Background
Protease-mediated processing of the Gag-Pro-Pol polyprotein is an essential step in the replication of retroviruses [
1]. The protease is activated concurrent with egress, or shortly thereafter, at which time it is released from the polyprotein via an autocatalytic reaction followed by proteolytic processing of the remainder of the polyprotein [
2]. The mechanism or trigger for activation of retroviral proteases is unclear. Strict regulation of the protease is required to prevent premature activation, which would inhibit virus assembly, budding and infectivity [
3‐
6]. This is especially important for betaretroviruses as the core assembles within an intracellular compartment before transport to the plasma membrane and release [
2]. Several studies have demonstrated that protein conformation/dimerization [
4,
7‐
9] and oxidation [
10,
11] play an integral role in protease activation, but activation of ovine betaretrovirus proteolytic processing has not been studied.
Inactive protease is detrimental to retroviral replication because in the absence of protease processing the virion will not convert to the mature metastable conformation required for the virus to become infectious. Indeed, there is a class of antiretroviral drugs designed to specifically inhibit the protease and consequently inhibit virus replication [
12,
13].
Enzootic nasal tumor virus (ENTV)-1 is an ovine betaretrovirus that is associated with enzootic nasal adenocarcinoma (ENA), a nasal tumor of sheep [
14]. Experiments showing transmission of ENA to a healthy lamb by inoculation with cell-free ENA tumor homogenate containing ENTV-1 antigens suggested that ENTV-1 is the causative agent of ENA, but did not completely fulfil Koch’s postulates [
14]. A factor limiting these experiments is an inability to produce high titer virus from cell culture since the virus cannot be propagated in vitro. In the study presented here, we sought to resolve this issue by constructing a molecular clone from which infectious ENTV-1 could be generated. Transfection of HEK 293T cells with the ENTV-1 molecular clone led to the production of virus particles, but processing of the Gag polyprotein was not observed. The protease could not be activated by treatment with a reducing agent but could be complemented with the JSRV Gag-Pro-Pol polyprotein producing virus particles with fully processed Gag. Mutagenesis of non-conserved amino acids within the protease domain failed to restore Gag polyprotein processing; however, removal of an additional proline residue from a polyproline tract in the matrix protein resulted in the production of fully mature virus particles. This is the first report demonstrating that ENTV-1 can be produced from a molecular clone providing the foundation for further studies investigating the pathogenesis and determinants of tissue tropism of this poorly understood virus.
Discussion
ENA is a relatively common disease of sheep, particularly in North America [
19,
28]. Despite this, little is known about the pathogenesis of the disease, including confirmation of ENTV-1 as the etiologic agent. Much of what is known about ENA and ENTV-1 is inferred from studies on OPA and JSRV. We endeavoured to address this problem and build upon our previous transmission experiments involving cell-free tumor homogenate by constructing a molecular clone of ENTV-1 from which virus could be generated for in vivo infection experiments. Failure of the original molecular clone to produce mature virions comprised of fully processed viral proteins presented a unique impediment to these experiments, but ultimately revealed important biological insights about ENTV-1.
Initially we thought that the defect in Gag polyprotein processing could be due to the requirement for a host cell protein found only in sheep cells or alternatively, due to the presence a restriction factor in human cells that inhibited maturation of ENTV-1 virions. To address this possibility, we attempted to produce recombinant ENTV-1 virions in both primary and continuous cell lines from sheep and goat; however, these cells were extremely resistant to transfection and did not yield sufficient amounts of virus for analysis, even after ultracentrifugation of virus-containing supernatant (data not shown). Indeed, it was difficult to observe ENTV-1 virions produced in HEK 293T cells in situations where the transfection efficiency was suboptimal; suggesting that virus production by this method is extremely inefficient. Since JSRV particles produced from HEK 293T cells are capable of polyprotein processing and the sequence identity between JSRV and ENTV-1 is relatively high, we concluded that it was unlikely that the defect in Gag polyprotein processing observed in virus particles produced from the ENTV-1 molecular clone was due to the absence of a critical host cell factor in the producer cells.
Co-transfection of the JSRV Gag-Pro-Pol expression vector with the ENTV-1 molecular clone generated virions with fully processed Gag (Fig.
2a). The antibody used in this experiment does not distinguish between JSRV and ENTV-1 capsid proteins so it was not possible to determine whether rescue of polyprotein processing was due to processing of ENTV-1 Gag by the JSRV protease or displacement of the ENTV-1 polyprotein by the JSRV polyprotein. The JSRV Gag-Pro-Pol expression vector contains additional elements (e.g., MPMV CTE and SV40 polyA) [
17] that promote much higher expression of Gag-Pro-Pol than the molecular clone so direct comparisons are not possible. Expression level differences were taken into account in the dosage response co-transfection experiments (Fig.
2b and c) in order to encourage heterologous Gag polyprotein co-packaging (ie. generation of virions with polyproteins from both JSRV and ENTV-1) [
29]. Incomplete rescue of processing, observed as partially processed Gag in lane 7 and 8 of Fig.
