HnRNP D activates production of HPV16 E1 and E6 mRNAs by promoting intron retention

Abstract Human papillomavirus type 16 (HPV16) E1 and E6 proteins are produced from mRNAs with retained introns, but it has been unclear how these mRNAs are generated. Here, we report that hnRNP D act as a splicing inhibitor of HPV16 E1/E2- and E6/E7-mRNAs thereby generating intron-containing E1- and E6-mRNAs, respectively. N- and C-termini of hnRNP D contributed to HPV16 mRNA splicing control differently. HnRNP D interacted with the components of splicing machinery and with HPV16 RNA to exert its inhibitory function. As a result, the cytoplasmic levels of intron-retained HPV16 mRNAs were increased in the presence of hnRNP D. Association of hnRNP D with HPV16 mRNAs in the cytoplasm was observed, and this may correlate with unexpected inhibition of HPV16 E1- and E6-mRNA translation. Notably, hnRNP D40 interacted with HPV16 mRNAs in an HPV16-driven tonsillar cancer cell line and in HPV16-immortalized human keratinocytes. Furthermore, knockdown of hnRNP D in HPV16-driven cervical cancer cells enhanced production of the HPV16 E7 oncoprotein. Our results suggest that hnRNP D plays significant roles in the regulation of HPV gene expression and HPV-associated cancer development.


INTRODUCTION
Human papillomaviruses (HPV) contain a double-stranded circular DNA genome of approximately 8 kb in size (1). HPV infections cause a range of disease from benign warts to invasive cancers, for example cervical cancer and tonsillar cancer (2). HPV type 16 (HPV16) is responsible for around 55% of all cervical cancers while the remainder is caused by other high risk (HR) HPV types (3). Cancer progression is due to an increased continuous expression of HPV oncoproteins E6 and E7 that inactivate tumor sup-pressor proteins p53 and pRb (4), respectively. E6 and E7 activate the cell cycle, inhibit apoptosis and cause genomic instability (5)(6)(7)(8)(9). The HPV16 E1 and E2 proteins are key factors during replication of HPV16 genomic DNA. E1 functions as DNA helicase whereas E2 has a multifunctional role including transcriptional regulation, initiation of HPV16 DNA replication, facilitation of HPV16 genome partitioning during mitosis and post-transcriptional control of HPV16 gene expression (10)(11)(12)(13). In contrast to E6 and E7, E2 has pro-apoptotic properties and is frequently inactivated when the HPV16 genome integrates in cellular chromosomes, a process that possibly enhances carcinogenesis (11,14). The HPV16 E4 and E5 proteins are essential for completion of the HPV16 replication cycle and E5 may contribute to carcinogenesis (15,16).
Since the HPV16 genome has two promoters only, alternative mRNA splicing plays a major role in the regulated expression of all HPV16 genes (17)(18)(19)(20)(21)(22). A complex pattern of alternatively spliced and polyadenylated HPV16 mRNAs is observed during the HPV16 life cycle. Therefore, it is not surprising that a number of cis-acting regulatory RNA elements and their cognate trans-acting factors control the HPV16 alternative splicing and polyadenylation. In addition, it has been shown that the levels of various RNA-binding proteins are altered during the progression of HPV16-infected cells to cervical cancer through a series of premalignant cervical intraepithelial lesions (23,24). Thus, it is of interest to identify cellular RNA-binding proteins that control HPV16 gene expression.
Heterogeneous nuclear ribonucleoproteins (hnRNPs) represent a large family of RNA-binding proteins (RBPs). The association of hnRNP proteins with pre-mRNAs is initiated co-transcriptionally at the nascent transcripts. Many of the RNA-binding proteins remain bound to the resulting mRNAs all the way to the ribosomes and shuttle back and forth between the nucleus and the cytoplasm, demonstrating that RNA-binding proteins are important determinants of pre-mRNA processing during the entire mRNA pathway including mRNA splicing, localization, translation and stability (25). hnRNPs are massively involved in alternative splicing and the canonical function of hnRNPs is performed through their binding to RNA elements adjacent to splice sites, thereby either repressing or supporting the assembly of the spliceosome complex on splice donor (SD) or splice acceptor (SA) sites. Alternatively, they affect the recruitment of other RNA-binding proteins such as Serine/Arginine (SR) rich proteins to exonic or intronic splicing enhancers or silencers (26). Furthermore, many hn-RNPs participate in more than one of these processes (27), e.g. regulation of both alternative splicing-and poly(A)-site usage by hnRNP H/F or L, alternative splicing and translation by hnRNP A1, or alternative splicing and RNA stability by hnRNP D.
The hnRNP family comprises at least 20 major RNAbinding proteins originally named alphabetically from A1 to U. These proteins share modular structures that include RNA recognition motifs (RRM) or quasi-RRMs, other motifs such as the K Homology (KH) domain and the arginine/glycine-rich RGG motif and other RNA-binding domains (RBD) present on a subset of hnRNPs (28). In some hnRNPs other auxiliary domains like glycine rich domains and proline-rich domains may be present (25,29). Although all hnRNPs display RNA-binding activity with a certain degree of specificity, an hnRNP protein may not bind exclusively to high-affinity binding sites. The RNAbinding specificity of hnRNPs is strongly influenced by the type and number of RNA-binding domains on the hn-RNP, which in turn generates both general and specific interactions with nucleic acids, as well as the primary and secondary structures of the target RNA (30). HnRNPs also possess nuclear localization sequences (NLSs) that bring the proteins to the nucleus, while other hnRNPs possess nucleocytoplasmic shuttling domains which allow hn-RNPs to shuttle between nuclear and cytoplasmic compartments (28). Given the important role of hnRNPs in mRNA metabolism, we speculate that HPV16 early gene expression is under control of a range of hnRNPs. Here we determined the effect of 13 different hnRNPs on HPV16 early mRNA splicing and we found that a number of the hnRNPs affected HPV16 early mRNA splicing, but in different ways. We focused our attention on the members of the cellular hnRNP D protein family. hnRNP D, also known as AUrich element binding factor 1 (AUF1) promotes the decay of many target mRNAs (31), but it was also reported to enhance the stability and affect translation of target transcripts. As a result, hnRNP D is involved in multiple cellular processes, including miRNA biogenesis (32), translational regulation (33,34), telomere maintenance (35,36), cell cycle control (37), apoptosis (38) and inflammatory responses (39) since hnRNP D target mRNAs encode proteins implicated in these processes. Reports on hnRNP D and mRNA splicing are relatively scarce but a regulatory function of hn-RNP D proteins in alternative splicing has been suggested (34). One example is that hnRNP D proteins as well as hn-RNP DL control their own expression by auto-or crossalternative splicing regulation (40). Another example suggests that hnRNP D together with neuronal members of the ELAVL protein family (nELAVLs) induce neuron-specific alternative splicing of the amyloid precursor protein (APP) (41).
In this manuscript, we show that hnRNP D proteins inhibit splicing and promote retention of the E1-and E6-encoding introns on the HPV16 early mRNAs, thereby specifically stimulating production of the partially spliced HPV16 E1-and E6-encoding mRNAs. Furthermore, hn-RNP D facilitated export of the partially spliced HPV16 E1 and E6 mRNAs to the cytoplasm. However, despite the fact that the HPV16 E1 and E6 mRNAs reached the cytoplasm in association with hnRNP D, HPV16 mRNAs were poorly translated. In conclusion, our results suggest that hnRNP D proteins inhibit HPV16 early mRNA splicing and enhance production of the partially spliced, intron-containing HPV16 E1 and E6 mRNAs, thereby playing an important role at the initial steps of the biogenesis of HPV16 E1 and E6 mRNA.
To construct HPV16 subgenomic plasmids pX656 and pX478, primer B97S (Supplementary Table S1) was used in combination with primer X656A or X478A to amplify HPV16 sequences that were digested with PteI and XhoI and inserted between the PteI and XhoI sites in CMV promoter driven empty vector pCL086 (53). HPV16 subgenomic plasmid pXH856F encoded HPV16 E6 and E7 genes (HPV16 nucleotides positions 104-855) (HPV16 nucleotide positions refer to the HPV16 reference sequence HPV16R (GeneBank: K02718.1)). E6 was fused to an HA tag sequence at the 5 -end and E7 to a flag tag sequence at the 3 -end. pXH856F was generated by insertion of DNA fragment PCR-amplified with pXH856F sense and anti-sense primer (Supplementary Table S1) and subcloned into pCL086 at PteI and XhoI sites. The 5 -splice site SD226 in pXH856F was mutated (T228C, A229C) creating pXH856SDmF. The mutations in SD226 were introduced by PCR mutagenesis with pXH856SDmF sense and antisense primers (Supplementary Table S1) in a two-step PCR amplification reaction using overlapping followed by subcloning into pXH856F. Plasmid p16E1-3xF encodes the HPV16 E1 gene (HPV16 nucleotides positions 865-2811) fused with a 3xFLAG tag sequence at the 3 -end and was constructed by PCR amplification using p16E1-3xF inverse sense and anti-sense primers (Supplementary Table S1) and subcloned into pCL086. HPV16 5 -splice sites SD880 and SD1302 on p16E1SDm-3xF were mutated (G881C and G1303C) to create p16E1SDm-3xF. The mutations were conducted by p16E1SDm-3xF sense and antisense primers (Supplementary Table S1) in two-step PCR amplification.
To construct HPV subgenomic plasmids pBELEN, pBE-LENdE1, pBELsluc plasmid was cut with restriction enzymes CsiI and XhoI, followed by blunting of sticky DNA ends with Klenow fragment and re-ligation, resulting in pBELEN. To construct pBELENdE1, pBELEN was cut with restriction enzymes AdeI and NsiI to delete sequences in the E1 coding region, followed by blunting of sticky DNA ends with Klenow fragment and re-ligation.

