Recovirus NS1-2 Has Viroporin Activity That Induces Aberrant Cellular Calcium Signaling To Facilitate Virus Replication

Tulane virus is one of many enteric caliciviruses that cause acute gastroenteritis and diarrheal disease. Globally, enteric caliciviruses affect both humans and animals and amass >65 billion dollars per year in treatment and health care-associated costs, thus imposing an enormous economic burden. Recent progress has resulted in several cultivation systems (B cells, enteroids, and zebrafish larvae) to study human noroviruses, but mechanistic insights into the viral factors and host pathways important for enteric calicivirus replication and infection are still largely lacking. Here, we used Tulane virus, a calicivirus that is biologically similar to human noroviruses and can be cultivated by conventional cell culture, to identify and functionally validate NS1-2 as an enteric calicivirus viroporin. Viroporin-mediated calcium signaling may be a broadly utilized pathway for enteric virus replication, and its existence within caliciviruses provides a novel approach to developing antivirals and comprehensive therapeutics for enteric calicivirus diarrheal disease outbreaks.

picornavirus-like superfamily of positive-sense RNA viruses, among which there is considerable positional homology of the cognate proteins of the nonstructural polyprotein (24,46,47). Within this rubric, the picornavirus 2AB region constitutes the positional homolog of the calicivirus NS1-2 protein, and several sequence motifs in NS1 are conserved in the 2A protein of some picornaviruses (24). While no functional homology between EV 2B and the NS2 region of NS1-2 has yet been identified, it is tempting to speculate that NS1-2 may have viroporin activity and dysregulate host Ca 2ϩ signaling analogous to that of EV 2B.
In this study, we investigated the role of Ca 2ϩ signaling in TV replication and whether TV NS1-2 has viroporin activity that can dysregulate Ca 2ϩ homeostasis. Using long-term live-cell Ca 2ϩ imaging, we sought to determine whether TV infection causes aberrant Ca 2ϩ signaling during infection and identify the cellular Ca 2ϩ pools critical for the TV-induced Ca 2ϩ signaling. Finally, we tested TV NS1-2 for viroporin activity and determined whether the putative NS1-2 viroporin domain caused aberrant Ca 2ϩ signaling similar to TV infection.

RESULTS
TV infection disrupts host calcium signaling kinetics in LLC-MK2 cells. Ca 2ϩ is a ubiquitous secondary messenger and many enteric viruses (e.g., RVs and EVs) require elevated cytosolic Ca 2ϩ to facilitate replication (31, 37-40, 43, 44). To determine whether TV causes aberrant Ca 2ϩ signaling like other enteric viruses, we examined whether Ca 2ϩ signaling dynamics changed during TV infection. We infected LLC-MK2 cells stably expressing GCaMP6s (MK2-G6s) with different infectious doses (multiplicities of infection [MOI] of 1, 5, and 10) or ␥-irradiated inactivated TV and performed live-cell fluorescence microscopy during the infection. GCaMP6s is a green fluorescent protein (GFP)-based genetically encoded Ca 2ϩ indicator that reports changes in cytosolic Ca 2ϩ as an increase in fluorescence (48). TV-infected MK2-G6s cells show increased cytoplasmic Ca 2ϩ levels ( Fig. 1A) beginning at roughly 8 h postinfection (HPI) (MOI of 10), and quantitation of the GCaMP6s signal shows a significant increase at 8 and 12 HPI (Fig. 1B). This is illustrated in the time-lapse movie of the infection (see Movie S1 in the supplemental material). The observed increase in Ca 2ϩ signaling coincides with the synthesis of TV nonstructural proteins, assessed by Western blotting using anti-Vpg (Fig. 1C, black arrowhead) and anti-TV (Fig. 1D, black arrowhead) antisera, which show increased TV protein production between 8 and 12 HPI, but no detection of Vpg or VP1 in mock lysates ( Fig. 1C and E). Further, based on a one-step growth curve, the increased cytosolic Ca 2ϩ also coincides with the onset of progeny virus production, which occurs between 6 and 8 HPI (Fig. 1F). The increases in cytosolic Ca 2ϩ were dynamic during TV infection (Movie S2). We noted that in infected cells, changes in cytosolic Ca 2ϩ occurred through an increased number of discrete Ca 2ϩ signals, much like what we recently observed in RV-infected cells (Fig. 1G) (66). We refer to these high-amplitude, transient Ca 2ϩ signals as "Ca 2ϩ spikes" and quantitated the number of Ca 2ϩ spikes per cell during infection. Compared to uninfected controls, TV-infected cells have significantly more Ca 2ϩ spikes/cell, but cells inoculated with ␥-irradiated TV did not exhibit increased Ca 2ϩ signaling (Fig. 1H). Together, these data indicate that increased Ca 2ϩ signaling requires replication-competent virus and occurs later during infection, well after entry has occurred. Additionally, Ca 2ϩ signaling in infected cells increases in an infectious-dose-dependent manner, saturating at an MOI of 5 (Fig. 1H). To visualize the aberrant Ca 2ϩ signaling induced by TV, we generated heatmaps plotting normalized GCaMP6s fluorescence over time (Fig. 1I). Heatmap data show an increased number and magnitude of Ca 2ϩ signals and that cytosolic Ca 2ϩ levels change earlier and more frequently throughout infection as the infectious dose increases (Fig. 1I). The heatmaps also show that MK2-G6s cells inoculated with ␥-irradiated TV do not have increased Ca 2ϩ signaling compared to mock-inoculated cells (Fig. 1I), consistent with the lack of increased Ca 2ϩ spikes (Fig. 1H). Taken together, these data suggest that, like other enteric viruses, TV disrupts host Ca 2ϩ signaling kinetics during infection.

