Human norovirus exhibits strain-specific sensitivity to host interferon pathways in human intestinal enteroids

To evaluate intestinal epithelial responses to HuNoV, we infected monolayers of two susceptible, secretor-positive jejunal (J2 and J11) cultures with GII.4 virus. Mock- and gamma-irradiated GII.4 (gGII.4) inoculated cultures were used as controls. We performed RNA sequencing (RNA-Seq) to determine transcriptional responses to GII.4 at 6, 10, and 24 h postinoculation (hpi) in comparison to mock-inoculated cultures (3 hpi) or inoculation with gGII.4 HuNoV (6, 10, and 24 hpi). We have previously shown that the genetic background of the HIE cultures impacts analyses of transcriptional signatures to viral infections, with responses first clustering by HIE line (36). Based on these findings, the genetic background of the HIEs (J2 or J11) was treated as a blocking covariate in a linear model to determine the differential gene expression profiles. Principal component analysis (PCA) showed that gGII.4-inoculated and GII.4-infected samples had a similar early response to mock infections (Fig. 1A). At later times, the transcriptome profiles are more varied with clear separation seen at the 24 hpi between GII.4-infected and gGII.4-inoculated cultures (Fig. 1A and SI Appendix, Fig. S1A).

In concordance with the PCA results, no significant changes in transcription were seen in both GII.4-infected and gGII.4-inoculated samples at 6 hpi compared to media only at 3 hpi (SI Appendix, Fig. S1B). However, a progression in gene activation occurred with time, with transcriptional responses detectable by 10 hpi (Fig. 1B and SI Appendix, Fig. S1B). By 24 hpi, 529 genes were significantly up-regulated in GII.4-infected HIEs when compared to the media controls (Fig. 1B, false discovery rate [FDR] ≤ 0.05 and a fold change [FC] of at least 2) and 266 genes were up-regulated when compared to gGII.4 at 24 hpi (SI Appendix, Table S1), with ISGs accounting for the majority of up-regulated genes. Since some changes in gene expression were also seen in gGII.4-inoculated cultures compared to media controls (Fig. 1B and SI Appendix, Fig. S1B) and we were interested in the kinetics of infection, we next determined the genes that are differentially expressed between GII.4-infected and gGII.4-inoculated samples and differentially changed with time (6 to 10 hpi and 10 to 24 hpi). There were no differentially detected genes between 6 and 10 hpi (Fig. 1C). Between 10 hpi and 24 hpi, 134 genes were up-regulated (FDR ≤ 0.05 and a FC of at least 2) in response to GII.4 compared to gGII.4 (Fig. 1 C and D).

Next, we used gene set enrichment analysis (GSEA) to determine the pathways involved in the response to GII.4 infection for different time intervals (37). Controlling for the effect of potential immune modulators in stool using gGII.4, our GSEA analysis revealed that the top pathways up-regulated by GII.4 with time were IFN signaling and antiviral responses linked to ISGs (Fig. 1E and SI Appendix, Table S1), with a clear time-dependent increase in IFN signaling. These findings were also substantiated when gene expression was examined pairwise at each time point with reference to the media control infection (SI Appendix, Fig. S1). In contrast to the increases in IFN signaling, a time-dependent reduction in ribosomal RNA (rRNA) processing and translation was seen following GII.4 infection when compared to gGII.4 inoculation (Fig. 1E and SI Appendix, Table S2).

In concordance with the GSEA results, we observed a robust, statistically significant increase in expression of ISGs such as IFI44L, MX1, IFIT1, IFIT3, IFITM1 and STAT1 in GII.4-infected HIEs at 24 hpi, as well as other antiviral genes including ADAR, EIF2AK2 (PKR) and OASL (Fig. 1F and SI Appendix, Table S1). Significant increases in the expression of type III IFN transcripts as well as IFNB but not IFNA were seen in both HIE lines compared to mock-infected cultures; this increase was not noted in gGII.4-inoculated HIEs. Although gGII.4 also induced some low-level ISG responses, these responses did not increase with time and are likely due to immune modulators present in the stool filtrate (SI Appendix, Fig. S1). To further evaluate indicators for activated IFN responses, we determined the effect of HuNoV infection on host long noncoding RNA (lncRNA) responses. LncRNAs are transcripts exceeding 200 nucleotides that are not translated into protein but are involved in gene regulation (38). A total of 192 lncRNAs were up-regulated following GII.4 infection at 24 hpi compared to mock-infected cultures and 48 lncRNAs were up-regulated when GII.4-infected samples were compared to gGII.4-inoculated samples at 24 hpi (SI Appendix, Fig. S2A and Table S3). Nine lncRNAs were differentially expressed between GII.4-infected and gGII.4-inoculated samples and differentially changed with time (10 to 24 hpi). While seven among the nine lncRNAs are nonannotated (SI Appendix, Fig. S2 B and C), two lncRNAs (NRIR and BISPR) are both associated with IFN response (39, 40). In summary, the transcriptome data show that GII.4 HuNoV induces a robust ISG response and alters IFN-associated host lncRNAs. The up-regulation of ISGs and increase in IFN signaling observed from GSEA support a role for IFN in the response to viral infection.

