Specific pathogen free (SPF) grade female MRL/lpr mice (6 weeks old) and C57/BL6 mice (6 weeks old) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China) and maintained in the SPF environment of the Zhejiang Chinese Medical University Laboratory Animal Research Center. The mice were housed in controlled environments with free access to food and water. The mice were weighed twice weekly, and the drug doses were adjusted accordingly. All animal experiments were performed according to the requirements of the Institutional Animal Care and Use Committee of China.

The time course and grouping information for the treatments is shown in Figure 1A. In the first 2 weeks, the mice were treated with a cocktail of broad-spectrum antibiotics, including ampicillin (0.2 g/L), vancomycin (0.1g/L), neomycin (0.2 g/L), and metronidazole (0.2 g/L), in their drinking water. A previous study reported that 2 weeks of antibiotics leads to intestinal dysbiosis (Aguilera et al., 2015). Then, a portion of the mice was orally gavaged with a 200 μL/day aliquot of fecal suspensions (initial dilution 5 g/mL feces) for 1 week. The fecal suspensions were prepared by resuspending the fresh feces of normal C57/BL6 (8 weeks old) in phosphate-buffered saline (PBS) as previously described (Ekmekciu et al., 2017). After the intervention of gut microbiota, the MRL/lpr mice with different types of gut microbiota were orally gavaged with 5 mg/kg prednisone every day for 4 weeks. The mice that did not receive FMT or prednisone treatments were subjected to oral gavages of water to match the stress of gavage manipulation every day. The entire experimental period was 7 weeks, starting 1 day after antibiotic administration. The sampling times were the 3rd, 5th, and 7th weeks. Each group contains five mice.

Colonic microbiota samples were collected within 10 min after euthanization. To avoid cross-contamination, each microbiota sample was collected by using a new pair of sterile tweezers. Total genomic DNA was extracted from each stool sample using a QIAamp^® DNA Stool Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocols. Harvested DNA was PCR amplified with broad-range bacterial primers targeting the V3-V4 region of the 16S rRNA gene as previously described (Yu et al., 2018). Subsequently, the amplicons were purified according to standard procedures, quantified, pooled and sequenced with the MiSeq Reagents Kit v3 (600 cycles, Illumina) according to the manufacturer’s instructions with 20% PhiX (Illumina). The sequencing reaction was conducted by Hangzhou Guhe Information and Technology Co., Ltd., Zhejiang, China.

The processing and quality filtering of the reads were performed by using scripts in Quantitative Insights into Microbial Ecology (QIIME, Version 1.9) (Caporaso et al., 2010). The clean reads were extracted from the raw-paired end reads according to previous studies (Yu et al., 2018). UCLUST was used to cluster sequencing reads into operational taxonomical units (OTUs based on > 97% identity) (Edgar, 2010). Bacterial taxonomy was assigned by using the SILVA (Quast et al., 2013) and NCBI databases (Sayers et al., 2019). The microbiota OTUs were imported into R version 3.4.3 and alpha and beta diversity metrics were computed using the “vegan” package. Alpha diversity included the Shannon and Chao1 diversity index, and beta diversity included unweighted UniFrac distance metrics. Beta diversity metrics were then visualized using Principal coordinate analyses (PCoA) in R version 3.4.3. OTUs with > 0.05% mean abundance in one sample and observed in > 10% of the samples were included in differential analyses. The linear discriminant analysis (LDA) effect size (LEfSe) method was carried out using Galaxy¹, with a set logarithmic LDA score of 2.0 and the standard test for the significant difference between the two groups (Kruskal–Wallis test and Wilcoxon rank-sum Test) (Segata et al., 2011).

Serum creatinine (Cr) and alanine aminotransferase (ALT) were measured via an enzymatic-colorimetric method using standard test kits on a TBA-40FR automated biochemical analyzer (Toshiba, Japan). Serum lactate dehydrogenase (LDH) activity was determined using the lactate dehydrogenase assay kit (Jiancheng, China) according to the manufacturer’s instructions. A multi-analyte ELISA was used to measure the levels of anti-dsDNA antibody (Shibayagi, Japan), IFN-α (Invitrogen, United States) and IL-6 (Invitrogen, United States) in serum according to the manufacturer’s instructions.

All results were presented as Mean ± SEM of data from at least three independent experiments. Differences between groups were evaluated with one-way ANOVA. Statistical analysis was performed using Prism 5.0 (GraphPad Software, United States). A significance level of p < 0.05 was considered statistically significant.

At 9 weeks-old of age, there were some differences in gut microbiota between MRL/lpr and C57/BL6 mice (Figures 1D,E, Supplementary Figure S1A, and Table 1). However, the MRL/lpr mice had significantly higher levels of autoantibodies and inflammatory factors than the C57/BL6 mice (Figure 2). After short-term antibiotics exposure, the alpha diversity of gut microbiota significantly reduced (Figures 1B,C) and the overall compositions of gut microbiota significantly changed (Figure 1D). At the phylum level, antibiotics caused significant decreases in Firmicutes and Bacteroidetes but significant increases in Proteobacteria and Verrucomicrobia (Figure 1E). At the genus level, antibiotics significantly downregulated 17 genera, including Bifidobacterium, Bacteroides, and Lactobacillus, and only two genera (Klebsiella and Proteus) were upregulated by antibiotics (Supplementary Figure S1B and Table 1). To renovate intestinal dysbiosis, the fecal microbiota from normal C57BL/6J mice was transplanted into the antibiotic exposure mice in the following 1 week. The alpha diversity and abundances of Firmicutes and Bacteroidetes were resorted after FMT (Figures 1B,C,E). Additionally, FMT could resort to the abundance of 10 genera changed by antibiotics, such as Bifidobacterium, Adlercreutzia, Bacteroides, Klebsiella, and Proteus (Supplementary Figure S1C and Table 1). However, the gut microbiota in FMT-treated mice was also inconsistent with that in the model mice (Figure 1D). Compared with the model mice, the abundance of three genera was higher, and the abundance of seven genera was lower in the FMT-treated mice (Supplementary Figure S1D and Table 1). In sum, this study successfully obtained three types of lupus mice with differently initial gut microbiota compositions before onset.