***
Intracellular Ca 2؉ is critical for TV replication. Since we observed aberrant Ca 2ϩ signaling during TV infection, we sought to determine whether Ca 2ϩ was involved in TV replication. To test this, we manipulated extracellular and intracellular Ca 2ϩ levels and determined the effects on TV yield. Doubling the extracellular Ca 2ϩ concentration (ϳ4 mM) did not affect TV yield ( Fig. 2A, right). In contrast, TV propagated in Ca 2ϩ -free media significantly reduced total yield ( Fig. 2A, middle). Interestingly, plaques of TV propagated in Ca 2ϩ -free media were significantly smaller than that propagated in normal media, even though the plaque assay titrations were performed in normal media ( Fig. 2C and D). Next, to investigate the role of intracellular Ca 2ϩ during infection, we treated LLC-MK2 cells with BAPTA-AM, which chelates cytosolic Ca 2ϩ and therefore buffers cytosolic Ca 2ϩ (49,50). TV replication in Ca 2ϩ -free media supplemented with BAPTA-AM (0 mM Ca 2ϩ ϩ BAPTA) was reduced up to 4 log units (Fig. 2B), which was a greater inhibition than Ca 2ϩ -free media alone ( Fig. 2A versus Fig. 2B). We next sought to determine whether intracellular Ca 2ϩ stores are important for TV replication by testing the effect of thapsigargin (TG) on TV replication. TG is an inhibitor of sarco/ endoplasmic reticulum (SERCA) Ca 2ϩ ATPase, which pumps cytosolic Ca 2ϩ into the ER to help maintain ER Ca 2ϩ stores. We treated TV-infected cells with TG and measured TV yield as described in Materials and Methods and found that TV replication is ϳ3 log units lower in TG-treated cells than in dimethyl sulfoxide (DMSO)-treated cells (Fig. 2B). Finally, we tested these different manipulations of extracellular or intracellular Ca 2ϩ on TV yield at different time points during infection (8,16, and 24 HPI) (Fig. 2E). These studies confirmed that reduction of extracellular Ca 2ϩ or treatment with TG significantly inhibited total virus replication; however, the rate of progeny virus production was not substantially reduced. Together, the replication assays demonstrate that intracellular Ca 2ϩ levels facilitate TV replication and that the ER Ca 2ϩ store is particularly important for robust virus production.
TV-induced Ca 2؉ signaling requires ER Ca 2؉ stores. We next sought to determine the effects that the manipulations to extracellular and intracellular Ca 2ϩ had on the TV-induced Ca 2ϩ signaling exhibited in Fig. 1. We altered extracellular and intracellular Ca 2ϩ concentrations as described in Materials and Methods and performed live Ca 2ϩ imaging of mock-infected and TV-infected MK2-G6s cells. TV-infected cells in 2 mM Ca 2ϩ (normal media) exhibited increased Ca 2ϩ signaling, as observed above (Fig. 3A). Supplementing media with additional extracellular Ca 2ϩ (4 mM Ca 2ϩ total) did not further increase the Ca 2ϩ spikes, but removing extracellular Ca 2ϩ abolished the TV-induced Ca 2ϩ spikes (Fig. 3A). Using heatmaps, we plotted the relative change in GCaMP6s fluorescence over time and observed increased signaling starting at ϳ8 HPI in both the 2 mM Ca 2ϩ and 4 mM Ca 2ϩ conditions (Fig. 3B). Further, the heatmaps show that infected cells in Ca 2ϩ -free media have a signaling profile that phenotypically mimics uninfected controls (Fig. 3B). Like the results obtained in replication assays, buffering cytoplasmic Ca 2ϩ using BAPTA-AM reduced the number of Ca 2ϩ spikes per cell to a level comparable to that of mock-infected cells ( Fig. 3C and Movie S3). Similarly, blocking the ER SERCA pump with TG significantly reduces TV-induced Ca 2ϩ signaling (Fig. 3D), supporting replication data and demonstrating that ER Ca 2ϩ stores are a critical source of Ca 2ϩ for enhancing replication. coincides with both structural and nonstructural protein synthesis. Mature Vpg in panel C is indicated by a black arrowhead, and the major band (open arrowhead) represents the Vpg-Pro precursor (ϳ30 kDa). L, lysate; ␣Vpg, anti-Vpg. (E) Western blot of mock lysates for structural protein VP1. (F) One-step growth curve for TV at a low MOI (MOI of 1) shows that virus replication is concomitant with viral protein synthesis (C and D) and with changes in Ca 2ϩ signaling (A). (G) Image from overlay of anti-Vpg staining (red) onto short (10-min) continuous imaging runs of TV-infected cells (MOI of 5) at 12 HPI. Accompanying Ca 2ϩ cell traces (right) show the dynamic increases in cytosolic Ca 2ϩ in infected cells. ROI, region of interest. (H) Compared to mock-infected cells, TV-infected cells have an increased number of Ca 2ϩ spikes per cell that increases in an infectious dose-dependent manner, saturating at an MOI of 5. IRR TV, gamma-irradiated TV. (I) Heatmap data suggest that Ca 2ϩ signaling increases with infectious dose and that a higher MOI disrupts host Ca 2ϩ signaling earlier in infection and sustains this aberrant Ca 2ϩ signaling throughout. Mock-infected and irradiated TV have similar heatmap profiles, suggesting that replication-competent virus is required to drive these changes in Ca 2ϩ signaling. Data are shown as means Ϯ standard deviations (SD) (error bars). Values that are significantly different are indicated by a bar and asterisks as follows: *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001; ****, P Ͻ 0.0001. Values that are not significantly different (NS) are also indicated. N Ն 3 for each experiment, except the one-step growth curve, which was N ϭ 2 with three replicates per experiment.
