Chondroitin Sulfate Proteoglycans Are De Facto Cellular Receptors for Human Papillomavirus 16 under High Serum Conditions

ABSTRACT

Human papillomaviruses (HPVs) are nonenveloped double-stranded DNA viruses that utilize heparan sulfate proteoglycans (HSPGs) as initial attachment factors prior to cell entry and infection. While extensively characterized, the selective interaction between HPV and HSPGs is generally studied using standard in vitro conditions, which fail to account for the effects that media additives, such as fetal bovine serum (FBS), can have on viral binding. As environmental conditions and growth factors associated with wound healing are thought to play a role in natural HPV infection, we sought to investigate the effects that serum or platelet extracts could have on the binding and infectivity of HPV. Here, we demonstrate that high concentrations of FBS and human serum greatly inhibit HPV16 binding, and that for FBS, this effect results from the obstruction of cell surface HSPGs by serum-derived heparin-binding proteins (HBPs). Surprisingly, we found that under these conditions, HPV particles utilize 6O-sulfated chondroitin sulfate proteoglycans (CSPGs) as initial binding receptors prior to infection. These findings were corroborated by small interfering RNA (siRNA)-mediated knockdown experiments, as well as through a cancer cell line screen, where we identified a strong association between viral binding in high serum and the expression of chondroitin sulfate biosynthesis genes. Furthermore, HPV binding in the presence of human platelet lysate also demonstrated an increased dependance on CSPGs, suggesting a possible role for these receptor proteoglycans in active wound healing environments. Overall, this work highlights the significant influence that serum/platelet factors can have on virus binding and identifies CSPGs as alternative cell attachment receptors for HPV.

IMPORTANCE Heparan sulfate proteoglycans (HSPGs) have previously been identified as primary attachment factors for the initial binding of human papillomaviruses (HPVs) prior to infection. Here, we demonstrate that in vitro, HPV binding to HSPGs is strongly dependent on the surrounding experimental conditions, including the concentration of fetal bovine serum (FBS). We found that high concentrations of FBS can block HSPG-binding sites and cause a dependence on 6O-sulfated chondroitin sulfate proteoglycans (CSPGs) as alternative initial viral receptors. Further, we demonstrate that use of a human-derived alternative to FBS, human platelet lysate, also occludes HSPG-dependent binding, causing a shift toward CSPGs for viral attachment. As HPV infection of basal epithelial cells is thought to occur at sites of microtrauma with exposure to high serum levels and platelet factors, these unexpected findings highlight a possible role for CSPGs as important cellular receptors for the binding and infectivity of HPV in vivo.

INTRODUCTION

Human papillomaviruses (HPVs) are small nonenveloped double-stranded DNA viruses that predominantly infect actively dividing basal cells of the stratified epithelium (1–3). While most often associated with nonmalignant lesions, a subset of HPVs, including HPV16, are known causative agents in the development of several kinds of tumors, including cervical cancer (1, 4, 5). HPV infection occurs through a complex, multistep process originating at sites of microtrauma, whereby the epithelial basement membrane becomes exposed (1, 5–7). Virus particles initially bind secreted heparan sulfate proteoglycans (HSPGs) located within the basement membrane before ultimately transferring to an undetermined secondary receptor on the cell surface, leading to internalization and nuclear transport (1, 6, 7). Importantly, in many in vitro models, HPV particles can bypass the need for basement membrane binding and instead are able to directly bind cellular HSPGs prior to the transfer to a secondary receptor (1). For example, seminal work investigating cellular HPV adsorption identified cell surface-localized HSPGs, such as syndecans, as the dominant receptors for the initial binding of HPV (8, 9). Studies such as these and many others utilizing HPV pseudovirus (PsV) reporters have uncovered a variety of key proteoglycan core proteins involved in viral adsorption and demonstrated that HPV binding is dependent on the extent and pattern of heparan sulfate (HS) chain sulfation, with preferences for 6O- and N-sulfation of N-acetylglucosamine (GlcNAc) (10, 11). Recently, HPV PsV was shown to selectively bind and infect a wide range of cancer cell lines through interactions with cell surface-localized HSPGs (12). As primary epithelial cell cultures fail to bind HPV PsV, it was suggested that cancer cell lines, in particular, contain specific HS structures and sulfation patterns that are especially adept in binding viral particles (12, 13). This finding has spurred the development of tumor-tropic HPV virus-like particle (VLP)-based anticancer drugs, such as AU-011 (belzupacap sarotalocan), which is currently in clinical trials for the treatment of uveal melanoma (14).

HS and chondroitin sulfate (CS) are polysaccharide chains that contain repeating glucuronic acid (GlcA) and hexosamine disaccharides and are two of the main glycosaminoglycans (GAGs) produced by cells, the others being hyaluronan, dermatan sulfate (also known as CS-B), and keratan sulfate. HSPGs and chondroitin sulfate proteoglycans (CSPGs) are formed via the covalent attachment and stepwise synthesis of HS or CS chains onto a proteoglycan core protein. These proteoglycans are then transported to the cell surface or are secreted to the extracellular matrix, where they participate in a variety of important cellular processes, such as cell growth, signal transduction, and adhesion. During polymerization, the growing HS or CS polysaccharides can be further modified through the epimerization of GlcA to iduronic acid (IdoA) as well as through site-specific sulfation by a variety of heparan or chondroitin sulfotransferases. HS contains four possible sulfation sites: NS-, 6O-, and 3O-sulfation of GlcNAc as well as 2O-sulfation of GlcA/IdoA, which are sulfated nonrandomly in stretches across the polysaccharide chain. Meanwhile, CS is commonly modified at three sites: 4O- and 6O-sulfation of N-acetylgalactosamine (GalNAc) in addition to 2O-sulfation of GlcA/IdoA. CS polysaccharides generally exist as one of four common sulfated forms and have been classically named based on their sulfation signatures: CS-A (4O-sulfated), CS-C (6O-sulfated), CS-D (2O- and 6O-sulfated), and CS-E (4O- and 6O-sulfated). The degree and site specificity of these sulfation signatures are crucial to the functions of both HS and CS, as they provide the proteoglycan with an enormous diversity of possible binding sites for molecules, such as growth factors (e.g., fibroblast growth factor [FGF], vascular endothelial growth factor [VEGF], transforming growth factor-β [TGF-β], etc.) and extracellular matrix (ECM) components (e.g., collagen, integrins, and fibronectin) as well as invading viruses (15–17). Herpes simplex virus 1 (HSV-1), for example, utilizes HS chains with 3O-sulfated GlcNAc for viral entry, while dozens of other viruses, such as dengue virus, vaccinia virus, and cytomegalovirus, use sulfated HSPGs as initial attachment receptors prior to infection (15).

