Influence of different glycoproteins and of the virion core on SERINC5 antiviral activity

Host plasma membrane protein SERINC5 is incorporated into budding retrovirus particles where it blocks subsequent entry into susceptible target cells. Three accessory proteins encoded by diverse retroviruses, HIV-1 Nef, EIAV S2, and MLV Glycogag, each independently disrupt SERINC5 antiviral activity, by redirecting SERINC5 from the site of virion assembly on the plasma membrane to an internal RAB7+ endosomal compartment. Pseudotyping retroviruses with particular glycoproteins, e.g., the vesicular stomatitis glycoprotein (VSV G), renders the infectivity of particles resistant to inhibition by virion-associated SERINC5. To better understand viral determinants for SERINC5-sensitivity, the effect of SERINC5 was assessed using HIV-1, MLV, and M-PMV virion cores, pseudotyped with glycoproteins from Arenavirus, Coronavirus, Filovirus, Rhabdovirus, Paramyxovirus, and Orthomyxovirus genera. Infectivity of particles, pseudotyped with HIV-1, amphotropic-MLV, or influenza virus glycoproteins, was decreased by SERINC5, whether the core was provided by HIV-1, MLV, or M-PMV. Particles generated by all three cores, and pseudotyped with glycoproteins from either avian leukosis virus-A, human endogenous retrovirus K (HERV-K), ecotropic-MLV, HTLV-1, Measles morbillivirus, lymphocytic choriomeningitis mammarenavirus (LCMV), Marburg virus, Ebola virus, severe acute respiratory syndrome-related coronavirus (SARS-CoV), or VSV, were insensitive to SERINC5. In contrast, particles pseudotyped with M-PMV, RD114, or rabies virus (RABV) glycoproteins were sensitive to SERINC5, but only with particular retroviral cores. Resistance to SERINC5 by particular glycoproteins did not correlate with reduced SERINC5 incorporation into particles or with the route of viral entry. These findings indicate that some non-retroviruses may be sensitive to SERINC5 and that, in addition to the viral glycoprotein, the retroviral core influences sensitivity to SERINC5. IMPORTANCE The importance of SERINC5 for inhibition of retroviruses is underscored by convergent evolution among three non-monophyletic retroviruses, each of which encodes a structurally unrelated SERINC5 inhibitor. One of these retroviruses causes tumors in mice, a second anemia in horses, and a third causes AIDS. SERINC5 is incorporated into retrovirus particles where it blocks entry into target cells, via a mechanism that is dependent on the viral glycoprotein. Here we demonstrate that retroviruses pseudotyped with glycoproteins from several non-retroviruses are also inhibited by SERINC5, suggesting that enveloped viruses other than retroviruses may also be inhibited by SERINC5. Additionally, we found that sensitivity to SERINC5 is determined by the retrovirus core, as well as by the glycoprotein. By better understanding how SERINC5 inhibits viruses we hope to extend fundamental understanding of virus replication and of the native role of SERINC5 in cells, and perhaps to advance the development of new antiviral strategies.


IMPORTANCE
The importance of SERINC5 for inhibition of retroviruses is underscored by convergent evolution among three non-monophyletic retroviruses, each of which encodes a structurally unrelated SERINC5 inhibitor. One of these retroviruses causes tumors in mice, a second anemia in horses, and a third causes AIDS. SERINC5 is incorporated into retrovirus particles where it blocks entry into target cells, via a mechanism that is dependent on the viral glycoprotein. Here we demonstrate that retroviruses pseudotyped with glycoproteins from several non-retroviruses are also inhibited by SERINC5, suggesting that enveloped viruses other than retroviruses may also be inhibited by SERINC5. Additionally, we found that sensitivity to SERINC5 is determined by the retrovirus core, as well as by the glycoprotein. By better understanding how SERINC5 inhibits viruses we hope to extend fundamental understanding of virus replication and of the native role of SERINC5 in cells, and perhaps to advance the development of new antiviral strategies.
