In this study, we used an siRNA approach to explore the complexity of the spliced 16E6E7 and 18E6E7 RNA isoforms and dissected their biological functions. We demonstrated that the majority of the viral oncoprotein E7 is derived from spliced E6*I mRNAs in cervical cancer cells, as well as in cells transfected with corresponding expression vectors. Full-length, bicistronic E6E7 mRNAs retaining intron 1 in the E6 coding region are responsible mainly for E6 production. Use of a synthetic, double-stranded, short RNA as siRNAs has been widely applied to the study of gene function (13
). RNAi mediated by introduced synthetic siRNA duplexes has been used to silence HPV oncogene expression (5
) with some success. However, lack of knowledge of bicistronic E6E7 RNA and its splicing in regulation of E6 and E7 translation make the results in those studies somewhat difficult to interpret. In this study, we designed all of our siRNAs based on our knowledge of the bicistronic 16E6E7 and 18E6E7 mRNA structures and alternative splicing. The siRNAs designed in this report have met at least six of eight criteria of siRNA design (25
) and are highly effective and oncogene specific compared to other published siRNAs. Importantly, we were able to provide direct evidence of production of HPV16 or HPV18 E7 from spliced E6*I mRNAs. Use of RNA splicing to regulate E6 and E7 expression in high-risk HPVs might be a selective advantage to these viruses in carcinogenesis. Our findings on E6 and E7 production from a separate RNA transcript also provided us with a novel therapeutic intervention of individual oncogenes. It has been debated for decades how a high-risk E7 oncoprotein could be expressed from a bicistronic E6E7 mRNA. The main reason for this debate is that the E6 ORF in a bicistronic E6E7 mRNA is too close to the E7 ORF. Considering that ribosomes scan an mRNA without an internal ribosome entry site sequence in a linear fashion starting from the RNA 5′ end, this limited space between the E6 ORF and the E7 ORF would not give enough room or time for a scanning ribosome to release all of its termination components and to reload all of the necessary translation components to reinitiate translation of the E7 ORF on the same bicistronic mRNA (19
). Because of this, two possible models have been proposed for how E7 could be translated: translation termination-reinitiation (29
) and leaky scanning-initiation (34
). Our previous report and data from this study favor translation termination-reinitiation as the mechanism for high-risk E7 production based on the following evidence. (i) E6*I mRNAs expressed either from E6*I cDNA constructs or from an actively spliced E6*I mRNA produce abundant E7. (ii) Full-length, intron 1-containing E6 mRNAs that translate little or no E7 are present in minimal amounts in most HPV16- or HPV18-positive cervical cancer cells (3
). (iii) E6*I mRNAs are abundant transcripts in almost all HPV16- or HPV18-positive cervical cancer cells examined (3
). However, we do not exclude the possibility that a small fraction of minimal leaky scanning is also involved in translation of E7 from spliced E6*I mRNAs.
The translation termination-reinitiation model states that when the ribosome reaches the termination site of an upstream ORF, the 60S ribosomal subunit is released while the 40S subunit remains bound to the mRNA, resumes scanning, and initiates another round of translation at a downstream AUG codon. For the downstream reinitiation event to occur, the 40S subunit must reacquire Met-tRNAi
; this is promoted by lengthening the intercistronic domain, which provides more time for Met-tRNAi
to bind (18
), or by increasing the concentration of eukaryotic initiation factor 2 (15
). The efficient expression of E7 from E6*I cDNA constructs (pTMF58 for HPV16 and pTMF59 for HPV18) and the constructs that transcribe actively spliced E6E7 pre-mRNAs fits into this model well, since splicing of the E6 intron increases the distance from 2 (HPV16) or 8 (HPV18) nt to >130 nt between the two ORFs. This hypothesis was further supported by the data in this study that shortening the intercistronic distance of the E6*I ORF from the E7 ORF blocks E7 translation.
The leaky scanning model states that, when the first AUG codon occurs in a strong context, ANNaugN or GNNaugG, all or almost all ribosomes stop and initiate at that point. However, when the first AUG resides in a very weak context, lacking both R in position −3 of the AUG and G in position +4 of the AUG, some ribosomes initiate at that point but most continue scanning and initiate farther downstream. This leaky scanning enables the production of two separately initiated proteins from one mRNA (19
). However, both the 16E6 and 18E6 AUGs are in a good context for ribosome recognition. The leaky scanning model proposed for E7 translation, which was mainly based on in vitro translation (34
) and in vivo expression assays using a vaccinia virus vector that bypasses nuclear RNA splicing events (33
), would therefore not explain why the plasmids pTMF54 (HPV16) and pTMF56 (HPV18), which transcribe the full-length E6E7 mRNAs with a 5′ ss mutation, expressed no detectable HPV16 and little HPV18 E7 (Fig. ). This argument against the leaky scanning model on high-risk E7 production was further supported by our observation that making the E6*I ORF in the same distance as the E6 ORF upstream of the E7 ORF in an E6*I mRNA prevents E7 translation (Fig. ). More obviously, the proposed leaky scanning model does not take account the E6*I mRNA as an RNA template for E7 translation, but instead argues for E7 production independent of RNA splicing (34
The intron 1-specific siRNAs in this study induced not only extremely high levels of p53 and p21 but also a partial accumulation of hypophosphorylated p105Rb
and cell cycle arrest at G1
. Phosphorylation of the tumor suppressor protein pRb is dependent on Cdk, and p21 inhibits Cdk activity and interacts with E2F. Therefore, increased p21 expression in cells treated with the intron 1-specific siRNA would prevent hypophosphorylated pRb from being phosphorylated by Cdk (8
) and therefore from promoting cell cycle progression. However, high-risk E7 in HPV16- or HPV18-positive cells, in addition to interacting with hypophosphorylated pRb proteins, also forms a physical complex with p21 and inhibits p21 action (11
). Conceivably, the profound accumulation of hypophosphorylated p105Rb
and substantial cell cycle arrest at G1
observed in the cells treated with the exon 2-specific siRNA was the outcome of all of those actions. It was noted (Fig. and ) that siRNA 198 and siRNA 220 can also induce p53 and p21, because their targeted regions are also the 3′ UTR of each E6 mRNA. The observed up-regulation of p53 and p21 expression by each exon 2-specific siRNA must therefore be a consequence of the degradation of the much less abundant E6 mRNAs, and a much higher percentage of the cells that were arrested at G1
phase in the siRNA 198-treated CaSki cells and in the siRNA 220-treated HeLa cells could be attributed to coexistence of more growth-suppressive pRb-E2F complexes and an enhanced expression of p21 in these cells.
The principle of how siRNA functions has been gradually elucidated and the pathway by which an siRNA mediates degradation of targeted mRNAs has been established. In general, siRNA-mediated RNA degradation occurs in an RNAi-induced silencing complex in the cytoplasmic mRNA decay center, P-bodies (20
). In this study, an intron 1-specific siRNA 219 was found, at a relatively high dose, to target HPV18 E6*I mRNAs for degradation in HeLa cells. Because E6*I lacks intron 1, the targeting must have occurred at the pre-mRNA level, before RNA splicing in the nucleus. Since the intron 1-specific siRNA 209 has only minimal activity in CaSki cells, this suggests that the observed nuclear activity varies from cell type to cell type, most likely due to differences in the nuclear membrane structures and other variables. While the manuscript was in preparation for publication, Robb et al. reported a similar observation of siRNA-mediated nuclear degradation of targeted nuclear RNAs (26
). Together, these independent findings clearly indicate that a functional siRNA pathway exists in the nucleus, at least in HeLa cells.