Binding of Human GII.4 Norovirus Virus-Like Particles to Carbohydrates of Romaine Lettuce Leaf Cell Wall Materials

Abstract

Norovirus (NoV) genogroup II genotype 4 (GII.4) strains are the dominant cause of the majority of food-borne outbreaks, including those that involve leafy greens, such as lettuce. Since human NoVs use carbohydrates of histo-blood group antigens as receptors/coreceptors, we examined the role of carbohydrates in the attachment of NoV to lettuce leaves by using virus-like particles (VLPs) of a human NoV/GII.4 strain. Immunofluorescence analysis showed that the VLPs attached to the leaf surface, especially to cut edges, stomata, and along minor veins. Binding was quantified using enzyme-linked immunosorbent assay (ELISA) performed on cell wall materials (CWM) from innermost younger leaves and outermost lamina of older leaves. The binding to CWM of older leaves was significantly (P < 0.05) higher (1.5- to 2-fold) than that to CWM of younger leaves. Disrupting the carbohydrates of CWM or porcine gastric mucin (PGM) (a carbohydrate control) using 100 mM sodium periodate (NaIO₄) significantly decreased the binding an average of 17% in younger leaves, 43% in older leaves, and 92% for PGM. In addition, lectins recognizing GalNAc, GlcNAc, and sialic acid at 100 μg/ml significantly decreased the binding an average of 41%, 33%, and 20% on CWM of older leaves but had no effect on younger leaves. Lectins recognizing α-d-Gal, α-d-Man/α-d-Glc, and α-l-Fuc showed significant inhibition on CWM of older leaves as well as that of younger leaves. All lectins, except for the lectin recognizing α-d-Gal, significantly inhibited NoV VLP binding to PGM. Collectively, our results indicate that NoV VLPs bind to lettuce CWM by utilizing multiple carbohydrate moieties. This binding may enhance virus persistence on the leaf surface and prevent effective decontamination.

INTRODUCTION

Food-borne illnesses are a recurrent problem worldwide (16), exposing the public to an increased risk of infection and causing major economic losses. In the United States alone, 9.4 million cases of food-borne illnesses occur annually (34), which are estimated to cost the economy $10 billion to $83 billion per year (32). Human enteric viruses cause the majority (59%, versus 39% and 2% by bacteria and parasites, respectively) of all food-borne illnesses in the United States (29, 34). Of these enteric viruses, noroviruses (NoVs) are considered the leading cause of food-borne illnesses in the United States, causing an estimated 5.5 million cases (58%) of these illnesses annually (2, 34).

NoVs are 28- to 35-nm-diameter, nonenveloped, single-stranded RNA viruses that are transmitted to humans mainly via the fecal-oral route. The virus has a low infectious dose (∼18 to 1,000 viral particles) and can persist for prolonged periods of time in the environment once introduced (15, 38). These characteristics facilitate virus spread through droplets, contaminated food, water, fomites, and person-to-person contact. NoVs cause outbreaks of acute viral gastroenteritis in all age groups, mainly within hospitals, cruise ships, the military, nursing homes, schools, and catered functions (45). Although the disease is usually self limiting, significant morbidity and mortality can occur among children, the elderly, and the immune compromised (34).

Food-borne outbreaks linked to consumption of fresh and ready-to-eat food products are increasingly recognized (2). Typical food items that have been implicated in norovirus outbreaks include poorly cooked meats or seafoods and ready-to-eat foods, such as fruits and vegetables (45). Many types of fruits and salad crops (e.g., grapes, raspberries, strawberries, and lettuce) are being recognized as vehicles for norovirus transmission (11, 35, 45, 46). The sources of produce contamination with NoVs vary and may include application of organic amendments, irrigation, or processing water and produce handling by infected people pre- or postharvesting (6, 45, 50). Therefore, prevention efforts focusing on good agricultural and industrial practices should enhance the control of food-borne NoV illnesses. Of particular importance are intervention strategies that focus on preventing the initial attachment of enteric pathogens to the surfaces of fruits and vegetables, which constitutes an important step in produce contamination. Many researchers have investigated the mechanisms underlying bacterial adherence to fruits and vegetables (reviewed in reference 2). They have reported that a number of factors contribute to bacterial attachment to vegetables, including the presence of pili, type III secretion systems, flagella, and the ability to form extensive biofilms. However, few studies have addressed the mechanisms by which food-borne viruses adhere to fruits and vegetables. Electrostatic forces have been suggested to be a contributing factor in attachment of surrogate viruses (echovirus 11 and feline calicivirus) to lettuce (48, 49). In addition, plant-associated factors that enhance virus attachment or entrapment include the presence of stomata and injuries on the surface of the plant (2, 50). Because attached viruses tend to be more difficult to remove by simple washes (2), understanding the mode of attachment can provide insights into suitable prevention strategies.

