Strain-specific requirements of susceptibility to rhesus enteric calicivirus infection

ABSTRACT

Recently, we identified the coxsackie and adenovirus receptor (CAR) as the entry receptor for rhesus enteric calicivirus (ReCV) isolate FT285 and demonstrated that co-expression of the CAR and the type B histo-blood group antigen (HBGA) is required to convert the resistant CHO cell line susceptible to infection. To address whether the CAR is also the functional entry receptor for other ReCV isolates and the requirement for specific HBGAs or other glycans, here we used a panel of recombinant CHO cell lines expressing the CAR and the type A, B, or H HBGAs alone or in combination. Infection studies with three diverse ReCV strains, the prototype GI.1 Tulane virus (TV), GI.2 ReCV-FT285, and GI.3 ReCV-FT7, identified that cell surface expression of the CAR is an absolute requirement for all three strains to promote susceptibility to infection, while the requirement for HBGAs varies among the strains. In addition to the CAR, ReCV-FT285 and TV require type A or B HBGAs for infection. In the absence of HBGAs, TV, but not Re-CV FT285, can also utilize sialic acids, while ReCV-FT7 infection is HBGA-independent and relies on CAR and sialic acid expression. In summary, we demonstrated strain-specific diversity of susceptibility requirements for ReCV infections and that CAR, type A and B HBGA, and sialic acid expression control susceptibility to infection with the three ReCV isolates studied. Our study also indicates that the correlation between in vitro HBGA binding and HBGAs required for infection is relatively high, but not absolute. This has direct implications for human noroviruses.

IMPORTANCE

Human noroviruses (HuNoVs) are important enteric pathogens. The lack of a robust HuNoV cell culture system is a bottleneck for HuNoV cell culture-based studies. Often, cell culture-adapted caliciviruses that rapidly replicate in conventional cell lines and recapitulate biological features of HuNoVs are utilized as surrogates. Particularly, rhesus enteric caliciviruses (ReCVs) display remarkable similarities, including the primate host, clinical manifestation of gastroenteritis, genetic/antigenic diversity, and reliance on histo-blood group antigens (HBGAs) for attachment. While the HuNoV entry receptor(s) is unknown, the coxsackie and adenovirus receptor (CAR) has recently been identified as the ReCV entry receptor. Here, we identified the CAR, the type A and B HBGAs, and sialic acids as critical cell surface molecules controlling susceptibility to ReCV infections. The CAR is required for all ReCV isolates studied. However, the requirement for the different carbohydrate molecules varies among different ReCV strains. Our findings have direct implications for HuNoVs.

INTRODUCTION

Human noroviruses (HuNoVs) are one of the leading causes of acute gastroenteritis in all age groups in developed and developing countries (1). The prototype Norwalk virus was discovered in the early 1970s (2), but the progress of HuNoV research was long-time hampered by the lack of a cell culture system or animal model. Recently, the establishment of HuNoV enteroid cultures significantly increased our ability to study HuNoVs (3). However, this system still has several limitations. At present, enteroid monolayers are infected with stool-derived inoculum, and while several HuNoV strains have been shown to replicate in enteroid cultures, passaging of HuNoVs is not yet achieved. A simple and robust cell culture system allowing the continuous propagation of HuNoVs and the production of virus stocks is still lacking.

To overcome some of the difficulties associated with HuNoVs, tissue culture-propagable caliciviruses, including murine noroviruses (MNVs) and recoviruses (ReCVs), are frequently used as HuNoV surrogates. These viruses rapidly replicate in conventional cell lines and recapitulate several features of HuNoVs. Particularly, ReCVs display remarkable biological similarities to HuNoVs, including the primate host, clinical manifestation of acute gastroenteritis in the infected host, genetic and antigenic diversity, and reliance on histo-blood group antigens (HBGAs) for attachment (4). While the HuNoV entry receptor(s) remains to be identified, the coxsackie and adenovirus receptor (CAR) has recently been identified as the ReCV entry receptor (5). It was demonstrated that infection of naturally resistant CHO cells with an ReCV strain (ReCV-FT285), which in in vitro assays showed interaction only with the type B HBGA, could infect and replicate in recombinant CHO cells co-expressing the CAR and the type B HBGA, while CHO cells expressing the CAR or the type B HBGA alone remained resistant to infection. Since both HuNoVs and ReCVs exhibit remarkable genetic and antigenic diversity and several HBGA binding types have been identified (A-binder, B-binder, AB binder, non-binder … etc.), we are interested in investigating whether CAR expression is also required for other ReCV strains and the correlation between HBGA binding types and the requirement for cell surface expression of corresponding HBGAs to promote susceptibility to infection. Besides the prototype Tulane virus, over 15 well-characterized cell culture-adapted ReCV isolates are available for such studies in our laboratory. These viruses represent three genotypes, at least four serotypes, and three different HBGA binding types (4). Here, we used a panel of recombinant CHO (rCHO) cell lines (Fig. 1) and three ReCV isolates representing the three different HBGA binding types to evaluate the role of the CAR, HBGAs, and possibly other cell surface molecules as susceptibility determinants for ReCV infections.

