INTRODUCTION
Receptor recognition by viruses is the first and essential step of viral infection. Understanding this process can provide critical insight into viral biology and, ultimately, the development of effective vaccines and antivirals.
Coronaviruses (CoVs), enveloped RNA viruses, infect a wide variety of species, causing various clinical conditions. The tissue and species specificities are determined by the presence of appropriate adhesion and entry receptors on the cell surface. Some alphacoronaviruses, e.g., transmissible gastroenteritis coronavirus (TGEV) and feline infectious peritonitis coronavirus (FIPV), employ aminopeptidase N, whereas human coronavirus NL63 (HCoV-NL63) utilizes angiotensin-converting enzyme 2 (ACE2) (
1–3). At the same time, ACE2 serves as a receptor for severe acute respiratory syndrome coronavirus (SARS-CoV), a betacoronavirus (
4). A number of receptors have been described for betacoronaviruses, including carcinoembryonic antigen-related cell adhesion molecule 1 for murine hepatitis virus (MHV), dipeptidyl peptidase 4 for Middle East respiratory syndrome coronavirus (MERS-CoV), and major histocompatibility complex class I for human coronavirus OC43 (
5–9). Furthermore, several alpha-, beta-, and gammacoronaviruses, e.g., TGEV, human coronavirus HKU1, HCoV-OC43, bovine respiratory coronavirus (BCoV), and avian infectious bronchitis virus (IBV), reportedly use sialic acids for initial attachment to the cell (summarized in reference
10).
The coronaviral particle consists of a dense core formed by a nucleocapsid (N) protein with viral genomic RNA and an envelope decorated with the membrane (M), envelope (E), and spike (S) proteins. Some coronaviruses contain other structural proteins, such as hemagglutinin esterase (HE) or accessory open reading frame (ORF) proteins (ORF3, ORF4a, and ORF7) (
59,
69). The S protein is a class I viral membrane glycoprotein responsible for the interaction with the entry receptor and fusion (
11).
HCoV-NL63 employs ACE2 as an entry receptor (
3). However, we recently reported that heparan sulfate (HS) proteoglycans (HSPGs) are required for effective adhesion of the virus to the cell surface and that such an interaction enhances the infection process (
12). HSPG binding was also demonstrated for MHV (
13), SARS-CoV (
14), and porcine epidemic diarrhea virus (PEDV) (
15).
HSPGs are glycoproteins that are ubiquitous at the surface of the mammalian cell. Binding to HSPG is the initial event promoting subsequent recognition of a secondary receptor by increasing the local concentration of pathogens or triggering conformational changes of proteins involved in viral entry. As an example, binding to HSPG induces structural rearrangements of proteins responsible for infection by adeno-associated virus 2 (
16), adenovirus types 2 and 5 (
17), human papillomavirus 16 (
18), and several herpesviruses (
19). Furthermore, the majority of oncogenic viruses (hepatitis B and C viruses, Kaposi’s sarcoma-associated herpesvirus, human papillomaviruses, Merkel cell polyomavirus, and human T cell lymphotropic virus type 1) initially attach to the HSPGs (
20). HSPGs can also enhance virulence by binding accessory viral factors necessary for viral replication. This is illustrated by HSPG binding of the Tat protein of human immunodeficiency virus, which after internalization activates transcription of the viral RNA (
21).
To better address the nature of the virus-HSPG interaction, we constructed viruslike particles (VLPs) composed of HCoV-NL63 proteins. VLPs structurally mimic the native virus and thus constitute a good model for studying virus-host interactions in the context of additional capsid components. Importantly, VLPs can be relatively easily tailored by molecular biotechnology techniques, facilitating the assessment of the role of individual viral proteins in receptor recognition.
In the present study, we demonstrate that HCoV-NL63 binds HSPG via the M protein, which is to some extent responsible for virus attachment. We thus show that viral entry is an outcome of the concerted action of the M and S proteins. The presented findings improve the understanding of viral biology and may facilitate the development of improved antivirals and neutralizing vaccines.