After the intervention of gut microbiota before onset, lupus activity showed being different in three types of MRL/lpr mice. As shown in Figure 2, the levels of 3 serum biochemical indexes important for liver and kidney function (ALT, Cr, and LDH) were significantly increased by the early and short-term antibiotics exposure. Although FMT could significantly alleviate the liver and kidney damage caused by antibiotics exposure, it did not eliminate the influence of antibiotics exposure (Figure 2). Anti-dsDNA autoantibodies are the most studied antibodies with lupus-related autoantibodies. The early and short-term antibiotics exposure caused significant increases in serum anti-dsDNA autoantibodies in MRL/lpr mice, and FMT reduced serum anti-dsDNA autoantibodies to levels before antibiotic exposure (Figure 2D). The effects of antibiotics and FMT on anti-dsDNA autoantibodies were consistent with their effects on inflammatory cytokines (IFN-α and IL-6) (Figures 2E,F).

In addition, this study also revealed the effects of the early and short-term antibiotics exposure and FMT on lupus progression. There were no differences in the developments of ALT and Cr, but some differences in the development of LDH, anti-dsDNA, IFN-α, and IL-6 were observed among the three groups (Figure 3). The levels of LDH and anti-dsDNA autoantibody were significantly increased from 9 to 13 weeks old in MRL/lpr mice of all groups. In addition, the anti-dsDNA levels in FMT-treated mice were the lowest among the three types of mice at 11 and 13 weeks old (Figure 3D). Interestingly, the level of IFN-α or IL-6 showed different development with anti-dsDNA among the three groups (Figures 3E,F). In the control mice, both IFN-α and IL-6 had significantly increased development from 9 to 13 weeks old. In the antibiotic-treated mice, IFN-α showed a significant decrease, and IL-6 showed an unobvious decrease from 9 to 13 weeks old (Figures 3E,F). In the FMT-treated mice, IFN-α showed a significant increase, but IL-6 showed an unobvious alteration from 9 to 13 weeks old (Figures 3E,F). In sum, early and short-term antibiotics exposure imposed no significant effects on lupus progression, but the following 1 week of FMT suppressed lupus progression in MRL/lpr mice from 9 to 13 weeks old.

In our previous study, interventions in the gut microbiota could affect the efficacy of prednisone on lupus (He et al., 2019). This study also demonstrated that the efficacy of prednisone was influenced by early and short-term intervention in gut microbiota by antibiotics exposure and FMT. In the control mice, prednisone showed significant therapeutic effects on lupus according to the decreases in ALT, Cr, LDH, anti-dsDNA, IFN-α, and IL-6 (Figure 4). However, prednisone could not alleviate lupus in both antibiotic-treated and FMT-treated mice (Figure 4). The level of anti-dsDNA autoantibody was reduced by prednisone in the early and short-term antibiotic exposure mice, but this effect was not statistically significant (Figure 4D). Similarly, the levels of IFN-α and IL-6 were not significantly influenced by prednisone in the early and short-term antibiotic exposure mice (Figures 4E,F). In the early and short-term FMT-treated mice, prednisone also caused no significant alteration in anti-dsDNA autoantibodies, IFN-α, and IL-6 (Figures 4D–F).

Simultaneously, the effects of prednisone on the gut microbiota were also different in the three types of MRL/lpr mice. Prednisone imposed no effects on alpha diversity (Chao1, Shannon) in control mice, decreased the Chao1 index in the control and early and short-term antibiotic exposure mice, and increased alpha diversity (Chao1, Shannon) in the early and short-term FMT-treated mice (Figures 5A,B). LEfSe analysis was performed to reveal microbial genera that were altered by prednisone in the three types of lupus mice (Supplementary Figures S2–S4). A Venn diagram showed the shared or unique altered genera of the gut microbiota in the three types of lupus mice. As shown in Figure 5C, a total of 6, 11, and 14 genera altered by prednisone were observed in control, antibiotic- and FMT-treated mice, respectively. Only two altered genera were shared by the three types of lupus mice: prednisone increased the abundance of (Prevotella) and Parabacteroides in control and antibiotic-treated mice but decreased these bacteria in FMT mice (Figure 5C). Additionally, the increase in Prevotella caused by prednisone was shared in control and antibiotic-treated mice. Five genera altered by prednisone were also shared in antibiotic- and FMT-treated mice but the effects of prednisone on the four shared altered genera (Dehalobacterium, Dorea, Sutterella, and Turicibacter) were different. No shared altered genera were observed in control and FMT-treated mice (Figure 5C). The uniquely altered genera in control mice were Proteus, Bilophila, and Klebsiella, which were all decreased by prednisone. The uniquely altered genera in antibiotic-treated mice were Allobaculum, Bifidobacterium, and Adlercreuzia, which were also all decreased by prednisone. Seven genera were uniquely altered by prednisone in FMT-treated mice, including decreased Bacteroides and Akkermansia and increased (Ruminococcus), Odoribacter, Coprococcus, Ruminococcus, and Desulfovibrio (Figure 5C).