HIV-1 Subtype C with PYxE Insertion Has Enhanced Binding of Gag-p6 to Host Cell Protein ALIX and Increased Replication Fitness

Genomic differences within HIV-1 subtypes is associated with various degrees of viral spread, disease progression, and clinical outcome. Viral budding is essential in the HIV-1 life cycle and mainly mediated through the interaction of Gag with host proteins. Two motifs within Gag-p6 mediate binding of host cell proteins and facilitate budding. HIV-1C has a natural deletion of one of these two motifs, resulting in an inability to bind to host cell protein ALIX. Previously, we have identified a tetrapeptide (PYxE) insertion at this deleted motif site in a subset of HIV-1C patients. Here, we report the incidence of PYxE insertions in three different HIV-1C cohorts, and the insertion restores the binding of Gag to ALIX. It also increases viral growth even in the presence of the antiretroviral drugs lopinavir and tenofovir alafenamide. Hence, PYxE insertion in HIV-1C might be biologically relevant for viruses and clinically significant among patients.

ABSTRACT

Human immunodeficiency virus type 1 subtype C (HIV-1C) has a natural deletion of a YPxL motif in its Gag-p6 late domain. This domain mediates the binding of Gag to host cell protein ALIX and subsequently facilitates viral budding. In a subset of HIV-1C-infected individuals, the tetrapeptide insertion PYxE has been identified at the deleted YPxL motif site. Here, we report the consequences of PYxE insertion on the interaction with ALIX and the relevance regarding replication fitness and drug sensitivity. In our three HIV-1C cohorts, PYKE and PYQE were most prevalent among PYxE variants. Through in silico predictions and in vitro experiments, we showed that HIV-1C Gag has an increased binding to ALIX when the PYxE motif is present. To go more into the clinical relevance of the PYxE insertion, we obtained patient-derived gag-pol sequences from HIV-1C_PYxEi viruses and inserted them in a reference HIV-1 sequence. Viral growth was increased, and the sensitivity to the protease inhibitor (PI) lopinavir (LPV) and nucleoside reverse transcriptase inhibitor tenofovir alafenamide (TAF) was decreased for some of the HIV-1C PYxE variants compared to that of wild-type variants. Our data suggest that PYxE insertion in Gag restores the ability of Gag to bind ALIX and correlates with enhanced viral fitness in the absence or presence of LPV and TAF. The high prevalence and increased replication fitness of the HIV-1C virus with PYxE insertion indicates the clinical importance of these viral variants.

IMPORTANCE Genomic differences within HIV-1 subtypes is associated with various degrees of viral spread, disease progression, and clinical outcome. Viral budding is essential in the HIV-1 life cycle and mainly mediated through the interaction of Gag with host proteins. Two motifs within Gag-p6 mediate binding of host cell proteins and facilitate budding. HIV-1C has a natural deletion of one of these two motifs, resulting in an inability to bind to host cell protein ALIX. Previously, we have identified a tetrapeptide (PYxE) insertion at this deleted motif site in a subset of HIV-1C patients. Here, we report the incidence of PYxE insertions in three different HIV-1C cohorts, and the insertion restores the binding of Gag to ALIX. It also increases viral growth even in the presence of the antiretroviral drugs lopinavir and tenofovir alafenamide. Hence, PYxE insertion in HIV-1C might be biologically relevant for viruses and clinically significant among patients.

