Abstract
Rotaviruses (RV) cause acute severe diarrhea and are highly infectious in homologous host species. However, RV infection does not cause appreciable intestinal inflammation and homologous RV replication resists the antiviral effects of different interferon (IFN) types. In suckling mice, we found that homologous murine EW RV triggers sustained induction of IFNs, but not the amplification of interferon-stimulated genes (ISGs) or inflammatory transcripts. IFNs can trigger both apoptosis and inflammatory necroptotic cell death pathways. However, during RV infection intestinal epithelial cells (IECs) are severely impaired in the cleavage of different caspases regulating intrinsic and extrinsic cell death. Inhibition of caspase cleavage occurs in both RV-infected and -uninfected bystander IECs. In contrast, RV induces cleavage of receptor-interacting protein 1 (RIP1), a master regulator of cell death and survival decisions. Comparison of IFN-resistant (homologous) and -sensitive (heterologous) RV strains in mice demonstrates that successful RV infection restricts IFN receptor (IFNR)-amplified intestinal inflammatory cytokine secretion. LPS-mediated intestinal damage, driven by IFNR-induced inflammatory necroptosis, is also prevented during ongoing homologous RV infection. We find that the ability of RV to negate IFNR-mediated antiviral and inflammatory functions may involve a two-pronged attack. First, RV infection induces resistance to IFNAR1-directed STAT1 activation in bystander cells and inhibits IFNR-feedforward amplification. Second, ectopic stimulation of different IFNRs during RV infection results in efficient negative feedback transcription, eliminating several host innate responses to infection. Our study reveals a novel mechanism underlying the ability of homologous RV to dismantle the antiviral and inflammatory arms of several intestinal IFN types.
Significance Interferons mediate distinct functions including amplification of antiviral and inflammatory genes and inflammatory necroptosis of cells. Here we show that rotavirus (RV) infection of small intestinal epithelial cells (IECs) subverts IFNR-mediated antiviral and inflammatory functions. Although RV sensitizes both infected and uninfected bystander IECs to acute necroptosis, it prevents intestinal inflammation and induces cleavage of a key necroptosis regulator, RIP1K. During RV infection, uninfected bystander cells resist IFN-mediated STAT1 phosphorylation and feedforward transcription is blocked. In addition, ectopic stimulation of IFNRs during RV infection triggers negative-feedback signaling and eliminates several host antiviral and inflammatory responses. Thus, perturbation of IFNR feedback, recently linked to IFN resistance in cancer and inflammatory pathologies, is also exploited by an acute viral pathogen, RV.
Introduction
Rotaviruses are non-enveloped segmented dsRNA viruses that cause severe dehydrating diarrhea in many mammals, resulting in an estimated 200,000 human infant deaths annually(1). Acute RV infection in a homologous host (the natural species in which the RV strain replicates to high titer and spreads from individual to individual) is an extremely efficient process; severe diarrhea can be initiated by exposure to a very small infectious inoculum (2). The host innate interferon (IFN) response is a multi-pronged line of defense against viral pathogens as its induction can activate hundreds of different antiviral genes and trigger inflammatory signals – thus eliminating infected cells, limiting the spread of virus infection to uninfected bystander cells, and shaping both the magnitude and quality of subsequent adaptive immune response (3–5). The induction of all three major IFN types (α/β, γ, and λ) is mounted during homologous RV infection of the small bowel(6–10), but the antiviral effects of the IFN response on homologous RV replication are modest at best(10). Although poorly studied, IFN-mediated inflammatory responses are also likely reduced by RV infection, which is associated with only mild intestinal inflammation(11, 12) compared to other acute diarrheal pathogens(13). How RV efficiently subverts the ability of different IFN types to execute their antiviral and inflammatory functions(14) is not well understood.
On the other hand, infection with heterologous RVs (RV not typically isolated from the infected host species) generally results in severely restricted intestinal replication and little or no sustained transmission to uninfected hosts(11, 15, 16). In suckling mice, the intestinal replication of certain heterologous RV strains is substantially restricted (∼104-fold) but can be partially rescued by deleting IFN receptors or the downstream convergent transcription factor STAT1, thus unmasking the key role of IFN in suppressing RV infectivity in heterologous host species(7, 10, 11, 16). Depletion of IFNRs in suckling mice can also result in an acute biliary inflammatory disease (often lethal) following infection with the heterologous rhesus RV strain (RRV)(11). Thus, identifying the mechanisms by which homologous RVs suppress host IFN antiviral and inflammatory responses in the gut is vital for developing more effective strategies to combat RV-associated morbidities and mortalities, and in understanding the pathogenic potential of naturally circulating RV strains.
Results
Homologous RV infection restricts intestinal IFN amplification responses
We compared the intestinal viral replication and antiviral responses between a replication-restricted, IFN-sensitive simian strain (RRV), and a replication-competent, IFN-resistant murine RV strain (EW)(10) at 12 and 72 hpi by qRT-PCR (Fig 1A-B). Both RV strains comparably induced the three major IFN-type transcripts (IFN-α/β, IFN-γ, and IFN-λ, Fig. 1A) at 12 hpi. Similar induction of IFN pathways by both RV strains is supported at the protein level by measurement of intestinal IFN-γ, which is produced in both EW and RRV RV-infected mouse pups at 24 hpi (Fig. 1C). Several IFN-stimulated genes (ISGs) were also transcriptionally induced at 12 hpi with similar efficiencies by both EW and RRV (IRF7, ISG15, IFIH1, DDX58, IRF9, IFIT2, and MX2).
(A-B) Transcriptional analysis of intestinal transcriptional responses to murine EW and simian RRV RV strains. Three-to five-day old suckling mice (N=4 per group) were infected with murine EW or simian RRV RV strains (or mock), and at 12 h or 72 h small intestines were collected for transcriptional analysis of host (A) and RV (B) genes. (C) Intestinal secretion of IFN-γ was measured by Luminex following RV infection at 24 hpi as in (A) (N=3 samples per group, each sample pooled from 3 pups; ND, not detected).
Despite inducing IFNs and various ISG transcripts with a similar or lower efficiency, intestinal EW RV NSP5 transcript levels were markedly higher than RRV at 12 hpi (by >104-fold, Fig. 1B). Surprisingly, at 72 hpi the induction of several IFN types and ISG transcripts continued to be comparable in mice infected with either murine or simian RV (Fig. 1A and Fig. S1). However, EW RV replication at 72 hpi remained substantially higher than RRV RV (by ∼104-fold, Fig. 1B and Fig. S2). The only exception to these findings was IL28A (encoding IFN-λ2), which was significantly higher (∼10 fold) in the EW RV-infected pups at 72 hpi. In terms of kinetics, the induction of type I IFNs (IFNA4 and IFNA5) and IRF7 transcripts by both RV strains significantly increased between 12 to 72 hpi while induction of IFN type II (IFNG) and type III (IL28A) declined during the infection time course. Of note, the 12 hour-induction of several interferon-stimulated gene and inflammatory transcriptional responses in EW RV infected pups was virtually eliminated by 72 hpi (IL6, CXCL10, ISG20, IFI203, PELI1, IFIH1, DDX58, IRF9, IFIT2, 72 hpi, Fig. 1A) resembling IFN amplification responses to replication-incompetent RRV RV. Thus, intestinal IFN amplification responses to EW and RRV do not mirror the marked differences in actual viral replication between these two RV types. A likely explanation for these results is that despite substantial continued EW RV replication in the gut and the induction of IFN genes during the first 3 days of infection, homologous murine RV has suppressed the amplification of intestinal ISGs and inflammatory genes by 72 hpi.