We hypothesized the contribution in antiviral pathways through ISGs might be primarily from the IFNL response, rather than IFNA or type II IFN (IFNG), because HIEs induce type III IFN in response to several viral infections (36, 41–46). Indeed, while increases in the expression of IFNB were seen along with type III IFNs in the gene expression (Fig. 1F), differential analysis showed very low expression of type I IFNs in HIEs. To validate the RNA-Seq results, we quantitated type I and type III IFN transcripts during HuNoV infection in J2 HIEs, which showed a more robust transcriptional response compared to J11 HIEs. Cultures inoculated with GII.4 virus treated with a neutralizing antiserum served as a negative control, and the TLR3 agonist poly(I:C) was a positive control to induce IFN pathways. We chose a TLR3 agonist because HIEs express only a subset of TLRs, and TLR3 is the only receptor up-regulated by HuNoV infection in both J2 and J11 HIEs (SI Appendix, Fig. S2D). GII.4 replication was detected starting at 12 hpi (Fig. 2A). As expected, no viral RNA was detected in the mock-inoculated or poly(I:C)-treated wells, and no viral replication occurred when the GII.4 inoculum was neutralized by the HuNoV antiserum (Fig. 2A). Type I IFN transcripts (IFNB1) were induced to low levels following poly(I:C) treatment but not induced after HuNoV inoculation (Fig. 2B). In contrast, the type III IFN transcripts (IFNL1) were highly induced by poly(I:C) treatment and GII.4 infection as early as at 1 and 12 hpi, respectively (Fig. 2C). The induction of type III IFN transcripts at 12 hpi was concurrent with HuNoV replication (Fig. 2A). IFNL1 induction was abrogated when the virus was neutralized by the antiserum, indicating the IFN responses were induced by HuNoV replication and not by other components of stool filtrate. In concordance with the RNA-Seq results, induction of ISG (IFI44L) was seen in GII.4-infected HIEs at 24 hpi but not when the virus infectivity was neutralized (Fig. 2D). These results were confirmed in two independent infected duodenal HIE lines. We next performed a Luminex assay to detect secreted IFN proteins from infected J2 and J11 HIEs at 96 hpi (Fig. 2 E and F). Moderate IFNL1 and high concentrations of IFNL2/3 were produced from HIEs infected with GII.4 or treated with poly(I:C), whereas the gamma-irradiated viruses only induced modest or low levels of IFNLs in each HIE line. The IFNB1 proteins were barely detectable in all groups. Responses varied in HIE lines with J2 HIEs being more responsive. In summary, our RT-qPCR and Luminex assay studies validated the RNA-Seq data and confirmed that HIEs respond to HuNoV replication and poly(I:C) treatment by inducing a predominant type III IFN response.

Knowing that HIEs elicit an IFN response against HuNoV infection, we next tested whether exogenously added IFNs or IFNs activated by poly(I:C) attenuate GII.4 HuNoV replication. Pretreatment with type I (IFNA1 and IFNB1) or type III IFNs (IFNL1, IFNL2, and IFNL3) significantly attenuated GII.4 HuNoV replication compared with untreated control (Fig. 3A). Pretreatment with type III IFNs showed a greater effect on GII.4 replication compared to treatment at 1 hpi (SI Appendix, Fig. S3A). Pretreatment of HIEs with poly(I:C) at 24, 12, or 6 h before infection marginally attenuated GII.4 HuNoV replication compared with the untreated control at 24 hpi (Fig. 3B). By contrast, poly(I:C) treatment after inoculation had a reduced effect compared with pretreatment groups (Fig. 3B). ISG IFI44L expression at 1 hpi (Fig. 3 C and D) was inversely proportional to GII.4 HuNoV replication, suggesting that the exogenously activated type I or type III IFN pathways were able to elicit antiviral responses to attenuate GII.4 HuNoV replication. This basal elevation of the ISGs at 1 hpi by IFN pretreatment (Fig. 3 C and D) suggests the ISGs might impede HuNoV infection during the entry step. Overall, these data demonstrate GII.4 HuNoV replication is reduced by type I or type III IFN pathways exogenously activated prior to infection.