Despite the similarity in their tissue localization and glycan chain structure, HSPGs, but not CSPGs, are generally considered the canonical initial binding receptors for HPV. This conclusion is based on numerous studies that utilize the enzymatic removal of HS and CS species, as well as binding competition assays with exogenous GAGs, to determine the specific nature of the virus-cell interaction (8, 9, 18–20). While vital for our current understanding of HPV biology in vitro, these foundational studies remain limited in either the directness of the binding measurement (e.g., use of bound cellular lysates) or in their relevance to physiological binding events (e.g., use of detached cells, binding temperatures at 4°C, etc.) and thus may be missing important alternative binding partners that may complement HSPGs during viral adsorption. Additionally, a wide variety of binding conditions have been used historically in both HPV adsorption and transduction assays, adding further variability in the interpretation and comparison of these results.

Previous work has shown that serum proteins and growth factors are crucial for the successful completion of natural wound healing (21). As HPV infection is thought to require disruption or wounding of the epithelial cell barrier and its subsequent repair (1, 22), we wished to investigate the impact of serum and other soluble factors present in the wound environment on the binding and infectivity of HPV. Using a high-content microscopy-based approach to directly measure HPV PsV binding, we report that high concentrations of human or bovine serum greatly reduce the cellular binding of HPV and, in the case of fetal bovine serum (FBS), block HSPGs, causing a dramatic shift in binding specificity toward CSPGs. Additionally, we show that human platelet lysate (HPL), a human-derived media additive containing substantial wound healing growth factors and proteins, also significantly impairs HPV binding and drives a heightened dependence on CSPGs as the initial viral-binding receptor. Overall, our work highlights the importance of experimental conditions when studying HPV binding and, in so doing, identifies a previously unappreciated role for CSPGs as alternative cellular receptors mediating HPV infection.

RESULTS

High concentrations of FBS and human serum impair HPV16 PsV binding and infectivity.

To directly assess the impact of serum concentration on HPV infectivity, we first created HPV16 PsV containing a nuclear-localized mCherry reporter. Using a high-content plate-based microscopy platform, transduction of this fluorescent viral reporter was quantified and used as a surrogate readout for overall PsV infection. Two model cancer cell lines were chosen for these studies, a cervical cancer line, HeLa, and an ovarian cancer cell line, OVCAR4, as both of these models have previously been shown to be highly transducible with HPV16 PsV (12). Infectivity of the PsV was assessed after the short-term (2 h) incubation of viral particles with cells in the presence of various concentrations of serum, followed by washing and replacement with cell growth medium for 48 h before imaging. Unexpectedly, in the presence of high concentrations of either human serum (HumS) or FBS, the infectivity of HPV PsV was markedly attenuated in both cell lines (Fig. 1A and B). As cell division is critical for the transduction of HPV (3), we examined the effect of short-term serum incubation on both cell proliferation and cell cycle progression. These studies determined that there were no differences in the overall percentage of 5-ethynyl-2′-deoxyuridine (EdU)-positive cells or cell cycle profiles, regardless of the serum concentration used, thus, ruling out alterations in cell growth rate as a driver of the decreased infectivity observed under elevated serum conditions (data not shown).

FIG 1.

Serum concentration alters magnitude of HPV16 PsV binding. (A, B) Percent infectivity of cell lines in various concentrations of FBS (A) or HumS (B) normalized to 2% serum conditions. (C, D) Binding assessment of HPV PsV particles on HeLa or OVCAR4 cell lines in the presence of various concentrations of FBS (C) or HumS (D). (E, F) Representative images of OVCAR4 cells from binding assays described in panels C and D, respectively. The top image is Alexa Fluor 488-labeled HPV PsV alone, and the bottom image is a merge of HPV PsV (green) and HCS CellMask whole-cell stain (purple). (G, H) Binding assessment of HPV PsV incubated in different types of bovine or human sera at 10% or 100% concentration in DMEM using HeLa (G) or OVCAR4 (H) cell lines.

Adsorption of HPV to the cell surface is typically the first step of the viral entry process required for the transduction of cultured cell lines. We therefore sought to examine possible defects in HPV PsV binding as a potential mechanism for the inhibition of infection seen in high serum. To assess the PsV binding capacity of cell lines, we used a microscopy-based approach, which avoids limitations of standard flow cytometry-based viral-binding assays and allows for the precise measurement of several relevant parameters, such as binding events per cell area, degree of PsV aggregation, and localization of PsV binding events. HPV PsV was generated as discussed above and then labeled with Alexa Fluor dyes for direct fluorescent visualization of viral binding events. Using this approach, we identified a substantial serum dose-dependent reduction in PsV binding, as measured by the number of binding spots per cell area, for both HeLa and OVCAR4 cell lines (Fig. 1C to F). Interestingly, high FBS conditions were more effective at blocking PsV binding than HumS and displayed a greater inhibitory effect on OVCAR4 cells than on HeLa cells. Additional serum formulations, including calf serum and plasma-derived human serum, were also tested under 10% and 100% conditions and again demonstrated significant reductions in PsV binding in high serum levels, further supporting the generalizability of the observations (Fig. 1G and H). Overall, these findings suggest that inhibition of HPV PsV infectivity in high serum results from decreased initial cellular binding rather than from downstream serum-induced effects on virion uncoating or trafficking.

HumS-induced PsV binding inhibition results from PsV particle aggregation.

Next, we wished to examine the mechanism of binding inhibition under high HumS and FBS conditions. Close examination of the images from the microscopy-based binding assays using HeLa and OVCAR4 cells revealed that viral particles that were bound in high HumS formed larger than typical aggregates or “clumps” (Fig. 2A). To provide quantitative evidence for this observation, morphologic image analyses were performed on the cell-bound PsV clumps and determined that viral binding in the presence of high HumS produced aggregates that were both much brighter and larger than seen under low HumS or any FBS conditions (Fig. 2B and C). Additionally, we found that this clumping effect is serum dose -dependent and also varies based on the quantity of HPV PsV particles used, indicating that the mechanism of PsV aggregation is similar to that seen for the formation of antibody-antigen complexes (23) (Fig. 3A). Consistent with this hypothesis, we determined that the relative binding inhibition caused by HumS is also dependent on the quantity of PsV used, correlating with changes in viral aggregation and not with changes in overall cell surface binding, as calculated by total spot intensity of cells (Fig. 3B and C). To note, these aggregates were not the result of nonspecific or HPV-specific antibodies present in the HumS, as PsV clumps were also found in the presence of Ig-depleted HumS (Fig. 3D). Lastly, to determine the nature of PsV aggregation, we examined the clumping of cell-bound PsV that was preincubated in either 1% or 100% HumS, followed by dilution into a 10% or 100% HumS binding buffer (Fig. 3E). These data demonstrate that PsV clumping is solely driven by preincubation in high HumS and is not reversed upon dilution into low serum binding conditions. Altogether, these findings suggest that factors in HumS, but not FBS, can induce the aggregation of HPV PsV in solution and that this clumping effect drives the reduction in PsV binding and infectivity seen under high HumS conditions.

