Report

Tropism for tuft cells determines immune promotion of norovirus pathogenesis

See allHide authors and affiliations

Science  13 Apr 2018:
Vol. 360, Issue 6385, pp. 204-208
DOI: 10.1126/science.aar3799

Aiding and abetting norovirus disease

Norovirus is highly infectious and usually causes transient, acute disease. In some individuals, norovirus persists and is associated with inflammatory bowel disorders. While investigating the cell tropism for murine norovirus, Wilen et al. discovered that a rare cell type, tuft cells, carrying the CD300lf receptor were the virus's specific target. Tuft cells proliferate in response to the type 2 cytokines interleukin-4 and interleukin-25, which thereby amplify norovirus infection. Moreover, infected tuft cells are resistant to immune clearance. This effect may explain the associated persistent disease symptoms that humans can suffer.

Science, this issue p. 204

Abstract

Complex interactions between host immunity and the microbiome regulate norovirus infection. However, the mechanism of host immune promotion of enteric virus infection remains obscure. The cellular tropism of noroviruses is also unknown. Recently, we identified CD300lf as a murine norovirus (MNoV) receptor. In this study, we have shown that tuft cells, a rare type of intestinal epithelial cell, express CD300lf and are the target cell for MNoV in the mouse intestine. We found that type 2 cytokines, which induce tuft cell proliferation, promote MNoV infection in vivo. These cytokines can replace the effect of commensal microbiota in promoting virus infection. Our work thus provides insight into how the immune system and microbes can coordinately promote enteric viral infection.

Human noroviruses (HNoVs) are the leading cause of acute viral gastroenteritis worldwide, causing up to 700 million infections and 200,000 deaths annually (1). Despite this disease burden, it is unknown what cell type(s) mediate transmission and, in some individuals, chronic infection (2). Murine norovirus (MNoV) represents a model for HNoV pathogenesis and immunity. More broadly, MNoV serves as a tractable system to uncover previously unidentified virus-host interactions such as the capacity of MNoV infection to trigger human-relevant pathology in genetically susceptible animals and the role of intestinal bacteria in promoting enteric viral infection (38). Identifying the cell tropism of MNoV could provide mechanistic insight into such phenomena and thereby shed light on enteric immunity and the genotype-phenotype relationship.

Norovirus tropism is not fully understood in either immunocompetent mice or humans. Recently, we showed that a small population of epithelial cells is the reservoir for chronic MNoV infection and that this epithelial cell tropism is determined in part by the MNoV nonstructural protein NS1 (9). However, the reason for selective intestinal epithelial cell infection, how infected cells differ from adjacent cells in the intestinal epithelium, and why we seldom observed adjacent infected epithelial cells are unknown. We recently identified CD300lf as a protein receptor for MNoV (10, 11). CD300lf is both necessary and sufficient for infection in vitro, and Cd300lf−/− animals are resistant to fecal-oral transmission of persistent MNoV infection (10). In this study, we used this finding to identify the target cell of MNoV in vivo.

Because MNoV readily replicates in explanted macrophages and dendritic cells that express CD300lf (12), we first sought to determine whether bone marrow–derived myeloid cells were responsible for infection by performing bone marrow transplants between Cd300lf−/− and wild-type (WT) littermates, followed by oral infection with MNoV strain CR6 (MNoVCR6). MNoVCR6 infection is characterized by robust fecal-oral transmission, persistent enteric infection resistant to adaptive immune clearance, and prolonged fecal shedding (13). Such persistent MNoV strains replicate predominantly in the distal small intestine and colon and can be detected in mesenteric lymph nodes (MLNs), but evidence of infection in the spleen is scant (14).

WT mice receiving WT bone marrow remained susceptible to MNoVCR6, but Cd300lf−/− mice receiving Cd300lf−/− bone marrow were resistant to MNoV infection, as measured by fecal shedding of MNoV (Fig. 1A) and tissue levels of viral genomes 21 days after infection (Fig. 1, B to E). Surprisingly, WT mice that received Cd300lf−/− bone marrow were susceptible to MNoVCR6, and Cd300lf−/− animals receiving WT bone marrow were resistant to infection. Viral titers in the ileum and colon correlated with those in feces (Fig. 1, A to C). WT mice that received either WT or Cd300lf−/− bone marrow transplants remained susceptible to MNoVCR6. Splenic infection was minimal in all groups examined, consistent with findings from prior studies with nonirradiated WT animals (Fig. 1D) (13). MNoVCR6 genomes were undetectable in the MLNs of Cd300lf−/− mice receiving either WT or Cd300lf−/− bone marrow, but viral genomes were detected at similar levels in the MLNs of WT mice receiving either WT or Cd300lf−/− bone marrow transplants (Fig. 1E). Thus, a recipient Cd300lf genotype was the primary determinant of MNoVCR6 intestinal replication and shedding, indicating that radiation-resistant cells were responsible for MNoVCR6 enteric infection. In contrast, both radiation-sensitive and -resistant cells contributed to infection with MNoV strain CW3 (MNoVCW3), which causes acute systemic infection (1416). The inability of MNoVCW3 to infect epithelial cells, to be shed in the feces, and to establish chronic infection maps to the viral NS1 protein, which is required to counteract interferon-λ (IFN-λ) signaling (5, 9, 13). We focused further efforts on identifying the tropism responsible for MNoVCR6 enteric infection and shedding.

