Learn more: PMC Disclaimer | PMC Copyright Notice
Serine/Arginine-Rich Splicing Factor 3 and Heterogeneous Nuclear Ribonucleoprotein A1 Regulate Alternative RNA Splicing and Gene Expression of Human Papillomavirus 18 through Two Functionally Distinguishable cis Elements
aAssociated Data
ABSTRACT
Human papillomavirus 18 (HPV18) is the second most common oncogenic HPV type associated with cervical, anogenital, and oropharyngeal cancers. Like other oncogenic HPVs, HPV18 encodes two major (one early and one late) polycistronic pre-mRNAs that are regulated by alternative RNA splicing to produce a repertoire of viral transcripts for the expression of individual viral genes. However, RNA cis-regulatory elements and trans-acting factors contributing to HPV18 alternative RNA splicing remain unknown. In this study, an exonic splicing enhancer (ESE) in the nucleotide (nt) 3520 to 3550 region in the HPV18 genome was identified and characterized for promotion of HPV18 929^3434 splicing and E1^E4 production through interaction with SRSF3, a host oncogenic splicing factor differentially expressed in epithelial cells and keratinocytes. Introduction of point mutations in the SRSF3-binding site or knockdown of SRSF3 expression in cells reduces 929^3434 splicing and E1^E4 production but activates other, minor 929^3465 and 929^3506 splicing. Knockdown of SRSF3 expression also enhances the expression of E2 and L1 mRNAs. An exonic splicing silencer (ESS) in the HPV18 nt 612 to 639 region was identified as being inhibitory to the 233^416 splicing of HPV18 E6E7 pre-mRNAs via binding to hnRNP A1, a well-characterized, abundantly and ubiquitously expressed RNA-binding protein. Introduction of point mutations into the hnRNP A1-binding site or knockdown of hnRNP A1 expression promoted 233^416 splicing and reduced E6 expression. These data provide the first evidence that the alternative RNA splicing of HPV18 pre-mRNAs is subject to regulation by viral RNA cis elements and host trans-acting splicing factors.
IMPORTANCE Expression of HPV18 genes is regulated by alternative RNA splicing of viral polycistronic pre-mRNAs to produce a repertoire of viral early and late transcripts. RNA cis elements and trans-acting factors contributing to HPV18 alternative RNA splicing have been discovered in this study for the first time. The identified ESS at the E7 open reading frame (ORF) prevents HPV18 233^416 splicing in the E6 ORF through interaction with a host splicing factor, hnRNP A1, and regulates E6 and E7 expression of the early E6E7 polycistronic pre-mRNA. The identified ESE at the E1^E4 ORF promotes HPV18 929^3434 splicing of both viral early and late pre-mRNAs and E1^E4 production through interaction with SRSF3. This study provides important observations on how alternative RNA splicing of HPV18 pre-mRNAs is subject to regulation by viral RNA cis elements and host splicing factors and offers potential therapeutic targets to overcome HPV-related cancer.
INTRODUCTION
High-risk human papillomavirus (HPV) is associated with more than 95% of cervical cancer, 40 to 90% of anogenital cancer, and 10 to 60% of oropharyngeal cancer with geographic variation (1). Two major polycistronic pre-mRNAs, early and late pre-mRNAs, are transcribed from the high-risk HPV genome. Alternative RNA splicing of the HPV polycistronic pre-mRNAs plays a crucial role in regulation of viral gene expression (2, 3). Although the molecular mechanisms that regulate alternative RNA splicing of bovine papillomavirus type 1 (BPV-1) (4,–11) and HPV16 pre-mRNA transcripts (11,–22) have been extensively studied in the past, a full transcription map of HPV18 in productive infection, the second most prevalent high-risk HPV genotype in association with cervical cancer (23), was constructed only recently (24), and the mechanistic regulation of HPV18 RNA splicing remains poorly investigated. HPV18 pre-mRNAs are transcribed mainly from a major early promoter, P55/102, or a major late promoter, P811, although a few other, weak promoters exist in the virus genome (24, 25). The HPV18 genome encodes at least five 5′ splice sites and eight 3′ splice sites. Productive HPV18 infection in keratinocyte-derived raft tissues leads to the generation of more than 15 isoforms of viral early and late RNAs through alternative RNA splicing by selection of alternative splice sites (Fig. 1) (24, 26, 27). With other, weak splice sites recently identified in HPV18-transfected U2OS cells (25), the updated HPV18 transcription map (Fig. 1) is one of the most complex transcription maps in human papillomaviruses.
Updated transcription map of HPV18. Early (P55/102) and late (P811) promoters, early (pAE at nt 4235) and late (pAL at nt 7278) polyadenylation signals, and cleavage sites (Cs4270 and Cs7299/7307) are indicated in the linearized diagram of the HPV18 genome. Other, minor promoters, P498/520/586 and P1193/1202/1207, identified both in productive HPV18 infection (24) and in U2OS cells transfected with an HPV18 plasmid (25), are also shown. The promoter P3036/3385, identified by in vitro transcription and chloramphenicol acetyltransferase (CAT) assays in HeLa cells (87) and in U2OS cells transfected with an HPV18 plasmid (25) but not confirmed in natural HPV18 infection, is represented by a dashed arrow. The numbers are the nucleotide positions in the reference HPV18 genome. LCR, long control region. The individual ORFs are shown above the linearized HPV18 genome, along with various RNA splicing isoforms below. The transcription map originated from the Zheng laboratory (24) and has been updated to include less abundant RNA isoforms from two recent publications (25, 26), with additional support from the current study. Exons (thick lines) and introns (thin lines) are illustrated for each RNA species derived from alternative promoter usage and alternative RNA splicing, with splice site positions numbered by nucleotide positions in the virus genome. The coding potentials for HPV18 viral proteins are shown on the right of each transcript. #, mRNAs reported only in transiently transfected U2OS cells (25) but not in HPV18-infected keratinocytes (24, 26).