FIG 2.

High concentrations of human serum induce HPV PsV aggregation. (A) Masked images of HeLa cells with HPV PsV particles bound in 100% FBS (left) or 100% HumS (right). (B) Density plot of bound HPV PsV spot intensity and spot area on HeLa cells in 100% FBS or 100% HumS. (C) Median spot intensity and spot area of PsV bound to HeLa cells in different FBS and HumS concentrations.

FIG 3.

Human serum induces clumping of HPV PsV in solution. (A) Median spot intensity of particles bound in different HumS binding concentrations, using various initial quantities of PsV. (B) PsV binding capacity of HeLa cells using different quantities of PsV in the presence of 10% or 100% HumS. (C) Whole-cell PsV spot intensity of HeLa cells after treatment with different quantities of HPV PsV in the presence of various doses of HumS. (D) Median intensity and area of HPV PsV spots bound to HeLa cells in the presence of naïve HumS or HumS preprecipitated with protein A, protein G, or protein L agarose beads. (E) Median spot intensity and area of HPV PsV particles that were preincubated (“Pre”) and bound (“Bind”) in indicated concentrations of HumS measured on HeLa (left) and OVCAR4 (right) cells.

FBS-derived heparin-binding proteins inhibit HPV PsV binding to cells.

To determine the FBS-induced blocking mechanism of HPV PsV binding, we again conducted preincubation experiments where PsV was incubated in 1% or 100% FBS, followed by dilution into a 10% or 100% FBS binding buffer. Unlike what was observed with HumS, we found that the inhibitory effect of FBS on PsV binding occurred only when cells were bound in the presence of 100% FBS (100% FBS binding buffer) and was independent of preincubation conditions (Fig. 4A). This finding implies that FBS factors act on the cell surface to block incoming PsV binding and not on the PsV particles themselves. FBS contains a vast array of proteins, glycans, antibodies, coagulation factors, hormones, and electrolytes along with many other undefined components (24). In an effort to identify the specific factor(s) responsible for inhibiting viral adsorption, we first examined the PsV blocking effects of FBS pretreated with high temperatures (75°C and 95°C) to denature susceptible proteins. Using OVCAR4 cells, we found that exposing 100% FBS to 75°C or 95°C for 15 min prior to its use in PsV-binding assays attenuated most of the previously observed inhibitory effects (Fig. 4B). Treatment of 100% FBS with proteinase K in addition to the high temperatures further reduced the inhibition, returning PsV binding to levels near those found under 10% FBS binding conditions. Notably, pretreatment of FBS with various concentrations of heparin also rescued cellular PsV binding, with higher concentrations of FBS requiring more heparin to fully quench its inhibitory effect (Fig. 4C). These data suggest that FBS-induced PsV blocking is caused by heat-labile proteins within serum, whose inhibitory effects can be blocked by pre-binding to heparin.

FIG 4.

Heparin-binding proteins contribute to FBS-induced inhibition of PsV binding. (A) Binding assessment of HPV PsV in different preincubation and binding buffers using HeLa (left) and OVCAR4 (right) cell lines. (B) HPV PsV binding assessment of OVCAR4 cells under conditions containing FBS that was either untreated, treated with heat, or treated with proteinase K (proK) followed by heat. (C) Binding of HPV PsV to OVCAR4 cells under conditions containing various concentrations of exogenous heparin and FBS. (D, E) Binding of HPV PsV to HeLa (D) and OVCAR4 (E) cells in the presence of naïve or heparin-agarose-treated FBS.

As heparan sulfate is known to be the predominant initial receptor for the cellular binding of HPV, we questioned whether some of the numerous heparin-binding proteins (HBPs) present within FBS could be responsible for preventing viral binding. To assess this possibility, we used heparin-agarose beads to precipitate and deplete FBS of serum factors that bind heparin. This HBP-depleted FBS was then used in PsV-binding assays of HeLa and OVCAR4 cells, where it was shown to have very limited inhibitory effects on PsV binding, even when used at 100% (Fig. 4D and E). To confirm that this effect was associated with the removal of heparin-binding proteins from the FBS, we performed mass spectrometry on the heparin-agarose-precipitated serum proteins. As expected, we identified an extensive list of known HBPs that are present in high quantities in FBS (Table 1). Together, these results strongly suggest that under high FBS conditions, serum-derived HBPs are present in sufficient quantities to bind to and occlude cell surface HSPGs, effectively blocking the binding of incoming HPV PsV particles.

TABLE 1.

Mass spectrometry results of FBS-derived, heparin-agarose-precipitated peptides

Gene ID	Description	Heparin, no. PSMsa	Heparin, abundance
SERPINC1	Antithrombin-III	2937	1.01E+10
APOB	Apolipoprotein B	1636	2.74E+09
SERPINA5	Plasma serine protease inhibitor	1149	2.34E+09
APOA1	Apolipoprotein A-I	400	1.67E+09
THBS1	Thrombospondin-1	699	1.67E+09
VTN	Vitronectin	514	1.30E+09
KNG1	Kininogen-1	298	8.85E+08
PPBP	C-X-C motif chemokine	186	7.23E+08
HBA	Hemoglobin subunit alpha	180	7.04E+08
FN1	Fibronectin	424	6.70E+08
F2	Prothrombin	297	6.28E+08
F13B	Coagulation factor XIII B chain	142	6.24E+08
APOE	Apolipoprotein E	197	5.80E+08
ITIH2	Inter-alpha-trypsin inhibitor heavy chain H2	184	4.10E+08
KLKB1	Plasma kallikrein	109	2.40E+08
F13A1	Coagulation factor XIII A chain	184	2.13E+08
CLEC3B	Tetranectin	67	2.00E+08
LGALS3BP	Galectin-3-binding protein	91	1.85E+08
APOD	Apolipoprotein D	57	1.83E+08
ECM1	Extracellular matrix protein 1	78	1.79E+08
F11	Coagulation factor XI	50	1.67E+08
SERPIND1	Serpin family D member 1	106	1.67E+08
KIF12	Alpha-1-microglobulin	82	1.36E+08
CHIA	Chitinase	30	1.24E+08
QSOX1	Sulfhydryl oxidase	95	1.20E+08
LCN2	Lipocalin 2	46	1.07E+08
CLU	Clusterin	39	1.02E+08
LTF	Lactotransferrin	80	9.81E+07
APOH	Apolipoprotein H	38	8.68E+07
ITIH1	Inter-alpha-trypsin inhibitor heavy chain H1	79	8.40E+07
LOC506828	Uncharacterized protein	80	8.05E+07
F5	Coagulation factor V	60	7.22E+07
HRG	Histidine-rich glycoprotein (fragments)	51	7.18E+07
KNG2	Kininogen-2	245	5.55E+07
HABP2	Hyaluronan-binding protein 2	46	5.45E+07
SERPINF2	Alpha-2-antiplasmin	65	5.45E+07
APOA2	Apolipoprotein A-II	33	5.23E+07
FBLN1	Fibulin-1	43	4.86E+07
GAPDH	Glyceraldehyde-3-phosphate dehydrogenase	58	4.69E+07
PRPF6	PRP6 homolog	13	3.96E+07
PCOLCE	Procollagen C-endopeptidase enhancer	22	3.96E+07
C1R	Complement subcomponent C1r	34	3.69E+07
ARHGAP21	Rho GTPase activating protein 21	8	3.60E+07
C1QTNF3	Adiponectin M	70	3.56E+07
RNASE4	Ribonuclease A family member 4	30	3.44E+07
C4BPA	C4b-binding protein alpha chain	39	3.07E+07
MST1	Hepatocyte growth factor-like protein	16	2.99E+07
OLFML3	Olfactomedin-like protein 3	24	2.65E+07
APOC3	Apolipoprotein C-III	7	2.57E+07
TUBA1B	Tubulin alpha-1B chain	39	2.27E+07

