Effect of HIV-1 Env on SERINC5 Antagonism

ABSTRACT

SERINC5 is able to restrict HIV-1 infection by drastically impairing the infectivity of viral particles. Studies have shown that the HIV-1 Nef protein counters SERINC5 through downregulating SERINC5 from the cell surface and preventing the virion incorporation of SERINC5. In addition, the Env proteins of some HIV-1 strains can also overcome SERINC5 inhibition. However, it is unclear how HIV-1 Env does so and why HIV-1 has two mechanisms to resist SERINC5 inhibition. The results of this study show that neither Env nor Nef prevents high levels of ectopic SERINC5 from being incorporated into HIV-1 particles, except that Env, but not Nef, is able to resist inhibition by virion-associated SERINC5. Testing of a panel of HIV-1 Env proteins from different subtypes revealed a high frequency of SERINC5-resistant Envs. Interestingly, although the SERINC5-bearing viruses were not inhibited by SERINC5 itself, they became more sensitive to the CCR5 inhibitor maraviroc and some neutralizing antibodies than the SERINC5-free viruses, which suggests a possible influence of SERINC5 on Env function. We conclude that HIV-1 Env is able to overcome SERINC5 without preventing SERINC5 virion incorporation.

IMPORTANCE HIV-1 Nef is known to enhance the infectivity of HIV-1 particles and to contribute to the maintenance of high viral loads in patients. However, the underlying molecular mechanism remained elusive until the recent discovery of the antiviral activity of SERINC5. SERINC5 profoundly inhibits HIV-1 but is antagonized by Nef, which prevents the incorporation of SERINC5 into viral particles. Here, we show that HIV-1 Env, but not Nef, is able to resist high levels of SERINC5 without excluding SERINC5 from incorporation into viral particles. However, the virion-associated SERINC5 renders HIV-1 more sensitive to some broadly neutralizing antibodies. It is possible that, under the pressure of some neutralizing antibodies in vivo, HIV-1 needs Nef to remove SERINC5 from viral particles, even though viral Env is able to resist virion-associated SERINC5.

INTRODUCTION

The HIV-1 Nef protein plays important roles in viral pathogenesis and disease progression. This is manifested by the association of Nef mutants with slow disease progression in both HIV-1 and simian immunodeficiency virus infections (1–3). Nef is known to downregulate important immune molecules on the surface of HIV-1-infected cells. Examples include Nef downregulation of cell surface CD4 (4, 5), which prevents reinfection and avoids antibody-dependent cell-mediated cytotoxicity (6, 7), and downregulation of cell surface major histocompatibility complex class I (MHC-I) (8), which protects the infected cells from killing by cytotoxic T cells (9). Nef interacts with several kinases and thereby interferes with various signaling pathways in T cells (10–12). Nef also enhances the infectivity of HIV-1 particles (13). The latter function of Nef is conserved across HIV-1 strains and contributes to the maintenance of a high viral load in patients (14, 15).

Since the initial report that Nef enhances HIV-1 infectivity in 1994 (13), the molecular mechanism remained elusive until the SERINC5 (serine incorporator 5) protein was discovered to profoundly impair HIV-1 infectivity and it was found that its activity is countered by Nef (16, 17). The human SERINC family has five members, all of which contain multiple transmembrane domains and may be involved in incorporating serine into phospholipids to produce phosphatidylserine and sphingolipids (18). Recent studies have shown that SERINC5 and, to a lesser extent, SERINC3 ablate HIV-1 infectivity by blocking viral entry, likely through restricting the expansion of the viral fusion pore and thus preventing the release of the viral core into the cytoplasm (16, 17). In addition to Nef, which antagonizes SERINC5 through downregulating cell surface SERINC5 and thus preventing SERINC5 incorporation into HIV-1 particles, the viral envelope (Env) protein of some HIV-1 strains has also been reported to overcome SERINC5 inhibition (16, 17). SERINC5 must exert inhibitory pressure on viruses other than HIV-1 as well, since the glycosylated Gag protein of murine leukemia virus and the glycoproteins of vesicular stomatitis virus (VSV) and Ebola virus all counteract SERINC5 (16, 17).

In contrast to Nef, much less is known about how HIV-1 Env overcomes SERINC5. It is also unclear why HIV-1 has evolved two means, Nef and Env, to resist SERINC5. The results of this study demonstrate that HIV-1 Env, but not Nef, is able to resist high levels of SERINC5 without excluding SERINC5 from incorporation into viral particles. In spite of the greater ability of Env than Nef to counter SERINC5, virion-associated SERINC5 renders HIV-1 more sensitive to some broadly neutralizing antibodies.

RESULTS

HIV-1 strains YU-2 and AD8-1 but not NL4-3 are resistant to high levels of ectopic SERINC5.