Fig. 1 Fecal-oral MNoV transmission requires radiation-resistant cells.

Reciprocal bone marrow transplants were performed among Cd300lf+/+ (WT) and Cd300lf−/− [knockout (KO)] littermates. Mice were then challenged perorally with MNoVCR6, which establishes persistent enteric infection in WT animals. (A) WT mice remained susceptible to MNoV, as measured by numbers of viral genome copies in feces at the indicated time points. In contrast, KO mice did not shed MNoVCR6 whether they received WT or KO bone marrow. (B to E) Twenty-one days postchallenge, MNoV viral genome loads were measured in the ileum (B), colon (C), spleen (D), and mesenteric lymph nodes (MLNs) (E). WT recipients had significantly more viral genomes than KO recipients. There was no significant difference between WT recipients of either WT or KO bone marrow. Fecal samples were analyzed by repeated-measures analysis of variance (ANOVA). Tissue samples were analyzed by one-way ANOVA. Significant differences for both fecal and tissue samples were relative to data for the WT donor–WT recipient (WT→WT) control as indicated. Means ± SEM are shown. NS, not significant; **P < 0.01; ***P < 0.001; ****P < 0.0001. LOD, limit of detection (represented by the dotted line in each graph). Data are pooled from three independent experiments. The numbers of mice per group are indicated in (A).

Consistent with our bone marrow transplant data, we recently determined that rare isolated intestinal epithelial cells were infected by MNoVCR6 during chronic infection, though the identity of the cells was not defined (9). Together with the bone marrow transplantation experiments described above, these findings indicate that a radiation-resistant epithelial cell must express the MNoV receptor (9). However, CD300lf is an immunoregulatory protein thought to be expressed on hematopoietic cells, particularly myeloid cells (17, 18). Expression of CD300lf on epithelial cells has not been described previously. We therefore performed immunofluorescence microscopy analysis of uninfected WT mice and observed a rare population of CD300lf-expressing cells throughout the ilea and colons (Fig. 2, A and B). Given the amphora-like morphology and the scarcity of CD300lf-expressing epithelial cells, we hypothesized that they were tuft cells, a rare chemosensory epithelial cell type in the hollow organs of mammals, including mice and humans (19). These cells, also known as brush, caveolated, multivesicular, or fibrillovesicular cells, contain a long apical “tuft” of microvilli, which protrudes into the intestinal lumen, and were recently discovered to be the primary source of interleukin-25 (IL-25), a cytokine that initiates a type 2 immune response upon intestinal helminth or parasite infection (2022). Indeed, all observed CD300lf+ epithelial cells expressed the tuft cell markers doublecortin-like kinase 1 (DCLK1) and cytokeratin 18 (CK18) (Fig. 2, A and B) (23). We also confirmed tuft cell–specific expression of Cd300lf transcripts in previously reported single-cell RNA sequencing (RNA-seq) data from mouse intestinal enteroids (24, 25). Next, we assessed CD300lf expression on intestinal epithelial cells (EpCAM+CD45) in a mouse line expressing a tuft cell–specific fluorescent reporter (Gfi1b-GFP) (26). There was near perfect concordance between Gfi1b-GFP expression and CD300lf expression in both the ileum and colon, confirming that tuft cells are distinct among epithelial cells in their expression of CD300lf (Fig. 2C).

Fig. 2 CD300lf is expressed on tuft cells but not on other intestinal epithelial cells.

(A and B) The MNoV receptor CD300lf is detectable on rare intestinal epithelial cells with morphology consistent with tuft cells. CD300lf colocalizes with tuft cell markers (A) DCLK1 and (B) CK18 in mouse ileum and colon. CD300lf is apically polarized toward the intestinal lumen. (C) CD300lf is expressed on Gfi1b-GFP+ tuft cells but not on other intestinal epithelial cells, as measured by flow cytometry. Events shown are Singlets+Live+CD45EpCAM+. Images and fluorescence-activated cell sorting plots are representative of one of at least three independent experiments. Dashed lines represent the epithelial barrier. White boxes in the overlay images correspond to the magnified inset images. Scale bars, 10 μm.