Despite the increased importance of human NoVs as food-borne pathogens, the inability to grow the virus in vitro has hindered research concerning virus transmission and control in food products. An alternative to using the live infectious virus would be the use of virus-like-particles (VLPs) of NoVs. These NoV VLPs are produced by the expression of the NoV capsid protein from recombinant baculoviruses in insect cells, which generate spontaneous self-assembled VLPs. NoV VLPs are morphologically similar to the native virions and contain the same viral antigenic determinants and binding sites; however, they lack nucleic acids and hence are noninfectious. These VLPs can be used as a surrogate to investigate NoV attachment to food surfaces. Therefore, the objective of this study was to evaluate the mode of attachment of NoV VLPs to lettuce leaves using extracted lettuce cell wall materials (CWM). We focused on carbohydrates of lettuce leaves for the following reasons: (i) human NoVs use carbohydrates of the histo-blood group antigens as receptors/coreceptors in humans (18), and (ii) carbohydrates constitute up to 90% of the plant cell wall dry weight (33). Additionally, we used a NoV strain belonging to genogroup II/genotype 4 (GII.4) because GII.4 strains are recognized as the predominant genotype in global food-borne NoV outbreaks (4). We used lettuce as a model salad crop for the following reasons: (i) it is consumed in large quantities in the United States with minimal processing, (ii) it has a large surface area for potential pathogen attachment, and (iii) it has been implicated in a number of food-borne NoV outbreaks (2, 12, 30).

MATERIALS AND METHODS

Production and purification of NoV VLPs and NoV antibodies.

The recombinant baculovirus (rBac) carrying the genes for the VP1 and VP2 structural proteins of the Hu/NoV/GII.4/HS66/2001/US strain (sample no. 191; GenBank accession no. EU105469) was generated by Karin Bok using the BaculoDirect baculovirus expression system (Invitrogen, Carlsbad, CA) and provided by Kim Green at NIH. The VLPs were generated using the above-described rBac in insect cells (Sf9) according to standard protocols (8). Briefly, Sf9 cells were cultured at a density of 3 × 10⁷ cells/flask (162 cm²), and rBac was inoculated at a multiple of infectivity (MOI) of 10, following which cells and media were harvested at 7 to 10 postinoculation days (PID). VLPs were pelleted through sucrose cushions and were further purified by CsCl gradient utracentrifugation in TNC buffer (10 mM Tris-HCl, 140 mM NaCl, 10 mM CaCl2, pH 7.4) containing protease inhibitors: 10 μM leupeptin, 100 μM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF), 1 μM pepstatin A, and 1 μM E-64 (Sigma-Aldrich, St. Louis, MO). The band at ∼1.34 g/cm³ was collected and washed with TNC buffer to remove CsCl. The collected VLPs were then suspended in TNC buffer. The protein concentration was quantified using the Bradford method, and the morphology of the VLPs was examined by electron microscopy. Finally, the antigenic reactivity of the VLPs was confirmed using immunoelectron microscopy and Western blotting. Hyperimmune serum against NoV VLPs was generated using guinea pigs according to an approved institutional animal care and use committee (IACUC) protocol (8). Total immunoglobulin G (12) from the hyperimmune serum was further purified using a Montage antibody purification kit (Millipore, Billerica, MA) according to the manufacturer's instructions. The protein concentrations of the purified IgG antibodies were quantified using the Bradford method, and removal of nonimmunoglobulin proteins was confirmed by using 10% SDS-PAGE.

Evaluation of NoV VLP attachment to lettuce leaves by using immunofluorescence microscopy.

Mature romaine lettuce heads (Lactuca sativa) were purchased from a local grocery store. Lettuce leaf pieces (5 by 5 mm²) were incubated with VLPs (10 μg/ml) in 100 μl of 0.01 M phosphate-buffered saline (PBS) (pH 7.4) for 1 h at room temperature (RT) under gentle shaking. The lettuce pieces were then washed 3 times, 5 min each, with PBS under shaking conditions. The pieces were fixed with 4% paraformaldehyde in PBS for 1 h at RT, followed by washing 3 times, 5 min each, and then blocking with PBS containing 10 mM NaN₃, 0.2% bovine serum albumin, and 5% normal goat serum. Several dilutions of the purified primary (anti-HS66 VLP guinea pig serum) and secondary (Alexa Fluor 488-conjugated goat anti-guinea pig IgG) antibodies (Invitrogen) were tested to determine the best working dilutions while avoiding nonspecific binding of the antibodies to the lettuce. Optimum dilutions for primary and secondary antibodies were 1:2,500 and 1:600, respectively. All incubation steps were performed at RT for 1 h. Following each antibody step, lettuce pieces were washed 5 times for 3 min each with PBS-Tween 20 (0.05%) (PBS-T) under shaking conditions. Confocal microscopy (Leica TCS-SP5; Leica Microsystems) was used to visualize the Alexa 488 green fluorescence specific for VLP binding and the plant chloroplast red autofluorescence using the 488 wavelength of the Argon excitation laser. The experiment was repeated three times, with duplicate lettuce pieces per treatment. Three controls were used, including lettuce pieces treated as above but without the VLPs or the primary or secondary antibodies.