The Pizzato group reported that Nef-negative and Nef-positive HIV-1 NL4-3 strains were equally inhibited by high levels of ectopic SERINC5 (16). In agreement with this observation, when we measured the infectivity of wild-type HIV-1 NL4-3 and NL4-3 carrying the nonfunctional NefG2A mutation, which were produced in the presence of increasing levels of SERINC5, both viruses were strongly inhibited in a SERINC5 dose-dependent manner (Fig. 1A). As a control, SERINC1 did not exhibit any anti-HIV-1 activity (Fig. 1A). Both the SERINC1 and SERINC5 proteins were well expressed in the transfected cells (Fig. 1B). We then asked whether there exist any HIV-1 strains that are able to resist these high levels of ectopically expressed SERINC5. Three primary HIV-1 strains, 89.6, YU-2, and AD8-1, as well as 10 transmitted founder viruses were tested by cotransfection together with SERINC5 DNA. The infectivity of these viruses was determined by infecting TZM-bl indicator cells. The results showed that, in contrast to NL4-3 and 89.6, which were profoundly inhibited by SERINC5, all transmitted founder viruses were much less inhibited and viruses AD8-1 and YU-2 showed complete resistance to SERINC5 (Fig. 1C).

FIG 1.

Susceptibility of different HIV-1 strains to inhibition by SERINC5. (A) Inhibition of NL4-3 and NL4-3(NefG2A) by different amounts of SERINC1 and SERINC5. HIV-1 DNA was cotransfected with increasing doses of SERINC1 or SERINC5 DNA. The infectivity of the viruses was determined by infecting TZM-bl cells. The results shown represent those from three independent experiments. RLU, relative light units. (B) Detection of SERINC1-Flag and SERINC5-Flag expression in cotransfected cells. Western blotting was performed using antibodies against the Flag tag, HIV-1 p24, and tubulin. The numbers on the left are molecular masses (in kilodaltons). (C) Inhibition of different HIV-1 strains by SERINC5. The DNA of different HIV-1 strains (500 ng) and SERINC5 DNA (25 ng) were cotransfected into HEK293T cells. The infectivity of the viruses was measured by infecting TZM-bl cells. (Top) The results of viral infectivity from one representative experiment are shown. (Bottom) The fold inhibition by SERINC5 from three independent experiments is summarized. (D) Alignment of the amino acid sequences of Nef from strains NL4-3, AD8-1, and YU-2. Conserved amino acids are indicated by asterisks; a colon indicates strong conservation of amino acids; a period indicates weak conservation. (E) Nef downregulates cell surface CD4 protein. Plasmid DNA that expressed Nef_NL4-3, Nef_AD8-1, and Nef_YU-2 and CD4 DNA were cotransfected into HEK293T cells. Cell surface CD4 protein was stained with PE-conjugated anti-CD4 antibody and detected by flow cytometry. (Left) The results of three experiments are summarized, with the level of CD4 expression in the absence of Nef being set equal to 100. P values were calculated using the Wilcoxon signed-rank test. (Right) The levels of Nef proteins were assessed by Western blotting. (F) The Nef proteins of strains NL4-3, AD8-1, and YU-2 did not overcome the inhibition by ectopic SERINC5. NL4-3(ΔNef) DNA was cotransfected with SERINC5 DNA as well as plasmid DNA that expressed Nef_NL4-3, Nef_AD8-1 or Nef_YU-2. (Left) The infectivity of the viruses was determined by infecting TZM-bl cells. (Right) The fold inhibition by SERINC5 from three independent transfections is summarized. *, P < 0.05; **, P < 0.01; NS, not significant.

It is possible that the observed resistance phenotype of AD8-1 and YU-2 is a result of the ability of their Nef proteins to counter high levels of SERINC5. To test this, we cloned the Nef genes of strains NL4-3, AD8-1, and YU-2 into expression vectors and first tested their function in downregulating cell surface CD4 (Fig. 1D and E). All three Nef proteins were expressed to similar levels (Fig. 1E). The Nef proteins of both strain NL4-3 (Nef_NL4-3) and strain AD8-1 (Nef_AD8-1) were equally effective in diminishing cell surface CD4, whereas the Nef of strain YU-2 (Nef_YU-2) was less effective, likely as a result of a sequence inserted in its N-terminal region (Fig. 1D and E) (19). We then cotransfected these Nef DNA clones with NL4-3(ΔNef) and SERINC5 DNA and observed that none of the three Nef proteins were able to markedly overcome the inhibition by the overexpressed SERINC5 (Fig. 1F). These data suggest that AD8-1 and YU-2 have a Nef-independent mechanism to resist high levels of ectopic SERICN5.

The V3 loop determines the ability of HIV-1 Env to counter SERINC5.