2b, indicates that the JSRV protease is processing the ENTV-1 polyprotein in trans in these virions but that there is a limit to which the JSRV protease can perform this function. Therefore, since the ENTV-1 polyproteins can be processed by the JSRV protease we concluded that the processing defect was due to a lack of ENTV-1 protease activity.
Interestingly, mutation of the pCMVENTV-1 protease to match that of ENTV-1NA4, the sequence from which it was derived, failed to rescue the processing defect. This was unexpected considering that the ENTV-1NA4 sequence was derived from nasal tumor tissue isolated from a sheep with ENA. Although only fully processed Gag was observed in virions purified from the ENTV-1NA4 tumor or any of the other ENTV-1NA tumors [
14], a processing defect could exist outside the context of transformed sheep cells. Since ENTV-1OVC was extracted from a tumor transmitted by cell-free tumor homogenate [
14], it was assumed that this sequence represented a replication-competent, infectious ENTV-1. It should be noted that homogenates from the ENTV-1OVC tumor contained unprocessed and partially processed Gag proteins which were thought to represent naked cores released from the cytoplasm of tumor cells due to mechanical disruption [
14]. It is possible that the processing defect observed in virions derived from the initial ENTV-1 molecular clone may be reflective of the phenotype of at least some circulating exogenous ENTV-1 particles. If that were the case, these virions would not be infectious unless complemented with an active protease, perhaps from endogenous betaretrovirus sequences [
30]. Expression of the endogenous ovine betaretroviruses is localized primarily to tissues of the reproductive tract [
31]; however, transcripts have been detected in other tissues such as the lung [
32]. Human endogenous retrovirus (HERV) sequences are upregulated in neoplastic tissues due to genomic instability and epigenetic changes during transformation [
33‐
35] and a similar phenomenon occurs in HIV-infected cells [
36]. Furthermore, HERV-K (a member of the
Betaretrovirus genus) Gag can co-assemble and co-package with HIV-1 Gag [
37]. Similarly, it has been shown that endogenous JSRV Gag can co-assemble with exogenous JSRV Gag and, in the presence of a particular mutation (R21W), restrict the release of exogenous JSRV particles in a transdominant fashion [
38,
39]. The high degree of similarity shared between ENTV-1, JSRV and endogenous ovine betaretroviruses [
19] suggests that since ENTV-1 can co-assemble with JSRV, it can likely also co-assemble with at least some of the endogenous ovine betaretrovirus Gag proteins. Therefore, it is conceivable that infection of sheep cells with ENTV-1 and subsequent transformation results in upregulation of endogenous betaretrovirus transcription, which in turn supplies polyproteins with active protease for co-packaging.
Since modification of the protease sequence of the defective molecular clone to match that of ENTV-1OVC failed to restore Gag polyprotein processing, we investigated sequence differences outside of the
pro reading frame that might be responsible for the defect. In HIV, regions upstream of the protease play a critical role in protease activation by influencing dimerization of the protease domain in the precursor [
40], which is requirement for activation [
41]. Of the ten ENTV-1 genomes we previously sequenced, four possessed a polyproline motif comprised of five consecutive prolines within the matrix region of Gag, which differed from the four consecutive prolines that comprised this motif in the other six ENTV-1 genomes, as well as in the genome of an ENTV-1 virus isolated in the UK. Despite being located in one of three variable regions in the ovine betaretrovirus genome, this polyproline motif is conserved within virus families and is consistently comprised of seven consecutive prolines in the case of JSRV and six consecutive prolines in the case of ENTV-2. Interestingly, this polyproline motif is absent from endogenous versions of JSRV [
42]. Since proline residues are known to introduce kinks in protein structures we theorized that the additional proline residue in the Gag-Pro-Pol polyprotein might prevent the protease from folding correctly, precluding dimerization and activation. Indeed, mutation of the molecular clone to restore the four amino acid long polyproline motif resulted in the recovery of mature virions containing fully processed Gag. Exactly why the addition of a proline residue to this motif had such a detrimental effect on polyprotein processing is not known; however, one might surmise that it could contribute to conformational changes in the polyprotein that alter protein-protein interactions, or as in the case of the Vpx protein of HIV-2, may reduce translation efficiency. Vpx proteins of the HIV-2/SIVsmm/SIVmac group encode a highly conserved polyproline motif (PPM) consisting of a hepta-proline stretch in the C-terminal region [
43,
44]. Recently, the PPM in HIV-2 Vpx was shown to be critical for its efficient expression in both eukaryotic and prokaryotic systems and this effect is determined by the context of PPM amino acid sequences, not the nucleotide sequences [
45]. Importantly, the number and position of consecutive prolines in this PPM were important for Vpx expression. PPMs are found in a large number of eukaryotic and prokaryotic proteins as well as in a wide range of human DNA and RNA viruses including herpesviruses, adenoviruses, and hepatitis viruses. However, whether these PPMs are responsible for efficient expression of the PPM-containing proteins or play some other as yet unidentified role in protein function is not yet known.