Extraction of plant CWM.

Extraction of plant cell wall material (CWM) was performed at 4°C according to the method of Hoffman et al. (17). Mature romaine lettuce heads (n = 3) were purchased from a local grocery store. Healthy plant leaves were sorted into innermost younger (∼2 to 6 cm) leaves (“YL”) and outermost older (∼20 to 25 cm) leaves (“OL”). The leaves were surface sterilized for 10 min in 0.1% sodium hypochlorite, hand washed thoroughly with sterile water, and then allowed to dry (21). The OLs were further cut to separate the midrib, also known as principal vein (“VL”), from the green leaf lamina (“GL”) (the green part of the leaf without the midrib). Briefly, 30 g of each of these tissues was extracted by grinding them in liquid nitrogen to fine powder. The powder was then homogenized in 80% ethanol using a Polytron homogenizer (Cole-Parmer Instruments), followed by centrifugation for 20 min at 2,500 × g. The resulting pellet was subjected to a series of ethanol washes (80% twice and 100% once) and then stirred for 2 h at 4°C in 60 ml methanol-chloroform (1:1 [vol/vol]). The pellet was collected by filtration using Whatman glass microfiber filters, grade GF/A (Sigma) and washed with ice-cold methanol-choloroform and finally with ice-cold acetone. The pellet was vacuum dried overnight at RT (yield, ∼0.4 g). The resulting CWM were stored at 4°C in a tightly sealed bottle under dark conditions.

Detection of NoV VLP binding to CWM and inhibition assays using ELISA.

To develop a CWM ELISA, an approach similar to that used by Tian et al. (40) was followed with modifications. Briefly, CWM were resuspended in PBS (pH 7.4) at 5 mg/ml, vortexed for 2 min, and centrifuged twice at 500 × g for 2 min, and the resulting supernatant was used to coat MaxiSorp 96-well plates (Thermo Fisher Scientific Inc., Rockford, IL). The plates were incubated at 4°C for 18 h. Following three washes with PBS-T, blocking was performed with 200 μl of PBS supplemented with 2% nonfat dry milk (blocking buffer) for 2 h at 37°C. Following blocking and washing, 50 μl of NoV VLPs diluted in PBS (1, 5, 10, 20, and 50 μg/ml) were added, and the plates were incubated for 1 h at RT. The optimum dilutions of the purified guinea pig anti-VLP primary antibody and the secondary antibody (goat anti-guinea pig IgG horseradish peroxidase [HRP] conjugated) (KPL, Gaithersburg, MD) were determined experimentally (1:1,600 and 1:4,000, respectively). Both antibodies were diluted in blocking buffer and incubated at 37°C for 1 h. Plates were washed four times with PBS-T following each step. Plates were read 60 min later at A₆₅₀ after incubation with the peroxidase substrate tetramethylbenzidine (TMB) (KPL). The reaction was stopped with 0.3 M sulfuric acid, and the plates were reread at 450 nm. Controls included the following: (i) wells coated with CWM but no VLPs added and (ii) PBS-coated wells with VLPs added.

Inhibition treatments.

For inhibition treatments, we focused on leaf lamina of older leaves (GL) for two reasons: (i) the leaf lamina represents the largest surface area of the lettuce head, and (ii) our results indicated that most of the NoV VLP binding occurred on leaf lamina of older leaves. We used CWM of young leaves for relative comparisons. To assay for carbohydrate involvement in binding of NoV VLPs to lettuce, carbohydrates of the CWM were oxidized using freshly made sodium periodate (NaIO₄) at 10, 50, or 100 mM in PBS (pH 7.0) for 1 h at RT under dark conditions. For specific inhibition of carbohydrates, monoclonal antibodies (MAbs) against various histo-blood group antigens (HBGA) (anti-A, anti-B [Immucor, Norcross, GA] and anti-B type 2, H type 1, Le^a type 1, Le^b type 1, Le^x type 2, and Le^y type 2 [BG3 to BG8, respectively] [Covance, Emeryville, CA]) diluted in PBS (1:25) and commercially available lectins (Sigma) that bind to a range of sugars (summarized in Table 1) were incubated with CWM after the blocking step for 1 h at 37°C. Lectins were diluted in TBS buffer (50 mM Tris-HCl, 150 mM NaCl supplemented with 2.5 mM MnCl₂, and CaCl₂) to a concentration of 10 and 100 μg/ml. Control (no lectin) wells were incubated with TBS buffer under the same conditions as above. To assay for protein involvement in the binding of NoV to lettuce, the CWM suspensions were boiled at 100°C for 5 min before coating the ELISA plates. In addition, to specifically test the involvement of proteins, total cell wall proteins from lettuce tissues were isolated (as described below) and used in ELISA to assay for NoV VLP binding. Controls included wells coated with the following: (i) boiled PBS or CWM with no VLPs added and (ii) CWM with no VLPs added but with lectin/MAbs or sodium periodate added to estimate nonspecific binding. None of these treatments increased nonspecific binding of the primary or secondary antibodies used in ELISA. Purified porcine gastric mucin (PGM) type III (Sigma) was used as a positive carbohydrate control, since it was previously reported to bind to NoV VLPs through the carbohydrate moieties but not the protein backbone (41). To apply the same concentrations of periodate tested on CWM, PGM was used at 5 ng/ml, a concentration experimentally determined to generate an absorbance value equal to that of the NoV VLPs binding to CWM of GL.