Since exogenously added IFNs reduce HuNoV replication, we tested whether attenuation of the IFN pathways could enhance HuNoV replication. We used Y136, a decoy receptor for type I and type III IFNs that is produced by the Yaba-like disease virus (47) to attenuate IFN pathways in HIEs. Replication of GII.4 HuNoV was not enhanced by Y136 treatment in J2 HIEs (Fig. 3E) even though the Y136 treatment abrogated the induction of ISG IFI44L (Fig. 3F). The attenuation of IFI44L indicates that IFN proteins are induced following GII.4 replication and are secreted from infected HIEs; however, although Y136 was able to attenuate IFN downstream responses via chelating secreted IFNs, GII.4 replication was not enhanced by chemically inhibiting the endogenous IFN responses in HIEs.

To further investigate whether endogenous type I or type III IFN pathways play a role in GII.4 HuNoV infection, we produced and evaluated whether the knockout (KO) of IFNAR1 (the type I IFN receptor subunit), IFNLR1 (the type III IFN receptor subunit), STAT1 (the master regulator of all three IFN pathways), or MAVS (the master regulator downstream of the TLR signaling pathways) alters the kinetics of GII.4 HuNoV replication in HIEs. First, we transduced J2 HIEs with lentiviral vectors carrying Streptococcus pyogenes-derived Cas9 (Sp-Cas9) and the small guide RNA (sgRNA) targeting the IFNAR1, IFNLR1, STAT1, or MAVS gene to generate the corresponding KO HIEs. Transduced HIEs were plated and expanded into single cell clones, and verified individually by genome sequencing, IFN treatment, and Western blotting. The IFNAR1-KO HIE contains a 6-bp deletion followed by a 1-bp insertion, indicating a 5-bp frameshift mutation in both alleles of the IFNAR1 gene (SI Appendix, Fig. S4A). The IFNLR1-KO HIE contains a 49-bp deletion as a frameshift mutation in both alleles of IFNLR1 gene (SI Appendix, Fig. S4C). Functional assessment of the KO lines showed that compared to the wild-type (WT) HIEs, treatment with the corresponding IFNs failed to induce ISGs in either IFNAR1-KO or IFNLR1-KO HIEs (SI Appendix, Fig. S4 B and D). Immunoblot and Western blot analyses showed that the STAT1-KO HIE does not express STAT1 protein (∼91-kDa band indicated with an asterisk and the splice variant of 84 kDa as a lower band) (SI Appendix, Fig. S4E) while the MAVS-KO HIE does not express MAVS protein (∼75-kDa band indicated with an asterisk) (SI Appendix, Fig. S4G). The basal level of IFI44L expression is lower in STAT1-KO HIEs compared with WT, and this ISG is not induced in mock-inoculated cells over 96 h (SI Appendix, Fig. S4F). Transfection of a STAT2 sgRNA-expressing plasmid into the STAT1-KO HIEs was used to produce a STAT1/STAT2-double-KO (DKO) HIE line, also verified by genome sequencing and Western blot analysis for STAT1 and STAT2 (SI Appendix, Fig. S4H) These CRISPR-Cas9-generated IFN pathway-deficient HIEs designated IFNAR1-KO, IFNLR1-KO, STAT1-KO, MAVS-KO, and STAT1/STAT2-DKO HIEs in a J2 WT background were then used to test for altered susceptibility to HuNoV infection and replication.

To assess the role of endogenous IFN pathways on GII.4 replication, we inoculated the WT and KO HIE lines with GII.4 and measured the kinetics of viral RNA replication at different times after infection by RT-qPCR. Infections were carried out at low multiplicity of infection (MOI) for these studies to allow for the evaluation of viral spreading over time. The replication kinetics of GII.4 HuNoV in all five KO HIEs were not enhanced compared with the WT HIEs (Fig. 4 A, C, and E). Significant induction of ISGs (as measured by IFI44L expression) in WT HIEs (Fig. 4 B, D, and F), was not altered by the loss of IFNAR1 but was decreased by the loss of IFNLR1 upon GII.4 infection (Fig. 4B), indicating HuNoV replication induces ISGs via type III IFN pathways. Further, the induction of ISGs was strongly attenuated in STAT1-KO and STAT1/2-DKO HIEs (Fig. 4 D and F, respectively) but only partially reduced in MAVS-KO HIEs upon GII.4 infection (Fig. 4D). These data indicate the induction of ISG IFI44L following GII.4 replication is STAT dependent, and partially MAVS dependent in HIEs.

