Variable infectivity and conserved engagement in cell-to-cell viral transfer by HIV-1 Env from Clade B transmitted founder clones

Abstract

HIV-1 transmission is usually initiated by a single viral strain called transmitted/ founder (T/F) virus. In in vitro models, HIV-1 can efficiently spread via cell-free and virological synapse (VS)-mediated cell-to-cell infection. Both modes of infection require the viral glycoprotein Envelope (Env). The efficiency with which T/F Envs initiate VS and mediate cell-to-cell infection has not been well characterized. Here we tested a panel of isogenic HIV-1 molecular clones that carry different Clade B T/F Envs. We found that despite variable infectivity among different Env clones in the two modes of infection, T/F Envs generally mediated efficient VS formation and subsequent cell-to-cell transfer. In contrast, in vitro infectivity of the T/F Env clones was more variable and strongly correlated with intrinsic fusogenicity of various Envs. We speculate that the conservation of cell-to-cell transfer by T/F Env is indicative of a biologically important function of Env.

INTRODUCTION

HIV-1 glycoprotein Envelope (Env) is a viral protein expressed on HIV-1 virions as well as the surface of HIV-1 infected cells. It is responsible for mediating viral attachment to CD4 receptors on target cells and driving viral membrane fusion (1). As the only surface viral protein, it is also the target of neutralizing antibodies that can block the spread of the virus. HIV-1 Env is synthesized as a 160kD precursor glycoprotein, gp160, which is subsequently enzymatically cleaved into two non-covalently bound subunits, gp120 and gp41 (2). The surface subunit (SU) gp120 is the receptor binding protein. Its interaction with the CD4 receptor triggers the membrane fusion reaction (3, 4). The trans-membrane (TM) subunit gp41 has an ~150 aminoacid-long cytoplasmic tail, which serves several important functions, including: cellular trafficking of Env, directing polarized budding of virus particles, incorporation of Env into virus particles, regulation of viral fusion, and endocytosis of Env. Trimers of gp120/gp41 heterodimer form the Env spike structure, which HIV-1 is completely dependent upon for host cell entry and HIV-1 infection (1, 4, 5).

HIV-1 infection can be initiated by both cell-free and cell-associated virus. Cell-free infection occurs when free HIV-1 particles are released from infected cells, and proceed to infect non-adjacent, uninfected CD4⁺ T cells. It provides a mechanism of viral dissemination to uninfected CD4⁺ T cells at a distance via lymphatic flow or blood circulation. In contrast, cell-to-cell infection is mediated by direct adhesive cell-cell contact structures that form between HIV-1 infected and uninfected CD4⁺ T cells. This direct adhesive connection is called the virological synapse (VS). The T cell VS has been characterized as an actin-dependent polarization of viral proteins Env and Gag on the infected cells and CD4 receptors that are recruited on uninfected target cells to the cell-to-cell contact region (6). Interactions between intercellular adhesion molecules (ICAM) 1 and 3 and LFA-1 may further stabilize the VS (7, 8). Cell-to-cell infection is found to be much more efficient in vitro compared to cell-free infection (9, 10) and enables the virus to resist certain classes of antiviral drugs (11, 12) as well as broadly neutralizing antibodies (bNAbs) in an epitope- and viral strain-dependent manner (13, 14).

Previous studies from our group and others support a model of cell-to-cell transmission, whereby HIV-1 initially transfers across the VS in a co-receptor independent manner into trypsin-resistant endocytic compartments within the HIV-1 uninfected target CD4⁺ T cells (9, 15–17). Subsequent viral fusion requires viral protease (PR)-dependent cleavage of viral protein Gag and maturation of the virus from within the target cell (15). Time-lapse live imaging studies suggest that interactions between Env and CD4 occur prior to the recruitment of Gag to the cell-cell contact region (18), indicating that Env initially functions as an adhesion molecule during formation of VS (19). Viral fusion events within endocytic compartments of individual target cells have also been observed (15). Previous studies in humanized mice, non-human primates and ex vivo human explants implicate cell-associated HIV-1 and SIV-1 in systemic viral dissemination (20) and mucosal transmission (21, 22). Although the extent to which cell-to-cell infection of HIV-1 occurs in vivo remains uncertain, recent studies indicate that cell-to-cell infection is operative in vivo in humanized mice, especially in the CD4⁺ cell-dense lymphoid tissues (23).

Chronically HIV-1-infected individuals typically harbor very diverse HIV-1 populations in their blood. However, during acute mucosal transmission, viral diversity within donors is severely reduced through a genetic bottleneck. Acute HIV-1 infection is mostly initiated with a single viral strain or, in rare cases, more than one closely related strain called transmitted / founder (T/F) viruses (24–27). T/F viruses are initially homogeneous during acute infection and diversify over time by accumulating errors from reverse transcriptase, Pol II, or innate cellular cytosine deaminases APOBEC 3 (24, 27, 28). Investigators have been interested in understanding phenotypic properties of T/F Envs that are associated with viral transmission. T/F Envs can be distinguished from chronic viruses by their co-receptor utilization (29, 30). T/F viruses of different subtypes display differential preference for CD4/CCR5 expression levels during viral entry (31). T/F viruses have also been reported to be more infectious (30, 32) and to package 1.9-fold more Env per particle compared to chronic viruses (32). They also display enhanced dendritic cell interactions and increased resistance to type 1 interferon (IFN) over chronic control viruses (32–34). This resistance to IFN is thought to be mediated by resistance to IFN induced trans-membrane proteins (IFITMs) (35–37), which have been shown to antagonize Env protein and alter infection in both cell-free and cell-to-cell routes (35, 37). Other studies suggest that T/F variants do not inherently replicate faster than related non-transmitted viruses from the same donor near the estimated time of transmission regardless of type 1 IFN (38). Interestingly, mutants that escaped from adaptive immune response were found to become less resistant to IFN, and tend to reduce their transmission potential (35). Another report from Oberle et al observed that T/F viruses were more sensitive to INF compared to non-transmitted viruses from the same HIV-1- infected donor, while no other phenotypic properties were observed including cell-to-cell transmission efficiency, replicative capacity, entry kinetics and sensitivity to entry inhibitors and neutralizing antibodies (39).

