in a reduction of the Firmicutes to Bacteroidetes (F/B) ratio, was observed (Figure 1A). We observed significant expansion of Bacteroidetes from approximate 1/3–1/2 of total bacteria reads and contraction of Firmicutes from approximate 2/3 to ½ of total bacteria reads in the b1/2-ARs KO chimera group (Figure 1B). No other change was observed at phylum level. Members of the alpha and delta proteobacteria in the Proteobacteria phylum, and Anaeroplasmataceae family in the Tenericutes phylum were depleted in the b1/2-ARs KO chimera (Figure 2A). Analyses of gut microbial community structure by unweighted UniFrac principal coordinate analysis (PCoA) revealed two separate clusters, indicating dissimilarities in microbial community structure between the b1/2-ARs KO chimeras and the C57 controls (Figure 1C). Furthermore, a number of genera, including Anaerotruncus, Coprococcus, Oscillibacter, and Desulfovibrio were significantly reduced in b1/2-ARs KO chimeras compared to C57 control chimeras (Figure 2B). Therefore, deficiency in b1/2-ARs signaling in the BM led to an altered gut microbiota, characterized by a shift from Firmicutes to Bacteroidetes dominance, a reduction in the relative abundance of Proteobacteria, and diverse changes at the genus level. Interestingly, no significant dysbiosis was observed in the b1/2-ARs KO chimeras. By contrast, the shifts in gut microbial composition is considered beneficial due to the depletion of pathogenic bacteria and alteration in F/B ratio.

Functionally, the end-products of anaerobic fermentation of the gut microbiota significantly contribute to the maintenance of intestinal and systemic homeostasis. Therefore, we grouped bacterial phylotypes at the genus level based on their primary fermented end-products (i.e., acetate, butyrate and lactate; Antharam et al., 2013; Yang et al., 2015). We did not observe significant alterations of total acetate, butyrate, and lactate-producing bacteria (Figure 3A) between C57 control and b1/2-ARs KO chimeras. However, hydrogen sulfide-producing bacteria, Desulfovibrio, was contracted in the b1/2-ARs KO chimeras (Figure 2B). HPLC analyses of fecal samples yielded no significant differences in butyrate levels between the KO and control chimeras, consistent with the bacterial analyses (Figure 3B). In the colon histological analysis, all sections from C57 and b1/2-ARs KO chimera animals were interpeted to be qualitatively normal, and no differences were detected between groups (Figure 3C). These indicate that the shift of gut microbial composition did not impact the major SCFA production and colon histology.

We have demonstrated that the loss of b1/2-ARs signaling in BM resulted in a shift in the gut microbiota. Due to the constant communication between intestine and the microbiota, we also measured the host colon transcriptome using a whole-genome microarray to determine global gene expression changes in the gut that may be involved in the microbiota shift. Based upon these data, we selected three abundantly expressed genes with significantly different expression levels (Appendix 1) between the C57 control and KO chimeras to validate the microarray data. Consistent with the microarray data, the expression levels of CFD, Cyp2e1, and Ms4a4b were decreased in the KO chimera mice (Figure 4A).

Subsequently, the microarray data were analyzed as described previously (Feswick et al., 2014; Ornostay et al., 2016). A vast number of differentially expressed probes in the distal colon of C57 vs. b1/2-ARs KO chimera mice were observed (N = 4,072, P < 0.05). All expression data are provided in Appendix 1. The two-way cluster is presented in Figure 4B to show the clustering of the individuals based on treatment. Thus, the cluster analyses using differentially expressed transcripts (unadjusted, P < 0.05) revealed that individual mouse from the two groups were separated based upon gene expression profiles. A core group of transcripts was uniquely different between the C57 control and the b1/2-ARs KO chimeras. Furthermore, microarray analyses identified three transcripts that showed significantly altered expression in the distal colon following an adjusted P-value (FDR = 5%): Killer Cell Lectin-Like Receptor Subfamily D, Member 1 (Klrd1), Membrane-Spanning 4-Domains, Subfamily A, Member 4A (Ms4a4b), and Casein Kinase 2, Alpha Prime Polypeptide (Csnk2a2). Ms4a4b is a CD20 homolog in T cells (Xu et al., 2010). Klrd1 is known as CD94 and is expressed on NK cells (Fang et al., 2011). Csnk2a2 is a protein kinase that promotes epithelial cell restitution (Song et al., 2000). Thus, these three differentially expressed genes are hypothesized to be important for both immune responses in the gut and communication with the microbiome. Herein, we observed significant upregulation of Csnk2a2 by 1.3-folds, while Klrd1 and Ms4a4b were downregulated ~4-fold in the distal colon of b1/2-ARs KO chimera mice (Figure 4C).

