Genetic Manipulation of Human Intestinal Enteroids Demonstrates the Necessity of a Functional Fucosyltransferase 2 Gene for Secretor-Dependent Human Norovirus Infection.

Several studies have demonstrated that secretor status is associated with susceptibility to human norovirus (HuNoV) infection; however, previous reports found that FUT2 expression is not sufficient to allow infection with HuNoV in a variety of continuous laboratory cell lines. Which cellular factor(s) regulates susceptibility to HuNoV infection remains unknown. We used genetic manipulation of HIE cultures to show that secretor status determined by FUT2 gene expression is necessary and sufficient to support HuNoV replication based on analyses of isogenic lines that lack or express FUT2. Fucosylation of HBGAs is critical for initial binding and for modification of another putative receptor(s) in HIEs needed for virus uptake or uncoating and necessary for successful infection by GI.1 and several GII HuNoV strains.

H uman noroviruses (HuNoVs) are a leading cause of nonbacterial gastroenteritis worldwide. From the first recognition in 1968 that a virus caused an outbreak of HuNoV gastroenteritis in an elementary school in Norwalk, Ohio (1), until 2016, there was no in vitro culture system of HuNoV in intestinal epithelial cells. A novel HuNoV culture system using human intestinal enteroids (HIEs) generated from stem cells isolated from human small intestinal crypts is now available and is being used worldwide to study virus replication, inactivation, and neutralizing antibodies (2)(3)(4)(5)(6)(7)(8)(9).
Secretor status is highly associated with infection and disease caused by many HuNoV strains based upon findings from human experimental infection studies and evaluation of acute gastroenteritis outbreaks (10)(11)(12)(13)(14). Fucosyltransferase 2 (FUT2) is an enzyme expressed in human epithelial cells that catalyzes ␣1,2-fucosylation of the terminal galactose, preferentially on glycan type 1 chain precursors. Glycosyltransferases coded for by FUT2 along with FUT3 and ABO genes determine the histo-blood group antigens (HBGAs) found on the epithelial cell surface (15,16). Persons who lack functional FUT2 alleles do not express ABH HBGAs on epithelial cells, are designated nonsecretors, and are highly resistant to gastroenteritis caused by some HuNoV strains such as GII.4 viruses. Similarly, HIEs derived from nonsecretor individuals are not susceptible to GII.4 HuNoV infection (2). HIEs and gastrointestinal epithelial cells of secretor-positive individuals also bind norovirus virus-like particles (VLPs) (17,21). However, while HIEs from secretors are permissive to HuNoV replication, FUT2 expression in conventional cancer-derived cell lines is not sufficient to make cells susceptible to HuNoV replication, suggesting that HBGAs function primarily as an initial attachment factor (18,19). These data also lead to questions on whether there are additional genetically determined differences in HuNoV susceptibility in addition to secretor status. In this study, we evaluated the direct association between FUT2 function and HuNoV infectivity using HIEs with or without FUT2 expression in the same genetic background.

RESULTS
Generation and characterization of isogenic HIE lines. We previously showed that GII.4 viruses can replicate in HIEs derived from secretor-positive individuals and a GII.3 strain can replicate in both secretor-positive and some secretor-negative lines, recapitulating observations seen in epidemiologic studies (2). We determined the FUT2 (Se) and FUT3 (Le) genotypes of jejunal HIE lines and selected two lines (J2 and J4); both lines express FUT3, eliminating Lewis status as a variable in these studies ( Table 1). The J2 cell line is heterozygous wild type (Se/se 428 [J2Fut2 ϩ/Ϫ ]); because the Se gene is autosomal dominant, the cells are secretor positive. The J4 line has a homozygous se 428 /se 428 (J4Fut2 Ϫ/Ϫ ) recessive mutation and is secretor negative (20). No other mutations associated with loss of or decreased FUT2 enzymatic function were observed in the J2 or J4 line. To better understand the importance of FUT2 in HuNoV infection, we generated isogenic knockout (KO) and knock-in (KI) HIE lines. A lentivirus-delivered CRISPR/Cas9 construct with a guide RNA targeting the FUT2 coding region was used to knock out FUT2 from J2 (J2Fut2 Ϫ/Ϫ ). Single cell selection under puromycin pressure was used to isolate a clonal population, and deletions of the gene in both alleles were confirmed by sequencing. To generate the KI FUT2 line, we transduced a functional FUT2 coding sequence driven by a cytomegalovirus (CMV) promoter in a lentivirus into J4 (J4Fut2 Ϫ/Ϫ/FUT2 ) cells.