Transfections
Transfections of HeLa cells and 293T cells were performed with Turbofect according to the manufacturer's protocol (Thermo Fisher Scientific). Briefly, a mixture of Turbofect:DNA ratio of 2:1 (l reagent: g DNA) for HeLa cells or 4:1 for 293T cells and DMEM without serum was incubated at room temperature for 20 min prior to dropwise addition to subconfluent cells. Transfections of nHFK were performed with ViaFect according to the manufacturer's protocol (Promega). In brief, a mixture of ViaFect:DNA ratio of 3:1 and EpiLife without serum was incubated at room temperature for 20 min prior to dropwise addition to subconfluent cells. Fluorescence images of EGFP set of plasmids transfected HeLa cells were acquired using Olympus CKX53 inverted microscope.

Nuclear and cytoplasmic extraction
Nuclear and cytoplasmic extracts were prepared from HeLa cells grown in 6 cm dishes at 24 h post-transfection. Cells Nucleic Acids Research, 2022, Vol. 50, No. 5 2785 were harvested by using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific) according to the manufacturer's protocol. In brief, cell pellets were resuspended in ice cold buffer CER I with protease inhibitors (Sigma Aldrich) and vortexed vigorously prior to incubation on ice for 10 min. Ice-cold buffer CER II was added to the samples that were vortexed vigorously and incubated on ice for one minute. After 5 min of maximum speed centrifugation, the supernatants were collected as cytoplasmic extracts. The pellets were washed once by PBS and collected as nuclear extracts.

RNA extraction and RT-PCR
Total RNA was extracted from transfected cells using TRI Reagent (Sigma Aldrich) and Direct-zol RNA MiniPrep (ZYMO Research) according to the manufacturer's protocols. Reverse transcription (RT) was performed in a 20 l reaction using random hexamers (Invitrogen) and reverse transcriptase (Invitrogen). One microliter of cDNA was subjected to PCR-amplification. cDNA representing HPV16 mRNAs spliced from HPV16 5 -splice site SD226 to 3 -splice site SA409 was amplified with RT-PCR primers 97S and 438AS (Supplementary Table S1) and cDNA representing HPV16 mRNAs spliced from 5 -splice site SD226 to 3 -splice sites SA409, SA526, or SA742 with RT-PCR primers 97S and 880AS (Supplementary Table S1). cDNA representing HPV16 mRNAs spliced from 5 -splice site SD880 to 3 -splice site SA2709 or SA3358 was amplified with RT-PCR primers 773S and E2AS/E2QAS or E4AS (Supplementary Table S1). cDNA representing HPV16 mRNAs spliced from 5 -splice site SD226 to 3 -splice site SA2709 or SA3358 were amplified with RT-PCR primers 97S and E2AS/E2QAS or E4AS, respectively (Supplementary Table S1). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA was amplified with primers GAPDHF and GAPDHR (Supplementary Table S1). Spliced actin and unspliced actin cDNAs were amplified with primers actin-s or actin-s1 and actin-a (Supplementary Table S1). cDNA representing mRNAs encoding each of the four isoforms of hnRNP D were amplified with hnRNP D mRNAspecific primers 2S and 7A (Supplementary Table S1). For location of these mRNAs see Figure 3A. To monitor recombination at the loxP sites in pHPV16AN, PCR was performed with primers 16S and 16A on DNA extracted from the transfected cells (this PCR yields a 366-nucleotide PCR fragment that is diagnostic for recombination at the LoxP sites). Examples of control PCR experiments performed on RNA samples in the absence of reverse transcriptase are shown in various figures. Primer pairs used for RT-PCR and RT-qPCR (described in next section) are summarized in Supplementary Table S4.

Real-time quantitative PCR (qPCR)
qPCR was performed in a final reaction volume of 20 l with 1 l of cDNA prepared as described above in a Min-iOpticon (Bio-Rad) using the SsoAdvanced SYBR Green Supermix (Bio-Rad) according to the manufacturer's instructions. To quantitate intron retained E6 mRNAs by RT-qPCR primers TotalE6F and 234AS were used (Supplementary Table S1). For primer location, see Figure 1C and Supplementary Figure S5. Spliced E2 mRNAs were quantitated by RT-qPCR using primers 773S and E2AS (Supplementary Table S1, for primer location, see Figure 1). For quantitation of E1 mRNA by RT-qPCR, primers 773S and E1AS were used (Supplementary Table S1, for primer location, see Figure 1). The expression levels of the mRNAs were determined from the threshold cycle (Ct), and the relative expression levels were calculated using the 2∧ Ct method. Results were normalized to GAPDH mRNA levels determined with primers GAPDHF and GAPDHR (Supplementary Table S1). mRNA quantification was performed in triplicates, and negative controls were included in each reaction. Melting curves were analyzed in each reaction.

Secreted luciferase assay
The Metridia longa secreted luciferase (sLuc) activity in the cell culture medium of transfected cells was monitored with the help of the Ready-To-Glow Secreted Luciferase Reporter assay according to the instructions of the manufacturer (Clontech) as described previously (43). In brief, 50 l of cell culture medium was added to 5 l of 0.5X Secreted Luciferase substrate/Reaction buffer in a 96-well plate and luminescence was determined in a Tristar LB941 Luminometer.

Western blotting
Cell extracts for western blotting were obtained by resuspending transfected cells in radioimmunoprecipitation assay (RIPA) buffer consisting of 50 mM Tris HCl pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 2 mM EDTA, 1 mM DTT and protease inhibitor (Sigma Aldrich), followed by centrifugation at full speed for 20 min and collection of the supernatants. Proteins were denatured by boiling in Laemmli buffer. After SDS-PAGE, the proteins on the gels were transferred onto nitrocellulose membranes, blocked with 5% nonfat dry milk in PBS containing 0.1% Tween 20 and stained with specific primary antibodies (Supplementary Table S2) to the indicated proteins followed by incubation with secondary antibody (Supplementary Table S2) conjugated with horseradish peroxidase and detection with chemiluminescence reagents.

ssRNA oligo pull down
Whole cell lysates of HeLa cells were prepared using cell lysis buffer consisting of 25 Supplementary Table S1. incubated with rotation for 1 h at room temperature followed by washing of beads 10 times with binding buffer using DynaMag 96 Side magnetic plate (Invitrogen). Proteins were eluted by boiling of the beads in Laemmli Buffer. Samples were subjected to SDS-PAGE followed by western blot analysis with the indicated antibodies (Supplementary Table S2).

Co-immunoprecipitation
Transfected Hela cells were lysed in cell lysis buffer as described above under 'ssRNA oligo pull down'. For immunoprecipitation, anti-flag antibody (M2, Sigma Aldrich) or IgG was added to Dynabeads protein G and incubated with cell lysates under gentle rocking at 4 • C overnight. The complexes were washed six times using cell lysis buffer and eluted by boiling in Laemmli buffer. Samples were subjected to SDS-PAGE followed by western blotting with specific primary antibodies (Supplementary Table S2).