INTRODUCTION

Human immunodeficiency virus type 1 (HIV-1) is a global threat, with an estimated 36.9 million HIV-1-infected individuals and 940,000 deaths in 2017. Virulence of HIV-1 is determined by its capacity to replicate within infected cells and its ability to infect new cells. Among different HIV-1 subtypes, more than 50% of all infections are caused by HIV-1 subtype C (HIV-1C), which is prevalent in South Africa, Zimbabwe, Mozambique, Botswana, Ethiopia, Eritrea, India, and in some parts of Brazil. However, HIV-1C has also become highly prevalent in several European countries, including Sweden (1, 2). The gag gene encodes a polyprotein that is proteolytically processed by viral and cellular proteases into six final products: p17 matrix protein (MA), p24 capsid protein (CA), spacer peptide 1 (SP1), p7 nucleocapsid protein (NC), spacer peptide 2 (SP2), and p6 protein (P6). These proteins are required for the assembly and release of new virions (3–6). The p6 late domain of Gag contains two conserved short peptide motifs that bind host factors to facilitate viral budding (6–8). The PTAP motif is present in both HIV-1 and HIV-2, and sometimes it is duplicated in HIV-1 genomes (9, 10). It mediates binding to ESCRT-I subunit tumor susceptibility gene 101 (TSG101) (11). The YPx_nL (where x_n represents a random number of any possible amino acids) motif binds to the ESCRT-III adaptor protein ALG-2-interacting protein X (ALIX) (12). However, this motif is deleted exclusively in HIV-1C (13). This deletion is associated with the loss of binding of p6 late domain to ALIX and a decrease in virus release from infected cells. Of note, HIV-2 lacks the YPx_nL motif but contains an alternative ALIX-binding motif, namely, PYKEVTEDL, that originates from simian immunodeficiency virus (SIV) in rhesus macaques (14).

Recently, we identified a tetrapeptide insertion, PYxE (where x represents a lysine [K], a glutamine [Q] or an arginine [R]), within the Gag protein in a subgroup of HIV-1C-infected individuals (C_PYxEi strains) (15). This C_PYxEi strain was preferentially identified in East African HIV-1C-infected patients but was less common among HIV-1C-infected patients from South Africa, India, and Germany (15). Furthermore, the PYxE insertion appears more frequent among patients on antiretroviral therapy, e.g., ritonavir-boosted protease inhibitors (PI), than in therapy-naive patients in India, and it is more frequent in therapy failure patients in South Africa (15, 16). Moreover, lower pretherapy CD4⁺ T cell counts, higher plasma viral loads, and reduced increase in CD4⁺ T cell counts were noted to be associated with PYxE insertion in HIV-1C patients compared to that of patients with wild-type HIV-1C (HIV-1C_WT) from East Africa (17). Furthermore, we observed that increased replication fitness in PYQE-inserted HIV-1C viruses was polymerase independent (17). A more recent study also claimed that Gag-protease is the major determinant of subtype differences in disease progression among HIV-1 subtypes (18).

As the tetrapeptide PYxE insertion was found at the site of the missing YPx_nL motif of HIV-1C Gag-p6, we hypothesized that this PYxE insertion restores the interaction of the Gag-p6 late domain with ALIX and thereby increases replication fitness and reduces sensitivity to PIs. A recent report showed that the PYRE insertion could rescue the viral growth of a PTAP-deleted HIV-1 variant, similar to the YPx_nL motif in the absence of a PTAP sequence (19). However, since the PTAP motif is always present in clinical HIV-1 isolates, it is not known if and how PYxE insertion affects HIV-1C pathogenicity in a clinically relevant context. To address this, we assessed the distribution and frequency of this PYxE insertion in different cohorts of HIV-1C-infected individuals. We also evaluated the ability of the PYxE motif to reconstitute the interaction of Gag with ALIX, its association with increased viral growth of clinical isolates, and its association with drug responses against all three drug classes: reverse transcriptase inhibitors (RTIs), PIs, and integrase strand transfer inhibitors (INSTIs).

(This article was submitted to an online preprint archive [20].)

RESULTS

Evolution of Gag-p6 late domain sequences in HIV-1B and HIV-1C.