Homologous rotavirus replication suppresses cleavage of several caspases in the intestine
The results above revealed that several IFNs continue to be transcriptionally induced in a sustained manner over 72 hrs during homologous RV infection. Exposure of cells to IFNs (IFN-α/β and -γ) in a sustained manner generally results in a sequential cleavage of the initiator caspase 8 and the executioner caspase 3, promoting apoptosis(17–20). How homologous RV contends with the potential for caspase-mediated cell death while inducing different intestinal IFN types (Fig. 1A, 1C) in a sustained manner is not well studied. Therefore, we next examined the expression of precursor (full-length, FL) and cleaved (Cl) forms of the initiator caspase 8, which is critical for extrinsic cell death(21), in murine EW RV-infected mice. Immunoblotting of small bowel IECs revealed that RV infection results in a substantially lowered ratio of Cl: FL caspase 8 isoforms (2.2 in mock vs. 0.5 in RV infected at 24 hpi, but not at 3 hpi) (Fig. 2A). The defective cleavage of caspase 8 (Fig.2A) would be expected to impair subsequent cleavage of the apoptotic executioner caspase 3, so we next examined the status of caspase 3 during RV infection. As anticipated, intestines from murine RV infected suckling mice also exhibited decreased cleavage of caspase 3 (Cl.: FL ratios of 1.4 and 0.4 in mock and EW RV infected, respectively, at 24 hpi but not 3hpi) (Fig. 2B). These unexpected results reveal that homologous RV infection inhibits intestinal caspases at the step of their zymogen activation.
(A-B) Analysis of caspase 8 and caspase 3 expression. Mouse pups were orally inoculated with EW (R) or mock-inoculated (M). At 3 and 24 hpi, bulk intestinal cells were harvested from whole small intestinal tissues and processed for SDS-PAGE. Gels were analyzed by western blot for precursor (FL) and cleaved (Cl.) isoforms of either caspase 8 (A) or caspase 3 (B), and other proteins. The normalized ratios of Cl. to FL caspases determined by densitometry are indicated below individual panels. (C) FACS analysis of caspase 8 and caspase 3 cleavage. Summary of flow cytometry analysis of CD45-CD44-ESA+ IECs following intracellular staining for KI67, and either cleaved caspase 8 or cleaved caspase 3, from EW RV infected or control pups at 24 hpi. (D) The percentage of IECs (CD45- CD44-ESA+) displaying cleaved caspase 8 or cleaved caspase 3 in mock and EW-infected mice at 24 hpi averaged from three experiments; error bars represent the standard errors of the means. (E) Intracellular staining for RV structural protein (VP6) and either cleaved caspase 8 (top panel) or cleaved caspase 3 (bottom panel) in CD45-CD44- ESA+ IECs following EW RV or mock infection at 24 hpi. (F) Immunoblotting analysis of small intestinal tissue from mouse pups infected with EW, RRV, or mock, and harvested at the indicated times. (G) Analysis of caspase 9 cleavage in cultured NIH 3T3 fibroblasts infected with an IFN-stimulatory RV strain, UK. 3T3 cells were infected with bovine UK RV (or mock) for the times indicated and then analyzed by immunoblotting. Numbers below blots indicate normalized ratios of Cl. to FL caspase 9. (H) Human HT-29 IECs were infected with porcine SB1-A RV (or mock) for the times indicated and analyzed by immunoblotting. Numbers below blots indicate normalized ratios of Cl. to FL caspase 8. (I) HT-29 cells infected with the RV strains indicated were analyzed for RIP1K, RIP3K, and VP6 at either 4 or 12 hpi by immunoblotting. (J) HT-29 cells were infected with SB1-A RV (or mock), or treated with TNF-alpha (10ng/ml) and cycloheximide (10μg/ml) for 4 h, and then analyzed for cell adherence or phosphatidyl serine marker expression. Data shown is representative of two or more independent experiments.
Cleavage of caspases 8 and 3 is inhibited in both rotavirus-infected and uninfected bystander villous enterocytes
Having observed that RV infection in vivo results in impaired caspase-8 and −3 processing in IECs, we next examined whether this process occurs specifically in the small intestinal villous epithelium(7). Using flow cytometry, we gated live villous IECs, using a previously described combination of cell surface markers (i.e. CD45-ESA+CD44-)(7, 22), and then measured expression of cleaved forms of either caspase 8 or caspase 3 in the RV infected and mock control pups (Fig. 2C). These experiments confirmed a substantial decrease in caspase 8 and caspase 3 cleavage in small intestinal villous enterocytes during RV infection (∼75% decreased caspase 8 and caspase 3 cleavage following infection).
We next examined whether caspase cleavage inhibition occurs solely in RV-infected mature villous IECs at 24 hpi using a panel of antibodies to define cell type, caspase activity, and the RV VP6 positivity (as a marker for RV infection status of cells) (Fig. 2D). Remarkably, efficient suppression of caspase-8 cleavage occurs in both VP6+ (infected), and in VP6- (i.e. uninfected ‘bystander’) mature villous cell population. The relative ratios of cells staining positive for cleaved caspase 8 in the different populations were: ∼3.0 (mock), 0.2 (bystanders), and 0.6 (RV infected) (Fig. 2D, top panels). The efficient inhibitory effect of RV infection on uninfected bystander (i.e. VP6-) cells was also observed following analysis of caspase 3 cleavage using a similar approach (Fig. 2D, relative caspase 3 cleavage ratios were 4.0, 0.3, and 0.9 in mock, VP6- bystander, and VP6+ RV-infected cells, respectively; Fig. 2D, bottom panels).
RV inhibition of caspase 8 and 3 cleavage in both RV infected and uninfected small bowel IECs would be expected to exert a significant impact on intestinal cell viability. Therefore, we examined cell death in two intestinal cellular compartments. First, viability analysis of total small intestinal single cell suspensions (i.e. ‘bulk’ level) revealed a modest overall decrease in intestinal cell death during EW RV infection (this decline was not statistically significant) (Fig. S3A-B). Second, a significant reduction in cell death was observed by TUNEL staining of cells in the intestinal non-hematopoietic (i.e. CD45-) compartment, which includes both epithelial and stromal cells(7), but not hematopoietic cells (Fig. S3C, p<0.001). Collectively, these results demonstrate that RV infection in vivo effectively prevents caspase cleavage and suppresses intestinal cell death in both infected and non-infected bystander cells.
