Enterobacter cloacae inhibits human norovirus infectivity in gnotobiotic pigs

Human noroviruses (HuNoVs) are the leading cause of epidemic gastroenteritis worldwide. Study of HuNoV biology has been hampered by the lack of an efficient cell culture system. Recently, enteric commensal bacteria Enterobacter cloacae has been recognized as a helper in HuNoV infection of B cells in vitro. To test the influences of E. cloacae on HuNoV infectivity and to determine whether HuNoV infects B cells in vivo, we colonized gnotobiotic pigs with E. cloacae and inoculated pigs with 2.74 × 104 genome copies of HuNoV. Compared to control pigs, reduced HuNoV shedding was observed in E. cloacae colonized pigs, characterized by significantly shorter duration of shedding in post-inoculation day 10 subgroup and lower cumulative shedding and peak shedding in individual pigs. Colonization of E. cloacae also reduced HuNoV titers in intestinal tissues and in blood. In both control and E. cloacae colonized pigs, HuNoV infection of enterocytes was confirmed, however infection of B cells was not observed in ileum, and the entire lamina propria in sections of duodenum, jejunum, and ileum were HuNoV-negative. In summary, E. cloacae inhibited HuNoV infectivity, and B cells were not a target cell type for HuNoV in gnotobiotic pigs, with or without E. cloacae colonization.

target cell type of HuNoV 20 . Therefore, in vivo evidence is essential to test the stimulatory role of E. cloacae and to confirm that B cells are a natural target for robust HuNoV infection and replication.
The neonatal Gn pigs recapitulate the hallmark features of the gastrointestinal tract in young children and have been widely used for enteric virus infection 21,22 . HuNoV pathogenesis studies and vaccine evaluations in Gn pigs have high translational relevance to those of humans 13,14,23 . Compared to chimpanzees (no longer available for biomedical research) and immunodeficient mice, Gn pigs are currently preferable for HuNoV infection study in many ways, such as the ability to become infected via oral route and resulting in diarrhea and fecal virus shedding 24 . In addition, the germ-free environment in the Gn system has enabled the studies of interaction between virus and specific bacterial strains [25][26][27] , as well as human microbiota 28,29 . In this study, via E. cloacae colonization in the well-established neonatal Gn pig model of HuNoV infection and diarrhea, we aimed to (i) elucidate the effects of E. cloacae on HuNoV infectivity in vivo, (ii) determine whether HuNoV infects B cells in vivo, and (iii) explore the mechanism of altered HuNoV infectivity in the presence of E. cloacae.

E. cloacae reduced HuNoV shedding but not diarrhea.
To evaluate the effects of E. cloacae on HuNoV infection and diseases in vivo, a group of Gn pigs were inoculated with three doses of E. cloacae on 3, 4, and 5 days of age. Together with another group of control pigs, all were inoculated with 2.74 × 10 4 genome copies of HuNoV GII.4/2006b at 6 days of age, which was post inoculation day 0 (PID0), then euthanized on PID3, PID7, or PID10. To confirm the colonization of E. cloacae in these Gn pigs, fecal E. cloacae shedding was determined daily from PID0 to euthanasia day. E. cloacae was detected in all treated pigs ( Fig. 1), whereas the control pigs remained sterile throughout this study (data not shown).
Fecal HuNoV shedding was monitored daily in both control and E. cloacae colonized Gn pigs after viral inoculation (Fig. 2a). In E. cloacae colonized pigs, virus shedding was clearly lower than that of control, as the shedding titers for all pigs on each day were below 5000 genome copies per gram of feces. The incidence of virus shedding showed that compared to control pigs, E. cloacae colonized pigs had a significantly shorter mean duration in PID10 subgroup (6.7 versus 5.5 days) ( Table 1). To further characterize the shedding titers, cumulative and peak shedding for individual pigs were calculated. The cumulative virus shedding in E. cloacae colonized pigs was significantly lower on PID7 and PID10 (Fig. 1b), and there was also a trend for lower peak shedding titers in E. cloacae colonized pigs (Fig. 1c). No significant difference in the incidence of diarrhea was observed among groups at any time points (Table 1), indicating E. cloacae colonization did not affect the incidence of diarrhea in Gn pigs after HuNoV infection. These data demonstrated that E. cloacae colonization inhibited HuNoV shedding in Gn pigs.