The sequence analysis of HIV-1B and South African and Indian HIV-1C Gag-p6 sequences (obtained from the Los Alamos Database) identified a conserved LYPx_nL motif in HIV-1B, which was missing from HIV-1C (Fig. 1a). We noted a unique evolution of the lysine (K) residue in the HIV-1B and HIV-1C Gag-p6 late domain, which is the target of ubiquitination. All the amino acids positions are per the standard HXB2 coordinates. While the K475 residue was conserved in both HIV-1B and -C, there was an evolution of a lysine residue at amino acid position 479 along with K481R mutation in the majority of HIV-1C sequences. Interestingly, the multiple-sequence alignment of HIV-1C_PYxEi strains identified conservation of the HIV-1C-specific K479 residue, but residue R481 in HIV-1C_WT changed to a lysine residue within HIV-1C_PYxEi strains and thus was identical to K481 in HIV-1B (Fig. 1a). When we analyzed Gag-p6 sequences from HIV-1C patients from the Swedish InfCare HIV cohort (n = 140), we found that 68% (95/140) of HIV-1C patients had a wild-type strain, whereas 32% (45/140) had a PYxE insertion; 22% (31/140) had a PYKE insertion, and 9% (13/140) had a PYQE insertion (Fig. 1b). All of the PYxE sequences in HIV-1C of patients from the Swedish cohort belonged to HIV-1C strains of Ethiopian and Eritrean origin. A similar trend was observed in the German HIV-1C cohort with 12% PYKE insertion (15/127) (Fig. 1c) and in the Ethiopian HIV-1C cohort with 37% (27/73) PYKE insertion (Fig. 1d). Thus, our analysis indicated that PYKE and PYQE are predominating in HIV-1C_PYxEi viruses of all cohorts analyzed. Subsequently, we restricted our analysis to these two variants.

FIG 1.

Consensus sequences and distribution of HIV-1 gag. (a) Aligned consensus sequences of gag from HIV-1 type B wild type (top), HIV-1 type C wild type (middle), and HIV-1 type C with a tetrapeptide insertion (bottom) from the Los Alamos database. Sequence logos with one-letter-coded amino acids were generated using WebLogo 3. (b to d) Distribution of HIV-1 type C with gag_wt, gag_PYKE, and gag_PYQE in patient cohorts from Sweden (b), Germany (c), and Ethiopia (d).

PYxE insertion enhances type C Gag: ALIX interaction.

Before testing the interaction between Gag and ALIX, we evaluated the cellular localization of both proteins. 293T cells were cotransfected with green fluorescent protein (GFP)-tagged ALIX with or without three codon-optimized variants of mCherry-tagged HIV-1C Gag, namely, wild-type Gag or Gag that includes the tetrapeptide insertion PYKE or PYQE. Confocal analysis showed cytoplasmic localization of ALIX and predominant plasma membrane localization of all three Gag variants (HIV-1C_WT, HIV-1C_PYQEi, and HIV-1C_PYKEi) as the complete Gag produces virus-like particles (Fig. 2). Thus, the tetrapeptide insertion did not cause a change or defect in the localization pattern of Gag proteins at the plasma membrane.

FIG 2.

Cellular localization of Gag variants and ALIX. Shown are confocal images of HEK293T cells cotransfected with the GFP-tagged ALIX and mCherry-tagged HIV-1C Gag variants (wt, PYKEi, and PYQEi) expression plasmids. Nuclei are visualized in blue, ALIX in green, and Gag in red. Bars, 20 μm.

Next, we evaluated the binding affinity of four HIV-1 Gag variants, HIV-1B_WT, HIV-1C_WT, HIV-1C_PYKEi, and HIV-1C_PYQEi, to ALIX (Fig. 3a). The late domain motifs’ peptides were selected based on crystal structures of PDB entries 2XS1 and 2R02. The binding affinity of the Gag peptide with ALIX was statistically significantly higher in HIV-1B_WT (32 ± 3 nM) than in any of the HIV-1C variants, i.e., HIV-1C_WT (290 ± 6 nM), HIV-1C_PYKEi (110 ± 4 nM), and HIV-1C_PYQEi (96 ± 4 nM) (Fig. 3b). However, statistically significantly higher binding affinity to ALIX was observed with HIV-1C_PYKEi and HIV-1C_PYQEi than with HIV-1C_WT.

FIG 3.

HIV-1C Gag with PYxE insertion has increased binding with ALIX. (a to d) Protein-peptide interaction using MST. The peptide sequences are given in the individual graphs. (e) The equilibrium dissociation constant (K_d) of the individual interactions. The smaller the K_d value, the greater the binding affinity of the ligand for its target. (f) HEK293T cells were cotransfected with GFP-tagged ALIX and Gag variants from HIV-1 type B or C. At 24 h posttransfection, cells were lysed and subjected to GFP pulldown. Input and pulldown samples were subjected to immunoblotting antibodies against GFP (ALIX), p55 (Gag), and β-actin. Data depicted are representative of at least two independent experiments.