Table 1.

Sugar and blood group specificities of lectins used in this study

Lectin code	Plant origin	Sugar specificitya	HBGA specificitya
BS-I	Bandeiraea simplicifolia	α-d-Gal ≫ GalNAc	B ≫ A
DBA	Dolichos biflorus	GalNAc	A (type 1)
LcH	Lens culinaris	α-d-Man/α-d-Glc	None
LEA	Lycopersicon esculentum	GlcNAc	None
MAA	Maackia amurensis	Sialic acid glycoconjugate	None
UEA-I	Ulex europaeus	α-l-Fuc	H type 2/Ley

^aAdapted from Handbook of Plant Lectins (47) and the work of Zakhour et al. (51).

Extraction of plant CWP.

Total plant cell wall proteins (CWP) were isolated at 4°C, following standard protocols (39). Briefly, 10 g of YL, OL, GL, and VL were cut into small pieces and homogenized by grinding in liquid nitrogen first and then in grinding buffer (50 mM sodium acetate, pH 5.5, 50 mM NaCl, and 30 mM ascorbic acid) containing 1× Halt protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific Inc.). The ground tissues were filtered using nylon mesh and sequentially washed with grinding buffer, 100 mM NaCl, sterile distilled water, acetone, and 10 mM sodium acetate (pH 5.5). CWPs were extracted twice using extraction buffer A (200 mM CaCl₂ in 50 mM sodium acetate, pH 5.5) for 45 min with gentle shaking. Supernatants were collected following filtration, and the remaining pellets were extracted again using buffer B (3 M LiCl in 50 mM sodium acetate, pH 5.5) for 18 h with gentle shaking. The resulting supernatants were collected following centrifugation at 1,000 × g for 10 min and then combined with the previous two supernatant fractions. CWPs were concentrated and desalted using Amicon Ultra-15 centrifugal filter units (Millipore) according to the manufacturer's directions. CWPs were quantified using the Bradford method, and the presence of proteins was confirmed by using 10% SDS-PAGE. ELISA plates were coated with CWP at 20 μg/ml in PBS. Controls included wells coated with CWP with no VLPs added and PBS with VLPs added.

Statistical analyses.

GraphPad Prism version 5 (GraphPad Software) was used for statistical analyses. One-way analysis of variance (ANOVA) followed by Tukey's post hoc test was used to determine significant differences between treatments and control within each tissue. Each treatment in each ELISA was run in quadruplicate, and the ELISA was repeated at least twice to determine assay consistency. Differences in means were considered significant when the P value was <0.05. Data were represented as means ± standard deviations.

RESULTS

NoV VLPs attached to lettuce leaves.

As confirmed by immunofluorescence (Fig. 1), NoV VLPs attached to lettuce pieces when applied to the plant surface at RT. The VLPs were observed across the whole surface area, occasionally along small veins within the leaf and as aggregates inside stomata. However, the cut edges of the lettuce leaf seemed to have the highest level of florescent signal, suggesting entrapment of the VLPs inside injured areas (Fig. 1A).

Fig 1.

Attachment of NoV VLPs to lettuce leaves. Confocal microscopy images of immunofluorescence performed on lettuce leaf pieces showing NoV VLPs attached to leaf surface and cut edges (A), (B) minor veins inside leaf (B), or as aggregates inside stomata (C) (shown under visible light in panel D). (E and F) Control non-NoV-VLP-treated lettuce pieces, showing leaf surface and cut edge (E) or minor leaf veins (F). Examples of immunofluorescence associated with NoV VLP binding are indicated by arrows.

NoV VLPs bound to lettuce CWM in a dose-dependent manner.