In this study, we investigated RNA cis-regulatory elements and host trans-acting factors involved in the regulation of HPV18 RNA splicing. We identified two functionally distinct cis-regulatory elements, one in the nucleotide (nt) 612 to 639 region being an exonic splicing silencer (ESS) in the regulation of HPV18 233^416 splicing and the other in the nt 3520 to 3550 region being an exonic splicing enhancer (ESE) in the regulation of HPV18 929^3434 splicing. We also identified viral ESS binding heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1) and the ESE binding to serine/arginine-rich splicing factor 3 (SRSF3). This is the first report of the identification and functional characterization of viral RNA cis-regulatory elements and host trans-acting factors in the regulation of HPV18 pre-mRNA splicing.
MATERIALS AND METHODS
Cell lines.
HEK293 cells (human embryonic kidney tissue-derived cells) and HeLa cells (HPV18-positive cervical cancer cells) were purchased from the American Type Culture Collection (ATCC) (Manassas, VA) and grown in Dulbecco's modified Eagle medium (Gibco, Thermo Fisher Scientific, Waltham, MA) supplemented with 10% fetal bovine serum (HyClone, GE Healthcare, Pittsburgh, PA), 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Primary human foreskin keratinocytes with the HPV18 episomal form (HFK18) were cultured as described previously in EpiLife calcium-free medium (Life Technologies, Thermo Scientific) supplemented with 5% fetal bovine serum (HyClone, GE Healthcare) and human keratinocyte growth supplement (Life Technologies, Thermo Scientific) in the presence of mitomycin C-treated J2 feeder cells (28,–30), and 2.5 mM CaCl2 was added to induce keratinocyte differentiation.
Plasmids.
HPV18 minigene plasmids, pMA31 (nt 103 to 967 with a wild-type [wt] ESS), pMA77 (nt 103 to 967 with a mutant [mt] ESS), pMA99 (nt 811 to 1067 and 3234 to 3565 with a wt ESE), pMA92 (nt 811 to 1067 and 3234 to 3565 with a mt ESE), and pMA35 (nt 811 to 1067 and 3234 to 3934 to express full-length E1^E4) were PCR amplified and subcloned into plasmid pEGFP-N1 at the XhoI/NotI sites. An HPV18 genomic DNA plasmid, provided by Harald zur Hausen and Ethel-Michele de Villiers, was used as a PCR template. Plasmid pZMZ-1 for chimeric doublesex (dsx) pre-mRNA preparation (4, 5) was used for insertion of the HPV18 ESE or its mutant by PCR. Point mutations were introduced by overlapping PCR (9). The primers used for plasmid construction are listed in Table S1 in the supplemental material. Plasmid transfection was conducted using LipoD293 DNA in vitro transfection reagent (SignaGen Laboratories, Gaithersburg, MD).
siRNAs.
Small interfering RNAs (siRNAs) targeting SRSF3 (si-SRSF3) (M-030081-00) and hnRNP A1 (M-008221-02) were purchased from GE Dharmacon (Lafayette, CO). Nontargeting control siRNA (si-NS) (D-001210-01; GE Dharmacon) served as a negative control. The siRNAs were transfected twice (for HEK293 and HeLa cells) or three times (for HFK18 cells) at an interval of 48 h with a LipoJet In Vitro Transfection kit (v. II) (SignaGen Laboratories) and analyzed 48 h after the last transfection. During the last 48 h of siRNA silencing, HFK18 cells were cultured under serum-free conditions in the presence of 2.5 mM CaCl2.
Antibodies.
The following mouse monoclonal antibodies were used for Western blotting: anti-hnRNP A1 (4B10; Santa Cruz Biotechnology, Dallas, TX), anti-hnRNP F (3H4; GeneTex, Inc., Irvine, CA), anti-SRSF1 (clone 96; Life Technologies, Thermo Fisher Scientific, Waltham, MA), anti-SRSF3 (7B4; ATCC), anti-phosphorylated pan-SR proteins (mAb104; ATCC), anti-β-actin (AC-15; Santa Cruz Biotechnology), anti-involucrin (SY-5; Sigma-Aldrich, St. Louis, MO), and anti-high-risk HPV E4 (FH1.1) (31).
In vitro splicing assays.
In vitro transcription and splicing assays were performed as previously described (32, 33). Briefly, templates for in vitro transcription were prepared by PCR amplification with the primers listed in Table S1 in the supplemental material, and [α-32P]rCTP-labeled pre-mRNAs were transcribed with T7 RNA polymerase. In vitro splicing assays were conducted by incubation of 4 ng of [α-32P]rCTP-labeled pre-mRNAs in a splicing reaction buffer containing 40% HeLa nuclear extract (Promega, Fitchburg, WI) in an MgCl2 concentration of 1.5 mM for dsx pre-mRNAs or 3 mM for the other pre-mRNAs at 30°C for the indicated times. The spliced products were then separated on a 6% denaturing polyacrylamide gel with 7.5 M urea. Autoradiography was captured with a phosphorimager screen (GE Healthcare, Little Chalfont, UK) and analyzed with a Molecular Dynamics Storm 860 PhosphorImager (GE Healthcare). The splicing percentage was calculated as described previously (32, 33).
