Characterization of Sialic Acid-Independent Simian Rotavirus Mutants in Viral Infection and Pathogenesis

ABSTRACT

Rotaviruses (RVs) are nonenveloped viruses that cause gastroenteritis in infants and young children. Sialic acid is an initial receptor, especially for animal RVs, including rhesus RV. Sialic acid binds to the VP8* subunit, a part of the outer capsid protein VP4 of RV. Although interactions between virus and glycan receptors influence tissue and host tropism and viral pathogenicity, research has long been limited to biochemical and structural studies due to the unavailability of an RV reverse genetics system. Here, we examined the importance of sialic acid in RV infections using recombinant RVs harboring mutations in sialic acid-binding sites in VP4 via a simian RV strain SA11-based reverse genetics system. RV VP4 mutants that could not bind to sialic acid had replicated to decreased viral titer in MA104 cells. Wild-type virus infectivity was reduced, while that of VP4 mutants was not affected in sialic acid-deficient cells. Unexpectedly, in vivo experiments demonstrated that VP4 mutants suppressed mouse pups’ weight gain and exacerbated diarrhea symptoms compared to wild-type viruses. Intestinal contents enhanced VP4 mutants’ infectivity. Thus, possibly via interactions with other unknown receptors and/or intestinal contents, VP4 mutants are more likely than wild-type viruses to proliferate in the murine intestine, causing diarrhea and weight loss. These results suggest that RVs binding sialic acid notably affect viral infection in vitro and viral pathogenesis in vivo.

IMPORTANCE Various studies have been conducted on the binding of VP8* and glycans, and the direct interaction between purified VP8* and glycans has been investigated by crystalline structure analyses. Here, we used a reverse genetics system to generate rotaviruses (RVs) with various VP4 mutants. The generated mutant strains clarified the importance of glycan binding in vitro and in vivo. Moreover, even when VP4 mutants could not bind to sialic acid, they were able to bind to an unknown receptor. As RVs evolve, pathogenicity can also be modified by easily altering the glycans to which VP4 binds.

INTRODUCTION

Rotaviruses (RVs) are responsible for most gastroenteritis cases in infants and young children, mainly in developing countries. An estimated 146,000 annual deaths are related to RV infection worldwide (1, 2).

The RV genome comprises 11 segments encoding six structural and six nonstructural proteins. The outer capsid proteins VP4 and VP7 are classified into the P and G genotypes, respectively, and these elicit neutralizing antibodies (3). The globally licensed RV vaccines Rotarix and RotaTeq are based on the G1, G2, G3, G4, and P[8] genotypes. Vaccines are reportedly effective in various genotypes, from prevalent to uncommon genotypes. The emergence of epidemic strains with different genotypes from vaccine strains has raised concerns about the effectiveness of vaccines in recent years (4).

Many infectious pathogens exploit diverse families of host glycans, such as histo-blood group antigens (HBGAs), sialoglycans, and glycosaminoglycans, for initial cell recognition and attachment (5). The RV outer capsid spike protein VP4 is cleaved by trypsin into two components: the VP5* and VP8* subunits. RV recognizes cell surface glycans by interacting with VP8* for initial cell attachment. After attaching to the host cell, VP5* and the outer capsid protein VP7 bind to integrins and heat shock cognate 70, enabling entry into cells (6–8). Most animal RVs recognize terminal sialic acids through the VP8* for cell attachment (9), whereas infection by most human RVs is insensitive to sialidase treatment (10, 11).

Recently, it was discovered that almost all human RVs, including P genotypes P[4], P[6], and P[8], recognize HBGAs as cell attachment receptors (12, 13). Sialic acid is anchored to VP8* of the rhesus RV via seven hydrogen bonds and six van der Waals contacts (14). The binding specificity between the virus and the glycan is determined by the affinity between the glycan and cleft between the four β-strands of VP8* (15). The crystal structure clarified that the VP8* cleft of animal RVs is narrower than that of human RV and could bind sialic acids (16). Previous studies have shown that isolated sialic acid-independent RRV variants are infectious without binding sialic acid (17). However, despite the analysis of the structural relationship between glycans and RV VP4s, the role of VP4 binding to glycans in the viral life cycle remains elusive. Here, to investigate the importance of sialic acid binding to VP8* in viral infection, sialic acid-independent SA11 mutants harboring amino acid mutations in the VP4 glycan binding pocket were generated and characterized, both in vitro and in vivo.

RESULTS

Generation of sialic acid-independent VP4 mutant viruses.

The simian RV strain SA11 is known to use sialic acid as an initial receptor (18), whereas human RV strains use other glycans, including HBGA and mucin core (19). To verify the glycan use of the simian and human RVs, the infectivity of recombinant strain SA11 (rSA11) and human RV recombinant strain Odelia (rOdelia) (20, 21) in MA104 cells treated with neuraminidase to remove cell surface sialic acids was examined. The infectivity of rSA11 was reduced by >90% by neuraminidase treatment, whereas the infectivity of rOdelia remained unaffected (Fig. 1A). In the binding assay by flow cytometry, the total number of cells binding to rSA11 decreased by 22% in neuraminidase-treated cells (Fig. 1B).