Next, we tested the effect of exogenous IFN treatment on GII.4 replication in the KO HIEs. Pretreatment of IFNA1 or IFNL1 significantly reduced GII.4 replication in WT and MAVS-KO HIEs, but not in STAT1-KO HIEs (SI Appendix, Fig. S5A). The induction of ISG IFI44L by IFNs was significantly lower in STAT1-KO HIEs than in the WT HIEs (SI Appendix, Fig. S5B), indicating HuNoV is sensitive to the JAK/STAT pathway induced by exogenous IFNs. We then evaluated whether the treatment of Y136 in IFNAR1, IFNLR1, STAT1, or MAVS-KO HIEs (SI Appendix, Fig. S6) compared to untreated groups would enhance GII.4 replication. In all KO lines tested, treatment with Y136 failed to enhance GII.4 replication (SI Appendix, Fig. S6 A and C) even though the ISG IFI44L expression was attenuated after treatment (SI Appendix, Fig. S6 B and D). Pretreatment with Y136 for 24 h before and after inoculation in STAT1-KO and MAVS-KO HIEs also failed to further enhance GII.4 replication (SI Appendix, Fig. S6E) although there was no induction of IFI44L (SI Appendix, Fig. S6F).

In summary, although the five IFN pathway-deficient HIE lines show attenuated ISG induction upon GII.4 infection, viral RNA replication kinetics are not altered in these KO lines, suggesting GII.4 replication is not sensitive to endogenous IFN responses during replication in jejunal HIEs.

Unlike GII.4, GII.3 HuNoV requires bile acids to infect HIEs, indicating that different strains of HuNoVs have different entry requirements and may show varied responses to host cellular pathways. Therefore, we assessed the effects of IFN pathways on GII.3 (TCH04-577) replication. Similar to GII.4 HuNoVs, GII.3 also induces strong type III IFN secretion detected by the Luminex assay (Fig. 2 E and F). Pretreatment of WT HIEs with IFNA1, IFNB1, IFNL1, and IFNL3 reduced GII.3 replication (Fig. 5A). The basal expression of ISG IFI44L is higher at 1 hpi compared to untreated HIEs (Fig. 5B) and corresponds to the reduced replication in these treated cultures. Replication of GII.3 HuNoV also was not enhanced by Y136 treatment (Fig. 5C) although IFI44L expression was reduced (Fig. 5D). We next inoculated WT and KO HIE lines with GII.3 and measured viral replication kinetics to determine whether GII.3 replication is similarly insensitive to endogenous IFN pathways as seen for GII.4 HuNoV replication. In contrast to GII.4, GII.3 HuNoV replication was increased 10 to 100-fold in STAT1-KO and IFNAR1-KO HIEs, respectively, compared with replication in WT or other KO HIE lines (Fig. 6 A and C), and the enhanced replication was accompanied by reduced ISG responses (Fig. 6 B and D), indicating that the endogenous IFN pathway is a restricting factor that blunts GII.3 HuNoV replication in HIEs. Although ISG expression was reduced, GII.3 replication was not enhanced in STAT1/2-DKO HIEs (Fig. 6 E and F).

We next performed fluorescent focus assays (FFAs) to evaluate whether the increased levels of GII.3 RNA replication were associated with increased viral spreading in the HIE cultures. Significantly higher numbers of GII.3-positive cells were detected in STAT1-KO monolayers compared with WT monolayers (Fig. 7 A and B) and clusters of GII.3-positive cells increased over time (Fig. 7C and SI Appendix, Fig. S7B), with cells being lost from the monolayer by 96 hpi compared with infections in WT HIEs (SI Appendix, Fig. S7B). To evaluate whether STAT1-KO HIEs are more susceptible to GII.3 infection, we next determined the tissue culture infectious dose (TCID₅₀) for GII.3 infection in WT and STAT1-KO HIEs by the Reed–Muench method (48). The TCID₅₀ for GII.3 infection of WT J2 HIEs was 7.94 × 10⁴ GEs/well, and the TCID₅₀ for STAT1-KO HIEs was 3.42 × 10³ GEs/well, suggesting that the latter genetically modified HIE line is more permissive for GII.3 infection (Fig. 7D). These results indicate that differences in GII.3 replication seen in the time-course experiments could be due to enhanced susceptibility of KO cells during the first round of infection as well as increased virus spreading during the subsequent rounds of infection in the STAT1-KO HIEs. In contrast, STAT1-KO HIEs were not more susceptible to GII.4 as determined by FFA (SI Appendix, Fig. S7A). In conclusion, GII.3 HuNoV replication is sensitive to host endogenous IFN pathways and knockout of STAT1 in particular increases viral replication and spreading, a response that is distinct from the unresponsiveness of GII.4 replication even in cultures with abrogated host IFN responses.