Fig 1.

HBGA and CAR expression profile of rCHO-CAR+ cell lines used in this study. Recombinant CHO cell lines were fixed with 3% PFA without permeabilization and stained with anti-CAR (3C100), anti-A, and anti- B blood group antigen antibodies and goat anti-mouse IgG-Alexa 488 or goat anti-mouse IgM-Alexa 594 secondary antibodies, respectively. The H antigen was detected by Ulex Europaeus Agglutinin I (UEA I)-Rhodamine (Vector Laboratories). Images of CHO cells and rCHO cell lines expressing only HBGAs are not shown but were previously published in (5).

RESULTS

rCHO cells expressing HBGAs are resistant to ReCV infection

rCHO cell lines expressing either the type A, B, or H HBGAs remained resistant to infection by all ReCVs used in this study. Thus, HBGA expression alone is not sufficient to promote susceptibility to ReCV infection (Fig. 2).

Fig 2.

Infection of rCHO cell lines expressing the type H, A, and B HBGAs with three different ReCVs. Virus titers are shown at 1 h (after incubation with 1 MOI ReCV) and at 72 h post-infection. The lack of increase of virus titers at 72 h indicated that none of the HBGA expressing cell lines supported ReCV virus replication. Each experiment has been repeated three times. Error bars represent ±SD. Statistical significance was not calculated since no virus replication was indicated.

ReCV-FT285 infection requires the co-expression of the CAR and either the type A or B HBGAs

In our previous study, we established that ReCV-FT285 infects recombinant rCHO cells that co-express the type B HBGA and CAR (rCHO-B+/CAR+), but rCHO cells expressing the type B HBGA or CAR alone remain resistant to infection (5). This correlated well with the results of our saliva-based studies that indicated an interaction between ReCV-FT285 and the type B, but not type H or A, HBGAs (4). Here, we confirmed the susceptibility of rCHO-B+/CAR+ cells, but unexpectedly rCHO-A+/CAR+ cells were also susceptible to ReCV-FT285 infection. As expected, rCHO-H+/CAR+and rCHO-CAR+ cells remained resistant to ReCV-FT285 infection, even at a high (10 MOI) infection dose (Fig. 3).

Fig 3.

Infection of CAR-expressing recombinant CHO cell lines with three different ReCVs. Virus titers are shown at 1 h and 72 h post-infection. An increase in virus titers at 72 h indicates infection and virus replication in the corresponding rCHO cell line. Each experiment has been repeated at least three times. Statistical significance between 1 h and 72 h titers was calculated by paired t-test (one-tailed). Error bars represent ±SD. *= P < .05.

Tulane virus and ReCV-FT7 infects all CAR-expressing rCHO cell lines

Surprisingly, all rCHO cell lines expressing the CAR were susceptible to both TV and ReCV-FT7 infection. However, TV [a type A and B HBGA binder (4, 6)] replicated at significantly higher titers in rCHO-A+/CAR+and rCHO-B+/CAR+ cells than in rCHO-H+/CAR+and rCHO-CAR+ cells. TV replication efficiency in the latter two cell lines was equivalent, indicating that the type H antigen has no role in susceptibility to TV infection (Fig. 3). On the other hand, ReCV-FT7, a non-binder strain (4), replicated at equally high titers in all four CAR-expressing rCHO cell lines, regardless of the presence or absence of any of the HBGAs, indicating that this strain is not utilizing HBGAs at all for attachment (Fig. 3). Based on these observations, the CAR alone or perhaps in interaction with some other cell surface component(s) promotes susceptibility of rCHO-CAR+ cells to TV and ReCV-FT7 infection.

Cell surface carbohydrates other than HBGAs are involved in ReCV-FT7 infection

To evaluate whether cell surface carbohydrate structures are still involved in promoting susceptibility to ReCV-FT7 infection, sodium periodate (NaIO₄) was used to oxidatively cleave cell surface carbohydrate moieties on rCHO-CAR+ cells. Pretreatment of rCHO-CAR+ cells with both 1 mM and 5 mM NaIO₄ significantly reduced ReCV-FT7 titers compared to the mock-treated control, while cell viability remained unaffected (Fig. 4). Mild NaIO₄ (1 mM) treatment selectively oxidizes the side chain of terminal sialic acids (7), and these results suggested the involvement of terminal sialic acid structures in ReCV-FT7 and perhaps TV infections of rCHO-CAR+ cells.

Fig 4.