FIG 1.

Infectivity of sialic acid-independent VP4 mutants in MA104 cells. (A) Immunofluorescence micrographs of neuraminidase-treated MA104 cells infected with rSA11 and rOdelia (left). Scale bar, 200 μm. MA104 cells treated with neuraminidase (0, 0.1, and 1.0 U/mL) for 30 min were infected with rSA11 and rOdelia (100 FFU/well) (right). At 16 h postinfection, cells were fixed, and virus-infected cells were detected by immunofluorescent assay. The number of virus-infected cells was counted, and the infectivity ratio is presented as a percentage relative to that in the neuraminidase-free control. (B) Flow cytometry analysis of the binding ability of rSA11 to neuraminidase-treated MA104 cells. MA104 cells were bound with 10 FFU/cell of rSA11 at 4°C for 1 h. The ratio of cells bound to viruses was analyzed by flow cytometry. (C) Crystal structural model of the VP8* of simian RV strain RRV (PDB ID 2P3K). Four β-strands comprising the glycan binding cleft are visualized in green. The amino acids for substitutions are indicated. (D) MA104 cells were infected with VP4 mutants with recombinant SA11 carrying VP4 mutants at an MOI of 0.01 FFU/cell. At 48 h postinfection, viral titers were examined. (E) Infectivity of VP4 mutants in MA104 cells treated with neuraminidase. (F) Growth kinetics of rSA11, rSA11_R101A, and rSA11_R101A/V143A/Y175A in MA104 cells. MA104 cells were infected with the viruses at an MOI of 0.01 FFU/cell and incubated. Viral titers from these experiments with triplicates were plotted as mean ± standard deviation (SD) (shown by error bars). Data are presented as the mean of three replicates. Significant differences are indicated as follows: *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, P < 0.001.

Previous structural analysis of the VP8* protein of the rhesus RV strain RRV revealed an interaction between the cleft formed by the four β-strands of VP8* and sialic acids. The four amino acids R101 (arginine at position 101), Y155, Y188, and S190 were expected to form strong hydrogen bonds with sialic acids (14). V143 is located at the edge of the cleft and is structurally close to the sialic acid-binding sites. Y175 stabilizes carbohydrate binding by interacting with Y188 (14). Since the amino acid sequences of SA11 VP4 are similar to those of RRV VP4, we generated SA11 viruses carrying VP4 with single or multiple alanine substitutions in amino acids R101, V143, Y155, Y175, and S190, which are located on the cleft formed by the four β-strands, to analyze the importance of the glycan binding cleft of VP8* in viral infection (Fig. 1C).

Viral replication of these VP4 mutants was lower than that of rSA11, implying that alanine substitutions in these amino acids impaired their interaction with sialic acid (Fig. 1D). More importantly, the infectivity of these VP4 mutants, except for rSA11_V143A, was resistant to neuraminidase treatment, suggesting that binding receptors other than sialic acid sustained viral infections (Fig. 1E). rSA11_R101A, which contains a mutation in amino acid R101 that leads to the formation of two hydrogen bonds in sialic acid (14), and rSA11_R101A/V143A/Y175A, which possesses the lowest viral titer (Fig. 1D), were selected for further experiments. The growth kinetics of rSA11_R101A and rSA11_R101A/V143A/Y175A were significantly lower than those of rSA11 in the early phase of infection, but they finally reached the same level as rSA11 at 72 h postinfection (Fig. 1F). These results indicate that amino acid mutations at sialic acid-binding sites within the VP8* cleft led to attenuated viral infectivity but could still bind to the cells independent of sialic acid.

Characterization of sialic acid-independent VP4 mutant viruses in sialic acid-deficient cells.

To verify the sialic acid-dependent infection of rSA11, we generated sialic acid-deficient HT29 cells by solute carrier family 35 member A1 (SLC35A1) gene knockout. Genome-wide CRISPR-Cas9 screening identified SLC35A1 as a gene involved in sialic acid biosynthesis, as a receptor for RV and reovirus (22, 23). Sialic acid is typically found in the terminal position of glycan chains attached to the cell surface and to secreted glycoproteins (24). The CMP-sialic acid biosynthesized within the nucleus is carried by the SLC35A1 to the medial- and trans-Golgi apparatus, where it is used for glycoprotein sialylation (25, 26). Therefore, by lacking SLC35A1, sialic acid is not transported to the cell surface, and the sialic acid-deficient state can be maintained. SLC35A1 knockout (SLC35A1-KO) HT29 cells were established by CRISPR-Cas9-mediated genome editing (27). Sequencing analysis showed the 37-bp deletion in SLC35A1 formed a stop codon 162 bp downstream from the start codon of SLC35A1. SLC35A1-KO cells were not stained by fluorescein isothiocyanate (FITC)-labeled elderberry bark lectin (SNA/EBL), indicating the loss of sialic acid on the cell surface (Fig. 2A). Treatment with neuraminidase did not reduce the infectivity of rSA11 in SLC35A1-KO cells, demonstrating that sialic acid is involved in the infection of the simian RV strain SA11 (Fig. 2A). Flow cytometric analysis revealed that rSA11 did not bind to the surface of SLC35A1-KO cells as much as was bound to the surface of control cells (Fig. 2B). Loss of SLC35A1 renders cells resistant to simian RV SA11 infection due to the absence of cell surface sialic acids. The infectivity of rSA11 VP4 mutants and rOdelia remained unaffected in SLC35A1-KO cells, while the infectivity of rSA11 decreased in SLC35A1-KO cells (Fig. 2C). These results were similar to those of the neuraminidase treatment test. The growth kinetics of mutants in SLC35A1-KO cells were analyzed; mutants could replicate to titers in SLC35A1-KO cells comparable to that of rSA11 in HT29 cells, whereas VP4 mutants in HT29 cells showed reduced titers (Fig. 2D to F). These results suggested that mutants modify the binding receptor and enable infection and replication by mutating the spike without sialic acid.