Previous studies have also identified the degree of occupancy of potential N-linked glycosylation sites (PNGS) and found that T/F Envs possess fewer PNGS and shorter variable loops (V1V2) compared to chronic Envs from the same infected individual (40–44). While shorter V1V2 and fewer PNGS have generally been associated with greater sensitivity to antibody neutralization (40, 45–51), there are conflicting reports on the neutralization susceptibility of T/F viruses versus corresponding chronic viruses (52–55).

Previous studies of T/F Envs in the context of either infectious molecular clones or isogenic wild type proviral clones have mostly focused on cell-free infection. Characterization of HIV-1 T/F Envs in cell-to-cell transmission remains limited (13, 14, 39, 56–59). In this study, we constructed a panel of fluorescent protein-expressing, replication-competent HIV-1 proviral molecular clones that express T/F Envs, and established flow cytometry-based assays to follow cell-to-cell transfer and subsequent viral fusion in T cell lines during HIV-1 cell-to-cell transmission. This panel of T/F Envs is geographically and genetically diverse clade B viral isolates (Table 1). We examined their infectivity in cell-free as well as cell-to-cell infection, and observed strong correlation of infectivity in the two modes of infection. Different T/F Envs exhibited a broad range of relative infectivity that significantly correlated with Env fusogenicity, while remaining similar in their ability to initiate VS-mediated transfer of virus between infected and uninfected CD4⁺ T cells. This panel of T/F Envs-expressing proviral molecular clones provides a valuable Clade B reference panel to characterize T/F Envs and to study their neutralization sensitivity against bNAbs with high sensitivity and wide dynamics.

Table 1.

Reference panel for Clade B HIV-1 T/F Env clones

Env clone name	Reagent Contributor1	Accession number	Origin	Tier	Subtype
6535.3	A	AY835438	Washington DC	1B	B
QH0692	A	AY835439	Trinidad	2	B
PVO4	A	AY835444	Italy	3	B
WITO	B	AY835451	Alabama	2	B
REJO	B	AY835449	Alabama	2	B
RHPA	B	AY835447	Tennessee	2	B
SC422661	A	AY835441	Trinidad	2	B

¹Contributors

A- Drs. David Montefiore, Feng Gao and Ming Li

B- Drs. B. H. Hahn and J. F. Salazar-Gonzalez

RESULTS

Characterization of fluorescent protein-expressing proviral molecular constructs with T/F Envs

To characterize infection efficiency of HIV-1 T/F viruses in cell-free and cell-to-cell transmission, a panel of genetically and geographically diverse full-length Clade B T/F Envs (ARP, Table 1) was cloned into a previously described NL4–3 based fluorescent protein-expressing replication competent proviral molecular clone NLCI (14, 60), substituting NL4–3 Env. These genetically and geographically diverse Envs were initially isolated from patients at different Fiebig stages (61) during acute infection, representing the majority of currently circulating viral strains in North America. They displayed diverse neutralization sensitivities when pseudotyped with env-deficient HIV-1 backbone, as measured by standard TZM-bl assay (62). Since the panel of T/F Envs is all R-tropic, NLCI_JRFL was generated similarly as the R-tropic chronic virus control. These viral constructs allowed us to compare the infectivity of cell-free versus cell-associated infection, where the only differences among the clones are in the Env glycoprotein.

When transfecting 293T cells, all the viral constructs efficiently expressed Envs as demonstrated by Western Blot of cell lysates of transfected 293T cells (Figure 1A). Viral production efficiency of T/F constructs was comparable to wild type NLCI_NL4–3 (Figure 1B). To determine if T/F Envs can be efficiently incorporated into virus particles, Western Blot with lysed virus particles was performed (Figure 1C). Quantitation of Western Blots of lysed virus particles showed that Env incorporation into chimeric constructs carrying T/F Envs or JRFL Env was lower compared to NLCI_NL4–3 (Figure 1D). WITO Env incorporation efficiency was much lower than others and 6535.3 Env was packaged into virus particles at very low level (Figure 1C and 1D). The majority of the gp160 precursors were processed into gp120, with efficiency comparable to wild type NLCI_NL4–3 (Figure 1E).

Figure 1. T/F Env-chimeric viral genomes preserve intact virus production and Env incorporation.

(A) T/F Envs expression in transfected 293T cells by Western Blot. (B) Viral production from 293T cells transfected with NLCI constructs bearing different Envs was quantitated by p24 ELISA. (C) Env incorporation into viral particles was examined by Western Blot. The amount of sample loaded was normalized to the sample p24 content. (D) Quantitation of Western Blots for Env incorporation efficiency into virus particles. (E) Quantitation of Western Blots for Env processing efficiency from gp160 precursor to gp120. (F) Vpu activity was assessed by measuring down modulation of surface Tetherin expression on Tetherin^high Jurkat cells nucleofected with T/F constructs. Gray shaded histogram represents a non-stained control. Blue histogram shows the Tetherin expression level in each sample and red curve showed Vpu deficient virus control. (G) Quantitation of Tetherin down modulation compared to ΔVpu construct.

Because of overlapping open reading frames in env and other viral genes, the panel of constructs with T/F Envs contains chimeric tat, rev and vpu. The robust production of cell-free virus indicates that tat and rev are functionally intact in these chimeric viruses. One primary function of HIV-1 Vpu is to down-modulate Tetherin expression on the surface of HIV-1-infected cells. Chimeric vpu generated in these clones was found to be capable of down-modulating cell-surface Tetherin expression in Tetherin-high Jurkat cells by 2 to 4-fold compared to NLCI-ΔVpu (63), to a level comparable to the parental NLCI_NL4–3 construct (Figure 1F,G). This suggested that chimeric vpu in T/F Env constructs was still functional.