As the peptides may not retain their conformational structures, as in the complete protein, we confirmed the binding by coimmunoprecipitation pulldown assays. GFP-tagged ALIX and HIV-1C Gag variants were transfected either alone or together in 293T cells. A plasmid encoding HIV-1B Gag was used together with GFP-ALIX as a positive control for Gag-ALIX interaction. Lysates were prepared 24 h posttransfection, GFP-tagged proteins were pulled down, and coimmunoprecipitated proteins were analyzed by immunoblotting (Fig. 3c). Pulldown of GFP-tagged ALIX was efficient, and the amount of HIV-1B Gag was increased when it was cotransfected with GFP-ALIX, confirming that HIV-1B Gag binds to ALIX. The HIV-1C Gag wild type was also detected when cotransfected with GFP-ALIX, but HIV-1C Gag PYKE was more prominently detected. The intensity of HIV-1C Gag PYQE was consistently lower than that of the PYKE variant but slightly increased compared to that of wild-type HIV-1C Gag. Thus, the tetrapeptide insertion PYxE in the HIV-1C Gag-p6 late domain promotes binding of Gag toward ALIX.

In silico analysis predicts binding of ALIX to the PYKE motif in Gag facilitated by ubiquitination.

As reported before, the lysine residues in Gag-p6 are subjected to ubiquitination (21). We first predicted the likelihood of ubiquitination of the lysine residue of PYKE (K_ins) compared to that of lysine residues adjacent to the PYKE motif. Indeed, K_ins was predicted to be more likely ubiquitinated than lysine residues in the vicinity of the PYKE motif (Fig. 4a). Since the PYKE motif is present at the missing ALIX-binding site and contains three charged amino acids that can facilitate interaction with other proteins, we used the PYKE insertion (amino acids 483 to 486) in the molecular modeling to predict the interaction of the Gag-p6 late domain with ALIX. Amino acids E482, Y484, and E486 of Gag could directly facilitate binding to ALIX (Fig. 4b and Table 1). Ubiquitin was predicted to bind K501 in ALIX and thereby affected the interaction between E486 in Gag-PYKE with K501 in ALIX. In addition, ubiquitin could also interact with E387 of ALIX and K_ins of Gag-PYKE through its residues K63 and E64, respectively. Thus, our in silico analysis predicted that the PYxE motif could restore the interaction of Gag with host cell ESCRT-III adaptor protein ALIX and that, in the case of a PYKE insertion, ubiquitination of the K_ins residue could further facilitate this interaction. We also tested the localization of ubiquitin by immunofluorescence. Since both Gag and ALIX are known to be ubiquitinated, it was not surprising that ubiquitin colocalized with ALIX and Gag (Fig. 4c).

FIG 4.

In silico binding prediction between PYKE inserted within HIV-1 type C Gag, ALIX, and ubiquitin. (a) Likelihood of ubiquitin binding to lysine residues in the vicinity of or within the PYKE motif. K_ins represents the lysine residue within the PYKE motif. (b) All proteins are presented as ribbon-like structures. The red ribbon represents Gag and its PYKE insertion, the blue ribbon ALIX and its Gag-binding sites, and the green ribbon ubiquitin. Amino acids involved in Gag-ALIX binding are shown in one-letter format. Interactions between amino acids are shown with black dotted lines. In silico docking was performed using Schrödinger software. (c) Confocal images of HEK293T cells transfected with the GFP-tagged ALIX and mCherry-tagged HIV-1C Gag PYQEi and HA-tagged ubiquitin expression plasmids. Nuclei are visualized in blue, ALIX in green, Gag in red, and ubiquitin in magenta. Bar, 20 μm.

TABLE 1.

List of amino acids that are involved in the interaction between Gag-P6, ALIX, and ubiquitin^a

Interacting atoms	Residue 1	Residue 2	Distance (Å)	Interaction type
OE1–NH1	E482 (LD)	R386 (ALIX)	2.96	Charge-charge
OE2–NH2	E482 (LD)	R386 (ALIX)	2.83	Charge-charge
OE1–NZ	E387 (ALIX)	K63 (Ub)	3.05	Charge-charge
OE2–NZ	E387 (ALIX)	K63 (Ub)	2.97	Charge-charge
OE2–NZ	E64 (Ub)	K485 (LD)	2.93	Charge-charge
OE2–NZ	E486 (LD)	K501 (ALIX)	2.95	Charge-charge
OE1–NZ	E16 (Ub)	K501 (ALIX)	3.12	Charge-charge
OH–OD1	Y484 (LD)	D506 (ALIX)	2.72	Hydrogen bond

^aE, glutamic acid; R, arginine; Y, tyrosine; D, aspartic acid; K, lysine; Ub, ubiquitin.

