3. Results
3.1 Virological findings
It is well known that bovine δPVs are the most important infectious agents involved in the etiopathogenesis of the majority of bovine urothelial tumours (Campo et al., 1992; Roperto et al., 2010a). E5 oncoprotein expression is correlated with the transformation of both mesenchymal and epithelial cells to form benign and malignant tumours (Suprynowicz et al., 2006). Therefore, we attempted to verify whether the δPV E5 oncoprotein was expressed in the examined samples. First, we detected E5 oncoprotein transcripts by RT-PCR, the sequencing of which showed 100% identity with BPV-2 and BPV-13 sequences deposited in GenBank (Accession numbers M20219.1 and JQ798171.1, respectively) (Supplemental Fig. S1). Furthermore, Western blot analysis revealed the expression of E5 oncoprotein, which showed that abortive infection, a prerequisite for immune evasion and cell proliferation, takes place (Fig. 1).
3.2 Expression of tripartite motif-containing 25 (TRIM25) and Riplet proteins.
As many viruses, including human papillomavirus, have E3 ubiquitin ligases as their targets (Ban et al., 2018), we wondered whether the bovine δPV E5 oncoprotein might interact with some ligases involved in the antiviral innate immune response mediated by RLRs, the ubiquitination of which appears to be a key post-translational modification. However, the molecular mechanisms of ubiquitin-mediated RIG-I and MDA5 activation remain to be fully understood (Shi et al., 2017; Oshiumi, 2020).
Many studies have reported that TRIM25 and Riplet are two essential E3 ubiquitin ligases for RIG-I signalling as they are known to ubiquitinate and activate RLRs (Oshiumi et al., 2013; Oshiumi, 2020).
Therefore, we investigated these two ligases by performing co-immunoprecipitation studies using anti-TRIM25 and anti-Riplet antibodies. The assay revealed the presence of E5 oncoprotein in anti-TRIM25 immunoprecipitates only, suggesting that the E5 oncoprotein of bovine δPVs interacts with TRIM25 but not with Riplet (Fig. 2). Our results are in line with in vitro studies performed on cells experimentally infected with HPV18, which showed that TRIM25 but not Riplet was a target of viral E6 oncoprotein (Chiang et al., 2018). We then investigated the expression levels of these two ligases. Western blot analysis of total extracts detected unmodified levels of Riplet expression (Fig. 3) and a statistically significant reduction in the expression of TRIM25 (Fig. 4). To understand whether the marked reduction in TRIM25 expression levels could be attributed to transcriptional events and/or increased protein degradation, we investigated the presence of TRIM25 transcripts by RT-PCR. Sequencing of the obtained cDNA amplicons showed 100% identity with bovine TRIM25 sequences deposited in GenBank (Accession number: NM_001100336.1) (Supplemental Fig. S2). Then, we performed a real-time PCR analysis on cDNA using specific primers for bovine TRIM25. This molecular assay did not show any variation in transcript expression in cells infected with bovine δPVs compared with cells from clinically normal cattle (Fig. 5). These results suggest that bovine δPVs interfere at the protein level rather than at the transcriptional level in reducing TRIM25 expression.
3.3. Expression levels of RIG-I and MDA5 and their downstream effectors
Expression of RLRs is ubiquitous and is typically maintained at low levels in resting cells, but is greatly increased after virus infection (Loo and Gale, 2011). Therefore, we decided to investigate RLR expression during spontaneous BPV infection.
We detected reduced expression levels of both RIG-I and MDA5 by Western blot analysis in urothelial cells infected by bovine δPVs compared with urothelial cells from clinically normal cattle (Fig. 6). We assumed that the levels of these proteins could be due to transcriptional reduction. Using specific primers for bovine RIG-I and MDA5, we carried out a real-time PCR. Sequencing of the transcript amplicons revealed cDNA fragments showing 100% identity with bovine RIG-I and MDA5 sequences deposited in GenBank (Accession numbers: XM_002689480.6 and XM_010802053.2, respectively) (Supplemental Fig. S3). Real-time PCR of cDNA revealed a statistically significant reduction in both RIG-I and MDA5 transcripts in δPV-positive cells compared with δPV-negative cells (Fig. 7). These results suggest that, like HPVs, bovine δPVs may interfere at the transcriptional level rather than at the protein level in reducing RIG-I and MDA5 expression to prevent their antiviral activities.