UV-crosslinking and immunoprecipitation (CLIP)
Transfected HeLa cells grown in 10 cm dishes were washed by ice-cold PBS followed by crosslinking twice with 0.4 J cm −2 UV light (254 nm) in a bio-link cross-linker (Biometra). Cytoplasmic extracts were prepared as described above. Whole cell lysates were prepared by resuspending cells in one ml of RIPA buffer and incubated on ice for 30 min with occasional vortexing to lyse cells. For immunoprecipitations, 2g of the anti-flag antibody (M2, Sigma Aldrich) or mouse IgG was incubated at 4 • C overnight in 0.5 ml of cell lysate. About 20 l of Dynabeads Protein G (10004D, Invitrogen) and 20 l Dynabeads Protein A (10001D, Invitrogen) were blocked with 1% BSA for 0.5 h, washed three times in RIPA buffer and then added to the antibody-protein mixtures followed by incubation for 1 h at 4 • C. The beads were washed three times with buffer I (50 mM Tris HCl pH 7.4, 300 mM NaCl, 0.5% NP-40, 1 mM EDTA, 1 mM DTT), three times with buffer II (50 mM Tris HCl pH 7.4, 800 mM NaCl, 0.5% NP-40, 1 mM EDTA, 1 mM DTT) and three times with buffer III (50 mM Tris HCl pH 7.4, 800 mM NaCl, 250 mM LiCl, 0.5% NP-40, 1 mM EDTA, 1 mM DTT). RNA was eluted by phenol/chloroform extraction and ethanol-precipitated and dissolved in 20 l of water. About 10 l of immunoprecipitated RNA was reverse transcribed using M-MLV reverse transcriptase (Invitrogen) and random primers (Invitrogen) according to the protocol of the manufacturer. Two microliters of cDNA were subjected to PCR amplification using HPV16-sepcific primer pairs 773S and E1AS, 773S and 438AS or E6F and E6R (Supplementary Table S1) as described above.

Ribonucleoprotein (RNP) immunoprecipitation (RIP) analysis
For immunoprecipitation of endogenous ribonucleoprotein (RNP) complexes (RIP analysis) from whole cell extracts, HN26 cells or 3310 cells were lysed in cell lysis buffer. The supernatants were incubated with anti-AUF1 antibody (Cell Signalling) or IgG (Millipore) (Supplementary Table   S2) overnight at 4 • C. Dynabeads Protein G (10004D, Invitrogen) + 20 l Dynabeads Protein A (10001D, Invitrogen) were added to the antibody-protein mixtures followed by incubation for 1 h at 4 • C. After washing of the beads six times using cell lysis buffer, RNA was eluted using Tri reagent and incubated with 20U of RNase-free DNase I for 1 h at 37 • C and subjected to RT-PCR using HPV16-sepcific primer pairs 773S and E4AS or 97S and 438AS (Supplementary Table S1) as described above.

siRNA library and siRNA transfections
ON-TARGETplus human hnRNP D siRNA SMARTpool consists of four siRNAs to hnRNPD (L-004079-00-0010) (Dharmacon). A scrambled negative control siRNA (D-001810-10-05) was also purchased from Dharmacon. Transfections were conducted with DharmaFECT1 (Dharmacon) according to the instructions of the manufacturer. The siRNA SMARTpool to hnRNP D or scrambled control siRNAs were transfected in duplicates into C33A2 or SiHa cells grown in 12-well plates for RNA extraction and RT-qPCR or in 6-well plates for protein extraction and western blotting. Cells were harvested at 48 h posttransfection for RNA extraction and RT-qPCR or 72 h post-transfection for protein extraction and western blotting performed as described above.

In vitro translation assay
In vitro translation was carried out with TNT(R) Quick Coupled Transcription/Translation Systems (Promega) according to the instructions of the manufacturer. In brief, 100 ng pET32a-HA-E6SDm or 200 ng pcDNA3.1(+)-E1SDm-3XFlag plasmid was translated in the absence or presence of 1 M recombinant hnRNP D protein (EUPROTEIN) or 1 M BSA. The 25 l reactions were incubated for 90 min at 30 • C. The translation reactions were analyzed by western blotting as described above. The Luciferase control RNA was also translated in the absence or presence of 1 M hn-RNP D or 1 M BSA. Luciferase activity in the translation reactions were monitored according to the instructions of the manufacturer using Tristar LB941 Luminometer. Recombinant hnRNP D and BSA were separated on SDS-PAGE followed by staining with Colloidal Blue Staining Kit (Invitrogen).

Quantitations
The softeware used to determine band intensity in western blots and RT-PCR gels is 'Image Lab 6.1.0' and quantitations were performed with the software 'Prism GraphPad 8.4.0'.

hnRNP A1, hnRNP D and hnRNP I inhibit splicing of HPV16 early mRNAs
To enhance our understanding of the regulated expression of the HPV16 early genes, we wished to identify hnRNP proteins that control splicing of HPV16 mR-NAs. We therefore utilized subgenomic HPV16 reporter plasmid pC97ELsLuc that encodes all HPV16 genes and can produce all alternatively spliced HPV16 mRNAs. The schematic representation of the HPV16 genome ( Figure  1A), pC97ELsLuc ( Figure 1B), the structures of the HPV16 alternatively spliced mRNAs it produces ( Figure 1C-E) and the location of the HPV16 RT-PCR primers are displayed in Figure 1. HPV16 plasmid pC97ELsLuc was cotransfected individually with plasmids expressing either hn-RNP A1, hnRNP A2, hnRNP AB, hnRNP C1, hnRNP D, hnRNP DL, hnRNP F, hnRNP G, hnRNP H, hn-RNP I, hnRNP L, hnRNP Q and hnRNP R, RNA was extracted and the various spliced HPV16 mRNAs were monitored by RT-PCR (Figure 2A-C). Quantitation of E6-and E7-encoding mRNA isoforms obtained in Figure  2A Figure 2D,E, respectively. We considered three observations to be of particular interest: (i) hnRNP A1 and hnRNP A2 affected splicing of E6/E7 mRNAs as previously described with different effects on splicing of E6 and E7 mRNAs. hnRNP A1 promoted production of intron-retained E6 mRNAs, while hnRNP A2 enhanced production of 226∧742-mRNAs (46)  . hnRNPs that inhibited E2 mRNA splicing also appeared to inhibit splicing between SD880 and SA3358 (E1∧E4), with the exception of hnRNP I ( Figure 2B). This may be explained by the presence of a long, uninterrupted polypyrimidine tract; eleven consecutive pyrimidine upstream of E2 splice site SA2709, while the E1∧E4 splice site SA3358 has no more than four consecutive pyrimidinenucleotides. Thus, HPV16 SA3358 would be a poor target for polypyrimidine tract binding protein/hnRNP I compared to HPV16 SA2709. All transfections have been repeated at least three times and the indicated RT-PCR products have been cut out from gels and subjected to sequencing to confirm their identities. The significant effects of hnRNP D on the majority of the HPV16 mRNAs warranted further studies on the effects of hnRNP D on HPV16 mRNA splicing.