Sodium periodate treatment of rCHO-CAR+monolayers significantly reduced ReCV-FT7 infection. Titers shown at 24 h post-infection. Each experiment has been repeated three times. Statistical significance was calculated by one-way ANOVA, multiple comparisons. Error bars represent ±SD. *= P < .05.

CAR glycosylation is not required for ReCV-FT7 susceptibility

CAR has two confirmed N-glycosylation sites at positions N106 and N201 (8). To address whether the oxidation of glycans on the CAR contributed to the reduced infection after NaIO₄ treatment, we treated rCHO-CAR+ cells with tunicamycin to block N-glycosylation. Tunicamycin treatment completely reduced the CAR to its un-glycosylated form without affecting cell viability, but it had no effect on ReCV infection (Fig. 5A and B). This indicates that the cell surface carbohydrates (presumably sialic acid) on rCHO-CAR+ cells that promote susceptibility to ReCV-FT7 infection are not part of CAR glycosylation. Furthermore, CAR glycosylation is not required for its function as an ReCV receptor.

Fig 5.

Tunicamycin treatment of CHO-CAR+monolayers reduced the CAR to its un-glycosylated form (A) but had no effect on ReCV-FT7 infection (B). MW: molecular weight marker. Titers shown at 24 h post-infection. Each experiment has been repeated three times. Statistical significance was calculated by paired t-test (two-tailed). Error bars represent ±SD. *= P < .05.

N-acetylneuraminic acid treatment does not inhibit but increases ReCV-FT7 infection

Binding of the TV to the sialic acid containing molecule N-acetylneuraminic acid (Neu5Ac), but not to N-glycolylneuraminic acid (Neu5Gc), has been previously reported (9). Thus, we anticipated that preincubation of ReCV-FT7 with Neu5Ac would result in reduced infectivity of rCHO-CAR+ cells due to the competition of Neu5Ac and the cell surface sialic acid molecules for the binding sites on the virion. Surprisingly, after Neu5Ac pretreatment, a clear, concentration-dependent increase in ReCV-FT7 infectivity was observed compared to the mock-treated cultures (Fig. 6). A similar increase in infectivity after preincubation of the TV and several other ReCVs with synthetic HBGAs has been reported in our previous studies (4, 7).

Fig 6.

N-acetylneuraminic acid (Neu5Ac) pre-treatment of ReCV-FT7 resulted in a concentration-dependent increase in infectivity. Titers shown at 24 h post-infection. Each experiment has been repeated three times. Statistical significance was calculated by one-way ANOVA, multiple comparisons. Error bars represent ±SD. *= P < .05.

rCHO-CAR+ cells are more sensitive than sialylation-deficient rLec2-CAR+ cells to ReCV infection

Lec2 cells are CHO cell glycosylation mutants with about 80% reduction in sialylation of glycoproteins and gangliosides compared to wild-type CHO cells (10). Comparison of virus growth in rCHO-CAR+and rLec2-CAR+ cells infected with TV or ReCV-FT7 revealed that both viruses replicated less efficiently in Lec2-CAR+ than in rCHO-CAR+ cells, indicating that the levels of cell surface sialic acids correlate with infectivity (Fig. 7).

Fig 7.

Infection of rCHO-CAR+and rLec2-CAR+ cells. Both TV and ReCV-FT7 infect sialic acid-deficient rLec2-CAR+ cells less efficiently than rCHO-CAR+ cells. Titers shown at 24 h post-infection. Each experiment has been repeated three times. Statistical significance was calculated by one-way ANOVA, multiple comparisons. Error bars represent ±SD. *= P < .05.

Maackia amurensis lectin II treatment of rCHO-CAR+ cells does not inhibit but increases ReCV-FT7 infectivity

CHO cells lack functional α2,6-sialyltransferase (ST6GAL1) and, therefore, have incomplete sialylation with only α2,3-linked sialic acid (11). Sambucus nigra lectin (SNL) and Maackia amurensis lectin II (MALII) are sialic acid-binding lectins that specifically bind to α2,6- and α2,3-linked sialic acids, respectively (12). Staining of rCHO-CAR+monolayers with CY3-conjugated SNA or biotinylated MALII and CY3-conjugated streptavidin confirmed the presence of only α2,3-linked sialic acids on these cells that could be significantly depleted by neuraminidase treatment (Fig. 8A). When rCHO-CAR+ cells were pre-incubated with MALII prior to infection, instead of a decrease, we observed an increase (~1 log) in ReCV-FT7 infectivity compared to the mock-treated cultures (Fig. 8B). This was unexpected; however, a similar observation has been previously reported where MALI pretreatment of LLC-MK2 cells increased TV infectivity, possibly through MALI interaction with TV and rapid internalization of the cell surface-bound lectin–virus complex (9).

Fig 8.