We performed RNA-Seq analysis on two differentiated jejunal HIE lines (J2 and J11), inoculated with mock, gamma-irradiated GII.4 or infectious GII.4 at high MOI infection (1.8 × 10⁸ genome equivalents/well). Total cellular RNAs were extracted from 5-d differentiated, mock-inoculated monolayers at 3 hpi, and GII.4- or gGII.4-inoculated monolayers at 6, 10, and 24 hpi using RNeasy Mini Kit (Qiagen). The RNA integrity was assessed on a 2100 Bioanalyzer (Agilent). A double-stranded cDNA library was created using 500 ng of total RNA (measured by picogreen), preparing the fragments for hybridization onto a flowcell. An ERCC RNA Spike-In Control Mix 1 was added to each sample according to the manufacturer’s protocol. First, cytoplasmic and mitochondrial ribosomal RNA was removed from the total RNA samples using Illumina’s Epidemiology-Specific RiboZero rRNA removal protocol. Then cDNA was generated using the fragmented and rRNA-depleted total RNA using random primers. During second-strand synthesis, deoxythymidine triphosphate (dTTP) was replaced with deoxyuridine triphosphate (dUTP) which quenches the second strand during amplification, thereby achieving strand specificity. Libraries were created from the cDNA by first blunt ending the fragments, attaching an adenosine to the 3′ end, and finally ligating unique adapters to the ends for PCR amplification. Paired end sequencing (2 × 100 bp) was performed on an Illumina HiSeq 2500 for a depth of 30 million paired end reads (a total of 60 million reads) per RNA sample. The fastq files generated during sequencing were quality trimmed using Trim Galore (70, 71). Trimmed reads were then aligned to GRCh38 using Hisat2 and then a count matrix was generated from the aligned reads using featureCounts (72). The RNA-Seq data files and raw count matrix files were deposited in the Gene Expression Omnibus (GSE150918).

The sgRNA sequences for several target innate genes including IFN receptors and pathway mediators are indicated in SI Appendix, Table S4. Each sgRNA sequence was cloned into the LentiCRISPR v2 plasmid (Addgene, 52961) or Hygro-LentiCRISPR v2 plasmid (purchased from the Cell-Based Assay Screening Service [C-BASS] core at BCM) by BsmBI digestion and ligation as previously described (73).

Lentiviruses which express Cas9 and sgRNA were packaged by cotransfecting pMD2.G (Research Resource Identifiers [RRID]: Addgene_12259), psPAX2 (RRID: Addgene_12260), and the designed lentiCRISPR v2 with sgRNA into HEK-293FT cells (ATCC, PTA-5077, RRID: CVCL_6911) using a polyethylenimine (PEI) transfection reagent (Sigma). The supernatant was harvested every 24 h until 72 h posttransfection and concentrated by polyethylene glycol (PEG)-6000 and sodium chloride precipitation, suspended in complete medium with growth factors [CMGF(+)], aliquoted, and then stored at −80 °C. HIE KO lines were generated following transfection and single-cell selection as described in SI Appendix, SI Materials and Methods.

For RT-qPCR, fold changes in mRNA expression for host genes were determined using the delta-delta-Ct method relative to the values in control samples as indicated in the figure legends, after normalization to the housekeeping gene GAPDH (74). Results are presented as means ± SD, unless stated otherwise. Comparisons between groups/cells were made using the two-tailed Student’s t test with adjusted P values using the Holm method for multiple comparison unless stated otherwise. Statistical significance was denoted with * (P < 0.05) in the figures and figure legends with indicated biological replicate numbers. Statistical analysis was performed in GraphPad Prism 8.