DISCUSSION
Cell infection is a complex process that involves viral attachment, which leads to viral enrichment on the cell surface and subsequent internalization. While initial binding is usually nonspecific and mediated by ubiquitous molecules, such as sugars or glycoconjugates (glycoproteins, glycolipids, and proteoglycans), the second step requires a highly specific interaction with the entry receptors, often involving their proteolytic processing. In some viruses, a single protein is responsible for both steps of this interaction, whereas in other viruses (i.e., paramyxoviruses), different structural elements of the virion are employed for the attachment to and fusion with the cellular membrane (
26). This sophisticated interplay of distinct receptor entities and their sequential engagement increase viral avidity and coordination in time for key events for efficient cellular uptake.
For coronaviruses, it is generally believed that S protein is responsible for both virion attachment and internalization (reviewed in references
27 and
28). This large (∼150-kDa) glycoprotein consists of three segments: an ectodomain, a transmembrane anchor, and a short endodomain. The ectodomain can be further divided into S1 and S2 subunits, although not all coronaviruses undergo enzymatic cleavage of the S ectodomain (
29). Different experimental approaches demonstrating that the S1 domain encompasses one or more receptor binding sites (RBSs), located at its C terminus, have been reported for almost all studied CoVs. While the C-terminal part of the S1 domain (S1-CTD) is highly divergent and responsible for the interaction with entry receptors, the more conserved N-terminal region of S1 (S1-NTD) is thought to function as a ligand for the initial attachment factors. Indeed, it was demonstrated that the S1-NTDs of several coronaviruses bind carbohydrates. Examples include the alphacoronavirus (TGEV), betacoronavirus (HCoV-HKU1 and BCoV), and gammacoronavirus (IBV) genera (
30–33). In contrast, the S2 domain presumably participates in the fusion of the cellular membrane with the viral envelope (
34). This was confirmed for a number of coronaviruses, including HCoV-229E (
35), SARS-CoV (
36), MHV (
37), and HCoV-NL63 (
38).
The structure of the HCoV-NL63 S protein and its role in viral infection were previously described (
3,
39–42). The receptor binding domain (RBD) within the S1 subunit and the heptad repeat region within the S2 subunit were mapped (
38). Some authors speculate that S1 also contains a domain responsible for binding to the attachment receptors on the cell surface, which triggers binding with ACE2 (
27,
41). Of note, it has been shown that purified full-length S protein of HCoV-NL63 exhibits surprisingly low-affinity binding to its putative receptor (
43). At the same time, there is evidence that the RBD of the HCoV-NL63 S protein binds ACE2 with a higher affinity than its full-length counterpart and with efficiency comparable to that of S1 of SARS-CoV (
40,
41,
44). These observations raise questions about the existence of an additional, S-independent stimulus prerequisite for this interaction.
In the present study, we provide evidence that the M protein may be responsible for the interaction of the virus with cellular HSPG. For this, we have used previously developed VLPs, produced in insect cells, which enable efficient and scalable expression of complex macromolecular structures. First, using confocal microscopy, we showed that adhesion to the cell surface was not abrogated in VLPs missing the S protein. Moreover, blocking the VLP-ACE2 interaction with specific antibodies did not affect adhesion. Next, using flow cytometry, we confirmed that adhesion is mediated by HSPG, since soluble HS blocks the interaction. To identify the ligand involved in this interaction, we decided to first screen the M protein for the presence of putative heparin binding domains. Using a simple peptide array overlay, we identified three regions that may potentially be responsible for such an interaction. The experimental data were consistent with the notion that these regions are rich in positively charged amino acids, lysine, and arginine (
45,
46). Subsequently, we verified this result using the predicted topology of the M protein and literature data. Only one site was predicted to localize on the virion surface, and consequently, we assumed that this particular site is responsible for the interaction. This assumption was then proven by an
in situ ELISA, which demonstrated that the region from aa 153 to 226 of the M protein binds to cellular HSPG.
Of note, topology prediction algorithms detected four transmembrane domains (TMDs) within the M protein, which is somewhat atypical for coronaviruses (
Fig. 6). The C terminus of the M protein is essential for the M-M, M-N, and M-S interactions, and it is thought to be hidden within the virion envelope (
6,
47–51). However, the exact region of the M protein responsible for these interactions has not been defined for most coronaviruses. Recently, it was suggested that this interaction might rely on structural motifs consisting of several residues dispersed throughout the protein (
52). Furthermore, we have previously demonstrated that a C-terminally tagged M protein is assembly competent for VLP formation, which supports the four-TMD model of the M protein and the resulting N
exo-C
exo topology. This conclusion was validated by the observation that HCoV-NL63 and NL63 VLP adhesion is hampered by an antibody raised against two peptides corresponding to the distal part of the M protein. Interestingly, this observation was consistent with works of Enjuanes and colleagues (
53,
54), who demonstrated a similar M protein topology of TGEV CoV belonging to the same genus (alphacoronavirus) as HCoV-NL63. Hence, we propose that such a unique organization of the NL63-HCoV M protein might be one explanation for its engagement in HCoV-NL63 infection.