The Env protein of some HIV-1 strains has been reported to resist SERINC5 inhibition (16, 17). We therefore measured SERINC5 inhibition of an NL(AD8Env) virus that had the NL4-3 Env replaced by the Env of AD8-1 (20). The results showed that NL(AD8Env) was as resistant to SERINC5 as AD8-1, suggesting that the AD8-1 Env is sufficient to render the otherwise sensitive NL4-3 virus resistant to high levels of SERINC5 (Fig. 2A). The V1 and V2 loops of Env have been shown to contribute to SERINC5 resistance (17). Accordingly, we replaced the V1, V2, and V3 loop sequences in NL4-3 Env with their counterparts in the AD8-1 Env. The resultant viruses, NL(AD8V1V2V3), which had all three V loops replaced, and NL(AD8V3), which had only the V3 loop replaced, were resistant to the overexpressed SERINC5 (Fig. 2A). The NL(AD8V1) and NL(AD8V2) viruses were not viable and were not tested for SERINC5 inhibition. We further investigated the role of the V3 loop in resisting SERINC5 inhibition by inserting into NL4-3 Env the V3 loop sequences from HIV-1 strains YU-2, RHPA, WITO, and THRO. The phenotypes of these V3 loop chimeric viruses, NL(YU2V3), NL(RHPAV3), NL(WITOV3), and NL(THROV3), respectively, recapitulated the SERINC5 resistance phenotype of the parental viruses from which the V3 loop was derived (Fig. 2B). We also generated V3 chimeric viruses that had a nonfunctional mutant NefG2A protein and observed that these chimeric viruses were also resistant to SERINC5 inhibition (Fig. 2C).

FIG 2.

HIV-1 Env resists SERINC5 inhibition. (A) SERINC5 DNA was cotransfected with NL4-3 DNA that bears either the Env of AD8-1 (AD8Env), the V1, V2, and V3 loop sequences of AD8-1 Env (AD8V1V2V3), or only the V3 loop sequence of AD8-1 Env (AD8V3). (Left) The infectivity of these viruses was determined by infecting TZM-bl cells. (Right) The fold inhibition by SERINC5 from three independent transfections is summarized. (B) HEK293T cells were transfected with SERINC5 DNA together with NL4-3 DNA that had the V3 loop sequence from either the AD8-1, YU-2, RHPA, WITO, or THRO virus strain. (Left) The infectivity of these viruses in the absence or presence of SERINC5 was determined by infecting TZM-bl cells. (Right) The fold inhibition by SERINC5 from three transfections is summarized. (C) SERINC5 inhibition of NL4-3(NefG2A) carrying the V3 loop from different viruses. (D) Infectivity of viruses that were produced from parental Jurkat cells, SERINC3/5-knockout Jurkat cells, and SERINC3/5-reconstituted KO cells. These three different Jurkat cell lines were infected with NL4-3, NL(AD8V3), and NL(YU2V3) viruses that expressed either the wild-type or the mutated Nef protein. Viruses that were produced from these Jurkat cells were used to infect TZM-bl cells. Viral infectivity was calculated by normalizing luciferase activity (in relative light units) to viral RT activity. The fold change for four infections was calculated, with the infectivity of each virus from the parental Jurkat cells being set equal to 1. *, P < 0.05; **, P < 0.01; ***, P < 0.001; NS, not significant.

We expected that resistance to high levels of ectopic SERINC5 should also enable viral resistance to endogenous SERINC5. We therefore used viruses NL4-3, NL(AD8V3), and NL(YU2V3) to infect Jurkat cells, SERINC3 and SERINC5 (SERINC3/5)-knockout (KO) Jurkat cells, and knockout Jurkat cells that were reconstituted to express SERINC3 and SERINC5 (17). The infectivity of the viruses that were produced from these Jurkat cells was determined by infecting reporter TZM-bl cells. As expected, Nef-mutated NL4-3 acquired an approximately 30-fold higher level of infectivity when it was produced from the SERINC3/5-knockout Jurkat cells than when it was produced from either control Jurkat cells or SERINC3/5-reconstituted Jurkat cells (Fig. 2D). In contrast, similar levels of viral infectivity were measured for the NL(AD8V3) or NL(YU2V3) viruses, regardless of whether the viruses were produced from the control, the SERINC3/5-knockout Jurkat cells, or the SERINC3/5-reconstituted Jurkat cells in the absence of functional Nef (Fig. 2D). We therefore conclude that the V3 loop represents a key determinant in Env that enables viral resistance to both ectopically expressed SERINC5 and endogenous SERINC5.

HIV-1 Env is as potent as the VSV G protein in overcoming SERINC5 restriction.

Next, we asked whether HIV-1 Env is superior to other viral antagonists, such as the VSV G protein, in countering SERINC5. To answer this question, we transfected HEK293T cells with ΔEnv/NefG2A NL4-3 viral DNA together with Env_HxB2, Env_YU-2, or VSV G protein DNA as well as different doses of SERINC5 DNA. In contrast to the profound SERINC5 inhibition of HIV-1 carrying the Env_HxB2 protein, Env_YU-2 resisted this inhibition as effectively as the VSV G protein (Fig. 3A). We then investigated to what extent this SERINC5 resistance activity is conserved in HIV-1 Env proteins across different HIV-1 strains. Cotransfection experiments were thus conducted with a panel of HIV-1 Env DNA clones of different HIV-1 subtypes. A broad range of sensitivity to SERINC5 inhibition, from complete resistance to 70-fold inhibition, was observed (Fig. 3B and C). The results also revealed that the Env proteins of subtype A, C, and D strains were much more resistant to SERINC5 inhibition than those of subtype B strains, which suggests a possible subtype-specific resistance of HIV-1 Env to SERINC5.

