Identification of a novel cellular target and a co-factor for norovirus infection – B cells & commensal bacteria

Abstract

Human noroviruses are a leading cause of gastroenteritis worldwide but research on these important enteric pathogens has long been restricted by their uncultivability. Extensive efforts to infect intestinal epithelial cells with murine and human noroviruses in vitro have been thus far unsuccessful while murine noroviruses efficiently and lytically infect innate immune cells including macrophages and dendritic cells. We have recently discovered that murine and human noroviruses infect B cells in vitro. The nature of B cell infection was distinct from innate immune cell infection in that mature B cells were infected noncytopathically in contrast to the lytic infection of macrophages and dendritic cells. Human norovirus infection of B cells was facilitated by commensal bacteria expressing an appropriate histo-blood group antigen. Importantly, we used the mouse model of norovirus infection to confirm that Peyer's patch B cells are infected, and that commensal bacteria stimulate infection, in vivo.

Keywords:

• enteric virus
• histo-blood group antigen
• intestinal microbiota
• norovirus

Introduction

Human noroviruses (HuNoVs) are a leading cause of gastroenteritis globally. In the United States alone, they annually cause 19–21 million symptomatic infections resulting in up to 71,000 hospitalizations and 800 deaths.¹ In fact, they have become the leading cause of severe childhood diarrhea since the introduction of a successful rotavirus vaccination program^2,3 and they are the predominant cause of foodborne disease outbreaks.⁴ In developing countries, HuNoVs are likely even more devastating, an idea underscored by an estimate that they cause over a million clinic visits and 200,000 fatalities in young children in impoverished parts of the globe.⁵

HuNoVs are nonenveloped positive-sense RNA viruses in the Caliciviridae family. Their genomes of ˜7.5 kb are divided into 3 open reading frames (ORFs)⁶: ORF1 encodes a large 200-kDa polyprotein that is cleaved by the viral protease into 6 mature nonstructural proteins, ORF2 encodes the major capsid protein called VP1, and ORF3 encodes a minor structural protein that localizes on the interior of the mature virion.⁷ The VP1 protein can self-assemble into virus-like particles (VLPs) that are structurally similar to native virions. Clinical trials testing the efficacy of HuNoV VLPs as vaccine candidates administered intranasally or intramuscularly have demonstrated modest short-term (up to 1 month) protection from developing severe disease,^8,9 results that are encouraging but do not address the critical questions of immunity duration or cross-reactive protective immunity. Development of potentially more effective live attenuated viral vaccines has long been hindered by an inability to culture HuNoVs. Our recent findings that a HuNoV can infect B cells in culture and that this infection is facilitated by commensal bacteria¹⁰ represent a major step forward although challenges remain. In this addendum we will provide a more detailed description of the implications of this new system.

The most logical target of an enteric pathogen is the intestinal epithelial cell (IEC) lining the gut wall. Since the discovery of the first HuNoV over 40 y ago by Kapikian et al.,¹¹ extensive efforts have been undertaken to culture these viruses in epithelial cells including IECs but thus far no published and reproducible data support infection of this cell type in vitro.¹² Murine NoVs (MuNoVs) efficiently and lytically replicate in macrophages and dendritic cells¹³ but efforts to infect these cell types with HuNoVs in culture have been unsuccessful.¹⁴ Several observations by our lab and other research groups led us to speculate that B cells represent another NoV target. First, nonstructural protein indicative of intracellular viral replication was observed in the B cell zones of STAT1^−/− mice infected with a MuNoV.¹⁵ Second, viral genome was detected in Peyer's patch lymphocytes of MuNoV-infected IL-10^−/− mice.¹⁶ These results suggest that NoVs can infect B cells at least under certain conditions of immunocompromise. Providing support that NoVs also infect B cells in immunocompetent hosts, wild-type mice had significantly higher MuNoV titers than B cell-deficient mice at early times post-infection.¹⁷ Finally, viral capsid-positive duodenal B cells were detected in chimpanzees infected with a HuNoV,¹⁸ suggesting that B cells are a common target of NoVs.