Tulane virus NS1-2 is targeted to the ER membrane. Our data indicate that TV activates aberrant Ca 2ϩ signaling involving the ER Ca 2ϩ store, much like the dysregulation of Ca 2ϩ homeostasis by other enteric viruses observed in RV and EV infections. Both RV and EV encode a viroporin, or viral ion channel, that targets the ER Ca 2ϩ store to activate aberrant Ca 2ϩ signaling pathways that are critical for virus replication (36,37,39,40,43,44). Viroporins are integral membrane proteins that have some common characteristics, including being oligomeric, having an amphipathic ␣-helix that serves as the channel lumen through the membrane, and a cluster of basic amino acid residues that facilitate insertion into the membrane (25,35,36,40,51). Previous work with NS1-2 from several different caliciviruses shows that it is membrane associated and localizes primarily to the ER (18)(19)(20)(21)25) and/or Golgi apparatus (20,21,23,24). Thus, we hypothesized that calicivirus NS1-2 could be a viroporin involved in the aberrant Ca 2ϩ signaling we observed during TV infection. Notably, the calicivirus NS2 domain is the positional homolog of the EV 2B viroporin (see Fig. S1A in the supple- mental material). This is potentially significant because previous studies have found conserved functional characteristics between the positional homologs of the other nonstructural proteins (21,23,24,47,(51)(52)(53)(54), and functional homology between EV 2AB and human norovirus (HuNoV) GII.4 NS1-2 (21,24). Additionally, when performing multiple-sequence alignments of other calicivirus NS1-2s, we found that the C-terminal domain (CTD) is highly conserved, particularly in the putative viroporin domain (Fig. S1B). To determine whether TV NS1-2 has viroporin-like characteristics, we examined TV NS1-2 for viroporin motifs. First, we performed a Kyte-Doolittle plot to detect hydrophobic regions of NS1-2 and an amphipathicity plot to identify potential amphipathic domains (Fig. 4A). We found that amino acids 195 to 215 (aa195-215) (Fig. 4A, dark green box) in the CTD of NS1-2 has a high amphipathic moment. We then used PSIPred (55) to model NS1-2 predicted secondary structure (Fig. 4B). Output from this analysis suggested that the NS1-2 CTD was predominantly comprised of ␣-helices (Fig. 4B, pink residues), and accompanying confidence scores for prediction of these C-terminal helices were Ն75% (Fig. S2). Interestingly, the region of peak amphipathicity (Fig. 4A) was located within one of the PSIPred helix predictions of the CTD (Fig. 4B, dark green bar) and contained clustered basic residues (blue asterisks), two key features of viroporins. Additionally, NS1-2 topology modeling identified two putative transmembrane domains (TMDs): the first (TMD1) from aa164-179, and the second (TMD2) from aa202-225 (Fig. 4C, top). The membrane topology schematic indicated that both TMD1 and TMD2 had predicted pore-lining regions within their helices (Fig. 4C,     (aa198-215), since TMD2 had the clustered basic residues common among viroporins. The helical wheel shows that TMD2 is highly amphipathic with clear polar and nonpolar faces to the helix (Fig. 4D). The calculated hydrophobic moment for TMD2 is 0.522, supporting the above amphipathicity predictions (Fig. 4A). Given the results of these computational studies, we predicted that NS1-2 TMD2 (aa195-215) is a viroporin domain and set out to test this prediction experimentally. First, we tested whether TV NS1-2 was an integral membrane protein and whether it localized to the ER similar to NS1-2 from other caliciviruses. To do so, we generated bacterial and mammalian expression vectors of full-length NS1-2. For mammalian expression vectors, we N-terminally fused full-length NS1-2 to mRuby3 (henceforth referred to as RFP-NS1-2 [RFP stands for red fluorescent protein]). From these constructs, we generated two truncation mutants of wild-type NS1-2 in both mammalian and bacterial expression vectors: the first, NS1-2 Δ176, was predicted to have TMD1 but lack the viroporin domain, and the second, NS1-2 Δ157, was predicted to lack both TMD1 and the VPD. We then transfected wild-type, full-length (WT) RFP-NS1-2, RFP-NS1-2 Δ157, and RFP-NS1-2 Δ176 into HEK 293FT cells and harvested cell suspensions next day. Samples after cell lysis, sonication, and fractionation were collected for SDS-PAGE Western blots. We found both Δ176 and WT TV NS1-2 in the total fraction (T) and membrane pellets (M), but not in the supernatant (S), suggesting that TMD1 mediates membrane association (Fig. 4E). Additionally, in the nonreducing, unboiled conditions used, oligomers of both Δ176 and WT RFP-NS1-2 were detected by Western blotting (Fig. 4E, black arrowheads). Similar results were obtained from membrane fractionation of analogous bacterially expressed NS1-2 constructs (Fig. S3). Using the mammalian expression vectors of RFP-NS1-2, we performed colocalization assays with fluorescent markers of the ER, Golgi apparatus, and mitochondria. RFP-NS1-2 showed no colocalization with the mitochondria or Golgi apparatus (Fig. 4F). In contrast, RFP-NS1-2 strongly colocalized with the ER-GFP marker (Fig. 4F), indicating that, like NS1-2 from other caliciviruses and EV 2B and RV NSP4, TV NS1-2 traffics to the ER membrane.

M D T S I D S V L S D T S P I S G A D V Q K L I F G N T Q P V S Y D R R P E P K L G Q V I V L D E G D C F H Y A I Y I E K G L L F S T G G M I G N G A F R L N G L T H P W G T L D L F G P E D K T F Y Q N K I G E K Y P Y S I T R S N C L H S I L Q T I G V S Y V H Y K D K P L P A Q F Y T H V Q D W N E N R Y D V G G T K S G W T Q Q L L E I V Y L I V K D V N W A K I C M D F K P L N L W H N W K T M K P T F K G V L A F L T R V A E L W G I N I S S L I T F L T S S L I P Q M D T S I D S V L S D T S P I S G A D V Q K L I F G N T Q P V S Y D R R P E P K L G Q V I V L D E G D C F H Y A I Y I E K G L L F S T G G M I G N G A F R L N G L T H P W G T L D L F G P E D K T F Y Q N K I G E K Y P Y S I T R S N C L H S I L Q T I G V S Y V H Y K D K P L P A Q F Y T H V Q D W N E N R Y D V G G T K S G W T Q Q L L E I V Y L I V K D V N W A K I C M D F K P L N L W H N W K T M K P T F K G V L A F L T R V A E L W G I N I S S L I T F L T S S L I P Q M D T S I D S V L S D T S P I S G A D V Q K L I F G N T Q P V S Y D R R P E P K L G Q V I V L D E G D C F H Y A I Y I E K G L L F S T G G M I G N G A F R L N G L T H P W G T L D L F G P E D K T F Y Q N K I G E K Y P Y S I T R S N C L H S I L Q T I G V S Y V H Y K D K P L P A Q F Y T H V Q D W N E N R Y D V G G T K S G W T Q Q L L E I V Y L I V K D V N W A K I C M D F K P L N L W H N W K T M K P T F K G V L A F L T R V A E L W G I N I S S L I T F L T S S L I