Infectivity of T/F viruses in TZM-bl assays

The infectivity of the viral clones was first measured by a single-round infection using the TZM-bl reporter cell line. Virus produced by transfecting 293T cells was used to infect TZM-bl cells in the presence and absence of 30 ug/ml DEAE-Dextran. The polycationic polymer, DEAE-Dextran, is thought to enhance adsorption of virions onto the cell surface and enhance viral infectivity by neutralizing charge repulsion between cell-free virus particles and cell-surface sialic acid (64). We observed that T/F viruses displayed a wide range of infectivity and that they were about 5 to 96-fold less infectious than lab-adapted strain NL4–3 and chronic viral strain NLCI_JRFL (Figure 2A). Cell-free infection of the T/F viruses was significantly increased in the presence of DEAE-Dextran (Figure 2A, C). Cell-to-cell infection assay was performed by co-culturing HIV-1 expressing donor cells and pre-seeded TZM-bl cells with and without the presence of DEAE-Dextran (Figure 2B). Acutely nucleofected Jurkat cells with NLCI constructs were utilized as HIV-1 expressing donor cells as previously described (14, 65). In contrast to cell-free infection, cell-to-cell infection of TZM-bl cells was not as responsive to DEAE-Dextran treatment (Figure 2B, C).

Figure 2. Infectivity of viruses with T/F Envs in TZM-bl cells.

(A) Infectivity of viruses was measured by of TZM-bl reporter cells in the presence and absence of 30ug/ml DEAE-Dextran. Viral input was all normalized to 4.5ng/well. (B) Cell-to-cell infection from HIV-1 expressing Jurkat donor cells to TZM-bl cell with and without the presence of DEAE-Dextran. Viral input was normalized to comparable level of nucleofection in Jurkat cells. (C) The effect of DEAE-Dextran on fold of increase in infection levels of TZM-bl cells by co-culturing with nucleofected Jurkat cells expressing HIV-1 carrying T/F Envs, over that by cell-free viruses.

Infectivity of T/F viruses in single round cell-free and cell-to-cell infections

While the TZM-bl reporter assay is a useful indicator of infection, epithelial cells are not natural targets of HIV-1. To assess their infectivity in physiologically relevant T cell lines, we utilized sensitive, flow cytometry-based, single-round cell-free and cell-to-cell infection assays. Cell-free infection assays were performed by mixing cell-free virus with CCR5-expressing T cells. In cell-to-cell infection, HIV-1-nucleofected Jurkat cells were used as donor cells, and a CCR5-expressing T cell line served as target cells, as previously described (14, 65).

Several CCR5-expressing T cell lines including MT4R5, CEM.NKR.CCR5, Molt4.CCR5 and A3.01.CCR5 were used as target cells in both cell-free and VS-mediated cell-to-cell infection assays. Transfected 293T cell-produced T/F Env-expressing viruses exhibited a broad range of infectivity in both cell-free (Figure 3A) and cell-to-cell infection in MT4R5 cells (Figure 3B). CEM.NKR.CCR5 cells showed similar infection levels as the MT4R5 cells in cell-free infection route when a minimum of 16ug/ml polybrene was present. A3.01.CCR5 cells were less infected, and Molt4.CCR5 cells were poorly infected (data not shown). In infection assays using T cell lines, lab-adapted strains and virus with T/F Env were normalized to the same amount of cell-free viral antigen or percentage of infected donor cells. We observed that viruses with T/F Envs were generally much less infectious, when challenged with comparable levels of viral antigen or infected cells, in comparison to lab-adapted strain NLCI_NL4–3 or chronic clone NLCI_JRFL in all cell lines tested (Figure 3A and B).

Figure 3. Single round infectivity of viruses with T/F Envs in different T cell lines.

(A) Single round cell-free infection levels of MT4R5 cells (left, 2ug/ml polybrene present), CEM.NKR.CCR5 cells (middle, 16ug/ml polybrene present), and A3.01.CCR5 cells (right, 2ug/ml polybrene present) by 4 ng/well viruses with T/F Envs. Error bars represent the SEMs from duplicates of two independent experiments. (B) Single round cell-to-cell infection levels of MT4R5 cells (left), CEM.NKR.CCR5 cells (middle) and A3.01.CCR5 cells (right) after co-culture with nucleofected Jurkat donor cells expressing HIV-1 with T/F Envs. Error bars represent the SEMs from duplicates of two independent experiments.

Recombinant viruses with T/F Envs are capable of multiple rounds of infection.

To investigate if the viruses were capable of productive infection, we also performed multiple-round infection assays using PHA-activated primary CD4⁺ T cells. Viruses with the same amount of p24 were used as cell-free inoculum (Figure 4A) and nucleofected Jurkat cells with comparable level of nucleofection efficiency were used as cell-associated inoculum (Figure 4B). Samples were collected every two days post infection. The percentage of infection in primary CD4⁺ T cells increased over time and plateaued at approximately four 4 days after infection, indicating that all viruses except clone 6535.3 were capable of multiple rounds of productive infection.

Figure 4. Multi-round infectivity of HIV-1 carrying T/F Envs in primary CD4+ T cells.

(A) Multi-round infection of activated primary CD4+ T cells initiated with transfected 293T cells produced cell-free viruses. (B) Multi-round infection of activated primary CD4+ T cells initiated by co-culturing with nucleofected Jurkat donor cells expressing HIV-1 that carry T/F Envs.