Murine Noroviruses Infect B cells in vitro

Because MuNoVs are cultivable in macrophages and dendritic cells, we first confirmed that they infect B cells in culture. Numerous MuNoV strains have been discovered since 2003 which differ in pathogenicity, persistence establishment, and immune response.¹⁹ Two genetically related but pathogenically distinct MuNoV strains called MNV-1 and MNV-3 were selected for our studies since differences in cell tropism could provide clues regarding the differences in overall infection outcome. Specifically, MNV-1 is more virulent than MNV-3 in wild-type and interferon-deficient mice²⁰; MNV-1 is cleared acutely whereas MNV-3 establishes a persistent infection in the colon^21,22; and MNV-3 elicits a more robust protective immune response than MNV-1.^17,23 Both MNV-1 and MNV-3 replicated efficiently and lytically in the RAW264.7 murine macrophage cell line but neither strain infected the CMT-93 intestinal epithelial cell line, as expected. Both virus strains also productively infected a mature B cell line called M12²⁴ and an immature B cell line called WEHI-231²⁵ although peak viral protein synthesis and progeny virion production were delayed ˜1 dpi compared to replication in RAW264.7 cells. There were several unexpected findings pertaining to MuNoV infection of cultured B cells that are worth further consideration.

While infection of RAW264.7 and WEHI-231 cells was cytopathic, we observed no CPE in M12 cultures infected with either MNV-1 or MNV-3. WEHI-231 cells displayed a virus strain-specific phenotype with regard to CPE, with MNV-1 causing a nearly complete loss in cell viability by 2 dpi but MNV-3 killing only 60% of the cells by this time point; in the MNV-3-infected cultures, the surviving cells actually began to divide at 3–4 dpi. Overall, M12 cells developed minimal to no CPE upon MuNoV infection whereas WEHI-231 cells displayed a greater sensitivity to MuNoV-induced cell death that was virus strain-specific. Based on the paradigm that nonenveloped viruses must lyse their target cells in order for progeny virions to be released extracellularly, the apparent noncytopathic nature of M12 cell infection was unexpected. However, recent findings pertaining to 2 other nonenveloped enteric viruses, hepatitis A virus (HAV) and hepatitis E virus (HEV), demonstrate a unique mode of viral egress in which classically nonenveloped viruses can be released from cells wrapped in cellular envelopes.^26,27 It is interesting to speculate that NoVs utilize a similar strategy to exit infected B cells in the absence of cell lysis. The difference in cytopathicity between M12 cells and other cell lines (e.g., RAW264.7, WEHI-231) may reflect differential cellular regulation of cytopathic versus noncytopathic NoV infection and could thus represent a useful tool in future studies probing putative alternate forms of viral egress. Likewise, the virus strain-specific difference in WEHI-231 cytopathicity could provide clues regarding viral regulation of egress mechanisms.

There was a direct correlation between MuNoV infectivity in the various cell lines and cytopathicity, with only 5–15% of M12 cells productively infected compared to 80–90% of RAW264.7 and WEHI-231 cells. The striking difference in infectivity between M12 and WEHI-231 cells was also surprising considering overall levels and kinetics of progeny virion production were similar between the 2 B cell lines. We have considered the possibility that progeny virus produced by M12 cells is more stable than virus produced by WEHI-231 cells but stability assays have not revealed any differences (unpublished data, Karst lab). Another possible explanation is that each infected cell in the M12 cultures produces significantly more virus compared to individual WEHI-231 cells but we have not yet probed this in further detail.