All four hnRNP D variants promote intron retention of HPV16 E1 and E6 mRNAs
We transfected plasmids expressing all four isoforms of hn-RNP D (hnRNP D37, D40, D42 and D45) ( Figure 3A) with HPV16 plasmid pC97ELsLuc and determined the effect on HPV16 mRNA splicing by RT-PCR ( Figure 1). All four isoforms of hnRNP D are ubiquitously expressed in a variety of cell lines including normal primary human foreskin keratinocytes (nHFK) at the level of RNA (Supplementary Figure S1F). The transfection of each hnRNP D isoform expressing plasmid revealed equal levels of individual isoform protein expression with an increase of 7-fold compared to the endogeneous protein level (Supplementary Figure S1G and H). The results revealed that all hn-RNP D proteins inhibited HPV16 E6/E7 mRNA splicing ( Figure 3B). All hnRNP D isoforms promoted production of intron-retained mRNAs encoding E1 at the expense of the spliced E2 mRNAs (880∧2709) ( Figure 3C). hnRNP D proteins also had an inhibitory effect on HPV16 E6/E7 mRNA splicing, which resulted in production of intronretained E6-encoding mRNAs at the expense of the spliced E7 mRNAs (226∧409 and 226∧526) ( Figure 3B). Furthermore, RT-PCR amplification with primer pair 97S+E1AS revealed the hnRNP D also enhanced production of an HPV16 mRNA that was intron-retained in both E6/E7and E1/E2-regions (Supplementary Figure S1A and B), suggesting that hnRNP D affected early steps in the splicing reaction or independently inhibited splice sites in the E1and E6-coding regions. Finally, the splicing inhibitory effect of hnRNP D on the HPV16 E6/E7 mRNAs (Supplementary Figure S1C) and the E1/E2 mRNAs (Supplementary Figure S1D), and to some extent E4 mRNAs (Supplementary Figure S1E), was dependent on the concentration of transfected hnRNP D plasmid. We concluded that hnRNP D promoted production of intron-retained HPV16 mRNAs encoding E1 or E6.
Next, we investigated if the effect of hnRNP D proteins on HPV16 intron-retained E1 mRNA production could be reproduced independently of upstream splice sites located in E6/E7 gene region. To this end, we used pBEL-sLuc ( Figure 3E) that encodes all HPV16 genes present in pC97ElsLuc except for the E6 and the E7 genes. Since the RT-PCR products representing intron-retained E1 mRNAs ( Figure 3C) could potentially originate from plasmid DNA that contaminated the RNA preparations, we performed RT-PCR in the absence or presence of reverse transcription (RT). As can be seen in Figure 3F, the bands representing intron-retained E1-encoding mRNAs were not detected in the absence of RT, nor were they easily detected in the absence of hnRNP D protein ( Figure 3F). Furthermore, primers located entirely within the E1 coding region (primers 1302S+2293AS) yielded similar results as the 773S and E2AS primers regarding detection of the intronretained E1-encoding mRNAs ( Figure 3G). RT-PCR with primers 773S+E1AS (located immediately downstream of SD880 and detecting intron-retained mRNAs, Figure 3E) also showed an increase in the presence of hnRNP D (Figure 3H). It also appeared that hnRNP D40 had a stronger splicing-inhibitory effect on the HPV16 mRNAs than the remaining members of the hnRNP D family ( Figure 3B-D). Thus, hnRNP D40 was used in the majority of the subsequent experiments. The effect of hnRNP D40 on intronretained E1 mRNA production was determined by RT-qPCR with primers 773S+E1AS revealing an increase of intron-retained E1 mRNAs with 4-fold in the presence of the hnRNP D40 ( Figure 3I). Furthermore, the ratio between intron-retained E1 mRNAs and spliced E2 mRNAs revealed increases of 8-fold from pC97ELsLuc ( Figure 3D) and 6-fold from pBELsLuc ( Figure 3J) in the presence of hnRNP D40. The enhancing effect of hnRNP D40 on production of intron-retained E1 mRNAs was further confirmed by comparisons between pC97ELsLuc and pBEL-sLuc (Supplementary Figure S2A  773S+E2AS) or by transfection with serially diluted hn-RNP D40 plasmid (Supplementary Figure S2D). Taken together, we concluded that hnRNP D proteins inhibited splicing of HPV16 early mRNAs and promoted retention of introns encoding E1 and E6, respectively. Retention of the E1 intron occurred independently of intron retention in the E6 coding region.

The N-terminal domain of hnRNP D40 contributes to induction of intron-retained HPV16 E1 and E6 mRNAs
In an effort to elucidate how hnRNP D40 inhibits HPV16 mRNA splicing, we investigated the effect of deletion mutations in hnRNP D40 ( Figure 4A). Analysis of FLAGtagged, wild-type and mutant hnRNP D40-proteins revealed that all mutants were expressed in the transfected      Figure 2D and E: (D) percentage intron-retained E6 mRNAs, (E) percentage intron-retained E1 mRNAs, (F) percentage 226∧409 mRNAs, (G) percentage 226∧526 mRNAs and (H) percentage 226∧742 mRNAs. Student's t-test was executed and obtained P values were displayed. n.s., no significance. (I and J) Effects of amino acid substitutions in the RG/RGG region of full-length hnRNP D40 on HPV16 mRNA splicing was monitored by HPV16 RT-PCR as described in (B). (K) Interactions between hnRNP D40 and cellular spliceosome factors U1-70K, U2AF65 and U2AF35 were investigated by co-immunoprecipitation (Co-IP) assay. Indicated FLAG-tagged, wild-type or mutant hnRNP D40 expression plasmid was transfected into HeLa cells. Whole cell extracts were subjected to IP by anti-FLAG antibody to purify the FLAG-protein interactome complex, followed by western blotting to detect the presence of endogenous cellular spliceosome factors in the complex. Band intensity was quantitated and IP efficiency was calculated as IP band intensity over corresponding Input band intensity. Fold change over WT D40 of IP efficiency is shown under each western blot image. (L) Interactions between hnRNP D40 and HPV16 mRNAs were investigated by UV-crosslinking and immunoprecipitation (CLIP) assay. HPV16 subgenomic plasmid pC97ELsLuc was cotransfected with FLAG-tagged wild-type or mutant hnRNP D40 expression plasmid into HeLa cells. UV-crosslinked whole cell extracts were subjected to IP by anti-FLAG antibody followed by extraction of RNA from the FLAG-protein interactome complex and RT-PCR using HPV16-specific RT-PCR primers, 97S+438AS or TotalE6F+TotalE6R. The schematic depiction of the location of the HPV16-specific RT-PCR primers is shown below the gel image. Fold change of CLIP efficiency of wild-type and mutant hnRNP D40s over wild-type D40 was calculated as described in (K). Figure 4B). To determine the subcellular localization of these hnRNP D40 proteins, a set of plasmids in which the hnRNP D40 open reading frame was fused to EGFP was also created. All hnRNP D40 proteins were localized to the nucleus, although mutant D8 differed from other mutants by showing stronger cytoplasmic than nuclear localization ( Figure 4C and Supplementary Figure  S3A). Wild-type hnRNP D40 and the N-terminally deleted proteins (D1, D2 and D5) were present primarily in the nucleus, whereas C-terminally deleted proteins (D3, D4, D7 and D9) also showed substantial cytoplasmic staining (Figure 4C and Supplementary Figure S3A). The increased cytoplasmic localization of the C-terminally deleted proteins was a result of deletion of the RGG-domain as substitutions of the arginines in four RGG-motifs for alanine had the same effect ( Figure 5A and Supplementary Figure S4B). We concluded that the nuclear localization of the mutant hnRNP D40 proteins allowed us to determine the effect of these proteins on HPV16 mRNA splicing.

HeLa cells (
Co-transfection of HPV16 reporter plasmid pC97ELsLuc with hnRNP D40 or either of the Nterminally deleted hnRNP D40 plasmids (D1, D2, D5 and D8) ( Figure 4D) revealed that all deletions reduced the ability of hnRNP D40 to promote production of HPV16 intron-retained E6 mRNAs ( Figure 4D,H) and intronretained E1 mRNAs ( Figure 4E,I). Quantitated percent of all isoforms is shown in Supplementary Figure S3B and C. Longer exposed images of Figure 4E,G are available in Supplementary Figure S3D. Quantitation showed that the reduction in splicing was statistically significant ( Figure  4H,I). More specifically, we found that deletion of the N-terminal Alanine-rich region of hnRNP D40 (mutant D5) reduced the ability of hnRNP D40 to promote production of intron-retained HPV16 E6 mRNAs ( Figure  4D,H). Deletion of the exon-2 coding region, in addition to the alanine-rich region, as in hnRNP D40 mutant D1 further reduced induction of intron-retained E6 mRNA ( Figure 4D,H) and intron-retained E1 mRNA ( Figure  4E, I) and unexpectedly activated splicing from SD226 to SA526 (226∧526) ( Figure 4D and Supplementary Figure  S3B). Deletion also of RRM1 domain, as in hnRNP D40 mutant D2, abolished induction of both intron-retained E6 mRNAs ( Figure 4D,H) and intron-retained E1 mRNAs ( Figure 4E, I) and unexpectedly activated splicing from SD226 to SA409 (226∧409) ( Figure 4D and Supplementary Figure S3B). Overexpression of a mutant hnRNP D40 protein consisting only of RRM1 and RRM2 domains (hnRNP D40 mutant D8) had only a minor effect on HPV16 mRNA splicing ( Figure 4D,E), which in this case may be explained by the preferential localization of D8 to the cell cytoplasm ( Figure 4C). We concluded that the N-terminus of hnRNP D40 contributed significantly to the inhibition of HPV16 early mRNA splicing and production of E1-and E6-mRNAs and that RRM1 domain was of particular importance.
Interestingly, we found that hnRNP D40 mutant D1 unexpectedly gained the ability to activate splicing to HPV16 3 -splice site SA526 (226∧526) (Figure 4D), while retaining some of its ability to inhibit splicing of both E6/E7 mRNAs and E1/E2 mRNAs ( Figure 4D,E). These results suggested that deletion of the N-terminal, alanine-rich region and the exon 2 coding sequence weakened the ability of hnRNP D40 to inhibit splicing, while retaining interactions with the splicing machinery, of which the latter was documented by the enhanced splicing to SA526. Hence, interactions of hn-RNP D40 with the splicing machinery and its ability to inhibit splicing could be separated. This interpretation was reinforced by the phenotype of mutant D2 that lost much of its inhibitory effect on the E6/E7 mRNA splicing ( Figure  4D,H) and all of its inhibitory effect on E1 mRNA splicing ( Figure 4E,I) but gained the ability to promote splicing to HPV16 SA409 (226∧409) ( Figure 4D). To determine if similar results were obtained in cells other than HeLa cells, we co-transfected pC97ELsLuc plasmid into 293T cells in the presence of hnRNP D40 or the two N-terminal deletions of hnRNP D40 D1 and D2. Similar results were obtained in the two cell lines (Supplementary Figure S4A): D1 and D2 had lost much of their splicing inhibitory function and activated alternative splicing of the HPV16 E6/E7 mR-NAs (Supplementary Figure S4A). RRM1 and RRM2 apparently played a decisive role in the selection of HPV16 splice site since the presence of RRM1 and RRM2 in D1 activated splicing to HPV16 SA526 (226∧526) ( Figure 4D), whereas RRM2 alone in D2 activated splicing to HPV16 SA409 (226∧409) ( Figure 4D). We concluded that hnRNP D40 RRM1 was essential for efficient splicing inhibition and induction of intron-retained HPV16 E1 and E6 mR-NAs and that the N-terminal alanine-rich region and the exon-2 region of hnRNP D40 contributed to splicing inhibition possibly via different mechanisms.