First, we confirmed the phenotype of the FUT2 KO and KI lines by evaluating HBGA expression using an enzyme-linked immunosorbent assay (ELISA) (Fig. 1). J2Fut2 ϩ/Ϫ cells expressed the secretor-positive glycans, Le b and B, as expected for this secretorpositive, Lewis-positive OB HIE line. KO of FUT2 altered the J2 phenotype such that the Le b and B glycans were no longer present, and only Le a was detected. J4Fut2 Ϫ/Ϫ cells expressed Le a but not Le b or other secretor glycans, confirming the secretor-negative genotype of this line. KI of FUT2 in J4 cells led to Le b instead of Le a expression.
We next used immunofluorescence microscopy with fluorescently labeled Ulex europaeus agglutinin-1 (UEA-1 lectin) to detect polarized cell surface expression of terminal ␣1,2-fucose (Fig. 2). Both J2Fut2 ϩ/Ϫ and J4Fut2 Ϫ/Ϫ/FUT2 cells had apical staining of UEA-1 lectin. This apical staining was lost in J2Fut2 Ϫ/Ϫ and J4Fut2 Ϫ/Ϫ cells ( Fig. 2A). These findings indicate that, as expected, terminal fucosylation of the glycan precursor in HIEs depends on the expression of the FUT2 gene. In the secretor-negative lines, unexpected internal cellular UEA-1 staining was present that was not observed in the secretor-positive lines. To confirm that the internal staining was due to specific UEA-1 recognition of ␣1,2-fucose and not off-target binding due to loss of the specific ligand, we preincubated the UEA-1 lectin with L-fucose prior to staining. When UEA-1 was preincubated with 10 mM L-fucose, staining was not detected in the secretornegative lines and was significantly reduced in the secretor-positive lines (Fig. 2B). After preincubation with 100 mM L-fucose, UEA-1 staining was not detected in either the secretor-negative or secretor-positive lines (Fig. 2C). Together, these data indicate that the apical and the internal staining observed with UEA-1 is due to specific interaction with ␣1,2-fucose. Table 1 summarizes the genotypic and phenotypic findings for the parental, KO and KI cell lines.
Effect of FUT2 expression on replication of HuNoV strains in isogenic HIE cell lines. To assess whether these isogenic cell lines support HuNoV replication, we evaluated virus binding and subsequent replication of GII.4, GII.3, GII.17, and GI.1 HuNoV strains that we previously demonstrated replicate in HIEs (2). Since all of these viruses but GII.4 require bile for replication, all infections were performed in the presence of glycochenodeoxycholic acid (GCDCA), a bile acid that supports replication (21).
First, we assessed the effect of KO of FUT2 from J2 cells with the four HuNoV strains (Fig. 3). After 1 or 2 h postinfection (hpi) and washing off the inoculum, GII.4, GII.3, and  (2). These findings support the requirement for functional FUT2 and secretor-positive HBGAs as an initial binding factor for several strains of HuNoV that is critical for replication. Next, we assayed the J4Fut2 Ϫ/Ϫ/FUT2 line to determine whether a genetically resistant HIE line can become susceptible with the expression of FUT2. We observed increased HuNoV binding at 1 or 2 hpi in the J4Fut2 Ϫ/Ϫ/FUT2 line compared with the parental nonsecretor J4Fut2 Ϫ/Ϫ line for all four HuNoV strains (Fig. 4); however, for GII.4, the increase was not significant. J4Fut2 Ϫ/Ϫ/FUT2 was permissive to infection by all four viruses, based upon increases in viral genome equivalents (GEs) at 24 hpi. Interestingly, although GE levels at 1 to 2 hpi were similar, we observed greater increases in GEs for GII.17 and GI.1 HuNoVs in J4Fut2 Ϫ/Ϫ/FUT2 cells than in J2Fut2 ϩ/Ϫ cells at 24 hpi. This suggests that additional factors that vary between individual HIE lines, aside from secretor status, may influence the replication of these two strains. Taken together, these results show that functional FUT2 is sufficient and critical for replication of multiple HuNoV strains and that GII.3 is capable of infection in some secretor-negative HIE lines.

DISCUSSION
Previous studies demonstrated that the HuNoV genome is capable of productive infection in transformed cell lines (e.g., Huh-7) following transfection. Overexpression of the FUT2 gene increased virus binding to the cells but did not make them susceptible to infection (19). These data indicated the presence of a block between virus binding and entry. The development of the HIE HuNoV infection model identified that enterocytes in these cultures are infected with HuNoVs and allowed the performance of studies to evaluate factors important for virus infection (2). The current study demonstrates that FUT2 alone is necessary and sufficient for the infection and replication of the secretor-dependent GI.1, GII.4, and GII.17 HuNoVs in HIE cells.