RIG-I and MDA5 interact with a mitochondrial adaptor, the mitochondrial antiviral signalling (MAVS) protein (Yoneyama et al., 2015; Oshiumi, 2020). It remains unclear how MAVS acts as a scaffold to assemble the signalosome in RLR-mediated antiviral signalling (Chen et al., 2018). Western blot analysis of MAVS expression revealed unmodified protein expression levels in both δPV-infected and healthy cells (data not shown). Our results are in line with experimental data showing that the expression levels of MAVS did not significantly vary in cells in which the E6 oncoprotein of HPV18 was shown to act as a RIG-I transcriptional repressor (Albertini et al., 2018). We then performed co-immunoprecipitation studies using an anti-MAVS antibody. Western blot analysis performed on the immunoprecipitates detected the presence of RIG-I and MDA5 as well as TRIM25, phosphorylated TANK-binding protease 1 (pTBK1), phosphorylated interferon regulatory factor 3 (IRF3), and Sec13, which is believed to be a positive regulator of MAVS (Chen et al., 2018) (Fig. 8). Western blot analysis performed on total extracts revealed a statistically significant reduction in the expression levels of Sec13 in δPV-infected cells compared with cells from clinically normal cattle (Fig. 9), which suggests that MAVS activation might be compromised in cells spontaneously infected with bovine δPVs. MAVS subsequently phosphorylates and activates TBK1 and IRF3, via an unknown mechanism, which results in the production of interferons as well as proinflammatory factors (Fang et al., 2017). Western blot analysis performed on anti-MAVS immunoprecipitates revealed the presence of pTBK1 and pIRF3, which suggests that MAVS forms a complex with pTBK1 and pIRF3 and plays a critical role in driving and coordinating synergistic functional activities of these downstream components. Moreover, we investigated the expression levels of TBK1 and IRF3 in total extracts by immunoblotting, which revealed statistically significant reduced levels of both proteins in cells infected with bovine δPVs compared with healthy cells (Fig. 10). Furthermore, western blot analysis revealed statistically significant reduced expression levels of pTBK1 (Fig. 11). TBK1 is activated via phosphorylation (Liu et al., 2015), which in turn phosphorylates and activates IRF3. Subsequently, IRF3 enters the nucleus to activate type 1 IFN (Fitzgerald et al., 2003; Fang et al., 2017). Altogether, our results suggest that the transcriptional downregulation of RIG-I and MDA5 in cells infected with bovine δPVs is responsible for an aberrant downstream signalling pathway, including TBK1/IRF3, which may lead to the impairment of the host antiviral response. Because of downregulated RLRs, an adequate innate immune response is not elicited against spontaneous bovine δPV infection, thus leading to persistent infection in the cells.
4. DISCUSSION
This study provides novel mechanistic insights into the role of E5 oncoprotein in dysregulating the host antiviral innate immune response in a spontaneous model of bovine papillomavirus disease. Our study showed, for the first time, that the E5 oncoprotein of bovine δPVs interacts with TRIM25, a key player in antiviral immunity (Koliopoulos et al., 2018), to hamper innate immune signalling pathway mediated by RIG-I and MDA5. These results are of interest as there are very limited, controversial in vivo studies based on the role of TRIM25 in RLR activation, which remains elusive (Hayman et al., 2019; Wang and Hur, 2020).
