TIM-mediated inhibition of HIV-1 release is antagonized by Nef but potentiated by SERINC proteins

Significance

TIM proteins inhibit release of HIV-1 and other enveloped viruses. However, it is currently unknown whether and how the virus counteracts this restriction. In this work, we demonstrate that Nef proteins of HIV-1 and other lentiviruses function as antagonists to overcome the TIM-mediated restriction. TIM-1 is more potent at inhibiting release of Nef-deficient HIV-1 relative to wild-type (WT) HIV-1, and ectopic expression of Nef relieves this restriction. Interestingly, we find that SERINC proteins potentiate TIM-mediated inhibition of HIV-1 release likely by stabilizing TIM-1 expression. Our work reveals a role for lentiviral Nef in antagonizing TIMs, in part through SERINCs.

Abstract

The T cell Ig and mucin domain (TIM) proteins inhibit release of HIV-1 and other enveloped viruses by interacting with cell- and virion-associated phosphatidylserine (PS). Here, we show that the Nef proteins of HIV-1 and other lentiviruses antagonize TIM-mediated restriction. TIM-1 more potently inhibits the release of Nef-deficient relative to Nef-expressing HIV-1, and ectopic expression of Nef relieves restriction. HIV-1 Nef does not down-regulate the overall level of TIM-1 expression, but promotes its internalization from the plasma membrane and sequesters its expression in intracellular compartments. Notably, Nef mutants defective in modulating membrane protein endocytic trafficking are incapable of antagonizing TIM-mediated inhibition of HIV-1 release. Intriguingly, depletion of SERINC3 or SERINC5 proteins in human peripheral blood mononuclear cells (PBMCs) attenuates TIM-1 restriction of HIV-1 release, in particular that of Nef-deficient viruses. In contrast, coexpression of SERINC3 or SERINC5 increases the expression of TIM-1 on the plasma membrane and potentiates TIM-mediated inhibition of HIV-1 production. Pulse-chase metabolic labeling reveals that the half-life of TIM-1 is extended by SERINC5 from <2 to ∼6 hours, suggesting that SERINC5 stabilizes the expression of TIM-1. Consistent with a role for SERINC protein in potentiating TIM-1 restriction, we find that MLV glycoGag and EIAV S2 proteins, which, like Nef, antagonize SERINC-mediated diminishment of HIV-1 infectivity, also effectively counteract TIM-mediated inhibition of HIV-1 release. Collectively, our work reveals a role of Nef in antagonizing TIM-1 and highlights the complex interplay between Nef and HIV-1 restriction by TIMs and SERINCs.

Human T cell Ig and mucin domain (TIM) proteins, which include TIM-1, TIM-3, and TIM-4, bind to phosphatidylserine (PS) via a conserved IgV domain and regulate the host immune response (1, 2). In a manner dependent on their expression patterns in different cell types, TIM-family proteins play distinct roles in cell proliferation, apoptosis, immune tolerance, and T-cell activation (2). Furthermore, TIM-1 polymorphisms have been associated with some allergic human diseases (3–5). Although expression of TIM proteins in target cells has been shown to promote entry of a wide range of enveloped viruses (6–8), we recently found that all three human TIM-family proteins also inhibit the release of HIV-1 and other enveloped viruses, including murine leukemia virus (MLV) and Ebola virus (EBOV). This inhibition of particle release is achieved by TIM binding to PS present on the surface of viral producer cells and newly budded virions (9).

Upon viral infection, cells produce type I interferons (IFNs) that up-regulate expression of hundreds of IFN-stimulated genes (ISGs). Many of these ISGs execute antiviral activities (10) and are collectively referred to as cellular “restriction factors” (11). As a countermeasure, many viruses, including HIV, have evolved effective strategies to overcome cellular restrictions (12, 13). For example, HIV-1 Vif binds to apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3 (APOBEC3), thereby inducing its proteasomal degradation and enhancing viral infectivity (14). To enable efficient virus release, HIV-1 Vpu counteracts tetherin (also known as BST2 and CD317), either by preventing its trafficking to the plasma membrane or by targeting it to endolysosomal compartments for degradation (15–17). The strategies and modes of action by which HIV-1 accessory proteins antagonize cellular restrictions have provided critical molecular and genetic insights into our understanding of virus–host interaction and coevolution (13, 18–20).

Unlike typical cellular restriction factors (12), the expression of TIM proteins is not induced by type I IFN (9). However, TIM proteins, especially TIM-1 and TIM-3, are expressed in activated human CD4⁺ T cells and monocyte-derived macrophages (MDMs), respectively (9, 21, 22), and they constitutively inhibit HIV-1 production (9). Similarly, SERINC proteins, recently discovered cellular “restriction factors” that diminish the infectivity of HIV and other lentiviruses (23, 24), are not induced by type I IFN, nor are they under positive selection (25). Because the antiviral function of SERINCs is specifically antagonized by the lentivirus Nef protein (23, 24, 26, 27), an important question is whether, and how, HIV-1 may counteract TIM-mediated restriction to facilitate viral production and replication. Here, we provide evidence that lentiviral Nef proteins are capable of antagonizing the ability of TIM-1 to inhibit viral release, in part by promoting TIM-1 internalization from the plasma membrane and sequestering TIM-1 within intracellular compartments. Interestingly, we find that SERINC proteins potentiate TIM-mediated inhibition of HIV-1 release, in part by stabilizing TIM-1 protein expression. Moreover, two other SERINC antagonists, that is, MLV glycoGag and equine infectious anemia virus (EIAV) S2 proteins (23, 24, 28–30), can also effectively relieve TIM-mediated inhibition of HIV-1 release, further supporting the involvement of SERINC proteins in TIM-mediated restriction. Our work unveils a mechanism by which lentiviral Nef proteins counteract restriction by TIM in part through SERINC to facilitate HIV-1 release and replication.

Results

HIV-1 Nef Antagonizes TIM-1–Mediated Restriction of Viral Release.