RT-PCR.
Total RNAs were extracted with TriPure reagent (Roche Diagnostics, Indianapolis, IN), treated with DNase Turbo (Invitrogen, Thermo Fisher Scientific), and applied for reverse transcription (RT)-PCR using murine leukemia virus (MuLV) reverse transcriptase with a random hexamer (Applied Biosystems, Thermo Fisher Scientific). The primers used in RT-PCR are listed in Table S1 in the supplemental material. Quantification of RT-PCR products was conducted with Image Lab 3.0 (Bio-Rad Laboratories, Hercules, CA).
Motif search and RNA pulldown assays.
Motif searching was conducted based on sequence homology to the SRSF3-binding sequences identified previously (34) or with SFmap software (http://sfmap.technion.ac.il/) (35, 36) to identify binding motifs found in systematic evolution of ligands by exponential enrichment (SELEX). RNA pulldown assays were performed as described previously (11, 34). Briefly, 5′ biotin-labeled RNA oligonucleotides were conjugated to Neutroavidin beads (Thermo Fisher Scientific) and incubated with HeLa cell extract at 4°C overnight. The proteins bound to the Neutroavidin beads were analyzed by Western blotting. The RNA oligonucleotides used in this study are shown in Table S1 in the supplemental material.
RESULTS
Reconstitution of in vitro nt 233^416 and nt 929^3434 splicing, the two major splicing events of HPV18 pre-mRNA.
To characterize HPV18 splicing regulation, we developed an in vitro RNA splicing system in the presence of HeLa nuclear extract and analyzed splice donor/acceptor usage between HPV18 positions 233^416 and 233^2779 and between 929^3434 and 3696^5613 (Fig. 2A). Individual pre-mRNAs were synthesized and 32P labeled by in vitro transcription. We compared the splicing reactions for each synthetic pre-mRNA with and without an 11-nt U1 binding site (U1bs) that enhances in vitro RNA splicing (Fig. 2A, pre-mRNAs 2, 4, 6, and 8) (32, 37). Among the eight pre-mRNAs analyzed (shown in Fig. 2B), RNA splicing activity was observed only for pre-mRNAs 2 (nt 233^416), 5 (nt 929^3434), and 6 (nt 929^3434). Notably, we found that for pre-mRNA 233^416, splicing was dependent on the U1bs at the RNA 3′ end in our splicing assay, since the same pre-mRNA (pre-mRNA 1) without a U1bs at its 3′ end did not exhibit any splicing (Fig. 2A and andB,B, compare pre-mRNA 1 to pre-mRNA 2). However, the 929^3434 splicing was found to be independent of the U1bs. Both pre-mRNA 5 without a U1b and pre-mRNA 6 with a U1b were equally spliced under our splicing conditions (Fig. 2B, lanes 16 to 21). This observation suggests the presence of a possible splicing enhancer that facilitates 929^3434 splicing. On the other hand, we failed to detect any splicing activity from pre-mRNAs 3 and 4, and 7 and 8, where splicing between 233^2779 and 3696^5613 might have been expected.
Construction of the in vitro splicing system for the major HPV18 alternative splicing sites. (A) Diagrams of the pre-mRNAs tested for in vitro splicing assays. A U1 binding motif of 11 nt (gray boxes) was attached to the 3′ ends of pre-mRNAs 2, 4, 6, and 8 to enhance in vitro splicing. Pre-mRNAs 3 to 8 have a truncated intron ∼350 nt in size for in vitro splicing assays. (B) Splicing gels from in vitro splicing assays. The splicing reactions were performed by incubating each 32P-labeled HPV18 pre-mRNA in panel A with HeLa nuclear extract at 30°C for 0, 1, and 2 h, and the splicing products were resolved on a 6% denaturing PAGE gel. The identities of unspliced pre-mRNA (a) and individual spliced products (b, b′, b″, and c) are indicated. The splicing efficiency (% spliced) was calculated from the splicing gels as described previously (32).
Identification of an ESE that promotes HPV18 929^3434 splicing.
To characterize the putative splicing enhancers that may contribute to pre-mRNA 5 929^3434 splicing, we examined three additional pre-mRNAs with truncations of various sizes from the RNA 3′ end for their 929^3434 splicing abilities (Fig. 3A). By comparing the splicing efficiencies of individual pre-mRNAs, we noticed that truncation of 154 nt from the pre-mRNA 5 3′ end (pre-mRNA A) resulted in loss of 929^3434 splicing ability (Fig. 3B, compare lanes 2 to 4 to lanes 5 to 7). On the other hand, truncation of 119 nt (pre-mRNA B) or 49 nt (pre-mRNA C) showed no or only a minimal effect on 929^3434 splicing (Fig. 3C, lanes 7, 9, and 10). Together, these observations suggest the presence of an ESE in the nt 3530 to 3635 region of the virus genome.
Identification of an ESE in the regulation of nt 929^3434 splicing of HPV18 pre-mRNAs. (A to C) Mapping of an ESE in regulation of in vitro splicing of HPV18 RNAs. HPV18 pre-mRNAs with successive truncations of the RNA 3′ end (A) were 32P labeled and incubated with HeLa nuclear extract at 30°C for the indicated times (B and C). The splicing products were resolved on a 6% denaturing PAGE gel. The identities of unspliced pre-mRNAs (a) and spliced products (b, b′, c, d, and e) on the splicing gels are indicated. (D and E) Enhancement of dsx pre-mRNA splicing by the identified HPV18 ESE. dsx pre-mRNAs 1 to 7 with a fragmented ESE ∼30 nt in size from the HPV18 nt 3520 to 3635 region at their 3′ ends (D) were evaluated for their in vitro splicing activities (E). (AAG)8 and a Py3 element (4) served as positive and negative controls, respectively. Details are shown in panels B and C.