In contrast to caspase 8-mediated extrinsic apoptosis, the initiator caspase 9 is cleaved during intrinsic apoptosis (23). Interestingly, suckling mice infected with EW RV also exhibited impaired cleavage of caspase 9 (Fig. 2E), as indicated by an accumulation of intestinal FL caspase 9 at 16hpi, but not 6hpi, when compared to either RRV or mock-infected controls (Fig. 2E). To examine the role of IFN in this process, we determined whether caspase 9 cleavage is blocked in cultured 3T3 fibroblasts infected with the bovine RV strain UK, which we previously showed triggers MAVS-dependent secretion of IFN-β in this cell line(24, 25). Infection of 3T3 cells with UK RV for 9 h suppressed caspase 9 cleavage and mediated an accumulation of the FL-caspase 9 zymogen. In contrast, increased Cl-caspase 9 and decreased FL-caspase 9 expression occurred at 16 hpi (Fig. 2F). These results confirmed that inhibitory effects of RV on intrinsic initiator caspase 9 activity do not require inhibition of IFN secretion and further indicated that caspase cleavage could be regulated temporally during RV infection.
Kinetic regulation of caspases by RV was further examined in an intestine-derived cell line by time-course analysis of caspase 8 cleavage in cultured HT-29 cells following infection with porcine RV strain SB1-A, which we previously showed prevented IFN-β secretion in this cell line(26). Compared to a mock-infected control, infection with SB1-A RV inhibited caspase 8 cleavage starting at 4 hpi but with maximal effect occurring at 8 hpi (Fig. 2G). Collectively, these studies indicate that RV broadly inhibits initiatory caspase cleavage events essential for either extrinsic or intrinsic cell death outcomes and regulates these processes temporally during the course of infection. The results also demonstrated that the RV regulatory effects on apoptotic signaling both in vitro (Fig. 2F-G) and in vivo (Fig. 2A-E) occur regardless of its ability to suppress IFN induction.
Several RV strains induce cleavage of the cell death regulator RIP1K
In response to sustained IFN exposure, caspase 8 cleavage results in apoptotic epithelial cell shedding with typically mild inflammatory consequences(20). In contrast, in cells that are defective in caspase 8 cleavage (such as IECs during EW RV infection in vivo, Fig. 2A-C), IFN stimulation generally directs a rapid and precipitous form of cell death by necroptosis that is acutely inflammatory(17, 27). Receptor interacting protein 1 kinase (RIP1K) is a crucial adaptor dual-function kinase at the crossroads of a cell’s decision to live or die in response to either apoptotic or necroptotic stimuli(28, 29). These opposing functions of RIP1K are primarily regulated by a proteolytic mechanism(30, 31). While overexpression of full-length human RIP1K triggers both apoptosis and necroptosis, caspase-mediated cleavage of the ∼75 kDa RIP1K (p75) at multiple sites within the kinase domain (KD) and interaction domain (ID) generates shorter fragments derived either from the RIP1K N-terminal (51 kDa, 37 kDa, and 25 kDa) or the C-terminal (39 kDa) regions. Both p51 and p39 RIP1K potently inhibit p75 RIP1K function and caspase 8 cleavage, preventing apoptotic and necroptotic cell death in response to sustained inflammatory stimuli(31).
As RV regulates caspase 8 cleavage in a temporal manner during infection (Fig. 2F-G), we examined whether RIP1K cleavage is also subjected to RV regulation. HT-29 cells were infected with representative RV strains differing in their modes of IFN regulation in vitro (IRF3-degrading simian RRV, NF-kB inhibitory porcine SB1-A and human WI-61, and NSP1-mutant simian SA11-5S RV strain that stimulates IFN-β secretion(32)) for either 4 or 12 h, and then analyzed by immunoblotting using an anti-(N-terminal) RIP1K antibody (Fig. 2H). When compared to mock-infected HT-29 cells, infection with the different RV strains resulted in enhanced p75 RIP1K expression at 4 hpi with varying efficiencies, and this increased expression was also be detected later during infection (at 12 hpi). In addition, RV infection also generated several shorter p51/p37/p25 RIP1K cleavage fragments, and strain-specific differences were apparent in the levels of these cleavage fragments. Notably, all RV strains including SA11-5S, triggered p51-RIP1K cleavage starting early during infection, with the exception of the human WI-61 RV strain infection where increased p51 RIP1K expression is primarily detectable by 12hpi. Moreover, simian (RRV and SA11-5S) and human (WI-61) RV strains induced additional cleavage of RIP1K (p37/p25 RIP1K - RRV and SA11-5S; p37 RIP1K - WI-61). Interestingly, SB1-A RV induces increased expression of both the (pro-death) full-length p75 RIP1K and the shorter (inhibitory) p51 RIP1K isoforms (Fig 2H, lanes 5-6). We thus examined how functionally antagonistic RIP1K (p75 and p51) fragments that are expressed during SB1-A infection influence cell death. We found that both cell adherence and phosphatidylserine expression (a marker for apoptotic cells) at 16 hpi in SB1-A RV-infected cells resembled mock-infected HT-29 cells, suggesting that increased p75 RIP1K expression does not trigger cell death. In contrast, HT-29 cells treated with a cocktail of tumor necrosis factor and cycloheximide exhibited substantial loss of cell adherence and increased apoptotic staining (Fig. 2I). These results reveal that RIP1K accumulation and its proteolysis are regulated during RV infection, and a shorter dominant-negative p51 regulatory RIP1K isoform is generated by many RVs early in infection that could potentially impair apoptosis and necroptosis.
Rotavirus restricts intestinal IFN-mediated inflammatory responses
In addition to promoting IFNR-mediated and caspase-dependent inflammatory death, different IFN types can also drive the production of inflammatory cytokines by an IFNR-directed positive feed-forward gene amplification process(3, 33). Therefore, we next examined whether, in addition to inhibiting caspase activation in vivo, homologous RV infection of suckling mice also suppresses the synthesis of IFNR-stimulated inflammatory target gene products. To address this possibility, we used an experimental design similar to Fig. 1A to measure production levels of different intestinal cytokines at 1 dpi in suckling mice infected with either homologous murine (EW strain) or heterologous simian (RRV strain) RVs (Fig. 3A). These studies reveal significant increases in intestinal production of a variety of cytokines in response to either homologous EW and/or heterologous RRV RV infection (RANTES, IP-10, IL1-A, EOTAXIN, GRO-A, MCP-3, MIP-1B, IL-4, GM-CSF, IL-15, LIF; Fig. 3A). Of these, a set of cytokines (MCP-3, IL-15, LIF, IL-10, MCP-1, MIP-2α, and IL-18) were increased comparably in response to either replication-competent murine (EW) or replication-restricted simian (RRV) RV. In contrast, a second set of predominantly pro-inflammatory cytokines (RANTES, IP-10, IL1-α, Eotaxin, GRO-α, MIP-1b, IL-6, IL-4, GM-CSF, and IL-17A) were primarily induced in the heterologous simian RRV-infected pups. These cytokines correlate with the poor intestinal replication of heterologous RVs in mice, although whether they contribute in any manner to RRV growth restriction remains to be determined. Strikingly, several differentially secreted cytokines in this set are reported to be IFN-α/β, IFN-γ, and/or IFN-λ mediated effectors (i.e. RANTES(34–36), IP-10(35, 37), Eotaxin(38), GRO-α(39, 40), and MIP-1b(36, 41, 42)). This suggests that successful intestinal replication of homologous murine RV is accompanied by a suppression of IFNR-amplified inflammatory cytokine responses(14).