FIG 3.

SERINC5 inhibition of HIV-1 bearing Env from different viral strains. (A) Env of YU-2 resists SERINC5 as effectively as the VSV G protein. ΔEnv/NefG2A NL4-3 DNA was cotransfected with different amounts of SERINC5 DNA and the Env DNA of either HIV-1 strain HXB2 or YU-2 or the VSV G protein. (Left) The infectivity of the viruses was determined by infecting TZM-bl cells. (Right) The fold inhibition by SERINC5 was also calculated. (B) SERINC5 inhibition of HIV-1 carrying Env from different viral strains. HEK293T cells were transfected with ΔEnv/NefG2A NL4-3 DNA, SERINC5 DNA, and Env DNA from different HIV-1 strains of different subtypes. The infectivity of the viruses in the absence or presence of SERINC5 was determined by infecting TZM-bl cells. (C) The fold inhibition by SERINC5 was calculated, and the averages from three experiments are presented.

SERINC5 sensitizes HIV-1 Env to inhibition by maraviroc and some neutralizing antibodies.

Studies have shown that Nef is able to prevent SERINC5 incorporation into HIV-1 particles (16, 17). To test whether HIV-1 Env employs the same mechanism to overcome SERINC5 inhibition, we cotransfected SERINC5 DNA with NL4-3, NL(AD8V3), or NL(WITOV3) viral DNA and measured the levels of virion-associated SERINC5 by Western blotting. The results showed that both wild-type NL4-3 and the Nef-mutated NL4-3(NefG2A) viruses carried similar levels of SERINC5 (Fig. 4A and B), indicating that Nef is unable to prevent the overexpressed SERINC5 from incorporation into virus particles. This explains the equally strong inhibition of both the wild-type and the Nef-mutated NL4-3 viruses by ectopic SERINC5. Interestingly, the SERINC5-resistant viruses NL(AD8V3) and NL(WITOV3), as well as the VSV G protein-pseudotyped viruses, also contained high levels of SERINC5 (Fig. 4A and B). Therefore, the NL(AD8V3) and NL(WITOV3) Env proteins, as well as the VSV G protein, act by resisting the inhibition of virion-associated SERINC5 rather than by preventing viral incorporation of SERINC5.

FIG 4.

Effect of virion-associated SERINC5 on inhibition of HIV-1 by maraviroc and T20. (A) Incorporation of SERINC5 (S5) into HIV-1 particles. SERINC5 DNA (25 ng) was transfected into HEK293T cells with ΔEnv NL4-3 DNA (500 ng) and VSV G protein DNA (25 ng) or with NL4-3 DNA that expressed an Env-bearing V3 loop from strain AD8-1 or WITO. Both wild-type Nef and the NefG2A mutant viruses were investigated. SERINC5 was transfected alone as a control for SERINC5 potentially associated with extracellular vesicles. Viral particles were harvested and subjected to Western blotting to detect virion-associated SERINC5. A representative Western blot is shown. The numbers on the left are molecular masses (in kilodaltons). (B) Relative levels of SERINC5 associated with HIV-1 particles. Protein band intensities were quantified using ImageJ software (NIH). The levels of SERINC5 in different cell lysates were adjusted by the levels of tubulin. The levels of SERINC5 in different virus samples were normalized to the levels of the viral CA protein, followed by further adjustment to the SERINC5 levels in the cell lysates. The final values represent the SERINC5 virion incorporation efficiency. The SERINC5 level in the wild-type NL4-3 virus is arbitrarily set equal to 1. The results shown are the averages from four independent cotransfection experiments. (C) Effect of SERINC5 on the inhibition of HIV-1 by maraviroc. NL(AD8V3) DNA was transfected into HEK293T cells with or without SERINC5 DNA to produce SERINC5-free or SERINC5-bearing virus particles. Viruses with the same amounts of RT activity were used to infect TZM-bl cells that had been pretreated with different concentrations of maraviroc. Viral infection was determined by measuring luciferase activity. The level of viral infectivity without maraviroc treatment was set equal to 100. (D) Effect of SERINC5 on the inhibition of HIV-1 infection by the fusion inhibitor T20. The same amounts of SERINC5-free and SERINC5-bearing NL(AD8V3) viruses were used to infect TZM-bl cells in the presence of different concentrations of T20. Viral infection was determined by measuring luciferase activity. The level of viral infectivity without T20 treatment was set equal to 100. The results shown are the averages from three independent infections.