Cellular inhibitory factors are frequently antagonized by lentiviral accessory proteins (e.g., Vif, Vpu, Vpr, Vpx, and Nef) or the viral envelope (Env) glycoprotein (12). To examine a possible role of HIV-1 accessory proteins in overcoming TIM-mediated inhibition of viral release, we cotransfected HEK293T cells with HIV-1 proviral DNA constructs containing intact or defective nef, vpu, vif, vpr, or env genes, together with increasing amounts of a TIM-1 expression plasmid. TIM-1 inhibited release of these HIV-1 variants, similarly to wild-type (WT) and in a dose-dependent manner, based on the quantification of Western blotting bands for the virion p24 versus the total HIV-1 Gag, which includes the virion p24 plus all cellular Gag fractions, that is, p24, p41, and Pr55 (Fig. 1 A and B). Notably, TIM-1 more potently inhibited the release of Δnef virus relative to WT virus or the other variants (Fig. 1 A and B), indicating that Nef likely counteracts TIM-1 inhibition of HIV-1 release. Expression of TIM-1 did not change HIV-1 Gag (Pr55) levels in viral producer cells, although, as we have reported previously, cell-associated p24 levels were increased in a TIM-1 dose-dependent manner (Fig. 1A), in part because of the accumulation of newly budded HIV-1 virions on the plasma membrane (9). Similar results were obtained using a WT HIV-1 NL4-3-IRES-eGFP construct and mutants thereof lacking vpu and/or nef genes (SI Appendix, Fig. S1 A and B).

Fig. 1.

HIV-1 Nef counteracts the inhibitory effect of TIM-1 on HIV-1 production. (A) Effect of TIM-1 on viral production of different proviral HIV-1 NL4-3 clones. HEK293T cells were transfected with HIV-1 proviral DNA plasmids encoding wild-type (WT) NL4-3 or nef, vpu, vif, vpr, or env-deficient NL4-3, along with indicated amounts of TIM-1 expression plasmid. Forty-eight hours posttransfection, Western blotting was performed to measure cell-associated Gag (Cells) and cell-free viral particles (VPs) by using an anti–HIV-1 p24 antibody. TIM-1 expression in cell lysates was determined by using an anti–TIM-1 antibody. The values for NL4-3 Δnef are boxed to highlight the differences. Virion-associated p24 and cellular Gag (Pr55, p41, and p24) were quantified by using Quantity One (Bio-Rad) software. Viral release efficiency was determined as the ratio of virion-associated p24 vs. the total Gag (i.e., virion p24 plus cellular Pr55, p41, and p24). Relative virion release (indicated as p24/total Gag) was calculated by setting the value of WT in the absence of TIM-1 to 1.00. (B) Relative release efficiency of HIV-1 and mutants was determined by averaging results of three independent experiments, with means and SDs shown. *P < 0.05; **P < 0.01; ****P < 0.0001. (C) Effect of ectopic expression of an exogenous Nef on the production of HIV-1 WT and Δnef particles. HEK293T cells were cotransfected with HIV-1 NL4-3 WT or Δnef proviral DNA along with increasing amounts of NL4-3 Nef expression plasmid (0.5 or 1.5 μg) in the presence or absence of TIM-1. Viral release was determined as described for A. Note that an anti-Nef antibody was used to detect both the endogenous Nef (En) encoded by the provirus and an exogenous Nef (Ex) expressed in trans from an expression plasmid. (D) Relative release efficiency of HIV-1 WT and Δnef in the presence and absence of TIM-1 and exogenous Nef was determined by averaging results of three independent experiments, with means and SDs shown. **P < 0.01; ****P < 0.0001. (E–H) Effect of TIM-3 knockdown on HIV-1 release in MDMs derived from three donors. Viral release was measured by quantifying the viral p24 level of the cell supernatants harvested at 48-h postinfection. The fold increase of HIV-1 release caused by TIM-3 shRNA knockdown relative to that of shRNA control was indicated. H shows a summary result of all three donors. *P < 0.05; **P < 0.01; ***P < 0.001.

We next assessed whether ectopic expression of an exogenous HIV-1 Nef protein can overcome TIM-1–mediated restriction of HIV-1 release, especially that of Δnef HIV-1. To this end, we transiently transfected HEK293T cells with increasing amounts (0.5 and 1.5 μg) of a plasmid encoding NL4-3 Nef, along with a proviral DNA construct encoding either NL4-3 WT or Δnef NL4-3, in the presence or absence of TIM-1. In the absence of TIM-1 expression, Nef caused only a small increase in virus release efficiency (Fig. 1 C and D); in contrast, in the presence of TIM-1, Nef overexpression in trans markedly rescued TIM-1–mediated inhibition of Δnef particle release while having only a small effect on WT particle release (Fig. 1 C and D, boxed in red). Interestingly, we found that, with a higher dose of exogenous Nef plasmid (e.g., 1.5 μg), TIM-1 levels were significantly up-regulated in both WT- and Δnef HIV-1–producing cells, despite the fact that viral production was increased for both WT and Δnef viruses (see details below).

MDMs express high levels of endogenous TIM-3 protein that intrinsically blocks HIV release (9). We next attempted to knock down TIM-3 in human MDMs derived from three healthy donors by shRNA and determined their effect on the release of HIV-1 WT and Δnef particles. We transduced MDMs with lentiviral shRNA vectors targeting human TIM-3 or scramble control, and following puromycin selection we infected cells with VSV-G–pseudotyped HIV-1 LAI particles, either lacking Env (labeled as “WT”) or lacking both Env and Nef (labeled as “Δnef”); the use of VSV-G pseudotypes was to ensure a comparable level of initial infection in WT and Δnef-producing cells so that de novo viral production could be compared (TIM proteins are also known to promote viral entry). In the absence of TIM-3 knockdown, the production of Δnef HIV-1 particles was ∼50% less than that of WT in MDMs of three donors (Fig. 1 E–H; P < 0.001). Notably, upon TIM-3 knockdown, the efficiency of which (∼70%) was determined by qPCR (SI Appendix, Fig. S1 C–F), we found that the level of HIV-1 release, especially that of Δnef particles, was significantly increased (P < 0.05 and P < 0.01 for both WT and Δnef, respectively); importantly, this led to an almost comparable level of release between Δnef and WT HIV-1 particles (Fig. 1 E–H), which strongly suggests that the Nef protein of HIV-1 NL4-3 counteracts the inhibitory effect of endogenous TIM-3 on HIV-1 release in human MDMs.

The TIM-1–Antagonizing Activity of Nef Is Conserved Among Primate Lentiviruses.