Viruses with T/F Envs are capable of efficient cell-to-cell transfer

Previous studies from our group and others suggested that during VS-mediated cell-to-cell transmission, HIV-1 is initially transferred across Env-CD4 dependent VS in a co-receptor independent manner into trypsin-resistant endocytic compartments, followed by viral fusion in response to viral maturation induced by cleavage of Gag by PR with delayed kinetics (9, 15, 18). Based on this two-step entry model, we proceeded to characterize cell-to-cell infection of T/F viruses by examining the efficiency of these viruses to transfer through VS from donor to target cells.

As the expression level of Env on the surface of HIV-1 infected cells may influence the efficiency of cell-to-cell infection, we first measured Env on the surface of HIV-1 expressing cells by antibody staining. T/F Envs were detected on the surface of Amaxa-nucleofected Jurkat cells by staining with the glycan-detecting monoclonal antibody 2G12 (Figure 5A, B). REJO and RHPA were non-reactive with the 2G12. In contrast to the 2G12 staining, pooled HIV-1⁺ patient IgG, VRC01 or b12 stained cells expressing the T/F clones weakly (data not shown).

Figure 5. HIV-1 with T/F Envs preserved efficient cell-to-cell transfer.

(A) Representative flow cytometry plot and histogram of cell surface HIV-1 Env on Jurkat cells expressing HIV-1 with different Envs (red line) compared with Δenv (filled gray). (B) Different Envs were detected on Jurkat donor cells expressing high levels of HIV-1 (indicated by mCherry expression level) with mAb 2G12. (C) Cell-to-cell transfer levels of MT4R5 cells and (D) CEM.NKR.CCR5 cells after co-culturing with nucleofected Jurkat cells expressing HIV-1 with T/F Envs. Error bars represent SEMs of duplicates from two independent experiments.

We then assessed the ability of the viral clones to engage in formation of VS and cell-to-cell viral transfer. The panel of T/F Envs was cloned into fluorescent virus clone HIV-1 Gag-iGFP (66), where GFP was inserted between MA and CA in frame, providing a sensitive fluorescent marker to indicate viral transfer between infected and uninfected cells. We used nucleofected Jurkat cells as donor cells and MT4R5 or CEM.NKR.CCR5 cells as target cells and performed the cell-to-cell transfer assay (65). Donor cells were normalized to the similar nucleofection efficiency by adjusting the amount of DNA in the nucleofection, or by diluting with un-transfected Jurkat cells. We found that the whole panel of T/F Envs displayed relatively comparable level of cell-to-cell transfer from nucleofected Jurkat cells to MT4R5 cells (Figure 5C) and CEM.NKR.CCR5 cells (Figure 5D), suggesting that the ability to engage in cell-to-cell transfer is similar among different T/F Envs. Surprisingly, even the poorly infectious clone 6535.3 was able to transfer at levels comparable with the more infectious clone QH0692. We note that the cell-to-cell transfer efficiency of Gag-iGFP-JRFL is still significantly higher than that observed in T/F constructs. However, the magnitude of difference in transfer efficiency between Gag-iGFP-JRFL and T/F constructs was about 4.6 to 26-fold lower than the difference observed in cell-to-cell infectivity with NLCI_JRFL relative to corresponding T/F Envs expressing constructs in CEM.NKR.CCR5 cells and as much as 5-fold lower in MT4R5 cells. Taken together, these results suggest that each T/F Env achieves sufficient cell-surface Env expression to mediate - similar and efficient synapse formation and cell-to-cell transfer of HIV-1.

Correlation of cell-free infectivity and cell-to-cell infectivity

When the relative infection efficiencies of cell-free infection and that of cell-to-cell infection were plotted against each other, we observed that the relative infection efficiencies of the two modes of infection were significantly correlated (Figure 6A, B). Interestingly, the T/F Envs generally possessed comparable abilities to initiate formation of VS and transfer of HIV-1 from donor to target cells. A significant correlation was not observed between the efficiency of cell-to-cell transfer and cell-to-cell infection (Figure 6C, D). The results presented suggest that the fusion step that follows cell-to-cell transfer is likely to be the rate-limiting step influencing the infectivity of various T/F Envs.

Figure 6. Relative infection efficiency of cell-to-cell transmission correlated with that in cell-free infection while not with the efficiency of cell-to-cell transfer.

The infection efficiencies of HIV-1 with different Envs in cell-to-cell transmission correlated with that in cell free infection (A) MT4R5 cells and (B) CEM.NKR.CCR5 cells. The infection efficiencies of HIV-1 with different Envs in cell-to-cell transmission were not correlated with the efficiency of cell-to-cell transfer in (C) MT4R5 cells and (D) CEM.NKR.CCR5 cells.

Env fusogenicity significantly correlated with viral infectivity in vitro

To directly measure fusogenicity of T/F Envs in cell-free and viral fusion after cell-to-cell transfer step, we employed a Cre-lox-mediated viral membrane fusion assay (67). We generated an indicator cell line that expresses a Cre-lox-activated genetic switch measuring the efficiency of Cre enzyme that is delivered to the target cell via viral membrane fusion of HIV-1 that packages large amounts of the Cre recombinase (67). A3.01.CCR5 RG cell line was generated by transduction with a retroviral vector containing the red-to-green switch cassette, in which dsRed was flanked by a pair of LoxP and followed by eGFP (Figure 7A) (67). We sub-cloned the panel of T/F Envs Gag-iCre construct, in which Cre is inserted in frame between MA and CA, preserving the protease cleavage sites (67). Upon viral fusion, Cre recombinase that was delivered from cell-free or cell-to-cell infection is introduced into target cells, bearing the Cre-sensitive genetic switch, thereby triggering a genetically encoded red-to-green color change (Figure 7A). The percentage of GFP-positive cells within target population provides a measure of viral membrane fusion following the addition of a similar input of cell-free or cell-associated virus (67). The relative abundance of GFP positive cells in target cells among different samples serves as a quantitative measure of the relative fusogenicity of each T/F Env. Env fusogenicity was measured in both cell-free virus particles (Figure 7B) as well as in cell-associated virus (Figure 7C). When Env fusogenicity was plotted against the relative infection efficiencies in cell-free and cell-to-cell infections, we observed that the relative infection efficiency correlated with Env fusogenicity in both routes of infection (Figure 7D, E). When Env fusogenicity was plotted against its cell-to-cell transfer efficiency, no significant correlation was observed (Figure 7F). These results suggest that the low infectivity of viruses carrying T/F Envs compared to lab-adapted virus is attributable to the intrinsically lower fusogenicity of T/F Envs. In our in vitro model, it is the intrinsic fusogenic potential of T/F Envs that ultimately determines cell-to-cell infectivity.