Given these findings, we assessed whether MNoVCR6 infects tuft cells. Immunofluorescence microscopy on intestines of WT mice infected with MNoVCR6 revealed rare cells expressing the MNoV nonstructural protein NS6/7 (Fig. 3A). These cells were in direct contact with the intestinal lumen and were observed in the surface epithelium of the colon and in both the villi and crypts of the ileum. All MNoV NS6/7-positive cells coexpressed the tuft cell marker DCLK1. No viral antigen–positive cells were observed in the lamina propria or immune cells. Similar histologic findings and viral tropism were identified in WT germ-free mice (fig. S1), indicating that intestinal bacteria are not required for either CD300lf expression by tuft cells or MNoVCR6 infection of tuft cells. To confirm and quantify MNoVCR6 infection of tuft cells, we performed flow cytometry analysis of colonic epithelial cells from infected Gfi1b-GFP mice. MNoVCR6 infection did not significantly reduce tuft cell frequency (Fig. 3B). Infected cells were defined as those expressing two independent viral nonstructural proteins (NS1/2 and NS6/7) (9). We observed 128 ± 33 (mean ± SEM) infected tuft cells per million live epithelial cells (EpCAM+CD45). We did not observe infection of non-tuft epithelial cells (Fig. 3, C and D). Overall, 1.4 ± 0.37% of Gfi1b-GFP+ tuft cells were MNoV infected. Together, our immunofluorescence and flow cytometric analyses indicate that tuft cells are the physiologic target cell of MNoV in WT animals. This finding likely explains why we did not observe clusters of infected cells in the intestine, as tuft cells are isolated from one another, being surrounded by other intestinal epithelial cells (9).

Fig. 3 MNoVCR6 specifically infects CD300lf-expressing intestinal tuft cells.

(A) MNoV nonstructural protein NS6/7 colocalizes with DCLK1 in the ilea and colons of WT mice infected with MNoVCR6 at 7 days postinfection. NS6/7 expression is punctate and cytoplasmic, consistent with the viral replication complex. (B) Flow cytometry analysis of intestinal epithelial cells (IECs) (Singlet+Live+CD45EpCAM+) from Gfi1b-GFP+ tuft cell reporter mice revealed similar frequencies of tuft cells in infected and uninfected mice. (C) A rare population of cells that coexpress the MNoV nonstructural proteins NS1/2 and NS6/7 was observed. These MNoV-positive cells are Gfi1b-GFP+ cells, demonstrating that they are tuft cells. (D) NS1/2+ NS6/7+ events were significantly enriched among GFP+ cells. NS1/2+ NS6/7+ events were at background levels among non-tuft cells. Data are pooled from three independent experiments with one to two mice per group. Shown are means ± SEM. NS, not significant; *P < 0.05; **P < 0.01. Dashed lines represent the epithelial barrier. White boxes in the overlay images correspond to the magnified inset images. Scale bars, 10 μm.

Given the role of tuft cells in type 2 immunity, we hypothesized that there might be an intimate relationship between type 2 immunity and enteric norovirus infection. The type 2 cytokines IL-4 and IL-25 induce tuft cell hyperplasia (2022). Therefore, we assessed whether these cytokines augmented MNoV transmission. WT mice were treated with IL-4, IL-25, or a phosphate-buffered saline (PBS) control prior to peroral challenge with a low dose [4.25 × 104 plaque-forming units (PFU) per mouse] of MNoVCR6 insufficient to establish robust infection in the majority of control mice. Both IL-4– and IL-25–treated animals were significantly more likely to be productively infected than PBS-treated animals, as measured by numbers of viral genome copies in the feces 7 days postchallenge (Fig. 4A). These results show that type 2 immune responses can enhance enteric viral transmission. We therefore asked whether type 2 cytokines affect MNoVCR6 fecal shedding during persistent infection. WT mice were challenged perorally with a high dose (106 PFU/mouse) of MNoVCR6 that is sufficient to infect all animals. After at least 21 days of infection, IL-4 or a PBS control was injected intraperitoneally and fecal shedding of virus was monitored. IL-4 significantly increased MNoVCR6 fecal shedding as detected 1 day after the second and final IL-4 injection (Fig. 4B). As in WT animals, IL-4 increased MNoV shedding in Rag1−/− and Ifnlr1−/− mice persistently infected with MNoV (Fig. 4B), indicating that cytokine promotion of infection was not caused by effects on T cells or B cells and is independent of IFN-λ–induced innate immune signaling, a potent regulator of intestinal norovirus infection (5). In addition, we demonstrated that IFN-λ treatment did not alter tuft cell abundance in the intestine (fig. S3).