Primary HIV-1 isolates are classified into M, N, O, and P groups that resulted from independent zoonotic transmissions and share distinct sequence homology and geographical distributions (31). We next examined whether the Nef proteins of different primary HIV-1 isolates, as well as the Nef proteins of HIV-2 and simian immunodeficiency virus (SIV), can also counteract TIM-1. Indeed, we observed that the Nef proteins of the HIV-1 M, N, O, and P isolates tested, as well as SIVcpz and SIVgor Nefs, all relieved the inhibitory effect of TIM-1 on release of HIV-1 Δnef particles, based on the Western blotting analyses of viral p24 levels (Fig. 2A) and quantifications of the viral p24 band intensity vs. that of total viral and cellular Gag (Fig. 2B). In agreement with our initial finding shown above, the Nef proteins of all HIV and SIV isolates also increased the total TIM-1 protein levels in the viral-producer cells (Fig. 2A).

Fig. 2.

The ability of Nef to antagonize TIM-1 is conserved among primate lentiviruses. (A) Effect of Nef proteins derived from different HIV-1 groups and SIV on HIV-1 production. HEK293T cells were transiently transfected with NL4-3 Δnef and TIM-1 plasmids, along with expression vectors encoding the Nef alleles derived from HIV-1 group M (NL4-3, JRCSF), N (YBF30, N-2693), O (O-13127, O-HJ162), P (P-14788, P-RBF168), as well as SIVgor (CP2139, CR8757) and SIVcpz (EK505, MB897). HIV-1 production was determined as described for Fig. 1A. (B) Effect of lentivirus Nef proteins on release of NL4-3 Δnef as determined by averaging results of three independent experiments. (C) Effect of SIVmac Nef on TIM-1 inhibition of SIVmac production. HEK293T cells were transfected with proviral DNAs encoding SIVmac239 WT or SIVmac239 Δnef, along with increasing amounts of TIM-1 expression plasmids (200 or 500 ng). Levels of cell-associated and virion-associated SIVmac239 Gag proteins were determined by Western blotting using anti-SIV p27 antibody. (D) Relative RT activities from experiments described in C. Viral particles released from transfected cells described in C were quantified by RT assay. Data are means ± SDs of three independent experiments. ***P < 0.001; ****P < 0.0001. (E) Effect of HIV-2 Nef on the production of HIV-2 particles in the presence or absence of TIM-1. Western blotting was conducted to measure the expression of HIV-2 Gag in the cell lysates and in the purified virions. (F) Relative RT activities from experiments described in E. Shown are means ± SDs of three independent experiments by comparing with the RT activity of HIV-2 WT alone. ****P < 0.0001.

We next determined whether the Nef proteins of other primate lentiviruses can antagonize the TIM-1–mediated inhibition of release of their cognate viral particles. We found that, similar to the results of different HIV-1 lineages shown above, the inhibitory effect of TIM-1 on the release of Δnef SIVmac (Fig. 2 C and D) and HIV-2 (Fig. 2 E and F) variants was much stronger compared with its inhibition of their respective WT counterparts. Thus, the antagonistic effect of Nef on TIM-mediated restriction is conserved across primate lentiviruses.

Nef Promotes TIM-1 Internalization and Sequesters TIM-1 in Intracellular Compartments.

HIV-1 accessory genes often antagonize host restriction factors by downregulating their expression or sequestrating them in intracellular compartments (32). As the first step to understand the possible mechanism by which Nef antagonizes TIM-1, we examined the steady-state level of TIM-1 expression on the plasma membrane, the kinetics of TIM-1 internalization, and its intracellular localization in the presence or absence of Nef. We found that, despite the significant increase in the total level of TIM-1 expression in transfected cells shown above (Figs. 1C and 2A), the surface expression of TIM-1 in Nef-expressing cells was relatively low compared with cells expressing TIM-1 alone (Fig. 3 A and B).

Fig. 3.

HIV-1 Nef promotes internalization of TIM-1 from the plasma membrane and sequesters TIM-1 in intracellular compartments. (A and B) HIV-1 Nef modestly down-regulates TIM-1 expression on the plasma membrane. HEK293T cells were transfected with a FLAG-tagged TIM-1 plasmid in the presence or absence of NL4-3 Nef (an equal DNA input for Nef and TIM-1 vector), and the expression of TIM-1 on the plasma membrane was determined by using an anti-FLAG antibody. A summary plot of the relative geometric means of TIM-1 from three independent experiments is shown in A. *P < 0.05. A representative flow cytometry profile demonstrating a modest decrease of TIM-1 in the presence of Nef is shown in B. (C and D) HIV-1 Nef sequesters TIM-1 in intracellular compartments. HEK293T cells cultured in chamber slides were transfected with pNL4-3ΔNef and the human TIM-1 expression vector in the absence or presence of a vector expressing HA-tagged Nef. At 24-h posttransfection, cells were permeabilized and incubated with primary and secondary antibodies to detect TIM-1 (green) and Nef (anti-HA, red). Cells were mounted with Vectashield mounting media with DAPI and examined with a Delta-Vision RT deconvolution microscope (C). Colocalization was quantified by calculating the Pearson correlation coefficients (R values) using the softWoRx colocalization module (D). (E–G) COS-7 cells cultured in chamber slides were transfected with a vector expressing HA-tagged Nef; cells were fixed, permeabilized, and incubated for 1 h with antibodies specific for TIM-1, HA, p62, or TGN46 appropriately diluted in 3% BSA–PBS. In this experiment, the primary antibodies were directly labeled with either Zenon Alexa Fluor 488 (TIM-1, green), Zenon Alexa Fluor 594 (p62 or TGN46, red), or Zenon Alexa Fluor 647 (anti-HA, blue) using the Zenon antibody labeling kit (Thermo Fisher). Colocalization between TIM-1 and p62 and between TIM-1 and Nef was performed; no colocalization between TIM-1 and TGN46 was detected. (H) Comparison of the internalization kinetics of TIM-1 in the presence of WT and Nef mutants. HEK293T cells stably expressing TIM-1 were transfected with plasmids encoding Nef-GFP (WT), Nef-G2A-GFP, Nef-D123A-GFP, or Vpr-YFP (negative control, Mock). The geometric means of the fluorescence intensity of TIM-1 at each time points were recorded, compared, and their relative values were plotted against time by setting the value of time point 0 for each Nef construct or mock as zero. Results are averaged percentages ± SDs of the internalized TIM-1 protein from three to five independent experiments. ***P < 0.001. (I) Effect of Nef mutants on TIM-1–mediated inhibition of HIV-1 release. HEK293T cells were cotransfected with NL4-3 Δnef proviral DNA plus plasmids encoding TIM-1, WT Nef, or Nef mutants. The relative viral release efficiency was determined as described in Fig. 1 and as indicated. A summary plot of the effect of Nef mutants on antagonizing TIM-1–mediated restriction of HIV-1 Δnef virus release from four independent experiments (SI Appendix, Fig. S4I).

