Learn more: PMC Disclaimer | PMC Copyright Notice
E6^E7, a Novel Splice Isoform Protein of Human Papillomavirus 16, Stabilizes Viral E6 and E7 Oncoproteins via HSP90 and GRP78
Associated Data
ABSTRACT
Transcripts of human papillomavirus 16 (HPV16) E6 and E7 oncogenes undergo alternative RNA splicing to produce multiple splice isoforms. However, the importance of these splice isoforms is poorly understood. Here we report a critical role of E6^E7, a novel isoform containing the 41 N-terminal amino acid (aa) residues of E6 and the 38 C-terminal aa residues of E7, in the regulation of E6 and E7 stability. Through mass spectrometric analysis, we identified that HSP90 and GRP78, which are frequently upregulated in cervical cancer tissues, are two E6^E7-interacting proteins responsible for the stability and function of E6^E7, E6, and E7. Although GRP78 and HSP90 do not bind each other, GRP78, but not HSP90, interacts with E6 and E7. E6^E7 protein, in addition to self-binding, interacts with E6 and E7 in the presence of GRP78 and HSP90, leading to the stabilization of E6 and E7 by prolonging the half-life of each protein. Knocking down E6^E7 expression in HPV16-positive CaSki cells by a splice junction-specific small interfering RNA (siRNA) destabilizes E6 and E7 and prevents cell growth. The same is true for the cells with a GRP78 knockdown or in the presence of an HSP90 inhibitor. Moreover, mapping and alignment analyses for splicing elements in 36 alpha-HPVs (α-HPVs) suggest the possible expression of E6^E7 mostly by other oncogenic or possibly oncogenic α-HPVs (HPV18, -30, -31, -39, -42, -45, -56, -59, -70, and -73). HPV18 E6^E7 is detectable in HPV18-positive HeLa cells and HPV18-infected raft tissues. All together, our data indicate that viral E6^E7 and cellular GRP78 or HSP90 might be novel targets for cervical cancer therapy.
IMPORTANCE
HPV16 is the most prevalent HPV genotype, being responsible for 60% of invasive cervical cancer cases worldwide. What makes HPV16 so potent in the development of cervical cancer remains a mystery. We discovered in this study that, besides producing two well-known oncoproteins, E6 and E7, seen in other high-risk HPVs, HPV16 produces E6^E7, a novel splice isoform of E6 and E7. E6^E7, in addition to self-interacting, binds cellular chaperone proteins, HSP90 and GRP78, and viral E6 and E7 to increase the steady-state levels and half-lives of viral oncoproteins, leading to cell proliferation. The splicing cis elements in the regulation of HPV16 E6^E7 production are highly conserved in 11 oncogenic or possibly oncogenic HPVs, and we confirmed the production of HPV18 E6^E7 in HPV18-infected cells. This study provides new insight into the mechanism of splicing, the interplay between different products of the polycistronic viral message, and the role of the host chaperones as they function.
INTRODUCTION
Human papillomaviruses (HPVs) are nonenveloped double-stranded DNA viruses which infect mucosal or skin keratinocytes. More than 120 HPV genotypes have been reported (1), and those responsible for malignant tumor formation are called high-risk or oncogenic HPVs, while those for benign anogenital warts are called low-risk or nononcogenic HPVs (2, 3). More than 95% of cervical cancers, 50 to 90% of other anogenital cancers, and 20 to 30% of oral and pharyngeal cancers are associated with persistent infection and host genomic integration of high-risk HPVs (3,–5). Among the major genotypes of high-risk HPVs, namely, HPV16, -18, -31, -33, -45, and -58, HPV16 is the most prevalent genotype; it is responsible for ~60% of cervical cancer cases worldwide (4, 6).
Oncogenic activities of high-risk HPVs are mediated by two viral oncoproteins, E6 and E7. E6 and E7 oncoproteins target tumor suppressor proteins, such as p53 and pRB, to induce cell proliferation, antiapoptosis, genome instability, and escape from innate immune systems (7,–9). In HPV16, these two viral oncogenes are transcribed from an early promoter, P97, as a single E6E7 polycistronic pre-mRNA, and its transcriptional level and translational efficiency are regulated by an alternative RNA splicing machinery of host cells (7, 10, 11). This polycistronic pre-mRNA contains two introns and three exons, with intron 1 in the E6 open reading frame (ORF) containing three alternative 5′ splice sites (ss) and three alternative 3′ ss. Thus, alternative RNA splicing of this E6 intron produces multiple splice isoforms of the E6E7 RNA (12). E6*I and E6*II are two major splice isoforms of E6 and have been noticed for decades (13,–15). E6*I RNA functions as an E7 mRNA for translation of E7 oncoprotein, while unspliced E6 RNA is responsible for full-length E6 expression (12, 16, 17). The role of E6*I protein in the regulation of cellular function and viral carcinogenesis remains under investigation (18,–21).
E6^E7 is another splice isoform produced by alternative splicing from a 5′ ss at nucleotide (nt) 226 to a 3′ ss at nt 742 of the polycistronic E6E7 pre-mRNA (16, 22). This RNA splicing causes the N-terminal coding region of the E6 ORF to be spliced in frame with the C-terminal coding region of the E7 ORF. Thus, this spliced RNA isoform encodes an E6^E7 fusion protein with a characteristic structure of the N-terminal half (41 aa residues) of E6 and the C-terminal half (38 aa residues) of E7 (Fig. 1A). E6^E7 was initially discovered through an in vitro splicing assay (16), and its existence was further confirmed in most HPV16-positive cervical cancer tissues and cell lines (12, 22). Although the consistent expression of E6^E7 in HPV16-positive cell lines and cervical cancer tissues indicates that its potential function is to maintain tumor cell growth, its function has never been explored.
Coimmunoprecipitation (co-IP) and LC-MS/MS analysis identified HSP90α, HSP90β, and GRP78 as E6^E7-interacting proteins. (A) Diagram of HPV16 E6^E7 amino acid residues. E6- and E7-derived amino acid residues are indicated in black and blue letters, respectively. The E6-derived nuclear localization signal (NLS) and E7-derived nuclear export signal (NES) are underlined. CXXC and LXXLL motifs are boxed. (B) Silver staining image of anti-FLAG-immunoprecipitated (α-FLAG) proteins from HEK293 cells transfected with a FLAG-E6^E7 expression vector or an empty vector for 48 h. Proteins identified by LC-MS/MS are indicated on the right. (C and D) Western blot (WB) verification of the specific binding of endogenous HSP90α, HSP90β, and GRP78, but not HSP70 and PRMT5, to FLAG-E6^E7 by anti-FLAG IP (C) or the binding of ectopic HA-HSP90β or Myc-GRP78 to FLAG-E6^E7 by anti-HA IP (D, top panel) or anti-Myc IP (D, lower panel). Cell lysate from an empty vector transfection served as an IP control in panel C, and Sepharose beads without antibody were used as an IP control in panel D. HEK293 cell lysates prepared at 48 h of transfection were used in all IP and Western blot assays. Western blotting was conducted with an antibody specific for an endogenous protein or a tag-specific antibody for an ectopic protein tag. (E) Relative mRNA levels of HSP90α, HSP90β, and GRP78 in normal cervix and cervical cancer tissues. *, P < 1 × 10−4 (Student’s t test between two groups). Normal cervix specimens, n = 5; cervical cancer tissue specimens, n = 40. (F) Western blot analysis of HSP90α, HSP90β, GRP78, and GRP78va (a GRP78 splice isoform) (31) for normal cervix specimens (lanes N1 and N2) and cervical cancer tissues (lanes T1, T2, T3, T4, and T5). β-actin was used as an internal loading control.
In this study, we report for the first time an important function of E6^E7 in cervical carcinogenesis. E6^E7 interacts with E6 or E7 and stabilizes E6 and E7 oncoproteins from proteasome-mediated degradation. The 90-kDa heat shock protein (HSP90) and 78-kDa glucose-regulated protein (GRP78) interact with E6^E7 and cooperatively contribute to its function. Furthermore, we found that E6^E7 is expressed not only in HPV16 but also in a subset of high-risk HPVs.