We conducted further analyses of the host intestinal gene expression to identify specific host IS cellular processes that may be associated with the observed microbiota changes in the b1/2-ARs KO chimeras. Strikingly, the majority (262) of cell processes were downregulated at the transcriptome level, while only 16 gene networks were upregulated in the distal colon of the b1/2-ARs KO chimera (Appendix 2). Given that the three most affected genes, Ms4a4b, Klrd and Csnk2a2, are closely related to IS, we focused our analyses on cell processes involved in IS regulation. The cell processes most affected by the b1/2-ARs KO chimera included macrophage adhesion, NK cell proliferation and differentiation, mast cell chemotaxis, and leukocyte function, all showing significant downregulation in the distal colon at the transcriptome level (P < 0.01; Appendix 2). As an example, Figure 5A summarizes gene networks associated with leukocyte function (fold change by −1.103, P < 0.01). In addition, epithelial cell interaction (P < 0.001), permeability (P < 0.01), proliferation (P < 0.0001), and recycling (P < 0.05) were also downregulated (by ~8%, or 1.08-fold) in the distal colon of the b1/2-ARs KO chimera.

Furthermore, gene set enrichment analyses suggested that the classical complement pathway, which is an important part of the IS that contributes to phagocytosis and enhances the ability of antibodies to clear pathogens, was also suppressed in the distal colon of b1/2-ARs KO chimera (Figure 5B). By contrast, there were very few cell processes increased in the colon of b1/2-ARs KO chimera, and these included stromal cell function, cell-cell recognition, mucosal secretion, and processes associated with peripheral nerve blood flow. These processes increased at the level of the transcriptome by 10–20% (Fold change of 1.1–1.2; Appendix 2). Here we point out that this is the median fold change of the entire network, and networks are based on 100s of genes in some cases. Thus, fold changes for a network, when many genes are involved, are expected to be smaller compared to any single gene.

Due to the observed changes in the colon IS, sub-network enrichment analyses were performed in order to identify possible associations with diseases described in the literature (Borody et al., 2015; Alipour et al., 2016; Degruttola et al., 2016). This analysis revealed a significant association between the observed transcriptome changes in our b1/2-ARs KO chimera and diseases of the intestine function (P < 0.01), barrier integrity (P < 0.001), secretion (P < 0.05), and motility (P < 0.0001), all associated with 5–12% (1.05–1.12-fold) downregulation of gene networks in the b1/2-ARs KO chimera (Figure 5C). This suggests that intestinal function may be altered following the b1/2-ARs KO BM transplantation. Moreover, the transcriptome analysis revealed an overall downregulation of gene networks associated with diseases related to bowel inflammation (Figure 5D). For example, gene networks associated with ileitis were suppressed ~8% (P < 0.05), and included Cd36, Il4, Cd44, Cd4, and Cd40 (Appendix 2). Gene networks related to enteritis (P < 0.05) and intestinal morphology (P < 0.05) were also decreased at the transcriptome level in the colon of b1/2-ARs KO chimera. Additionally, a transcriptional network of inflammatory stress molecules was decreased by 6% (1.06-fold; P < 0.01). This network included genes such as Nos3, Il1b, Fos, Rora, Tlr4, and Fosl1, among others. A network related to inflammatory lesions (e.g., transcripts including Gf1, Ccl5, Cd8a, Il1b, Il1a, Il4, Il18, Cd4, Il15, Cx3cl1, Pparg, and Il6, among others) were also decreased by 8% (1.08-fold; P < 0.05). Lastly, obesity networks (e.g., transcripts including Retn, Lpl, Lep, Esr1, Lipc, Hsd11b1, Gpt, Lipe, Vldlr, and Il6, among others, fold change −1.042, P < 0.05) were also decreased by 4% (1.04-fold). All enriched disease networks are reported in Appendix 2.

We report an overall suppression of the immune system in the gut of b1/2-ARs KO chimera. Since the suppressed transcripts, MS4a4b and Klrd1, are primarily expressed in T cells and NK cells, respectively, we proposed that the reduction in these gene transcripts in the colon would be associated with a decreased presence of infiltrating CD4⁺ T and NK cells in the colon of b1/2-ARs KO chimera. Thus, we employed immunofluorescence to quantify the relative frequencies of CD4⁺ T cells, CD11b⁺ macrophages, and CD94⁺ NK cells in the colon. Although there were no apparent qualitative histologic differences between the C57 and b1/2 ARs KO chimeras, we observed a reduced contribution of CD4⁺ T cells (by 61.8% or 1.62-fold; Figures 6A,D), CD11b⁺ macrophages (by 65.8%, Figures 6B,D), and CD94⁺ NK cells (by 75.2%, Figures 6C,D) within the colonic lamina propria of the b1/2-ARs KO chimera, in line with the observed reduction in immune responses at the transcriptome level.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.