FIG 2.

VP4 mutant growth rates in SLC35A1-KO cells that lack cell surface sialic acids. (A) Analysis of sialic acid expression. Formalin-fixed HT29 and SLC35A1-KO cells were stained by 10 μg/mL FITC-labeled elderberry bark lectin (SNA/EBL) recognizing the terminal sialic acid. Scale bar, 5 μm (left). HT29 cells and SLC35A1-KO cells treated with neuraminidase from Clostridium perfringens (0, 0.1, and 1.0 U/mL) for 30 min at 37°C were infected with rSA11 (100 FFU/well). At 16 h postinfection, cells were fixed, and virus-infected cells were detected by immunofluorescent assay. The number of virus-infected cells was counted, and the infectivity ratio is presented as a percentage relative to infectivity in the neuraminidase-free control (right). (B) SLC35A1-KO cells were bound with 10 FFU/cell of rSA11 at 4°C for 1 h. The ratio of cells bound to viruses was analyzed by flow cytometry. (C) SLC35A1-KO cells were infected with rSA11, mutants, and rOdelia (100 FFU/well). Infected cells were fixed and counted at 16 h. The infectivity ratio is presented as a percentage relative to infectivity of HT29 cells normalized to 1. (D to F) rSA11 (D), rSA11_R101A (E), and rSA11_R101A/V143A/Y175A (F) growth kinetics in HT29 and SLC35A1-KO cells. Cells were infected with the viruses at an MOI of 0.01 FFU/cell and incubated. Data are given as the mean of three replicates. Values are means ± SD. Significant differences are indicated as follows: *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, P < 0.001.

Effect of HBGAs on infection of sialic acid-independent VP4 mutants.

Since the results thus far indicate that rSA11 mutants use unidentified binding receptor molecules, we attempted to identify the glycans to which the mutants bound in cultured cells. Unlike simian RVs, human RVs utilize HBGAs as binding receptors. HBGAs are carbohydrates linked to proteins or lipids of the red blood cells and mucosal epithelial surfaces of the urinary, respiratory, and gastrointestinal tracts (28, 29). HBGAs are synthesized by the addition of monosaccharides to the precursor glycan through glycosyltransferases encoded by ABO, Lewis, and secretor genes. Previously, it was shown that VP8* from the DS1 P[4] strain bound to H-type I HBGA and A-type HBGA, and VP8* from the Wa P[8] strain bound to H-type I HBGA (12, 30).

To confirm these results using a reverse genetics approach, recombinant SA11 strains carrying VP4 of DS1 P[4] or VP4 of Odelia P[8] (rSA11_P[4] DS1-VP4 and rSA11_P[8] Odelia-VP4, respectively) were generated. For the infectivity assays, HT29 cells expressing A-type HBGA (31) and Caco-2 cells expressing H-type I HBGA (32) were treated with anti-A-type or H-type I HBGA antibodies. Infection of rSA11_P[4] DS1-VP4 was inhibited by anti-A-type HBGA and anti-H-type I antibodies (Fig. 3A and B). Similarly, infectivity of rSA11_P[8] Odelia-VP4, which is the same genotype as the Wa strain, decreased in the presence of anti-H-type I HBGA antibodies (Fig. 3A and B). The infectivity of both rSA11_P[4] DS1-VP4 and rSA11_P[8] Odelia-VP4 was unaffected by neuraminidase treatment (Fig. 3C). In contrast, the infectivity of rSA11 was reduced by treatment with neuraminidase but not by the application of anti-A-type and anti-H-type I HBGA antibodies. These results confirmed the previous studies that P[4] and P[8] VP4 bind cells via HBGAs, while simian RV does not use HBGAs as binding receptors (16). Furthermore, these results substantiated the concept of the RV infection mechanism, that is, binding specificity to glycan receptors exclusively depends on the VP4 protein.

FIG 3.