It has to be mentioned that some coronaviruses acquire the ability to bind HSPG in the course of cell culture adaptation, as described for MHV and IBV (
13,
55–57). To exclude this possibility, we compared the M protein sequence used in the present study (Amsterdam I strain) with those of clinical isolates, and no differences in this region were identified (not shown). We believe that the engagement of the M protein in cell attachment is hence a natural characteristic of HCoV-NL63.
The involvement of proteins other than the S protein in carbohydrate binding by coronaviruses has been previously described. For instance, HE of an MHV strain binds cellular sialic acids (
58). HE also appears to mediate the attachment of BCoV and HCoV-OC43 to the cell surface (
33,
59,
60). More generally, the involvement of multiple capsid proteins was described for a number of viruses, i.e., influenza virus (paramyxoviruses) or herpesviruses (reviewed in references
61 and
62). It is also worth mentioning that the presented results do not preclude the participation of S protein in the recognition of cellular HSPG during HCoV-NL63 infection.
The observations presented here provide new insight into the understanding of cell receptors and their interplay with the viral ligands during HCoV-NL63 infection. Considering the role of the M protein, one could suggest novel strategies for inhibiting or preventing infection by this and potentially other coronaviruses. Specifically, some experimental anticoronaviral vaccines that induce anti-S protein humoral responses cause serious adverse effects, probably related to an antibody-dependent enhancement mechanism (
63–65). In this context, vaccines, immunotherapeutics, or antivirals developed to inhibit the interaction between the M protein and cellular HSPG may offer an interesting alternative.
MATERIALS AND METHODS
Cell lines.
Sf9 (Spodoptera frugiperda; ATCC CRL-1711) and HF (High Five) (Trichoplusia ni; ATCC CRL-7701) cells were cultured in ESF medium (Expression Systems, CA, USA) supplemented with 2% FBS (fetal bovine serum; Thermo Fisher Scientific, Poland), 100 μg/ml streptomycin, 100 IU/ml penicillin, 10 μg/ml gentamicin, and 0.25 μg/ml amphotericin B. The culture was maintained in a humidified incubator at 27°C. Sf9 cells were used for baculovirus (BV) generation and amplification, while HF cells were used for recombinant protein expression.
LLC-Mk2 cells (Macaca mulatta kidney epithelial cells; ATCC CCL-7) were maintained in minimal essential medium (MEM) (2 parts Hanks’ MEM and 1 part Earle’s MEM; Thermo Fisher Scientific, Poland) supplemented with 3% FBS, 100 μg/ml streptomycin, 100 IU/ml penicillin, and 5 μg/ml ciprofloxacin. The culture was maintained at 37°C under 5% CO2.
VLP production.
VLPs composed of membrane (M), envelope (E), and spike (S) proteins of HCoV-NL63 were produced as described previously (
22), but nucleocapsid (N) protein was additionally included for this work. For this, the N gene was subcloned from pET Duet (
23) to pFastBac Dual, under the control of the polyhedrin promoter. Recombinant baculoviruses (rBVs) were generated using the Bac-to-Bac system (Thermo Fisher Scientific, Poland) and titrated using a plaque assay. Subsequently, HF cells were coinfected with (M+E) bicistronic rBV, N monocistronic rBV, and S monocistronic rBV at a multiplicity of infection (MOI) of 4 and cultured for 72 h. Secreted VLPs were then harvested by centrifugation (5,000 ×
g for 30 min).
For purification of VLPs, the harvested culture medium was diluted (1:1) with binding buffer (20 mM K2HPO4-KH2PO4 [pH 6.2], 70 mM NaCl) and loaded onto a 5-ml heparin HT column (GE Healthcare, Poland) using an Äkta fast-performance liquid chromatography (FPLC) system (Äkta, Sweden). Before purification, the column was equilibrated with binding buffer. Proteins were eluted with a linear NaCl gradient (50 mM to 2 M NaCl in binding buffer), and collected peak fractions were analyzed using SDS-PAGE and Western blotting.