We next investigated whether some HIV-1 Envs, such as the NL(AD8V3) Env, even though they are resistant to virion-associated SERINC5, might be affected by SERINC5 so that Env becomes vulnerable to some inhibitory pressures. We first measured the responses of SERINC5-free and SERINC5-bearing NL(AD8V3) viruses to the CCR5 inhibitor maraviroc and the fusion inhibitor enfuvirtide (T20). Much stronger inhibition of SERINC5-bearing NL(AD8V3) than SERINC5-free viruses by maraviroc but not by T20 was observed (Fig. 4C and D), which suggests that virion-associated SERINC5 might have interfered with the HIV-1 Env usage of CCR5 as the coreceptor for entry and, as a result, sensitized HIV-1 to the CCR5 antagonist maraviroc.

One possible impact of virion-associated SERINC5 on Env is a change in the Env conformation, which can be detected using antibodies that recognize specific epitopes on Env. We therefore tested a panel of neutralizing antibodies for their ability to inhibit SERINC5-free and SERINC5-bearing viruses (Table 1). Among the antibodies tested, three antibodies, 35O22, 4E10, and 10E8, inhibited SERINC5-bearing NL(AD8V3) or NL(NefG2A/AD8V3) viruses to a greater degree than they inhibited the SERINC5-free virus (Fig. 5A; Table 2). We further tested 4E10 inhibition of the NL(NefG2A/WITOV3), NL(NefG2A/RHPAV3), and NL(NefG2A/YU2V3) viruses and observed stronger inhibition of these viruses when SERINC5 was cotransfected (Fig. 5B; Table 2). In support of these data, viruses that were produced from SERINC3/5-knockout Jurkat cells exhibited greater resistance to the 35O22 antibody than viruses from either the parental Jurkat cells or the SERINC3/5-reconstituted cells (Fig. 5C; Table 3), which demonstrates the negative impact of endogenous SERINC3/5 proteins on the response of HIV-1 to inhibition by neutralizing antibody 35O22. Therefore, the virion-associated SERINC5 may have altered the accessibility of certain epitopes in HIV-1 Env to some neutralizing antibodies, including those targeting gp41 (such as 4E10 and 10E8) or its interface with gp120 (such as 35O22).

TABLE 1.

Summary of neutralizing antibodies used in this study

Neutralizing antibody	Target or function	Reference
VRC03	CD4 binding site	30
17b	Binds to a CD4-induced discontinuous epitope	31
7H6	MPERa	26
447-52D	V3 loop	32
PG16	V1/V2 loop	33
10-1074	V3 loop	34
2F5	gp41 epitope ELDKWA	27
Z13e1	MPER	28
4E10	gp41 epitope NWFDIT	25
10E8	MPER	26
35O22	gp41/gp120 interface	29

^aMPER, membrane-proximal external region.

FIG 5.

Effect of SERINC5 on inhibition of HIV-1 by neutralizing antibodies. (A) NL(AD8V3) viruses with either wild-type Nef or the NefG2A mutant were produced by transfecting HEK293T cells with or without SERINC5 DNA. Viruses with the same levels of RT activity were incubated with different concentrations of each neutralizing antibody before they were used to infect TZM-bl cells. Viral infection was determined by measuring luciferase activity. The level of viral infectivity without neutralizing antibody treatment was set equal to 100. Details about the neutralizing antibodies that were used are presented in Table 1. The results shown represent those from three independent infection experiments. The 50% inhibitory concentration values for antibodies 35O22, 4E10, and 10E8 were calculated from three independent experiments, and the results are summarized in Table 2. (B) Responses of NL(NefG2A/WITOV3), NL(NefG2A/RHPAV3), and NL(NefG2A/YU2V3) viruses to inhibition by the 4E10 neutralizing antibody in the absence or presence of SERINC5. The 50% inhibitory concentration values were calculated, and the results are shown in Table 2. (C) The NL(AD8V3), NL(NefG2A/AD8V3), and NL(NefG2A/WITOV3) viruses were produced by infecting Jurkat cells, SERINC3/5 (S3/S5)-knockout Jurkat cells, or SERINC3/5-reconstituted Jurkat cells. Viruses with the same amounts of RT were incubated with the neutralizing antibody 35O22 for 1 h before they were used to infect TZM-bl cells. Viral infection was determined by measuring the luciferase activity in the TZM-bl cell lysates. The 50% inhibitory concentration values were calculated from three independent infection experiments, and the results are summarized in Table 3.

TABLE 2.