E. cloacae reduced HuNoV titers in intestinal tissues and blood. Previous studies showed that
HuNoV (GII.4) antigen was observed in duodenum and jejunum of Gn pigs [13][14][15] . In this study, analysis of the tissue samples confirmed the existence of HuNoV genomes in duodenum and jejunum, and viral genomes were also detectable in ileum for both control and E. cloacae colonized pigs (Fig. 3a,b). Compared to control pigs, on PID3, virus titers were significantly lower in duodenum and ileum of E. cloacae colonized pigs (Fig. 3a). On PID10, virus titer was significantly lower in ileum of E. cloacae colonized pigs (Fig. 3b).
Previously, HuNoV was detected in blood in humans with gastroenteritis 30,31 , as well as in Gn calves and pigs after HuNoV challenge 12,13 . In this study, viral genomes were present in plasma in 4 of 6 control pigs (66.7%) and 5 of 8 (62.5%) E. cloacae colonized pigs, and virus titer on PID3 was significantly lower in E. cloacae colonized pigs (Fig. 3c). In addition, viral genomes were also detectable in whole blood cells in 2 of 6 control pigs (33.3%) and in 1 of 8 E. cloacae colonized pigs (12.5%) (Fig. 3d). Taken together, the reduced HuNoV titers in intestinal tissues and in blood indicated the inhibitive role of E. cloacae for HuNoV infection in Gn pigs. HuNoV antigen was observed in enterocytes but not in B cells. To confirm HuNoV infection of enterocytes in Gn pigs and to identify virus-infected cells in E. cloacae colonized pigs, immunohistochemistry (IHC) was performed to detect the viral capsid protein VP1 on sections of small intestine from both groups. As expected, HuNoV antigen was observed in enterocytes of duodenum and jejunum from control pigs. In E. cloacae colonized pigs, both duodenal and jejunal enterocytes were also the only positive cells, whereas cells of the lamina propria were negative for both groups (Fig. 4).
To determine whether HuNoV infects B cells in the presence of E. cloacae in Gn pigs, we isolated total mononuclear cells (MNC) from ileum and spleen, then performed qRT-PCR to detect viral genomes. 1 of 6 control  Table 1. Data are presented as mean with individual animal data points (b,c). Statistical significance was determined by Student's t-test. NS, not significant, *P < 0.05.

Table 1. Incidence of clinical signs and fecal virus shedding in Gn pigs after HuNoV GII.4 2006b infection.
All pigs were inoculated with a HuNoV GII.4 2006b variant 092895 at 6 days of age. Daily rectal swabs were collected after inoculation to assess diarrhea and HuNoV shedding by Taqman qRT-PCR. a Pigs with daily fecal scores of ≥ 2 were considered diarrheic. Fecal consistency was scored as follows: 0, solid; 1, semisolid; 2, pasty; 3, semiliquid; and 4, liquid. b Fecal consistency score or virus shedding calculation included all the pigs in each group from PID1 to PID3, PID7, or PID10. If no diarrhea or virus shedding was observed, the days to onset were recorded as 1 day longer than each timepoint (4, 8, or 11) and the duration days were recorded as 0 for statistical analysis. * Fisher's exact test or ** Student's t-test was used for statistical analysis. Groups with significant differences (P < 0.05) were indicated with letters A and B.
pigs had detectable viral genomes in ileal MNC (Fig. 5a), and 3 of 6 control pigs had detectable viral genomes in splenic MNC (Fig. 5b), although the titers were as low as 20 genomic copies in 10 7 MNC. However, no viral genomes were detected in ileal or splenic MNC in E. cloacae colonized pigs, presumably resulting from the lower HuNoV infection in these pigs. Furthermore, IHC for both viral antigen and B cells was performed in ileum, because ileum is populated with B cells in conventional and Gn pigs 32 , and viral genomes were detected in ileum (Fig. 3a). B cells were observed in the lamina propria, as probed by primary antibody targeting cellular marker CD79, while HuNoV antigen was observed only in enterocytes. Thus, there was no signal co-localization in sections from control and E. cloacae colonized pigs, even though B cells could be located in close proximity to HuNoV positive cells (Fig. 5c). In all, these data suggest that B cells are not a HuNoV target in Gn pigs, with or without E. cloacae colonization.