RESULTS
HSP90α, HSP90β, and GRP78 are HPV16 E6^E7-interacting proteins.
To uncover the function of E6^E7, we first investigated proteins that interact with FLAG-tagged E6^E7 (FLAG-E6^E7) by anti-FLAG immunoprecipitation (IP) and nano-liquid chromatography–tandem mass spectrometry (LC-MS/MS). Bands of specific proteins pulled down by FLAG IP in FLAG-E6^E7-expressing cells, compared with protein bands in control vector-transfected cells in a silver-stained gel, were analyzed by LC-MS/MS and subsequently identified as HSP90α, HSP90β, and GRP78 (Fig. 1B; see also Table S1 in the supplemental material). Co-IP and Western blotting confirmed the specific interactions of these three proteins with HPV16 E6^E7 (Fig. 1C and D), while other proteins, including HSP70 and PRMT5, failed to be validated (Fig. 1C). HSP90α, HSP90β, and GRP78 are host chaperone proteins that regulate the stability of their specific client proteins by assisting their noncovalent folding and assembly. HSP90α, HSP90β, and GRP78 promote tumor progression by specifically targeting oncogenic proteins, such as HER2 (23), Raf-1 (24), and Akt (25), and are frequently up-regulated in most cervical cancer tissues (Fig. 1E and F) and other types of cancers (26, 27). E6^E7 carries a nuclear localization signal (NLS) from the N-terminal half of E6 (28) and a nuclear export signal (NES) from the C-terminal half of E7 (29) (Fig. 1A). However, the majority of HPV16 E6^E7 in transfected HEK293 cells is found in the cytoplasm, as has been seen for HSP90 and GRP78 (30,–33), with a small fraction in the nucleus (see Fig. S1A in the supplemental material). Subsequent efforts failed to characterize E6^E7 in destabilizing p53 and pRB or in preventing cell proliferation by overexpression in HPV-negative HEK293 or HCT116 cells (Fig. S1B and S1C).
HSP90β and GRP78 promote the protein stability of HPV16 E6, E7, and E6^E7, and E6^E7 augments this function through protein-protein interactions.
Given that HSP90α, HSP90β, and GRP78 are E6^E7-interacting proteins, we speculated that these chaperone molecules may affect the steady-state level of E6^E7. The levels of E6^E7 protein and mRNA expression were examined in HEK293 cells with or without coexpression of HSP90β or GRP78 from available expression vectors. Although the transfection efficiencies of the plasmids in these groups were similar as determined by neomycin phosphotransferase II (NPT II) expression from the neomycin resistance gene in the plasmid, we found an approximately >5-fold increase in the level of E6^E7 protein, without a change of the E6^E7 mRNA level, when coexpressed HSP90β or GRP78 was ~40% or ~5-fold above its respective endogenous level (Fig. 2A; see Fig. S2A in the supplemental material). Since E6^E7 has the N-terminal half of E6 fused with the C-terminal half of E7, coexpression of HSP90β or GRP78 was also examined in parallel for the steady-state levels of HPV16 E6 and E7 in HEK293 cells. Both HSP90β and GRP78 were found to stabilize HPV16 E6 (Fig. 2B) and E7 (Fig. 2C) proteins but not their corresponding mRNAs. However, compared with GRP78 or HSP90β, E6 appears to be a better responder to E6^E7 than does E7 by transient expression assay. HSP90β and GRP78 were also found to stabilize E6*I protein, a truncated isoform of E6 (Fig. 2D). In either case, HSP90β and GRP78 exhibited no effect on cellular β-actin protein and GAPDH mRNA (Fig. 2A to D).
HPV16 E6^E7, HSP90, and GRP78 promote the protein stability of HPV16 E6 and E7. (A to E) ΗSP90β and GRP78 increase the protein but not mRNA levels of HPV16 E6^E7, E6, and E7 in HEK293 cells. Following cotransfection with the indicated vectors in each panel for 48 h, HEK293 cells were analyzed by Western blotting for protein expression levels of FLAG-E6^E7 (A), GFP-E6 (B), GFP-E7 (C), GFP-E6*I (D), or GFP (E) and by Northern blotting for the expression levels of corresponding mRNAs. An empty vector without any tagged protein expression served as a control in each transfection, and neomycin phosphotransferase II (NPT II, a neomycin resistance gene product) served as a control for plasmid transfection and expression efficiency. LE, longer exposure; SE, shorter exposure. After normalization with β-actin in a Western blot or with GAPDH in a Northern blot, the relative (fold) changes in the levels of the corresponding protein or mRNA in HEK293 cells cotransfected with HA-HSP90β, Myc-GRP78, or HA-E6^E7 over the levels in cells transfected with an empty control vector are shown at the bottom of each panel in bar graphs. Error bars indicate standard deviations from two different blots. (F) E6^7 increases the E7 protein level in primary human foreskin keratinocytes. The keratinocytes were cotransfected with 4D-Nucleofector and GFP-E6 plus an HA-E6^E7 expression vector or an empty control vector for 48 h and analyzed by Western blotting. (G) HA-HSP90β, Myc-GRP78, and HA-E6^E7 increase GFP-E7 protein level in HeLa cells. HeLa cells were cotransfected with the indicated vectors for 48 h and analyzed by Western blotting. (H) The effect of HPV16 E6^E7 on E6 and E7 stability relies on HSP90. HEK293 cells were transfected twice, with a 48-h interval, with siRNAs specific for HSP90α and -β isoforms or a nontargeting control siRNA (−) for 4 days. During the second siRNA transfection, the cells were cotransfected with an HA-E6^E7 expression vector or a control vector (p3×FLAG-CMV14) in combination with a GFP-E6 or -E7 expression vector for 24 h and analyzed by Western blotting for HSP90α and -β knockdown efficiency with a pan-HSP90 antibody, E6^E7 with an anti-HA antibody, or HPV16 E6 or E7 with an anti-GFP antibody. β-actin served as a loading control. (I to K) HA-E6^E7 increased protein levels, and the levels of MG132-stabilized GFP-E6, GFP-E7, and FLAG-E6^E7 are comparable in HEK293 cells at 48 h of transfection. The cell cotransfections were conducted with a GFP-E6 (I), GFP-E7 (J), or FLAG-E6^E7 (K) expression vector along with an HA-E6^E7 expression vector or control vector. For MG132 treatment, the cells transfected with a GFP-E6, GFP-E7, or FLAG-E6^E7 expression vector were treated with MG132 (10 µM) or an equivalent amount of dimethyl sulfoxide (DMSO) 6 h prior to being harvested for Western blotting with an anti-GFP or anti-FLAG antibody. β-actin served as a sample loading control.
Unexpectedly, the most drastic changes in the protein steady-state levels of E6 (~10-fold), E7 (~5-fold), and E6*I (~10-fold) were found when they were coexpressed with E6^E7 in the presence of endogenous HSP90β and GRP78 in HEK293 cells (Fig. 2B to D). These changes were protein target specific because the coexpression of HSP90β, GRP78, or E6^E7 did not affect the level of green fluorescent protein (GFP) or mRNA (Fig. 2E) and could be reproducible with E7 in human primary foreskin keratinocytes (Fig. 2F) and HeLa cells (Fig. 2G). The effect of E6^E7 on the E6 or E7 protein level in the cells could be greatly reduced when endogenous HSP90 was knocked down, resulting in a low level of E6^E7 expression (Fig. 2H). Data suggest that E6^E7 requires endogenous chaperones for its expression and activity. By using a proteasome inhibitor, MG132, we found that MG132 treatment of HEK293 cells for 6 h could increase HPV16 E6, E7, and E6^E7 to comparable levels, as observed from E6^E7 coexpression (Fig. 2I to K). The smaller size of an additional uncharacterized E6 band appeared in the presence of hemagglutinin (HA)-E6^E7 but not in the presence of MG132 (Fig. 2I), suggesting its correlation with E6^E7 expression.