Fig. 4 Tuft cell tropism determines transkingdom interactions of MNoV.

(A) WT mice were injected intraperitoneally with PBS, IL-4, or IL-25 prior to peroral challenge with a low dose (4.25 × 104 PFU) of MNoVCR6. Both IL-4 and IL-25 increase MNoV transmission, as measured by detection of MNoV genomes in feces 7 days postinfection. The numbers above each column reflect the number of infected animals relative to the total number of animals per group (chi-square test, <0.0015). (B) WT, Rag1−/−, and Ifnlr1−/− mice chronically infected with a high dose (106 PFU) of MNoVCR6 for 21 days were administered PBS or IL-4. MNoV fecal shedding significantly increased in WT, Rag1−/−, and Ifnlr1−/− mice after IL-4 injection (24 days postinfection) compared with that after PBS administration. (C) IL-4 enhancement of MNoV fecal shedding during chronic infection requires Il4rα expression on VillinCre-expressing epithelial cells. (D) Broad-spectrum antibiotics (vancomycin, neomycin, ampicillin, and metronidazole), which prevent MNoVCR6 infection, significantly reduce tuft cell–specific gene transcripts as measured by RNA-seq in the colon but not in the ileum. NES, normalized enrichment score; FDR, false discovery rate. (E and F) DCLK1+ tuft cells were quantified by immunofluorescence microscopy. Antibiotics reduce DCLK1+ tuft cells in the colon but not in the ileum. IL-4 and IL-25 increase DCLK1+ tuft cells in the ileum but not in the colon. (G) Antibiotic pretreatment prevents MNoVCR6 infection. This antiviral state can be reversed with IL-4 or IL-25 administration prior to MNoVCR6 challenge. Shown are means ± SEM. NS, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Data in mouse experiments are pooled from at least three independent experiments with two to six mice per group, except for the Ifnlr1−/− study, in which data are pooled from two independent experiments. Each dot in (E) and (F) represents the tuft cell frequency in one mouse. At least 10 independent low-power images were averaged per mouse. Data were analyzed by the Mann-Whitney U test unless otherwise indicated.

The murine parasite Trichinella spiralis induces type 2 inflammation and augments MNoVCR6 infection (7). The mechanism of action was previously hypothesized to be increased viral replication in alternatively activated macrophages exposed to type 2 cytokines such as IL-4 and IL-13 (7). However, in this study we showed that tuft cells and not macrophages are the target cell for MNoVCR6. Thus, we tested whether the enhanced MNoVCR6 infection resulting from IL-4 treatment was mediated by effects of this cytokine on epithelial cells. To test this hypothesis, we generated epithelial cell–specific IL-4 receptor α (Il4rα) conditional knockout mice (Il4rαf/f × VillinCre) (27). Mice were infected with MNoV for at least 21 days, after which IL-4 was administered. IL-4 enhanced MNoVCR6 shedding in Il4rαf/f × VillinCre animals but not Il4rαf/f × VillinCre+ animals, demonstrating that IL-4 signals through its receptor on epithelial cells (Fig. 4C) (22). These data suggest that IL-4 promotes norovirus infection via effects on tuft cells, the only epithelial cells infected with the virus.