Effect of HBGAs on infection of simian RV VP4 mutants. (A to C) Infectivity of rSA11 and monoreassortant viruses with VP4 genes of human RV on the genomic backbone of SA11. Cells were infected with the virus (100 FFU/well). After 16 h, infected cells were fixed and counted. The infectivity ratio is presented as percentage of infectivity relative to that in the control. (A) Infectivity of viruses in HT29 cells incubated with anti-A-type HBGA antibody at 4°C for 1 h. (B) Infectivity of viruses in Caco-2 cells incubated with anti-H-type I HBGA antibody. (C) Infectivity of viruses in neuraminidase treated MA104 cells. (D and E) Infectivity of rSA11, rSA11 VP4 mutants, and rSA11_P[4] DS-1-VP4 treated with (D) anti-A-type HBGA antibody and (E) anti-H-type I HBGA antibody in HT29 and Caco-2 cells. Data are given as the mean of three replicates. Values are means ± SD. Significant differences are indicated as follows: *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, P < 0.001.

Using the same inhibition assay, the glycan receptor usage of the rSA11 VP4 mutants was explored. The rSA11 VP4 mutant infectivity did not change in HT29 cells treated with the anti-A-type HBGA antibody. rSA11_R101A infectivity was inhibited by anti-H-type I HBGA antibody in a dose-dependent manner, whereas rSA11_R101A/V143A/Y175A infectivity was uninhibited (Fig. 3D and E). These results suggest that glycan usage by SA11 may be altered by a single amino acid substitution (i.e., R101A). Furthermore, as the infectivity of rSA11_R101A/V143A/Y175A was unaffected by HBGA antibody treatment, it was considered that glycan-binding specificity was determined by multiple amino acids, and unspecified molecules might be involved in virus-cell interactions in rSA11 infection.

Sialic acid-independent VP4 mutant viruses have increased pathogenicity in mice.

To assess the replication capacity of rSA11 VP4 mutants in vivo, 3-day-old inbred mice were orally inoculated (1 × 10⁵ focus-forming units [FFU]/head or 1 × 10⁶ FFU/head) with rSA11, rSA11_R101A, or rSA11_R101A/V143A/Y175A. Mouse pups infected with VP4 mutants exhibited significantly lower body weights than those infected with rSA11 (Fig. 4A and D). Diarrhea induced by inoculation with 1 × 10⁵ FFU/head of rSA11_R101A and rSA11_R101A/V143A/Y175A lasted longer than that of rSA11, and higher diarrhea scores were recorded (Fig. 4B). In contrast, mice inoculated with 1 × 10⁶ FFU/head exhibited similar diarrhea intensity between animals infected with rSA11 and VP4 mutants (Fig. 4E). This is due to severe dehydration and the fact that diarrhea was difficult to determine in the 1 × 10⁶ FFU/head-inoculated group. One mouse in each group infected with the VP4 mutants (1 × 10⁵ FFU/head) died (Fig. 4C). Furthermore, mice infected with VP4 mutants (1 × 10⁶ FFU/head) had higher mortality than mice infected with rSA11 (37% for rSA11 versus 58% for rSA11_R101A and 66% for rSA11_R101A/V143A/Y175A) (Fig. 4F). Additionally, mouse pups inoculated with VP4 mutants were smaller in size (Fig. 4G). To investigate whether pathogenicity and viral replication are correlated, mouse pups were infected with 1 × 10⁵ FFU/head of rSA11 and VP4 mutants, and viral gene copies were measured at the early stage of infection. The intestinal tract was divided into upper (small intestine) and lower (cecum and colon) parts; levels of viral replication of VP4 mutants were always higher than those of rSA11 in the upper and lower parts (Fig. 4H and I). Similarly, 3-week-old mice were infected with VP4 mutants, rSA11 was nearly undetectable in their small intestine and cecum, whereas VP4 mutants were detected (Fig. 4J and K). However, rSA11_R101A was detected only in the small intestine, whereas the rSA11_R101A/V143A/Y175A mutant was detected in the small intestine and cecum. These results suggest that mutations at sialic acid-binding sites can make RV more virulent and increase viral replication in a mouse model. Additionally, rSA11_R101A/V143A/Y175A virus was more likely to replicate in the intestinal tract than rSA11 and rSA11_R101A in adult mice.

FIG 4.

Pathogenicity of sialic acid-independent VP4 mutants in mice. Three-day-old BALB/cAJcl mice were inoculated with (A to C) 1 × 10⁵ FFU/head (n = 7 to 9 mice/group) or (D to F) 1 × 10⁶ FFU/head (n = 10 to 12 mice/group) of rSA11 and VP4 mutants. Mice were monitored for 7 days after inoculation. (A and D) Body weight; (B and E) diarrhea score; (C and F) Kaplan-Meier survival curves; (G) appearance of mouse pups (1 × 10⁵ FFU/head) on day 5 postinfection. Mouse pups were inoculated with PBS, inactivated rSA11_R101A, rSA11, rSA11_ R101A, and rSA11_ R101A/V143A/Y175A. (H and I) Three-day-old BALB/cAJcl mice were inoculated with 1 × 10⁵ FFU/head (n = 4 to 8 mice/group). After 12 and 24 h of viral infection, the intestines were removed, divided into (H) upper (small intestine) and (I) lower (cecum and colon) parts, frozen, thawed, and homogenized. Viral RNA was extracted from mixtures. (J and K) Three-week-old BALB/cAJcl mice were inoculated with rSA11 and mutants at 1 × 10⁵ FFU/head (n = 5 to 6 mice/group). Two days after infection, the mice were euthanized, and their intestines were collected. The intestinal tissues were frozen, thawed, and homogenized. RNA was extracted from the supernatant obtained by centrifuging the mixture. Viral genome copies of the (J) small intestine and (K) cecum were measured using qRT-PCR. Data are presented as the mean of five or more replicates. Values are means ± SD. Significant differences are indicated as follows: *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, P < 0.001.