We next performed immunostaining and imaging analyses and observed that, in the absence of Nef (HIV-1 ΔNef), TIM-1 was predominantly expressed on the plasma membrane of transfected HEK293T cells (Fig. 3C, Bottom); however, when Nef was coexpressed (HIV-1 ΔNef + Nef), a significant portion of TIM-1 was detected within intracellular compartments (Fig. 3C, Top; Pearson coefficient for Nef/TIM-1 colocalization, ∼0.74; Fig. 3D). Importantly, Nef also strongly colocalized with endogenous TIM-1 in an internal compartment in COS-7 cells, which express high levels of TIM-1; the internal compartment in which Nef and TIM-1 colocalized in COS-7 cells contained the autophagy marker p62 (Pearson coefficient for TIM-1 and p62, 0.94) (Fig. 3 E and G) but not the TGN marker TGN-46 (Fig. 3F). Consistent with the increased overall levels of exogenous TIM-1 in Nef-expressing HEK293T cells (see above), Nef expression markedly increased the levels of endogenous TIM-1 in COS-7 cells, as examined by Western blotting (SI Appendix, Fig. S2A). In transfected HEK293T cells, TIM-1 also significantly colocalized with p62 (Pearson coefficient for TIM-1/p62 colocalization, ∼0.73; SI Appendix, Fig. S2 B and F), an autophagy-related marker, but not with other intracellular compartmental markers CD63, LAMP-1, and TGN46 (SI Appendix, Fig. S2 C–F).

Analysis of the TIM-1 internalization kinetics showed that HIV-1 Nef substantially enhanced the rate of TIM-1 endocytosis from the plasma membrane (Fig. 3H; see SI Appendix, Fig. S3 for flow cytometry profiles). We also examined two Nef mutants, G2A and D123A, in which a myristoylation acceptor glycine (G) and an acidic residue (D) were respectively replaced by an alanine (A) resulting in a defect in membrane targeting (33) or AP2 binding (34). These mutants had a more modest effect on modulating TIM-1 internalization (Fig. 3H and SI Appendix, Fig. S3) and sequestration (SI Appendix, Fig. S4 C and E) and were unable to antagonize TIM-1–mediated inhibition of HIV-1 production (Fig. 3I and SI Appendix, Fig. S4I).

We examined additional Nef mutants for their effect on TIM-1–mediated inhibition of HIV-1 release and observed that, like WT Nef, PP75AA and EE156QQ Nef fully counteracted the ability of TIM-1 to inhibit the release of HIV-1 Δnef (Fig. 3I and SI Appendix, Fig. S4I). The PP75 motif is located in the proline-rich SH3-binding domain and is known to be important for Nef⁺ virus growth but not CD4 down-regulation (35), whereas the EE156 motif is located in the flexible loop region crucial for β-COP binding and CD4 down-regulation (36). Interestingly, the Nef LL165AA mutant, which is deficient for AP2 binding and CD4 down-regulation (37), was still able to counteract the TIM-1–mediated inhibition of HIV-1 Δnef release, although less efficiently compared with WT, PP75AA, and EE156QQ (Fig. 3I and SI Appendix, Fig. S4I). Immunostaining and imaging assays showed that the Nef PP75AA and EE156QQ mutants sequester TIM-1 as efficiently as the WT, whereas the Nef LL165AA mutant exhibits a much-reduced ability to sequester TIM-1 (SI Appendix, Fig. S4 D, F, and G). These imaging data correlated with the differential abilities of the Nef mutants to antagonize TIM-1–mediated inhibition of viral production (SI Appendix, Fig. S4 H and I). Overall, these results demonstrate that expression of WT Nef protein increases TIM-1 internalization from the plasma membrane and likely perturbs its endocytic trafficking; this leads to the accumulation and stabilization of TIM-1 in intracellular compartments. This internal retention of TIM-1 by Nef reduces the capability of TIM-1 to block HIV-1 release. These data are reminiscent of the Vpu-mediated increase in overall tetherin levels in COS cells despite the ability of Vpu to counteract the inhibitory effect of tetherin on virus release in this cell line (38).

SERINC Proteins Potentiate Inhibition of HIV-1 Release by Stabilizing TIM-1, the Effect of Which Is Counteracted by Nef.

Recent studies have demonstrated that Nef antagonizes SERINC-mediated inhibition of HIV-1 infectivity by downregulating SERINCs from the plasma membrane, resulting in decreased SERINC incorporation into virions and therefore enhanced viral infectivity (23, 24). Given that both SERINCs and TIMs are antagonized by HIV-1 Nef, we sought to explore a possible interplay between SERINC and TIM proteins during HIV-1 production. We first used shRNA lentiviral transduction to deplete SERINC3 (two shRNA clones; SI Appendix, Fig. S5A) in HEK293T cells, which are known to express an endogenous albeit low level of SERINC3 (SI Appendix, Fig. S5B) (23, 24). Depletion of SERINC3 more efficiently overcame the inhibition of TIM-1 on HIV-1 Δnef production (by approximately threefold) compared with that of HIV-1 WT (∼30%) (Fig. 4A and SI Appendix, Fig. S5C). As would be expected, the infectivity of HIV-1 Δnef was more dramatically rescued by knocking down SERINC3, approximately by fourfold, compared with that of WT, which was ∼30% (Fig. 4B and SI Appendix, Fig. S5D). These results suggested that endogenous SERINC3 protein in HEK293T cells potentiates the overexpressed TIM-1 protein-mediated inhibition of HIV-1 release, particularly that of Nef-deficient virus.