Prior work showed that the bacterial microbiome is required for efficient establishment of enteric MNoVCR6 infection (6). Specifically, broad-spectrum antibiotics that deplete intestinal bacteria prevent MNoVCR6 transmission and persistent infection (6). The mechanism for this effect is incompletely understood. We therefore asked whether antibiotic treatment affected expression of tuft cell–specific transcripts. RNA-seq was performed on control and antibiotic-treated mice, and the expression of a list of curated tuft cell genes was used to assess differences in tuft cell–specific genes (24). Antibiotic treatment resulted in a decrease in tuft cell–specific gene expression in the colon (normalized enrichment score, 2.23; P < 0.001; false discovery rate, <0.001) (Fig. 4D); changes in tuft cell genes did not reach statistical significance in the ileum. Consistent with the RNA-seq gene set enrichment analysis, antibiotics decreased DCLK1+ cells in the colon but not the ileum (Fig. 4, E and F). IL-4 and IL-25 induced tuft cell hyperplasia in the ilea of antibiotic-treated mice, whereas colonic tuft cells were not increased in number by IL-4 or IL-25 (Fig. 4, E and F). These findings indicate that both type 2 cytokines and intestinal bacteria regulate tuft cells, albeit in a tissue-specific manner (22). The observation that intestinal bacteria contribute to tuft cell regulation in vivo raised the question of whether the antiviral role of antibiotics could be overcome with administration of type 2 cytokines that act on epithelial cells to control MNoV infection. WT mice were pretreated with antibiotics for 2 weeks prior to challenge with a high dose (106 PFU) of MNoVCR6. Consistent with prior findings, antibiotics significantly reduced MNoVCR6 infection (Fig. 4G) (6). IL-4 or IL-25 administration, prior to MNoVCR6 challenge, rescues viral infection in antibiotic-treated mice. Both IL-4 and IL-25 significantly increased both the proportion of mice infected with virus and the magnitude of fecal shedding (Fig. 4G). The differential regulation of tuft cells by type 2 cytokines and antibiotics in the ileum and colon, respectively, suggests that a threshold number of tuft cells may matter more than the anatomic location of tuft cells within the intestine.

Here, we have identified intestinal tuft cells as the physiologic target cell of MNoV. This discovery has important implications for our understanding of transkingdom interactions and the pathogenesis of persistent intestinal infection. Norovirus infection triggers inflammatory bowel disease–like phenotypes in genetically susceptible hosts (3, 4). Now that we have identified the tropism of norovirus for tuft cells, a question to consider is whether tuft cells regulate inflammatory bowel disease–like phenotypes. Tuft cell tropism also links the proviral effects of helminths and commensal bacteria, which increase tuft cells in the ileum and colon, respectively (2022). Noroviruses can persist in the intestine for months in both mice and humans (2, 2830). This persistent infection is resistant to both antibody and CD8+ T cell–mediated clearance, yet the mechanism of immune evasion is unknown (31). Our identification of MNoV tropism for tuft cells suggests that tuft cells represent an immune-privileged site for enteric viral infection in mice. It is possible that other viruses also infect tuft cells, enabling these viruses to take advantage of type 2 immune responses to promote infection.

Supplementary Materials

www.sciencemag.org/content/360/6385/204/suppl/DC1

Materials and Methods

Figs. S1 to S3

Table S1

References (3239)

References and Notes

Acknowledgments: We acknowledge J. Guo, X. Zhang, A. Orvedahl, A. Mayer, T. Schaiff, M. Artyomov, G. Broze, and N. Lasky for helpful discussions and technical support. We also thank W. Beatty at the Molecular Microbiology Imaging Facility and E. Lantelme and D. Brinja at the Flow Cytometry Core at Washington University School of Medicine. Funding: This work was supported by NIH grants K08 AI128043 (C.B.W.); K22 AI127846 (M.T.B.); K99 DK116666 (R.C.O.); T32 AI007163 (C.C.Y.); R01 AI127552 and U19 AI109725 (H.W.V.); and DK093668, DK103788, HL123340, and R01 AI121244 (J.A.N. and K.C.). In addition, S.L. was supported by the Basic Sciences Research Program through the National Research Foundation of Korea, funded by the Ministry of Education (grant NRF-2016R1A6A3A03012352). M.T.B. was supported by Digestive Diseases Research Core Centers grant P30 DK052574. J.A.N. was supported by a Vilcek Fellowship and a Sir Keith Murdoch Fellowship. Author contributions: C.B.W., S.L., L.L.H., M.R.M., D.R.B., C.C.Y., Y.-T.W., D.A.M., D.K., J.A.N., K.C., and M.T.B. performed experiments. C.B.W., S.L., R.C.O., C.A.H., K.C., P.M.A., M.R.H., S.A.H., M.V.L.C., M.T.B., and H.W.V. designed the project. C.B.W., S.L., L.L.H., R.C.O., C.D., B.L.H., T.F., J.R.B., H.W.V., and S.A.H. analyzed data. C.B.W. wrote the paper. All authors read and edited the manuscript. Competing interests: C.B.W., R.C.O., and H.W.V. are inventors on a patent application submitted by Washington University entitled “Receptor for norovirus and uses thereof” (U.S. provisional application 62/301,965). M.V.L.C. is an employee of Genentech, a for-profit institution. Data and materials availability: The data from this study are tabulated in the main paper and supplementary materials. All reagents are available from H.W.V. under a material transfer agreement with Washington University. All data and code to understand and assess the conclusions of this research are available in the main text and supplementary materials and via the European Nucleotide Archive under accession no. PRJEB23132.
View Abstract

Stay Connected to Science

Navigate This Article