TV NS1-2 has viroporin activity that disrupts Ca 2؉ signaling. Since our predictive modeling suggested that NS1-2 met the biophysical requirements for a viroporin and our live-cell Ca 2ϩ imaging data exhibited large changes in cytosolic Ca 2ϩ during TV infection, we tested whether NS1-2 has viroporin activity. We performed the Escherichia coli lysis assay, which is a classical viroporin functional assay, wherein viroporin expression by E. coli BL21(DE3)pLysS results in permeabilization of the inner membrane, resulting in T7 lysozyme-mediated cell lysis (42). This assay has been used to identify and initially characterize many viroporins (37,57,58). We expressed full-length HisNS1-2 in BL21(DE3)pLysS cells and measured optical density (OD) over time after protein induction with IPTG. For the lysis assay, strong viroporin activity is characterized by large decreases in OD over time, whereas no viroporin activity is characterized by increases in OD over time. Our results show that induced NS1-2 has strong viroporin activity, similar to that of RV NSP4, our positive control for viroporin activity (Fig. 5A). We see no changes in OD over time for uninduced NS1-2, indicating that histidinetagged NS1-2 (HisNS1-2) viroporin activity correlated with protein expression, detected by immunoblotting for the 6ϫHis tag (Fig. 5B). We then asked whether recombinant insertion and orientation where the putative VPD (aa195-212) comprises the pore-lining helix (bottom right). (D) Helical wheel plot generated from the NS1-2 amphipathic segment (dark green bar) shows clustered basic residues (blue circles) and a hydrophobic moment of 0.522 from aa198-215, coinciding with the putative VPD. (E) Mammalian expressed full-length RFP-NS1-2 and RFP NS1-2 Δ176 are membrane associated, but RFP NS1-2 Δ157 is not. Both the total fraction (T) and membrane pellets (M) extracted with 1% SDS contain RFP-NS1-2 and Δ176, but centrifuged supernatant (S) does not, suggesting that RFP-NS1-2 and Δ176 are membrane-associated proteins. In contrast, the supernatant contains RFP-NS1-2 Δ157. Further, immunoblot assays run under nonreducing conditions show that full-length RFP-NS1-2 and Δ176 oligomerize (black arrowheads). No detection of NS1-2 observed in transfection control lysates. L, lysate; mo␣myc, anti-myc monoclonal antibody. (F) Cotransfection experiments using intracellular markers for predominant intracellular Ca 2ϩ stores mitochondria (Mito), Golgi apparatus, and endoplasmic reticulum (ER) to determine whether TV NS1-2 associated with any intracellular organelle(s). Based on deconvolution microscopy data, RFP-NS1-2 localized to the ER (right), but not with the Golgi apparatus (middle). RFP-NS1-2 did not localize to the mitochondria (left) (N Ն 2). N Ն 3 for immunoblot experiments. Movie S4). As described above, we quantitated the number of Ca 2ϩ spikes and confirmed that recombinant expression of RFP-NS1-2 increased the number of Ca 2ϩ spikes per cell approximately twofold, similar to that of EV 2B and RV NSP4 (Fig. 5D). Taken together, our results demonstrate that TV NS1-2 has viroporin activity in the lysis assay, similar to bona fide viroporins, and causes aberrant host Ca 2ϩ signaling when expressed in mammalian cells. NS1-2 viroporin activity maps to the putative viroporin domain. Our computational studies above identified a putative TV NS1-2 VPD from aa195-212. To determine whether the NS1-2 viroporin activity maps to this putative VPD, we generated C-terminal truncation mutants in bacterial expression vectors with deletions after aa212 (A212-Δ), after aa194 (W194-Δ), or after aa176 (D176-Δ) and characterized them in the lysis assay (Fig. 6A). We found that the A212-Δ truncation (red) had strong lysis activity comparable to full-length NS1-2 (black) (Fig. 6B). In contrast, the D176-Δ truncation (blue) exhibited no lysis activity, comparable to uninduced NS1-2 (gray) (Fig. 6B). Immunoblot analysis confirmed that protein expression correlated with viroporin activity and that the impaired activity of W194-Δ was not due to lower expression levels, since the expression was comparable to that of full-length protein and A212-Δ (Fig. 6C). Since the W194-Δ truncation (green) had impaired viroporin activity, this suggests that the VPD functionally extends to aa177-212.
Next, we characterized truncation mutants for their activation of aberrant Ca 2ϩ signaling in MK2-G6s cells. Since recombinant expression of full-length RFP-NS1-2 induced aberrant Ca 2ϩ signaling (Fig. 5D), we tested whether truncating the putative viroporin domain alone (Δ176) or both TMDs (Δ157) would compromise NS1-2-induced Ca 2ϩ signaling (Fig. 6D). First, we examined the subcellular distributions and expression levels of the constructs. While the full-length and Δ176 truncation both appeared reticular, the Δ157 truncation had cytoplasmic distribution, consistent with it lacking both TMDs (Fig. 6E). Immunoblot analysis shows that the expression of both truncations was much greater than that of full-length NS1-2 (Fig. 6F, left blots), and by loading less lysate, we can better resolve the 2-kDa size difference in the Δ157 and Δ176 truncations (Fig. 6F, right blots). Next, we examined whether these truncations could induce Ca 2ϩ signaling by long-term live-cell Ca 2ϩ imaging in MK2-G6s cells. Individual cell traces illustrate that neither the Δ157 nor Δ176 truncation dramatically increased Ca 2ϩ signaling similar to full-length RFP-NS1-2 (Fig. 6G). Quantitation of the Ca 2ϩ spikes per cell showed that while both truncations exhibited higher Ca 2ϩ signaling than RFP alone (Fig. 6H), the amplitude of these spikes was significantly reduced compared to full-length RFP-NS1-2 (Fig. 6I). The significant reduction in the number and amplitude of Ca 2ϩ spikes/cell for both mutants highlights the critical importance of an intact VPD for disrupting host Ca 2ϩ signaling. Together this work demonstrates that TV NS1-2 is an ER-targeted viroporin that induces aberrant Ca 2ϩ signaling.