Nested C-terminal deletions in hnRNP D gradually reduced ability of hnRNP D to inhibit HPV16 mRNA splicing
In contrast to N-terminal deletions of hnRNP D40, either of the C-terminally deleted hnRNP D40 plasmids D3, D4, D7 and D9 ( Figure 4A) promoted production of intronretained E6 mRNAs ( Figure 4F, J), although less efficiently than wild-type hnRNP D40. Inhibition of splicing to SA409 and in particular to SA742 was gradually reduced with larger C-terminal deletions in hnRNP D40 ( Figure 4F and Supplementary Figure S3B). Even plasmid D9 that contained only the N-terminal alanine-rich region and the exon 2 coding region retained some splicing inhibitory activity ( Figure 4F,J). Similarly, splicing inhibition of the HPV16 E1/E2 mRNAs was gradually lost with increasing size of hnRNP D40 C-terminal deletions ( Figure 4G,K). Similar results were obtained with the two-C-terminal deletions D3 and D4 in other cell line (Supplementary Figure S4A), corroborating the observation that C-terminal deletions reduced the inhibitory effect on alternative splicing of HPV16 E6/E7 mRNAs. Taken together, these results supported the idea that the N-terminal region of hnRNP D40 played an important role in inhibition of HPV16 E1/E2 and E6/E7 mRNA splicing and indicated that this effect was enhanced by the C-terminus, suggesting that the C-terminus of hn-RNP D40 may interact with the splicing machinery.

The RGG-domain of hnRNP D40 contributes to its ability to inhibit HPV16 mRNA splicing
In an effort to understand how hnRNP D40 interacted with the splicing machinery, we utilized hnRNP D40 mutants D1 and D2 for further mutagenesis. D1 and D2 have Nterminal deletions but retain either both RRM1 and RRM2 (D1), or only RRM2 (D2). Both D1 and D2 contain an intact C-terminus with the Q-rich region and the RG/RGGrich 'RGG'-region ( Figure 5A). We separately introduced point mutations in the glutamine-rich region (Q) and in the RGG-region in D1 and D2 ( Figure 5A). Substitutions of arginine for alanine in four RG/RGG motifs in the RGGregion in D1 (D1-AGG) did not affect the ability of D1 to promote production of intron-retained E6 mRNA (Figure 5B,D) but abrogated the splicing enhancing effect to HPV16 SA526 (226∧526) (Figure 5B,F) and reduced the ability of D1 to promote production of intron-retained E1 mRNAs ( Figure 5C, E). In contrast, substitutions of all six glutamines in the Q-rich region for alanine (D1-Q6A), did not affect the ability of D1 to enhance splicing to HPV16 SA526 ( Figure 5B,F), nor did they affect the ability of D1 to promote production of intron-retained E1 mRNAs (Figure 5C,E). We concluded that the hnRNP D40 RGG-region was important for enhancement of E6 mRNA splicing to SA526, as well as for inhibition of E1 mRNA splicing. These results were supported by the analysis of hnRNP D40-D2 with the same mutations. Mutations in the RGG domain in D2 (D2-AGG) ( Figure 5A) alleviated splicing enhancement of SA409 ( Figure 5B,G). Interestingly, both D1-AGG and D2-AGG restored HPV16 E6 mRNA splicing to HPV16 SA742 (226∧742) compared to their corresponding parental mutants, D1 and D2, respectively (Figure 5B,H), which is in line with the observation in Figure 4F that sequential C-terminal deletions restored 226∧742. The D2-AGG mutations did not substantially affect production of intron-retained E1 mRNAs ( Figure 5C,E) or intronretained E6 mRNAs ( Figure 5B,D). Similarly to D1-Q6A and D1, D2-Q6A and D2 had similar phenotype ( Figure  5F, G, H). Our results demonstrated that the RGG-domain contributed to the ability of D1 and D2 to activate splicing to alternative HPV16 splice sites in the absence of the splicing inhibitory part in the N-terminus of full-length hnRNP D40. We speculated that the C-terminus of hnRNP D40 interacted with the splicing machinery and that the inhibition of splicing and the interactions with the splicing machinery were two separate properties of hnRNP D40. Finally, substitutions of arginine for alanine in the RGG-region in the context of full-length hnRNP D40 showed a reduction of the splicing inhibitory effect of hnRNP D40 on E6 and E1 mRNAs ( Figure 5I,J). The effects in the context of the fulllength protein were relatively subtle but reproducible and significant (Supplementary Figure S4C). Taken together, our results suggested that the C-terminal RGG-domain of hnRNP D40 contributed to the splicing inhibitory function of hnRNP D40 by interacting with the splicing machinery, while the inhibitory effect on splicing was mediated by the N-terminus of hnRNP D40.

hnRNP D40 interacts with complex A components of the splicing machinery
To determine if hnRNP D40 interacted with the splicing machinery, we investigated if hnRNP D40 co-immunoprecipitated components of the cellular spliceosome. HeLa cells were transfected with plasmids expressing full-length (D40) or D1 or D2 mutants of hnRNP D40. Cell extracts were prepared and subjected to immunoprecipitation by anti-FLAG antibody followed by western blotting with antibodies to spliceosome components U1-70K, U2AF65 or U2AF35, or with anti-FLAG antibody. Our results revealed that wild-type hnRNP D40 co-immunoprecipitated cellular U1 snRNP component U1-70K and U2 auxiliary factors U2AF65 and U2AF35 ( Figure 5K). The two N-terminal mutants of hnRNP D40, D1 and D2, retained the interactions with cellular U1 snRNP components U1-70K and U2AF65 and U2AF35, albeit less efficiently than full-length hnRNP D40 ( Figure  5K). Furthermore, the branch-point adenosine recognizing complex component, splicing factor 3b (SF3b) and one of the translation regulators, cytosolic poly(A)-binding protein-1 (PABP-C1) interacted with full-length hnRNP D40 as well as with the N-terminal mutants D1 and D2 (Supplementary Figure S4D). These results not only demonstrated that the C-terminus of hnRNP D40 was sufficient for interactions with these spliceosomal complex A components as well as with downstream RNA-processing regulators but also revealed that the N-terminus of hnRNP D40 contributed to the efficiency of these interactions. We concluded that hnRNP D40 protein interacted with complex A components of the splicing machinery.

hnRNP D40 interacts with HPV16 mRNAs in an hnRNP D40 RRM1-dependent manner
Next, we investigated if hnRNP D40 or deletion mutants thereof (D1 and D2) interacted with HPV16 RNA. Plasmids expressing FLAG-tagged hnRNP D40, D1 or D2 were co-transfected with HPV16 reporter plasmid pC97ELsLuc into HeLa cells followed by UV-crosslinking and Immunoprecipitation (CLIP) with affinity purification of FLAG tagged hnRNP D40-RNA complexes. FLAGantibody-immunoprecipitated UV-crosslinked RNA was extracted and subjected to RT-PCR with HPV16 specific primers ( Figure 5L). The results revealed that immunoprecipitation of wild-type hnRNP D40 also immunoprecipitated HPV16 RNA detected by HPV16 primer pair TotalE6F+TotalE6R that amplifies cDNA representing all HPV16 mRNAs transcribed from pC97ELsLuc (Figure 5L). Analysis of the HPV16 RNA with primer pair 97S+438AS that discriminates between intron-retained and spliced HPV16 E6 mRNAs revealed that hnRNP D40 interacted with both intron-retained and spliced E6 mR-NAs ( Figure 5L). Deletion of N-terminal alanine-rich and exon-2 region (D1 mutant) did not affect the interactions with HPV16 RNA but deletion of RRM1 did (D2 mutant) ( Figure 5L). We concluded that hnRNP D40 interacts with HPV16 mRNAs directly in an hnRNP D40 RRM1dependent manner.