Binding of NoV VLPs to CWM extracted from lettuce OL and YL was dose dependent as shown by ELISAs (Fig. 2A). Increasing the VLP concentration, starting at 1 μg/ml, significantly increased the binding of VLPs to OL and YL. At all tested VLP concentrations, the binding to lettuce OL and YL CWM was significantly higher than that to PBS-coated wells. However, NoV VLP binding to OL was significantly more (1.5- 2-fold) than that to YL at all tested NoV VLP concentrations. To determine whether NoV VLPs bind preferentially to different parts of the OL, the green part of the leaf was separated from the principal vein and CWM was extracted from each. Similar to the trend observed for OL and YL, NoV VLP binding to GL was dose dependent, and at all tested concentrations the binding was significantly higher than that to PBS-coated wells. In contrast to results for GL, NoV VLP binding to VL was not dose dependent; however, the level of binding was significantly higher than that for PBS-coated wells. Interestingly, NoV VLPs bound at significantly higher levels (1.5- to 4.5-fold) to the leaf GL than to the VL CWM (Fig. 2B).

Fig 2.

Quantification, using ELISA, of NoV VLPs binding to lettuce cell wall materials. The absorbance reading at 450 nm, shown in the y axis, represents NoV VLP binding to CWM extracted from lettuce OL and YL (A) or GL and VL (B). NoV VLP binding to PBS was used as a control. At each VLP concentration, VLP binding to OL, YL, GL, and VL was significantly higher than binding to PBS-coated wells. Only for OL, YL, and GL, the binding was dose dependent: increasing the VLP concentration significantly increased NoV VLP binding (P < 0.05). Error bars represent standard deviations.

Carbohydrates were involved in NoV VLP binding to lettuce leaf CWM.

Oxidizing carbohydrate moieties of CWM of OL by sodium periodate significantly inhibited, in a dose-dependent manner, the binding of NoV VLPs (average of 38.5% ± 5.9% reduction at 100 mM NaIO₄) (Fig. 3A). However, NoV VLP binding to YL was relatively less affected by carbohydrate oxidation (average 17.3% ± 2.8% decrease at 100 mM NaIO₄), and there were no significant differences between the different sodium periodate concentrations (Fig. 3B). To assess the involvement of proteins in NoV VLP binding, the CWM was boiled prior to coating of the ELISA plates. Boiling had no effect on NoV VLP binding to CWM of OL (Fig. 3A), but it significantly decreased (25% ± 4.9%) the binding to CWM of YL (Fig. 3B). The GL and VL parts of OL were each assessed in a similar way. Boiling and sodium periodate treatments significantly decreased (19.1% ± 4.7% and 43.2% ± 9.1%, respectively) the binding of NoV VLPs to CWM of GL (Fig. 3C); however, neither caused a significant decrease in NoV VLP binding to CWM of VL (Fig. 3D). Since binding to the VL was not affected by either treatment, its removal from OL to obtain the GL part revealed the involvement of proteins in NoV VLP binding to older lettuce leaves. As expected, VLP binding to PGM was completely inhibited after periodate oxidation at all concentrations, while boiling had no effect (Fig. 3E).

Fig 3.

Effect of sodium periodate and boiling on NoV VLP binding to cell wall materials. Effect of pretreatment of OL (A), YL (B), GL (C), or VL (D) CWMs or PGM with sodium periodate (at 10, 50, or 100 mM) or boiling (100°C for 5 min) (E) on NoV VLP binding. The mean absorbance values of NoV VLP binding to each CWM preparation and to PGM were transformed to 100%, and the absorbance values of the different treatments were adjusted accordingly. Means with different letters differ significantly (P < 0.05). Error bars represent standard deviations.

NoV VLPs bound weakly to lettuce CWPs.

Although boiling treatment denatures proteins, to further investigate the role of proteins in NoV VLP binding to lettuce, total CWPs were extracted separately from the different lettuce parts and tested for VLP binding by ELISA. Interestingly, NoV VLPs starting at 1 μg/ml showed significant binding to CWP extracted from all tissues compared to results for control non-VLP-treated wells (Fig. 4). There were no significant differences in VLP binding between OL and YL and between GL and VL at each of the tested VLP concentrations. Binding was not dose dependent, since VLP binding to CWP did not increase significantly with increasing VLP concentrations, but at each tested concentration, the binding was significantly higher than that for control non-VLP-treated wells or PBS control wells.

Fig 4.

Binding of NoV VLPs to lettuce cell wall proteins. Binding of NoV VLPs (1, 5, 10, 20, or 50 μg/ml) to CWP extracted from OL and YL (A) or GL and VL (B). Binding of NoV VLPs to PBS was used as a control. At each VLP concentration, VLP binding to OL, YL, GL, and VL was significantly higher than binding to PBS-coated wells. No statistically significant differences were noted between NoV VLP binding to CWP of OL and that of YL or between that to GL and to VL (P > 0.05). Error bars represent standard deviations.