(A) Secretion of intestinal cytokines during RV infection of suckling mice. Suckling mice were infected as in Figure 1, and small intestinal homogenates were analyzed at 24 hpi for the indicated cytokine protein levels (N=3 samples per group, each sample pooled from 3 pups and measured in duplicate). (B) Schematic of reported endotoxin-mediated inflammation that is IFNR-stimulated and negatively regulated by cleaved caspase 8. (C) Effect of homologous RV infection in suckling mice on endotoxin-mediated intestinal damage. Suckling mice were mock- or EW RV-infected and 3 d later were parenterally administered purified endotoxin (15μg/pup) for 6 h before harvesting small intestines for histological examination by hematoxylin and eosin staining. Data shown is representative of 2 pups.
In order to confirm whether homologous murine RV suppresses IFNR-amplified inflammation, we used a suckling mouse model of endotoxemia, in which LPS-induced inflammation and severe intestinal damage are reported to occur primarily through IFNR-mediated inflammatory cytokine expression(43–45) and necroptotic cell death(17, 27, 46). In prior work, we found that murine RV infection of the small bowel protected suckling mice from endotoxin-induced mortality(8). However, whether or not EW RV replication exerts a local protective effect in the LPS-challenged murine small bowel was not evaluated. Here, we specifically examined if EW RV infection prevents small intestinal inflammatory damage during endotoxemia (Fig 3B). Suckling mice orally infected with EW RV (or mock) for 72h were systemically administered endotoxin and resulting intestinal damage examined 6h later. In mock-infected mice, parenteral administration of endotoxin led to severe intestinal necrosis, injury, and tissue destruction after 6h, as expected (Fig. 3C)(47). In contrast, suckling mice infected with murine EW RV (for 3d) prior to endotoxin challenge exhibit minimal small intestinal damage (Fig. 3C). These results are consistent with the conclusion that homologous RV effectively prevents IFN receptor-amplified small intestinal inflammatory damage due to necroptosis or inflammatory gene expression. The results also indicate, somewhat surprisingly, that efficient caspase 8 inhibition in the small intestine during EW RV-infection (Fig. 2, C, D) does not sensitize mice to LPS-induced inflammation.
Different RV strains potentiate IFN resistance in uninfected HT-29 bystander cells
Cleaved caspase 8 is a key negative regulator of IFN (and LPS)-mediated necroptosis and acute inflammation(17), and IEC-specific caspase 8 deletion results in spontaneous intestinal inflammation(48), possibly triggered by constitutive tonic IFN signaling(27, 46). Inhibition of caspase 8 cleavage is thus considered to be a potent priming event for acute IFN-directed necroptotic inflammation in the gut(17, 21, 46). Although murine RV infection reduces the cleavage of caspase 8 isoforms in both infected and bystander IECs (Fig. 2D), and induces various intestinal IFNs in vivo(7, 9) (Fig. 1A), these sequelae do not translate into appreciable ISG amplification, intestinal inflammation or tissue damage (Figs. 1A, 3A-C), cell death (Fig. S3), or viral growth restriction (Fig. 1B)(7, 10, 11). To better understand this apparent paradox, we examined whether RV infection might block IFN receptor (IFNR) signaling in both infected and bystander cells in order to effectively prevent IFN effector functions. We previously reported that in cell culture, several RV strains mediate efficient lysosomal degradation of endogenous IFNAR1, IFNGR1, and IFNLR1 subunits in individual RV+ infected cells, but not in uninfected RV-bystander cells(8).
We postulated that RV might employ a different strategy to also disable IFNR functions in uninfected bystander cells. In previous in vitro studies, we showed that a porcine RV strain (SB1-A) (but not several other RV strains) inhibited IFNR-mediated STAT1-Y701 phosphorylation in RV-bystander cells(26), following a relatively short (i.e. 30-minute) stimulation with high concentrations of purified IFNβ (i.e. up to 2,000 IU/ml). Here, we more extensively examined whether other RV strains might also suppress IFN responsiveness in bystander cells, but with a lower efficiency than the SB1-A porcine strain. This re-examination used an extended administration time for exogenous IFN stimulation prior to STAT1-pY701 analysis, thus better accounting for possible lowered efficiencies of IFNR inhibition by other RV strains (Fig. 4A). Human HT-29 cells were infected with several RV strains (simian RRV and SA11, porcine SB1-A, bovine UK, and simian SA11 mono-reassortant SOF, encoding a porcine NSP1) for 12h and cells subsequently stimulated for 6h with purified IFN-β (500 IU/ml) before analysis of STAT1-pY701 expression in individual RV-VP6+ and VP6- HT-29 cells. Stimulation of mock-infected cells with IFN-β activated STAT1-Y701 phosphorylation and this activation was effectively inhibited by all tested RV strains in RV infected-VP6+ cells as we previously reported(26). More importantly, STAT1-Y701 activation was also significantly inhibited in RV-VP6- bystander cells by the other RV strains (Fig. 4A). These results thus reveal that all the RV strains tested inhibit IFN receptor signaling in bystander cellular populations in vitro, although their relative efficiencies appear to vary across the strains.
(A) Different RV strains trigger resistance to IFN-mediated STAT1-Y701 phosphorylation in bystander cells. HT-29 cells were infected with the RV strains indicated and 12 h later, stimulated with 500 IU/ml of purified IFNβ for 6 h before FACS analysis. Numbers in red indicate fold increases in STAT1-pY701 following IFN stimulation in RV VP6+ (infected) and VP6-(bystander) populations. Data is representative of two independent experiments. (B-C) Effect of RV infection on dsRNA- and IFN receptor-mediated transcriptional activity. Schematic at the top depicts the experimental approach used. HT-29 cells were infected with RV (RRV strain, at an MOI of ∼0.5) or mock-infected and at 6 hpi, cells were either transfected with 2μg of purified dsRNA or were stimulated with 500 IU/ml of purified IFNβ. At 12 hpi cells were analyzed by qRT-PCR for expression of host (B) and RV (C) transcripts. Data represents two independent experiments. (D) Schematic summarizing differential inhibition of dsRNA- and IFN receptor-stimulated transcription by RV. Dashed red line indicates possible negative feedback signaling upon ectopic IFNR stimulation during RV infection, solid red lines indicate RV inhibition of IFNRs by receptor degradation (in RV-infected cells(8)) and by blockage of IFNR-mediated STAT1 activation and transcription (in RV bystanders).