hnRNP D40 interacts with the entire intronic region of the E6 coding region as well as with a previously identified HPV16 splicing silencer in E7 region
Having established that hnRNP D40 interacts with HPV16 mRNAs ( Figure 5L) and with components of the splicing machinery ( Figure 5K) thereby inhibiting HPV16 early Nucleic Acids Research, 2022, Vol. 50, No. 5 2795 mRNA splicing and promoting intron retention and production of intron-retained HPV16 E1 and E6 mRNAs, we wished to identify the HPV16 sequences that were targeted by hnRNP D40. For this purpose, we focused on the E6/E7 coding region as it is shorter than the E1 coding region. To this end, we performed RNA-mediated pull downs of hnRNP D40 with overlapping, biotinylated RNA oligonucleotides that covered the E6-and E7-coding regions from HPV16 nucleotide position 178 to 875 ( Figure  6A,B). This experimental approach was chosen since hn-RNP D40 may not necessarily bind directly to HPV16 mR-NAs. Cell extracts were prepared from HeLa cells transfected with FLAG-tagged hnRNP D40. These extracts were subjected to RNA-mediated protein pull downs with the RNA oligos indicated in Figure 6A,B, followed by western blotting with anti-FLAG antibody ( Figure 6A,B). The results revealed that hnRNP D40 was specifically pulled down by a subset of the RNA oligos. Of particular interest was that all RNA oligos spanning the E6-encoding intron between HPV16 5 -splice site SD226 and 3 -splice site SA409 pulled down hnRNP D40 ( Figure 6A,B, highlighted with a double-headed arrow). Furthermore, RNA oligos 579-604 and 594-620 that spanned a previously identified splicing silencer (46) also pulled down hnRNP D40 ( Figure  6A,B). Additional RNA oligos that pulled down hnRNP D40 did not coincide with known RNA elements except for oligos 721-755 that encoded HPV16 3 -splice site SA742 ( Figure 6A,B). We concluded that interactions of hnRNP D40 with a previously identified splicing silencer located between HPV16 nucleotide positions 594 and 604, as well as with the intronic sequences in the E6 coding region (243-445), correlated with the ability of hnRNP D40 to inhibit splicing between SD226 and SA409 and to promote retention of the intron between the same two splice sites.

hnRNP D40 promotes intron retention of HPV16 E6 mRNAs independently of the HPV16 E1 sequences
Since hnRNP D40 interacted with sequences in the E6 coding region, we investigated if hnRNP D40 could inhibit E6 mRNA splicing and induce intron retention of the E6 mR-NAs in the absence of E1 sequences. We had previously showed that hnRNP D40 inhibited splicing and induced intron retention of the E1/E2 mRNAs in the absence of E6 and E7 sequences (see effect of hnRNP D40 on pBEL-sLuc ( Figure 3E-J, Supplementary Figure S2)). To this end we co-transfected hnRNP D40 plasmid with HPV16 plasmid pX656 that starts at HPV16 nucleotide position 97 and ends at HPV16 position 656 and does not overlap with the previously used HPV16 plasmid pBELsLuc (Supplementary Figure S5A and B). hnRNP D40 inhibited splicing of HPV16 E6/E7 mRNAs produced from pX656, albeit less efficiently than it inhibited splicing of E6/E7 mRNAs produced from pC97ELsLuc ( Figure 6C,D), demonstrating that hnRNP D40 can inhibit splicing in the HPV16 E6/E7coding region independently of E1 sequences. The pX656 contained a previously identified splicing silencer located between HPV16 positions 594 and 604 in addition to the E6-encoding intron with HPV16 splice sites SD226, SA409 and SA526, but the pull-down experiments displayed in Figure 6A demonstrated that the major sites of interactions be-tween hnRNP D40 and HPV16 E6/E7 mRNAs were within the E6 intron between SD226 and SA409. Thus, our results predicted that a smaller plasmid containing only a part of the E6 coding region encompassing SD226, SA409 and the E6 intron (without splicing silencer in the E7 coding region) should also respond to hnRNP D40 overexpression. HPV16 plasmid pX478 that contains HPV16 sequences from HPV16 nucleotide position 97 to position 478 (Supplementary Figure S5C) was transfected into HeLa cells in the absence or presence of hnRNP D40 expression plasmid. RT-PCR analysis of HPV16 mRNAs revealed that hnRNP D40 inhibited splicing and promoted production of intron-retained HPV16 mRNAs from this minimal HPV16 pX478 expression plasmid ( Figure 6C,D). As expected, hn-RNP D40 deletion mutant D1 also inhibited splicing of the HPV16 mRNAs produced from plasmid pX478 but less efficiently than wild-type hnRNP D40 ( Figure 6C,D). We concluded that hnRNP D40 promoted intron retention of HPV16 E6 mRNAs independently of HPV16 E1 sequences.

hnRNP D40 increases the levels of HPV16 intron-retained E6 mRNAs in the cytoplasm
Our results also suggested that hnRNP D40 should increase E6 protein production at the expense of E7 protein production as a result of its splicing inhibitory function of HPV16 E6/E7 mRNAs. Analysis of HPV16 E6 and E7 protein production from pC97ELsLuc transfected in the absence or presence of hnRNP D40 expression plasmid unexpectedly revealed that overexpression of hnRNP D40 suppressed production of both E6 and E7 protein ( Figure 6E). Similar results were obtained with hnRNP D40 mutant D1 ( Figure 6E). hnRNP D40 also inhibited E6 and E7 protein production from a smaller HPV16 expression plasmid named pXH856F that encodes only E6 and E7 ( Figure  6F and Supplementary Figure S5D). This effect was independent of the splicing process itself since overexpression of hnRNP D40 inhibited E6 and E7 protein production also from a version of the pXH856F plasmid in which the major HPV16 5 splice site SD226 was mutationally inactivated, plasmid pXH856SDmF ( Figure 6F and Supplementary Figure S5E). These results suggested that hnRNP D40 was associated with HPV16 mRNAs also in the absence of RNA splicing and as such, could potentially affect other RNA processing steps, thereby explaining the reduction in E6 protein levels despite the increase in the levels of intronretained E6 mRNAs.
To determine if hnRNP D40 affected export of intronretained E6 mRNAs, we analyzed the subcellular distribution of the intron-retained E6 mRNAs produced from the splicing deficient plasmid pXH856SDmF in the absence or presence of hnRNP D40. The intron-retained E6 mR-NAs produced from pXH856SDmF were present in the nucleus and in the cytoplasm ( Figure 6G). Unexpectedly, the levels and subcellular distribution of E6 mRNAs were similar in the absence or presence of hnRNP D40 ( Figure  6G,H). RT-PCR of intron-retained and spliced actin RNA served as a control for proper fractionation of the transfected cells ( Figure 6G). We concluded that the nuclear export of intron-retained E6 mRNAs transcribed from the splicing deficient plasmid pXH856SDmF was not consid-  Figure 6. hnRNP D40 interacted with HPV16 E6E7 RNA sequences in vitro and influenced HPV16 E6E7 mRNA processing. (A) RNA oligonucleotidemediated pull-down assay using sequential short RNA oligos spanning the entire HPV16 E6-and E7-coding regions. The numbers in each RNA-oligo name represent HPV16 nucleotide positions and refer to nucleotide numbers in the reference HPV16 genome (GenBank: K02718.1). Thick double-headed arrow: intronic region of E6/E7 mRNAs located between SD226 and SA409. (B) Schematic representation of RNA splicing in the HPV16 E6 and E7 coding region. Intron-retained E6 mRNA is depicted as well as spliced HPV16 mRNAs 226∧409, 226∧526 and 226∧742. Box with S: previously identified E6/E7 mRNA splicing silencer element (46). RNA oligo sequences are available in Supplementary Table S3 Figure S5. (G) Subcellular distribution of intron-retained E6 mRNAs produced from pXH856SDmF in the absence (−) or presence (+) of hnRNP D40 plasmid. Nuclear and cytoplasmic fractions were prepared from the transfected cells and RNA was extracted and subjected to HPV16 RT-PCR using primers TotalE6F and 757AS (for primer location, see Figure 1C and Supplementary Figure S5E). Cell fractionation was validated by RT-PCR analysis of unspliced actin RNA located exclusively in the nuclear fraction while spliced actin mRNAs found in both fractions. (H) qPCR using cDNA samples from (G) to evaluate levels of intron-retained E6 mRNAs in nuclear and cytoplasmic fractions in the absence or presence of hnRNP D40. Primers TotalE6F and 234AS were used (for primer location, see Figure 1C and Supplementary Figure S5E erably affected by hnRNP D40 (Figure 6G,H). Thus, inefficient nuclear mRNA export of intron-retained E6 mRNA in the presence of hnRNP D40 could not explain the significant inhibition of E6 protein production exerted by hn-RNP D40 ( Figure 6F). To exclude the possibility that the mutation of HPV16 5 -splice site SD226 in pXH856SDmF affected the RNA export regulation mediated by hnRNP D40, the effect of hnRNP D40 on the subcellular distribution of E6 and E7 mRNAs was investigated using pXH856F encoding wild-type SD226. The inhibitory effect of hnRNP D40 on HPV16 E6/E7 mRNA splicing was observed however the effect appeared to be mainly against 226∧409 (Figure 6I). We observed that the intron-retained E6 mRNAs were to a large extent restricted to the nuclear fraction while the levels of the spliced, HPV16 226∧409-mRNAs were higher in the cytoplasm than in the nucleus ( Figure 6I, top panel, lanes 1-4). In the presence of hnRNP D40, primarily intron-retained E6 mRNAs were detected both in the nucleus and cytoplasm ( Figure 6I, top panel, lanes 5-8).
The results were confirmed using a primer pair that only detected intron-retained E6 mRNAs and mRNAs spliced between splice sites SD226 and SA409 (226∧409) ( Figure  6I, second panel), demonstrating a statistically significant increase of cytoplasmic intron-retained E6 distribution in the presence of hnRNP D40 ( Figure 6N). These results indicated that hnRNP D40 did not inhibit the nuclear export of intron-retained E6 mRNAs. Thus, inhibition of nuclear export of intron-retained E6 mRNAs could not explain the reduction of E6 protein levels in the presence of hnRNP D40 observed in Figure 6E and F. This observation was consistent with results obtained using HPV16 subgenomic plasmid pC97ELsLuc ( Figure 6J, K and O). Furthermore, we showed that hnRNP D40 enhanced the levels of cytoplasmic intron-retained E6 mRNAs also from the smaller HPV16 subgenomic plasmid pX478 ( Figure 6M). This increase was statistically significant ( Figure 6P). Taken together, we concluded hnRNP D40 had multiple impacts and affected various steps of HPV16 E6 and E7 mRNA processing: hnRNP D40 interacted with E6/E7 mRNAs at multiple positions, it inhibited E6/E7 mRNA splicing independently of the E1 region, and increased levels of intronretained E6 mRNAs in the cytoplasm. However, hnRNP D40 also significantly reduced the levels of E6 protein produced from the intron-retained E6 mRNAs.

