Human papillomavirus capsids preferentially bind and infect tumor cells

Abstract

We previously determined that human papillomavirus (HPV) virus-like particles (VLPs) and pseudovirions (PsV) did not, respectively, bind to or infect intact epithelium of the cervicovaginal tract. However, they strongly bound heparan sulfate proteoglycans (HSPG) on the basement membrane of disrupted epithelium and infected the keratinocytes that subsequently entered the disrupted site. We here report that HPV capsids (VLP and PsV) have the same restricted tropism for a wide variety of disrupted epithelial and mesothelial tissues, whereas intact tissues remain resistant to binding. However, the HPV capsids directly bind and infect most tumor-derived cell lines in vitro and have analogous tumor-specific properties in vivo, after local or intravenous injection, using orthotopic models for human ovarian and lung cancer, respectively. The pseudovirions also specifically infected implanted primary human ovarian tumors. Heparin and ι-carrageenan blocked binding and infection of all tumor lines tested, implying that tumor cell binding is HSPG-dependent. A survey using a panel of modified heparins indicates that N-sulfation and, to a lesser degree, O-6 sulfation of the surface HSPG on the tumors are important for HPV binding. Therefore, it appears that tumor cells consistently evolve HSPG modification patterns that mimic the pattern normally found on the basement membrane but not on the apical surfaces of normal epithelial or mesothelial cells. Consequently, appropriately modified HPV VLPs and/or PsV could be useful reagents to detect and potentially treat a remarkably broad spectrum of cancers.

Keywords: heparan sulfate proteoglycans; human papillomavirus; tumor tropism.

Figure 1

Normal tissues are refractory to HPV binding in vivo unless traumatized. HPV VLPs coupled to Alexa Fluor 488 (AF488) were administered intraperitoneally into naïve animals or animals that had been pre-treated with 2% nonoxynol-9 (N9) via the same route. After 24hr, the tissues were harvested, frozen, sectioned and VLPs were imaged directly by detecting AF488 signal microscopically (Blue=DAPI, Green=VLP). The first row depicts tissues from animals that did not receive N9 (panels, a, c, e, g and i) and the second row depicts tissues from animals after N9 pre-treatment (b, d, f, h and j). Images are representative of 3 animals.

Figure 2

HPV-16 infects many human tumor cell lines. A panel of 56 human tumor cell lines were infected with HPV16-GFP PsV, and GFP signal was measured 72hr later. HeLa cells served as controls and were used to calculate the relative infection of each cell type and these values are plotted (see Table S1 for raw data values). Cell lines are ordered as follows: Leukemia/Lymphoid (K-562, RPMI 8226, CCRF-CEM, MOLT-4, HL-60); Melanoma (SK-MEL-2, UACC-62, SK-MEL-5, MALME-3M, LOXIMVI, UACC-257, M14, SK-MEL-28); Central Nervous System (CNS; SNB-19, U251, SF-268, SF-539, SNB-75, SF-295); Colon (HCC-2998, HCT-15, KM12, COLO205, HCT-116, HT29, SW-620); Renal (TK-10, CAKI-1, A498, 786-0, SN12C, ACHN, UO-31, RXF-393); Lung (NCI-H522, NCI-H23, NCI-H322M, NCI-H460, EKVX, A549, HOP-62, HOP-92, NCI-H226); Ovarian (OVCAR-3, OVCAR-4, IGROV1, OVCAR-8, OVCAR-5, SK-OV-3); Breast (T47D; MCF7, HS578T, BT-549, MDA-MB-231); Prostate (PC-3, DU-145).

Figure 3

HPV16 VLP binding to and PsV infection of tumor cell lines is competitively inhibited by heparin. (a) Flow histograms of 2µg/ml AF488 VLPs bound to cells in the absence (black line) or presence (gray line) of 1mg/ml heparin (negative control, dashed line). Tumor cell lines tested: Cervical (HeLa, SiHa, C 33A, CaSki); Head and Neck (CAL-33, HSC-3, UPCI SCC-90, UPCI SCC-154); Melanoma (SK-MEL 2, SK-MEL 5, SK-MEL 28, UACC-62); Lung (NCI-H23, NCI-H322M, NCI-H460, NCI-H522); Ovarian (SK-OV-3, A2780, OVCAR-3, OVCAR-4); Uroepithelial (J82, RT112, T24, UMUC-5). (b) Dose response inhibition of binding using modified heparins and heparinases on SK-OV-3 (ovarian, solid black line), NCI-H460 (lung, solid gray line) and SK-MEL-2 (melanoma, dashed black line) (experiment performed in triplicate, data representative of two experiments). Dose response of heparin inhibition of PsV infection of (c) ovarian, (d) lung and (e) melanoma cancer cell line panels. Heparin concentration begins with 1mg/ml and is serially diluted three-fold (experiment performed in triplicate, data representative of two experiments).

Figure 4

ι-Carrageenan blocks binding of HPV to SHIN-3 DSR ovarian tumors in vivo. Non-tumored and SHIN-3 DSR tumored mice were examined for AF488 VLP binding (a) and HPV16-Luc PsV infection (b). (a) Animals received an intraperitoneal injection of PBS, 100µg AF488 VLP or 100µg AF488 VLP+1% ι-carrageenan. 2hr post-injection, the small intestines was removed and placed in a “wheel” formation for ex vivo fluorescent imaging. White, Red (tumor), and Green (VLP) channels were collected and individual as well as merged images are pictured. (b) Tumored and non-tumored animals received an intraperitoneal injection of 1×10^8 IU of HPV16-Luc PsV in 100µl PBS +/− 1% ι-carrageenan. 48hr later, luciferin substrate was administered and bioluminescence was measured. A region of interest was drawn around the peritoneal cavity and the average radiance was calculated. All data are representative of n ≥5/group.

Figure 5

HPV pseudoviruses infect established xenograft human ovarian tumors but not healthy tissue in vivo. Human ovarian tumors passaged only in immune-compromised animals were subcutaneously implanted. When tumors were between 7–15mm in diameter, animals received an intratumoral injection of 50µl PBS or 50µl 2×10^7 IU HPV16-Luc (a) or RFP (b) PsV. Non-tumored animals received a subcutaneous injection in their hind flank. Luminescence values are reported as average radiance (a) and micrographs (b) depict RFP expression (Red=RFP; Blue=DAPI) as a readout for PsV infection. n= 2–11/group.

Figure 6

HPV pseudoviruses infect orthotopic lung tumors after systemic administration (IV). Non-tumored or orthotopically H460-GFP tumored animals received intravenous injection of 25µg or 100µg each of HPV16-Luc 48hr before imaging (a) and (b) and HPV16-RFP 72hr before imaging (c). Whole body luminescent images were first acquired (a) followed by imaging of individual organs ((b) depicts lungs) (signal decreases rapidly upon euthanasia and dissection; Fig. S5 shows raw images of all organs). After imaging, tissues were frozen, sectioned and examined by fluorescent microscopy (c). Images are merged channels, Blue=DAPI, Red=PsV infection and Green=tumor (faint in some images). T= tumor, N= normal. n=7–10/group.