HIV-1 is not the only virus inhibited by SERINC5. SIVs lacking nef are also inhibited by SERINC5 and SIV nefs counteract this inhibition (17) with a potency that is proportional to the prevalence of SIV in wild primate populations (21). Two examples of convergent evolution of anti-SERINC function by virally encoded proteins are found outside of primate immunodeficiency viruses. Murine leukemia virus (MLV) Glycogag and equine infectious anemia virus (EIAV) S2 are viral antagonists of SERINC5 activity, and neither share sequence or structural homology with Nef, nor to each other (17,(22)(23)(24).
The mechanism by which virion-associated SERINC5 inhibits HIV-1 entry is unknown.
The block is manifest after virion attachment to target cells, apparently at the stage of fusion pore expansion; virion contents mix with target cell cytoplasm but virion core transfer to the cytoplasm is inhibited (17,19). Otherwise isogenic virions pseudotyped with HIV-1 Env glycoproteins from different HIV-1 isolates exhibit a range of dependency on Nef and of sensitivity to SERINC5 (25,26). SERINC5 increases HIV-1 sensitivity to antibodies and peptides targeting the membrane-proximal external region of gp41, suggesting that it somehow alters the conformation of the HIV-1 glycoprotein near the virion membrane (19,25). Importantly, HIV-1 particles pseudotyped with vesicular stomatitis virus (VSV) G or Ebola virus glycoprotein are resistant to SERINC5 antiviral activity (17,18,24). These initial observations suggest a correlation between the location of viral fusion and sensitivity to SERINC5 activity, with glycoproteins that mediate fusion at the cell surface (Env from HIV-1 and amphotropic MLV [A-MLV]) being sensitive and those that mediate fusion in endo-lysosomal compartments (VSV-G and Ebola GP) being resistant (17,24). Taken together these results indicate that the virion glycoprotein is a viral determinant of sensitivity to SERINC5.
SERINC5 is a multipass transmembrane that localizes almost exclusively to the plasma membrane (17,18). As such, in the absence of counter-measures, all enveloped viruses would be expected to encounter SERINC5 during viral egress, and to potentially be subject to its antiviral effects. We sought to address the breadth of SERINC5 antiviral activity and assess whether the route of entry impacts the sensitivity of viral glycoproteins to the antiviral effects of SERINC5. To do so, we investigated whether the co-expression of SERINC5 during viral production could inhibit a variety of glycoprotein pseudotypes of HIV, MLV, or M-PMV cores.
Using this system, we tested the sensitivity of a number of retroviral Envs as well as representative glycoproteins from the Arenavirus, Coronavirus, Filovirus, Rhabdovirus, Paramyxovirus, and Orthomyxovirus genera. Consistent with previous findings, we observed that glycoprotein is a major determinant of SERINC5 sensitivity. While many glycoproteins were universally insensitive to the antiviral effects of SERINC5, the glycoproteins from NL4.3, A-MLV, and influenza were inhibited by SERINC5 in all viral core pseudotypes tested. No correlation was observed between SERINC5 sensitivity and the route of viral entry mediated by the viral glycoprotein. Unexpectedly, we also observed that sensitivity to SERINC5 antiviral activity for M-PMV, RD114, and rabies virus (RABV) glycoproteins depended on the retroviral core onto which they were pseudotyped. Our findings reveal that an interplay between virion core and glycoprotein determines the sensitivity to SERINC5 antiviral activity.

RESULTS
To determine which viral glycoproteins are sensitive to the antiviral activity of SERINC5 we assessed infectivity of pseudotyped GFP-expressing lentiviral vectors produced in the presence or absence of SERINC5. Included in this panel was a diverse selection of retroviral Env glycoproteins, including those from human immunodeficiency virus-1 (HIV-1), avian leukosis Similar to the findings of others (17,18,24), we observed that SERINC5 causes a greater than 100-fold reduction in viral infectivity for HIV-1 and A-MLV pseudotypes, while no significant reduction was observed for EBOV and VSV pseudotypes ( Fig. 1A and Table 1).