Fig. 4.

SERINC3 and SERINC5 proteins potentiate TIM-mediated inhibition of HIV-1 production. (A) Effect of knockdown of SERINC3 on HIV-1 production. HEK293T cells stably expressing control shRNA or SERINC3 shRNAs (two clones) were transfected with proviral DNA plasmids encoding NL4-3 WT or Δnef together with TIM-1 plasmid. Viral production was evaluated by examining RT activity. Shown are means and SDs of three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001. (B) Effect of SERINC3 knockdown on HIV-1 infectivity as determined by infecting HeLa-TZM cells. *P < 0.05; ***P < 0.001. (C) Effect of coexpression of SERINC3 or SERINC5 with TIM-1 on HIV-1 production. HEK293T cells were transfected with plasmids encoding NL4-3 WT or ΔNef and a TIM-1 plasmid plus increasing amounts (200 and 500 ng) of pBJ-SERINC3-HA or SERINC5-HA plasmids. Forty-eight hours posttransfection, Western blotting was performed to examine Gag and SERINC (HA-tagged) expression in the cell lysates and purified virions using specific primary antibodies. (D) Effect of SERINC3 or SERINC5 coexpression with TIM-1 on HIV-1 release, as quantified by viral p24 vs. the total cellular Gag. Data are means ± SDs of three independent experiments relative to that of WT NL4-3 alone (“Vector”). (E) Effect of SERINC3 and SERINC5 on the expression of TIM-1 on the cell surface as determined by flow cytometry using an anti–TIM-1 antibody. (F–J) Effect of SERINC3 and SERINC5 knockdown on HIV-1 WT and ΔNef release in PBMCs and CD4⁺ T cells. The experimental design is shown in F. PBMCs (G and H) or CD4⁺ T cells (I and J) were transduced with VSV-G–pseudotyped lentiviral vectors encoding either control shRNA or shRNA targeting SERINC3 or SERINC5. Cells were then infected with HIV-1 LAI Δenv (WT) or LAI ΔenvΔnef (ΔNef) bearing VSV-G. The RT activities of newly produced HIV-1 virions were determined. Shown are means ± SDs of three independent experiments relative to the RT activity from PBMCs or CD4⁺ T cells transduced with control shRNA and infected with WT HIV-1. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; NS, not significant.

To further investigate the role of SERINC proteins in TIM-1–mediated inhibition of HIV-1 release, we cotransfected HEK293T cells with different amounts of SERINC3 or SERINC5 plasmids, along with a constant dose of TIM-1. We observed that coexpression of SERINC3 or SERINC5 with TIM-1 further decreased the production of HIV-1 WT and Δnef particles that was inhibited by TIM-1 (Fig. 4 C and D), which correlated with the increased expression of TIM-1 on the cell surface as determined by flow cytometry (Fig. 4E and SI Appendix, Fig. S5 E and F). Importantly, SERINC3 or SERINC5 alone did not inhibit HIV-1 release (even increased viral production in some cases) (SI Appendix, Fig. S5G). We next examined a series of Nef mutants harboring EDAA, LLAA, ΔCAW, or Δ12–39 mutations, some of which have recently been shown to less efficiently antagonize SERINCs (39). We observed that these Nef mutants exhibited an intermediate counteracting effect against TIM-1 relative to WT and Δnef (SI Appendix, Fig. S6 A and B). Altogether, these results suggest that SERINC proteins cooperate with TIM-1 to inhibit HIV-1 release, and that Nef likely antagonizes TIM-1, in part through SERINCs.

To determine the role of endogenous SERINCs in potentiating TIM-1 in CD4-positive T cells, we knocked down SERINC3 and SERINC5 either in activated human peripheral blood mononuclear cells (PBMCs) or purified CD4⁺ T cells known to express endogenous TIM-1 and TIM-3 proteins that intrinsically inhibit HIV-1 release (9). To avoid the potential confounding effect of SERINC knockdown on HIV-1 infectivity in multiple rounds of replication, we applied HIV-1 LAI Δenv (expressing Nef) and LAIΔenvΔnef (not expressing Nef) bearing VSV-G to determine the effect of SERINC3 or SERINC5 knockdown on a single-round HIV-1 production (see the procedure depicted in Fig. 4F). Relative to shRNA control-transduced cells, we observed roughly a threefold increase of HIV-1 production in SERINC3 or SERINC5-depleted PBMCs (donors 1 and 2) or CD4⁺ T cells (donors 3 and 4), again more prominent for HIV-1 LAIΔenvΔnef compared with LAIΔenv (Fig. 4 G–J). Notably, in three of the four donors examined (donors 1, 2, and 4), the production of LAIΔenvΔnef particles was ∼25–50% less than that of LAIΔenv particles, suggesting that Nef counteracts SERINC-mediated inhibition of HIV-1 LAI production (Fig. 4 G, H, and J). The knockdown efficiency of SERINC3 and SERINC5 in PBMCs and CD4⁺ T cells of these donors was about twofold to approximately threefold, based on quantitative RT-PCR (qRT-PCR) assays shown in SI Appendix, Fig. S7 A–D. Overall, data from these experiments revealed that endogenous SERINC3 and SERINC5 in human CD4⁺ T cells potentiate TIM-mediated inhibition of HIV-1 release, especially that of Nef-deficient viruses.

SERINC5 Stabilizes TIM-1.

One notable observation from the above experiments was that overexpression of SERINC3 and SERINC5 up-regulated the TIM-1 expression level in transfected HEK293T cells (Fig. 4C) and that knockdown of SERINC3 or SERINC5 decreased the TIM-1 expression in human PBMCs (SI Appendix, Fig. S7E). To understand the underlying mechanism, we performed pulse-chase labeling assays in HEK293T cells and compared the half-life of TIM-1 in the presence or absence of SERINC5 (Fig. 5). We found that SERINC5 greatly stabilized TIM-1 expression, extending its half-life from <2 to ∼5 h (Fig. 5). This likely accounts for the up-regulation of TIM-1 by SERINC5 (Fig. 4C) and also explains, at least in part, the potentiation of SERINCs on TIM-mediated inhibition of HIV-1 release. It remains to be determined how knockdown of SERINCs also down-modulates the RNA level of TIMs (SI Appendix, Fig. S7E).