Recent reports showed that proteins with acidic LXXLL motifs bind to and stabilize HPV16 E6 oncoprotein (34, 35) and that the hydrophobic surfaces and two CXXC zinc-binding motifs in HPV16 E7 are responsible for E7 homodimerization (36). Because those structures are retained in HPV16 E6^E7 (Fig. 1A), we proposed that E6^E7 might interact with E6 and E7. Indeed, E6^E7 was found to bind both HPV16 E6 and E7 oncoproteins or vice versa by co-IP (Fig. 3A; see also Fig. S2B, left panel, in the supplemental material). Similarly, GRP78, which interacts with E6^E7 (Fig. 1C and D), was capable of pulling down both E6 and E7 or vice versa by co-IP in the absence of E6^E7 (Fig. 3B; see Fig. S2B, right panel, in the supplemental material). In contrast, HSP90β, which also interacts with viral E6^E7 (Fig. 1C and D), showed no binding activity to HPV16 E6 or E7 (Fig. 3C) nor to GRP78 (Fig. S2C), indicating that the effect of HSP90β coexpression on the steady-state level of E6 or E7 is indirect.
HPV16 E6 and E7 oncoproteins are proteins that interact with E6^E7 and GRP78. (A to C) HPV16 E6 and E7 interact with E6^E7 (A) and GRP78 (B), but not with HSP90β(C). HEK293 cells were transfected with FLAG-E6^E7 (A), Myc-GRP78 (B), or HA-HSP90β (C) along with GFP-E6 or -E7 (A to C) for 48 h. The cell lysates were immunoprecipitated with the corresponding antibody, as indicated. Rabbit IgG was used as a negative control for each anti-GFP IP, and Sepharose beads without antibody served as controls for anti-FLAG, anti-HA, and anti-Myc IP. The interacting proteins in coimmunoprecipitations or in the input were examined by Western blotting (WB) with anti-GFP for GFP-E6 or -E7 (A to C), anti-FLAG for FLAG-E6^E7 (A), anti-Myc for Myc-GRP78 (B), or anti-HA antibody for HA-HSP90β(C). hc, IgG heavy chain. (D) E6^E7 is a self-interacting protein. HEK293 cells were cotransfected with a FLAG-E6^E7 and an HA-E6^E7 expression vector or an empty (−) control vector for 48 h. The cell lysates were blotted for the expression level of each protein (input panel) and immunoprecipitated with anti-FLAG antibody. The proteins pulled down by IP were blotted with an anti-FLAG antibody for FLAG-E6^E7 or anti-HA antibody for HA-E6^E7. β-Tubulin served as a sample loading control.
The self-interaction of E6^E7 was confirmed by co-IP of HA-E6^E7 and FLAG-E6^E7 (Fig. 3D). In addition, the coexpression of HA-E6^E7 and FLAG-E6^E7 was found to greatly increase the level of each protein, indicating that the self-interaction of E6^E7 is important for its in vivo accumulation, as previously reported for HPV16 E6 and E7 (34, 37).
HPV16 E6^E7 increases the half-lives of E6 and E7.
To investigate how HPV16 E6^E7 increases the steady-state level of E6 or E7 when coexpressed in HEK 293 cells, we transfected HEK293 cells with HPV16 E6 or E7 in the presence or absence of E6^E7 and then treated the cells with cycloheximide (CHX) in a time course manner, followed by Western blotting of the cell lysates for E6 or E7 protein. CHX is an inhibitor of protein biosynthesis in eukaryotic organisms by blocking protein translational elongation (38), which enables us to compare the half-lives of E6 and E7 protein in the presence and absence of E6^E7. As shown in Fig. 4, the half-life of HPV16 E6 is ~45 min, and that of E7 is ~56 min. In these carefully controlled CHX experiments with NPT II as a transfection efficiency control and β-tubulin as a loading control for each sample, we found that E6^E7 is capable of extending the half-life of E6 to ~119 min and that of E7 to ~154 min. These data indicate that E6^E7 interacts with and stabilizes E6 or E7 protein.
HPV16 E6^E7 stabilizes E6 and E7 and prolongs the protein half-life. HEK293 cells were cotransfected with a GFP-E6 (A) or GFP-E7 (B) expression vector together with an HA-E6^E7 or an empty control vector. After 16 h of cotransfection, the cells were treated with 0.1 mg/ml of cycloheximide (CHX) for the indicated times (0, 0.5, 1, 2, 4, and 8 h) before the sample collection for Western blotting with an anti-GFP antibody, anti-β-tubulin for sample loading, or anti-NPT II antibody for plasmid transfection efficiency. The half-life (t1/2) of GFP-E6 or GFP-E7 was determined from a line plot analysis according to the following formulas, with its expression level at time zero being set to 100% and the two lines (x, y) crossing the 50% decay point (y + 0.5): for GFP-E6 with the control vector, y = −0.9583x + 1.2217; for GFP-E6 with HA-E6^E7, y = −0.3683 + 1.2291; for GFP-E7 with the control vector, y = −0.6238x + 1.0828; and for GFP-E7 with HA-E6^E7, y = −0.2802x + 1.2179, where x is the CHX treatment time (h) and y is the relative protein expression levels of GFP-E6 (A) or GFP-E7 (B). Black squares, protein expression level of GFP-E6 (A) or GFP-E7 (B) in HEK293 cells cotransfected with a control vector; red squares, protein expression level of GFP-E6 (A) or GFP-E7 (B) in HEK293 cells cotransfected with an HA-E6^E7 expression vector.
HSP90β, GRP78, and E6^E7 are required to maintain the steady-state level of E6 and E7 in HPV16-positve cervical cancer cells.
We next looked into the functional regulation of E6 and E7 oncoproteins by endogenous HSP90β, GRP78, and E6^E7 in HPV16-infected cervical cancer cells. We first applied 17-N-allylamino-17-demethoxygeldanamycin (17-AAG), an HSP90 ATPase inhibitor for both HSP90α and HSP90β that is in clinical trials for many types of cancer (27). In the presence of 17-AAG, HPV16-positive CaSki cells showed a decreased level of HPV16 E6 expression, with the increased stability of p53 indicative of E6 (17, 39, 40) and E7 (Fig. 5A) reduction but with no change in the expression of full-length E6 (unspliced), E7 (E6*I), or E6^E7 mRNA (Fig. 5B). This was expected because HSP90 and E6^E7 together were found to stabilize viral E6 and E7 better than HSP90 or E6^E7 alone (Fig. 2B and C), despite the finding that HSP90 itself does not interact with E6 or E7 (Fig. 3C). Consistently with the reduced expression of E6 and E7 oncoproteins, 17-AAG treatment blocked the growth of CaSki cells (Fig. 5C). GRP78 knockdown in CaSki cells also reduced the protein levels of both E6 (indicated by p53 increase) and E7 (Fig. 5D) and inhibited CaSki cell growth (Fig. 5E). All together, these data indicate that HSP90 and GRP78 are two chaperone proteins important for maintaining the steady-state levels of viral E6 and E7 oncoproteins for their oncogenic properties.