Sequence analysis showed that this region is enriched with AG motifs (3579-GAGCGGAGAAGCAG-3590). Our previous studies and others indicated that an AG-rich sequence may function as an ESE (4, 5, 38,–40). By introduction of a point mutation (G to U) into the 3579 to 3590 region or by deletion analysis of this AG-rich region, we found that the AG-rich region has no effect on RNA 929^3434 splicing in vitro and thus does not function as an ESE (data not shown).
We then examined other regions that might promote RNA 929^3434 splicing in a Drosophila dsx exon 3 and 4 pre-mRNA, where splicing depends on an ESE (4, 5, 7, 41) (Fig. 3D). Various parts of the nt 3520 to 3635 region were attached to the 3′ end of the cassette dsx exon 4, along with an (AAG)8 as a positive control and a Py3 sequence as a negative control (4, 41), and examined for their potential ESE activities. As shown in Fig. 3E, we found that the region between nt 3520 and 3550 promoted dsx RNA splicing as efficiently as the (AAG)8 positive control (Fig. 3E, compare lane 12 to lane 10), while all the other regions did not affect or only minimally affected dsx RNA splicing (Fig. 3E, lanes 13 to 16). The dsx pre-mRNA containing a Py3 sequence as a negative control displayed no RNA splicing (Fig. 3E, lane 11). When taken together, these data indicate that the nt 3520 to 3550 region functions as an ESE that promotes HPV18 929^3434 splicing.
The HPV18 ESE contains an SRSF3-binding site that regulates its ESE activity.
To understand how the ESE functions in the promotion of RNA splicing, we analyzed the ESE sequence by searching for potential binding sites of trans-acting splicing factors that might regulate ESE function (11, 34, 42,–45). As a result, we discovered that the identified ESE contains two putative SRSF3-binding motifs—SRSF3a, from nt 3522 to 3527, and SRSF3b, from nt 3539 to 3544—and one putative SRSF1-binding motif, from nt 3540 to 3546 (Fig. 4A). Interestingly, the SRSF1 motif overlaps the SRSF3b motif. To look at this further, we introduced point mutations to disrupt each binding motif, or both, in the ESE (Fig. 4A) and examined their binding potentials for individual SR proteins by RNA pulldown assays. As shown in Fig. 4B, we demonstrated that the wt ESE binds only SRSF3, but not SRSF1 or other members of the classical SR protein family. The mt ESE with disruption of either one or both of the SRSF3-binding motifs lost its ability to associate with SRSF3. The data indicate that an SRSF3-binding site in the identified ESE consists of two SRSF3-binding motifs for stable SRSF3 avidity.
Efficient in vitro RNA splicing requires an HPV18 ESE to interact with SRSF3. (A) The identified ESE 3520 to 3550 region in the HPV18 genome contains putative binding sites for SRSF3 (shaded boxes) and SRSF1 (open boxes). The RNA oligonucleotides derived from this region with a wild-type (18ESE wt) or mutant (18ESE mt-1, -2, and -3) sequence (mutated nucleotides are underlined) were used for RNA pulldown assays (34). (B) 18ESE wt binds SRSF3 only. RNA pulldown assays were performed by mixing individual 5′-biotin-conjugated RNAs with HeLa extract. The pulldown products were blotted by using anti-SRSF1, anti-SRSF3, and a mAb104 antibody that recognizes phosphorylated SR proteins. (C and D) In vitro RNA splicing of dsx pre-mRNAs requires interaction of the ESE and SRSF3. Six versions of the HPV18 ESE (C) were examined in dsx exon 3 and 4 pre-mRNAs for their splicing activities in a 2-h splicing reaction. The splicing efficiency (% spliced) was calculated from the splicing gel (32). The identities of spliced products are shown on the right.
We further examined the functional importance of the SRSF3-binding motifs in the ESE by using the dsx pre-mRNA. To do this, we fused the wt ESE; mutant ESEs mt-1, mt-2, and mt-3 (Fig. 4A); and a short version containing only SRSF3a or SRSF3b of the ESE to the dsx pre-mRNA exon 4 (Fig. 4C). When used in in vitro splicing assays, we found that, while the wt ESE promoted splicing of dsx pre-mRNA as expected (Fig. 4D, pre-mRNA 1), all three mt ESEs lacking SRSF3-binding activities exhibited no splicing enhancement activity (Fig. 4D, pre-mRNAs 2 to 4), with the two short versions of the ESE displaying more than 50% reduction in ESE activity for dsx pre-mRNA splicing (Fig. 4D, pre-mRNAs 5 and 6). The finding that a short version of an ESE retained some reduced splicing activity suggests a single SRSF3-binding motif does have weak binding affinity for SRSF3 to exert such weak splicing activity, but this weak binding affinity for SRSF3 is not sustained for stringent washing steps in the RNA pulldown assay (Fig. 4B).
The HPV18 ESE promotes viral 929^3434 splicing in cells via its interaction with SRSF3 to regulate E1^E4 production.