E5 oncoprotein did not appear to influence the transcriptional activity of TRIM25; therefore, it is conceivable that E5 oncoprotein enhanced TRIM25 proteasomal degradation, which may hinder the activation of RIG-I and MDA5. It is well known that TRIM25 ubiquitinates and activates RLRs in a dose-dependent manner (Gack et al., 2007). Our results appear to be corroborated by experimental studies that showed that HPV oncoproteins could enhance the proteasomal degradation of TRIM25 (Chiang et al., 2018). Furthermore, our study suggested the existence of multiple evasion mechanisms based on bovine δPV-mediated inhibition of key components of the RLR pathways. Indeed, E5-expressing cells showed a marked reduction in the transcriptional activity of both RIG-I and MDA5. Reduced RIG-I and MDA5 mRNA levels detected by real-time PCR suggested that some proteins of bovine δPVs could downregulate the transcriptional activity of RIG-I and MDA5, which allowed δPVs to impair the innate antiviral response, a prerequisite for persistent infection. Our results appeared to be strengthened by experimental data from in vitrostudies in which HPV oncoproteins have been shown to act as transcriptional repressors of RIG-I and MDA5 to impair the viral host response during persistent infection (Reiser et al., 2011; Albertini et al., 2018). RLRs catalyse the conversion of MAVS fibrils to prion-like aggregates. Although MAVS activation is a complex, multistep process, this conformational change of MAVS is essential for the recruitment of downstream signalling molecules (Hou et al., 2011). Not much is known about the mechanism(s) of how MAVS functions in antiviral signalling pathways (Chen et al., 2018); therefore, the activation mechanism of MAVS downstream pathways remains elusive (Zhu et al., 2019). In our study, MAVS expression levels did not vary significantly. Many viruses block RLR-mediated immune signalling thus inhibiting host antiviral response without modifying MAVS expression levels (Zhang et al., 2020). It is conceivable that in our spontaneous model of PV infection, the marked reduction in the expression levels of RIG-I and MDA5 may be responsible for the loss of conformational changes thus compromising the activation of MAVS, which is necessary for activating and propagating the antiviral signalling cascade. In addition, we found reduced expression levels of Sec13, which may contribute to further attenuation of MAVS downstream signalling. It has been suggested that Sec13 facilitates MAVS aggregation and ubiquitination and is thus required for RLR-MAVS-related antiviral responses (Chen et al., 2018). It has been shown that Sec13 expression correlates with MAVS activation. Indeed, the overexpression of Sec13 increases MAVS activation, whereas Sec13 downregulation attenuates MAVS activation (Chen et al., 2018). In vitro studies have shown that MAVS may serve as a scaffold to facilitate the interaction of TBK1 with IRF3 (Liu et al., 2015). MAVS has been shown to activate the transcription factor IRF3 through TBK1 (Fang et al., 2017). We found a marked reduction in the expression levels of total and phosphorylated TBK1, which may result in perturbation of IRF3 activation as TBK1 plays a crucial role in allowing efficient IRF3 phosphorylation in the IFN-producing pathways that require MAVS as the adaptor protein (Fang et al., 2017). Many viruses inhibit RIG-I/MAVS signalling by blocking TBK1 phosphorylation (Darlympe et al., 2015). It is conceivable that the E5 oncoprotein of bovine δPVs is a key player involved in the downregulation of TBK1 activation. Low expression of TBK1 has been shown to markedly reduce IFN1 induction (Seth et al., 2005) and proinflammatory macrophage (M1) polarisation (Stone et al., 2019). Furthermore, we found reduced expression levels of IRF3, which may hamper their interaction network, a critical step in the production of IFNs (Ding et al., 2014; Liu et al., 2015).
Bovine δPVs must escape innate immune surveillance to establish persistent infections and viral proteins may manipulate this process through several mechanisms. This study showed that similar to human PVs, bovine PVs perturb the RLR-mediated innate immune signalling pathway through the viral E oncoprotein, which is encoded in the early stages of PV infection. This perturbation results in an abnormal host antiviral response, which allows PVs to continue their infectious cycle leading to persistent viral infection. Bovine δPVs reduce the levels of the DNA sensors that can recognise BPVs, which can hamper pTBK1 signalling as well as the production of IFNs, similar to human PVs (Hong and Laimins, 2017). IFN production plays a crucial role in the immune response against PV infection as IFNs promote the clearance of latent PV episomes in persistently infected cells (Westrich et al., 2017) and/or rapid reduction in PV episome copies per cell (Herdman et al., 2006). It has been suggested that basal cells in the initial infection usually contain low levels (around 100 copies per cell) of human and bovine PV episomes (Turek et al., 2002; Groves and Colemans, 2015). Animal cells that fail to resolve their infection and retain oncogene expression for years can facilitate tumourigenesis by BPVs (Doorbar, 2006).
In conclusion, bovine δPVs must escape innate immune surveillance to establish persistent infections, and viral proteins manipulate this process through several mechanisms. Despite the importance, molecular mechanisms for many bovine δPV oncoproteins remain poorly characterised, in part due to challenges in identifying their substrates. Therefore, further investigations aimed to clarify the functional role of viral oncoproteins at the intersection of immune evasion and aberrant proliferation of cells persistently infected by bovine δPVs, warrant future research.