E. cloacae promoted gut immunity. Probiotic lactobacilli colonization enhanced gut immunity in Gn pigs by promoting intestinal T cell and B cell responses, as well as the secretion of intestinal immunoglobulins 25,26 . Lactobacilli protected against rotavirus diarrhea and also functioned as adjuvants for rotavirus vaccine by promoting protective immune responses 27,33,34 . We hypothesized that E. cloacae colonization might promote gut immunity in Gn pigs as well, which in return would inhibit HuNoV infectivity. To determine the role of E. cloacae in intestinal immune system development, we compared the sizes of ileal Peyer's patches (IPP) for both groups.
In E. cloacae colonized pigs, IPP were significantly larger and more developed than those of control pigs on 6 days and 13 days post inoculation of E. cloacae (Fig. 6a). The IPP in E. cloacae colonized pigs were characterized by significantly greater width from muscularis mucosae to muscularis associated with gut associated lymphoid tissue (GALT), width of GALT, height and width of follicles (Fig. 6b).

Discussion
The lack of a robust cell culture and animal model for HuNoVs propagation has long eluded study of HuNoV biology and development of antiviral therapies. Among the in vitro models, HuNoVs reverse genetics systems containing subgenomic or genomic RNA have been established 35,36 , but the virus-like particle or virion production was inefficient. Reports of HuNoVs infection and replication observed in 3D intestinal epithelial cells were not reproducible 37,38 . Although HuNoV infection has been observed in chimpanzees 11 , the species is no longer available for the entire biomedical research due to ethical concerns. The BALB/c Rag/ɣc −/− mouse was recognized as a HuNoV infection model via intraperitoneal inoculation 16 , but it is not suitable for HuNoV propagation due to the low robustness of replication and the lack of virus shedding. Given the limitations of current models, clinical fecal  18,19 . Hence, we expanded our Gn pig model with E. cloacae colonization, which should enhance HuNoV infection and replication. Our original objective was to develop a pig model for effective HuNoV propagation. Surprisingly, E. cloacae inhibited HuNoV infectivity in Gn pigs instead. First, fecal virus shedding has been used as direct evidence to characterize the status of HuNoV infection in vivo. Our data showed that the incidence of shedding and the shedding amount in E. cloacae colonized pigs were less than those of control pigs, including lower cumulative shedding and peak shedding in individual pigs (Table 1 and Fig. 2). Second, the small intestine has been shown to be the primary HuNoV infection site in Gn pigs [13][14][15] . We also confirmed the existence of viral genomes in all sections of small intestine in both groups, and the titers were significantly lower in E. cloacae colonized pigs. In addition, lower viremia was observed in E. cloacae colonized pigs (Fig. 3). As HuNoV genomes were present in whole blood cells, it is likely that those were virions captured by phagocytes and translocated through blood circulation, since ileal and splenic MNC had detectable viral genomes as well. Third, E. cloacae was expected to facilitate HuNoV infection of B cells based on the in vitro model, but no HuNoV-positive B cells were observed in ileum sections even though the adjacent enterocytes were HuNoV positive. In addition, the entire lamina propria in sections of duodenum, jejunum, and ileum were HuNoV-negative, in both control and E. cloacae colonized pigs.
Enteric bacteria can bind to HuNoV via surface HBGA 17 . It is likely that HBGA-expressing E. cloacae serves as a blockade between HuNoV and target cells in vivo instead of facilitating virion attachment, and thus leads to lower rates of infection. Both inhibition and enhancement of HuNoV P particles attachment via such binding have been observed in vitro for other probiotic bacteria, such as Lactobacillus casei BL23 and Escherichia coli Nissle 1917 39 . Our data suggested that inhibition of attachment by E. cloacae was the scenario in pigs in vivo, resulting in decreased HuNoV infectivity. Another reason for the lower HuNoV infectivity might be due to the enhanced development of gut immunity by E. cloacae colonization. We observed that colonization of E. cloacae in Gn pigs stimulated the development of IPP. Further and longer term studies will be required to determine the magnitude of immune responses to HuNoV infection between the two groups.