These observations indicate that restriction by SERINC5 is not dictated by how the viral glycoprotein mediates fusion, as fusion mediated by influenza (27) or by RABV (28)  Next we tested a panel of filoviral glycoproteins for sensitivity to SERINC5 restriction.
All of these glycoproteins require proteolytic processing (32,33) following internalization into the target cell and utilize the lysosomal protein NPC1 to initiate viral fusion (34,35). In addition to the EBOV and MARV glycoproteins tested in Fig  with an infectivity-enhancing derivative (GP-A82V) that arose during the outbreak (36,37). As shown in Fig. 1B, none of the filoviral glycoproteins were inhibited >10-fold in the presence of SERINC5. However, there may be modest differences in sensitivity to SERINC5 activity, specifically RESTV and TAFV GP appear slightly more sensitive (4.3-and 2.9-fold, respectively) to SERINC5 inhibition compared to either Mayinga or Makona Ebola virus glycoproteins (1.65-and 1.2-fold, respectively).
HIV-1 Nef, MLV glygoGag, and EIAV S2 counteract SERINC5 antiviral activity by removing SERINC5 protein from the cell surface and relocalizing it to an endosomal compartment (17,18,22). The ability of a viral glycoprotein to re-localize a normally plasma membrane localized antiviral protein has been previously shown for HIV-2 Env and human BST2 (38). Thus, we reasoned that viral glycoproteins may confer resistance to SERINC5 activity by re-localizing SERINC5 to an internal membrane compartment. To test this, we compared SERINC5 incorporation into HIV-1 virus-like particles (VLPs) pseudotyped with the various glycoproteins shown in Fig 1A. We found that HIV-1 VLPs universally incorporated SERINC5 irrespective of the viral glycoprotein present ( Figure 2). In replicate blotting, only HERV-K Env showed a consistently lower level of SERINC5 incorporation into viral particles (data not shown). However, this observation is likely to be caused by pleiotropic effects of cells transfected with this glycoprotein, as cell growth was significantly reduced compared to other transfections, and reduced levels of Gag and GFP were also observed ( Fig. 2, data not shown).
Regardless, no direct correlation between SERINC5 exclusion from virions and resistance to its antiviral effects was evident.
A previous report indicated that MLV virions pseudotyped with RD114 Env are susceptible to the antiviral effects of SERINC5 (24). However, in the presence of SERINC5 we only observed a modest ~4.5-fold reduction in viral titer of RD114 pseudotyped HIV-1 virions (Fig. 1A). In response to this discrepancy, we sought to determine if the viral core modulates susceptibility to SERINC5 antiviral activity. Thus, we tested the same panel of glycoproteins for SERINC5 sensitivity when pseudotyped on different virion cores. First, we tested the SERINC5 sensitivity of the same panel of glycoproteins as in Fig. 1A on MLV viral cores ( Figure 3A and Table 1). We observed that the glycoproteins sensitive to SERINC5 restriction on HIV-1 cores (HIV-1, A-MLV, Flu, and Rabies) were also restricted when pseudotyped on MLV cores.
Additionally, we observed that M-PMV Env was sensitive to SERINC restriction when pseudotyped onto MLV cores, whereas it was not when pseudotyped onto HIV-1 cores.
Returning to the initial impetus for exploring different cores, we observed a ~7.5-fold reduction of infectivity for RD114 pseudotyped MLV cores when produced in the presence of SERINC5, which is similar to the magnitude of the inhibitory effect reported by Ahi et al. (24).