^aPSM, peptide spectrum-match.

HPV PsV binds to 6O-sulfated CS under high FBS conditions.

Despite the dominant blocking effect of the FBS, HPV PsV was still able to bind and infect HeLa cells to a substantial degree in high serum (Fig. 1A and C). This was in contrast to the OVCAR4 cell line, which had essentially no cellular PsV binding or infection at FBS levels greater than 30%. We therefore wished to further examine the residual binding events seen with HeLa cells under high FBS conditions and to better understand the differences between the two cell lines in their susceptibility to HPV PsV binding and infection. First, we used specific glycosidases to remove HS and CS from HeLa cells prior to binding with PsV. As expected, under 10% FBS conditions, we found a strong dependence on HS for PsV binding, as shown by a greater than 75% reduction in binding capacity after treatment with either heparinase I (HepI) or heparinase III (HepIII) (Fig. 5A). However, in 100% FBS, neither HepI nor HepIII treatment had any appreciable effect on overall binding when used alone or in combination (Fig. 5B). Instead, treatment with chondroitinase ABC (ChABC) or hyaluronidase (HAase) resulted in a significant reduction of PsV binding in 100% FBS. As HAase, despite the name, is known to cleave CS species as well as hyaluronan in vitro, we concluded that PsV binding under high FBS conditions is likely dependent on CS and not HS. Importantly, while individual treatment with ChABC or HAase had no effect on PsV binding in 10% FBS, when these enzymes were used in combination with HepI and HepIII, a near complete loss in binding was demonstrated. This additivity between HS- and CS-degrading enzymes suggests that while HSPGs are the predominant binding partners for HPV, CSPGs may represent alternative initial receptors for incoming PsV, even under low serum conditions.

FIG 5.

FBS factors act on the cell surface to change binding specificity of HPV PsV. (A, B) HPV PsV binding comparison of HeLa cells after treatment with different glycosidases under 10% (A) and 100% (B) FBS conditions; HepI, heparinase I; HepIII, heparinase III; HAase, hyaluronidase; ChABC, chondroitinase ABC. (C, D) Binding comparison of HPV PsV on HeLa cells in the presence of desulfated heparins (C) and chondroitin sulfate species (D) normalized to untreated control, for 2% (top), 10% (middle), and 100% (bottom) FBS binding conditions; De-2OS, 2O-desulfated heparin; De-6OS, 6O-desulfated heparin; De-NS, NS-desulfated heparin; De-NS/ReA, NS-desulfated/reacetylated heparin.

To better understand the specific interactions between PsV and CS/HS under high serum conditions, we performed competition assays in which HPV PsV was preincubated with exogenous GAGs prior to HeLa cell binding. Differentially sulfated heparins were used as a proxy for HS chains and enabled assessment of the importance of sulfate position on PsV binding. As previously shown (12), heparin was a potent inhibitor of HPV binding under 2% and 10% FBS conditions, with 6O- and NS-sulfation as the most important sulfate positions for this inhibitory effect (Fig. 5C). In 100% FBS, however, preincubation of PsV with the modified heparins led first to a 3- to 4-fold increase in PsV binding at low heparin concentrations, followed by binding inhibition at high heparin concentrations. As discussed previously, this effect is likely driven by the ability of exogenous heparin to bind serum HBPs and thus quench their PsV-occluding effects on cellular HSPGs. The PsV blocking of CS was assessed using the four canonical sulfated CS species (CS-A, CS-C, CS-D, and CS-E) as well as dermatan sulfate (CS-B) under 2%, 10%, and 100% FBS conditions. Despite exhibiting only minor inhibitory effects in 2% or 10% FBS, we found that GalNAc(6S)-containing CS-C and CS-D were extremely potent inhibitors of PsV binding in 100% FBS, with 50% inhibitory concentration (IC₅₀) values in the high nanograms per milliliter range (Fig. 5D). This is in contrast to CS-A (4O-sulfated) and CS-B (2O- and 4O-sulfated), which had no effect on PsV binding in 100% FBS. CS-E, which possesses both 4O- and 6O-sulfated GalNAc, was also able to block binding under all FBS conditions but required high concentrations to do so. Additionally, as seen with heparin, CS-E promoted PsV binding when used at lower doses in 100% FBS, suggesting possible interactions of this GAG with inhibitory serum proteins. Overall, these data suggest that PsV binding to HeLa cells under high serum conditions is dependent on CS, not HS, and that 6O-sulfation of GalNAc is likely an important determinate mediating this interaction.

CSPGs represent HPV PsV receptors across a variety of cancer cell lines.