As different IFNR types normally expressed in RV bystander cells(8) are likely to be functionally impaired (Fig. 4A), we next examined how RV inhibition of IFNR-mediated STAT1 phosphorylation influences downstream antiviral and inflammatory gene transcriptional functions. Several ISGs and inflammatory transcripts upregulated by IFNR-ligand stimulation can also be induced if cells are stimulated with intracellular dsRNA, primarily through NF-kB and IRF3 signaling pathways(49). Thus, here we compared the ability of RV to regulate cellular ISG and inflammatory transcriptional amplification following stimulation with either exogenous IFN-β or intracellular long dsRNA (Fig. 4B). Control HT-29 cells stimulated with IFN-β or dsRNA triggered both overlapping (TNF, TNFAIP3, HERC5, IFI6, IFIT3, IFIT1, IFNB1, RSAD2) and unique (TLR2, IL26, IL21, SERPING by IFN-β; IL29, IFNG by dsRNA only) sets of transcriptional responses. Regardless of how transcriptional stimulation was carried out, RV infection effectively prevented further transcriptional upregulation of HERC5, IFNG, IL22, IL21, IFI6, IFIT1, IFIT3, and RSAD2. In contrast, we identified a set of pro-inflammatory transcripts that was specifically blocked by RV in response to exogenous IFN-β, but not to intracellular dsRNA (TLR2, IL26, TNF, IRF8, TNFAIP3, IL29, IL1B). Among the transcripts measured, only SERPING was efficiently induced by IFNAR1 stimulation during RV infection, though with a lowered efficiency than in uninfected HT-29 cells. The stimulus-specific RV inhibition of transcription was not due to RV replication sensitivity, since it was unaltered under either the IFN-β or dsRNA stimulation conditions used (Fig. 4C). In addition, the induction of certain transcripts was specifically repressed by ectopic IFNAR1 stimulation during RV infection (HERC5, IFNG, and IL22, statistically significant p<0.01; IRF8, IL29, and IL1B, not statistically significant). These results demonstrate that RV effectively inhibits IFN-β-receptor mediated transcription of selected ISGs and inflammatory genes. In addition, they indicate that ectopic IFNAR1 stimulation during the course of RV infection can elicit negative feedback inhibition of transcription (Fig. 4B, 4D). Such feedback-based regulation(50) is likely to be particularly relevant to antiviral and inflammatory functions in uninfected RV-bystander cells that, unlike a RV+ infected cell, express IFNRs (8) (Fig. 4D).
Murine RV infection re-purposes IFNR-directed intestinal antiviral and inflammatory signaling
The stimulation of IFNGR1 and IFNAR1 is reported to suppress STAT1 via negative feedback signaling under certain conditions, fostering IFN-resistance in the context of cancer and chronic viral infection(33, 50). We examined whether intestinal IFNR-mediated transcription is altered by homologous RV in vivo to suppress IFN antiviral and inflammatory signaling (Fig. 1A and 3A). Following infection of suckling mice with murine EW RV for 12h, we parenterally administered purified IFN-γ (or PBS), and collected small intestines after an additional 12h for transcriptional analysis (Fig. 5A). In mock-infected control pups, IFN-γ stimulation upregulated several intestinal transcripts (STAT1, ISG15, IFI204, IL10RA, IL1RN, statistically significant; S100A8, BIRC5, REG3G, not significant; Figs. 5B-5D). In contrast, transcription of CCL5 was significantly repressed by ectopic IFNGR1 stimulation. Similar transcriptional responses are also upregulated by EW RV infection, but with magnitudes significantly larger than obtained by IFN-γ stimulation alone. A set of antiviral (DHX58, DDX58, MX2) and inflammatory (TRAIL, NOS2, CXCL10) transcripts was significantly upregulated by RV, but not by IFN-γ stimulation. Remarkably, ectopic stimulation of the IFNGR1 in EW RV-infected mice either significantly suppressed (STAT1, ISG15, TRAIL, and CXCL10) or completely eliminated (DHX58, DDX58, MX2, CCL5, and NOS2) these intestinal antiviral and inflammatory responses (Fig 5B-C). In contrast, IFN-γ stimulated, EW RV-infected pups upregulated anti-inflammatory and proliferative gene transcription (S100A8, BIRC5, ARG2, REG3G, MKI67, and IL1RN) (Fig. 5D). These transcripts were not significantly induced by either IFN-γ stimulation or by EW RV-infection alone, and synergistically increased when IFNGR1 stimulation was carried out during EW RV infection. To ascertain whether the effects of IFN stimulation in EW RV-infected pups could be explained by altered RV replication levels, we measured intestinal RV RNA by qRT-PCR. The results demonstrate that EW RV replication was not significantly altered following IFN-γ stimulation (Fig. 5E). These findings suggest that during homologous RV infection, intestinal IFNR function can be diverted towards suppressing antiviral and inflammatory responses and promoting an anti-inflammatory program, revealing a potential mechanism by which uninfected RV bystander cells resist the antiviral and inflammatory effects of IFN secreted by IEC and enteric hematopoietic cells(7, 51).
(A) Schematic of the experimental approach used. (B-E) Transcriptional analysis of different assigned categories of intestinal transcripts following ectopic stimulation of the type II IFN receptor. Mice were infected with EW RV (or mock) (N=3-5 pups per group) and at 12 hpi were administered purified murine IFNγ (1μg/pup, or PBS, i.p.). Twelve hours later mice were sacrificed, and small intestinal RNA was analyzed by qRT-PCR for different host (B-D) or RV NSP5 gene (E) transcripts.
Discussion
Interferons are pleiotropic innate immune effectors that constitute a primary line of defense against viral pathogens. By signaling through their cognate receptors, different IFN types potentiate the expression of numerous and multi-functional antiviral and inflammatory genes(4, 5). In addition, IFNR stimulation can also induce apoptotic and necroptotic forms of cell death(17–19, 46), possibly to limit viral replication and spread. In its homologous host species, RV replication is only marginally restricted by the induction of different IFNs and robust viral replication is not associated with substantial inflammation(7, 8, 10, 11, 16). However, heterologous RV replication is highly restricted by each of the three major IFN types(10). How homologous RVs manage to avoid the antiviral and inflammatory actions of IFNs is still not fully understood.
We found that in suckling mice, mucosal innate immune transcriptional responses to the virulent and highly infectious murine EW RV strain closely resembled those to a replication-restricted heterologous simian strain (Fig.1A). The similarity in innate responses - including IFN type transcription, IFN-γ production, and amplification of ISGs or inflammatory gene transcripts over time (Fig. 1C) - did not correlate well with the very different replication levels of the EW vs. RRV (approximately 10,000-fold difference, Fig. 1B). Different IFN type transcripts remained significantly elevated at 72 hpi in EW RV-infected pups, indicative of a sustained innate immune response to ongoing RV replication. However, we did not observe amplification of several ISGs (i.e. ISG15, ISG20, IFI203, IFIH1, DDX58, IRF9, and MX2 (Fig. 1A) during the later stages of EW infection. These results indicated that although homologous RV replication and IFN induction occur in a sustained manner during infection, this does not effectively lead to sustained amplification of multiple antiviral and inflammatory genes.