MAL II binds to cell surface sialic acids on rCHO-CAR+ cells, but instead of blocking, it increases ReCV infectivity. (A) Monolayers were fixed with 3% PFA without permeabilization, blocked with HBS-Ca₂ −3% BSA, and stained with SNA-CY3 (α−2,6-linked sialic acid-specific lectin) or MALII-biotin (α−2,3-linked sialic acid-specific lectin) diluted in HBS-Ca₂ −1% BSA (20 µg/mL) overnight at 4°C. The MALII-treated wells were incubated with streptavidin-CY3 (10 µg/mL) for 1 hour at 37°C, before mounting and fluorescence imaging. (B) MALII pretreatment of rCHO-CAR+ cells enhances ReCV-FT7 infectivity. Monolayers were pre-incubated with MALII diluted in HBS-Ca₂ (100 µg/mL) for 2 hours at 37°C before infection with ReCV-FT7. Titers shown at 48 hours post-infection. Each experiment has been repeated three times. Statistical significance was calculated by one-way ANOVA, multiple comparisons. Error bars represent ±SD. *= P < .05.

Neuraminidase treatment of rCHO-CAR+ cells reduces ReCV-FT7 infectivity

Vibrio cholerae neuraminidase (VCN) is a sialidase that cleaves α2,3-, α2,6-, and α2,8-linked sialic acids with high efficiency. ReCV-FT7 infectivity was evaluated after treating rCHO-CAR+ cells by different concentrations (100–400 mU/mL) of VCN. ReCV-FT7 infectivity was reduced in a concentration-dependent manner, with a significant reduction at 200 and 400 mU/mL VCN (Fig. 9). Based on the trypan blue exclusion assay, there were no significant cell viability differences between the mock- or VCN-treated cell monolayers. However, cytotoxicity only observable by metabolic assays, but not by dye exclusion assays, has been reported after VCN treatment of tumor cells (13). To exclude the possibility that toxicity of VCN treatment causes reduced infectivity, we used the CellTiter-Glo Luminescent Cell Viability assay to evaluate ATP levels in these cultures. This test indicated a significant reduction in the metabolic activity of the VCN-treated cultures compared to the mock-treated control. However, while there was no difference between the metabolic activity among cultures treated with different VCN concentrations, a significant concentration-dependent reduction in infectivity was evident, clearly indicating the involvement of sialic acid in ReCV-FT7 infectivity (Fig. 9).

Fig 9.

Neuraminidase pre-treatment of CHO-CAR+ cells reduces ReCV-FT7 infectivity in a concentration-dependent manner. (A). Vibrio cholerae neuraminidase treatment at ≥200 mU/mL of rCHO-CAR+ cells significantly reduced ReCV-FT7 infectivity. (B) While a metabolic assay indicated 30% reduction in cell viability in all VCN-treated rCHO-CAR+cultures compared to the mock-treated culture, cell viability among the VCN-treated groups was almost identical (± 1%). Thus, infectivity differences among the VCN-treated groups were not due to viability differences but rather due to the difference in the amount of cell surface sialic acid removed. Values shown at 48 h post-infection. Each experiment has been repeated three times. Statistical significance was calculated by one-way ANOVA, multiple comparisons. Error bars represent ±SD. *= P < .05.

DISCUSSION

We used a panel of recombinant CHO cell lines and three diverse ReCV strains to evaluate the role of the CAR and specific HBGAs in promoting susceptibility to ReCV infections. CHO cells lack CAR and HBGA expression and are resistant to ReCV infection. However, when CHO cells are transfected with full-length ReCV genomic RNA, they produce infectious virus. Thus, CHO cells are able to support intracellular steps of the replication but are not able to support viral entry (permissive but not susceptible) (5). CHO cells allow the individual or combined expression of the CAR and HBGAs from expression vectors without any background from natural expression and can be used to dissect the role of suspected susceptibility determinants individually or in combination. Based on our previous studies, we selected three diverse genogroup I (GI) ReCV isolates from our collection, each representing a different genotype, serotype, and HBGA-binding type (4, 6). We clearly demonstrated that similarly to HuNoVs, the expression of HBGAs alone cannot promote susceptibility to ReCV infection, and that the CAR is a functional receptor for all three ReCVs studied (Fig. 2 and 3). Based on in vitro binding and blocking assays, we expected that ReCV-FT285 would only infect rCHO cells co-expressing the CAR and the type B HBGA. However, this strain also infected rCHO cells with CAR and type A HBGA expression. The ability of ReCV-FT285 to utilize type A HBGA has not been indicated by any assay in our previous studies, showing that while in vitro binding and blocking assays are good indicators of virus–host interactions, they are not absolute. While the mechanism behind this is not known, this observation could have implications for antiviral strategies and is worthy of further investigation. Some explanation could be that the sensitivity of the in vitro binding assay is above the concentration of ligands needed for biological functional activity or the contribution of other biological substances that are absent in the binding assay.