Fig. 5.

SERINC5 stabilizes TIM-1 as determined by pulse-chase labeling assay. HEK293T cells were transfected with pCIneo-FLAG-TIM-1 in the presence or absence of pBJ-SERINC5-HA plasmid. Twenty-four hours after transfection, cells were subjected to pulse-labeling for 1 h and chased for indicated periods of time. Cell were lysed at 24-h posttransfection, and lysates were immunoprecipitated using an anti-FLAG antibody and resolved by SDS/PAGE. (A) A representative image of the pulse-chase labeling experiment. Proteins were detected by PhosphorImager analysis, and the TIM intensity in the presence or absence of SERINC5 was quantified by using Quantity One (Bio-Rad) software. Relative TIM-1 intensity was determined by setting the level of TIM-1 (time point 0) without SERINC5 to 1.00. (B) A summary plot of the half-life of TIM-1 in the absence or presence of SERINC5. Results are from five independent experiments.

Similar to Lentiviral Nef Proteins, MLV glycoGag and EIAV S2 Proteins also Counteract TIM-1.

In addition to HIV-1 Nef, MLV glycoGag and EIAV S2 proteins have also been shown to antagonize SERINCs and enhance HIV-1 infectivity (23, 24, 28–30). We therefore asked whether expression of these two proteins would overcome TIM-mediated inhibition of HIV-1 release similarly to HIV-1 Nef. Indeed, we observed that expression of MLV glycoGag restored release of HIV-1 particles that was inhibited by TIM-1 and that the effects were more pronounced for Δnef compared with WT (Fig. 6 A and B). Moreover, MLV glycoGag, like HIV-1 Nef (Fig. 1C), up-regulated the TIM-1 expression levels in transfected cells (Fig. 6A). Similar results were obtained for EIAV S2 (Fig. 6 C and D). Importantly, neither MLV glycoGag nor EIAV S2 alone had an effect on HIV-1 release (Fig. 6 E and F). Results for direct comparisons between HIV-1 Nef, MLV glycoGag, and EIAV S2 for counteracting TIM-1 are shown in SI Appendix, Fig. S8. Overall, these results support a model by which HIV-1 Nef, MLV glycoGag, and EIAV S2 share a common mechanism for antagonizing TIM-mediated inhibition of HIV-1 release, which is in part through SERINCs.

Fig. 6.

MLV glycoGag and EIAV S2 antagonize TIM-1–mediated inhibition of HIV-1 production. (A) Effect of MLV glycoGag on HIV-1 production. HEK293T cells were transiently transfected with plasmids expressing WT or Δnef NL4-3 (1 μg each) and TIM-1 (200 ng), together with increasing amounts of glycoGag plasmid (200 or 500 ng). Viral release was determined as described in Fig. 1C. (B) Averaged HIV-1 release efficiency was determined from results of three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001. (C) Effect of EIAV S2 on HIV-1 production. The experimental procedure was the same as described for MLV glycoGag, except that 200 ng of an EIAV S2 plasmid was used. (D) Effect of EIAV S2 on TIM-1–mediated inhibition of HIV-1 release was determined from results of three independent experiments. *P < 0.05; ****P < 0.0001. (E and F) MLV glycoGag and EIAV S2 alone do not inhibit HIV-1 release. The experiments were performed as described in A and C, except that no TIM-1 plasmid was transfected; results from one representative experiment are shown.

Discussion

HIV-1 Nef is a small, multifunctional protein that plays critical roles in AIDS pathogenesis. Among other functions, Nef modulates cell surface expression of various receptors, including CD4, MHC-I, CD8, CD28, and CD80, to promote viral immune evasion (reviewed in ref. 40). Consistent with its role in AIDS pathogenesis, SIVmac variants harboring deleted nef genes are less pathogenic in rhesus macaques (41). Similarly, infection with nef-defective HIV-1 strains is associated with low viral loads and delayed progression to AIDS in infected humans (42–44). Another important function of Nef is to antagonize cellular restriction factors that intrinsically inhibit lentivirus replication. For example, the Nef proteins of HIV-1 group O and most SIV strains are capable of counteracting tetherin by downregulating or sequestrating this antiviral protein, thereby enhancing lentiviral particle release (45–48). Recently, the Nef proteins of some HIV-1 M group strains were also shown to counteract tetherin (49). Here, we provide evidence that the Nef proteins of primate lentiviruses are capable of overcoming the inhibitory effect of TIM proteins on the release of HIV-1, HIV-2, and SIVs. Mechanistically, we find that Nef promotes the internalization of TIM-1 from the plasma membrane and sequesters TIM-1 in intracellular compartments. While the exact nature of these compartments remains to be determined, our preliminary data showed that Nef and TIM-1 are colocalized with p62, an autophagy-related marker, but not with TGN46, a marker for the cellular trans-Golgi network. Another intriguing finding of this study is that SERINC proteins, which themselves do not block HIV-1 release but impair viral infectivity in a manner antagonized by HIV-1 Nef, MLV glycoGag, and EIAV S2 (23, 24, 28–30, 50), potentiate TIM-mediated restriction of virion release in part by stabilizing TIM-1, revealing an intriguing interplay between SERINCs and TIMs at the late stage of HIV-1 replication.

How does Nef antagonize the function of TIM to inhibit HIV-1 release? Given our previous finding that the TIM–PS interaction is essential for TIM-mediated inhibition of HIV-1 release (9), we have considered the following models: First, Nef may directly down-regulate TIM expression in the cell, including its expression on the cell surface. Second, Nef may act directly on PS, either by inhibiting its synthesis, trafficking, flipping to the outer leaflet of the plasma membrane, and/or reducing its incorporations into virions. Third, Nef may directly disrupt TIM–PS interaction, in particular as it occurs in the extracellular space where the budded viral particles accumulate when TIMs are present. Last, Nef could act on some other protein(s) and/or lipid(s), thus indirectly counteracting the TIM-mediated inhibition of viral release. In this work, we have performed a series of experiments to test these possibilities. We excluded the possible effect of Nef on PS by examining the PS levels on the plasma membrane as well as in the viral particles by using FITC-labeled Annexin V or a nanoparticle plasmon coupling-based quantification (51). We also excluded the possibility that HIV-1 Nef could directly disrupt the TIM-1–PS interaction by performing an in vitro ELISA using a soluble form of TIM-1 and liposome enriched with PS. We reasoned that the interference by Nef of the TIM–PS interaction is topologically unfavorable, because Nef is essentially an intracellular protein despite the fact that it can be secreted into culture media and has been detected in exosomes (52, 53). Although more sensitive assays will be needed to precisely define the TIM–PS interaction in virus-producing cells and released viral particles, our current data strongly suggest that mechanisms unrelated to PS (see below) are involved in Nef antagonism of TIM-mediated restriction of lentiviral production.