Noroviruses exhibit aberrant Ca 2؉ signaling during infection and expression of NS1-2. Many aspects of HuNoV pathogenesis remain unknown, but elevation of cytosolic Ca 2ϩ is implicated in many other enteric virus infections (31,37,38,59,66). The identification of aberrant Ca 2ϩ signaling by TV and viroporin activity of NS1-2 could provide new insights into HuNoV pathogenesis if this activity is also evident in noroviruses. Thus, we wanted to know whether the aberrant Ca 2ϩ signaling observed was specific to TV or shared among noroviruses. To test this, we infected GCaMP6sexpressing BV-2 cells with MNV-1 CW1 at an MOI of 1, 5, or 10 and performed long-term Ca 2ϩ imaging, as described in Materials and Methods. Like TV infection, Ca 2ϩ signaling in MNV-infected cells increases concomitant with infectious dose (Fig. 7A) and manifests as an increase in dynamic Ca 2ϩ signaling (Movie S5). Interestingly, mockinoculated BV2-GCaMP6s exhibited a greater number of Ca 2ϩ spikes than mockinoculated MK2-GCaMP6s cells, but this is likely due to differences in basal Ca 2ϩ signaling between immune and epithelial cells (60)(61)(62).
We next sought to determine whether the NS2 viroporin function we discovered in TV NS1-2 was conserved in the NS2 of any other calicivirus. The multiple-sequence  aberrant Ca 2ϩ signaling during infection and disrupt host Ca 2ϩ signaling through production and expression of the nonstructural protein NS1-2.

DISCUSSION
As obligate intracellular pathogens, viruses are adept at exploiting host pathways to facilitate replication. Viruses from many different taxonomic families activate aberrant Ca 2ϩ signaling because Ca 2ϩ signals are used by all cells to regulate a vast array of cellular functions. Therefore, this represents a powerful strategy to reconfigure host cell physiology via targeted disruption of host Ca 2ϩ homeostasis. The overarching goal of this study was to determine whether dysregulation of Ca 2ϩ signaling is a characteristic of caliciviruses and whether this is due to the production of a viroporin protein similar to picornaviruses. To address these questions, we studied TV, as a model calicivirus, The exploitation of host Ca 2ϩ signaling to facilitate virus replication is a common feature of many viruses (31). Our finding that TV coopts Ca 2ϩ signaling is consistent with previous studies showing that elevated Ca 2ϩ levels are important for picornavirus replication, especially since caliciviruses and picornaviruses utilize a similar replication strategy (44). Similar to other Ca 2ϩ -disrupting viruses, TV also induces aberrant Ca 2ϩ signaling peak of virus replication, many hours after cell entry. This is consistent with the reduced virus yield in media with reduced extracellular Ca 2ϩ or treatments to buffer cytosolic Ca 2ϩ (BAPTA-AM) or block refilling of ER Ca 2ϩ stores (TG). Further, as we recently reported for RV infection, the TV-induced increase in cytosolic Ca 2ϩ manifests as many discrete Ca 2ϩ signals rather than a monophasic increase in Ca 2ϩ over the infection (66). This raises the following questions. (i) What cellular pathways are activated by this Ca 2ϩ signaling? (ii) How do they benefit TV replication? Both RV and EV have been shown to exploit Ca 2ϩ signaling to activate the biosynthetic early stages of autophagy, which facilitates virus replication through rearrangement of cellular membranes to form replication complexes (67). MNV infection of primary macrophages or the RAW264.7 cell line activates autophagy, but in contrast to RV and EV, autophagy limits MNV replication (68). Thus, it remains to be determined whether autophagy plays a role in calicivirus replication complex assembly or whether Ca 2ϩ signaling regulates autophagy activation during calicivirus infection. Further, elevated Ca 2ϩ signaling may serve to modulate cellular apoptotic responses. Strong monophasic increases in cytosolic Ca 2ϩ activate apoptosis through mitochondrial Ca 2ϩ overload, but transient and oscillatory Ca 2ϩ fluxes serve as prosurvival signals (69). Activation of apoptosis has been seen in norovirus-and feline calicivirus-infected cells, and caspase activation is critical for cleavage and release of MNV NS1 from NS1-2, which in turn modulates cellular innate immune responses (19,22). Additionally, previous work with MNV-1 CR6 shows that NS1-2 from this norovirus is cleaved by caspase-3 during late infection. Apoptotic induction coincided with viral egress, suggesting that activation of apoptosis and cleavage of NS1-2 by caspase-3 occur to facilitate viral spread after viral replication and virion assembly (17). Thus, increased transient Ca 2ϩ signaling may serve to counteract apoptosis activation until necessary to help prolong cell viability and maximize virus replication.
Within the superfamily of picornavirus-like positive-sense RNA viruses, there is positional homology between the ORF1 nonstructural proteins of caliciviruses (and likely astroviruses) and the P2-P3 nonstructural proteins of picornaviruses (24,46,47). We used this framework to determine whether TV NS1-2 exhibited viroporin activity, since the positional homolog, the picornavirus 2B protein, is a well-established Ca 2ϩconducting viroporin (35,45). We found that TV NS1-2 has viroporin activity, similar to 2B and RV NSP4, and the viroporin activity mapped to the integral membrane NS2 domains. Since both the N and C termini are likely oriented in the cytoplasm, NS1-2 is classified as a class IIB viroporin, similar to the picornavirus 2B proteins (35,45). This topology is supported by the cytosolic accessibility of the NS1 domain and the need for the C terminus to also be localized in the cytosol to enable cleavage by the NS6 protease. Further, the similarity of Ca 2ϩ signaling induced by TV and HuNoV RFP-NS1-2 raises the question of whether NS1-2 viroporin activity is conserved throughout the Caliciviridae family. Though 2B and NS1-2 lack appreciable primary sequence homology, this is not surprising because viroporins, even from the same virus family, often share only the common viroporin motifs (i.e., [i] having an amphipathic ␣-helix, [ii] having a cluster of basic residues, and [iii] being oligomeric) (34,39,40). We found that among NS1-2 from different caliciviruses, these characteristic features are conserved, so we predict that viroporin activity of NS1-2 is a common function. Furthermore, since blunting cytosolic Ca 2ϩ signaling with BAPTA-AM reduced TV replication, blocking NS1-2 viroporin activity with mutations or drugs should also reduce replication. This is supported by a previous study showing that recombinant coxsackie B3 virus with mutations of the 2B viroporin exhibited significantly impaired replication or was completely replication deficient (70). Analogous studies can be done using the TV reverse genetics system once residues critical for viroporin activity are identified through mutagenesis screens of the TV NS1-2 viroporin domain we mapped in this study.
