The Bristol isolate of C. elegans, N2 strain and the RNAi deficient mutant WM27 rde-1 (ne219) was used in this study. All worm strains were maintained on nematode growth medium (NGM) plates at 20°C and seeded with the Escherichia coli (E. coli) bacteria strain OP50 unless otherwise indicated (Brenner, 1974). Serial passage of both C. elegans N2 and rde-1 (ne219) strain was conducted by transferring one square centimeter of agar with mixed stage animals onto a fresh OP50 seeded NGM plate.

Eggs were obtained by incubating freshly starved larvae on new 90 mm diameter seeded NGM plates with basic hypochlorite solution as previously described (Lewis and Fleming, 1995). Synchronous cultures were achieved by allowing the purified eggs to hatch overnight in M9 buffer without bacteria. The larval stage 1 (L1) animals were then washed by M9 buffer and diluted to 1000 worms per ml for Orsay virus infection.

Orsay virus was maintained using the JU1580 isolate of C. elegans at room temperature and the virus filtrate was prepared following a protocol described previously (Felix et al., 2011). To prepare Orsay virus inoculum, infected JU1580 worms were washed off from fresh slightly starved 90 mm plates using M9 buffer, 3 ml per plate. After centrifugation twice at 21000 g for 5 min (4°C), the virus-containing supernatant was then filtered through a 0.22-μm filter unit (Merck Millipore, Germany). For all strains, 10 young adults were inoculated on 55 mm culture plates. At the same time, 40 μl of infectious filtrate was pipetted onto the bacterial lawn. The cultures were incubated at 20°C for 4 days. Infections were performed in four biological replicates.

Four days after infection, all worms were collected in M9 and then divided into two parallels. The worms were treated based on the method described in the literatures (Alegado and Tan, 2008; Niu et al., 2016). The worms were surface sterilized by washes with 100 mM levamisole and incubated in M9 buffer containing 100 mM levamisole and 100 mg/ml gentamicin. Levamisole treatment has been demonstrated to inhibit pharyngeal pumping and defecation. The treatment helps to prevent gentamicin entering into the intestinal lumen and the release of luminal bacteria. After incubation, the adult worms were washed with a levamisole solution to remove the gentamicin and homogenized with M9 containing 1% Triton X-100 to recover bacteria within the worm intestine.

DNA was extracted using the DNeasy^® Blood&Tissue Kit (QIAGEN, Germany) according to the manufacturer’s instructions. The genomic DNA was evaluated with a NanoDrop spectrophotometer (Thermo Fisher Scientific, United States), using an A260/A280 ratio between 1.8 and 2.0 as a criterion for quality control and agarose gel electrophoresis, respectively. DNA samples were stored at -20°C until use. Total RNA was prepared using Invitrogen TRIzol (Thermo Fisher Scientific, United States), by following the manufacturer’s instructions.

cDNA was generated from 1 μg total RNA applying random primers using Superscript III (Thermo Fisher Scientific, United States), and cDNA was diluted to 1:100 for qRT-PCR analysis. qRT-PCR was performed using SYBR Green realtime PCR master Mix (TOYOBO, Japan). The amplification was performed on a CFX96 Touch^TM Deep Well Real-Time PCR Detection System (Bio-Rad Laboratories, Inc., United States). Each sample was normalized to ama-1, and then viral RNA1 (primers GW194: 59 ACC TCA CAA CTG CCA TCT ACA and GW195: 59 GAC GCT TCC AAG ATT GGT ATT GGT) levels were compared to those uninfected animals as negative controls.

Accumulation of Orsay virus and the intestinal cell infection phenotype in N2 and rde-1 (ne219) mutant animals were detected by single molecular fluorescent in situ hybridization (smFISH), which was performed as previously described (Raj et al., 2008; Raj and Tyagi, 2010; Ji and van Oudenaarden, 2012; Franz et al., 2014). We used the set of custom Stellaris^TM Molecule FISH Probes (Biosearch Technologies, Inc., United States) labeled with Quasars^® 670 Dye for the Orsay virus RNA1 molecule in this experiment. The oligonucleotide sequences were reported by Franz et al. (2014). Color images through an Olympus BX51 DIC microscope were acquired by a Canon T1 DSLR connected to an LM Microscope adapter.