As expected, TV infected rCHO cells co-expressing CAR and the type A or B HBGAs. However, it also infected, but with lower efficiency, rCHO cells expressing only the CAR or CAR and the type H HBGA. Since the role of the type H HBGA in TV infection has not been indicated by any previous assay and the expression of the type H antigen did not increase the infectivity of rCHO-CAR+ cells compared to rCHO cells expressing only the CAR, it is most likely that the type H antigen has no role in TV infection.

Finally, ReCV-FT7 infected all CAR-expressing rCHO cell lines with the same efficiency, regardless of the co-expression or lack of any of the HBGAs. This is in accordance with the non-binder status of this strain and further indicates that HBGAs do not play a role in susceptibility to ReCV-FT7 infection.

The finding that TV and ReCV-FT7 were able to infect rCHO cells expressing the CAR only raised the question of whether the CAR alone is able to promote susceptibility to infection for these strains or if cell surface molecules other than HBGAs are still involved.

The most widely involved cell surface molecule in viral attachment is sialic acid. Several caliciviruses have been reported to utilize sialic acid-containing structures for attachment, including HuNoVs and the TV (9, 14–17). Treatment of rCHO-CAR+ cells with both 1 mM and 5 mM NaIO₄ significantly reduced ReCV-FT7 titers compared to the mock-treated control (Fig. 4), suggesting the involvement of terminal sialic acid structures in ReCV-FT7 infectivity. Since the CAR is a glycoprotein, we tested whether the reducing effect of NaIO₄ treatment on ReCV replication was the result of oxidative reduction of glycans on the CAR. The function of glycoproteins is mainly determined by the terminal sugar residues on the oligosaccharide chains and by the amino acid residues that the oligosaccharide chain attaches to the protein (N- vs O-glycosylation) (18). The CAR contains two validated N-glycosylation sites, one on each of its two extracellular immunoglobulin (Ig)-like domains (N106 in D1 and N201 in D2) (8). We used tunicamycin to block N-linked glycosylation in rCHO-CAR+ cells. Tunicamycin treatment reduced the CAR from its 46-kDa glycosylated form to its 40-kDa non-glycosylated form but had no effect on ReCV-FT7 infectivity (Fig. 5A and B), indicating that CAR glycosylation is not important to its function as an ReCV receptor. Moreover, since tunicamycin non-selectively interferes with all N-linked glycosylation of the cell, it is most likely that glycans, including terminal sialic acid linked to glycoproteins with N-linkage, are not involved in the susceptibility of rCHO-CAR+ cells to ReCV-FT7 infectivity.

We also compared TV and ReCV-FT7 infectivity in two CAR-expressing recombinant cell lines rCHO-CAR+and rLec2-CAR+. Both TV and ReCV-FT7 showed lower infectivity in the sialylation-deficient rLec2-CAR+ than in rCHO-CAR+ cells, further indicating the role of sialic acid in ReCV susceptibility (Fig. 7).

Sialic acids are commonly found at the terminal ends of carbohydrate chains of mucins, glycoproteins, glycolipids, and gangliosides. There are more than 50 structural variations of sialic acids in nature, but the most prevalent sialic acids found in mammalian cells are N-acetylneuraminic acid (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc). Glycoproteins expressed in CHO cells contain predominantly Neu5Ac, while depending on the CHO cell line, the Neu5Gc levels vary between 1 and 13% (19, 20). In an ELISA-based binding assay, TV showed strong binding to Neu5Ac but did not bind to Neu5Gc (9). We expected that preincubation of ReCV-FT7 with Neu5Ac before infection would result in reduced infectivity due to competition for sialic acid binding sites on the viral capsid. Surprisingly, instead of a blocking effect, we observed a concentration-dependent increase in infectivity (Fig. 6). Previously, we described a similar enhancement in ReCV infectivity by synthetic HBGAs (4, 7), and enhancement of a GII.4 HuNoV infectivity in B cell cultures by HBGA-expressing bacteria or synthetic type H HBGA has also been reported (21). The enhancement mechanism is not understood, although we suspect a conformational activation of the receptor-binding site upon HBGA/sialic acid binding to the virion (4). Structural studies for mapping the HBGA/sialic acid and CAR-binding sites and analysis of structural changes in the entry receptor binding site upon carbohydrate binding will be needed to identify the exact mechanism. Nevertheless, concentration-dependent increase in ReCV infectivity by NeuAc indicates an interaction between NeuAc and ReCV-FT7.