Next, we sought to extend our work to other model cancer cell lines and uncover genetic determinants responsible for the differences in PsV binding seen between HeLa and OVCAR4 cells under high serum conditions. Screening 20 unique lines from the NCI 60 cancer panel, we measured PsV binding capacity in 10% and 100% FBS. From this screen, we found a broad range in PsV binding ability throughout the cell panel, with five lines, including HeLa cells, demonstrating a high binding capacity in 100% FBS (deemed “High Binders”: UACC62, A498, HeLa, HS578T, and SKMEL28) and four lines, including OVCAR4 cells, exhibiting no binding (deemed “Low Binders”: OVCAR4, SW620, T47D, and HCT15) (Fig. 6A). RNA-sequencing data from the Cancer Cell Line Encyclopedia (CCLE) (25) were used to compare expression signatures between these groups, looking for differentially expressed genes that might contribute to increased PsV binding under high serum conditions. Filtering based on Gene Ontology gene sets (26, 27) (Aminoglycan biosynthetic process GO:0006023) as well as known proteoglycan genes, we found that high binding cells had increased expression of genes related to chondroitin sulfate biosynthesis and sulfation as well as increased expression of CSPG core proteins (Fig. 6B). These genes also correlated strongly with PsV binding capacity when mapped across all cell lines tested (Fig. 6C). Hierarchical clustering across CSPG biosynthesis genes successfully grouped the nonbinding cells and identified gene expression patterns defining the medium- and high-binding groups (Fig. 6D). By contrast, clustering based on HSPG-related genes was unable to differentiate between the binding groups, again supporting the importance of CS, and not HS, in mediating PsV binding under high serum conditions (data not shown).

FIG 6.

Expression of CSPG-related genes correlates with HPV PsV binding under high serum conditions across a variety of cancer cell lines. (A) HPV PsV binding capacity of cancer cell lines under 100% FBS binding conditions. The dashed line indicates binding capacity of HeLa cells in 10% FBS. FBS binding groups (High, Med, and Low) were determined based on magnitude of PsV binding in 100% FBS and used for downstream analyses. (B) Differential gene expression of GAG-related genes, comparing High versus Low binders from panel A. RNA-sequencing expression data were acquired from the Cancer Cell Line Encyclopedia (CCLE) (25); padj, adjusted P value. (C) Spearman correlation of GAG-related gene expression and PsV binding capacity in 100% FBS for cells in panel A. (D) Heatmap of RNA-sequencing expression data for genes involved in CS polymerization/elongation and sulfation and common CSPG core proteins across cells in panel A.

To further validate these findings, a library of small interfering RNAs (siRNAs) targeting HSPG- and CSPG-related genes was used to treat HeLa cells prior to PsV binding assessment. Under 10% FBS conditions, a number of HS synthesis genes were shown to be important for mediating PsV binding, including HS elongation/polymerization (EXT1, EXT2, and EXTL1) and sulfation (NDST1, NDST2, and HS6ST2) processes (Fig. 7A). However, knockdown of these same genes had little effect on PsV binding in 100% FBS (Fig. 7B). Instead, knockdown of many CS synthesis genes resulted in a substantial loss in PsV binding, including elongation/polymerization factors (XYLT2, CSGALNACT1, CHSY1, CHSY3, and CHPF) and the sulfotransferase CHST11. Interestingly, knockdown of HS initiation enzymes EXTL2 and EXTL3 resulted in improved PsV binding under high serum conditions. As both HS and CS compete for the same protein-tetrasaccharide linker for chain initiation, these findings are expected, as more CS is produced in the absence of HS chain synthesis (28). A variety of other HS sulfotransferases were also implicated (either positively or negatively) in PsV binding at 100% FBS. This suggests additional unexplored connections and possible compensatory mechanisms between HS and CS sulfation pathways. Of note, siRNA knockdown of common proteoglycan genes had no effect on the cellular binding of PsV, regardless of serum concentration (data not shown). This is consistent with previous reports that HPV binding is independent of proteoglycan core proteins, instead relying on the specificity and degree of GAG sulfation. Overall, these results identify a number of key CS biosynthesis genes as significant contributors to HPV PsV binding and further validate the important role for CSPGs as de facto viral receptors in the presence of high serum levels.

FIG 7.

Chondroitin sulfate biosynthesis and sulfation enzymes mediate HPV PsV binding under high serum conditions. (A, B) HPV PsV binding capacity of HeLa cells under 10% FBS (A) and 100% FBS (B) conditions after treatment with a panel of siRNAs targeting GAG-related genes; data are normalized to nontargeting siRNA controls (dashed black line). Increased (green points) and decreased (red points) binding was determined as those points significantly outside the binding range of control-treated cells (red dashed lines).

Human platelet lysate impairs HPV PsV binding and infectivity by altering target specificity on the cell surface.

To expand our findings using more human-relevant conditions, we utilized human platelet lysate (HPL) as a culture media additive instead of FBS. HPL contains numerous growth factors and proteins involved in the wound healing process and is commonly used as a human-derived substitute for FBS in the culturing of mesenchymal stem cells for clinical applications (24, 29). As HPV infection is thought to occur at sites of microtrauma, we wished to test the effects of HPL on PsV binding and infection. We found that HPL, even when used at low concentrations, resulted in a significant loss in PsV binding to HeLa and OVCAR4 cells (Fig. 8A). While PsV clumping did occur at high HPL concentrations, this alone could not explain the substantial loss in whole-cell PsV binding capacity observed across the cell lines (Fig. 8B to D).

FIG 8.

Human platelet lysate inhibits HPV PsV binding. (A) Binding assessment of HPV PsV to HeLa and OVCAR4 cell lines in the presence of various concentrations of HPL. (B) Density plots of bound PsV spot intensity and spot area on HeLa cells in 5% and 0.5% HPL. (C) Median spot intensity of PsV binding events on cell lines in various concentrations of HPL. (D) Whole-cell spot intensity of PsV binding to cell lines in HPL from experiment in panel C.

As HeLa cells were moderately more resistant to the effects of HPL, we wondered if a shift from HS to CS could also be occurring under these conditions, similar to that found in high levels of FBS. To test this possibility, we again used both enzymatic degradation of HS and CS chains as well as competition assays with exogenous GAGs in the presence of HPL. As expected, removal of HS through HepI and HepIII treatment resulted in a significant loss of binding at both 0.5% and 10% HPL, demonstrating a dependance on HS for PsV binding (Fig. 9A and B). In contrast, ChABC and HAase had no effect on binding at 0.5% HPL but at 10% HPL resulted in a greater than 50% reduction in overall PsV binding. Further, the combination of ChABC or HAase with a heparinase (either HepI or HepIII) resulted in a complete loss of binding at 10% HPL, indicating that while HSPGs play a predominant role in PsV binding in very low HPL, both HSPGs and CSPGs are responsible for the viral-binding events seen under 10% HPL conditions. Consistent with these findings, PsV competition assays with modified heparins and CS species were performed, this time in 10% HPL. As seen in 100% FBS, we found that treatment with the different heparin species resulted in an initial increase in PsV binding at low heparin concentrations, followed by a subsequent decrease in binding at high heparin concentrations (Fig. 9C). Additionally, treatment with CS-C and CS-D resulted in very potent inhibition of PsV binding, again similar to that observed in 100% FBS, highlighting a role for 6O-sulfated CSPGs as important attachment factors in 10% HPL (Fig. 9D). Lastly, to determine if the binding reduction seen upon increasing doses of HPL corresponded with a decrease in infectivity, HPV PsV was added to cells in the presence of HPL for 2 h at 37°C, followed by washing and a 48-h incubation under standard growth medium conditions. A substantial inhibition of infectivity was observed upon increasing doses of HPL and strongly correlated with the binding results seen previously for both HeLa and OVCAR4 cell lines (Fig. 9E). Overall, these data suggest that HPL, even at doses as low as 3%, can dramatically impair HPV PsV binding by causing the formation of viral aggregates as well as by blocking cell surface HSPGs and ultimately results in a potent inhibition of viral transduction.