Virus attachment and entry into the infected cell are complex processes, and the exact role of fucosylated molecules in HuNoV entry remains to be determined. It is likely that initial binding of secretor-dependent HuNoVs to HIEs is regulated by fucosylation, with the attachment factors being fucosylated. However, it is unclear whether the fucosylated molecules serve only as the initial attachment factor that then facilitates interaction with a virus-specific receptor or whether the fucosylated molecules function as a virus receptor themselves; examples of both mechanisms have been described previously with nonfucosylated glycans (22). For example, reovirus binding to its proteinaceous JAM-A receptor is enhanced by initial binding to sialic acid residues (23), while human coronaviruses OC43 and HKU1 bind to 9-O-acetylated sialic acid receptors (24) and influenza viruses bind to terminal sialic acid receptors (25). If fucosylated molecules serve only as attachment factors, then the virus receptor(s) for secretordependent HuNoVs is present in both secretor-positive and secretor-negative HIEs and is not expressed in other nonsusceptible, secretor-positive cancer-derived or trans-  We observed a striking phenotypic difference in FUT2-expressing HIE cells compared to cells not expressing FUT2. UEA-1 detection of ␣1,2-fucose was observed exclusively at the apical surface in FUT2-positive cells. Unexpectedly, in the secretor-negative lines, there was internal cellular staining by UEA-1. The staining was due to UEA-1-specific interactions with ␣1,2-fucose, indicating that these glycan structures are present but unable to transit to the surfaces of the secretor-negative HIE cultures. There is evidence that FUT2 expression leads to surface expression of ␣1,2-fucosylated molecules. This was shown in specific-pathogen-free mice, where the lumen of the small intestine is mostly lacking in surface fucosylation, and intraperitoneal injection of lipopolysaccharide (LPS) stimulates expression of ␣1,2-fucosylated molecules at the surface in a FUT2-dependent manner (27,28). There may be additional fucosyltransferase enzymes present in our HIEs capable of adding fucose to glycoproteins or glycolipids in the absence of FUT2 activity. However, these alternatively fucosylated molecules are unable to transit to the cell surface. Fucosyltransferase 1 (FUT1), expressed in erythroid cells and some other tissues, is also capable of adding ␣1,2-fucose, although preferentially on glycan chains other than those found in intestinal epithelia. In the parental J2 enteroid line used in this study, FUT1 transcripts are expressed by transcriptome sequencing (RNA-seq), and the fragment per kilobase per million (FPKM) value is  Fig. 3. Total well RNA was extracted, and GEs were determined by RT-qPCR. Each data bar represents the mean for three wells of inoculated HIE monolayers. Error bars denote SD. Each experiment was performed two or more times, with three technical replicates in each experiment. Numbers above the bars indicate log 10 fold change comparing GEs at 24 hpi to 1 or 2 hpi. Significance was determined by two-way ANOVA with post hoc analysis using Tukey's test (*, P Ͻ 0.05; n.s., not significant).
ϳ60-fold lower than that of FUT2. In bovine coronary venular endothelial cells, a lotus lectin (LTL) that also recognizes ␣1,2-fucosylated molecules detects cytoplasmic tubule structures, and this staining is lost after depletion of both FUT1 and FUT2 (29). Further studies showed that LTL costained with a Golgi marker in newly derived primary human fibroblasts from oral mucosa, but this Golgi colocalization was lost and LTL was instead detected in tubule structures as the fibroblasts were passaged. In our system, the presence of some FUT1 expression may allow for ␣1,2-fucose-containing structures intracellularly. Future studies will determine the identity and subcellular location of the fucosylated molecules present in the FUT2-negative HIE lines and whether FUT1 is required for their presence.
HIEs provide an excellent tool for future studies on intestinal enzymes involved in glycosylation and how glycosylation alters glycoprotein localization. An association of enteric commensals and pathogens with host secretor status has led to increased recognition of secretor glycans being susceptibility factors important in infection and disease outcomes (30,31). The exact role played by the glycans in these infections is not fully understood. Our isogenic, physiologically active HIE lines should be helpful to determine whether fucosylation plays a role in microbe binding, entry, or postentry processes that may affect the epithelial responses to infection, and it will be interesting to understand different outcomes of infection with the different microbes.