NoV VLPs bound specifically to cell wall carbohydrates of lettuce leaves.

Because sodium periodate inhibited the binding of NoV VLPs to lettuce leaf CWM and to confirm the involvement of carbohydrates in the binding, specific inhibition of carbohydrate moieties was investigated using specific MAbs against carbohydrates of human HBGA and carbohydrate binding lectins (Table 1). None of the MAbs inhibited the binding of NoV VLPs to GL or YL (data not shown). Only antibody against the A antigen significantly inhibited (29.2% ± 7.1% decrease) NoV VLP binding to PGM (data not shown).

All lectins (Table 1) at 10 μg/ml significantly inhibited the binding of NoV VLPs to GL, reducing the binding on average by ∼20% (Fig. 5A). Increasing the lectin concentration to 100 μg/ml showed a dose effect, further decreasing the binding with BS-I (42.5% ± 4.4%), DBA (41.3% ± 9.7%), LcH (51.3% ± 4.3%), LEA (33.1% ± 9.1%), and UEA-I (43.4% ± 3%), while MAA did not show any further significantly decreased binding. In contrast, none of the lectins at 10 μg/ml inhibited the binding of NoV VLPs to YL (Fig. 5B). Increasing the lectin concentration to 100 μg/ml significantly decreased the binding with BS-I (29.1% ± 2.4%), LcH (24.4% ± 3.4%), and UEA-I (14.5% ± 2.9%), while no significant inhibition was shown with DBA, LEA, and MAA. PGM was used as a positive control since it is known to contain hexosamine, hexoses, fucose, and sialic acid. At 10 μg/ml, lectin DBA, UEA-I, and MAA significantly inhibited the binding of NoV VLPs to PGM by 11.9% ± 4%, 23.3% ± 2.3%, and 8.2% ± 2.2%, respectively (Fig. 5C). At 100 μg/ml, there was an increased inhibition of NoV VLP binding by DBA and UEA-I (17.9% ± 2.3% and 31.8% ± 1.2%, respectively), while MAA inhibition did not increase further. Lectin LcH and LEA showed significant inhibition (7.7% ± 4% and 12.1% ± 2.7%, respectively) only at a higher concentration (100 μg/ml).

Fig 5.

Carbohydrate-binding lectins inhibit binding of NoV VLPs to lettuce cell wall materials. Percent binding of NoV VLPs to GL (A), YL (B), or PGM (C) and its dose-dependent inhibition using lectins (10 and 100 μg/ml) are shown. Lectin labeling along the x axis is consistent with Table 1. Significant differences in comparison to results for CWM given control treatment (None) are denoted by asterisks. Error bars represent standard deviations.

DISCUSSION

Although it is known that NoVs can attach to lettuce leaves (14, 49), the underlying factors that facilitate this interaction are not clear. In the current study, we demonstrated that a NoV-lettuce interaction is mediated by various carbohydrate moieties that are present in the plant cell walls. First, using immunofluorescence, we demonstrated that NoV VLPs can attach to the surfaces of the lettuce leaves, stomata, along minor veins, and it can be concentrated inside cut edges (Fig. 1). These results corroborate and extend findings of previous studies that used murine NoV as a surrogate for human NoV and the VLPs of NoV Norwalk strain belonging to GI.1 (14, 49). This finding is important, since damage to lettuce leaves is common under field conditions (3) and shredded lettuce is now being widely used in packaged salads (10). Viral particles attached to damaged surfaces or open cuts can be refractory to removal by washing and sanitization, allowing their persistence and transmission to consumers.

Second, we showed for the first time that VLPs of human NoV GII.4, the predominant NoV genotype globally, not only attach to surfaces and damaged areas but also specifically bind to lettuce plant leaves mainly through the carbohydrates of the cell wall (Fig. 5). Furthermore, minor binding to cell wall proteins was also observed (Fig. 4). A recent study showed that NoV GI.1 could bind to unknown proteinaceous components released from the surface of lettuce (14). However, since the authors prepared the lettuce extract by crushing the leaves with a cell spreader and collecting the supernatant after high-speed centrifugation, it is possible that their method released some surface or intracellular protein(s) or other proteins of bacteria that were attached to the surface. In our study, we specifically isolated the cell wall materials from lettuce leaves using an established method that is commonly used in elucidating the structure of cell wall polysaccharides (17, 21). This method selects for carbohydrates, because proteins have poor solubility in ethanol; however, some intracellular and CWP molecules that are covalently linked to carbohydrates are coextracted (13). Therefore, to assay for NoV VLP binding to CWP, we used an alternative method that specifically selects for CWP. The latter method uses salt in the grinding buffer, which prevents loose association of cytoplasmic proteins to the cell walls (39). Efforts to enhance binding of NoV VLPs to CWP by increasing the CWP coating concentration (data not shown) or VLP concentrations (Fig. 4) did not increase the binding, suggesting that NoV VLPs bind weakly or nonspecifically to the isolated lettuce CWP. Consequently, and since the plant cell wall is primarily composed of polysaccharides (up to 90% of dry weight), with a lesser amount of proteins (2 to 10%) (33), and GII.4 NoVs bind to carbohydrates of HBGAs (18), we focused on elucidating the binding of NoV VLPs to carbohydrates.