DNA libraries were constructed according to the standard protocol provided by Illumina. Quality control of libraries were performed before sequencing. Libraries were sequenced using the Illumina Miseq instrument (Illumina, United States) with 2 × 300 base pairs (bp) paired-end (PE) sequencing.

mRNA was concentrated from total RNA and cDNA libraries were constructed using mRNA. After validation and the quality assessment of each library on gel electrophoresis and bioanalyzer (Agilent Technologies, United States), the libraries were sequenced on the Illumina Hiseq platform (Illumina, United States) with 2 × 150bp paired-end (PE) sequencing.

The raw data generated by sequencing were analyzed using the pipeline tools in QIIME (Caporaso et al., 2010). Sequences were demultiplexed using customized Perl scripts, based on the presence of the unique barcodes assigned to each sample.

Data were de-noised by removing reads with average quality values <30. After removing both the barcodes and primer sequences, the filter-passed reads were clustered into Operational Taxonomic Units (OTUs) at a 97% similarity cutoff using the de novo OTU-picking strategy. Taxonomies were assigned according to the Greengenes database.

Following construction of the OTU table and phylogenetic tree, microbial richness (Chao1, observed species, and phylogenetic diversity (PD)), diversity (Shannon Index and Simpson Index), and evenness (Equitability, Simpson diversity index) estimators were calculated. Besides, microbial structure was evaluated by analyzing weighted UniFrac distance, unweighted UniFrac distance, and Bray–Curtis distance metrics (Lozupone and Knight, 2005). The relative abundances of microbial taxa at the phylum, class, order, family, genus, and species levels were calculated and compared.

Raw sequencing reads were trimmed using the following criteria: (1) reads containing adapters were removed; (2) reads containing greater than 10% N bases were removed; and (3) reads containing greater than 50% low-quality reads (Q ≤ 5) were removed. After which, the resulting filtered reads were mapped and annotated. Gene expression was quantified as Reads Per Kilo bases per Million reads (RPKM) (Mortazavi et al., 2008). DESeq (Anders and Huber, 2010) was used to perform statistical analyses of the gene expression profiles.

The data were further analyzed with the following statistical methods. (1) Differences in alpha- and beta-diversity were evaluated using the non-parametric two-sample t-tests with 1000 permutations. (2) Differences in relative abundances of individual OTUs and microbial taxa (phylum, class, order, family, genus, and species) were analyzed using LEfSe (Segata, 2011) with default parameters [P < 0.05, Linear discriminant analysis (LDA) score > 2.0].

To monitor the composition of the intestinal microbiota of C. elegans upon Orsay virus infection, we analyzed 16S ribosomal DNA library data of infected and non-infected worms. The laboratory reference strain N2 and an rde-1 (ne219) mutant deficient in the primary Argonaute protein RDE-1 were used in this study.

Quantitative real-time polymerase chain reaction (qRT-PCR) and single molecular fluorescence in situ hybridization (smFISH) were performed to verify Orsay virus infection of C. elegans. rde-1 (ne219) mutant animals were more sensitive to infection with Orsay virus than were N2 animals, as measured by viral load (Figure 1A). Previous work suggested that C. elegans intestinal cells are structurally altered by Orsay virus infection and the viral particles are mainly localized to intestinal cells (Felix et al., 2011; Franz et al., 2014). Using smFISH, the Orsay virus RNA1 segment was detected in intestinal cells of adult animals. Orsay virus-infected rde-1 (ne219) mutant (rde-1_OrV) animals showed a much greater proportion of infected intestinal cells compared to N2 animals (Figure 1B).

Orsay-infected N2 (N2_OrV) animals showed no significant difference in intestinal microbial diversity compared with uninfected animals (Figures 2, 3, P > 0.05). According to the weighted UniFrac distance measurements, a clear separation was found between uninfected N2 (N2_mock) animals and uninfected rde-1 (ne219) mutant (rde-1_mock) animals. The weighted UniFrac distance in N2_mock group was greater than in rde-1_mock group, and this demonstrated that the microbial diversity in N2 animals was significantly greater than in rde-1 (ne219) mutant animals. The variation of microbial characteristics was similar for rde-1 (ne219) mutant animals but significantly greater for N2 animals (Figure 2 and Supplementary Table S1). Uninfected N2 animals showed the greatest bacterial community diversity, which was considered to be an indicator of improved health, whereas the rde-1 mutant tended to host similar microbial communities.