IC₅₀s of neutralizing antibodies against HIV-1 containing ectopic SERINC5^a

Virus	IC50 (μg/ml) for neutralizing antibody:
	4E10		35O22		10E8
	−S5	+S5	−S5	+S5	−S5	+S5
NL(AD8V3)	>10	0.50 ± 0.04	>3	0.11 ± 0.01	>1	0.53 ± 0.02
NL(NefG2A/AD8V3)	>10	0.06 ± 0.05	>3	0.15 ± 0.02	>1	0.25 ± 0.09
NL(NefG2A/WITOV3)	7.84 ± 0.40	0.11 ± 0.09
NL(NefG2A/RHPA3)	9.22 ± 0.12	0.22 ± 0.10
NL(NefG2A/YU2V3)	7.81 ± 0.35	0.18 ± 0.20

^aIC₅₀, 50% inhibitory concentration; −S5, virus not bearing SERINC5; +S5, virus bearing SERINC5.

TABLE 3.

IC₅₀ of 35O22 neutralizing antibody against HIV-1 from Jurkat cells with or without SERINC3/5 expression^a

Virus	IC50 (μg/ml) for the following Jurkat cells:
	Parental	S3/S5 double KO	S3/S5 double KO, reconstituted
NL(AD8V3)	0.10 ± 0.12	>3	0.15 ± 0.20
NL(NefG2A/AD8V3)	0.21 ± 0.02	>3	0.41 ± 0.09
NL(NefG2A/WITOV3)	0.14 ± 0.09	2.6 ± 0.17	0.24 ± 0.03

^aIC₅₀, 50% inhibitory concentration; S3/S5, SERINC3/5.

DISCUSSION

The results of our study show that neither HIV-1 Nef nor Env is able to prevent high levels of ectopic SERINC5 from incorporation into virus particles. However, some HIV-1 Env proteins, but not Nef, resist the inhibition of virion-associated SERINC5, indicating that Env and Nef counteract SERINC5 by different mechanisms. Since Nef is known to enhance HIV-1 infectivity and can exclude SERINC5 from HIV-1 virions (16, 17), the inability of Nef to counter the ectopic SERINC5 indicates that a much higher level of ectopic SERINC5 than endogenous SERINC5 was used in this study. Nonetheless, experimentation with ectopic SERINC5 led to the finding of the greater ability of HIV-1 Env than Nef to overcome SERINC5, albeit by distinct mechanisms. Given that the envelope glycoproteins of VSV and Ebola virus also resist SERINC5 inhibition (16, 17), it is expected that more viral envelope proteins will be found to be refractory to SERINC5.

We mapped the HIV-1 Env determinant of SERINC5 resistance activity to the V3 loop. We were unable to examine the role of the V1 and V2 loops in the Env resistance of SERINC5. However, studies from other groups have suggested that the V1 and V2 loops allow Env to counter SERINC5 (17, 21). Since the V1 and V2 loops fold into a pocket in which the V3 loop resides, it is conceivable that V1, V2, and V3, as an interdependent structural entity modulating Env stability and the Env conformation as well as coreceptor usage, could function together to counter SERINC5.

Since SERINC5 impedes HIV-1 entry, likely through restricting the expansion of the viral fusion pore (16, 17), it is possible that some HIV-1 Env proteins, such as Env_AD8-1 and Env_YU-2, have an entry function strong enough to overcome this SERINC5 restriction, whereas those HIV-1 Env proteins with a relatively weaker entry function are inhibited by SERINC5, which may have led to the wide range of responses of different HIV-1 Env proteins to SERINC5 inhibition. In line with this possibility, when maraviroc was used to diminish the cell surface level of CCR5 that can be engaged by HIV-1 Env and, as a result, delay viral entry, otherwise resistant HIV-1 becomes sensitive to the inhibition of the virion-associated SERINC5.

In spite of the resistance of HIV-1 Env to SERINC5, the inability of Env to prevent SERINC5 from incorporation into virus particles allows the virion-associated SERINC5 to have the opportunity to impact the Env conformation. One consequence of this impact is the increased vulnerability of Env to inhibition by some neutralizing antibodies. This effect of SERINC5 on Env might help explain why the function of Nef to exclude SERINC5 from HIV-1 particles has been conserved across all HIV-1 strains. The Pizzato group previously showed that Nef renders HIV-1 refractory to neutralizing antibodies, including 4E10, which targets the MPER region of gp41, and that this function of Nef is independent of its ability to enhance virion infectivity (22). In light of our observation that virion-associated SERINC5 sensitizes HIV-1 to neutralizing antibodies, including 4E10, we suggest that one mechanism by which Nef protects HIV-1 from attack by some anti-gp41 antibodies may involve SERINC5 downregulation.

In summary, the results of our study demonstrate that HIV-1 Env is capable of resisting virion-associated SERINC5. This SERINC5 resistance function is also shared by envelope proteins of other viruses, including VSV. However, the sensitization of HIV-1 Env to some neutralizing antibodies by virion-associated SERINC5 may have pressured the virus to exclude SERINC5 from virion incorporation using Nef.