In response to sustained IFN exposure, apoptosis can occur via proteolytic activation of the initiator caspase 8(19, 20). IFN-stimulated apoptosis is a relatively slow process, likely mediated by IFN-stimulated expression of intermediary apoptotic ligands (e.g., FasL and TRAIL)(18–20), and typically has mild inflammatory consequences. In addition to triggering apoptosis, proteolytic cleavage of pro-caspase 8 can suppress necroptosis - a more inflammatory form of cell death(17, 52). Necroptosis is rapidly activated when cells are deficient for caspase 8, or in the presence of caspase inhibitors, following exposure to inflammatory ligands like TNF-α or FasL(21). Ligand stimulation of the IFN-α/β and -γ receptors can also precipitate rapid cellular necroptosis, leading to acute inflammation(17, 21). This process is mediated by currently unknown but critical IFNR-dependent effectors (termed constitutively IFN-regulated effectors of necroptosis, CIRENs)(46). In addition, IFNs can amplify the expression of inflammatory cytokine genes encoding IFN-sensitive response promoter elements. Both IFN-mediated necroptosis and inflammatory gene expression require an intact IFNR signaling pathway and are critical mediators of acute intestinal inflammation in response to endotoxin(17, 21, 43, 46, 53). Of note, before this set of experiments, the regulated cell death and inflammatory arms of IFN receptor signaling had not been well studied in the context of RV infection.
Our results suggest that homologous RV infection initiates a two-pronged set of events to subvert and delay IFN-mediated cell death functions and potentially allow more time for RV replication within IECs. An estimated 1400 cells/villous/day are normally shed to maintain intestinal homeostasis in an adult mouse(54). We found that a key step in epithelial cell turnover, the cleavage of caspases 8 and 3, was significantly inhibited during homologous RV infection in both virus-infected and uninfected bystander mature villous IECs (Fig. 2, S3). This blockage of caspase activation may serve to prevent or diminish IFN-accelerated apoptosis and thus enhance viral replication and spread to uninfected IECs. The inhibition of caspase 3 activity could itself result from RV-inhibited cleavage of the upstream initiator caspase 8 (Fig. 2A, 2C-D), which mediates caspase 3 activation by the extrinsic death pathway and is inhibited in both RV-infected and bystander IECs (Fig. 2D). In addition, homologous murine RV also hampers homeostatic cleavage of another initiator caspase (caspase 9) that is crucial for executing intrinsic cell death pathways. The ability of RV to regulate multiple types of caspases in vivo is novel and, to the best of our knowledge, represents an ingenious viral strategy to broadly regulate and delay several effectors of intestinal cell death. The inhibition of caspase cleavage also occurs during RV infection in vitro and is temporally regulated (Fig. 2F, G).
The mechanism by which cleavage of the different types of caspases is regulated by RV is presently unknown. Nevertheless, analysis of a key caspase 8-regulated signaling effector, RIP1K(28, 30, 53), in RV-infected HT-29 cells provides possible insight into underlying mechanisms. Specifically, we found that in vitro different strains of RV promoted the cleavage of full-length p75 RIP1K into shorter fragments (i.e. p51, p37, and p25) (Fig. 2H). The cleavage of RIP1K into the p51 fragment is caspase 8-mediated and generates a negative regulator that can prevent apoptotic and necroptotic cell death in cells constantly exposed to inflammatory stimuli(31). In contrast to the regulation of RIP1K by caspase 8, under certain conditions (e.g. ER stress) RIP1K is required for proper caspase 8 cleavage and apoptosis(55). This suggests the existence of a regulatory feedback of RIP1K on caspase 8 activity. One possible explanation of our findings could be that one or more RIP1K cleavage products generated early during RV infection (Fig. 2H) regulate caspase 8 cleavage and cell death during ER stress conditions that occur later in RV infection(56–59). Time-course analyses of caspase 8 and RIP1K cleavage during RV infection (Fig 2G-H) indicated that both these processes are impaired temporally, but do not mediate appreciable levels of cell death or apoptotic markers even at relatively late times of infection (Fig. 2I). It will be interesting to determine whether these RIP1K regulatory processes occur in RV bystander cells and to identify the viral and host effectors involved.
Regardless of the exact underlying mechanisms, impaired caspase 8 cleavage in both RV infected and bystander IECs (Fig. 2D) creates a potentially problematic situation in the intestinal epithelium(21). IEC-specific caspase 8 deletion by itself is sufficient to trigger spontaneous intestinal inflammatory lesions, resulting in terminal ileitis and colitis(48, 60). As noted above, caspase 8 cleavage inhibition is also a key priming event that triggers IFN receptor-mediated necroptosis, acute inflammation, and endotoxin-induced tissue damage(17-19, 27, 46, 53, 60). Paradoxically, we found that murine EW RV infection significantly protected suckling mice from subsequent lethal endotoxemia (100% vs. 0% survival, at 24h post-endotoxin exposure)(8). Since endotoxin-mediated damage is potentiated by IFN receptor-stimulated inflammatory gene expression(44) and necroptotic cell death(27), it seemed plausible to postulate that homologous RV infection locally represses IFN-mediated inflammatory functions in the small intestine. Our histopathologic findings (Fig. 3C) confirm that, at the local small bowel level, EW RV-infected murine pups are substantially protected from severe inflammatory damage due to exogenous endotoxin exposure. An interesting question raised by these findings, which we are currently examining, is whether similar protective effects of RV on inflammation can occur at extra-intestinal sites. Homologous RV infection also blocked the IFN receptor-stimulated inflammatory gene expression program, as indicated by the finding that intestinal secretion of several IFN-stimulated inflammatory cytokines (RANTES(34–36), IP-10(35, 37), Eotaxin(38), GRO-α(39, 40), MIP-1b(36, 41, 42)) were all significantly lower in the murine RV-infected pups compared to a replication-restricted heterologous simian RV, despite substantially higher replication levels of the murine strain (Fig. 3A).
Rotavirus efficiently degrades multiple IFNR types in infected cells(8), which is predicted to directly limit both IFN-mediated inflammatory cell death and gene amplification by depleting surface IFNR expression. The RV protein(s) that mediate IFNR degradation have not been identified, although the RV-encoded non-structural protein NSP1 would seem to be the most likely candidate due to its ability to inhibit STAT1 signaling and cause the degradation of both IRF3 and bTrCP(26, 32, 61). In this study we examined whether IFNR function in bystander cells was also inhibited by RV. We previously reported that a porcine RV strain (SB1-A) prevents STAT1-Y701 phosphorylation in uninfected RV-bystander HT-29 cells following a relatively short (i.e. 30 min) ectopic IFN exposure(26). Whether non-porcine RVs also prevent IFN-mediated signaling in bystander cells was not explored further using altered IFN stimulation conditions. Here, using a more prolonged IFN stimulation (i.e. 6 h) combined with FACS analysis, we demonstrate that several non-porcine RV strains, shown previously to mediate IFNR degradation in infected cells, also induce effective resistance to IFN-directed STAT1-pY701 in bystander cells (Fig. 4A).