To further verify the identified HPV18 ESE function in cells, we transfected HEK293 cells with an HPV18 minigene harboring a wt (plasmid pMA99) or mt (plasmid pMA92) ESE (Fig. 5A). RT-PCR analysis of total cell RNA prepared 24 h after transfection revealed that the pMA99 pre-mRNA containing a wt ESE exhibited efficient 929^3434 splicing, as expected, whereas the pMA92 pre-mRNA bearing a mt ESE showed almost no 929^3434 splicing (Fig. 5B, compare lanes 2 and 3). These data are consistent with the results obtained from the in vitro splicing assays (Fig. 4D). Moreover, we found that siRNA knockdown of SRSF3 expression reduced the 929^3434 splicing of the pMA99 pre-mRNA by ∼50% (Fig. 5B, compare lanes 2 and 4), confirming SRSF3 as a trans-acting splicing factor for the ESE function in mammalian cells.
SRSF3 promotes HPV18 929^3434 splicing for E1^E4 expression. (A) Diagram of HPV18 minigenes (pMA99 and pMA92) used for 929^3434 splicing. Plasmid pMA99 has a wt ESE containing two SRSF3-binding sites, and pMA92 has mutant ESE lacking SRSF3-binding sites, as shown in Fig. 4A. Both minigenes have a truncated intron of 339 nt and are under the control of a cytomegalovirus (CMV) immediate-early promoter (PCMV IE) and a simian virus 40 (SV40) polyadenylation signal (pASV40) for their expression in mammalian cells. F1 and R1 are forward and reverse primers used for RT-PCR. The dots in the ESE sequence represent nucleotides without mutations. (B) Knockdown of SRSF3 expression in HEK293 cells or disruption of SRSF3-binding sites in HPV18 ESE significantly reduces the 929^3434 splicing of HPV18 minigene RNA in HEK293 cells. Total RNA from HEK293 cells transfected twice with si-NS or si-SRSF3 for 96 h (at an interval of 48 h) and once with the indicated HPV18 minigene for 24 h was used for RT-PCR, with GAPDH (glyceraldehyde-3-phosphate dehydrogenase) serving as a loading control. (C to F) Knockdown of SRSF3 expression in HEK293 cells reduces E1^E4 splicing (929^3434 splicing) and E1^E4 protein production but activates selection of other weak 3′ splice sites downstream of the major 3434 3′ splice site. Plasmid pMA35, an expanded version of pMA99, contains an E1^E4 ORF (C) and was used to evaluate the effect of SRSF3 knockdown on E1^E4 splicing and protein expression (D to F). Total RNA and protein from HEK293 cells with or without SRSF3 knockdown and pMA35 transfection as described above were separately prepared. The RNA samples were used for RT-PCR (D) with the primer pair F1 and R2 (indicated in panel C), and the protein samples were blotted for the E1^E4 protein by using an anti-HPV18 E1^E4 antibody (F). GAPDH RNA served as an internal loading control for RT-PCR (D), and β-actin served as a sample loading control for Western blotting (F). RT-PCR products 1, 2, and 3 (D) were gel purified, cloned, and sequenced and represent the respective products of 929^3434, 929^3465, and 929^3506 splicing (E). (D) The asterisk indicates a heterogeneous double-stranded DNA band with one strand from the 929^3434 products hybridized with another strand from the 929^3465 or 929^3605 product during PCR annealing (Fig. 6). RT− indicates no reverse transcriptase in RT-PCR.
Because most HPV18 late transcripts use 929^3434 splicing to produce viral E1^E4 protein, we next transfected HEK293 cells with an HPV18 E1^E4 open reading frame (ORF)-containing minigene (plasmid pMA35) (Fig. 5C) and compared its 929^3434 splicing efficiency in cells with and without SRSF3 knockdown. As shown in Fig. 5D, knockdown of SRSF3 in HEK293 cells reduced 929^3434 splicing of the pMA35 pre-mRNA by 50% (Fig. 5D, compare lane 3 to lane 2 for band 1) but activated alternative 929^3465 and 929^3506 splicing to produce two smaller RT-PCR products (Fig. 5D, band 2 and band 3 in lane 3). A larger RT-PCR band above the 929^3434 splicing product (band 1) was identified as a heteroduplex derived from two RT-PCR products during the heating and annealing reactions, with one forming a single-DNA-strand loop resulting from the 929^3434 strand being annealed to a 929^3465 or 929^3605 DNA strand (Fig. 6). Gel purification, cloning, and sequencing of the two smaller RT-PCR bands confirmed the two alternative splicing events (Fig. 5E). The nt 3465 3′ splice site and the nt 3506 3′ splice site in the HPV18 genome have been previously identified as alternative 3′ splice sites in productive HPV18 infection in keratinocytes (24, 26). As expected, reduction of the HPV18 929^3434 splicing in HEK293 cells by SRSF3 knockdown also led to a significant reduction in the production of HPV18 E1^E4 protein (Fig. 5F). Together, these data clearly indicate that SRSF3 interaction with the HPV18 ESE is needed for HPV18 929^3434 splicing and E1^E4 production.
Formation of heteroduplex DNA molecules containing a single-stranded loop from two RT-PCR products. (A) Agarose gel profile of the heteroduplex molecules. The gel-purified RT-PCR products from 929^3434 (lanes 2, 6, 10, and 14) and 929^3465 (lanes 3 and 7) or 929^3506 (lanes 11 and 15) splicing were mixed (lanes 4, 8, 12, and 16), heated at 100°C for 5 min (lanes 8 and 16), and annealed at 25°C for 1 h and then at 4°C for 1 h before loading onto an agarose gel for electrophoresis. The individual products or their mixtures without heating and annealing (lanes 2 to 4 and 10 to 12) served as controls. The band (lanes 8 and 16) appearing above the nt 929^3434 product after heating and annealing of the mixed products is the heteroduplex molecules containing single-stranded loops confirmable by TA cloning and sequencing. Lanes 1, 5, 9, and 13 are 100-bp DNA ladders. (B) Schematic of the RT-PCR products and their heteroduplex. The primer pair used for amplification is shown on the right. The identity of the mRNA from which the product was derived is shown on the left alongside the expected size of the product.