Terminal sialic acid is most frequently attached to galactose in α2,6- or α2,3-linkages. It was reported that Sambucus nigra lectin (SNL) that binds to α2,6-linked sialic acid reduced TV infectivity in LLC-MK2 cells, indicating that TV can utilize α2,6-linked sialic acid for attachment (9). The same study also demonstrated that preincubation of LLC-MK2 cells with the α2,3-linked sialic acid-specific Maackia amurensis leucoagglutinin (MAL I) did not block but enhanced TV infectivity, possibly through an interaction with TV and rapid internalization of MAL I-TV complexes. CHO cells lack functional α2,6-sialyltransferase (ST6GAL1) and, therefore, have incomplete sialylation with only α2,3-linked sialic acid (11). Since rCHO-CAR+ cells are susceptible to ReCV-FT7 and TV infection and blocking N-glycosylation did not affect the infectivity, we postulated that terminal α2,3-linked sialic acid on O-glycosylated glycoproteins or glycolipids may support attachment of these strains to rCHO-CAR+ cells. We used Maackia amurensis hemagglutinin (MAL II), that binds to O-linked glycans containing the trisaccharide Siaα2–3Galβ1–3GalNAc, to preincubate CHO-CAR+ cells prior to infection (12). Similar to what was reported with MAL I for the TV in LLC-MK2 cells, MAL II did not block but increased ReCV-FT7 infectivity in rCHO-CAR+ cells (Fig. 8). Thus, Maackia amurensis lectins did not provide a clear evidence for the involvement of α2,3-linked sialic acid in TV or ReCV-FT7 susceptibility in either study. Enhancement of viral infectivity by lectins has been reported for mainly enveloped viral pathogens and human lectins. Most likely, the interaction between glycans on viral glycoproteins of enveloped viruses and cell surface lectins drives enhanced internalization and infectivity (22, 23). The involvement of bacterial or plant lectins in enhanced viral infectivity has also been reported, where lectins most likely enhance attachment and entry by forming a bridge between viral glycoproteins and cell surface glycans (24). Since caliciviruses are non-enveloped and have only one major capsid protein without evidence of glycosylation, the mechanism for lectin enhancement of infectivity cannot be explained by either mechanism and needs further investigation.

Finally, treatment of the rCHO-CAR+monolayer with Vibrio cholerae neuraminidase at ≥200 mU/mL concentrations significantly reduced ReCV-FT7 infectivity (Fig. 9). Vibrio cholerae neuraminidase cleaves α2,3-, α2,6-, and α2,8-linked sialic acids. Since CHO cells only express α2,3-linked sialic acid (Fig. 8A), our result clearly established the role of α2,3-linked sialic acid in ReCV-FT7 infectivity. In LLC-MK2 cells, the role of α2,6-linked sialic acid in TV infectivity was clearly established by a significant reduction of infectivity with an α2,6-linked sialic acid-specific lectin (SNA) and after pretreatment with Vibrio cholerae and Arthrobacter ureafaciens neuraminidases that cleave α2,3-, α2,6-, and α2,8-linked sialic acids. Since an α2,3-linked sialic acid-specific sialidase (e.g., sialidase S) was not used in that study, and the lectin (MAL-I) did not block but promoted TV entry (9), the involvement of terminal α2,3-linked sialic acid in TV infectivity was not clearly established. In this study, we focused on the non-binder strain ReCV-FT7. Since CHO cells lack functional α2,6-sialyltransferase (ST6GAL1), the involvement of terminal α2,6-linked sialic acids in ReCV-FT7 infection remains to be evaluated. The detailed evaluation of TV infectivity in rCHO-CAR+ cells and studies to identify whether terminal sialic acids on N-linked, O-linked glycans (glycoproteins) or on glycolipids support ReCV attachment in rCHO-CAR+ cells are ongoing.

In summary, we demonstrated that the CAR in combination with HBGAs or/and terminal α2,3-linked sialic acid controls susceptibility of naturally resistant CHO cells to ReCV infection. Based on the requirement for these cell surface molecules to initiate infection, the three ReCV strains used in this study represented three novel groups. HBGAs are not able to support infection, while the CAR is an absolute requirement for all three strains. Group 1 (ReCV-FT285) requires the cell surface expression of the CAR and HBGA but cannot utilize sialic acid, group 2 (ReCV-FT7) requires the CAR and sialic acid but not HBGAs, and group 3 (TV) requires the CAR and is able to utilize both HBGAs and sialic acid to promote susceptibility to infection. The requirement for specific carbohydrates for infection was in a general correlation with the in vitro HBGA binding types of the strains, but this correlation is not absolute since ReCV-FT285, a type B HBGA binder, was also able to utilize the type A HBGA for infection. A similar observation had been made previously in a volunteer challenge study with the prototype Norwalk virus (25). While in saliva-based binding assays the Norwalk virus VLPs bound to saliva from group O and A secretors, binding to saliva from group B secretors was insignificant (similar to binding to saliva from non-secretors) (26). Hence, while none of the non-secretors became infected after Norwalk virus challenge, 43% of the seven type B volunteers developed infection (25). These indicate that in vitro binding assays do not fully represent the requirements for infection.