MATERIALS AND METHODS

Plasmids.

pNL4-3, pYU-2, and a panel of infectious molecular clones of transmitted founder viruses were obtained from the NIH AIDS Reagent Program. pAD8-1 and pNL(AD8Env) proviral DNA clones were kindly provided by Eric O. Freed (20). NL4-3 V3 chimeric plasmids were generated by inserting into NL4-3 DNA a synthesized env fragment of NL4-3 containing the V3 regions of different HIV-1 strains, including AD8-1, YU-2, RHPA, WITO, and THRO. The NL4-3 ΔEnv mutation was engineered by replacing 2 amino acids at positions 39 and 40 into two consecutive termination codons through site-directed mutagenesis. The NL4-3 ΔNef mutation was generated by replacing Nef codons 31 to 33 into three consecutive termination codons. The HIV-1 Env-expressing DNA clones Env_HxB2 and Env_YU-2 were kindly provided by Joseph Sodroski. Panels of HIV-1 Env-expressing plasmids, including HIV-1 subtype A, C, and D Env clones, were obtained from the NIH AIDS Reagent Program (catalog number 11947) (23). The panel of SGA HIV-1 subtype B clones was obtained from the NIH AIDS Reagent Program (catalog number 11663) (24). The cDNAs of SERINC genes were purchased from OriGene (SERINC1, catalog number RC206001; SERINC2, catalog number RC210091; SERINC3, catalog number RC202866; SERINC4, catalog number RC216546; SERINC5, catalog number RC230125). SERINC DNA sequences were amplified and cloned into the pQCXIP retroviral expression vector (catalog number 631516; Clontech). A Flag tag was added to the C terminus of each SERINC protein. Nef DNA was amplified from NL4-3, AD8-1, and YU-2 proviral DNA and inserted into pQCXIP. A hemagglutinin (HA) tag was added to the C terminus of each Nef protein. pMX-hCD4 (catalog number 16416; Addgene) expresses human CD4.

Cell lines.

Parental Jurkat cells, SERINC3/5-knockout Jurkat cells, and SERINC3/5-reconstituted KO cells were kindly provided by Heinrich Gottlinger (17). Parental Jurkat cells and SERINC3/5-knockout Jurkat cells were grown in RPMI containing 5% fetal bovine serum (FBS). SERINC3/5-reconstituted KO cells were grown in RPMI containing 5% FBS in the presence of 2 μg/ml puromycin (Sigma) and 150 μg/ml hygromycin B (Roche Diagnostics).

Virus production.

HIV-1 was produced by transfecting cells of the human embryonic kidney cell line HEK293T with HIV-1 proviral DNA. Viruses in the supernatants were clarified by centrifugation in a CS-6R centrifuge (Beckman Coulter) at 3,000 rpm for 25 min at 4°C. The amounts of viruses were determined by measuring viral reverse transcriptase (RT) activity. To investigate the effect of SERINC5 on HIV-1 infectivity, 500 ng of HIV-1 proviral DNA was cotransfected with different amounts of SERINC5 DNA into HEK293T cells that were seeded in 6-well plates. In experiments in which HIV-1 Env was supplied in trans from an Env-expressing plasmid to the NL4-3(ΔEnv) virus, after testing different doses of Env plasmid DNA, an amount of 25 ng Env DNA produced an infectivity similar to that of wild-type NL4-3 and was thus used in the NL4-3(ΔEnv) and Env DNA cotransfection experiments.

For virus production in parental Jurkat cells, SERINC3/5-knockout Jurkat cells, and SERINC3/5-reconstituted KO cells, DNA clones of NL4-3, NL(AD8V3), and NL(YU2V3), which express either wild-type Nef or the mutated NefG2A protein, were cotransfected with the VSV G protein DNA into HEK293T cells. Forty-eight hours later, viruses were harvested and concentrated by ultracentrifugation (29,000 rpm, 1 h) before they were used to infect the three Jurkat cell lines described above. After 24 h, the infected cells were washed 3 times with phosphate-buffered saline to remove free viruses and were resuspended in fresh medium. The supernatant was harvested 24 h later, and the amounts of viruses were determined by measuring viral RT activity.

Measuring viral infectivity.

Viral infectivity was measured by infecting TZM-bl indicator cells, which contain an HIV-1 LTR-luciferase expression cassette. These cells were obtained from the NIH AIDS Reagent Program (catalog number 8129). TZM-bl cells were first seeded into 24-well plates (40,000 cells per well) before being infected with HIV-1. At 48 h after viral infection, the TZM-bl cells were lysed in 1× passive lysis buffer (catalog number E1941; Promega). Cell lysates were mixed with luciferase substrate (catalog number E4530; Promega), and luciferase activity was measured using a luminometer. The levels of luciferase activity were normalized by the relative quantities of viral RT activity, and the results represent the infectivity of the virus particles.

To measure the inhibition of HIV-1 infection by the CCR5 inhibitor maraviroc (catalog number 11580; NIH AIDS Reagent Program), TZM-bl cells were incubated with different concentrations of maraviroc for 1 h at 37°C before they were infected with HIV-1. The fusion inhibitor T20 (catalog number 9845; NIH AIDS Reagent Program) was first mixed with HIV-1 and then immediately used to infect TZM-bl cells.