To investigate the significance of SRSF3 in the regulation of HPV18 RNA splicing in keratinocytes, we knocked down SRSF3 expression in an HPV18-immortalized human foreskin keratinocyte line (HFK18). Consistent with the results from the HPV18 minigene-transfected HEK293 cells, we found that SRSF3 knockdown in HFK18 cells suppressed 929^3434 splicing and activated 929^3465 and 929^3506 splicing of HPV18 pre-mRNAs (Fig. 7A and andB),B), although it was a bit less dramatic than might be expected from HEK293 cells (Fig. 5D), perhaps due to less knockdown efficiency of SRSF3 for HFK18 cells. Moreover, SRSF3 knockdown in HFK18 cells was found to promote alternative RNA splicing of HPV18 pre-mRNAs from nt 929 to 2779 for HPV18 E2 expression and from nt 3696 to 5613 for HPV18 L1 expression. Interestingly, SRSF3 knockdown was also found to increase the expression of involucrin, a keratinocyte differentiation marker (Fig. 7C), suggesting that the decreased expression of SRSF3 promotes keratinocyte differentiation. These data are consistent with our early findings that reduced expression of SRSF3 correlates with increased keratinocyte differentiation and L1 expression in papillomavirus-infected tissues and that differentially expressed SRSF3 controls the HPV early-to-late switch (11).
Knockdown of SRSF3 expression in keratinocytes reduces 929^3434 splicing but activates alternative splicing of other, weak 3′ splice sites and L1 expression during HPV18 infection. (A) Diagram showing HPV18 late pre-mRNA with major alternative splice sites. HPV18 ORFs (E1^E4, E2, and L1) are shown at the top, with the primers used in RT-PCR indicated by arrows (F1, F2, R2, R3, and R4) below the pre-mRNA. (B) Knockdown of SRSF3 activates alternative RNA splicing of HPV18 pre-mRNAs in HFK18 cells containing an episomal HPV18 genome, an HPV18-immortalized keratinocyte line derived from primary foreskin keratinocytes. Total RNA isolated from HFK18 cells transfected three times with si-NS or si-SRSF3, with an interval between transfections of 48 h, was used for RT-PCR, and the spliced RNA products from 929^3434, 929^3465, 929^3506, E2, and L1 were detected with a primer set indicated in each gel panel. *, a heterogeneous double-stranded DNA band as described in the legend to Fig. 5D. GAPDH RNA served as an internal loading control for RT-PCR. (C) SRSF3 knockdown appears to induce the expression of involucrin, a keratinocyte differentiation marker. HFK18 cells were cultured in growth media containing a low (0.17 mM) or high (2.5 mM) concentration of Ca2+ and were examined by Western blotting for involucrin as a differentiation marker. β-Actin served as a loading control. To test the SRSF3 knockdown effect, HFK18 cells grown in low-Ca2+ (0.17 mM) medium were transfected twice with si-SRSF3 or si-NS for 96 h, at an interval of 48 h, and maintained in low- or high-Ca2+ culture medium for a total of 4 days. Involucrin and β-actin were then blotted.
Identification of an ESS in regulation of HPV18 233^416 splicing via interaction with hnRNP A1.
Considering that HPV18 233^416 splicing is essential for the expression of HPV16 E6 and E7 oncoproteins (46), we explored whether a potential splicing cis element(s) exists in the HPV18 E6E7 bicistronic pre-mRNA in the regulation of 233^416 splicing. To do this, a series of HPV18 E6E7 pre-mRNAs with successive 3′ extensions of exon 2 were prepared in vitro and analyzed for their in vitro splicing efficiencies (Fig. 8A). As shown in Fig. 8A and andB,B, the first three pre-mRNAs, pre-mRNAs 1, 2, and 3, all exhibited efficient 233^416 splicing, with a splicing efficiency of ∼27 to 30% (lanes 1 to 3). Pre-mRNA 4, which had an additional 43-nt extension in its exon 2 over pre-mRNA 3, showed a splicing efficiency of ∼17%, about ∼11% splicing reduction compared to that of pre-mRNA 3 (lane 4), whereas further extension of exon 2 to include a nt 610 to 639 region in pre-mRNA 5 severely inhibited this splicing, and only ∼4% of pre-mRNA 5 underwent 233^416 splicing (lane 5). There was no RNA splicing for the remaining pre-mRNAs, with further extension beyond the nt 639 position (pre-mRNAs 6 to 8 [lanes 6 to 8]). These results suggest that the nt 611 to 639 region is most inhibitory for 233^416 splicing. Subsequently, we designated this region an ESS.
Identification of an ESS in regulation of HPV18 233^416 splicing. (A) Diagrams of pre-mRNAs 1 to 8 with successive exon 2 extensions. Each pre-mRNA's 3′ end was attached to a U1 binding motif (11 nt) (gray boxes) to enhance its splicing efficiency (20, 32). The numbers are the nucleotide positions in the virus genome. (B) Splicing gels from in vitro splicing assays. The splicing reactions were performed by incubating each 32P-labeled HPV18 pre-mRNA in panel A with HeLa nuclear extract at 30°C for 2 h, and the splicing products were resolved on a 6% denaturing PAGE gel. The identities of unspliced pre-mRNAs and individual spliced products are indicated on the left. The splicing efficiency (% spliced) was calculated from the splicing gel as described (32).