While the HuNoV entry receptor is yet to be discovered, the role of HBGAs and sialic acid has also been implicated in HuNoV infections. Targeting cellular components of viral entry is an effective antiviral strategy. The knowledge gained on ReCV susceptibility determinants and the research tools developed in this study could have direct implications for HuNoV research.

MATERIALS AND METHODS

Virus strains

Three genogroup I (GI) ReCV isolates, the prototype Tulane virus (TV, NC_043512, GI.1, type A and B binder), ReCV-FT285 (KC662366, GI.2, type B binder), and ReCV-FT7 (KC662368, GI.3, nonbinder), were used in this study. Virus stocks were grown in LLC-MK2 cells, filtered through 0.2-µm vacuum filter units, titrated (TCID₅₀/mL), aliquoted, and stored at −80°C until used. The HBGA binding type was determined in our previous studies and confirmed here by a saliva-blocking assay as described previously (4). This assay had a 100% correlation with saliva or synthetic oligosaccharide binding assays (ELISA).

Cell lines and antibodies

LLC-MK2 cells were maintained in M199 medium supplemented with 10% FBS and penicillin/streptomycin/amphotericin B (P/S/A). Control and recombinant Chinese hamster ovary (CHO) cell lines expressing the type H and B HBGAs were kindly provided by Dr. Jacques Le Pendu. CHO-Lec2 cells (CRL-1736) were purchased from ATCC. CHO cell lines were cultured in alpha-MEM supplemented with 10% FBS and P/S/A. Expression vectors pcDNA3.1/neo(+) carrying the human alpha-1–3-N-acetylgalactosaminyltransferase cDNA and pcDNA3.1/zeo(+) carrying the human CXADR cDNA were used to create the type A HBGA-expressing CHO cell line and the hCAR-expressing CHO cell lines, respectively. After several rounds of selection and cloning, the expression of corresponding antigens was evaluated by Western blot and immunofluorescence microscopy (Fig. 1). Anti-CAR monoclonal antibodies RmcB (Millipore Sigma, Cat# 05–644), 3C100 (Santa Cruz Biotechnology, Cat# sc-70493), and anti-A (ABO1) and anti-B (ABO2) monoclonal antibodies (Diagast) were used with corresponding fluorophore-labeled secondary antibodies to detect CAR, type A and type B HBGA expression, while Ulex Europeaus Agglutinin-Rhodamine (Vector Laboratories, RL-1062–2) was used for the detection of the type O HBGA (H-antigen). In Western blots, the CAR was detected by a rabbit anti-CAR polyclonal antibody (Thermo Fisher Scientific, PA5-31175) and HRP-conjugated goat anti-rabbit IgG secondary antibody.

Chemicals and enzymes

Sodium periodate (Avantor), tunicamycin (Avantor), N-acetylneuramic acid (NANA) (Millipore Sigma), Vibrio cholerae neuraminidase (Millipore Sigma), Ulex europaeus agglutinin (Vector Laboratories), Sambucus nigra lectin, and Maackia amurensis lectin II (Vector Laboratories) were used in this study.

Infection of recombinant CHO cells

Control and recombinant CHO cell monolayers were seeded onto 24-well plates (1 × 10⁵ cells/well) and incubated overnight at 37°C. The next day, the culture medium was aspirated, and cell monolayers were overlayed with 300 µL of the culture medium containing 1 MOI of the corresponding ReCV strain. Plates were incubated for 1 h at 37°C. Wells were washed three times with DPBS and overlaid with 0.5 mL of the culture medium. One plate was transferred to −20°C immediately (1 h timepoint) and replicated plates at corresponding time points 24 h, 48 h, and 72 h) post-infection. Plates were subjected to three freeze–thaw cycles, and culture media from individual wells was harvested. Cell debris was removed by centrifugation, and supernatants were titrated on 96-well plates seeded with LLC-MK2 cells as described below.

Virus titration

LLC-MK2 cells (1 × 10⁴ cells/well) seeded in 96-well plates the day prior were inoculated with ten-fold serial dilutions of samples (100 µL/well, 3–4 wells/dilution). Plates were incubated for 5 days and stained with crystal violet. Virus titers (TCID₅₀) were calculated using the Reed and Muench method (27).

Sodium periodate treatment

rCHO-CAR+monolayers were rinsed with HEPES-buffered saline (10 mM HEPES, 0.15 M NaCl, pH 7) (HBS) and incubated with HBS (mock) or 1 and 5 mM NaIO₄ in HBS for 30 min at 4°C. The unreacted periodate was blocked with 0.5% glycerol in HBS. Plates were washed twice in HBS, overlaid with 1 MOI of ReCV in culture media, and incubated for 1 hour at 37°C. The unbound virus was removed by washing twice with DPBS, culture medium was added, and plates were incubated in a CO₂ (5%) incubator at 37°C. At 24 hours post-infection, the supernatants were harvested and processed for virus titration.