Lentiviral transduction of HIEs. A cell suspension was prepared from three-dimensional (3D) undifferentiated jejunal HIEs cultivated as previously described (2,33). After trypsinization and pelleting of the cells at 300 ϫ g, the resulting cell pellet was suspended at a concentration of 3 ϫ 10 5 cells per ml of concentrated lentivirus supplemented with 10 M Rho-associated protein kinase (ROCK) inhibitor Y-27632 (catalog no. Y0503; Sigma) and 8 g/ml Polybrene (catalog no. TR-1003-G; EMD Millipore). The mixture was plated in one well of a 48-well plate. The plate was then centrifuged for 1 h at 300 ϫ g at room temperature (RT). After spinoculation, the lentivirus solution was removed, the cells were washed once with CMGF(Ϫ) medium, centrifuged again, embedded in 30-l Matrigel plug, and incubated at 37°C and 5% CO 2 in the presence of CMGF(ϩ) medium with ROCK inhibitor for recovery. Five days postransduction, the cells were treated with puromycin (2 g/ml) or Blasticidin S (5 g/ml) until mock-treated cells were completely dead. Single cells were isolated by serial dilution in 96-well plates for sgRNAtransduced HIEs, and deletion of the gene was confirmed by sequencing of genomic DNA from each single cell clone using primers BO-119 and BO-120 ( Table 2) that amplified the portion of the FUT2 gene targeted by the sgRNA. HIE HBGA phenotyping. Differentiated HIE cells were suspended in phosphate-buffered saline (PBS) and boiled for 5 min. After incubation of the boiled supernatants on vinyl, flat-bottomed, 96-well plates (ThermoFisher Scientific) for 4 h at RT and blocking with 10% Carnation instant nonfat dry milk overnight at 4°C, anti-Le a Gamma-clone (Immucor), anti-Le b (BG-6; Biolegend), anti-A type Gamma-clone (Immucor), or anti-B type Gamma-clone (Immucor) was used as the primary antibody, and horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (Sigma) was used as the secondary antibody. 3,3=,5,5=-Tetramethylbenzidine (TMB) (KPL) was used as a HRP substrate, and the reaction was stopped with 1 M phosphoric acid (Fisher Scientific) after a 10-min incubation at RT. The absorbance was determined at 450 nm with a Spectramax 190 plate reader (Molecular Devices). Pooled saliva samples from persons who collectively express each histo-blood group antigen (HBGA) type evaluated was used as a positive control (34). Saliva from an individual negative for the HBGAs was used as a negative control.
Immunofluorescence and quantitation. HIE monolayers were grown in glass bottom plates (catalog no. 655892; Greiner Bio-One), differentiated for 5 days, and fixed with 4% paraformaldehyde (PFA) for 20 min at RT. HIE monolayers were permeabilized and blocked with 5% bovine serum albumin (BSA) in 0.1% Triton X-100 in PBS for 30 min at RT. All the subsequent steps were performed in PBS plus 0.1% Triton X-100. HBGAs and cell boundaries were detected after overnight incubation at 4°C with rhodamine-labeled Ulex europaeus agglutinin-1 (UEA-1) (1:1,000) (catalog no. RL-1062; Vector Laboratories) and Alexa Fluor 647 phalloidin (ThermoFisher Scientific), respectively. For the fucose inhibition assay, the UEA-I was incubated with different concentrations of L-fucose at RT for 1 h prior to staining. Nuclei were stained with 4=,6=-diamidino-2-phenylindole (DAPI) (300 nM) for 5 min at RT. Orthogonal 5-m-thick sections of the sample were captured using a Zeiss LSM 510 confocal microscope. For quantifying fluorescence intensity, five fields per well were analyzed. The fluorescence threshold in these images was set in Image J. Mean fluorescence data from 15 identical regions of interest (ROIs) per field were collected. Comparisons between treatment groups were made using a Student's t test. P values of Ͻ0.05 were considered statistically significant.
In /BCM723-02, 6.9 ϫ 10 5 GEs/well) as described previously (2). Each infection was performed in triplicate wells for each time point, and conditions were tested in at least two independent experiments. Inocula were removed, and monolayers were washed twice with CMGF(Ϫ) medium to remove unbound virus. Differentiation INT medium (100 l supplemented with 500 M glycochenodeoxycholic acid [GCDCA] [Sigma]) was then added to each well, and the cultures were incubated at 37°C for the indicated time points. RNA was extracted from each well using the KingFisher Flex Purification system and MAgMAX-96 viral RNA isolation kit. RNA extracted at 1 to 2 hpi was used as a baseline to determine the amount of input virus that remained associated with cells after the infected cultures were washed to remove unbound virus. Replication of virus was determined by HuNoV RNA levels, which were quantified using a standard curve based on a recombinant HuNoV RNA transcript, and replication of virus was determined by assessing changes in virus GE levels at 24 hpi from baseline.
Statistics. All statistical analyses were performed on GraphPad Prism (GraphPad Software, La Jolla, California USA). Samples with RNA levels below the limit of detection of the reverse transcriptionquantitative PCR (RT-qPCR) assay were assigned a value that is one-half the limit of detection of the assay. Comparisons between infection time point groups or infected cell lines were made using two-way analysis of variance (ANOVA) with Tukey's test for post hoc analyses. P values of Ͻ0.05 were considered statistically significant.

ACKNOWLEDGMENTS
This work was supported in part by Public Health Service grants AI-057788 and P30 DK 56338 and contract HHSN2722017000381 from the National Institutes of Health, by CPRIT RP160283-Baylor College of Medicine Comprehensive Cancer Training Program, and by the John S. Dunn Research Foundation.