SDS-PAGE and Western blotting.
Insect cells or culture media were harvested, resuspended in denaturing buffer to final concentrations of 1.5% SDS and 2.5% β-mercaptoethanol, and resolved by Laemmli SDS-PAGE using 4-to-20% gradient precast gels (Bio-Rad, Poland). A PageRuler Prestained Plus protein ladder (Thermo Fisher Scientific, Poland) was used in this study as a protein size marker. Gels were stained with Coomassie brilliant blue or subjected to electrotransfer in buffer containing 25 mM Tris, 192 mM glycine, and 20% methanol onto an activated polyvinylidene difluoride (PVDF) membrane. Following transfer, the membrane was blocked with 5% skim milk in Tris-buffered saline supplemented with 0.05% Tween 20 (TBS-T), followed by 1.5 h of incubation with rabbit polyclonal anti-M serum (1:15,000; kindly provided by Lia van der Hoek and generated by rabbit immunization with peptides spanning aa 180 to 195 and aa 212 to 226 of the M protein), mouse monoclonal anti-N antibody (1:1,000; Ingenansa, Spain), and mouse polyclonal anti-S serum (1:250; Eurogentec, Belgium) and 1 h of incubation with anti-rabbit (1:20,000; Dako, Denmark) and anti-mouse (1:20,000; Dako, Denmark) secondary antibodies conjugated with horseradish peroxidase (HRP), respectively. The signal was developed using the Immobilon Western chemiluminescent HRP substrate (Millipore, Poland) and visualized by exposing the membrane to an X-ray film (Thermo Fisher Scientific, Poland).
Confocal microscopy.
For assessment of protein colocalization in insect cells, HF cells were grown in 6-well culture plates on glass coverslips coated with 0.01% poly-l-ornithine (Sigma-Aldrich, Poland). Cells were infected with rBVs at an MOI of 1, fixed at 48 h postinfection with 4% formaldehyde, permeabilized with 0.2% Triton X-100 in phosphate-buffered saline (PBS), and blocked for 1 h, at room temperature with 5% bovine serum albumin (BSA) in PBS. Expression of M and N proteins was detected with rabbit polyclonal anti-M serum (the same as described above; 1:1,000) and mouse monoclonal anti-N antibody (the same as described above; 1:2,000), respectively, followed by detection with Alexa fluorophore secondary antibodies at a 1:400 dilution (Santa Cruz Biotechnology, USA). Cell nuclei were stained with DAPI (4′,6′-diamidino-2-phenylindole) (0.1 g/ml in PBS; Sigma-Aldrich, Poland). Coverslips were mounted on glass slides with Prolong diamond antifade mountant (Sigma-Aldrich, Poland).
For transduction and adhesion analyses of VLPs, LLC-Mk2 cells were grown to 80% confluence for 48 h in 6-well culture plates on glass coverslips. Cells were then washed with ice-cold PBS and inoculated with 1 ml of purified VLPs. Next, LLC-Mk2 cells were incubated for 2 h at 4°C and subsequently for 90 min at 32°C under 5% CO2 and further washed three times with PBS. Subsequently, cells were fixed, permeabilized, and stained with anti-M polyclonal serum, as described above. Additionally, actin filaments were visualized with phalloidin conjugated with Alexa 647 (0.132 μM; Sigma-Aldrich, Poland).
To verify the role of the ACE2 protein during VLP entry, LLC-Mk2 cells were incubated with anti-ACE2 polyclonal antibodies (catalog number AF933; R&D Systems) or an appropriate isotype control antibody (catalog number GTX35039; GeneTex) for 1 h at 37°C (5 μg/ml). The anti-ACE2 antibody concentration was determined based on HCoV-NL63 neutralization experiments in LLC-Mk2 cells (not shown). Furthermore, cells were overlaid with purified MEN or MENS VLPs and incubated for 2 h at 4°C and subsequently for 90 min at 32°C under 5% CO2 in the presence of antibodies. Next, cells were washed three times with PBS, fixed, permeabilized, and stained with anti-M polyclonal serum, as described above.