Our Western blotting analyses consistently show that lentiviral Nef proteins increase, rather than decrease, the total levels of TIM-1 expression in transfected cells or endogenous TIM-1 in COS-7 cells. This up-regulation of TIM-1 by Nef could be due to the transcriptional activity of Nef that activates NF-κB (54, 55) or, alternatively, due to a slowed TIM-1 degradation during the protein biosynthesis and trafficking process. In this regard, our preliminary results suggesting autophagy is associated with Nef and TIM-1 colocalization are very interesting. We found that the internalization rate of TIM-1 is increased when Nef is ectopically expressed and Nef mutants deficient for membrane targeting (G2A) and/or endocytic trafficking (D123A) are incapable of efficiently promoting TIM-1 internalization or antagonizing TIM-1. Consistent with this finding, we observed that TIM-1 is sequestered within intracellular compartments by Nef, in sharp contrast to the predominant plasma membrane localization of TIM-1 when Nef is absent. Moreover, Nef mutants deficient in modulating TIM-1 internalization and trafficking are unable to antagonize the ability of TIM-1 to inhibit HIV-1 release. Hence, Nef appears to modulate the endocytic trafficking of TIM-1 to counteract its inhibitory effect on HIV-1 release. Additional experiments will be needed to dissect exactly how Nef modulates the trafficking of TIM-1, although this may be related to the reported interactions between Nef with the cellular clathrin adaptor protein 2 (AP-2) (34, 56–58).

SERINCs are multitransmembrane proteins that are normally expressed on the plasma membrane; the restrictive activity of SERINCs on HIV-1 infectivity is counteracted by Nef via Nef-mediated down-regulation of SERINCs from the plasma membrane (23, 24). In this work, we tested the hypothesis that SERINCs could act on TIMs, thereby indirectly counteracting TIM-mediated inhibition of HIV-1 release. Indeed, we find that knockdown of endogenous SERINC3 in HEK293T cells, or depletion of SERINC3 and SERINC5 in human PBMCs or purified CD4⁺ T cells, enhances the production of HIV-1 particles, particularly in the absence of Nef. Consistent with this finding, we showed that ectopic expression of SERINC3 or SERINC5 enhances the ability of TIM-1 to block HIV-1 release, correlating with an up-regulation of TIM-1 expression by SERINCs, including its expression on the cell surface. Additional lines of evidence that support a role for SERINCs in Nef-dependent antagonism of TIM restriction include the following: (i) Nef mutants deficient in antagonizing SERINCs, such as G2A, D123A, LL165AA, are less potent in counteracting TIMs (Fig. 3I and SI Appendix, Fig. S6 A and B); (ii) MLV glycoGag and EIAV S2 proteins, which effectively antagonize SERINCs, counteract TIM-mediated suppression of Nef-deficient lentivirus particle release (Fig. 6 A–D and SI Appendix, Fig. S8 A and B); (iii) knockdown of SERINCs in PBMCs and CD4⁺ T cells down-modulates TIM-1 and TIM-3 expression (SI Appendix, Fig. S7E); and, finally, (iv) coexpression of SERINC5 in HEK293T cells stabilizes TIM-1 expression (Fig. 5 A and B). Thus, SERINCs appear to contribute to TIM-mediated inhibition of HIV-1 release by stabilizing TIMs, and lentiviral Nef proteins counteract TIMs in part through acting on SERINCs.

We must emphasize that, as has been previously reported (23, 24), SERINC proteins alone do not affect HIV-1 release (SI Appendix, Fig. S5G), and the potentiating effect of SERINCs on TIM-mediated inhibition of lentivirus release is dependent on the levels of TIM proteins in viral-producer cells. This explains why Nef does not significantly affect HIV-1 release in HEK293T cells, which express a very low level of TIMs (9). However, when TIM-1 is ectopically expressed in HEK293T cells, knockdown of the endogenous SERINC3 led to an increase in HIV-1 production, especially in the absence of Nef (Fig. 4A). In human PBMCs and purified CD4⁺ T cells, which are known to express endogenous levels of both TIMs and SERINCs, we observed that knockdown of SERINC3 or SERINC5 led to enhanced HIV-1 production, more so for Δnef than for WT (Fig. 4 G–J). While SERINCs could modulate PS on the plasma membrane and/or in the virions, given that SERINCs have been previously implicated in the biosynthesis of serine-derived lipids including PS (59), our preliminary results show no significant difference in the levels of PS between SERINC-expressing and SERINC-depleted cells and in produced HIV-1 virions, consistent with a recent report (39). Instead, we found that SERINC5 greatly stabilizes TIM-1 protein in coexpressing cells, extending the half-life of TIM-1 from <2 h to up to 6 h (Fig. 5). This explains, at least in part, the potentiating effect of SERINCs on TIM-mediated inhibition of HIV-1 release. The link between TIMs and SERINCs, as well as the capability of Nef to antagonize both of these proteins, as revealed in this work (see working model in SI Appendix, Fig. S9), highlights an interesting but complex interplay between the HIV accessory protein Nef and host restriction by TIMs and SERINCs.

Materials and Methods

Plasmids and Constructs.