Differences in viral growth of clinical strains of HIV-1C with PYxE motif.

Finally, we assessed the consequence of the PYxE insertion on HIV-1C replication fitness using clinical strains obtained from HIV-1C-infected patients. gag-pol sequences were cloned from one wild-type HIV-1C virus with a single PTAP motif and no PYxE insertion (PT01), three HIV-1C strains with a PYKE insertion (PT04-06), and one with a PYQE insertion (PT07). We also included two HIV-1C strains with a PTAP duplication (PT02 and PT03), as it has been shown to increase the viral fitness and affect the drug sensitivity (Fig. 5a) (22, 23). Then these sequences were cloned into the genome of a reference virus (NL4-3, HIV-1B wild type), replacing its gag-pol sequence. The various recombinant viruses were tested for viral replication in an ex vivo viral growth kinetic assay and compared to the reference virus in the MT-4 cell line (Fig. 5b) and primary CD4⁺ T cells isolated from donor blood (Fig. 5c). In MT-4 cells, a statistically smaller amount of HIV-1C_WT virus was observed in the supernatant than that for the HIV-1C_PYxEi and HIV-1B reference viruses 3 days postinfection. At day 5 postinfection, the amount of HIV-1C_WT virus was still reduced compared to those of the other viruses. The amount of HIV-1C_WT in the supernatant was highest at day 7 postinfection. The reduction of HIV-1C_PYxEi and HIV-1B viruses at day 7 postinfection is most likely caused by the death of the virus-producing cells, which was observed microscopically. Similar data were observed in CD4⁺ T cells, where HIV-1C_WT, compared to all other viruses, showed the smallest amount of virus in the supernatant 7 days postinfection (Fig. 5c). Our data that viral replication for the HIV-1C wild-type clone was reduced compared to that of the HIV-1B reference virus are in concordance with a previous report (24). More importantly, our data showed that when a PTAP duplication or a PYxE insertion was present, viral growth was increased compared to that of the HIV-1C wild-type variant and was comparable to that of the HIV-1B wild type. This correlation suggests that HIV-1C strains with a PYxE insertion have a growth advantage compared to wild-type HIV-1C. This is also in line with our earlier study that HIV-1C_PYQEi viruses are more replication competent as well as more pathogenic (17).

FIG 5.

PYxE insertion increases viral growth and drug sensitivity toward protease inhibitor drug lopinavir. (a) gag-pol sequences were amplified from HIV-1 type C patient isolates and cloned into HIV-1 molecular clone pNL4-3. Recombinant viruses were propagated in HEK293T cells, and MOIs were determined in TZM-bl reporter cells. Viral growth assay was performed in MT-4 cells (b) or purified primary CD4⁺ T cells (c). Cells were infected with the various recombinant viruses at an MOI of 0.05, and supernatants were harvested on days 0, 3, 5, and 7. Viral loads in supernatants were measured by HIV-1 p24 analysis on the HIV COBAS 8000 platform. Data represent the means ± standard deviations from triplicates. Clinical data regarding viral load at initiation of therapy (d), CD4⁺ T cell count at initiation of therapy (e), and nadir CD4⁺ T cell count (f) were collected from the Swedish HIV-1C cohort, and patients were grouped as HIV-1C_PYxEi-infected (n = 45) or HIV-1C_WT-infected individuals (n = 95).

We next checked whether the ex vivo data corroborate with the clinical data. In the Swedish HIV-1C cohort (n = 140), we checked the nadir CD4⁺ T-cell count, CD4⁺ T-cell count at the initiation of therapy, and viral load at initiation of therapy. The viral load was significantly higher in HIV-1C_PYxEi-infected individuals than in the individuals who were infected with wild-type HIV-1C viruses (Fig. 5d). The CD4⁺ T-cell count at the initiation of therapy (Fig. 5e) and nadir CD4⁺ T-cell count (Fig. 5f) were also statistically lower in the HIV-1C_PYxEi-infected individuals than in individuals infected with wild-type HIV-1C viruses. These data further strengthen our ex vivo findings.