The binding between E. cloacae and HuNoV may also play a role in virus retention in vivo. We observed that virus titer in duodenum decreased from PID3 to PID10 in control pigs, but it appeared to be unchanged or even increased in E. cloacae colonized pigs (Fig. 3a,b). E. cloacae is naturally resistant to broad-spectrum antibiotics 40 , thus it could be dominant in gut microbiota in immunocompromised patients, who also require antibiotic therapy to manage microbial infections in many cases 41 . Increased E. cloacae might in return contribute to persistence of HuNoV infection in immunocompromised hosts by virus retention.
In summary, colonization of E. cloacae in neonatal Gn pigs inhibited HuNoV infectivity, including reduced virus shedding, lower viral genome titers in intestinal tissues and in blood. HuNoV infection of B cells was not observed in duodenum, jejunum, or ileum in either control or E. cloacae colonized pigs. To our knowledge, this is the first in vivo study to evaluate the effects of E. cloacae on HuNoV infectivity, and our study paves the way for future studies of the interaction between HuNoV and enteric bacteria in vivo.  locations including all pigs in each group were measured to characterize the IPP size, including width from muscularis mucosae to muscularis (mm to m) associated with gut associated lymphoid tissue (GALT), width of GALT, height and width of follicles. Statistical significance was determined by Mann-Whitney test. *P < 0.05; **P < 0.01; ***P < 0.001. Error bars denote SEM.

Virus and bacterium.
Scientific RepoRts | 6:25017 | DOI: 10.1038/srep25017 approved by the institutional review boards of the Cincinnati Children's Hospital Medical Center (IRB number: 2008-1131), and informed consent was obtained from parents or child. The stool was processed as inoculum and stored in our laboratory for studies of HuNoV infection in Gn pigs 14 . Enterobacter cloacae was purchased from ATCC (ATCC 13047) and grown in nutrient broth overnight at 30 °C with shaking at 250 rpm. Overnight cultures containing 15% glycerol were stored in − 80 °C freezer for colonization of Gn pigs. The final concentration of E. cloacae was measured in serial dilutions by enumeration of colony forming unit grown on nutrient broth agar plates.

E. cloacae colonization and HuNoV inoculation of Gn pigs. Near-term Yorkshire cross breed pigs
were derived by hysterectomy and maintained in Gn pig isolators as described previously 42 . A subset of pigs was orally inoculated with 10 4 CFU of E. cloacae daily on 3, 4, and 5 days of age to initiate colonization, which was monitored via testing fecal shedding. Control pigs received a diluent only. All pigs were orally inoculated with 2.74 × 10 4 viral RNA copies of HuNoV at 6 days of age. Four ml of 200 mM sodium bicarbonate was given 15 min prior to HuNoV inoculation to reduce gastric acidity. Pigs were euthanized on PID3, PID7, or PID10 for collection of blood, intestinal contents, and tissues. Forty cm of distal ileum and whole spleen organ were collected for the isolation of MNC 22 . Pigs used in this study were HBGA-typed to be A + and/or H + , and sterility was confirmed one day before E. cloacae inoculation for all pigs, as well as on euthanasia day for control pigs, as previously described 14 . All animal experimental procedures were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee at Virginia Tech (IACUC protocol: 14-108-CVM).