The pCIneo vector that expresses human TIM-1 with an N-terminal FLAG tag has been previously described (9). Molecular clones of HIV-1 NL4-3 and SIVmac239 were obtained from the National Institutes of Health (NIH) AIDS Reagent Program. HIV-1 proviral constructs NL4-3 Δnef (60), Δvpu (61), Δvif (62), Δvpr (63), and Δenv (64) were generated by using PCR-based mutagenesis based on the NL4-3 backbone. HIV-1 LAI Δenv, ΔenvΔnef, SIVmac239 Δnef, and HIV-2 Rod9 Δnef plasmids were gifts from Michael Emerman, Fred Hutchinson Cancer Research Center, Seattle. The HIV-2 Rod9 plasmid was obtained from the NIH AIDS Reagent Program. The HIV-1 NL4-3 Nef expression plasmid was obtained from Yong-Hui Zheng, Michigan State University, East Lansing, MI. Plasmids encoding HIV-1 NL4-3 IRES-eGFP WT and mutants Δnef, Δvpu, ΔvpuΔnef, as well as nef alleles of HIV-1 groups M, N, O, and P, and SIVs were kindly provided by Frank Kirchhoff and Daniel Sauter, Institute of Molecular Virology, Ulm University Medical Center, Ulm, Germany (46, 48, 65, 66). The human SERINC3, SERINC5, and MLV glycoGag expression plasmids were kindly provided by Heinrich Göttlinger, University of Massachusetts, Worcester, MA or Massimo Pizzato, University of Trento, Trento, Italy. The plasmids encoding Nef mutants (G2A, PP75AA, D123A, EE156QQ, and LL165AA, either fused with IRES-GFP or tagged with HA) were obtained from Massimo Pizzato. The TIM-1–YFP construct was obtained from Jose Casasnovas, National Center for Biotechnology, Madrid. Plasmids encoding TIM-3 shRNA, SERINC3 shRNA, SERINC5 shRNA, and control shRNA were purchased from Sigma. HIV-1 NL4-3 proviral DNA encoding Nef mutants EDAA, LLAA, ΔCAW, and Δ12–39 was kindly provided by Oliver Fackler, University Hospital Heidelberg, Heidelberg (39). The EIAV S2 expression vector was obtained from Frederick Fuller, North Carolina State University, Raleigh, NC (67).

Virus Production and Infection.

HEK293T cells were transfected with proviral DNA plasmids encoding HIV-1, SIV, or HIV-2, along with a TIM-1 expression plasmid in the absence or presence of HIV-1 Nef, MLV glycoGag, or EIAV S2 DNA by using calcium-phosphate. Twenty-four hours posttransfection, the supernatants were harvested and clarified through 0.2-µm filters. Virus production was quantified by measuring RT activity as previously described (9). HIV-1 release efficiency was also measured by quantifying the viral p24 signal on Western blots compared with total viral plus cellular Gag proteins. Alternatively, viral p24 levels were quantified by using an ELISA kit (catalog #5421; ABL). Virus infectivity was examined by infecting HeLa-TZM-bl cells and firefly luciferase activity was measured 48 h after infection according to the manufacturer’s instructions (Promega). For virus production in PBMCs and CD4⁺ T cells, cells were infected with HIV-1 WT or Δnef bearing VSV-G for 6 h. Cells were then washed three times with PBS and maintained for an additional 18 h; HIV-1 release in the supernatant was monitored by measuring RT activity.

Western Blotting.

Western blotting was performed as previously described (68, 69). In particular, we used a RIPA buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.1% SDS), which disrupts membrane-associated proteins, to lyse cells. The supernatants containing HIV-1 were harvested, and viral particles were concentrated by ultracentrifugation at 32,000 rpm (Sorvall; Discovery 100SE, Rotor TH641) for 2 h at 4 °C. Cell lysates and purified virions were dissolved in 5× sample buffer, resolved on 10% SDS/PAGE gel, and probed by antibodies against HIV-1 p24 (also for SIV and HIV-2 Gag), HIV-1 Nef, or TIM-1.

Knockdown of SERINC3 or SERINC5 in PBMCs and CD4⁺ T Cells.

PBMCs were from healthy, anonymous donors and were stimulated with 2 μg/mL PHA and 5 U/mL IL-2 for 3 d. Subsequently, PBMCs or purified CD4⁺ T cells were transduced with lentiviral vectors expressing SERINC3 or SERINC5 shRNA. The transduced PBMCs and CD4⁺ cells were selected with 1 µg/mL puromycin for 1–3 d before infection by HIV-1. The SERINC knockdown efficiency in PBMCs and CD4⁺ T cells was determined by measuring the RNA level of SERINC3 or SERINC5 by qRT-PCR.

Internalization Assay.

The internalization assay was performed as previously described (48). Briefly, HEK293T cells were transiently transfected with plasmids encoding a FLAG-tagged TIM-1 and Nef (including Nef mutants). Forty-eight hours posttransfection, the TIM-1 expression level on the cell surface was determined by staining with anti-FLAG and FITC-conjugated secondary anti-mouse antibodies at 4 °C for 1 h. After three washes with PBS, the internalization of TIM-1 was initiated by incubating cells at 37 °C for indicated periods of time. Cells were then split into two portions: one-half of the cells were washed with pH 2.0 buffer for 1 min; another half were left unwashed. Cells were then fixed with 3.7% formaldehyde and analyzed together by flow cytometry using anti-FLAG and FITC-conjugated secondary antibodies at 4 °C for 1 h. The percentages of internalized TIM-1 were calculated by dividing the intracellular fluorescent intensity by the total fluorescent intensity.

Pulse-Chase Labeling.

HEK293T cells were transfected with a TIM-1 expression vector (pCIneo-FLAG-TIM-1) in the presence or absence of SERINC5 expression vector (pBJ-SERINC5-HA). Twenty-four hours posttransfection, cells were starved with DMEM without cystine and methionine (MP Biomedicals) for 30 min and pulse-labeled for 1 h in DMEM containing 62.5 μCi of ³⁵S-Met/Cys. Cells were then chase-labeled for 0, 0.5, 1, 2, 4, and 6 h, followed by lysis in RIPA buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.1% SDS). The harvested cell lysates were immunoprecipitated with anti-FLAG beads (Sigma) and were resolved by SDS/PAGE. The phosphorimages were collected by Typhoon FLA 9000 (GE Healthcare) and quantified by Quantity One software (Bio-Rad).

Statistical Analysis.

All statistical analyses were carried out in GraphPad Prism 5, with Student’s t tests or one-way ANOVA, unless otherwise noted. Typically, data from at least three independent experiments were used for the analysis.

Cells and Reagents, qRT-PCR, and Immunofluorescence Microscopy.

Materials and methods for cells and reagents, qRT-PCR, and immunofluorescence microscopy can be found in SI Appendix.