Figure 7. Infectivity significantly correlated with Env fusogenicity in both modes of transmission.

(A) A3.01.CCR5 RG cells were generated by transducing A3.01.CCR5 cells with MSCV floxR-G cassette that expresses dsRed and switched to GFP in response to Cre recombinase. Env fusogenicity was measured with in both cell-free virus particles (B) and cell- associated virus (C). The infection efficiencies of HIV-1 with different Envs in cell-free infection (D) and cell-to-cell infection (E) significantly correlated with the corresponding fusogenicity measured in A3.01.CCR5 RG cells in Cre-induced fusion assay, while not with the efficiency of cell-to-cell transfer (F).

Since these recombinant viral clones pair a T/F Env with NL4–3 molecular backbone, it is possible that incompatibility of the Env with the NL4–3 backbone may affect early events of viral infection up to the point of viral gene expression. To determine if complementation of low fusogenicity of T/F Envs in viral entry would rescue the defects in infectivity of T/F viruses, we pseudotyped the NLCI constructs with VSV-G. We infected CD4^- A2.01 cells and CD4⁺ CCR5⁺ A3.01.CCR5 cells, respectively, with non-pseudotyped or VSV-G-pseudotyped cell-free viruses. With the same amount of viral input, the original panel of NLCI viruses displayed a broad range of infectivity as described earlier, whereas all VSV-G pseudotyped viruses infected to similar levels indicating that the infectivity differences are predominantly due to the env gene that is used (Figure 8). These results suggested that weaker infectivity of viruses with T/F Envs may be attributed to the intrinsically low fusogenicity of the Env glycoprotein, again indicating that viral fusion, rather than the cell-to-cell transfer through VS, is the rate-limiting step during T/F virus infection, in both cell-free and cell-to-cell routes.

Figure 8. Cell-free single round infectivity of VSV-G pseudotyped viruses in A2.01 and A3.01.CCR5 cells.

Cell-free infection levels of A2.01 cells (A) and A3.01.CCR5 cells (B) by 5ng/well of viruses with T/F Envs (left), compared to 5ng/well of viruses pseudotyped with VSV-G (right). Error bars represent SEMs of duplicate wells from two independent experiments.

DISCUSSION

Cell-to-cell infection between HIV-1 infected cells and uninfected CD4⁺ T cells has been studied using different systems (6, 13, 39, 56, 57, 59, 68). Most previous studies of HIV-1 cell-free and cell-to-cell infection have utilized the lab-adapted HIV-1 strain NL4–3. Compared to lab-adapted viral strains, T/F viruses are less well characterized. Recent work on T/F viruses has sought to determine the properties that are selected for during newly acquired infection (26, 39, 54, 69–72). As the only surface viral protein that mediates virus infection, as well as being the major target for cellular and humoral immune responses, HIV-1 Env is likely to possess important transmission-related properties that are discernable from chronic strains, such as resistance to IFN and IFN-induced trans-membrane proteins (IFITMs) (33, 35, 37). Although disparate results have been reported as to the relative resistance of T/F against IFN compared to non-transmitted viruses (38, 39), IFITMs have been reported to restrict HIV-1 infection in both cell-free as well as cell-to-cell infection to varying degrees by directly interacting with Env, inhibiting Env processing or incorporation (35, 37).

In this study, we developed a panel of fluorescent protein-expressing infectious proviral molecular clones of HIV-1 that carry Clade B T/F Envs, enabling us to characterize and directly compare T/F viruses where the only genetic difference is the sequence that encodes the Env glycoprotein. We utilized a T cell-to-T cell transmission assay to study cell-free versus cell-to-cell infection efficiencies. We observed that the panel of viruses with T/F Envs displayed a wide range of infectivity in cell-free and cell-to-cell infection assays and were also less infectious when compared to lab strain NLCI_NL4–3 and chronic viral strain NLCI_JRFL in single round infection assays using various different T cell lines (Figure 3A and B) as well as in standard TZM-bl assays (Figure 2A, B). The robust infectivity of the lab strain NL4–3, as well as chronic strain JRFL, may relate to its history of being extensively passaged in vitro. The panel of viruses with various T/F Envs also exhibited different levels of replication during multiple rounds of productive infection (Figure 4) except for the 6535.3 Env, which did not replicate well in culture, perhaps due to its poor incorporation into viral particles (Figure 1C, D).

A previous study of some of the same T/F Envs by Ochsenbauer et al (69), found that T/F viruses, including RHPA, REJO and WITO, displayed similar replication kinetics in primary CD4⁺ lymphocytes at day 10 after infection measured by p24 production. In the present study, however, we observed a significant range in replication efficiency among different T/F strains, as well as between T/F strains and lab-adapted / chronic strain as early as two days after infection.