HSP90, GRP78, and E6^E7 are required to maintain a steady-state level of E6 and E7 in HPV16-positive CaSki cells. (A to C) A functional HSP90 is required for the stability of HPV16 E6 and E7 and proliferation of cervical cancer cells. HPV16-positive CaSki cells treated with 5 µM 17-AAG or DMSO for 48 h were examined by Western blotting (A), RT-PCR (B), and a cell proliferation assay (C). p53 was used to indicate E6. GAPDH served as a loading control in RT-PCR. (D and E) Knockdown of GRP78 expression in CaSki cells destabilizes E6 and E7 and inhibits cell growth. CaSki cells treated twice with 40 nM nontargeting siRNA (si-NS) or GRP78 siRNA (si-GRP78) for 96 h were examined by Western blotting (D) and a cell proliferation assay (E). (F to I) Knockdown of E6^E7 expression in CaSki cells destabilizes viral E6 and E7 and prevents cell growth. An E6^E7-specific siRNA (si-E6^E7) for the nt 226-to-nt 742 splice junction (F) was designed to silence E6^E7 expression without affecting other E6 splice isoform RNAs, as shown by RT-PCR (G). CaSki cells transfected twice with 40 nM si-NS or si-E6^E7 at 48-h intervals for 96 h were examined for E6 (p53) and E7 expression (H) and cell proliferation (I). See the details in panels A to C. (J and K) Overexpression of E6^E7 in CaSki cells increases E7 stability (J) and promotes cell proliferation (K). CaSki cells were transfected twice (for Western blotting at day 5) or three times (for cell proliferation at day 7) with a FLAG-E6^E7 or an empty vector at 24-h intervals. (L and M) Overexpression of E6^E7 has no effect on C33A, an HPV-negative cervical cancer cell line containing mutations in both p53 and pRB. C33A cells transfected with a FLAG-E6^E7 or an empty vector as described for panels J and K served as a cell line control to CaSki cells, and results were analyzed by Western blotting (L) and a cell proliferation assay (M) at days 5 and 7, respectively. *, P < 0.05; **, P < 0.01. Nonsignificance (N/S), P ≥ 0.05 by Student’s t test (C and E, I and K, and M). (A, D, H, J, and L) β-Actin served as a sample loading control.
Functional regulation of HPV16 E6 and E7 oncoproteins by E6^E7 was investigated by specific knockdown of E6^E7 expression, although it is difficult for us to detect the E6^E7 protein as well as the E6, E6*I, and E6*II proteins in cervical cancer cell lines (see Fig. S3 in the supplemental material). We designed a small interfering RNA (siRNA) targeting to the splice junction region covering the nt 226 5′ ss and nt 742 3′ ss (si-E6^E7) (Fig. 5F) to knock down E6^E7 expression in HPV16-positive CaSki cells. We found that this si-E6^E7 exhibited a high efficiency and specificity in knocking down E6^E7 RNA (nt 226 to 742) in CaSki cells, without affecting full-length E6 RNA (unspliced), other alternatively spliced isoform RNAs, such as E6*I and E6*II, from HPV16 early transcripts, or cellular GAPDH (glyceraldehyde-3-phosphate dehydrogenase) RNA (Fig. 5G). Western blotting showed that the protein levels of both E6 and E7 in CaSki cells were decreased by the knockdown of E6^E7 (Fig. 5H, where E6’s decrease is indicated by p53’s increase), resulting in cell proliferation retardation (Fig. 5I). Data suggest that the expression of E6^E7 is important for the steady-state levels of both E6 and E7 oncoproteins in CaSki cells. This assumption was further validated by overexpression of E6^E7 in CaSki cells, where E7 oncoprotein could be stabilized upon E6^E7 overexpression, leading to a decrease in pRB protein level and an increase in CaSki cell proliferation (Fig. 5J and K). E6^E7 overexpression displayed no effect on the growth of C33A cells, a cervical cancer cell line without HPV infection (Fig. 5L and M) but carrying mutant p53 and pRB (41).
Conservation of E6^E7 among other high-risk HPVs.
We next investigated alternative RNA splicing to produce E6^E7 RNA from polycistronic early pre-mRNAs of other high-risk HPVs. The first step of RNA splicing by the cellular splicing machinery is to recognize a branch point sequence (BPS) and a polypyrimidine tract upstream of a 3′ ss, respectively, by SF1 and U2AF65, and of a 5′ ss by snRNP U1, followed by the stable association of U2 snRNP to the BPS to proceed to catalytic steps (42, 43). Although the production of the HPV16 E6^E7 splice isoform takes place from a 5′ ss at nt 226 to a 3′ ss at nt 742, the BPS governing the first step of splicing was unknown. By using a lariat reverse transcription (RT)-PCR technique in the presence of SuperScript II reverse transcriptase, we were able to amplify a splicing intermediate lariat structure in which the 5′ phosphate of the guanosine at the 5′ end of the intron is linked to the 2′ hydroxyl group of the adenosine at the BPS branch site (BS) to form a 5′-to-2′ phosphodiester linkage (44, 45) (Fig. 6A). Since this 5′-2′ link makes the lariat in a circular form and the SuperScript II reverse transcriptase may read through the 5′-2′ phosphodiester bond, two paired primers in opposite directions designed to detect the lariats would be able to amplify a product running through the link with a nucleotide substitution at the branch site (Fig. 6A, right panel). This was achieved by using HeLa nuclear extract for in vitro RNA splicing of HPV16 E6^E7 RNA (Fig. 6B, left panel) and then lariat RT-PCR of the in vitro-spliced products (Fig. 6B, right panel). Gel purification, cloning, and sequencing of the single nested-PCR product (Fig. 6B, right panel) identified two adenosines, one at nt 716 (4/10 clones) and the other at nt 718 (4/10 clones); these are two major alternative branch site adenosines for splicing of the HPV16 E6^E7 RNA. We also found that an adenosine either at nt 719 or at nt 721 might serve as a minor site (Fig. 6C; see Fig. S4 in the supplemental material).
Mapping of RNA cis elements for splicing of HPV16 E6^E7 pre-mRNA. (A) Diagram of a lariat RT-PCR strategy to map the BPS responsible for splicing of E6^E7 pre-mRNA, with the indicated primers (F1 to F2 and R1 to R4). (B) Validation of in vitro splicing (2 h) of E6^E7 pre-mRNA by RT-PCR. The left panel indicates the fully spliced products at a size of 136 nt. The right panel indicates the products of lariat RT-PCR at a size of ~200 nt. Combinations of primers used in RT-PCR are indicated in the box below. Lane M, molecular size markers. (C) Summary of the mapped branch sites by lariat RT-PCR. (D) Introduction of the A-to-G mutation (mt-1, mt-2, mt-3, and mt-4) into the mapped branch sites for in vitro RNA splicing. Nucleotides identical to the wild-type (wt) nucleotide are indicated by dots. (E) Reconstitution of E6^E7 RNA splicing in vitro. Following the splicing reaction in HeLa cell nuclear extract for the indicated times (h), the spliced products were analyzed in a 6% polyacrylamide gel with 7.5 M urea. Identities of each band are indicated on the right. (F) E6^E7 pre-mRNAs with wt or mutant branch sites were used for a 2-h reaction of in vitro splicing. Relative E6^E7 splicing efficiencies (percentages) were calculated as described previously (12) and are indicated in the bottom. The identities of each band are indicated at the right. (G) Diagram of the minigene vectors to express E6^E7 pre-mRNAs with a wt or mutant branch site (mt-1, mt-2, or mt-3). (H) RT-PCR was performed on total RNA of HEK293 cells transfected with E6^E7 minigenes or a control vector for 24 h to detect unspliced and spliced E6^E7 mRNAs (for the upper panel, primers F3 and R1 were used). GAPDH served as a loading control (lower panel). The identity of each band is indicated at the right. *, a nonspecific amplicon. Relative E6^E7 splicing efficiencies (percentages) are indicated at the bottom.
Subsequently, we introduced point mutations (A→G) at these positions (Fig. 6D) and compared their effects on the in vitro splicing of 32P-labeled HPV16 E6^E7 pre-mRNA (Fig. 6E). As shown in Fig. 6F, an A-to-G mutation at nt 716 (lanes mt-1) suppressed the splicing efficiency of E6^E7 pre-mRNA by 68%, but introduction of the same mutation at nt 718 reduced the splicing efficiency by only 30% (lanes mt-2). However, simultaneous introduction of the A-to-G mutation at both nt 716 and nt 718 (lanes mt-3) or together with the A-to-G mutations at nt 719 and nt 721 (lanes mt-4) completely blocked the in vitro splicing of E6^E7 pre-mRNA. Based on these observations, we conclude that splicing of HPV16 E6^E7 pre-mRNA takes place by use of two alternative branch sites, of which nt 716A serves as a major site and nt 718A serves as a minor site. The same results were also seen by in vivo splicing assays in HEK293 cells transfected with an HPV16 E6^E7 minigene (Fig. 6G and H). We further confirmed that the A-to-G mutation at nt 716 (lanes mt-1) completely abrogated E6^E7 splicing in HEK293 cells but that the same mutation at nt 718 (lanes mt-2), although it also reduced splicing greatly, retained 20% of the splicing efficiency of the wild-type (wt) construct (Fig. 6H). As expected, a combination of both mutations in the two mapped branch site adenosines of the HPV16 E6^E7 minigene (mt-3) were detrimental to in vivo E6^E7 splicing (Fig. 6H).