The binding of NoV VLPs to lettuce CWM was inhibited by periodate oxidation (Fig. 3), providing the first evidence of the involvement of carbohydrates in the binding, since periodate cleaves the C—C bond in the ring structure of polysaccharides (23, 37). Further evidence was demonstrated by specific inhibition of NoV VLPs binding to lettuce leaf CWM using six lectins whose carbohydrate and HBGA specificities are known (Table 1). Our results showed that multiple carbohydrate moieties, mainly α-d-Gal, GalNAc, Man/Glc, Fuc, GlcNAc, and sialic acid, could be involved in the binding of NoV VLPs to lettuce older leaf CWM (Fig. 5A). We used the glycoprotein PGM as a positive control for the following reasons: (i) it was previously reported to bind to NoV VLPs through the carbohydrates moieties but not the protein backbone (41); (ii) the chemical composition of mucin is known to contain approximately 37% hexosamine (GalNAc and GlcNAc), 27% total hexoses (galactose and mannose), 20% protein, 10% fucose, and 6% sialic acid (41, 44). We have also quantified the HBGA content of purified PGM type III using ELISA and found that it reacts with BG1 (anti-A), BG4 (anti-H type 1), BG6 (anti-Le^b), and BG8 (anti-Le^y) (data not shown). These results (with the exception of Le^y) are consistent with the HBGA content of crude extract of PGM type II (43); however, the presence of HBGA Le^y in the PGM used in this study could be due to the different types of mucin used, since not all mucins contain the same types of HBGA (7). Previous investigation found that GI.1 NoV binding to PGM type II could be inhibited with MAb against A, H type 1, and Le^b antigens (42). In our study, only BG1 (anti-A) but not BG4 (anti-H type 1), BG6 (anti-Le^b), or BG8 (anti-Le^y) inhibited binding of GII.4 NoV VLPs to PGM. In addition, of the HBGA-recognizing lectins, only DBA (anti-A) and UEA-I (anti H type 2/Le^y) but not BS-I (anti-B) inhibited the binding. The results suggest that GII.4 NoV VLPs bind to PGM type III through the A and H type 2 antigens but not H type 1 and Le^b, as was previously reported for GI.1 NoVs and PGM type II (42). Collectively, we showed that multiple carbohydrate moieties, namely, Fuc, GalNAc, Man/Glc, GlcNAc, and sialic acid, but not α-d-Gal, are involved in binding of GII.4 NoV VLPs to PGM type III (Fig. 5C). This finding is important because it demonstrated that our GII.4 NoV VLP strain can bind to multiple monosaccharide moieties; thus, it is not surprising that it utilized multiple monosaccharide moieties in its binding to lettuce CWM. Since similar carbohydrate moieties were utilized in binding of NoV VLPs to PGM, as in the case of lettuce CWM (in addition to α-d-Gal), this suggested that CWM share similar carbohydrate moieties with the carbohydrate component of PGM. In competition assays, preincubation of PGM with the NoV VLPs significantly reduced the binding of the VLPs to CWM of lettuce, while preincubating VLPs with CWM did not significantly decrease the VLP binding to PGM (data not shown). This further suggested that CWM share similar carbohydrate moieties with the carbohydrate component of PGM and that the affinity of VLPs for commercially purified PGM is much higher than that for crude extract of lettuce CWM. Our results indicate that the VLPs of the human NoV GII.4/HS66 strain bind to lettuce cell wall material mainly through the carbohydrates.