FIG 9.

Human platelet lysate alters binding target specificity and infectivity of HPV PsV. (A, B) Binding of HPV PsV after pretreatment of HeLa cells with glycosidases in 0.5% (A) or 10% (B) HPL; data are normalized to untreated controls; HepI, heparinase I; HepIII, heparinase III; HAase, hyaluronidase; ChABC, chondroitinase ABC. (C) HPV PsV binding of HeLa cells in 10% HPL upon treatment with modified heparins; data are normalized to untreated controls; De-2OS, 2O-desulfated heparin; De-6OS, 6O-desulfated heparin; De-NS, NS-desulfated heparin; De-NS/ReA, NS-desulfated/re-acetylated heparin. (D) HPV PsV binding of HeLa cells in 10% HPL upon treatment with chondroitin sulfate species; data are normalized to untreated controls. (E) HPV PsV infectivity of HeLa and OVCAR4 cells in the presence of various concentrations of HPL.

DISCUSSION

Here, we show that CSPGs, as a product of their associated CS chains, can function as initial receptors for HPV16 PsV binding in a large subset of cancer cell lines. This interaction is most pronounced when HS/HSPGs, the canonical HPV-binding receptors, are enzymatically removed or are otherwise occupied and is limited to CS-C or CS-D chains containing 6O-sulfated GalNAc. Due to the considerable levels of heparin-binding proteins in FBS and HPL, these cell culture media additives, when used at high concentrations, can actively block HSPGs on the cell surface, occluding the HS-dependent binding of incoming viral particles. As a result, cellular CSPGs can act as de facto primary receptors for the initial binding of HPV PsV prior to cell entry and transduction (Fig. 10). Because natural HPV infection of basal epithelial cells occurs at sites of wound healing surrounded by expectedly high levels of serum and platelet factors, our results suggest a possible role for CSPGs as viral attachment factors in these environments. We have previously reported that HepIII treatment inhibited HPV16 PsV infection by approximately 90% in a mouse cervicovaginal challenge model, suggesting that HSPGs serve as the primary in vivo viral receptor (30). However, the extension of these results to humans must be done with caution, given that different mechanisms of inhibition are seen for bovine and human serum in this study and the lack of available data regarding the composition of mouse serum or mouse platelet lysate.

FIG 10.

The inhibitory effects of FBS or HPL on the target specificity and binding of HPV PsV. Model of HPV16 PsV binding under conditions of low and high FBS or HPL.

Studies investigating the mechanism of HPV binding were initiated many years ago and have utilized a variety of different viral formulations (e.g PsV, quasivirus, authentic virus, etc.), binding conditions, and measurement techniques (7–12, 18, 19). We wished to build upon this work to examine the role that the surrounding binding conditions can play in determining the specificity of virus-cell interactions. Making use of a high-content microscopy platform enabled us to quantify PsV binding on intact cell monolayers (instead of detached cells) and allowed for visualization of many other parameters of viral binding, such as distribution and aggregation, which could be relevant for HPV entry and infection. Fetal bovine serum is the most common cell culture media additive used in labs worldwide, and thus, low concentrations of FBS (1 to 10%) have been routinely used in studies of HPV binding and infection. Our work suggests that the concentration of FBS should be carefully monitored in future binding experiments, as we show that high FBS levels can profoundly influence the glycan binding specificity of HPV on model cell lines. Human serum was also tested in our PsV attachment studies and at high doses was shown to greatly inhibit both PsV binding and transduction. Yet, unlike FBS, we found that this inhibition occurred via the induction of PsV aggregation, independent of serum antibodies, and not through major shifts in cellular binding specificity. These findings depict an alternative mode of viral inhibition, which needs to be addressed in future studies investigating HPV transduction.

While HumS is generally not sufficient for the culturing of most human cell lines (due to a deficiency of necessary chemokines and nutrients), HPL contains significant amounts of growth factors and proteins present in settings of inflammation or wound healing and provides a comparable human-derived alternative to FBS. We found that HPL is an extremely potent inhibitor of HPV PsV binding at concentrations as low as 3% and acts through a similar mechanism to FBS by blocking viral attachment to cellular HSPGs. While minor resistance to the effects of HPL is observed in HeLa cells compared to OVCAR4s, this effect is overcome upon increasing the HPL concentration, indicating that CSPG binding receptors may also be occluded by the high levels of HBPs or other factors present within the platelet extracts. The propagation of cell lines in human platelet lysate-supplemented media generally requires between 5 and 20% HPL for sustained growth (29). As these conditions far exceed the amount needed for inhibiting HPV binding and transduction, caution should be used in the interpretation of data if HPL is to be used as a media additive in future virus studies. Interestingly, human platelet lysate, like HumS, also induced the formation of viral aggregates when preincubated with PsV prior to cellular attachment. While this effect was not the sole mediator driving the inhibition of PsV binding events, it suggests the presence of a similar substrate within both serum and platelet granules that can act as an aggregator of PsV particles. These findings have potential implications for the development and application of VLP-drug conjugates, such as AU-011. Current intravitreous treatments of AU-011 provide relatively direct trafficking to choroidal melanomas; however, any future studies involving intravenous delivery will need to be monitored for particle aggregation, as this will likely affect extravasation and distribution of the drug to tumors.

The concept of CSPGs as possible HPV receptors has been hypothesized and tested previously, where it was demonstrated that exogenous CS can indeed act as a weak inhibitor of HPV binding (19). However, this inhibitory effect was found to be 10-fold lower than that of heparin and was limited only to CS-C. Our work using HeLa cells under low serum conditions supports these findings, as CS-C was shown to moderately impair HPV PsV binding when used at high doses. Additionally, we showed that the enzymatic removal of CS had little effect on overall binding unless used in combination with a heparinase, again highlighting the dominant role that HSPGs play over CSPGs in the binding of HPV PsV in low serum. However, we demonstrate for the first time that under conditions of high serum/platelet factors, these roles are reversed, with CSPGs as the primary interacting partner for incoming viral particles. Further, through binding competition assays, we found that 6O-sulfation of GalNAc, present in CS-C and CS-D, is important for these CSPG-mediated PsV binding events. This result converges well with the inhibitory effects of CS-C, but not 4O-sulfated CS-A, seen in previous studies investigating HPV PsV binding and infection (19, 31).