To understand how the identified ESS might function in regulation of the 233^416 splicing of HPV18 E6E7 bicistronic pre-mRNAs, the ESS region was first searched for putative splicing factor binding motifs with SFmap software (http://sfmap.technion.ac.il/) (35, 36). We identified two potential hnRNP A1 binding motifs, one at nt 615 to 620 and the other at nt 632 to 638, and one putative hnRNP F binding motif at nt 621 to 626 within the identified ESS (Fig. 9A). Subsequent validation by RNA pulldown assays with wt or mt (mt-1, -2, and -3) ESS RNA oligonucleotides revealed the binding of hnRNP A1 to the 632 to 638 motif, but not to the 615 to 620 motif, because both wt and mt-1, but not mt-2, ESS RNA oligonucleotides were capable of binding to hnRNP A1 (Fig. 9A and andB).B). We did not see the ESS interacting with hnRNP F or any other classical SR proteins (Fig. 9B).
Binding of hnRNP A1 to the identified ESS is responsible for regulation of HPV18 233^416 splicing. (A) RNA sequences of the identified ESS and its mutations. Predicted binding sites for hnRNP F and hnRNP A1 are indicated by open and shaded boxes, respectively. The wt and mt (mt-1, mt-2, and mt-3) RNA oligonucleotides (mutated nucleotides are underlined) were used for RNA pulldown assays. (B) HPV18 ESE binds hnRNP A1. RNA pulldown assays were performed by mixing each RNA oligonucleotide in panel A with HeLa cell extract. The pulldown products were then blotted using anti-hnRNP A1, anti-hnRNP F, and mAb104 antibodies. (C) Diagram of HPV18 minigenes bearing a wt (pMA31) or mt (pMA77) ESS for HPV18 233^416 splicing assays in HEK293 cells. Both minigenes containing intact E6 and E7 ORFs are under the control of a CMV IE promoter (PCMV IE) and an SV40 polyadenylation signal (pASV40). The horizontal arrows represent primers (F3, F4, and R5) used in RT-PCR. The numbers below the diagram are the nucleotide positions in the virus genome. The wt and mt ESS sequences are detailed below the diagram. (D and E) hnRNP A1 regulates HPV18 233^416 splicing through the ESS. HEK293 cells without (D) or with (E) hnRNP A1 knockdown were transfected with pMA31 or pMA77 for 24 h before extraction of total RNA for RT-PCR analysis using the primer pair F3 plus R5 (D) or F4 plus R5 (E). GAPDH RNA served as a loading control. (F) Diagram of HPV18 E6E7 polycistronic pre-mRNA derived from the viral early promoter P55/102 and polyadenylated at a viral early pA site (pAE). The numbers on the E6 and E7 ORFs and the diagram are the nucleotide positions in the virus genome. F4 and R5 are two primers used for RT-PCR analysis. (G) Knockdown of hnRNP A1 in HeLa cells promotes HPV18 233^416 splicing and reduces E6 intron retention. Total RNA from HeLa cells transfected twice with si-NS or si-SRSF3 for 96 h, at an interval of 48 h, was examined by RT-PCR. GAPDH RNA served as a loading control. (H) The knockdown efficiency of hnRNP A1 in HEK293 or HeLa cells was evaluated by Western blotting. β-Actin served as a loading control. RT− in panels D and E and G and H indicates no reverse transcriptase in RT-PCR.
We then evaluated the function of the ESS-hnRNP A1 interaction in HPV18 233^416 splicing in human cells and during HPV18 infection. To do this, we transfected HEK293 cells with an HPV18 E6E7 expression plasmid harboring a wt (plasmid pMA31) or mt (plasmid pMA77) ESS (Fig. 9C). The mt ESS in pMA77 has the mt-3 mutation and does not bind hnRNP A1 (Fig. 9A and andB).B). RT-PCR analyses of total cell RNA prepared 24 h after transfection revealed that disruption of the hnRNP A1-binding site in pMA77 enhanced 233^416 splicing by ∼45% compared to what was obtained from the wt ESS-containing pMA31(Fig. 9D). Western blotting of transfected HCT116 cells bearing wt p53 showed that the increased 233^416 splicing from the pMA77-transfected cells at 48 h resulted in reduction of HPV18 E6 expression and stabilization of p53 by 2-fold compared to that from the pMA31-transfected cells (data not shown). On the other hand, knockdown of hnRNP A1 expression in HEK293 cells was also consistently found to increase the 233^416 splicing of the E6E7 bicistronic pre-mRNAs and therefore reduced its intron retention by ∼50% from that in the cells without hnRNP A1 knockdown (Fig. 9E). A similar result was seen in HPV18-infected HeLa cells with or without hnRNP A1 knockdown (Fig. 9F to toH).H). In HPV18-infected HeLa cells, RT-PCR analyses of total cell RNA revealed that hnRNP A1 knockdown promoted HPV18 233^416 splicing, leading to reduction of its intron retention by up to 70% from that in the cells without knockdown of hnRNP A1 (Fig. 9G and andH).H). Altogether, these data show that the identified ESS in the E7 ORF region regulates HPV18 233^416 splicing by interaction with hnRNP A1.