Effect of intestinal contents on the infectivity of sialic acid-independent mutants.

The enhanced infectivity and replication of rSA11_R101A and rSA11_R101A/V143A/Y175A in the murine model were unexpected because the infectivity of these VP4 mutants in cell lines was significantly lower than that of rSA11. These results suggest that the in vivo and in vitro replication mechanisms differ. Previous studies regarding the association between intestinal contents, including microbiota, and RV infection have been conducted. Some probiotic bacteria are able to inhibit the binding of the virus to intestinal cells by blocking viral receptors and binding to viruses on the surface (33). However, some microbiotas also promote viral infections (34). To examine the effect of the intestinal microbiota on the enhancement of RV infectivity, rSA11 and rSA11 VP4 mutants were mixed with intestinal contents from the lower small intestine or cecum, incubated at 37°C, and inoculated into MA104 cells. The infectivity of rSA11_R101A and rSA11_R101A/V143A/Y175A incubated with cecal contents at 37°C significantly increased compared to that of rSA11. However, incubation with cecal contents heated at 95°C for 10 min prior to mixing did not affect RV infectivity (Fig. 5A). Incubation with feces of the small intestine did not increase the infectivity but rather decreased the infectivity of the rSA11 and rSA11 VP4 mutants (Fig. 5B). To examine the involvement of the microbiota in the enhancement of viral infection, cecal contents obtained from mice treated with penicillin for 2 weeks were prepared. The enhancement of infectivity of rSA11_R101A and rSA11_R101A/V143A/Y175A was completely abolished by penicillin treatment, indicating that bacteria sensitive to penicillin may be involved in viral infection and promote sialic acid-independent infection of VP4 mutant viruses in mice (Fig. 5C).

FIG 5.

Effect of cecal contents on infectivity of sialic acid-independent mutants. (A) Infectivity assays of rSA11 and mutants incubated with cecal contents. Cecal contents were collected from adult BALB/cAJcl mice. Cecal contents were frozen and thawed once and used as they are or after being incubated at 95°C for 10 min. Viruses (1 × 10⁵ FFU/mL) were incubated with cecal contents at 37°C for 1 h. Viral mixtures were diluted to 1,000 FFU/mL and added to MA104 cells in 96-well plates. After 16 h of incubation, cells were fixed, and the number of virus-infected cells was counted. The infectivity ratio was presented as the percentage relative to that in control. (B) Infectivity of rSA11 and mutants incubated with small intestinal contents; (C) infectivity of rSA11 and mutants incubated with cecal contents from penicillin-treated mice. Data are given as the mean of three replicates. Values are means ± SD. Significant differences are indicated as follows: *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, P < 0.001.

DISCUSSION

Here, we examined the effect of amino acid mutations in the glycan-binding domain of the VP4 spike protein on viral infectivity by analyzing recombinant simian RVs generated using a reverse genetics system. RVs enter cells through two steps mediated by the interaction of VP8* and VP5* with sialic acid and sialic acid-independent receptors (35). The activation of VP4 in the progeny virus is induced by trypsin cleavage to VP8* and VP5*, which leads to conformational rearrangement on the virion surface (36). VP8* forms the head domain of the VP4 spikes and has a glycan receptor-binding domain. Crystal structure analysis revealed that the glycan-binding domain of VP8* comprised a cleft-like structure formed by four β-strands. Generally, the VP8* of human RV, especially of genotype P[4] and P[8] strains, has a wide cleft that interacts with HBGA or gangliosides containing internal sialic acid, and the VP8* of animal RVs has a narrow cleft that interacts with the terminal sialic acid of glycans (37, 38). Thus, it was proposed that wide or narrow clefts of human and animal RVs are glycan usage determinants (16). Meanwhile, the VP4 genotype P[14] derived from the human RV HAL1166 strain, which binds to the A-type HBGA, has a narrow cleft of VP8* like animal RVs (31).

The glycan-binding ability of various genotypes of VP4 proteins has been determined exclusively using wild-type viruses and purified proteins. Here, we generated VP4 mutants of RVs by introducing amino acid mutations in the VP8*cleft and examined the contribution of the glycan-binding domain to virus infectivity. The results demonstrated that amino acids in the cleft of VP8*, which were found to bind to sialic acid in previous reports (14), individually contribute to binding to sialic acid and viral infectivity. However, VP4 mutants unexpectedly maintained infectivity in cell lines, indicating that sialic acids were not the sole glycan receptor for animal RV, but other unidentified glycans may be involved. The H-type I HBGA-specific antibody blocked rSA11_R101A infection, but rSA11 infection remained unaffected. This result indicated that VP4 with the R101A mutation acquired the ability to bind to H-type I HBGA. However, the VP4 triple mutant virus rSA11_R101A/V143A/Y175A did not bind to A-type and H-type I HBGAs, suggesting that V143 and Y175 are involved in H-type I binding. Additionally, the glycan to which VP8* binds can be easily changed by mutating only a few amino acids. Further experiments are required to confirm the involvement of HBGA and identify other receptor molecules.