Finally, M12 cells were persistently infected with MNV-1 and MNV-3. This persistent infection was productive since constant levels of infectious virus were detected in the culture supernatants through 25 cell passages. Moreover, the low infectivity remained stable throughout the M12 passaging study, suggesting that host or viral factors maintain a specific level of infection in the culture through undefined mechanisms. WEHI-231 cells that survived the initial cytopathic effect during MNV-3 infection also became persistently infected, demonstrating that persistence establishment by NoVs in cultured B cells is not unique to the M12 cell line. Because MNV-3 establishes a persistent colonic infection whereas MNV-1 is cleared acutely,^21,22 it is possible that this differential outcome of WEHI-231 infection mimics in vivo MNV-1 and MNV-3 infection outcomes of an intestinal B cell subset representing a persistent NoV reservoir.

Murine Noroviruses Infect B cells in vivo

Because MuNoVs can be easily studied in their natural host, we were able to confirm the in vivo relevance of B cell infection by NoVs. While previous studies suggested that B cells were targeted by NoVs in vivo,^15-18 we performed experiments to provide unequivocal proof. MNV-1 and MNV-3 titers were reduced in the distal ileums at 0.5 and 1 dpi, and in the mesenteric lymph nodes (MLNs) that drain the small intestine at 1 dpi, but not in the colons of B cell-deficient µMT mice compared to wild-type B6 mice. Studies with light-sensitive neutral red-labeled MNV-1 confirmed that in vivo viral replication was reduced in the absence of B cells. The reduced titers in the small, but not large, intestine suggested that Peyer's patch B cells were specifically infected since Peyer's patches are localized along the small intestine. Confirming this, viral genomes were detected in CD19⁺ B cells purified from Peyer's patches of MNV-1- and MNV-3-infected B6 mice. It should be noted that bulk Peyer's patch cells contained significantly more viral genome than purified B cells, consistent with additional cellular targets such as macrophages and dendritic cells. Further proving that B cells are bona fide NoV targets in vivo, viral nonstructural protein was detected in CD19⁺ and B220⁺ Peyer's patch B cells of MNV-1-infected STAT1^−/− mice. Collective data thus suggest that B cells in Peyer's patches and MLNs are infected by NoVs during in vivo infections.

A GII.4-Sydney Human norovirus infects B cells in vitro

Based on our findings that MuNoVs infect B cells in vitro and in vivo, we probed whether HuNoVs share this cell tropism. HuNoVs are genetically diverse but a majority of infections globally are caused by strains segregating in a single genetic cluster referred to as the genogroup II, genotype 4 (GII.4) cluster. Every 2–4 y a new pandemic GII.4 strain emerges and rapidly spreads across the globe and evidence suggests that these pandemic strains arise due to immune-driven evolution once herd immunity controls a previous pandemic strain.²⁸ The most recent GII.4 pandemic strain is the GII.4-Sydney strain which emerged in 2012.²⁹ When a GII.4-Sydney HuNoV-positive stool sample was applied to the human B cell line BJAB, we detected significant albeit modest increases in viral genome copy number. As expected, viral genome replication was completely ablated by UV treatment and a polyclonal α-VP1 antibody fully neutralized infectivity. Further confirming viral replication in these cells, we detected nonstructural and structural protein. Most importantly, lysates from the infected cells supported infection of naïve BJAB cells, demonstrating productive infection of progeny virions.

Similar to MuNoV infection of M12 cells, we failed to detect CPE in HuNoV-infected BJAB cultures microscopically or using trypan blue exclusion (unpublished data, Karst lab). Although both MuNoV and HuNoV infection of certain B cell lines is thus apparently noncytopathic, GII.4-Sydney HuNoV did not establish persistence in BJAB cells as MuNoVs did in M12 cells. Considering that MuNoV persistence establishment in cultured B cells was cell type- and virus strain-dependent, it is unclear whether this reflects an inherent difference in murine vs. human NoV B cell infections or a difference in the B cell lines or specific virus strains used in the 2 systems. Regardless, both MuNoVs and HuNoVs can establish prolonged infections associated with extended periods of fecal shedding even during asymptomatic infections.¹⁹ Thus, there must be in vivo reservoir of prolonged or persistent infection which may in fact be intestinal B cells and may be associated with the ability of the viruses to noncytopathically infect this cell type.