To assess synapse formation and the transfer of virus during HIV-1 cell-to-cell infection (15), we measured the efficiency with which the T/F Envs could initiate transfer of HIV-1 from infected donor cells to uninfected target cells. Interestingly, although the panel of viruses with T/F Envs displayed a wide range of cell-to-cell infection efficiencies, they were capable of similarly efficient cell-to-cell transfer into target cells (Figure 5C, D). Even the least infectious clone with 6535.3 Env was able to mediate VS formation as well as viral transfer comparable to the more infectious clones, such as QH0692. We also note that the levels of cell-to-cell transfer of virus were surprisingly insensitive to the variable levels of Env surface expression. We observed that T/F Envs were expressed on the surface of infected donor cells at different levels, as shown by surface Env staining with mAb 2G12 (Figure 5B), however, they all maintained the ability to engage in VS formation and subsequent cell-to-cell transfer. Taken together, these results suggest that Env expression levels may not be the rate-limiting factor in cell-to-cell transfer of HIV-1, and that cell-to-cell transfer is a conserved feature of T/F Env, independent of their infectivity in vitro.

Interestingly, we also observed that the relative infection efficiency in cell-free infection significantly correlated with that in cell-to-cell infection (Figure 6A, B), while neither of them displayed significant correlation with the efficiency of cell-to-cell transfer (Figure 6C, D). When fusogenicity of T/F Envs was measured using viral membrane fusion assay (67), we found that they directly correlated with the viral infectivity (Figure 7D, E). VSV-G pseudo-typing complemented the low fusogenicity of T/F Envs, and enabled the viruses to infect target cells with relatively comparable efficiency (Figure 8). This suggests that low infectivity in cell-to-cell infection may be attributed to intrinsically low fusogenicity of T/F Envs.

We interpret these results within a context of a “two-step entry” model (15), where a viral transfer step precedes viral membrane fusion. Cell-to-cell infection requires that Envs expressed on the surface of infected cells maintain their ability to interact with CD4 and engage in VS formation to allow transfer of virus in a fusion-independent manner. While the T/F Env generally showed similar transfer efficiency, fusion potential of T/F Envs can exhibit greater variation. During cell-to-cell infection, once transferred virus particles are associated with target cells, they may be able to infect with different efficiency depending on fusogenic potentials of the different Envs.

We also observe that viruses with T/F Envs exhibited lower infectivity in vitro compared to lab-adapted strain such as NL4–3 or JRFL Recent studies identified different conformational stages of mature unliganded Env trimers by assessing dynamics of Env using single molecule fluorescence resonance energy transfer (smFRET) (73). It has been observed that lab-adapted NL4–3 Env trimers tend to adopt an “open” conformation and spontaneously access high-energy states, which are associated with CD4 and co-receptor binding. However, greater proportions of Envs from primary isolates have been found to adopt a “closed” and much less dynamic ground state conformation (73, 74), reducing the ability to interact with CD4 and co-receptors (73). The strain from which lab-adapted clone NL4–3 was derived was extensively passaged in T cell lines without immune pressures (75), and therefore may have undergone selection for greater fusogenicity and infectivity. The “closed” conformation of T/F Envs, on the other hand, may result from maintaining a balance of the ability to establish infection and to evade immune pressures during viral transmission in vivo. Based on this model and the results we presented here, we suggest that the lower infectivity of T/F viruses is due to their intrinsically low fusogenicity and may be associated with selection pressure such as IFN during the process of viral transmission (33).

Viruses bearing various T/F Envs are similarly capable of efficient viral transfer from infected donor cells to uninfected target cells. We speculate that the conservation of cell-to-cell transfer by T/F Env is indicative of a biologically important function of Env. However, for both routes of viral transmission the intrinsic fusogenicity of T/F Envs determines viral infectivity. This dependency may be part of a cost-benefit evolution of the virus, and may allow for evasion of immune surveillance, or resistance to IFN and IFN-induced host factors such as IFITMs. Additional studies are required to determine the role it may play in acute HIV-1 infection.

MATERIAL AND METHODS

Viral constructs

A panel of full-length Clade B primary isolate Transmitted/Founder (T/F) HIV-1 Env (ARP Cat #11227, Drs. David Montefiori and Beatrice Hahn) were cloned into pNL4–3 based NLCI_NL4–3 backbone in place of NL4–3 (75) Env (Table 1). NLCI_NL4–3 is a proviral molecular clone with mCherry in the nef locus. Nef expression is restored by internal ribosome entry site (IRES) (60). NLCI_JRFL was generated using full-length JRFL Env, a chronic CCR5-tropic Env. Gag-iGFP and Gag-iCre, as previously described (66, 67), have GFP or Cre inserted into Gag in frame between MA and CA domains. The same panel of env was also cloned into Gag-iGFP and Gag-iCre respectively, replacing NL4–3 env. All constructs were generated using overlap PCR with restriction enzymes EcoRI and MluI. PCR amplified sequences were verified by Sanger sequencing.

Cells and cell culture.

Human cell lines Jurkat E6–1 and CEM.NKR-CCR5 were obtained from Dr. Arthur Weiss and Dr. Alexandra Trkola respectively from the NIH AIDS Reagent Program (ARP). Cells were maintained in RPMI 1640 medium containing 10% fetal bovine serum (FBS), 100U/ml penicillin, 100ug/ml streptomycin, and 2 mM glutamine (complete RPMI medium) and were passaged regularly and maintained at density of below 1 X 10⁶/ml. MT4R5 cell line was obtained from Dr. James E. Robinson, and cultured in complete RPMI medium supplemented with 2ug/ml puromycin. A3.01.CCR5 cell line was obtained from Dr. Robert McLinden from ARP. Propagation medium of these cells was complete RPMI supplemented with 1mg/ml G418. A2.01 cells were obtained from Dr. Thomas Folks from APR, and were cultured in complete RPMI medium. Cell lines 293T and TZM-bl cell lines were obtained from ATCC and maintained in Dulbecco’s Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS), 100U/ml penicillin, 100ug/ml streptomycin, and 2 mM glutamine. Primary CD4⁺ T cells were obtained from human peripheral blood through the New York Blood Center and isolated by negative selection with a Miltenyi CD4⁺ T cell isolation kit II (Miltenyi Biotec). Unactivated CD4⁺ T cells were maintained in 10U/ml interleukin (IL-2, ARP). Primary CD4⁺ T cells were activated by co-culturing with irradiated allogeneic peripheral blood mononuclear cells (PBMC) in complete RPMI medium with 100 U/ml IL2 and 2 ug/ml phytohemagglutinin (PHA) and used 2 to 3 days after activation. Stable cell line A3.01.CCR5 RG was generated by transduction with retrovirus MSCV containing a WPRE vector where dsRed was flanked by LoxP and followed by Cre-activated enhanced GFP (67) (plasmid 32702; Addgene, Sadelain lab and Clevers lab).