Since the replication of VP4 mutant viruses was significantly reduced in cell lines, the increased pathogenicity of VP4 mutants in the mouse model was an unexpected result. Unlike the viral infection model in cell culture, numerous molecules derived from food digestives, microbiota, and intestinal epithelial cells are potentially associated with viral infection. We investigated the contribution of intestinal microbiota to the increased pathogenicity of VP4 mutant viruses and found that sialic acid-independent rSA11 VP4 mutant infections were enhanced by microbiota-rich cecal contents, while rSA11 remained unaffected. The infectivity enhancement of VP4 mutant viruses was abolished when penicillin-treated cecal samples were used. These results suggest the involvement of the intestinal microbiota, specifically bacteria, in the altered infectivity of VP4 mutant viruses. The intestinal microbiota is associated with RV infection in various ways. Several studies have demonstrated that probiotic bacteria protect the host from viral infection by promoting the intestinal barrier and mucosal immunity (39, 40). Microbiota inhibit virus receptors by binding to the virus, eliminating the virus in feces, and interfering with the binding of the virus to target cells. Regarding RV, Escherichia coli Nissle 1917 protects gnotobiotic pigs from RVs by regulating immune responses (33, 41, 42). A human G1P[8] RV strain and Ruminococcus gauvreauii bind to each other by HBGA-like substances produced by bacteria (43). While there is evidence that bacteria block viral infections, there are also reports that bacteria promote viral infections. In antibiotic-treated mice, intestinal RV and norovirus titers were reduced compared to those in control mice (44, 45). Here, the exact mechanisms underlying the enhanced pathogenicity of SA11 VP4 mutants could not be determined. The altered sensitivity of rSA11 and rSA11_R101A or rSA11_R101A/V143A/Y175A to intestinal contents may indicate that the intestinal microbiota is associated with different RV strains in various ways.

Moreover, RVs are associated with intestinal mucus in the gastrointestinal tract. Sialic acid is added to the terminal glycan of mature mucin proteins, and terminal sialylated glycans are important for mucus integrity and protection against bacterial proteolytic degradation (46–48). Mucin inhibits several viral infections. For example, in influenza virus infections, the tracheal mucosa has sialylated decoys and mimics sialoglycan receptors on the cell surface to enable protection against the influenza virus by being trapped in the mucus layer (49, 50). Furthermore, human immunodeficiency virus (HIV) can be inhibited by human milk, which is rich in MUC1 and MUC4, and mucins are also abundant in the gastrointestinal tract (51). Murine intestinal mucins have also been shown to be potent inhibitors of RV (52). Thus, it could be hypothesized that the sialic acid-dependent SA11 was trapped in the intestinal mucosal barrier due to interactions with sialylated glycans. Since the sialic acid is located at the terminal of glycan moiety, it is presumably easier to recognize and bind to relative to other glycan receptors. Therefore, SA11 has the advantage of being easily adsorbed to the cell surface via sialic acid. However, our experiments suggest the requirement of a different approach to facilitate infection. Moreover, the fact that only a small amount of rSA11_R101A was detected in the cecum of 3-week-old mice despite the increased infectivity when mixed with cecal contents may be attributed to the virus being trapped en route from the small intestine to the cecum. In contrast, sialic acid-independent mutants, especially the three-point mutants (rSA11_R101A/V143A/Y175A), may escape from sialic acid and bind to other glycans. It is possible that mutants bind glycans close to intestinal cells and infect them more easily than the wild type.

In conclusion, we showed that amino acid mutations in the glycan-binding domain of VP4 spike proteins alter the binding capability of the virus to glycan receptors. The in vitro results obtained were consistent with those of previous studies on recombinant VP4 proteins. We demonstrated the different outcomes of amino acid mutations in the VP4 glycan-binding domain in cell culture and animal models. This result not only confirmed the previous conclusions but also implied the complex involvement of multiple glycans in the in vivo infection steps of RVs.

MATERIALS AND METHODS

Cells.

Monkey kidney epithelial MA104, human colon cancer cell line HT29, and Caco-2 cells were cultured in Dulbecco's modified Eagle’s medium (DMEM) (Nacalai Tesque, Kyoto, Japan) supplemented with 5% (vol/vol) fetal bovine serum (FBS) (Gibco, Waltham, MA, USA), 100 U/mL penicillin (Nacalai Tesque), and 100 μg/mL streptomycin (Nacalai Tesque).

Virus preparation.