The increased Ca 2ϩ signaling observed in TV-infected cells is phenotypically similar to that induced by recombinant expression of full-length NS1-2, but the Ca 2ϩ signaling is abrogated by truncation of the viroporin domain. Further, NS1-2 primarily localized to the ER, which is a major intracellular Ca 2ϩ storage organelle. Thus, our model predicts that NS1-2 directly releases Ca 2ϩ from the ER; however, it is likely that both NS1-2 and activation of host Ca 2ϩ signaling pathways contribute to the observed Ca 2ϩ signals. Ca 2ϩ signals from NS1-2 require it to directly conduct Ca 2ϩ and have a high enough conductance that the ER Ca 2ϩ release event can be detected by a fluorescent Ca 2ϩ indicator, yet these unitary events are challenging to detect even for large channels like the inositol trisphosphate receptor (IP3R) (71). Future studies using patch clamp electrophysiology are needed to confirm that NS1-2 conducts Ca 2ϩ and determine its conductivity. Nevertheless, based on the similarities between NS1-2 and other Ca 2ϩ -conducting viroporins, EV 2B and RV NSP4, NS1-2 viroporin activity would reduce ER Ca 2ϩ levels, and this in turn will activate host Ca 2ϩ signaling pathways. First, the moderately increased steady-state cytosolic Ca 2ϩ levels could foster more ER Ca 2ϩ release by potentiating the IP3R Ca 2ϩ release channel (72). Second, reduced ER Ca 2ϩ levels activate the store-operated Ca 2ϩ entry (SOCE) pathway, wherein decreased ER Ca 2ϩ levels activate the ER Ca 2ϩ sensing protein stromal interaction molecule 1 (STIM1). Activated STIM1 translocates to ER microdomains adjacent to the plasma membrane and opens Ca 2ϩ influx channels, like Orai1, to elevate cytosolic Ca 2ϩ (32,33). This Ca 2ϩ influx, in concert with SERCA, helps to refill ER stores for continued signaling.
HuNoV and human sapoviruses cause outbreaks of acute gastroenteritis (AGE) and are a major cause of foodborne illnesses. However, the molecular mechanisms of how these caliciviruses cause vomiting and diarrhea, the chief symptoms of AGE, have not been characterized. The dysregulation of Ca 2ϩ signaling by TV may provide insights into the pathophysiology of enteric caliciviruses. Both IP 3 -mediated ER Ca 2ϩ release and SOCE have been shown to activate chloride secretion from epithelial cells (73,74). In studies of other viroporins, the viroporin-induced elevated cytosolic Ca 2ϩ induces cytoskeleton rearrangement, leading to disassembly of tight junctions and loss of barrier integrity (40). Hyperactivation of chloride secretion and loss of tight junctions would contribute to excess fluid secretion and diarrhea. In our study, we have shown that dysregulated Ca 2ϩ signaling is a feature of calicivirus infection using TV. Additionally, our data with recombinant GII.3 NS1-2 shows aberrant Ca 2ϩ signaling at the onset of expression similar to what we observe with recombinant expression of TV NS1-2 (Fig. 7). This suggests that HuNoV NS1-2 may be functioning as a viroporin, similar to TV NS1-2. Thus, future studies can further examine the role of aberrant Ca 2ϩ signaling in calicivirus pathophysiology using human intestinal enteroid cultures that support the replication of many HuNoV strains (4).
In summary, we have shown that TV activates aberrant Ca 2ϩ singling during infection, and cellular Ca 2ϩ is critical for robust TV replication. Further, we found that the NS2 domain of the NS1-2 nonstructural protein is a viroporin that alone induces Ca 2ϩ signaling similar to TV infection. Together, these results indicate that NS1-2 is functionally analogous to EV 2B and RV NSP4. While little is known about the function(s) of NS1-2, and particularly the NS2 domain of NS1-2, the similarity with other Ca 2ϩconducting viroporins may provide a broader insight for understanding NS1-2 functions. Finally, antiviral drugs against viroporins have been developed for influenza virus M2 and HIV Vpu (35). Thus, the NS1-2 viroporin may be a viable antiviral drug target against caliciviruses.

MATERIALS AND METHODS
Cell lines, GECI lentiviruses, and viruses. All experiments were performed in LLC-MK2 cells. Lentivirus packaging and recombinant protein expression for Western blot lysate production was performed in HEK293FT cells (ATCC CRL-3216). Cell lines were grown in high-glucose Dulbecco modified Eagle medium (DMEM) (catalog no. D6429; Sigma) containing 10% fetal bovine serum (FBS) (Corning lot no. 35010167) and antibiotic/antimycotic (Invitrogen), and maintained at 37°C with 5% CO 2 . Lentivirus packaging in HEK293FT cells was performed as previously described (42). Briefly, LLC-MK2 cells were transduced with a lentivirus vector encoding GCaMP6s 1 day after seeding (ϳ85% confluence). We confirmed positive expression of GCaMP6s 48 to 72 h after transduction and then passaged cells 1:2 and added hygromycin (100 g/ml) for selection of the LLC-MK2-GCaMP6s cell lines, henceforth referred to as MK2-G6s. We determined GCaMP6s activity and dynamic range using thapsigargin (TG) (0.5 M). Tulane virus (TV) stocks were made in-house by infecting cells with an MOI of 0.01 and harvesting at ϳ95% cytopathic effect (CPE). Virus titer was determined by plaque assay. Irradiated virus controls were made by gamma-irradiating TV stocks for 19 h. MNV-1 CW1 virus was a kind gift from Herbert Virgin, and BV2 cells were a kind gift from Christiane Wobus. MNV-1 stocks were made by infecting BV2 cells at an MOI of 0.01 and harvesting at ϳ95% CPE. Virus titers were determined by plaque assay on BV2 cells using a similar protocol as for TV plaque assays except the final overlay was 1.2% Avicel.