Fluorescent images were acquired using a Zeiss LSM 710 confocal microscope (Carl Zeiss Microscopy GmbH), deconvolved using AutoQuant X3 software, and processed in ImageJ Fiji (National Institutes of Health, Bethesda, MD, USA) (
66).
The number of particles and number of cells were quantified using the built-in ImageJ Fiji tool “3D Objects Counter.” The numbers of internalized particles were counted manually from orthogonal views. For this study, the actin cortex was assumed to indicate the cell surface.
Electron microscopy.
Samples were prepared as described previously (
22). Briefly, purified VLPs were fixed in Karnovsky solution and loaded onto copper grids coated with a support film (Formvar 15/95E; Sigma-Aldrich, St. Louis, MO, USA). After drying, the material was stained with uranyl acetate (Polyscience, Inc., Warrington, PA, USA) and lead citrate (Sigma-Aldrich, St. Louis, MO, USA). Subsequently, the grids were washed with water and dried in air at room temperature. Ultrastructural observations were performed by using a Hitachi H500 transmission electron microscope at an accelerating voltage of 75 kV.
Flow cytometry.
To evaluate adhesion of VLPs to LLC-Mk2 cells, cells were grown for 48 h to reach 100% confluence in 6-well culture plates on glass coverslips. Cells were washed twice with PBS and incubated with purified VLPs, iodixanol-concentrated HCoV-NL63 (
12,
67), mock supplemented with heparan sulfate (HS) (1 mg/ml; Sigma-Aldrich, Poland), or PBS for 4 h at 4°C. The cells were then washed three times with PBS, fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 in PBS, and incubated overnight in 5% bovine serum albumin and 0.5% Tween 20 in PBS. To examine HCoV-NL63 or VLP adhesion, cells were incubated for 2 h at room temperature with a mouse monoclonal anti-N antibody (the same as described above; 1:1,000 in 2.5% BSA with 0.5% Tween 20 in PBS), followed by 1 h of incubation with an Alexa Fluor 488-labeled goat anti-mouse antibody (1:400). Cells were then washed, mechanically detached from the glass coverslips, resuspended in PBS, and analyzed by flow cytometry (FACSCalibur; Becton, Dickinson). Data were processed using CellQuest software (Becton, Dickinson) and FlowJo V10.
Expression of the C-terminal domain of the M protein.
The region coding for the C-terminal fragment (aa 153 to 226) of the M protein was PCR amplified (5′ primer CCA gga tcc gGA TGG CCA TAA GAT TGC TAC TCG TG and 3′ primer GCA ctc gag TTA GAT TAA ATG AAG CAA CTT), digested with BamHI and XhoI enzymes (Thermo Fisher Scientific, Poland), gel purified, and cloned into the pET Duet plasmid in a manner to include the 6×His tag in frame at the N terminus. The sequence (6×His-M153–226) was verified by sequencing. The Escherichia coli BL21 strain was transformed with the recombinant plasmid and precultured overnight at 37°C. LB medium (1 liter; BioShop-LabEmpire, Poland) was inoculated with the preculture, induced with isopropyl-β-d-thiogalactopyranoside (IPTG) (0.5 mM; BioShop-LabEmpire, Poland) at an optical density of 0.6, and harvested at 4 h postinduction. Bacterial cell pellets were subsequently resuspended in 50 ml of lysis buffer (50 mM Tris [pH 7.5], 5 mM urea, 250 mM NaCl, 5 mM dithiothreitol [DTT], 1 mM EDTA, 1% Triton X-100) and subjected to 2 cycles of Cell Disruptor (Constant Systems, UK) operation at 25 lb/in2. Lysates were then centrifuged (40 min at 5,000 × g), and the supernatant was diluted (1:1) with immobilized-metal affinity chromatography (IMAC) binding buffer (20 mM K2HPO4-KH2PO4 [pH 7.4], 500 mM NaCl, 20 mM imidazole). Diluted supernatants were loaded onto a 1-ml IMAC column (GE Healthcare, Poland) charged with Ni2+ and connected to an Äkta FPLC system (Äkta, Sweden) and preequilibrated with binding buffer. Proteins were eluted with 500 mM imidazole in IMAC binding buffer. Collected peak fractions were pooled and fractionated again to remove imidazole and excess NaCl. For this purpose, a 26/10 desalting column (GE Healthcare, Poland) (preequilibrated with 20 mM Na2HPO4-NaH2PO4 [pH 7.7], 250 mM NaCl, and 5% glycerol) was used. Purified 6×His-M153–226 protein was analyzed by SDS-PAGE and Western blotting.