To date, HPV16, among all known HPVs, appears to be the only genotype to express E6^E7. However, we found that the E6^E7-splicing cis elements (5′ ss, BS, polypyrimidine tract, and 3′ ss) identified in HPV16 are highly conserved among 11 of 36 α-HPVs (13 from high-risk, 4 from possibly high-risk, 8 from low-risk, and 11 from risk-undetermined HPVs), of which 8 are high-risk HPVs (HPV16, -18, -31, -39, -45, -56, -59, and -73), 1 is possibly a high-risk HPV (HPV30), and 2 are low-risk HPVs (HPV42 and HPV70) (Fig. S5A and S6). Interestingly, this alignment analysis also showed that HPV30, -42, and -70, like HPV16 and HPV18, contain an intron in the E6 ORF (Fig. S5B), which is characteristic of high-risk, but not of low-risk, HPVs (7). The other remaining HPVs analyzed, including high-risk HPV33, -35, -51, -52, and -58 (46), appear to lack either a 5′ ss, a 3′ ss, or a visible BS for E6^E7 splicing in the corresponding regions of HPV16, indicating that they express no E6^E7 (Fig. S5A). In the HPVs with the conserved E6^E7-splicing cis elements, the N-terminal half of the E6 ORF could hypothetically be spliced in frame into the C-terminal half of the E7 ORF, as seen in HPV16 (Fig. S6). Based on these analyses, we examined and confirmed by RT-PCR the expression of HPV18 E6^E7 in HPV18-positive HeLa cells but not in HEK293 cells or in HPV16-positive CaSki cells (Fig. 7A and B). Cloning and sequencing of the RT-PCR products further revealed that HPV18 E6^E7 is a splicing product from nt 233 to nt 791 (Fig. 7C). In addition to its expression in cervical cancer cell lines, E6^E7 expression from raft tissues with productive HPV16 or HPV18 infection (47) could be verified by RT-PCR (Fig. 7D). However, we were unable to detect HPV18 E6^E7 protein by using an antibody against the HPV18 E6 N terminus that recognizes ectopically expressed HPV18 E6 but not E6*I in HEK293 cells or native HPV18 E6 and E6*I from HPV18-infected raft tissues or HeLa cells.
Expression of HPV18 E6^E7 in an HPV18-infected cell line and raft tissues. (A) Diagram of alternative RNA splicing from the nt 233 5′ ss to the nt 791 3′ ss to produce E6^E7 mRNA from HPV18 early transcripts. The ORFs of the HPV18 E6 and E7 oncogenes are indicated at the top. Primers specific for HPV18 E6^E7 mRNA detection by RT-PCR are indicated below the pre-mRNA. (B) Detection of HPV18 E6^E7 from HPV18-positive HeLa cells and HPV16 E6^E7 from HPV16-positive CaSki or SiHa cells by RT-PCR. HPV-negative HEK293 cells and GAPDH mRNA served as controls. (C) Determination of the HPV18 E6^E7 splice junction by sequencing of RT-PCR products gel purified from the experiment whose results are shown in panel B. (D) Detection of HPV18 E6^E7 and HPV16 E6^E7 in the HVK raft tissues with the corresponding virus infection. (E) E6^E7 plays a central role in HSP90 and GRP78 regulation of HPV16 E6 and E7 protein stability through protein-protein interactions. HSP90 and GRP78 are also involved in the stability of HPV16 E6 and E7, either indirectly (dashed arrow) or directly (solid arrow).
Quantification of E6 and its splice variants E6*I, E6*II, and E6^E7 in HPV16-positive cervical cancer cell lines and tissues.
The HPV16 early pre-mRNA derived from its early promoter P97 is a polycistronic transcript bearing two introns and three exons. In HPV16-infected cells, this viral pre-mRNA undergoes extensive alternative RNA splicing in order to express other viral ORFs downstream. One of the two introns is positioned in the E6 ORF and contains three alternative 5′ ss (nt 191, 221, and 226) and three alternative 3′ ss (nt 409, 526, and 742). Although the nt 226 5′ ss and the nt 409 3′ ss are two splice sites preferentially selected over the other splice sites, crossing over the intron to excise a minimal length of the intron in RNA splicing (12), splicing of the intron in the E6 ORF will disrupt the integrity of the E6 ORF and prevent the expression of full-length E6 protein. Thus, we quantified the relative usage of E6^E7 over that of unspliced E6 and two other well-described E6 splice isoforms, E6*I derived from splicing the nt 226 5′ ss to the nt 409 3′ ss and E6*II derived from splicing the nt 226 5′ ss to the nt 526 3′ ss, in HPV16-positive cervical cancer cells and cervical cancer tissues by splicing junction-specific TaqMan real-time RT-PCR (Fig. S7). As shown in Tables 1 and S2, we found that the copy number of HPV16 E6^E7 mRNA, as with E6, E6*I, and E6*II, varies from one cell line or tumor tissue to another and that its proportion is much lower (0.2% to 18.4%) than those of the unspliced, full-length E6 RNA (13.2% to 55.9%), E6*I (29.5% to 79.9%), and E6*II (4.6% to 34.6%) in the examined cervical cancer cell lines or tissues.
TABLE 1
Relative quantification of E6^E7 alternative-splicing productsa
| Sample name | Sample type | % retentionb | % of an E6 splice isoform from: | ||
|---|---|---|---|---|---|
| nt 226 to 409 | nt 226 to 526 | nt 226 to 742 | |||
| CaSki | Cervical cancer cell line | 18.4 | 73.6 | 7.6 | 0.5 |
| SiHa | Cervical cancer cell line | 54.3 | 39.8 | 5.6 | 0.3 |
| 20861 | CIN cell line | 13.2 | 79.9 | 6.6 | 0.2 |
| 20863 | CIN cell line | 38.3 | 56.8 | 4.6 | 0.4 |
| T1 | Cervical cancer | 17.0 | 40.2 | 24.4 | 18.4 |
| T2 | Cervical cancer | 28.4 | 57.0 | 13.7 | 0.9 |
| T3 | Cervical cancer | 55.2 | 37.4 | 7.1 | 0.2 |
| T4 | Cervical cancer | 18.4 | 53.8 | 26.9 | 0.9 |
| T5 | Cervical cancer | 19.5 | 45.7 | 34.6 | 0.2 |
| T6 | Cervical cancer | 42.0 | 45.8 | 11.6 | 0.7 |
| T7 | Cervical cancer | 55.9 | 33.2 | 10.4 | 0.4 |
| T8 | Cervical cancer | 24.2 | 29.5 | 32.8 | 13.5 |
| T9 | Cervical cancer | 20.2 | 62.0 | 16.3 | 1.5 |
DISCUSSION
In this study, we showed a crucial role of HPV16 E6^E7 in stabilizing E6 and E7 oncoproteins and evidenced its contribution to oncogenicity in cooperation with E6 and E7, although E6^E7 itself does not independently affect the stability of p53 or pRB. Our data also revealed that the stability of E6^E7, E6, and E7 requires HSP90 and GRP78, two host chaperones frequently overexpressed in cancer cells. E6^E7 interacts with GRP78, HSP90, and HPV16 E6 and E7. GRP78, but not HSP90β, interacts with E6 and E7. We found that the role of E6^E7 appears critical in stabilizing E6 and E7, with E6 being a better responder than GRP78 or HSP90β by transient expression. However, the E6^E7 enhancement of E6 or E7 stability becomes minimal in cells with a reduced expression of GRP78 or HSP90, where E6^E7 itself is unstable. We could not see greater loss of the protein stability when both chaperones were knocked down or inhibited. Instead, inhibition or knockdown of HSP90 increases the expression of GRP78 (48, 49). Because HSP90 does not interact with E6, E7, or GRP78, the effect of HSP90 on E6 and E7 stability must be exerted through other cofactors. In general, HSP90 collaborates with a large set of cochaperones in assembling a functional chaperone machinery to assist in client protein folding to the native state (50). Other chaperone proteins (HSP70, HSP40, and cyclophilin) in the regulation of HPV entry and genome replication have been reported in HPV11 (51), HPV16 (52), and HPV31 (53).