hnRNP D40 increased the levels of HPV16 intron-retained E1 mRNAs in the cytoplasm
In addition to the intron-retained E6 mRNAs produced from pC97ELsLuc, intron-retained E1 mRNAs transcribed from pC97ELsLuc were also increased in both nucleus and cytoplasm ( Figure 6L). To confirm this effect also in the absence of E6/E7 sequences on the mRNAs, we used HPV16 subgenomic reporter plasmid pBELsLuc that lacks E6E7 genes ( Figure 3E). pBELsLuc was transfected into HeLa cells in the absence or presence of hnRNP D40 plasmid followed by fractionation into nuclear and cytoplasmic fractions prior to RNA extraction and RT-PCR. As can be seen in Figure 7A, spliced E2 mRNA (880∧2709) levels were reduced in the presence of hnRNP D40, as expected, but they were present in both nuclear and cytoplasmic fractions, al-though predominantly in the cytoplasm. In sharp contrast to the spliced E2 mRNAs, the intron-retained E1 mRNAs were detected primarily in the presence of hnRNP D40 (Figure 7A). In the presence of hnRNP D40, the intron-retained E1 mRNAs were present in both the nuclear and cytoplasmic fractions ( Figure 7A), demonstrating that hnRNP D40 not only inhibited splicing of the HPV16 E1/E2 mRNAs and promoted intron retention to generate intron-retained E1 mRNAs, it also allowed these mRNAs to be exported to the cytoplasm. A similar effect was observed with hnRNP A1 ( Figure 7A) an hnRNP protein evolutionarily close to hnRNP D protein family (54) and initially shown to inhibit HPV16 E6/E7 and E1/E2 mRNA splicing (Figure 2A,C). Actin mRNAs served as controls for cellular fractionation: unspliced actin mRNAs were present primarily in the nuclear fraction, whereas spliced actin mRNAs were detected in both fractions ( Figure 7B). Nuclear restricted Lamin B and SRSF2 and the shuttling SRSF1 protein served as additional controls for cellular fractionation (Supplementary Figure S6). We also performed RT-qPCR with primers 773s and E1AS on the cytoplasmic fractions, a primer pair that is specific for intron-retained HPV16 E1 mRNAs ( Figure  1C). The levels of HPV16 intron-retained E1 mRNAs in the cytoplasm increased more than four-fold in the presence of hnRNP D40 ( Figure 7C). We concluded that hnRNP D40 inhibited HPV16 E1/E2 mRNA splicing, promoted intron retention to generate intron-retained E1 mRNAs and enhanced the appearance of the E1 mRNAs in the cytoplasm independently of E6/E7 sequences on the mRNAs. The HPV16 subgenomic plasmid pBELEN encodes only the HPV 16 E1 gene including all splice sites in the E1 coding region (Supplementary Figure S7B). pBELEN responded to hnRNP D40 by enhanced production of intronretained E1 mRNA (Supplementary Figure S7D). RT-PCR analysis of HPV16 mRNAs produced from pBELEN in the absence or presence of hnRNP D40, revealed that the level of HPV16 intron-retained E1 mRNA increased in both nucleus and cytoplasm, whereas the levels of HPV16 spliced E2 mRNAs (880∧2709) were reduced in the presence of hnRNP D40 (Supplementary Figure S7E). This result indicated that redistribution of intron-retained E1 mRNA to the cytoplasm was enhanced in the presence of hnRNP D40 (Supplementary Figure S7F). For further confirmation, we generated a plasmid that produced a shorter intron-retained E1 mRNA that would be easier to detect and quantitate by RT-PCR since the predicted size of the entire intronretained E1 mRNA produced from pBELEN is relatively big (>2 kb). We therefore introduced a deletion in the E1 coding region in pBELEN, between the HPV16 major splice sites SD880 and SA2709, to generate a smaller plasmid named pBELENdE1 (Supplementary Figure S7C). Detection of intron-retained 'E1' mRNAs was considerably improved when pBELENdE1 was used ( Figure 7D lanes 1-4, compared to Figure 7A). RT-PCR analysis of HPV16 mR-NAs produced from pBELENdE1 in the absence or presence of hnRNP D40, revealed that the levels of HPV16 intron-retained mRNAs were increased in both nucleus and cytoplasm ( Figure 7D), an increase that was statistically significant ( Figure 7E), whereas the levels of HPV16 spliced E2 mRNAs (880∧2709) were reduced in the presence of hnRNP D40, and were located primarily in the cytoplas-  Figure 7. hnRNP D40 upregulated levels of HPV16 intron-retained E1 mRNAs in the cytoplasm and interacts with these mRNAs in the cytoplasmic fraction. (A) HeLa cells were transfected with pBELsluc plasmid in the absence (−) or presence of plasmids expressing either hnRNP D40 or hnRNP A1. Nuclear (N) and cytoplasmic (C) fractions were prepared from the transfected cells and RNA was extracted and subjected to HPV16-specific RT-PCR using primer pair 773S+E2AS. (B) Cellular fractionation was validated by analysis by RT-PCR of unspliced and spliced actin mRNAs. (C) Levels of HPV16 intron-retained E1 mRNAs in the cytoplasmic fraction was determined by RT-PCR with primers 773S+E1AS. Quantification of electrophoresed RT-PCR products were performed as described and normalized to RT-PCR products representing cytoplasmic spliced actin mRNA. Fold over control (−) is shown. (D) Effect hnRNP D40 on subcellular localization of HPV16 intron-retained E1 mRNAs and spliced E2 mRNAs produced from HPV16 subgenomic plasmid pBELENdE1 (for schematic representation of pBELENdE1, see Supplementary Figure S7C). pBELENdE1 was transfected into HeLa cells in the absence (−) or presence (+) of hnRNP D40 plasmid. RT-PCR was performed on nuclear or cytoplasmic RNA using HPV16-specific primers pair 773S+E2Xba that detects HPV16 intron-retained E1 mRNAs as well as spliced mRNAs (880∧2709). (E) Percentage cytoplasmic intronretained HPV16 E1 mRNAs over a total sum of nuclear and cytoplasmic intron-retained E1 mRNAs in absence (−) or presence of hnRNP D40 from (D). (F) Effect of hnRNP D40 on HPV16 E1 protein levels was determined by western blotting on extracts form HeLa cells transfected with HPV16 E1-FLAG expressing plasmid p16E1-3xF (wild type) or p16E1SDm-3xF (harboring mutations at splice donors SD880 and SD1302 in the E1 gene thereby abolishing E2 mRNA splicing), in the absence (−) or presence (+) of hnRNP D40 expressing plasmid. Western blotting using anti-FLAG antibody (M2). (G) The levels of HPV16 intron-retained E1 mRNAs produced by p16E1SDm-3xF in nuclear and cytoplasmic fractions, in the absence (−) or presence (+) of hnRNP D40 expression plasmid, was determined by HPV16-specific RT-PCR primers primer pairs F-E1-1+E2Xba and PstI-SD880mE1-F+E2Xba (primer positions are displayed in Supplementary Figure S7). (H) Overexpression of hnRNP D40 does not affect production of sLuc from a CMV-promoter driven sLuc gene. (I) HeLa cells were transfected with indicated, FLAG-tagged hnRNP D expression plasmids. Nuclear and cytoplasmic extracts were subjected to Western blotting with anti-FLAG antibody or anti-tubulin or anti-histone antibody to control for subcellular fractionation. (J) Association between HPV16 E1 mRNAs and hnRNP D40 protein in the cytoplasm demonstrated by CLIP assay on cytoplasmic extracts from HeLa cells transfected with HPV16 subgenomic plasmid pC97ELsLuc in the absence or presence of FLAG-hnRNP D40 expressing plasmid. Cytoplasmic extracts were subjected to immunoprecipitation with anti-FLAG antibody to purify FLAG-hnRNP D40: RNA complexes. RNA in the ribonucleoprotein (RNP) complex was extracted and subjected to RT-PCR using primers 773S and E1AS that specifically detect HPV16 intron-retained E1 mRNAs. mic fraction ( Figure 7D). Of particular interest was that the proportion of cytoplasmic intron-retained HPV16 mR-NAs was significantly increased in the presence of hnRNP D40 ( Figure 7E). These results confirmed that hnRNP D40 inhibited HPV16 mRNA splicing and promoted intron retention to generate intron-retained HPV16 E1 mRNAs that were exported to the cytoplasm.
Finally, we wished to determine if the increase of intronretained E1 mRNAs in the cytoplasm caused by hnRNP D40 also resulted in increased E1 protein levels. We generated an HPV16 E1 expression plasmid in which the E1 gene was fused with three-times FLAG sequence at its 3 -end named p16E1-3xF (Supplementary Figure S7G). This plasmid contained intact HPV16 5 -splice sites SD880 and SD1302 and had the potential to produce an intronretained mRNA encoding a 73kD, 3xFLAG-tagged HPV16 E1 protein, as well as mRNAs spliced to HPV16 3 -splice site SA2709. In a second version of this plasmid, SD880 and SD1302 were mutationally inactivated to generate plasmid p16E1SDm-3xF (Supplementary Figure S7H). As expected, HPV16 E1 protein production was greatly improved from p16E1SDm-3xF compared to p16E1-3xF ( Figure 7F), but overexpression of hnRNP D40 decreased E1 protein levels from both plasmid p16E1-3xF and p16E1SDm-3xF ( Figure 7F). Analysis of the mRNAs produced by p16E1-3xF and p16E1SDm-3xF revealed that intron-retained HPV16 E1 mRNA levels transcribed from p16E1SDm-3xF were increased in the cytoplasm in the presence of hnRNP D40 ( Figure 7G). The results were confirmed by using different sets of primer pairs (the primer positions are indicated in Supplementary Figure S7H). The results revealed that the levels of intron-retained E1 mRNAs increased in the presence of hnRNP D40 even though HPV16 5 -splice sites SD880 and SD1302 had been inactivated ( Figure 7G), which indicated that hnRNP D40 also functioned independently of splicing. This effect was not due to transcriptional activation by hnRNP D40 since hnRNP D40 does not affect transcription from CMV-promoter driven plasmids ( Figure  7H). Taken together, these results were reminiscent of the effect of hnRNP D40 on the HPV16 E6 mRNA and E6 protein levels ( Figure 6). Our results indicated that hnRNP D40, in addition to its splicing inhibitory function, interacted with HPV16 E1 mRNAs also in the absence of RNA splicing and promoted the appearance of intron-retained HPV16 E1 mRNAs in the cytoplasm. However, these mR-NAs were poorly translated into E1 protein and the results suggested that hnRNP D40 negatively interfered with mRNA translation. Indeed, translation in vitro of both E6 and E1 mRNAs into E6 and E1 proteins was inhibited by recombinant hnRNP D protein but not by bovine serum albumin (BSA) (Supplementary Figure S8A, B, D and E). This translational inhibition was specific for HPV16 E6 and E1 mRNAs since luciferase mRNA translation was unaffected by the presence of recombinant hnRNP D protein (Supplementary Figure S8C).