In humans and gnotobiotic pigs, it has been shown that NoVs bind to carbohydrates of HBGAs, which serve as an attachment factor (7, 18, 27). HGBAs are carbohydrates containing structurally related oligosaccharides. Human NoV strains belonging to the GII.4 genotype account for 85.8% of the global NoV food-borne outbreaks (4) and bind to more HBGAs than most other genotypes, contributing to their worldwide transmission (36). In our study, although three of the lectins (DBA, BS-I, and UEA-I) that inhibited NoV VLP binding to CWM are known to recognize blood group A type 1, B, and H type 2/Le^y, respectively, MAbs against these HBGAs (BG1, BG2, and BG8) were unable to inhibit the binding of NoV to CWM. In contrast to MAbs, which recognize specific epitopes, lectins recognize not only the terminal monosaccharide residues but also more complex internal sugar moieties (28), suggesting that the backbone carrying the terminal carbohydrate has a role in the binding of GII.4 NoV VLPs to lettuce CWM. This idea is supported by a previous study showing that in contrast to binding of MAbs to HBGA, which requires only the terminal carbohydrate moieties, binding of NoV to HBGA requires both the backbone and carbohydrate moieties (19). Recently, GII.4 NoVs have been found to bind to the digestive tracts of oysters through carbohydrate structures that are similar to the A-type HBGA (25, 40). The specific binding allows the virus to concentrate inside the digestive tract and resist removal through depuration, which might explain why NoV outbreaks frequently occur in shellfish (9, 26). A similar situation could potentially occur on cut edges of shredded or damaged lettuce leaves, since NoV VLPs not only adsorb or entrap inside open cuts but also could bind specifically to exposed cell wall carbohydrates, as we showed here. The specific binding would allow the accumulation of the viral particles inside open cuts and hence resist removal by simple washing.

Our data indicated that binding of human NoV VLPs to old lettuce leaf CWM was significantly higher (1.5- to 2-fold) than that to young leaf CWM at room temperature. Furthermore, most of the binding of NoV VLPs on older leaves occurred as a result of the CWM carbohydrates of the leaf lamina, not the main vein. Since leaf lamina represents the majority of the lettuce head surface area, we focused on leaf lamina of older leaves and used young leaf CWM for relative comparisons. Periodate oxidation had significantly higher inhibitory effects on binding of old versus young (43.2% ± 9.1% versus 17.3% ± 2.8%, respectively) leaf CWM to NoV VLPs (Fig. 3), suggesting that carbohydrates play a larger role in the binding of NoV VLPs to old leaves than in that to young leaves. This result was further supported by differences in lectin inhibitory effects on VLP binding to CWM extracted from young versus old leaves. For example, three of the six lectins did not show any inhibition of the binding at any concentration, suggesting that these sugar moieties (GalNAc, GlcNAc, and sialic acid) are not involved in the binding of NoV VLPs to young leaf CWM. Second, only at a higher lectin dose (100 μg/ml), three lectins inhibited binding, suggesting that Gal, Man/Glc, and Fuc sugar moieties were involved in the binding of NoV VLPs to young leaf CWM. We demonstrated that this GII.4 NoV strain has the capacity to bind to multiple sugar moieties by using PGM as a positive control. Therefore, the difference in sugars utilized in the binding of NoV VLPs to young versus old leaf CWM could be the result of different sugar abundances or different sugar composition. This idea is supported by findings of previous studies showing that the plant cell wall changes continuously in response to various developmental stages (5, 13, 33). Specifically, the cell wall polysaccharides are known to vary in quantity and structure between different organs and different morphological parts, such as roots, stems (1, 5, 31), and early versus late emerging leaves (20). Therefore, it is possible that the CWM of these contrasting leaves have different sugars or different abundances of the same sugar, thus allowing NoV VLPs to bind to the different sugar profiles. No studies have examined the attachment of enteric viruses on leaves of different ages; however, studies that have examined the growth of food-borne pathogenic bacteria on lettuce have also shown that growth of these bacteria was higher on old leaves than on young leaves (22, 24). Although our results cannot be directly compared to those in these studies, the difference in growth of these bacteria could reflect leaf-age-specific factors leading to differences in the initial attachment and subsequent differences in growth. Collectively, we demonstrated that NoV VLPs bind at significantly higher levels to old lettuce leaf CWM than to young lettuce leaf CWM. In addition, our results suggest that NoV VLPs utilize different sugar profiles on old leaf versus young leaf CWM. Although young leaves are enclosed within the normal developing lettuce head under field conditions, which would protect them from contact with the virus, during industrial processing these leaves become exposed. Based on our findings, older leaves may be associated with a greater risk of contamination with NoVs than younger leaves, and their removal prior to sale may result in increased safety of the marketed lettuce.

In conclusion, we have demonstrated, for the first time, the ability of human NoV GII.4 VLPs to attach to the lettuce leaf surface and damaged cut edges and to specifically bind, mainly through the carbohydrates, to the cell wall materials isolated from lettuce leaves. Given the low infectious dose of the virus, this specific binding may enhance the virus persistence and increase the risk for human exposure. We have also shown that binding differs between different leaf ages and different regions within the leaf (lamina versus midrib). Since cuts (as in shredded lettuce or field damage) breach the cell wall integrity, exposing more carbohydrates, i.e., binding sites, our results emphasize the need for preventive efforts that reduce initial contact of virus with lettuce and require improved washing and sanitization strategies.