It is known that E6 and E7 are unstable unless they are associated with appropriate proteins. Recent studies indicated that E6 can be stabilized by heterodimerization with E6-AP through an acidic LXXLL motif (34). Other studies also suggested that both E6 and E7 may be stabilized through homodimerization in biochemical assays, which are mediated by two separated CXXC zinc-binding motifs within E6 and E7 proteins (36, 37, 54). Although HPV16 E6^E7, which retains two zinc-binding motifs, interacts with both HPV16 E6 and E7, how this protein-protein interaction contributes to stabilize E6 and E7 remains unknown. By interaction with E6 or E7, E6^E7 perhaps serves as a bridge to recruit HSP90 for better assembly of a functional chaperone machinery to facilitate E6 or E7 folding. Alternatively, E6^E7, by interaction with HSP90 and GRP78, may simply serve as a cofactor to promote chaperone activities, as reported for other HSP90 cofactors (50). Why both chaperone proteins are required for stabilization to occur remains to be understood. One possibility is that they are required for the different transitional states of the newly synthesized protein. Nevertheless, our study indicates that E6^E7 plays a central role in HSP90’s and GRP78’s regulation of the stability of E6 and E7 (Fig. 7E). Since HSP90 and GRP78 also stabilize E6^E7, this study provides the further evidence of a positive-feedback loop in infected cells to promote the steady-state levels of E6 and E7 oncoproteins.
Early studies of bovine papillomavirus 1 (BPV-1)-transformed mouse C127 cells showed that the BPV-1 E6 ORF could be spliced to the E7 ORF in frame, leading to the production of a novel E6^E7 mRNA to encode a hypothetical E6^E7 fusion protein with a size of 183 aa residues (71 aa residues from E6 and 112 aa residues from E7) (55, 56). Subsequent investigation of a cDNA derived from the E6^E7 mRNA showed no transformation activity in mouse C127 cells (56). Although the in-frame RNA splicing of the E6 ORF to the E7 ORF was identified in HPV16 in 2004 (16, 22), whether such a splicing event exists in other oncogenic HPVs was unknown, and its regulation by RNA cis elements was poorly understood. In this report, we mapped the branch point and polypyrimidine tract for the selection of the nt 742 3′ ss to produce E6^E7 mRNA and successfully identified the nt 716 and nt 718 adenosines as two alternative branch sites in the BPS to guide the usage of this 3′ ss. Subsequent clustering analysis revealed the conservation of these RNA cis elements from early HPV16 to HPV18, -30, -31, -39, -42, -45, -56, -59, -70, and -73 transcripts (Fig. S5). We confirmed this production of E6^E7 in HPV18-infected cells and cervical cancer cell lines by RT-PCR. HPV42 and HPV70 are two low-risk HPVs but could be detected in cervical intraepithelial neoplasia (CIN) lesions and anogenital cancer (57,–61). Our analysis indicates that both HPV42 and HPV70 contain an E6 intron leading to production of E6*I, a characteristic of high-risk HPVs (7). Thus, the presence of viral E6^E7 in HPV16, HPV18, or other high-risk HPV infection to stabilize viral E6 and E7 might provide oncogenic advantages to these HPVs over the other HPVs lacking E6^E7 expression in cervical-lesion progression.
An intriguing question in this study is how a low level of E6^E7 transcripts relative to E6- and E7-encoding transcripts could translate a sufficient level of protein to alter the level of E6 or E7 protein. Both E6 and E7 have very short half-lives of less than 60 min, but in the presence of viral E6^E7, these proteins’ half-lives may be increased to ~120 min for E6 and ~150 min for E7 (Fig. 4), indicating that the viral E6^E7 functions as a potent protein to promote the stability of E6 and E7. We confirmed this in HPV16-positive CaSki cells in which the expression of E6^E7, despite being at protein level undetectable by Western blotting with antibodies against HPV16 E6 (Fig. S3), is important for the stability of E6 and E7, because siRNA knockdown of E6^E7 expression in CaSki cells may trigger the instability of both E6 and E7 (Fig. 5). HPV16 E6 is another protein notorious for its difficulty of detection in HPV16-positive cell lines (Fig. S3), but it is very potent in the induction of p53 degradation (17, 39). Together, our data indicate that a potent protein may function well at a very low level and thus does not need to be highly expressed.
Although therapeutic vaccines specifically targeting oncogenic E6 or E7 have been under development (62,–64), the current treatment of cervical cancer is dependent on surgical procedures and conventional chemoradiotherapy. The major difficulty of targeting E6 and E7 is that E6 and E7 oncogenic activities are mediated through protein-protein interactions rather than enzymatic activities. Inhibition of E6 and E7 expression through siRNA or small peptides appeared effective in vitro, but these efforts were not designed for in vivo preclinical trials, and the siRNA or small peptides are difficult to deliver (17, 65, 66). However, our findings in this study provide a possible new strategy to induce viral E6 and E7 instability by using HSP90 and GRP78 inhibitors for the treatment of cervical cancers. It has been known that overexpressed HSP90 in tumor cells promotes the stability of many oncoproteins, such as Her-2, Akt, Raf-1, Cdk4, Cdk6, and Src (27). Small-molecule inhibitors that block the activity of HSP90 ATPase, which is essential for the function of HSP90, are currently under evaluation in clinical trials for various types of cancer (67,–71) but not for cervical cancer. Thus, our observations that HSP90 and GRP78 regulate viral E6 and E7 stability undoubtedly provide a scientific foundation for future treatment of cervical cancer by using chaperone inhibitors.
MATERIALS AND METHODS
Plasmids, oligonucleotide primers, antibodies, and inhibitors.
All mammalian expression vectors and oligonucleotide primers used in this report are summarized in Table S3 in the supplemental material. The siRNA duplex of 5′ GACGUGAG/UGUGACUCUAUU 3′ and 5′ UAGAGUCACA/CUCACGUCGUU 3′ was used as si-E6^E7 (oMA63). Monoclonal anti-HPV16 E7 (ED17) and anti-β-actin (Ac-15) antibodies and polyclonal anti-HPV18 E6 N-terminal region (N-17) and anti-GRP78 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-pan-HSP90 polyclonal antibody, anti-HSP90α monoclonal antibody (D1A7), anti-HSP90β monoclonal antibody (D3F2), anti-HSP70 monoclonal antibody (D69), and anti-PRMT5 polyclonal antibody were purchased from Cell Signaling Technology (Danvers, MA). Anti-FLAG (M2) monoclonal antibody, anti-β-tubulin monoclonal antibody (Tub2.1), and anti-c-Myc monoclonal antibody (9E10) were purchased from Sigma-Aldrich (St. Louis, MO). Anti-GFP polyclonal antibody, anti-GFP monoclonal antibody (JL-8), and anti-pRB antibody (G3-245) were from BD Biosciences (San Jose, CA, USA). Anti-p53 monoclonal antibody (DO-1) was purchased from Merck KGaA (Darmstadt, Germany). Anti-neomycin phosphotransferase II rabbit polyclonal antibody was from EMD Millipore (Billerica, MA, USA). Anti-HA (3F10) rat and (HA-7) mouse monoclonal antibodies were purchased from Roche Diagnostics (Basel, Switzerland) and Sigma-Aldrich, respectively. Anti-HPV16 E6 mouse monoclonal antibodies 6F4 and 3F8 were purchased from Euromedex (Souffelweyersheim, France). 17-AAG (17-N-allylamino-17-demethoxygeldanamycin, tanespimycin) was purchased from Merck KGaA. MG132 and cycloheximide were purchased from Sigma-Aldrich.