hnRNP D40 is associated with HPV16 mRNAs in the cytoplasm
The results presented above suggested that hnRNP D not only inhibited HPV16 mRNA splicing but also might have the ability to accompany the HPV16 mRNAs to the cytoplasm. As already shown in Figure 4C, overexpressed hnRNP D40-EGFP was primarily located in the nucleus. However, hnRNP D is known to be a shuttling protein and small amounts of hnRNP D40 could potentially be present in the cytoplasm. We therefore transfected HeLa cells with a plasmid expressing a FLAG-tagged hnRNP D40 protein, fractionated the cells and analyzed nuclear and cytoplasmic fractions for FLAG-tagged hnRNP D40 by western blotting. The results revealed that the FLAGtagged hnRNP D40 produced from the expression plasmid could be detected in both nuclear and cytoplasmic compartments ( Figure 7I), whereas tubulin was detected primarily in the cytoplasmic-and histone in the nuclearfractions ( Figure 7I). To determine if hnRNP D40 was associated with the intron-retained HPV16 E1 mRNAs in the cytoplasm, we performed a CLIP assay on cytoplasmic extracts from HeLa cells transfected with HPV16 reporter plasmid pC97ELsLuc in the absence or presence of transfected pFLAG-hnRNP D40 plasmid. As can be seen from the results, intron-retained HPV16 mRNAs could be specifically amplified from the UV-crosslinked RNA-protein complexes immunoprecipitated with anti-FLAG antibody ( Figure 7J). These results demonstrated that HPV16 intronretained E1 mRNAs were associated with hnRNP D40 in the cell cytoplasm, suggesting that hnRNP D40 accompanied the intron-retained HPV16 mRNAs from the nucleus to the cytoplasm where it potentially interfered with HPV16 mRNA translation in a negative manner. hnRNP D40 promotes intron retention and production of E1 and E6 mRNAs produced from episomal HPV16 genomes in human primary keratinocytes Finally, we wished to determine if hnRNP D40 was actively inhibiting splicing also of mRNAs produced from complete, episomal HPV16 genomes. We therefore used plasmid pHPV16AN that encodes the entire HPV16 genome ( Figure 8A) (43). The pHPV16AN plasmid was cotransfected with cre-recombinase-expressing plasmid to release the HPV16 genome between the loxP sites and generate the episomal form of the HPV16 genome ( Figure 8A). Transfections were performed in the absence or presence of hn-RNP D40 expression plasmid. As can be seen in Figure  8B, hnRNP D40 inhibited splicing of the E6/E7 mRNAs and promoted production of intron-retained E6 mRNAs ( Figure 8B). A relatively minor inhibitory effect on splicing from SD880 to SA3358 was observed ( Figure 8C). Importantly, the inhibitory effect on the E2 mRNAs was readily observed, followed by intron retention and generation of intron-retained E1 mRNAs ( Figure 8D). Since transfection of cells with pHPV16AN is relatively inefficient, the larger, intron-retained E1 mRNAs may be relatively inefficiently amplified with primer pair 773S and E2AS. Therefore, we also monitored the levels of intron-retained E1 mR-NAs with primers 773S and E1AS ( Figure 1E) that also detected an increase in the levels of E1 mRNAs in the presence of hnRNP D40 ( Figure 8E). These results were reproduced by transfection of pHPV16AN in the absence or presence of hnRNP D40 plasmid in human primary keratinocytes. As can be seen, hnRNP D40 inhibited splic-