The ability of various RV strains to block IFN signaling in bystander cells is a second prong of the viral strategy to subvert antiviral and inflammatory functions mediated through the IFNRs. The effector functions of different IFNs are precisely titrated by a sensitive receptor-centered negative-feedback modulating mechanism(62–64), ensuring that excessive inflammatory damage to the host does not occur by uncontrolled feed-forward IFNR signaling. Dysregulation of this equilibrium can lead to either IFN resistance exemplified by cancer cells(65), or to acute inflammatory pathologies(66),such as sepsis. Previously, we demonstrated that parenteral IFNγ administration results in efficient STAT1-pY701 activation in small intestinal epithelial cells, a process that is significantly suppressed during homologous EW RV infection in suckling mice(8). A similar perturbation of IFNR-mediated STAT1 activation by murine EW RV occurred if purified IFN-β or IFN-λ was used to stimulate their cognate receptors(8). Earlier studies also identified a STAT1-independent pathway that regulates IFNγ-stimulated gene expression(67–70). More recently, the production of inflammatory cytokines in response to toll-like receptor (TLR) stimulation was significantly suppressed by ectopic IFNγ and IFNβ stimulation of STAT1-null cells, by re-purposing IFNGR1 and IFNAR1 signaling toward feedback inhibition(50). In this study, we examined whether such alternate IFNR signaling could also be usurped by RV to broadly suppress innate antiviral and inflammatory responses.
Several transcripts upregulated during the IFNR feed-forward amplification phase can also be efficiently induced by intracellular dsRNA through signal transduction of different pattern recognition receptors, including the RIG-I-like-receptors (RLRs), toll-like-receptors (TLRs), and the dsRNA-dependent protein kinase (PKR)(3). We therefore compared the ability of dsRNA- and IFNβ-stimulation to upregulate antiviral and inflammatory gene expression during RV infection of HT-29 cells (Fig. 4B). As RV+ infected HT-29 cells are efficiently depleted of different IFNRs, an assumption in our experiment was that IFNβ-stimulated transcription occurs via the IFNRs expressed on uninfected RV-bystanders, although we did not directly demonstrate this. The results clearly indicated an efficient inhibition of feed-forward transcriptional responses to IFNR stimulation by RV, when compared to intracellular dsRNA stimulation (Fig. 4B). Interestingly, IFNβ stimulation during the course of RV infection repressed HERC5, IFNG, IL22, IRF8, IL29, and IL1B transcription, suggesting that IFNAR1 stimulation (presumably of uninfected RV-bystander cells) might also trigger an inhibitory feed-back transcription mode (Fig. 4C).
In subsequent experiments, we quantified intestinal IFNR-stimulated transcription during homologous EW RV (or mock) infection by ectopically stimulating the IFNGR1 in suckling mice (Fig. 5A). These results demonstrated that, in contrast to a feed-forward ‘priming’ transcriptional mode in mock controls, IFNγ stimulation during EW RV infection potentiates decreased ISG and inflammatory transcripts (STAT1, ISG15, IFI204, DHX58, DDX58, MX2, CCL5, TRAIL, NOS2, and CXCL10) (Fig. 5B-C). In addition, we found that ectopic IFNGR1 stimulation during EW RV infection (but not IFNγ stimulation or EW RV alone) synergistically increases the transcription of anti-inflammatory transcripts (Fig. 5D). Interestingly, several of these upregulated genes (i.e. S100A8, BIRC5, ARG2, REG3G, and MKI67) are targets of the STAT3 signaling pathway, whose activation is reported to suppress STAT1-mediated inflammatory and antiviral programs due to IFNR stimulation(33, 68–70). We are currently examining whether STAT3 activation occurs during RV infection and mediates IFNR feedback signaling.
Collectively, our findings demonstrate that RV effectively reprograms IFNR-relayed transcription towards suppressing antiviral and inflammatory amplification (Fig. 3, 4, 5), a process whose efficiency is expected to be enhanced by IFNs generated in the local intestinal milieu during enteric RV infection (Fig. 1). The ability of RV infection to reprogram IFNR transcriptional function identifies an underlying mechanism by which other IFNR functions, such as caspase cleavage-mediated cell death could also be controlled by RV.
MATERIALS AND METHODS
Viruses and reagents
The tissue culture-adapted RV strains (simian RRV, SA11-4F, SA11-5S, human WI-61, porcine SB1-A, bovine UK) were propagated in African green monkey kidney cells (MA104) as previously described (71). Virus titers were determined by plaque assays in MA104 cells. HT-29 cells were plated in 6-well or 24-well cluster plates and completely confluent monolayers were infected 2-3d later as described. The non-tissue culture-adapted murine EW RV strain was prepared as an intestinal homogenate from RV-infected BALB/C suckling mice. The diarrheal dose (DD50) was determined as previously described (16). Human intestinal epithelial HT29 and embryonic kidney HEK293 cells were purchased from the American Type Culture Collection (ATCC) and maintained in Advanced D-MEM Ham’s F12 (Cellgro) or D-MEM medium (D-MEM, Cellgro) containing 10% fetal calf serum (Invitrogen) supplemented with non-essential amino acids, glutamine, penicillin, and streptomycin, respectively. Purified human IFN-β (PBL), human IFN-λ (PBL), and human IFN-γ (Millipore) were used for stimulation of cells. Purified carrier-free human IFN-β (PBL), murine IFN-γ, human TNF-α (R&D System, Minneapolis, MN), LPS (Sigma) were used. Cycloheximide (Sigma) and high-molecular weight LyoVec poly(I:C) (Invivogen) were used at 10 μg/mL and 2.5 μg/mL, respectively.
Mouse infection and tissue harvest
All mice were housed in the VA Palo Alto Health Care System (VAPAHCS) Veterinary Medicine Unit and all experimentation followed VAPAHCS Institutional Animal Care and Use Committee approved protocols. 3-5 day old Sv129 pups were infected via oral gavage with 104 DD50 of murine EW RV, 107 PFU of the simian RRV, or mock infected with 1X M199 media (Gibco) as described (16). (16). At the indicated time-points, the small intestines were removed. Sections were stored at −80 °C or fixed in 4% paraformaldehyde for processing and subsequent analyses. The protocol for IEC isolation, as previously described (72), was slightly modified. Briefly, the intestines were removed and opened longitudinally in RPMI media supplemented with penicillin, streptomycin, and L-glutamine (Gibco). The intestines were washed, cut into smaller pieces, and placed in 1X PBS containing 5 mM EDTA and 5 mM DTT (Sigma-Aldrich). Tissues from 3-4 pups (per infection group) were pooled at this point. The tissues were then placed on a shaker for 10 minutes at room temperature. Released cells were subsequently filtered through a 100-μm mesh filter into fresh RPMI FBS media (RPMI supplemented with 10% fetal bovine serum, penicillin, streptomycin, and L-glutamine) and immediately placed on ice. The remaining tissues were then incubated with fresh RPMI FBS media and rocked for an additional 10 minutes, followed by a final wash in RPMI FBS media for an hour. Cells were filtered, washed, and resuspended in PBS for further analyses.