We analyzed microbial diversity (alpha diversity) to determine whether RDE-1 deficiency and Orsay virus infection were associated with the intestinal microflora of C. elegans. The microbial diversity of N2 animals was not significantly affected by Orsay virus infection, and RDE-1 deficiency was associated with decreased microbial diversity, which decreased further after virus infection. OTU richness did not differ significantly between N2_mock and N2_OrV animals, but was lower in rde-1 (ne219) mutant animals compared with N2 animals. Orsay virus-infected rde-1 (ne219) mutant animals had lower OTU richness than uninfected animals (Figures 3A,D). The Shannon diversity indices showed that rde-1 (ne219) mutant animals tended to harbor fewer diverse bacterial communities than did N2 animals, and that this decrease in diversity was exacerbated by Orsay virus infection (Figure 3E). The equitability index and Simpson index (measures of phylogenetic diversity, PD) indicated that the microbial communities of uninfected N2 animals exhibited the greatest diversity, and those of Orsay virus-infected rde-1 (ne219) mutant animals exhibited the lowest (Figures 3B,C,F). Therefore, N2 animals exhibited higher levels of biodiversity and unevenness estimations compared to RDE-1-deficient animals. This indicated that RNAi plays a role in maintaining microbiomes during virus infection and the defective RNAi response in rde-1 (ne219) animals made the animals unable to maintain their microbiome when challenged with virus infection.

To evaluate the response of the gut microbiome to RDE-1 deficiency and virus infection, we compared the microbial clades of infected and uninfected N2 and rde-1 mutant animals using LEfSe. The bacterial compositions differed between uninfected N2 and rde-1 (ne219) mutant animals. In the latter, the gut microbiome was significantly enriched in Actinobacteria, including Leucobacter and Pseudoclavibacter; Proteobacteria, including Acinetobacter and Burkholderia; and Alphaproteobacteria, specifically, Ochrobactrum (Figure 4B and Supplementary Figure S1A). In contrast, the intestinal microbiome of uninfected N2 animals was enriched in Bacteroidetes, including Myroides and Sphingobacterium (Figure 4B and Supplementary Figure S1A).

To identify the microbial taxa that respond to Orsay virus infection, we examined the microbiomes of infected and uninfected rde-1 (ne219) mutant animals. Orsay virus infection resulted in marked enrichment in the genera Sphingobacterium, Brevundimonas, and Trabulsiella (Figure 4A and Supplementary Figure S1B). However, compared with Orsay virus-infected N2 animals, the relative abundances of Rhodococcus, Ochrobactrum, Tetrathiobacter, Delftia, Trabulsiella, and Acinetobacter were higher in rde-1_OrV animals (Figure 4C and Supplementary Figure S1C), while the relative abundances of Pseudochrobactrum, Phyllobacterium, Agrobacterium, Shinella, and Stenotrophomonas were higher in N2_OrV animals (Figure 4C and Supplementary Figure S1C). These results suggest that Orsay virus infection and RDE-1 deficiency exert marked effects on the composition of the gut microbiome.

To assess the relationship between the intestinal microbiome and gene expression in C. elegans after Orsay virus infection, we compared the mRNA profiles of mock and Orsay virus-infected N2 and rde-1 (ne219) animals.

The cluster analysis indicated that a large set of genes was associated with the difference between infected and uninfected rde-1 (ne219) mutant animals (Figure 5A). We searched for genes that were significantly differently expressed (fold-change > 2 and P < 1 × 10^-4, Student’s t-test) between uninfected and infected rde-1 (ne219) animals. A total of 144 genes were specifically induced or repressed by Orsay virus infection (Figure 5A and Supplementary Table S2). These 144 genes showed markedly different expression levels; 126 were up-regulated and 18 down-regulated upon infection (Figure 5C and Supplementary Table S2). However, only 17 larger, differently expressed genes were found between uninfected and infected N2 animals; 14 genes were up-regulated and three down-regulated upon infection (Figure 5C and Supplementary Table S3). Gene ontology analysis showed that terms involved in innate immune response or defense response were enriched in the list of upregulated genes in infected compared with uninfected rde-1 (ne219) animals (Figure 5B and Supplementary Figure S2). In contrast, genes downregulated in rde-1 (ne219) animals were mainly involved with lipid metabolism, lipid transport and fatty acid metabolism (Figure 5B and Supplementary Figure S3).