Due to observed differences in SERINC5 sensitivity with M-PMV pseudotypes of HIV-1 and MLV cores, we next tested for SERINC5 antiviral activity against our panel of glycoproteins pseudotyped onto M-PMV cores ( Figure 3B and Table 1

DISCUSSION
Initial reports indicated that the viral glycoprotein is a determinant of sensitivity to SERINC5 antiviral activity (17,18,25,26) and suggested that viral glycoproteins which mediate fusion via a pH-dependent, endocytic entry pathway are resistant to SERINC5 antiviral activity (17,24). Here, to examine these issues further, pseudotypes using glycoproteins from diverse families of enveloped viruses were assessed for sensitivity to restriction by SERINC5. We observed that SERINC5 restricted virions pseudotyped with glycoproteins from several retroviruses (HIV-1, A-MLV, RD114, and M-PMV), influenza A (Orthomyxoviridae), and rabies (Rhabdoviridae). To our knowledge, this is the first time antiviral activity of SERINC5 has been described for a non-retroviral glycoprotein. As the glycoproteins of these viruses were studied as retroviral pseudotypes, it remains to be established if the infectivity of authentic influenza or rabies viruses are affected by SERINC5, or other SERINC family members.
Additionally, our observation with influenza A and rabies glycoproteins demonstrates that mediating entry via an endocytic route does not, in itself, protect from the antiviral effects of SERINC5.
While Env glycoproteins from the retroviruses HIV-1, MLV, and RD114 have all previously been found to be inhibited by SERINC5 (17,18), we now report that M-PMV glycoprotein is SERINC5-sensitive as well. Interestingly, we saw a ~100-fold reduction in infectivity of autologously pseudotyped M-PMV cores when produced in the presence of SERINC5. This observation was unexpected given that lenti-and gammaretroviruses encode accessory factors that counteract SERINC5 activity. And yet, functionally intact M-PMV (the only viral gene known to be missing from the GFP-expressing M-PMV vector is Env, which is complemented in trans during the transfection) was sensitive to the antiviral effects of human SERINC5.
Surprisingly, we observed that particular glycoproteins displayed different sensitivity to the antiviral effects of SERINC5 depending on the viral core onto which they were pseudotyped.
For instance, rabies virus glycoprotein was inhibited by SERINC5 when on HIV-1 or MLV cores, but insensitive to SERINC5 when on M-PMV cores. In contrast, M-PMV glycoprotein was sensitive to SERINC5 restriction when on MLV or M-PMV cores, but resistant when on HIV-1 cores. Additionally, a ~17-fold SERINC5-mediated inhibition was observed for M-PMV cores pseudotyped with RD114 Env, while 7.5-fold and 4.3-fold inhibitions were observed for RD114 pseudotypes of MLV and HIV cores, respectively. A previous report demonstrated similar magnitude inhibition for RD114-pseudotyped MLV by endogenous SERINC activity (24). Regardless, RD114 showed little sensitivity to SERINC5 when pseudotyped onto HIV-1 cores. Neutralization by monoclonal antibodies that target the membrane-proximal domain of HIV-1 glycoprotein is altered by the presence of SERINC5 (19,25). Given that MA, the membrane proximal domain of gag makes contacts with the retroviral TM (40,41), one can imagine that SERINC5 has the potential to influence interactions between MA and TM in the HIV-1 virion that are essential for infectivity. In similar fashion, SERINC5 might influence retroviral core interactions by the heterologous glycoproteins tested here, for which SERINC5 restriction activity was core-dependent, i.e., the rabies virus, MPMV, and RD114 glycoproteins.

MATERIALS AND METHODS
Plasmid DNA. Plasmids used in this study are described in Table 1, including Addgene or NIH AIDS Reagent Program code numbers (where applicable), where full plasmid sequences can be obtained. A pcDNA3.1 based vector bearing codon-optimized pNL4-3 env with a cytoplasmic tail truncation after residue 710 (HXB2 residue 712), similar to that previously described (42), was generated using standard cloning techniques and is available from Addgene.   Volumes of lysate corresponding to equal protein content were combined 1:1 with 2x Laemmli buffer containing 50 mM TCEP and incubated at room temp for 5 minutes.
One half of the denatured viral pellet and approximately 8 μ g protein from cellular lysates were run on 4-15% gradient acrylamide gels, and transferred to nitrocellulose membranes.