Mouse stimulation and Luminex assays
To ectopically stimulate IFNGR1, suckling mice infected with 104 DD50 of murine EW RV (or mock-infected) were intraperitoneally administered purified murine IFN-γ (1μg IFN-γ per pup) or IFN-λ2 at 12 hpi and sacrificed 12 h later for analysis of transcripts by qRT-PCR. For endotoxin stimulation, 3-5 day old 129sv mice were orally inoculated by gavage with 104 DD50 of murine RV-EW. At 3 dpi, mice were intraperitoneally injected with PBS or purified LPS (Sigma, 10 mg/kg body weight) and sacrificed 6 h later for analysis of paraformaldehyde-fixed and paraffin embedded tissue sections by hematoxylin and eosin staining. For measurement of intestinal cytokines, mouse intestines were harvested and weighed pooled tissues were homogenized in ice-cold PBS containing a cocktail of protease and phosphatase inhibitors (Sigma) using a handheld micropestle (Thermo Fisher). Lysates were clarified by two sequential rounds of centrifugation at 10,000X G for 10 min, and analyzed for cytokine levels at the Human Immune Monitoring Center at Stanford University. Mouse 38-plex kits were purchased from eBiosciences/Affymetrix and used according to the manufacturer’s recommendations. Plates were read using a Luminex 200 instrument with a lower bound of 50 beads per sample per cytokine. Custom assay Control beads by Radix Biosolutions are added to all wells.
Western blot analysis
Preparation of cell lysates and immunoblotting in vitro was performed as described. For mouse samples, EDTA/DTT-treated IECs were lysed in radioimmunoprecipitation assay buffer (RIPA) supplemented with protease and phosphatase inhibitors (Roche). Equal volumes of 2X Laemmli buffer (Sigma-Aldrich) were added to each sample and analyzed by immunoblotting. Blots were probed with the following primary antibodies: anti-caspase 3, anti-caspase 8, anti-caspase 9, anti-cleaved caspase 3, anti-cleaved caspase 8, anti-cleaved caspase 7, anti-PARP, anti-GAPDH, anti-survivin, anti-STAT-3, anti-p-STAT-3 (Y705) (all from Cell Signaling Technologies) and an anti-RV capsid antibody (from Santa Cruz Biotechnologies). The blots were subsequently incubated with either anti-rabbit or anti-mouse secondary antibodies conjugated to horseradish peroxidase (HRP), incubated with ECL Detection reagent, and exposed to film (GE Healthcare). All densitometry quantifications of protein blots were analyzed using ImageJ software analysis and normalized to GAPDH protein bands.
Transcriptional analysis
For in vivo analysis of transcripts, mice were sacrificed and sections of the small intestine were collected on ice in Trizol (Life Technologies). For in vitro studies, cells were first washed in PBS and then lysed in Trizol. Total RNA was extracted following the manufacturer’s instructions and subjected to DNAse digestion before use in qRT-PCR. Synthesis of cDNA and subsequent microfluidics PCR on the Fluidigm platform was done as described earlier (73). Serial 10-fold dilutions of mouse or human reference RNA (Agilent) were run in duplicate for each PCR run.
Flow cytometry analysis
For flow cytometry analysis of caspase expression and cell viability in vivo, mouse intestines were harvested as described above and equivalent numbers of EDTA/DTT-treated IECs were stained for cell viability using Invitrogen’s LIVE/DEAD Fixable Aqua Dead Cell Stain Kit following the manufacturer’s protocol. Cells were blocked with purified rat anti-mouse CD16/CD32 Fc Block (BD Biosciences) and stained for the following murine surface markers: anti-CD45 (clone 30-F11) v450 (BD Biosciences), anti-CD326 (Ep-CAM/ESA, clone G8.8) PE (Biolegend), and anti-CD44 (clone IM7) PerCP/Cy5.5 (Biolegend) as described earlier (7). Cells were then fixed and permeabilized with BD Cytofix/Cytoperm Fixation/ Permeabilization buffer according to the manufacture’s protocol and subsequently used for intracellular protein staining. To perform intracellular protein staining, cells were blocked with BD Fc Block and incubated with unconjugated antibodies to cleaved caspase 8 (Cell Signaling Technologies) or cleaved caspase 3 (Cell Signaling Technologies) along with conjugated antibodies to RV VP6 (1E11-Texas Red) and Ki67 (Alexa Fluor 647, BD Biosciences). Subsequently, cells were incubated with a secondary goat anti-rabbit-Alexa Fluor 488 antibody (Molecular Probes) for detection of cleaved caspases. All data was acquired using a LSRII (BD Biosciences) and analyzed with FlowJo software (Treestar Inc). Staining and flow cytometry analysis of HT-29 cells was performed as described.
TUNEL and apoptosis assays
For TUNEL, following surface staining of intestinal epithelial cells (as described above), cells were fixed and permeabilized for subsequent determination of DNA fragmentation using Roche’s In Situ Cell Death Detection Kit-Fluorescein (following the manufacturer’s protocol). Briefly, cells were fixed in 2% PFA and permeabilized in 0.1% Triton X-100 in 0.1% sodium citrate buffer and incubated with the terminal deoxynucleotidyl transferase (TdT), along with fluorescein-dUTP labeling solution for 1 hour at 37 °C. As a positive control, permeabilized IECs were treated with DNase I (Ambion) for 30 min at 37 °C (to induce DNA fragmentation) prior to the TUNEL reaction. An additional negative control (labeling reactions carried out without TdT enzyme) was also included. Incorporation of labeled dUTP was analyzed on a LSRII flow cytometer. To assay apoptosis marker expression in HT-29 cells, cells were washed with PBS, stained with Apopxin Green (1:100 dilution, Abcam) at room temperature for 10 minutes, and washed before examination by fluorescence microscopy.
Statistical analysis
Analysis of variance (ANOVA) and 2-tailed student’s t-tests were performed using GraphPad InSTAT software. Figure legends indicate the specific tests used and P values <0.05 were considered significant. All error bars represent the standard error of the mean.
ACKNOWLEDGEMENT
This work was supported by grants 1RO1 AI021362, R56 AI125249, and 1RO1 AI1125249 from the National Institutes of Health, and 1IO 1BX000158-01A1 from the Veterans Affairs. We thank Catherine Cruz for carrying out initial studies and helpful scientific discussions during the study.