MA104 cells were cultured in Corning 100-mm by 20-mm dishes (Corning, Corning, NY, USA) to confluent monolayers and were infected with RV in DMEM supplemented with 0.5 μg/mL porcine pancreatic trypsin (Sigma-Aldrich, St. Louis, MO, USA). Cell culture media were collected 24 to 36 h postinfection; triple-layered particles (TLPs) were purified by freeze-thawing, ultracentrifuge pelleting at 109,400 × g for 90 min at 12°C, and cesium chloride gradient centrifugation at 148,900 × g for 17 h at 12°C.

Generation of RV mutants and VP4 of human RV monoreassortant viruses.

The RV rescue systems have been previously described (53). A VP4 mutant plasmid was generated by standard site-directed mutagenesis by using KOD mutagenesis kit (Toyobo, Osaka, Japan) following the manufacturer's protocol. Ten plasmids, each carrying an SA11 gene segment (0.25 μg of each) or encoding VP4 of SA11 (GenBank accession no., LC178567.1) wild type or mutated at sialic binding sites, VP4 of P[8]Odelia (GenBank accession no. LC485134.1), or P[4]DS1 (GenBank accession no. HQ650119.1) (0.25 μg of each), as well as expression plasmids pCAG-NSP2SA11, pCAG-NSP5SA11, pCAG-D1R, pCAG-D12L (0.25 μg of each), and pCAG-FAST (0.005 μg) were transfected to BHK-T7 cells in a 12-well plate (1.7 × 10⁵ cells/well) using 2 μL TransIT-LT1 (Mirus, Marietta, GA, USA) per μg of plasmid. After 48 h, MA104 cells (5 × 10⁴ cells/well) were added to transfected BHK-T7 cells and incubated at 37°C for 3 days with 0.5 μg/mL porcine pancreatic trypsin (Sigma-Aldrich). After incubation, cells were lysed by freezing/thawing and added to MA104 cells for viral replication. All plasmids were confirmed by DNA sequencing.

Rotavirus growth kinetics.

A monolayer of MA104 cells (1 × 10⁵ cells/well) or HT29 cells (3 × 10⁵ cells/well) in 24-well plates was infected with viruses at a multiplicity of infection (MOI) of 0.01 FFU/cell. After 1 h, cells were washed twice with DMEM and cultured in DMEM supplemented with 0.5 μg/mL porcine pancreatic trypsin (Sigma-Aldrich) at 37°C in 95% (vol/vol) air with 5% (vol/vol) CO₂. The cells were frozen at −80°C at 0, 12, 24, 48, and 72 h postinfection. After freeze-thawing, the viral titer was determined using a focus assay.

Generation of an HT29 SLC35A1-KO cell line.

Guide RNAs (gRNAs) were designed by inputting a part of the SLC35A1 sequence (GenBank accession no. NM_006416.5) into the CRISPR design tool CRISPOR (54). We selected a gRNA (forward [F] primer, 5′-CACCGATTATTCAAGTTATACTGCT-3′; reverse [R] primer, 5′-AAACAGCAGTATAACTTGAATAATC-3′) with the highest specificity and efficiency scores. Synthetic gRNA oligonucleotides were cloned into gRNA/Cas9-expressing pLentiCRISPRv2 (Addgene no. 52961) at BsmBI restriction sites (55). Lentiviral vectors were produced by cotransfecting 293 T cells with a cocktail of lentiCRISPRv2, the Gag-Pol packaging plasmid pCMV-delta R8.2 (Addgene no. 12263), and the vesicular stomatitis virus G protein (VSV-G) expression vector pCMV-VSV-G (Addgene no. 8454) using polyethylenimine. After 48 h, the supernatant was collected and cleaned using a 0.45-μm-pore low-protein-binding Durapore polyvinylidene difluoride membrane filter (Merck Millipore, Burlington, MA, USA). HT29 cells were seeded at a density of 5 × 10⁴ cells in 6-well plates. A lentiviral vector suspension containing 16 μg/mL of Polybrene (Sigma-Aldrich) was added to HT29 cells. For antibiotic selection, the HT29 cells were treated with 0.1 μg/mL of puromycin in DMEM supplemented with 5% FBS. After 2 weeks of puromycin selection, single-cell clones were isolated using cloning rings. The knockout of SLC35A1 in HT29 cells had no significant effect on cell viability and growth.

Immunofluorescence microscopy.

HT29 cells were grown on coverslips and fixed with 3.7% formaldehyde. Cells were washed twice with PBS and stained with 10 μg/mL FITC-labeled elderberry bark lectin (SNA/EBL) (Thermo Fisher Scientific, Waltham, MA, USA) and Hoechst stain in phosphate-buffered saline (PBS) with 0.5% bovine serum albumin (BSA). Images were acquired with a confocal microscope (Nikon A1RHD25; Nikon, Tokyo, Japan).

Analysis of binding RV to MA104 and HT29 cells by flow cytometry.