Assessment of diarrhea and HuNoV shedding. Pig feces were collected daily by rectal swabs following HuNoV inoculation to assess diarrhea and fecal virus shedding. Fecal consistency was scored as follows: 0, solid; 1, semisolid; 2, pasty; 3, semiliquid; and 4, liquid. Pigs with daily fecal scores of 2 or greater were considered diarrheic. Virus shedding was determined as described previously 14 . Briefly, pig feces on rectal swabs were released by swirling in 1 ml PBS and processed as solution, 250 μ l of which was prepared for total RNA isolation using TRIzol LS (Thermo Fisher Scientific). The RNA pellet was resuspended in 40 μ l RNase-free water, and 5 μ l RNA was used in a TaqMan qRT-PCR reaction to detect HuNoV genomes following the manufacturer's instructions in the SensiFAST Probe No-ROX One-Step Kit (Bioline).
Detection of HuNoV genome in tissues and blood. Small intestinal tissues were collected during necropsies, directly frozen in liquid nitrogen, and then stored in − 80 °C freezer. 40 to 60 mg of frozen tissues were thawed at room temperature, washed in 1 ml PBS once, then homogenized in 0.2 ml TRIzol LS using Bullet Blender with 100 mg of 1.0 mm Zirconium Oxide beads (Next Advance). Total RNA from the homogenized tissues was isolated by adding 0.55 ml TRIzol LS. Blood was collected immediately after euthanasia, and 30% of ACD was added to prevent coagulation. Plasma and whole blood cells were separated by centrifugation at 2000g for 5 min. Plasma from 250 μ l blood and whole blood cells from 50 μ l blood for each pig were used for RNA isolation. 1 × 10 7 MNC from ileum and spleen were used for RNA isolation. For these samples, RNA was isolated using 750 μ l TRIzol LS following the manufacturer's instructions, and the HuNoV genome copies were determined by TaqMan qRT-PCR as described above.
Immunohistochemistry. Small intestinal tissues were collected upon euthanasia, fixed in 4% paraformaldehyde overnight, embedded in paraffin, sectioned at 5 μ m, and placed on positively charged glass slides. Tissue slides were deparaffinized and rehydrated by washing in a graded ethanol series. For enzymatic antigen retrieval, slides were digested in 80 μ g/ml proteinase K solution (Sigma Aldrich) for 30 min at 37 °C, followed by washing with tris-buffered saline (TBS). For blocking, slides were incubated in TBS containing 10% normal pig serum and 1% BSA for 2 h at room temperature. For IHC of duodenum and jejunum, slides were incubated with a goat anti-HuNoV GII.4 VLP polyclonal antibody diluted in TBS containing 1% BSA overnight at 4 °C, then washed with TBS containing 0.025% Triton X-100, and incubated with Alexa Fluor 488-labeled donkey anti-goat secondary antibody (A-11055; Thermo Fisher Scientific; 1:500) diluted in TBS containing 1% BSA for 1 h at room temperature. For IHC of ileum, mouse anti-CD79 (VP-C366; Vector Laboratories; 1:1000) and Alexa Fluor 546-labeled donkey anti-mouse secondary antibody (A10036; Thermo Fisher Scientific; 1:500) were also included in the two incubation steps above. Finally, slides were washed in TBS and mounted in Vectashield containing 4,6-diamidino-2-phenylindole (DAPI) for counterstaining cell nuclei (Vector Laboratories). Images were acquired on a Zeiss LSM 880 confocal laser scanning microscope in Fralin Imaging Center at Virginia Tech. Ileum Peyer's patches histopathology. Sections of ileum were prepared as described above, and H&E staining was performed routinely. A pathologist was blinded to identification of the samples and evaluated the IPP histopathology using a light microscope with an ocular micrometer. To characterize the IPP sizes, muscularis mucosae to muscularis (mm to m) associated with gut associated lymphoid tissue (GALT), width of GALT, height and width of follicles were measured. For each parameter, 12 random locations including all pigs in each group were measured.
Statistics. Pigs (male and female) were randomly divided to control group and E. cloacae group, pigs infected with HuNoV in each group were randomly assigned to be euthanized on PID3, PID7, or PID10. To assess clinical signs and virus shedding, pigs euthanized on PID10 contributed data to PID3 and PID7 subgroups, as did PID7 contribute data to PID3 subgroup. Statistical significance was determined with analyses specified in figure legends and table notes using GraphPad Prism 6.0 (GraphPad Software). P value < 0.05 was indicated as statistically significant.