Because NoVs enter the host in the gut lumen, they must possess a mechanism to traverse the intestinal epithelium to access their target immune cells. Consistent with this, HuNoVs can be internalized by enterocytes in culture^30-32 and MuNoVs can be transcytosed across a confluent epithelial monolayer to target underlying permissive macrophages³³; in all of these studies, no evidence of viral replication in the enterocytes themselves was noted. Indeed, we confirmed that the GII.4-Sydney HuNoV could cross a confluent intestinal epithelial monolayer to infect underlying B cells in the basal chamber of a transwell system. Many other types of viruses can be transcytosed across a polarized epithelium often in the absence of viral replication in the epithelial cells (e.g.,^34-39) so this appears to be a common strategy for viruses to breach barriers along mucosal surfaces. Important areas of future research will be to determine whether the enterocytes transcytosing NoVs recognize pathogen-associated molecular patterns (PAMPs) and contribute to the antiviral immune response; and whether antiviral immune responses that prevent transcytosis are important in controlling NoV infection. Related to the latter idea, a recent report demonstrated that the type III interferon (IFN), or IFN-λ, response is critical to control persistent enteric infection by MuNoVs in a manner requiring IFN-λ receptor expression on nonhematopoietic cells⁴⁰; thus, one could speculate that IFN-λ-mediated signaling suppresses transcytosis of NoVs and this prevents persistence establishment.

Commensal Bacteria Facilitate Norovirus Infections

Filtration of the GII.4-Sydney HuNoV-positive stool sample through a 0.2-μm membrane significantly reduced viral genome replication in direct B cell infection and in the transwell system, revealing a filterable co-factor for HuNoV infection of B cells. Because we used unprocessed stool as a source of virus, we reasoned that the most likely candidate for a filterable co-factor was commensal bacteria. In selecting a specific commensal bacteria to test for stimulatory activity, we were guided by the knowledge that HuNoVs bind HBGAs in a virus strain-specific manner^30,41 and that certain bacteria express glycans indistinguishable from human HBGAs.⁴² In fact, Miura et al. recently revealed that HBGA-expressing commensal bacteria such as Enterobacter cloacae can bind HuNoV VLPs.⁴³ Thus, we hypothesized that HBGA-expressing commensal bacteria in the unfiltered virus-positive stool sample facilitated HuNoV infection of B cells. Indeed, H antigen-expressing E. cloacae stimulated viral infection of B cells in a dose-dependent manner whereas H antigen-negative Escherichia coli and free lipopolysaccharide (LPS) failed to stimulate infection. The bacteria used in these experiments were heat-killed prior to incubation with the filtered stool inoculum so bacterial viability is not required for its stimulatory activity. Providing further evidence that the HBGA specifically stimulates infection, synthetic H antigen also rescued infectivity of filtered virus inoculum. Although we do not yet know the precise mechanism by which a HBGA (either free or bacterial-bound) stimulates HuNoV infection of B cells, we determined that it facilitates viral binding to the surface of B cells. Based on a previous study showing that LPS stimulates poliovirus binding to its receptor on permissive cells,⁴⁴ we speculate that the HBGA plays a similar role in bridging NoV attachment to a yet-to-be-identified entry receptor expressed on B cells.