Despite a strong binding preference for CS-C over CS-A under high serum conditions, our siRNA screen of HS and CS biosynthetic enzymes identified a known CS 4O-sulfotransferase, CHST11, as a crucial mediator of PsV binding while failing to discern a role for CS 6O-sulfotransferases CHST3 or CHST7. These seemingly conflicting findings can be explained by the substantial redundancy within the CS biosynthesis pathway. CHST11 is one of three known chondroitin sulfate 4O-sulfotransferases, the others being CHST12 and CHST13. While CHST11 may play a redundant role in CS 4O-sulfation, it has been demonstrated that this enzyme is crucial for the maintenance of CS polymerization and overall chain length (32–34). These insights provide important context for our siRNA screen findings, as CHST11, but not CHST12 or CHST13, was implicated in PsV binding under high serum conditions. Thus, it is likely that CHST11 influences PsV binding through its effect on CS polymerization and overall chain elongation and not solely via its 4O-sulfotransferase activity. Additionally, as CHST3 and CHST7 are both expressed at appreciable levels in HeLa cells, the knockdown of either of these genes individually may have been insufficient in preventing CS-C production and could explain why PsV binding was not impaired. Further studies involving the double knockdown or knockout of these sulfotransferases will need to be performed to further investigate these claims.

The work presented here has some limitations that require further investigation. As the composition of heparin-binding proteins and other factors surrounding the basement membrane and basal epithelium during wounding are not easily measured, and likely vary based on degree of trauma, it is unclear how closely the serum or platelet lysate used in our assays reflects physiological viral-binding conditions. While the wide effective range that we demonstrate for HPL and serum factors partially mitigates these concerns, subsequent in vivo studies will be needed to determine the exact components present in a wounding environment. Additionally, while a variety of cancer cell lines were used in this study to aid in uncovering genetic determinants of HPV binding and further define the tumor-targeting properties of HPV VLP-based drugs, these lines may not accurately represent the natural interaction of HPV with basal epithelial cells or the basement membrane. Future work will need to test the effect of different serum and platelet factor conditions on HPV binding to these important viral targets. Lastly, the process by which CS-bound PsV ultimately enters and infects cells remains to be elucidated. HS binding is known to induce a conformational change of the HPV particle, which promotes the transfer to a secondary entry receptor. Further studies will be needed to determine if CS binding can induce similar structural changes in the viral particle or if, instead, CS-bound HPV uses HSPGs as intermediate receptors prior to entry into cells.

Overall, we demonstrate that under conditions of high serum or platelet lysate, the cellular binding and infectivity of HPV is dramatically impaired. This surprising observation results from a shift in dependance from HSPGs to CSPGs for initial viral attachment and highlights an unexpected role for CS in the binding and entry of HPV. While inhibition in fluids or factors known to be present at sites of HPV infection may seem evolutionarily disadvantageous, there could be a number of benefits, including a refined tropism for susceptible host cells in vivo or avoiding immune recognition as a result of hematological spread. This work also emphasizes the importance of experimental conditions when studying HPV binding interactions. Care should be taken when choosing buffers and media additives for a given application. Lastly, as FBS and other serum products are used extensively in cell-based assays across the field of virology, these findings extend beyond HPV studies and could be of importance to work involving other HSPG-binding viruses, especially those that utilize the bloodstream for viral dissemination.

MATERIALS AND METHODS

Cell culture and reagents.

HeLa cells were acquired from ATCC and were grown in Dulbecco’s modified Eagle’s medium (DMEM) + 10% FBS. OVCAR4 cells and the NCI 60 cancer cell panel were acquired from the Developmental Therapeutics Program at the National Cancer Institute (Frederick, MD) and were grown in RPMI 1640 + 5% FBS. All cell lines were grown at 37°C and subcultured every 3 to 4 days. FBS (F2442), human serum (HumS) (H6914), and plasma-derived human serum (H4522) were purchased from Millipore-Sigma. Calf serum was acquired from Thermo Fisher Scientific (16170086). Human platelet lysate (HPL) used was from Stem Cell Technologies (06960). Glycosidases HepI (H2519), HepIII (H8891), HAase (H3506), and ChABC (C3667) were all purchased from Millipore-Sigma, as were heparin (H4784), CS-A (C9819), CS-B (C3788), and CS-C (C4384). Modified heparins were acquired from Galen Laboratory Supplies (2O-desulfated heparin [De-2OS; DSH001], 6O-desulfated heparin [De-6OS; DSH002], NS-desulfated heparin [De-NS; DSH003], NS-desulfated/reacetylated heparin [De-NS/Re-Acetyl; DSH004]) as were CS-D (CD-D) and CS-E (CS-EII). All modified heparins and chondroitin sulfate species were diluted in phosphate-buffered saline (PBS) to a concentration of 5 mg/mL before further dilution in binding media for use in PsV binding assays. For all binding and infection studies, FBS and HumS dilutions described were made using DMEM (Gibco), while HPL dilutions were made in RPMI 1640 (Gibco) to avoid previously observed coagulation in high-calcium-containing buffers, such as DMEM.

HPV16 PsV production.

HPV16 PsVs were generated as described previously (35); in brief, 293TT cells were transfected with p16sheLL and a pmCherry-NLS reporter plasmid (a gift from Martin Offterdinger, Addgene 39319) (36) and were grown for 48 h prior to harvest and lysis. PsV lysate was allowed to mature in a 9.5 mM MgCl₂ buffer containing 0.5% Triton, 0.1% benzonase nuclease, and 25 mM sodium phosphate at 37°C overnight prior to separation on an OptiPrep density gradient (Sigma D1556). Gradients were centrifuged at 50,000 rpm for 3.5 h prior to fractionation and loading on a polyacrylamide gel for protein detection and quantification. PsV-containing fractions were pooled and run against bovine serum albumin (BSA) loading controls for quantification of L1 protein content via gel imaging using an Amersham Imager 600 (GE Healthcare Life Sciences). PsV titering was performed on HeLa cells prior to studies under serum-containing conditions. Fluorescent labeling of HPV PsV was performed as described previously (12) using Alexa Fluor 488 or 647 protein labeling kits from Thermo Fisher (A10235 and A20173). PsV was then purified using an OptiPrep gradient and quantified as described above.

High-throughput assessment of HPV16 PsV infectivity and binding.