Replication assays. LLC-MK2 cells were seeded at 125,000 cells/well in 24-well plates (Costar 3524; Corning) and inoculated the next day with TV at an MOI of 1 for 1 h. Inoculum was removed, and cell medium was replaced containing different extracellular Ca 2ϩ conditions (0 mM Ca 2ϩ , 4 mM Ca 2ϩ ), intracellular Ca 2ϩ chelator 50 M 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acidacetoxymethyl ester (BAPTA-AM), or the sarco/endoplasmic reticulum calcium ATPase (SERCA) blocker thapsigargin (TG). Ca 2ϩ -free DMEM was purchased from Gibco (catalog no. 21068-028). Standard high-glucose DMEM (Sigma) has 1.8 mM CaCl 2 , which we refer to as "2 mM Ca 2ϩ ," and media with 4 mM Ca 2ϩ was made by adding 2 mM CaCl 2 to the standard high-glucose DMEM (Sigma). We maintained TV-infected cells under these conditions until the positive control (normal media) had ϳ90% CPE. Progeny virus was harvested by three freeze/thaw cycles, and the virus yield was determined by plaque assay. For plaque assays, cells are seeded at 75,000 cells/well in 24-well plates, and 2 days after seeding, the cells were inoculated for 1 h with 10-fold serial dilutions of the sample. Then, we removed the inoculum and added the overlay. Overlays for plaque assays were made by mixing equal parts of 1.2% Avicel (FMC Corporation) and 2ϫ DMEM (Gibco). Plaque assays were harvested at 72 h and fixed and stained with crystal violet (3% solution) to visualize plaques. Titer is represented as plaque-forming units per milliliter (PFU/ml). To compare plaque size, images of wells were analyzed using Nikon Elements software to measure the longest diameter, and the resulting data were graphed using GraphPad Prism software.
One-step growth curves. One-step growth curves for TV were performed using a modified protocol from previous reports (11,15). Briefly, LLC-MK2 cells were inoculated with TV at an MOI of 1 in serum-free DMEM (0% FBS DMEM). At 1 h postinfection (HPI), the inoculum was removed and replaced with 0% FBS DMEM. Cells were harvested at 0, 4,6,8,10,12,16,20,24, and 28 HPI, and virus yield was determined by plaque assay. Each biological replicate was performed in duplicate.
Long-term Ca 2؉ imaging experiments. Calcium imaging experiments were set up by adapting a protocol detailed in previous reports (66). For TV infections, MK2-G6s cells were seeded at 78,500 cells/well in 15 -slide 8-well chambers (Ibidi, Germany) and infected the next day with TV at the indicated MOI. After 1 h, the inoculum was removed and replaced with FluoroBrite DMEM (Gibco). For studies involving pharmacological compounds, the FluoroBrite DMEM was mixed with dimethyl sulfoxide (DMSO) (0.1%; vehicle control) or the indicated pharmacological compounds dissolved in DMSO. For MNV-1 infections, BV2-G6s cells were seeded at 150,000 cells/well in 15 -slide 8-well chambers and infected the next day with MNV-1 strain CW1 at the indicated MOI. After 1 h, the inoculum was removed and replaced with FluoroBrite DMEM as described above. Then the slide was mounted into a stage-top environmental chamber (Okolab H301-Mini) maintained at 37°C with humidity control and 5% CO 2 . Time-lapse live-cell Ca 2ϩ imaging experiments were conducted from ϳ2 HPI until ϳ18 to 24 HPI on a Nikon TiE epifluorescence microscope using a Spectrax LED light source (Lumencor) and a 20ϫ Plan Apo (numerical aperture, 0.75) objective. Images were acquired at 1 or 2 images/well point/minute. Images were acquired and analyzed using the NIS elements advanced research software package (Nikon). Prior to image analysis, background camera noise was subtracted from the images using an averaged file of 10 no-light camera images. Cells that underwent division during the imaging run were excluded from analysis. Intracellular Ca 2ϩ signaling over time was quantified by calculating the number of Ca 2ϩ spikes per cell. This was determined as follows: raw fluorescence intensity values were measured from individual cells using Nikon software, then exported to Microsoft Excel to normalize the fluorescence to the first image (F/F 0 ). The Ca 2ϩ spikes were calculated by subtracting each normalized fluorescence measurement from the previous measurement to determine the change in GCaMP6s fluorescence (ΔF) between each time point. Ca 2ϩ signals with a ΔF magnitude of Ͼ5% were counted as Ca 2ϩ spikes. For each condition tested, Ca 2ϩ spikes in Ն30 cells were determined.
Heatmap generation. To generate heatmaps of the normalized GCaMP6s fluorescence over time for long-term Ca 2ϩ imaging experiments, we used the TidyR (75) and ggplot2 (76) packages available through R studio. Normalized GCaMP6s data from Excel was used to create an R-compatible file (.csv) containing the normalized fluorescence and the acquisition time data for the data set, and the file was imported into R. We used the TidyR package to organize data into a format accessible by ggplot2. We then used ggplot2 to generate heatmaps.
Prediction of viroporin motifs in silico. We used the Hydropathy Analysis program at the Transporter Classification Database to generate Kyte and Doolittle Hydropathy and Amphipathic moment plots to identify putative viroporin motifs within full-length TV NS1-2 (77). Secondary structure, membrane topology, and membrane integration predictions were performed using PSIPred prediction analysis suite (website http://bioinf.cs.ucl.ac.uk/introduction/) (55). Helical wheel plots to identify clustered basic residues within the putative viroporin domain were generated using the PepWheel analysis program at Heliquest (website http://heliquest.ipmc.cnrs.fr/) (56).
Expression vectors. E. coli expression constructs for the lysis assay were generated via ligationindependent cloning (LIC) using the pET46-Ek/LIC kit (MilliporeSigma, Darmstadt, Germany). The pET46-Ek/LIC constructs all have an N-terminal six-histidine tag. Mammalian expression vectors were generated by inserting c-myc tag and mRuby3 red fluorescent protein (RFP) upstream of full-length NS1-2 and then subcloning this into the pTagRFP-N vector in place of TagRFP (Epoch Life Sciences, Missouri City, TX). This construct is referred to as RFP-NS1-2. The NS1-2(Δ176) and NS1-2(Δ157) truncation mutations in both bacterial and mammalian expression vectors were generated using the NEB Q5 site-directed mutagenesis kit (New England Biolabs, Ipswich, MA). Primer sequences used for the bacterial and mammalian expression vectors are listed in Fig. S4 in the supplemental material. The sequences of all constructs were verified using universal primers specific to the construct backbone (GENEWIZ, South Plainfield, NJ). The mammalian expression vector for EV 2B was generated by cloning the 2B from enterovirus 71 upstream into pTagRFP-N, and the construction of the NSP4-TagRFP expression vector was previously described (78).