p24 enzyme linked immunosorbent assay (ELISA)

P24 ELISA was performed following a modified version of a previously published protocol (76). Costar 3922 flat bottom high binding plates were pre-coated with 50 ug/ml (in 0.1M NaHCO₃) anti-p24 capturing antibody (Aalto D7320) over night at room temperature. Coated plate was washed with 1x TBST, blocked with 2% nonfat milk (Lab Scientific) for 1 hour and washed again in TBST. HIV-1-containing supernatant was lysed and serially diluted with 1% empigen, and added to treated plates. P24 standard of known concentration was serially diluted in the same solution and incubated in the same plate. After 3-hour incubation at room temperature, the plate was washed 4 times with 0.05% Tris-buffered saline and Tween 20 (TBST) and incubated for 1 additional hour with 0.5 ug/ml alkaline phosphatase conjugated mouse anti-HIV p24 antibody (CLINIQA) in TBST with 20% sheep serum. After extensive washing with TBST, 50 ul of Sapphire Substrate (Tropix) was added into each well and incubated for 20 minutes before the luminescence was quantitated on Fluo Star Optima plate reader. Linear regression standard curve and sample analysis was performed using Prism (Graphpad software inc.).

Western Blot

Virus supernatant from transfected 293T cells was collected and quantitated by p24 ELISA. Virus was pelleted by ultracentrifugation through 6% Opti-prep density gradient medium (Sigma Aldrich) and re-suspended in PBS and quantitated by p24 ELISA. Both of concentrated virus and transfected 293T cells were lysed with 1% Tris-Triton lysis buffer supplemented with 1x protease inhibitor (Roche). Lysed 293T cells were centrifuged at 13,000 rpm for 3 min at 4°C. Total protein concentration in cell lysates was determined by Coomassie Plus assay kit (Thermo Scientific). Cell lysates and viral particle lysates were normalized according to protein concentration and p24 amount respectively. All samples were denatured and reduced before being separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Protein was transferred to Polyvinylidene difluoride (PVDF) membrane (GE Healthcare) for antibody detection. Polyclonal sheep anti-gp120 antibody (Dr. Micheal Phelan, ARP) was used to probe gp120 and gp160; Pooled Clade B HIV-1 infected patient IgG was used to probe all HIV-1 viral proteins (Dr. Luiz Barbosa, ARP). Horseradish peroxidase (HRP) conjugated anti-sheep or anti-human antibodies (Jackson ImmunoResearch Laboratories, Inc) were used as secondary antibodies. Chemiluminescence signal was detected by SuperSignal west Femto Maximum Sensitivity substrate (ThermoFisher Scientific) and visualized with FluorChem E imaging system (ProteinSimple). Signal intensity was quantitated with Image J (Wayne Rasband, National Institutes of Health).

Surface Tetherin staining

Nucleofection was used to transfect NLCI HIV-1 constructs in Tetherin high Jurkat cells (63) (Amaxa Biosystems). Viable cells were purified from transfected populations through ficoll paque (GE healthcare) density gradient purification 24 hours post transfection. Anti-tetherin APC antibody (Biolegend) was used at concentration of 5 ug/ml in PBS with 2% FBS for 30 minute at 4°C. Cells were fixed with 2% paraformaldehyde prior to analysis by flow cytometry.

TZM-bl assay

Cell-free virus particles were produced by transfection of 293T cells using Polyjet (SignaGen) with NLCI based primary isolates plasmid DNA. Cell-free virus particles were harvested 48 hours after transfection of 293T cells, centrifuged at 1000g, filtered with membrane with pore size of 0.45 um and stored in −80°C freezer. Viral supernatant was quantified by p24 enzyme linked immunosorbent assay. 4.5 ng of viruses were used to infect 2 × 10⁴ pre-seeded TZM-bl cells in flat bottom 96 well plates. HIV-1 virus-containing supernatant was added to each well and incubated for 48 hours in the presence or absence of 30 ug/ml of DEAE-Dextran. 48 hour post infection, TZM-bl cells were washed and lysed with cell culture lysis reagent (Promega) and 20 ul of each sample was read on Fluo Star Optima plate reader with injection of 100 ul of Luciferase Assay Reagent (Promega).

Cell-to-cell infection of TZM-bl cells was performed by overlaying 2 × 10⁴ nucleofected Jurkat cells (described above) on top of the same number of pre-seeded TZM-bl cells in flat bottom 96 well plate in the presence or absence of 2ug/ml DEAE-Dextran. 48 hours after infection, samples were collected and read the same way as in the cell-free assay.

Single round cell-free infection assay in T cells

Cell-free virus particles were prepared as described above. Different viruses were normalized to the same input of p24 antigen. Virus containing 4.5 ng p24 was used infect 1.5 × 10⁵CCR5 expressing T cells in each well of flat bottom 96 well plates. 2 ug/ml of polybrene was present in the infection assay to facilitate cell-free infection. In the case of A3.01.CCR5 cells, 16 ug/ml of polybrene was used. To ensure measurement of a single round of infection, at 18 hours post infection, medium was replaced with fresh medium containing 10uM azidothymidine (AZT; ARP). At 48 hours post infection, cells were treated with tryspin-EDTA (Gibco) to remove surface-attached viral particles, neutralized with complete RPMI medium, washed with phosphate-buffered saline (PBS, Sigma) and fixed with 2% paraformaldehyde. mCherry fluorescence signal from infected cells were detected by flow cytometry using BD LSRFortessa flow cytometer (BD Biosciences) and analyzed with Flowjo (Tree Star, Inc).

