The question remained as to why
The question remained as to why hCrm1 was functional and mCrm1 was not in terms of Rev and HIV mRNA nuclear export, and previously we proposed three possibilities (Elinav et al., 2012). One was that human voltage gated sodium channel had a positively acting factor that somehow stabilized the hCrm1 and Rev-RRE complex (but did not do so for mCrm1); a second was that murine cells had a negatively acting factor that disrupted the mCrm1 and Rev-RRE complex (but did not do so for hCrm1). Lastly, we hypothesized that hCrm1 simply interacted more favorably or strongly with Rev-RRE complex, compared to mCrm1. We decided to test the third model by evaluating hCrm1 and mCrm1 interaction with lentiviral Rev proteins, first biochemically and then genetically, the latter using a mammalian two-hybrid system in both human and murine cells.
Results To test whether there is a differential interaction between hCrm1 and mCrm1 and Rev-RRE complex, we first turned to a biochemical method initially developed by Cullen and colleagues (Bogerd et al., 1998), using purified, bacterially expressed HIV Rev-GST and Ran-GTP proteins, along with in vitro transcribed HIV RRE RNA and in vitro translated hCrm1 or mCrm1 proteins. Both Rev-GST and RevM10-GST fusions were purified using glutathione beads as ~45kD proteins (Fig. 1). When incubated with Ran-GTP and either in vitro translated, HA epitope-tagged hCrm1or mCrm1, only HA-hCrm1 was precipitated, but only in the presence of HIV Rev-GST fusion and RRE RNA (Fig. 1). This is consistent with prior work showing ~2× stronger binding of purified hCrm1 compared to mCrm1 to HIV-1 Rev (Booth et al., 2014). As expected, RevM10-GST fusion failed to precipitate or bind either HA-Crm1 protein, in the presence or absence of RRE. To confirm and extend those findings of a differential interaction between hCrm1 and mCrm1 and Rev-RRE, we first attempted to use the yeast two and three hybrid systems, but failed to observe a detectable interaction between any of the Crm1s and Rev, in the presence or absence of HIV RRE RNA. We then turned to the luciferase complementation system in which protein interaction between amino and carboxy terminal firefly luciferase (FFLUC) fusion proteins results in detectable FFLUC activity by relative light units (RLUs) (Luker et al., 2004). Using this system we were able to detect a strong HIV Rev-Rev genetic interaction but not between Rev and any Crm1 fusion protein. Because the luciferase complementation system may only allow readout when each fusion partner is small enough to permit enzymatic catalysis and full-length Crm1 is greater than 100 kD in size, we next attempted the mammalian two-hybrid system, knowing that we could also separately transfect in HIV RRE RNA in plasmid form, driven by either RNAPII or III promoters. Based on these results, we tested the functionality of the Gal4DBD-Crm1 fusions in a functional assay, transfecting them into murine B78 cells that had an integrated HIV reporter vector encoding both a truncated form of human cyclin T1 and blasticidin resistance (bsd) (Coskun et al., 2007), along with VSV G expression construct and an HIV vector encoding a truncated form of human cyclin T1 and eYFP. The B78 cells express murine Crm1, and this assay allows us to test the functionality of other Crm1 constructs (Fig. 2A). Human cyclin T1 is required since expression of both eYFP and bsd in the vector used are dependent upon the HIV long terminal repeat and the presence of Tat. As anticipated, the non-fused, full-length hCrm1 had the greatest activity in terms of infectious virus release, as measured by the number of blasticidin-resistant colonies on HOS cell targets (Fig. 2B). The Gal4DBD-hCrm1 fusion had ~25% of the activity of the non-fused hCrm1 but was significantly more active than the Gal4DBD-mCrm1 fusion and the Gal4DBD-hCrm1 411–412-414 mutant. As reported previously (Aligeti et al., 2014), the 2xNES-Rev-mCherry expression plasmid gave roughly 50% of the number of colonies, in the absence of hCrm1 (Fig. 2B). As anticipated, similar results were obtained using eYFP as a flow cytometric readout on the HOS targets (not shown).