Transfection experiments. MK2-G6s cells were seeded in 15 -slide 8-well chambers (Ibidi, Germany) and at 85% confluence transfected with mammalian expression constructs in Opti-MEM (ThermoFisher) and Lipofectamine 2000 (Invitrogen). Transfection was optimized so cells received 400 ng of plasmid DNA and 0.5 l of Lipofectamine 2000 per well. Trichostatin A (TSA) (10 M) was added from 1 to 3 h posttransfection. TSA is a histone deacetylase (HDAC) inhibitor used to increase expression from the vectors (79)(80)(81). Time-lapse Ca 2ϩ imaging was performed beginning 8 h posttransfection to capture expression kinetics and up to 24 h posttransfection to measure changes in Ca 2ϩ signaling during expression of the RFP-tagged proteins.
E. coli lysis assay. E. coli lysis assays were performed as previously described (42). Briefly, pET46-Ek/LIC constructs of the full-length TV NS1-2 and truncation mutants were transformed into E. coli BL21(DE3)pLysS cells. Transformations were plated on LB containing 1% glucose, 100 g/ml ampicillin, and 35 g/ml chloramphenicol and grown at 37°C overnight. Isolated colonies were picked the next day and cultured overnight in liquid LB containing 1% glucose, 100 g/ml ampicillin, and 35 g/ml chloramphenicol at 37°C in an orbital shaker at 250 rpm. The next day, overnight cultures were subcultured by 1:100 dilution into 200 ml LB containing 1% glucose, 100 g/ml ampicillin, and 35 g/ml chloramphenicol. Subcultures were grown at 37°C in an orbital shaker at 250 rpm for ϳ3 h to an optical density at 630 nm (OD 630 ) between 0.3 and 0.5 and then induced with 1 mM isopropyl-␤-D-thiogalactopyranoside (IPTG). Absorbance measurements at 630 nm (OD 630 ) were taken every 10 min for 90 min and normalized to the induction OD 630 to determine the percent growth or lysis over time after induction. Each experiment was performed Ն3 times. Protein expression was determined by SDS-PAGE using a 4 to 20% Tris-glycine gel (Bio-Rad, Hercules, CA) and Western blotting for the six-histidine tag. An uninduced culture served as the negative control for viroporin activity and NS1-2 synthesis.
Membrane association experiment. Membrane association experiments were performed using a modified protocol from previously reported experiments (37,42). For bacterial membrane association, we collected lysed membranes from a 200-ml induced culture. For mammalian membrane association experiments, we collected lysed membranes in 500 l of radioimmunoprecipitation assay (RIPA) buffer with protease inhibitor from a transfected well of a six-well plate. Lysed membranes were centrifuged at 21,000 ϫ g for 20 min, and supernatants were decanted. Pellets were resuspended in phosphatebuffered saline (PBS) and sonicated three times for 1 min on ice. Total lysate was collected after sonication. The membranes were then pelleted by ultracentrifugation at 49,000 ϫ g for 1 h using a TLA-100.3 rotor in an Optima TL ultracentrifuge (Beckman Coulter, Indianapolis, IN), and the supernatant was collected for the soluble fraction. Finally, the membrane fraction pellet was resuspended in PBS containing 1% SDS to solubilize membrane proteins. Samples from the total lysate, soluble fraction, and membrane fractions were analyzed by Western blotting.
Production of TV and Vpg antisera. For the anti-TV antisera to detect VP1, adult male and female CD-1 mice (purchased from the Center for Comparative Medicine, Baylor College of Medicine) were immunized five times with CsCl 2 gradient-purified TV at 10 g/dose in AddaVax adjuvant (InvivoGen). Immunizations were given at 3-week intervals. For the anti-Vpg antisera, adult BALB/c were immunized three times with 10 to 20 g of purified Vpg expressed in E. coli per dose. The priming dose was given in Freund's complete adjuvant, and the subsequent boosts were given in Freund's incomplete adjuvant. Figure S4 shows immunoblot analysis of the antisera. All experiments were performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals (83).
Immunoblot analysis. Samples were prepared using procedures adapted from reference 37. Briefly, samples were mixed with 5ϫ sample buffer containing 2-mercaptonethanol and boiled for 10 min at 100°C. Samples were then run on a 4 to 20% Tris-glycine gel (Bio-Rad, Hercules CA) and transferred onto a nitrocellulose membrane using the Transblot Turbo transfer system (Bio-Rad, Hercules, CA). To detect the bacterial constructs of NS1-2 and NSP4, we used the mouse anti-His tag monoclonal antibody at 1:1,000 (Genscript, Piscataway, NJ). To detect mammalian expression constructs of NS1-2, we used the mouse anti-c-Myc monoclonal antibody (clone 9E10) at 1:1,000 (R&D Systems, MN). To detect TV structural protein VP1, we used the mouse anti-TV polyclonal antibody we made in-house by hyperimmunizing CD1 mice with purified TV particles. To detect TV nonstructural protein Vpg, we used the mouse anti-Vpg polyclonal antibody made by hyperimmunizing mice with bacterially expressed and purified Vpg. For loading control of mammalian cell lysates, we used the mouse anti-glyceraldehyde-3phosphate dehydrogenase (anti-GAPDH) at 1:3,000 (Novus Biologicals, CO). For secondary detection of all primary antibodies used in these experiments, we used alkaline phosphatase-conjugated goat anti-mouse IgG at 1:2,000 (Southern Biotech, Birmingham, AL) and visualized using alkaline phosphatase substrate (Tris-base, nitro blue tetrazolium [NBT], 5-bromo-4-chloro-3-indolyl phosphate [BCIP]). We used a PageRuler 10-to 180-kDa prestained protein ladder for all of our Western blots (ThermoFisher).
Statistical analysis. Statistical analyses were completed using GraphPad Prism (version 7.03). Data in this article are presented as means Ϯ standard deviations. Unless otherwise noted, all experiments in this article were performed in biological triplicate, with at least two technical duplicates per biological replicate, when applicable. We performed column statistics to collect descriptive statistics and to determine the normality of the data sets. We then used the unpaired Student's t test for data sets with a parametric distribution or a Mann-Whitney test for data sets with a nonparametric distribution. Differences were determined statistically significant if the P value was Ͻ0.05. Authors had access to the data for this article, and all authors approved the final article.
Data availability. RConsole code for the heatmaps generated in this paper is available upon request.