HPV16 PsV infectivity was determined via transduction of the pmCherry-NLS fluorescent reporter. Cells were seeded at 4 × 10³ cells per well in 96-well imaging plates (PerkinElmer 6055300) and incubated for 24 h prior to the addition of HPV PsV. Then, 100 μL of 500 ng/mL HPV PsV (for a total of 50 ng L1/well) was prediluted in various serum or platelet lysate concentrations and then added to wells, followed by a 2-h incubation at 37°C. Cells were then washed three times with DMEM + 10% FBS followed by incubation in growth medium at 37°C for 48 h. Next, cells were washed with PBS and fixed with 4% paraformaldehyde (PFA) for 15 min at room temperature (RT) prior to nuclear staining with Hoechst 33342 (Thermo Fisher Scientific, H3570). Cells were then washed and imaged using a CellVoyager CV8000 high-content microscope (Yokogawa) with the 10× objective for downstream image processing and analysis. HPV PsV binding was determined via the fluorescent detection of Alexa Fluor 488- or 647-labeled viral particles; 7.5 × 10³ cells were seeded per well in 96-well imaging plates and grown for 24 h prior to the addition of PsV. PsV was prediluted in experimental binding buffers at 250 ng/mL (unless otherwise stated) and incubated at RT for 30 min. Then, 100 μL of the PsV dilutions (for a total of 25 ng L1/well) was added to the preplated cells, incubated for 1 h at 37°C, and washed three times with PBS prior to fixing with 4% PFA for 15 min at RT. Next, cells were permeabilized with 0.1% Triton X-100 for 15 min at RT and stained for 30 min with HCS CellMask green or deep red stains (Thermo Fisher Scientific H32714 and H32721, respectively) and Hoechst 33342, followed by imaging using a 40× water objective and a CellVoyager CV8000 high-content microscope. Preincubation binding studies using PsV followed the protocol above but were preincubated at 10× concentration (2.5 μg L1/mL) in the designated preincubation buffer and then further diluted 1/10 (final concentration of 250 ng L1/mL) into the designated binding buffer. Ig depletion of HumS was performed via incubation of HumS with protein A (Thermo Fisher Scientific, 22810), protein G (Sigma, 11719416001), or protein L (Thermo Fisher Scientific, 20510) agarose beads. Agarose bead precipitation of HumS was performed using 0.5 mL of agarose slurry in 2.5 mL of HumS. Reactions were rotated overnight at 4°C, followed by two rounds of centrifugation to sufficiently remove agarose beads from solution before use in PsV-binding studies.

Image analyses were performed using Columbus Image Analysis software (PerkinElmer) using pipelines constructed and validated on controlled optimization plates. Cellular nuclei and cytoplasm were detected and measured based on Hoechst 33342 and HCS CellMask stains, respectively. mCherry-positive cells were determined based on median fluorescent intensity of mCherry within the nuclear mask, with positivity thresholds determined per plate based on untransduced cellular controls. Percent infectivity was then calculated from the ratio of mCherry-positive cells to total cells imaged. Binding events were detected via a spot identification analysis pipeline using channels corresponding to the Alexa Fluor 488 or 647 PsV and quantified for size and fluorescent intensity. Cell area was determined from the cytoplasmic mask and allowed for the calculation of median binding events per squared pixel (px²).

Enzyme/heat treatment, heparin-agarose bead precipitation, and mass spectrometry of FBS proteins.

Heat treatment of FBS was performed for 15 min using standard heat blocks set to 75°C or 95°C. Treated FBS was then allowed to cool and was centrifuged to remove precipitate prior to use in binding assays. Proteinase K (Qiagen, 19131) treatment was performed using 5 ng of proteinase K per mL of FBS and incubated for 1 h at 37°C, followed by heat treatment at 75°C or 95°C and subsequent use in PsV binding experiments. Heparin-agarose (Sigma, H6508) and control agarose (protein G, Sigma, 11719416001; GlcNAc-agarose, Sigma, A2278) bead precipitation of FBS was performed using 0.5 mL of agarose slurry in 2.5 mL of FBS. Reactions were rotated overnight at 4°C, followed by two rounds of centrifugation to sufficiently remove agarose beads from solution before use in PsV binding studies. Agarose pellets were further washed and collected for analysis by mass spectrometry. Using a bovine peptide database, quantification of individual peptides was calculated, and abundance was determined for peptides appearing in heparin agarose-treated FBS but not in the protein G or GlcNAc-agarose bead controls.

Glycosidase treatment and exogenous GAG competition assays.

Glycosidases were initially resuspended in manufacturer-recommended buffers (HepI/HepIII/ChABC: PBS + 50 mM NaCl + 4 mM CaCl₂ + 0.01% BSA; HAase: 20 mM sodium phosphate buffer + 77 mM NaCl + 0.01% BSA), followed by dilution in DMEM without FBS at 2 U/mL HepI, 1 U/mL HepIII, 1 U/mL ChABC, and 200 U/mL HAase. One hundred microliters of these dilutions was added to preseeded HeLa cells (7.5 × 10³/well) on 96-well imaging plates, which were then incubated at 37°C for 1.5 h. Wells were then aspirated of glycosidase treatment, and HPV PsV was bound and quantified as described above. Competition assays with exogenous GAGs were performed on HeLa cells in binding buffers containing defined FBS or HPL levels. Both PsV and GAGs were prediluted in binding buffer and combined and incubated at RT for 30 min. One hundred microliters of the PsV/GAG dilutions was then added to cells for binding assessment, as described above.

Gene expression data and analysis.

RNA-sequencing expression data for the cancer cell line panel were acquired from the Cancer Cell Line Encyclopedia (CCLE) (25) as raw counts and were processed and analyzed in R via the DESeq2 package. Both pheatmap and ggplot2 packages were used for downstream hierarchical clustering and graphing.

siRNA library screen.

The GAG-targeting siRNA library was manually constructed and purchased from Horizon Discovery Biosciences using ON-TARGETplus Smart pools containing four unique siRNAs against each gene. siRNAs were diluted in 1× siRNA buffer (Horizon Discovery) to a final concentration of 2 μM. HeLa cells were preplated at 4 × 10³ cells per well of a 96-well imaging plate and were grown for 24 h prior to transfection with 5 nM siRNA and 0.2 μL of Dharmafect1 transfection reagent (Horizon Discovery, T-2001-03) per well. Cells were then grown for 72 h prior to PsV binding assessment.

Data analysis, graphing, and statistics.

Data were analyzed and plotted using R (gene expression correlation, volcano plots, hierarchical clustering, heatmaps, and spot density plots) or GraphPad prism (bar/line plots and nonlinear regression fitting). Unless otherwise stated, data points represent the mean of three independent biological replicates, with error bars representing the standard deviation of the mean. Statistical significance was determined by Student’s t test performed using GraphPad prism and denoted as the following: *, P < 0.05; **, P< 0.01; ***, P < 0.001; ****, P < 0.0001.