A variety of other enteric viruses exploit commensal bacteria to stimulate their in vivo infectivity.^45-47 Based on our results with a HuNoV, we were interested to determine the biological significance of commensal bacterial interactions during a natural NoV infection. Indeed, antibiotic depletion of the intestinal microbiota prior to MNV-1 and MNV-3 infections resulted in significant reductions in acute virus titers. This reflected decreased viral replication in contrast to increased clearance, as indicated by infections with light sensitive neutral red-labeled virus. Because virus titers were reduced but not eliminated completely, we presume that bacteria stimulate MuNoV infection but are not absolutely required. Since the publication of our original article, Kernbauer et al. reported similar reductions in MuNoV titers in germ-free mice⁴⁸ and Baldridge et al. reported that commensal bacteria facilitate persistent MuNoV infections.⁴⁹ Thus, commensal bacteria stimulate HuNoV infection of cultured B cells as well as acute and persistent MuNoV infection in vivo. These data strongly argue that NoVs have evolved common strategies to use bacterial ligands to enhance their infectivity although the strategies used by MuNoVs and HuNoVs are likely not identical. For example, MuNoVs can efficiently infect macrophages, dendritic cells, and B cells in vitro in the absence of bacteria. Moreover, mice do not express HBGAs so MuNoVs must use distinct bacterial ligands. Regardless, the mouse model will be invaluable in dissecting the general in vivo relevance of these interactions to overall NoV infection.

A Working Model of Norovirus Infection

Based on available data regarding NoV enteric infections, we have developed a model whereby NoVs bind specific glycans expressed by commensal bacteria in the gut lumen; the virus-bacteria complex is transcytosed across the polarized intestinal epithelium; and the glycan stimulates NoV infection of underlying intestinal immune cells.⁵⁰ There are many open questions regarding this model.

1. Pertaining to glycan usage, because the HBGAs used by HuNoVs are also expressed on host enterocytes and secreted into the gut lumen it is possible that the virus uses both host-derived and bacterial-expressed glycans to stimulate immune cell infection. It will be interesting in future studies to determine the contribution of host versus bacterial glycans. Moreover, a possible implication of viral binding to glycans produced by the host (irrespective of whether they are host-derived or bacterial-derived) is that the virion masks its capsid antigens with self antigens and thus this process facilitates viral evasion of immune recognition.

2. Pertaining to general exploitation of commensal bacteria, it is probable that bacterial stimulation of NoV infection is multifactorial. For example, LPS binding by poliovirus enhances the attachment of the virus to its receptor on permissive cells and it also stabilizes the virus particles which may enhance viral transmission.⁴⁴ There is also evidence in other model systems that commensal bacteria skew the immune response to the virus^46,47 so it will be critical to test whether this is true for NoVs.

3. It will be important in future studies to determine the impact of intestinal B cell infection on the overall immune response to NoVs. For example, HuNoVs do not elicit long-lasting immunity^6,51,52; is it possible that viral infection of Peyer's patch B cells as well as infection of intestinal macrophages and dendritic cells is directly related to the suboptimal memory immune response to NoVs?

4. Our studies investigated only a single HuNoV strain, the current pandemic GII.4-Sydney strain. It will be critical moving forward to determine whether other strains similarly target B cells and whether they exploit commensal bacteria for this infection.

Utility of a Human norovirus Cell Culture System

Development of an in vitro culture system for HuNoVs should facilitate an investigation of the basic mechanisms of viral replication, although the modest level of replication in the current system represents a remaining challenge. Continuing efforts to enhance the robustness of this system are critical and ongoing. The availability of a robust HuNoV cultivation system will also enable the development of methods for measuring immune correlates of protection (e.g., antibody neutralization assays), development of live attenuated viral vaccine candidates, and testing of putative antiviral compounds in vitro. In fact, our finding that a polyclonal α-VP1 antibody could fully neutralize viral infectivity provides proof-of-principle that this new cell culture system can be used to assess true neutralizing activity of α-HuNoV antibodies. Overall, our recent findings expand the cellular targets of NoVs to now include innate and adaptive immune cells. Moreover, our work together with studies by Kernbauer et al.⁴⁸ and Baldridge et al.⁴⁹ illuminate a previously unappreciated role for commensal bacteria as co-factors for NoV infection.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.