In situ ELISA.
LLC-Mk2 cells were seeded into a 96-well plate 48 h prior to the assay and incubated at 37°C under 5% CO2. Purified 6×His-M153–226 protein (initial concentration of 0.15 mg/ml) was 2-fold serially diluted with PBS and transferred to a plate with confluent LLC-Mk2 cells. As a control, 6×His-N HKU1 was prepared in the same manner and added to the cells at the same molar ratio. After 2 h of incubation at 32°C under 5% CO2, the culture plate was inverted to remove unbound material, and cells were washed 4 times with PBS. Next, unspecific binding sites on the plate were blocked with 5% bovine serum albumin in PBS for 1 h, and subsequently, samples were incubated for 90 min with anti-His antibody conjugated with HRP (1:10,000; Sigma, Poland). The signal was visualized with the 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (OptiEIA; Becton, Dickinson, USA), and the reaction was stopped with 1 M HCl. The absorbance was measured at a wavelength (λ) of 450 nm using a Tecan Infinite 200 Pro microplate reader. All measurements were performed in triplicate, and the background signal (from control wells) was subtracted. Another in situ ELISA was performed to evaluate 6×His-M153–226 protein binding to LLC-Mk2 cells in the presence or absence of soluble HS. Briefly, 6×His-M153–226 protein was preincubated for 30 min at room temperature with HS (1 mg/ml in PBS) and transferred to a plate with confluent LLC-Mk2 cells. As a control, 6×His-M153–226 (diluted in the same manner) was added to cells. After 2 h of incubation at 32°C under 5% CO2, the culture plate was processed as described above.
Peptide array.
A cellulose membrane displaying 54 immobilized peptides covering the whole membrane protein (15 amino acids each, with a sliding window with a 4-aa step) was designed and purchased from JPT Peptides GmbH (Germany). The cellulose membrane was activated in methanol for 5 min, washed three times with TBS-T, and blocked for 2 h with 2.5% skimmed milk (in TBS-T). Next, the membrane was incubated for 3 h with 10 ml of heparin conjugated with biotin (10 μg/ml in PBS; Sigma-Aldrich, Poland), washed 4 times in TBS-T, and incubated for 2 h with 5 ml of streptavidin-HRP (2 μg/ml in PBS; Sigma-Aldrich, Poland). Following extensive washing (4 times) with TBS-T, the signal was developed using the Immobilon Western chemiluminescent HRP substrate (Millipore, Poland) and visualized by exposing the membrane to an X-ray film (Thermo Fisher Scientific, Poland).
Topology prediction.
The TMHMM server, v. 2.0, for prediction of transmembrane helices in proteins (
68) was used to evaluate M protein topology.
Pseudoneutralization assay.
To verify if anti-M antibody inhibits binding to target cells, purified MEN and MENS VLPs and iodixanol-concentrated HCoV-NL63 were incubated for 1 h at room temperature with an equal volume of rabbit polyclonal anti-M serum at 1:10 or PBS. Next, samples were added to LLC-Mk2 cells seeded on glass coverslips 48 h earlier and grown to 80% confluence and washed with PBS directly before the experiment. After 4 h of incubation at 4°C, cells were washed three times with PBS, fixed, and stained with anti-N antibody for confocal microscopy analysis, as described above. As a control, an identical experiment was conducted with preimmune rabbit serum. Additionally, the effect of anti-M serum on HCoV-OC43 adhesion to HRT-18G cells was examined in a corresponding experiment.
The number of particles and number of cells were quantified using the ImageJ Fiji 3D Objects Counter tool.
Statistical analysis.
All data were tested for normality using a Shapiro-Wilk test and a Mann-Whitney test (for pseudoneutralization assays), a Kruskal-Wallis test with Dunn’s multiple-comparison test (for the anti-ACE2 antibody effect on VLP adhesion to cells), or one-way analysis of variance (ANOVA) with a Dunnett multiple-comparison test (for the HS effect on VLPs adhesion to cells). P values of <0.05 were considered significant. All graphs were prepared using GraphPad Prism 7.