In vitro splicing and lariat RT-PCR.
In vitro transcription and an in vitro splicing assay of HPV16 E6E7 pre-mRNA (nt 104 to 881) with a U1 binding site at the 3′ end were performed as previously described with modifications (12). Sixty nanograms of pre-mRNA was applied for in vitro splicing in the presence of 40% HeLa nuclear extract and 3 mM MgCl2 at 30°C for the times indicated in Fig. 6B. The RT reaction with SuperScript II (Life Technologies, Carlsbad, CA) and PCR by AmpliTaq (Life Technologies) were performed with primers indicated in Fig. 6A and B for the ethanol-precipitated, in vitro-spliced RNA products. The lariat RT-PCR products were then subcloned into the pCR2.1 TOPO vector (Life Technologies) and sequenced. In vitro splicing assays were performed on 4.0 ng of 32P-labeled pre-mRNAs in the presence of 40% HeLa nuclear extract and 3 mM MgCl2 at 30°C for the times indicated in Fig. 6, followed by separation with 6% polyacrylamide gel with 7.5 M urea and exposure to a PhosphorImager screen. The image was captured using a Molecular Dynamics PhosphorImager Storm 860 and analyzed with ImageQuant software.
RT-PCR and TaqMan real-time RT-PCR.
RT-PCR was performed as described before (12). Briefly, total RNA was treated with Turbo DNase (Roche), followed by RT with a random hexamer primer and Moloney murine leukemia virus (MuLV) reverse transcriptase (Life Technologies). PCR was subsequently performed with AmpliTaq (Life Technologies). For real-time RT-PCR, a TaqMan probe with 5′-6-carboxyfluorecein (FAM) and 3′-carboxytetramethylrhodamine (TAMRA) and oligonucleotide primers were designed from the conserved sequence regions among HPV16 subtypes (Table S3). Total RNA (400 ng) from cervical cancer cells or tissues was used for each real-time RT-PCR, and relative copy numbers of each splice isoform were determined from the cycle threshold (CT) value of the isoform RNA against a standard curve created from the corresponding cDNA plasmids.
Transfection and immunoprecipitation.
Plasmid transfections (2 µg for each plasmid) were performed with FuGENE HD transfection reagent (Roche Diagnostics) for CaSki and C33A cells in a 6-cm dish and with lipod293 transfection reagent (version II) (SignaGen Laboratories, Gaithersburg, MD) for HEK293 and HeLa cells. siRNA (40 nM) was transfected with a LipoJet transfection kit (SignaGen Laboratories), with a 48-h interval if the second transfections were needed for an efficient knockdown. For human primary keratinocytes, 2 µg of each plasmid was transfected for 1 × 106 cells with a 4D-Nucleofector system (Lonza Cologne GmbH, Cologne, Germany) according to the nucleofection protocol designed for human keratinocytes. For the immunoprecipitation, cells were collected in RIPA buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 0.25% sodium deoxycholate, 0.5% NP-40). Following sonication, the cell lysate was treated with RQ1 DNase (Promega, Fitchburg, WI) and RNase A (Life Technologies) and incubated with antibody-conjugated Sepharose 4B (Sigma-Aldrich) for 1 h at 4°C. Precipitated products were then washed with RIPA buffer and eluted with a 2.5× SDS protein gel loading solution containing 10% β-mercaptoethanol for Western blotting.
Cell lines and cervical tissues.
CaSki cells and SiHa cells are cervical cancer cell lines with an integrated HPV16 genome (200 to 300 copies/cell and 1 to 2 copies/cell, respectively). HeLa cells are a cervical cancer cell line with an integrated HPV18 genome (10 to 50 copies/cell). W12-derived subclone cell lines, 20861 and 20863 cells, originated from a CIN I lesion and contain an integrated (20861 cells) and episomal (20863 cells) HPV16 genome (72). Primary human foreskin keratinocytes (C-001-5C) were purchased from Life Technologies and grown in the presence of J2 feeder cells (73). HPV-negative cervical cancer cell line C33A (mutant p53, mutant pRB), colon cancer cell line HCT116 (mutant ras, wt p53, wt pRB), and adenoviral E1b55k- and E1A-positive human embryonic kidney cell line HEK293 (wt p53, wt pRB) were also used in this study. Total RNAs from HPV16-positive cervical tissues (samples T1 to T9 in Table 1) and raft cultures with HPV16 or HPV18 infection were the RNAs left over from our previous studies (47). The mRNA expression profiling derived from 5 normal cervix samples and 40 cervical cancer tissue samples in Fig. 1E were obtained from Oncomine 3.0 (74). Protein extracts from normal cervix (lanes N1 and N2 in Fig. 1F) and HPV16-positive cervical cancer tissues (lanes T1 to T5 in Fig. 1F) were purchased from US Biomax (Rockville, MD).
Northern blotting.
Northern blotting was performed as previously described (75). Briefly, each 5 µg of total RNA was separated in 1% agarose gel with 1× MOPS (morpholinepropanesulfonic acid) buffer with formaldehyde. Separated RNAs were then capillary transferred onto a nylon membrane and cross-linked by UV. Membranes were then prehybridized and incubated with 32P-labeled probes overnight at 42°C. The following probes were used: oZMZ220 for the detection of HPV16 E6 and E6*I, oZMZ380 for HPV16 E7 and E6^E7, oZMZ296 for enhanced green fluorescent protein (EGFP), and oZMZ270 for GAPDH (Table S2). After exposure to a PhosphorImager screen, the radioactivity was captured and analyzed as described above.
Cycloheximide treatment.
For the cycloheximide-chase study, HEK293 cells at 1.5 × 106 in a 6-cm plate were transfected with 2 µg of pZMZ70 (GFP-E6) or pZMZ74 (GFP-E7) or with 2 µg of the pCMV3×FLAG14 empty vector or pMA48 (E6^E7-HA) for 16 h. Culture medium was then replaced with cycloheximide (0.1 mg/ml)- or 10 µM MG132-containing medium for the times indicated in Fig. 4. Cells were then collected in 2.5× SDS protein sample buffer containing 10% β-mercaptoethanol for Western blotting.
WST-8 cell proliferation assay.
To detect dehydrogenase activities in living cells, a WST-8 cell proliferation assay was performed with cell counting kit 8 from Dojindo Molecular Technologies (Rockville, MD), as described previously (76, 77). Briefly, cells were incubated in culture medium with 10% WST-8 cells for 40 min at 37°C. The cell culture media were then measured in triplicate at 450-nm absorbances, and the cell viability was calculated.
LC-MS/MS analysis.
Cell lysates of HEK293 cells transfected with pMA15 (FLAG-E6^E7) or the p3×FLAG CMV14 (Sigma) empty vector for 48 h were treated with RQ1 DNase (Promega) and RNase A (Life Technologies) before immunoprecipitation with anti-FLAG beads (Sigma). Immunoprecipitation products were separated in 4% to 20% Tris-glycine gel (Bio-Rad) and silver stained with SilverQuest (Life Technologies). Specific protein bands in IP products of FLAG-E6^E7 pulldown experiments were excised from the gel with silver staining. Following destaining, excised gel fragments were dry frosted and rehydrated with trypsin solution. Trypsin-digested peptides were then purified for LC-MS/MS analysis by a service provider, ProtTech (Phoenixville, PA).
Immunocytostaining and confocal microscopy.