MA104 cell monolayers were incubated with 1.0 U/mL of neuraminidase from Clostridium perfringens (previously Clostridium welchii) (Sigma-Aldrich) for 30 min at 37°C. Neuraminidase-treated MA104 and HT29 cells were mixed with purified SA11 TLPs at an MOI of 10 FFU/cell for 1 h at 4°C. The cells were pelleted (400 × g, 5 min) and washed three times with ice-cold PBS. The cells were incubated with mouse anti-SA11 VP4 antibody for 30 min at 4°C. Anti-SA11 VP4 antibody was generated as previously described and observed to bind to R441 in SA11 VP4 (56, 57). After washing, the cells were incubated with anti-mouse IgG antibody conjugated to Alexa Fluor 488. After 30 min of incubation in the dark at 4°C, the cells were washed three with PBS before 300:1 of propidium iodide (PI) was added. Cells were analyzed using an Attune NxT acoustic focusing cytometer (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Inhibition assays on RV infection.

Inhibition assays were performed on MA104, HT29, and Caco-2 cells using neuraminidase and HBGA antibodies. Monoclonal antibodies against blood group A or H-type I antigen were purchased from Covance. MA104 cells (2 × 10⁵ cells/mL), HT29 cells (6 × 10⁵ cells/mL), or Caco-2 cell monolayers (4 × 10⁵ cells/mL) were grown in 96-well plates. Neuraminidase treatment was performed at 37°C for 30 min or increasing concentrations of antibodies were allowed to bind to the cells at 4°C for 1 h before 100 FFU of the virus was added per well. The inoculum was removed, and the cells were washed with DMEM and incubated for 16 h at 37°C. Virus titers were determined by staining the cells with a rabbit anti-RV NSP4 antibody and Alexa Fluor 488-labeled goat anti-rabbit secondary antibody (Biotium, San Francisco, CA, USA). Rabbit anti-NSP4 antiserum was raised against a synthetic SA11 NSP4 peptide spanning amino acid residues 158 to 171 (Eurofins Genomics, Tokyo, Japan). The infectivity ratio is presented as the percentage of infectivity relative to that of the control.

Mice.

In total, 24 pregnant female BALB/cAJcl mice and 16 3-week-old male BALB/cAJcl mice were purchased from Japan SLC Co., Ltd. Mice were maintained under standard housing conditions with a 12-h dark and light cycle.

Animal experiments.

Neonatal mice were divided into groups of five or three. On day 3, the pups were inoculated by oral gavage with 10 μL RV (1 × 10⁵ FFU/head or 1 × 10⁶ FFU/head) or PBS as a control. rSA11_R101A was inactivated by UV irradiation. From days 3 to 9, body weight was measured, and the diarrhea score was analyzed as previously described (58). To investigate the viral replication in mouse pups, neonatal mice were divided into six groups. Three-day-old mice were inoculated by oral gavage with 10 μL RVs (1 × 10⁵ FFU/head). After 12 h and 24 h, the intestinal tract was removed and frozen, and thawed to extract the viral RNA. Male BALB/cAJcl 3-week-old mice were inoculated with 100 μL of RVs (1 × 10⁵ FFU/head). After 2 days, the mice were euthanized, and the small intestine and cecum were removed. The small intestine and cecum were washed with PBS to remove intestinal contents, frozen, and thawed to extract the viral RNA.

Real-time quantitative reverse transcription-PCR.

Real-time quantitative reverse transcription PCR (qRT-PCR) was performed as follows. Total RNA was extracted from the small intestine and cecum tissues using TRIzol reagent (Life Technologies, Carlsbad, CA, USA), according to the manufacturer’s instructions. Reverse transcription and amplification of the SA11 VP1 genome segment were performed using ReverTra ace (TOYOBO, Osaka, Japan) with the primers F (5′-AGGCAAACCATTGGAGGCAGAC-3′) and R (5′-CCAACCAGAACATGACTGCATT-3′). For RV SA11 VP1 detection, the cDNA was analyzed using TaqPath qPCR master mix (Thermo Fisher) with the VP1 probe 5′–6-carboxyfluorescein (FAM)–TCCAACAGCGGAGGAATATACGGAC–6-carboxytetramethylrhodamine (TAMRA)–3′ in a reverse transcription-PCR (RT-PCR) cycler (TaKaRa PCR thermal cycler Dice real-time system III; TaKaRa, Shiga, Japan).

Infectivity of RVs incubated with intestinal contents.

Intestinal contents from the small intestine and cecum were collected with PBS from adult BALB/cAJcl mice or mice treated with penicillin and were immediately frozen. The cecal contents were incubated for inactivation at 95°C for 10 min. Confluent MA104 monolayers were grown in 96-well plates. RVs were diluted to 1 × 10⁵ FFU/mL and incubated with intestinal contents for 1 h at 37°C. A diluted viral mixture (100 FFU) was added to each well and incubated for 16 h at 37°C. Virus titers in formalin-fixed cell monolayers were determined by staining the cells with rabbit anti-RV NSP4 antibody and Alexa Fluor 488-labeled goat anti-rabbit secondary antibody (Biotium). Virus infectivity was expressed as the percentage of FFU in control wells incubated with the virus-PBS mixture.

Statistical analyses.

Statistical analyses were performed using Prism software version 9 (GraphPad Software). Student's t test and analysis of variance (ANOVA) were used to determine statistical significance as appropriate.