Virus particle analysis.

To detect the incorporation of SERINC5 into HIV-1 particles, HEK293T cells were transfected with HIV-1 proviral DNA together with SERINC5 DNA. At 48 h posttransfection, culture supernatants were first clarified by passage through a 0.2-μm-pore-size filter (VWR) to remove the cell debris. HIV-1 particles were pelleted through 20% sucrose by ultracentrifugation in an Optima L-100XP ultracentrifuge (Beckman Coulter) at 35,000 rpm for 1 h at 4°C. The pelleted virus particles were suspended in phosphate-buffered saline. The amounts of viruses were determined by measuring viral RT activity. Viruses with the same amounts of viral RT activity were examined by Western blotting using antibodies against HIV-1 p24 and Flag (to detect SERINC5-Flag).

Western blotting.

Transfected cells were lysed in Cytobuster protein extraction reagent (catalog number 71009; EMD Millipore Novagen) containing protease inhibitors (catalog number 11836153001; Roche) on ice for 20 min. After clarification by centrifugation, cell lysates were mixed with 4× Laemmli buffer. Protein samples were separated by electrophoresis in SDS-polyacrylamide gels, followed by transfer onto a polyvinylidene difluoride (PVDF) membrane (catalog number 3010040; Roche). The membranes were blocked in 5% skim milk (in phosphate-buffered saline) containing 0.1% Tween 20 (catalog number TWN510; BioShop) for 1 h at room temperature. The membranes were incubated with primary antibodies, including rabbit anti-p24 antibody (catalog number SAB3500946; Sigma-Aldrich), sheep anti-gp120 (catalog number 11710; NIH AIDS Reagent Program), rabbit anti-HA (catalog number H6908; Sigma-Aldrich), mouse anti-Flag (catalog number F1804-1MG; Sigma-Aldrich), mouse antitubulin (catalog number sc-23948; Santa Cruz Biotechnology), and mouse anti-HIV-1 Nef (catalog number 3689; NIH AIDS Reagent Program). After the membranes were washed, they were incubated with secondary horseradish peroxide (HRP)-conjugated antibodies consisting of either donkey anti-rabbit IgG (catalog number NA934V; GE Healthcare Life Science), sheep anti-mouse IgG (catalog number NA931; GE Healthcare Life Science), or rabbit anti-sheep IgG (catalog number 618620; Invitrogen). The membranes were treated with enhanced chemiluminescence (ECL) reagents (catalog number NEL105001EA; PerkinElmer), and the signals were visualized by exposure to X-ray films. The intensities of the protein bands in the Western blots were determined using ImageJ software (NIH).

Immunostaining of cell surface CD4.

HEK293T cells were cotransfected with 50 ng of Nef-HA DNA and 50 ng of pMX-hCD4 DNA. At 48 h after transfection, the cells were harvested and washed twice in phosphate-buffered saline containing 2% fetal bovine serum (FBS). Samples were fixed in 2% paraformaldehyde (PFA) and incubated on ice for 20 min, followed by incubation with phycoerythrin (PE)-conjugated anti-human CD4 antibody (catalog number 555342; BD Pharmingen) for 1 h. The cells were washed with phosphate-buffered saline containing 2% FBS, and cell surface CD4 was detected using a flow cytometer. The flow cytometry data were analyzed using FlowJo software.

Antibody neutralization assay.

We obtained from the NIH AIDS Reagent Program a panel of HIV-1 neutralizing antibodies, including 4E10 (catalog number 10091) (25), 10E8 (catalog number 12294) (26), 2F5 (catalog number 1475) (27), Z13e1 (catalog number 11557) (28), 35O22 (catalog number 12586) (29), VRC03 (catalog number 12032) (30), 17b (catalog number 4091) (31), 7H6 (catalog number 12295) (26), 447-52D (catalog number 4020) (32), PG16 (catalog number 12150) (33), and 10-1074 (catalog number 12477) (34). Viruses were incubated with different concentrations of each antibody for 1 h at 37°C and then used to infect TZM-bl cells. At 48 h after infection, TZM-bl cells were harvested and the levels of luciferase activity in the cell lysates were measured.

For neutralization assays with viruses that were produced from parental Jurkat cells, SERINC3/5-knockout Jurkat cells, and SERINC3/5-reconstituted KO cells, the same amount of virus (as measured from the viral RT activity) was incubated with different concentrations of neutralizing antibody 35O22 for 1 h at 37°C before infecting TZM-bl cells. Forty-eight hours later, the luciferase activity from the cell lysate was measured.

Statistical analysis.

The P values of parametric data sets were calculated by an unpaired, two-tailed Student's t test. The P values of normalized data sets were calculated using the Wilcoxon signed-rank test.