Cells were fixed with 4% PFA in phosphate-buffered saline (PBS), permeabilized by 0.1% Triton X-100, and incubated with 5% bovine serum albumin (BSA) for nonspecific blocking. Subsequently, cells were incubated with a mouse anti-FLAG (M2) monoclonal antibody, followed by incubation with anti-mouse IgG-Alexa 488 and Hoechst stain. Fluorescent and differential interference contrast (DIC) images were obtained with a Zeiss LSM510 Meta confocal microscope (Carl Zeiss MicroImaging, Inc., Thornwood, NY).
SUPPLEMENTAL MATERIAL
Table S2
Summarized data from HPV16 E6 splice isoform-specific real-time RT-PCR.
Table S3
List of plasmids (A) and DNA oligonucleotides (B) used in this study.
Figure S1
Cytoplasmic E6^E7 expression in HPV16-negative cells does not affect the stability of p53 and pRB proteins or cell proliferation. (A) HPV16 E6^E7 is present mainly in the cytoplasm of HEK293 cells. The cells were transfected with a FLAG-E6^E7 expression vector for 48 h and immunostained with mouse anti-FLAG and then anti-mouse IgG-Alexa 488 (green) for E6^E7 or stained with Hoechst dye (blue) for DNA. Scale bars, 10 µm. (B and C) HPV16 E6^E7 overexpressed in HEK293 and HCT116 cells does not affect p53 and pRB stability or cell proliferation. Both cell types, which express wild-type p53 and pRB, were transfected with a control empty vector or a FLAG-E6^E7 expression vector. The cells were analyzed by Western blotting 72 h after transfection (B) and counted in triplicate in WST-8 assays (C). SE, shorter exposure; LE, longer exposure. Download
Figure S2
(A) Expression levels of exogenous versus endogenous GRP78 and HSP90β proteins in HEK293 cells. HEK293 cells transfected with Myc-GRP78 (left panel) or HA-HSP90β (right panel) were analyzed by Western blotting with an anti-GRP78 or anti-HSP90β antibody. The total amount of exogenous Myc-GRP78 or HA-HSP90β protein was calculated relative to the endogenous protein in the cells transfected with an empty vector 48 h after transfection of the corresponding plasmids. (B) HPV16 E6 and E7 preferentially bind to GRP78 over E6^E7. Densitometric analyses were performed on 10% of input cell lysates, and coimmunoprecipitated products were blotted with the corresponding tag-specific antibodies for ectopic protein tags. The density ratio of each immunoprecipitated product over the protein in the input sample (IP/input) was calculated and is shown to the right of each Western blot. IP/input values were then compared between FLAG-E6^E7 (left panel) or Myc-GRP78 (right panel) and GFP-E6 or -E7 to determine the relative affinities of E6 and E7 for each protein. (C) HSP90 and GRP78 do not interact with each other. HEK293 cells were transfected with an HA-HSP90β or Myc-GRP78 expression vector, and 24 h after transfection, the cell lysates were prepared for co-IP using an anti-HA antibody to pull down endogenous GRP78 (left panel) or using an anti-Myc antibody to pull down endogenous HSP90α/β (right panel). Unconjugated Sepharose beads served as an IP control. Western blotting was conducted with an antibody specific for the endogenous protein or a tag-specific antibody for the ectopic protein tag. Download
Figure S3
HPV16 E6, E6*I, and E6^E7 are undetectable in HPV16-positive CaSki and SiHa cells by Western blotting (WB) with a Euromedex anti-E6 monoclonal antibody, 6F4, which has been used in Western blotting to detect HPV16 E6 (78–80). The 6F4 antibody, which recognizes an epitope from 7 to 17 aa residues of HPV16 E6, was used for Western blot analyses of E6^E7 expression in HEK293 cells transfected with a FLAG-E6^E7 expression vector or a control vector 48 h after transfection; for E6 and E6*I expression in CaSki cells transfected with si-NS or si-E6^E7 for 4 days (see the details in Fig. 5F to I), and in SiHa cells. HPV-negative C33A and HPV18-positive HeLa cell lysates served as negative controls. The 6F4 antibody at a 1:1,000 or 1:200 dilution detected E6^E7 protein ectopically expressed in HEK293 cells but not the endogenous E6^E7, E6, and E6*I proteins in CaSki or SiHa cells in these assays. Download
Figure S4
Mapping of the branch point sequence for splicing of HPV16 E6^E7 by cloning and sequencing. Gel-purified RT-PCR products from the splicing intermediate lariats of E6^E7 RNA were TA cloned and sequenced. The sequence junction at the 5′-to-2′ linkage between an identified branch site (nt 716A, nt 718A, nt 719A, or nt 721A) and the G at nt 227 at the 5′ end of the intronic sequence is divided by a vertical line in each panel. A nucleotide substitution (A to T or A to G) at each mapped branch site is indicated above each sequencing chromatogram. Download
Figure S5
Sequence alignments of putative splicing cis elements for E6^E7 splicing among 36 α-HPVs. (A) Sequence alignments are shown, with the indicated corresponding sequences (red) of the 5′ ss GU, branch site (BS) at nt 716 A/nt 718A, and 3′ ss AG in HPV16 E6^E7 RNA. PPT, polypyrimidine tract between the BS and 3′ ss; HR, high risk; HR?, possibly high risk; LR, low risk; ND, risk not determined. The viral RNAs with predicted splicing signals to produce a putative in-frame E6^E7 are highlighted in yellow. The 3′ ss AG dinucleotides in the RNA used for splicing that do not produce an in-frame protein are indicated with bold black. The RNAs of HPV27 and HPV57 with conserved splicing signals in these regions were excluded from the production of E6^E7 because the dinucleotides immediately upstream of the 5′ ss are not an AG dinucleotide, which deviates from the HPV16 5′ ss region. An AG dinucleotide at HPV56 nt 771 to 772 is not assigned as a putative 3′ ss because a G preceding the AG dinucleotide blocks its usage for RNA splicing (81). (B) HPV30, -42, and -70 are predicted to produce an E6*I mRNA, similar to HPV16 and HPV18, by splicing from an E6 5′ ss, as shown in panel A, to an E6 intron 3′ ss. Shown in this panel are their RNA 3′ ss elements (left), including a branch point sequence (BPS), polypyrimidine tract, and a 3′ ss and their putative E6*I peptide sequences (right). Download
Figure S6
Amino acid sequences of (E6 and E7) and putative E6^E7 proteins in α-HPVs. Predicted amino acid residues of E6^E7 in each HPV genotype are shown with the N-terminal E6 portion in black and the C-terminal E7 portion in blue. Underlined sequences are putative zinc binding (CXXC) motifs, and the sequences in red are LXXLL motifs. Download
Figure S7
Quantification of RNA isoforms of HPV16 early transcripts by real-time RT-PCR. (A) Diagram of a TaqMan probe and RT-PCR primer sets used for quantification of the E6 splice variants. Primers were designed within regions conserved among HPV16 subtypes. (B) Specific amplification of individual splicing RNA isoforms from total cell RNA of HPV16-positive CaSki cells by RT-PCR using the primer sets shown in panel A. (C) TaqMan RT-PCR quantification of individual RNA isoforms produced by intron 1 retention (E6), nt 226 to 409 (E6*I), nt 226 to 526 (E6*II), or nt 226 to 742 (E6^E7) splicing (filled circles) are plotted along with a standard curve created from the corresponding cDNAs, showing the correlation of copy number (log2) to the CT value. Download
ACKNOWLEDGMENTS
We thank Craig Meyers of Penn State University for the raft tissues infected with HPV16 and HPV18 and Xing Xie and Yang Li of Zhejiang University for total RNAs extracted from anonymized excess cervical cancer tissues that were no longer needed for diagnostic and clinical purposes. We also thank Jeffrey Strathern of NCI and Yihong Ye of NIDDK for their critical comments in the course of this study.
Footnotes
Citation Ajiro M, Zheng Z-M. 2015. E6^E7, a novel splice isoform protein of human papillomavirus 16, stabilizes viral E6 and E7 oncoproteins via HSP90 and GRP78. mBio 6(1):e02068-14. doi:10.1128/mBio.02068-14.