Tunicamycin treatment

rCHO-CAR+monolayers were incubated in culture media supplemented with 1.5 and 3 µg/mL tunicamycin for 24 hours. Plates were washed twice with DPBS, overlaid with culture media containing 1 MOI of ReCV, and incubated for 1 hour at 37°C. Unbound virus was removed by washing twice with DPBS, and monolayers were overlaid with culture media and incubated for 24 hours. Plates were subjected to three freeze–thaw cycles, and culture media from individual wells were harvested. Cell debris was removed by centrifugation, and supernatants were titrated on 96-well plates seeded with LLC-MK2 cells.

N-acetyl-neuraminic acid treatment

ReCV-FT7 was pre-incubated with various concentrations (40–180 µM) of the sialic acid containing molecule N-acetyl-neuraminic acid (NANA) for 1 hour at 37°C, and the virus/NANA mixture was used to inoculate rCHO-CAR+monolayers at 1 MOI for 1 hour. Unbound virus was removed by washing twice with DPBS, plates were cultured and harvested 24 hours post-infection, and virus titrated as described previously.

Maackia amurensis lectin II treatment

rCHO-CAR+monolayers were pre-incubated with MALII diluted in HBS supplemented with 0.2 mM CaCl₂ (HBS-Ca_2, pH 5.5) (100 µg/mL) for 2 hours at 37°C before infection with ReCV-FT7. Virus titers were determined at 48 hours post-infection.

Neuraminidase treatment

rCHO-CAR+cell monolayers were treated with Vibrio cholerae neuraminidase (VCNA) to biochemically remove cell surface sialic acid. Monolayers in 96-well plates were rinsed with HBS-Ca₂ and then incubated with 100 µL HBS-Ca₂ supplemented with 100–400 mU/mL of VCNA for 1 hour at 37°C. Mock-treated monolayers were incubated with buffer only. Cells were washed with DPBS, overlaid with 1 MOI of ReCV for 1 hour at 37°C, washed twice to remove unbound virus, and incubated for 48 h. Virus titers were determined as described above.

Cell viability

Cell viability in all treated and mock cultures was evaluated by trypan blue exclusion assay. CellTiter-Glo Luminescent Cell Viability assay (Promega) was used to evaluate the metabolic activity of VCN-treated cells.

Immunofluorescence

Cell monolayers were grown on chamber slides, fixed with 3% paraformaldehyde (PFA) without permeabilization, and blocked with PBS-10% goat serum. Primary antibodies diluted in PBS-1% goat serum-0.1% Tween 20 were applied, and slides were incubated in a humidifier box overnight at 4°C. After washing three times with PBS-0.1% Tween 20 (PBS-T, pH 7.4), slides were overlaid with the corresponding Alexa dye-labeled secondary antibody and incubated for 1 hour at 37°C. Slides were washed three times with PBS-T, cover slips were mounted with ProLong Diamond Antifade Mountant with DAPI (Thermo Fisher Scientific), and slides were cured overnight in the dark at room temperature. The following day, the slides were sealed with nail polish and stored at room temperature in slide holders. Images were captured on an Evos FLoid Cell Imaging Station (Life Technologies).

Western blot analysis

Cells grown in T-25 flasks were washed with ice-cold PBS and lysed in 0.3 mL Pierce RIPA buffer (Thermo Fisher Scientific). The lysates were incubated on ice for 30 min with vortexing every 5 min. The insoluble material was removed by centrifugation at 12,000 rpm, for 15 min, at 37°C, and samples were aliquoted and stored at −80°C until use. Protein concentration was determined by the Bradford protein assay (Bio-Rad). For analysis, samples (~20 µg protein) were mixed with an equal volume of 2X Laemmli sample buffer (Bio-Rad), boiled for 5 min, separated on 4–20% Mini-Protean TGX precast protein gels (Bio-Rad), and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blocked with 5% nonfat dried milk in PBS-T at room temperature for 3 hours, rinsed with PBS-T, overlaid with primary antibody (0.25 µg/mL) diluted in blocking buffer, and incubated with agitation at 4°C overnight followed by three washes in PBS-T. Membranes were incubated with HRP-conjugated secondary antibody for 3 hours at room temperature and washed three times with PBS-T. Signals were detected with the SuperSignal West Pico PLUS chemiluminescent substrate (Thermo Fisher Scientific) using a ChemiDoc MP imaging system (BioRad). Subsequently, all membranes were stripped using Restore PLUS Western Blot stripping buffer (Thermo Fisher Scientific) and re-probed with mouse monoclonal antibody to GAPDH.

Statistical analyses

Experiments were performed in at least three independent replicates. Differences between virus titers were evaluated for statistical significance by one-way ANOVA or unpaired t-test using the GraphPad software (Dotmatics). A P-value ≤ 0.05 was considered statistically significant.