DISCUSSION
It is well known that papillomavirus gene expression relies heavily on alternative RNA splicing, which is regulated by viral RNA cis elements and host cell spliceosomes (3, 47, 48). These viral RNA cis elements (ESE and ESS) were originally discovered in bovine papillomavirus late pre-mRNAs (4, 5, 8, 9) and later in HPV16 pre-mRNAs (12, 14, 21, 22). In general, the ESE and ESS are bipartite elements (4, 49), commonly located downstream of a suboptimal 3′ splice site to regulate selection of an alternative upstream splice site by interaction with various cellular splicing factors (5, 6, 11, 37, 48, 50,–52). In this report, we identified an ESE within the E4 ORF region that can regulate both HPV18 early and late gene expression and an ESS in the E7 ORF region that can regulate the expression of HPV18 E6 and E7 and other viral early genes.
Host splicing factors and some viral proteins responsible for the regulation of BPV-1 (5, 6, 47) and HPV16 (11, 18, 48, 53) RNA splicing have been extensively studied. In this report, we further reveal that the host splicing factors responsible for two major splicing events in HPV18 are hnRNP A1 and SRSF3. hnRNP A1 binds to the newly identified ESS in the E7 ORF region and suppresses HPV18 233^416 splicing, whereas SRSF3 interacts with the newly identified ESE and promotes the 929^3434 splicing of HPV18 pre-mRNAs (Fig. 10). Although these observations are convincing, there are perhaps other, unidentified splicing factors that are also involved in regulation of these two splicing events, because knocking down either SRSF3 or hnRNP A1 from cells or disruption of either SRSF3- or hnRNP A1-binding sites in HPV18 transcripts exhibited only partial effects. Other splicing factors have been found to regulate alternative splicing of BPV-1 and HPV16 pre-mRNAs by interaction with the ESE or ESS (11, 47, 48). Various studies have shown that ESS functions through hnRNP A1 (54,–56) and other splicing factors (6, 37, 57, 58). In HPV16, hnRNP A1 binds to an ESS in the L1 coding region and regulates HPV16 L1 RNA splicing (12, 22). hnRNP A1 and A2 together were also proposed to regulate HPV16 E6 intron splicing (226^409 splicing) (19), but the mechanism of this regulation remains unknown.
Regulation of HPV18 pre-mRNA splicing by an hnRNP A1-dependent ESS and an SRSF3-dependent ESE. HPV18 early pre-mRNA (top) and late pre-mRNA (bottom) are regulated by multiple alternative splicing events. The identified ESS in the nt 612 to 639 region binds hnRNP A1 to reduce nt 233^416 splicing of HPV18 E6E7 pre-mRNA for the production of E6 mRNA. The identified ESE in the nt 3520 to 3550 region binds SRSF3 to promote HPV18 929^3434 splicing for expression of viral early genes and E1^E4. Loss of SRSF3 binding to this region or reduction of SRSF3 expression activates selection of other, minor 3′ splice sites upstream or downstream of the nt 3434 3′ splice site.
SRSF3 affects several steps of viral gene expression during virus infection (11, 59,–65). In HPV16, SRSF3 binds to an ESE in the E4 ORF region and controls the viral early-to-late switch as a result of its differential expression in keratinocytes (11). Undifferentiated keratinocytes express more SRSF3 and support a higher level of viral early gene expression, including HPV16 E6 and E7. In contrast, differentiated keratinocytes express much lower levels of SRSF3 but more viral L1 and L2. In HPV18, we found that the identified ESE in the E4 ORF region binds SRSF3 and promotes HPV18 929^3434 splicing. Similar to HPV16, this SRSF3-ESE interaction in HPV18 pre-mRNA also suppresses L1 expression by inhibition of 3696^5613 splicing. Hence, these observations indicate that both HPV16 and HPV18, which are members of the genus Alphapapillomavirus, share an evolutionary, phylogenetical pathway to regulate their gene expression at the posttranscriptional level during keratinocyte differentiation. Interestingly, we found that SRSF3 requires two intact SRSF3-binding motifs in the HPV18 ESE for stable binding, with knockdown of SRSF3 in keratinocytes appearing to promote cell differentiation. The mechanisms that underlie the latter observation remain to be further investigated.
In addition to its regulation of viral gene expression, SRSF3 has been characterized as an oncogenic factor (66,–68), and it effects global change of the gene expression of hundreds of mammalian genes to maintain cell homeostasis (34, 39, 69,–71). Upregulation and functional association with various types of cancer have been reported for SRSF3 (66,–68, 72) and hnRNP A1 (73,–76), which we have identified here as two crucial trans-acting factors for alternative splicing of HPV18 pre-mRNAs. Thus, our observations indicate that the manipulation of alternative splicing regulation may be a potential therapeutic target for HPV-related cancer. While a chemical inhibitor that targets splicing factors themselves has not been developed yet, the recent development of kinase inhibitors working upstream of splicing factors, such as CDC2-like kinase (CLK) (77,–79) or SR protein kinase (SRPK) (80,–82) for SR protein phosphorylation, may provide a way to block splicing factor activity. Alternatively, splicing cis elements may also be targeted, given the recent development of modified therapeutic oligonucleotides able to modulate splicing regulation (83,–86). Looking forward, the identification and characterization of the HPV18 ESE and ESS in this report may in due course offer potential therapeutic targets to overcome HPV-related cancer.
ACKNOWLEDGMENTS
We thank Craig Meyers of Penn State University for providing the HFK18 cells used in this study.
This research was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.
We have no conflicts of interest to declare.
Footnotes
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JVI.00965-16.