Single round cell-to-cell infection assay in T cells

Cell-associated viral inoculum was provided by Jurkat cells nucleofected with the panel of NLCI constructs that carry primary isolate Envs. Nucleofected donor cells were cultured overnight, after which dead cells were removed by Ficoll paque (GE healthcare) density gradient centrifugation. Nucleofected donor Jurkat cells and target CCR5-expressing CD4⁺ T cells were dye-labeled with 5uM cell proliferation dye eFluor 670 (eBiosciences) and 10 uM cell proliferation dye eFluor 450 (eBiosciences) respectively. At the time of co-culture, the percentage of HIV-expressing donor cells was adjusted to a similar fraction of all cells, by adjusting amount of DNA used in nucleofection or by diluting donor cells with non-nucleofected Jurkat cells. 1.5 × 10⁵ donor cells were co-cultured with same number of HIV-1 naïve target cells in each well of round bottom 96 well plates. To measure single round infection, medium was replaced with fresh medium containing 10 uM AZT about 18 hours after the co-culture. At 40 to 48 hours post infection cells were treated with trypsin-EDTA to remove surface-attached viral particles and to disrupt donor-target doublets. Cells were then neutralized with complete RPMI medium, washed with PBS and fixed with 2% paraformaldehyde. Samples were then analyzed with LSR Fortessa flow cytometer (BD Biosciences) and the Flowjo software (Tree Star, inc). mCherry fluorescence signal from efluor450 labeled target population represent cells that were infected with HIV-1 through cell-to-cell transmission.

Surface Env mAb staining.

Jurkat cells were nucleofected with T/F Envs carrying NLCI constructs as described previously in cell-to-cell infection assay. Viable cells were separated by ficoll gradient centrifugation, washed and placed into V bottom 96 well plate at density of 1.5 × 10⁵ per well and stained with 10 ug/ml HIV-1 monoclonal antibody 2G12 at 4°C for 45 min followed by 2 ug/ml Alexa Fluor (AF) 647 conjugated goat anti-human IgG (Life technologies) at 4°C for 30 min. Samples were then washed and fixed for FACS. Average of median fluorescence intensity was calculated from two independent experiments.

To quantitate cell-surface Env expression level, we examined the mean fluorescence intensity (MFI) of mCherry high population, corresponding to HIV-1 infected cells that express both early and late HIV-1 genes, including Env. The level of surface Env binding by each anti-HIV mAb was calculated as relative MFI, which is a ratio of MFI of Env-AF647 over secondary antibody alone control.

Relative MFI =(MFI Env-AF647 - MFI secondary antibody control) / MFI secondary antibody alone control.

Cell-to-cell transfer assay

Cell-associated viral inoculum was prepared by nucleofecting Jurkat cells with the panel of Gag-iGFP constructs that carry T/F Envs. Nucleofected donor Jurkat cells and target CCR5-expressing CD4⁺ T cells were dye-labeled with cell proliferation dye eFluor 670 and eFluor 450 respectively. At the time of co-culture, the percentage of HIV-expressing donor cells was adjusted to a similar fraction of all cells, by adjusting amount of DNA used in nucleofection or by diluting donor cells with non-nucleofected Jurkat cells. 1.5 × 10⁵ donor cells were co-cultured with same number of HIV-1 naïve target cells in each well of round bottom 96 well plates. 3 hours after the co-culture, cells were treated with trypsin-EDTA to remove surface-attached viral particles and to disrupt donor-target doublets. Cells were then neutralized with complete RPMI medium, washed with PBS and fixed with 2% paraformaldehyde. Samples were then analyzed with LSRFortessa flow cytometer (BD Biosciences) and Flowjo (Tree Star, inc). GFP+ signal from efluor450 labeled target population represent cells that obtained viral transfer from HIV-1 expressing donor cells.

Multiple rounds of infection

To examine the ability of the viruses to undergo multi-rounds of infection, we initiated the infection with either cell-free viruses produced from transfected 293T cells (2.5ng p24) or by nucleofected Jurkat cells as described in the single round cell-to-cell infection assay. Primary CD4+ T cells were isolated and activated as previously described, labeled with proliferative dye efluor450 and used as target cells in a multiround infection assay. We collected samples every two days and split cells to maintain optimal density. Samples were then analyzed with LSRFortessa flow cytometer (BD Biosciences) and Flowjo (Tree Star, inc). mCherry+ signal from efluor450 labeled target population represent cells infected with HIV-1.

Cell-free and cell-to-cell fusion assay

Cell-free virus particles that package Cre recombinase were produced by transfecting 293T cells and quantified using p24 ELISA. 4.5 ng of each different virus was used to infect 1.5×10⁵ A3.01.CCR5 RG reporter cells in flat bottom 96 well plate format. After 40 hour of incubation at 37°C, cells were washed with PBS, trypsinized to remove surface-attached free virus particles, neutralized with complete RPMI medium, washed again with PBS and fixed with 2% paraformaldehyde for flow cytometry.

For measurement of fusion in cell-to-cell infection, Jurkat cells nucleofected with Gag-iCre constructs that carry different Clade B T/F Envs were used as donor cells, and prepared as previously described (67). A3.01.CCR5 RG cells were used as target cells. The donor and target cells were labeled with cell proliferative dye eFluor670 and eFluor450 respectively. 1.5×10⁵ donor cells and the same number of target cells were mixed in each well of round bottom 96-well plate. After a 40-hour incubation at 37°C, we collected samples as previously described for flow cytometry analysis. GFP+ signal from eFluor450-labeled target population represent cells in which viral fusion events have occurred.