Effect of PYxE strains in susceptibility toward antiretroviral drugs.

Several reports have shown that a PYxE insertion is more frequent among HIV-1C therapy failure patients (16, 25). To examine whether recombinant viral clones without any known drug resistance mutations to RTIs, PIs, and INSTIs responded differently to various antivirals, the target cells were infected with individual viral clones in the presence of a specific drug. The viral replication was assessed, and the 50% effective concentration (EC₅₀) was calculated and fold change of the EC₅₀ was noted (Fig. 6). No differences in EC₅₀ values were observed among the various recombinant viral clones in the presence of the nonnucleoside RTIs (NNRTIs) efavirenz (EFV), rilpivirine (RPV), and etravirine (ETR) or the INSTIs raltegravir (RAL), elvitegravir (EVG), dolutegravir (DTG), and cabotegravir (CAB).

FIG 6.

PYxE insertion increases drug sensitivity toward protease inhibitor drug lopinavir. TZM-bl reporter cells were infected with the individual recombinant HIV at an MOI of 0.05 and cultured in the presence of individual antiretroviral drugs at different dilutions. EC₅₀ values for each virus with each drug were determined, and EC₅₀ fold changes (FC) were calculated compared to those of the pNL4-3 reference virus. PI, protease inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; INSTI, integrase strand transfer inhibitor; ATV, atazanavir sulfate; DRV, darunavir ethanoate; LPV, lopinavir; AZT, azidothymidine; TAF, tenofovir alafenamide; EFV, efavirenz; RPV, rilpivirine; ETR, etravirine; RAL, raltegravir; EVG, elvitegravir; DTG, dolutegravir; CAB, cabotegravir. The FC data presented in the heat map were the means from three individual experiments.

In contrast, with the PI lopinavir (LPV), one out of two viral clones with a PTAP duplication and two out of four with PYxE insertion showed a >4-fold decreased sensitivity toward the drug. However, there was no change in susceptibility against DRV and ATV. One of the other viral clones with a PYxE insertion showed a 3.5-fold decreased sensitivity to the nucleoside reverse transcriptase inhibitor (NRTI) tenofovir alafenamide (TAF). Thus, our data indicate that PYxE insertion alters sensitivity to the PI LPV and NRTI TAF, which may be due to the increased replication fitness. However, this is not universal across all the HIV-1_PYxEi strains and could be due to coevolution of other amino acids in the Gag protein.

DISCUSSION

Genetic diversity among HIV-1 strains and their adaptation can provide advantages for individual viral strains in different biological contexts. Here, we described the distribution of HIV-1C containing tetrapeptide PYxE insertion within Gag and assessed the binding efficiency of Gag with the host cell protein ALIX. Our results showed that the PYxE insertion enhances the binding capacity of HIV-1C Gag to ALIX in a potentially ubiquitin-dependent or -independent manner. Also, the insertion increased the replication fitness of the virus in both an in vitro model using a T cell line as well as in primary CD4⁺ T cells and seemed to alter the sensitivity against the PI LPV and NRTI TAF in some of the strains.

Although the PYxE insertion is quite prevalent among HIV-1C from Ethiopian and Eritrean individuals, it is not as prevalent in South African or Indian HIV-1C strains (16, 25). The 12-bp sequence encoding the PYxE motif is not present in non-subtype C M group HIV-1 sequences but is present within the Gag-p6 late domains of SIV_mac239, SIV_smE543, and their close relative HIV-2 (14, 26). As HIV-1C has a natural deletion of the YPx_nL motif and insertion of PYxE has replication advantages, we hypothesize that these strains have evolved through a recombination event of HIV-1C with either SIV or HIV-2. It is well described that the HIV-1 epidemic in Ethiopia geographically clusters very strongly (27), and the Ethiopian HIV-1C has been proposed to originate from either a single lineage or multiple descendants (28). However, to explain why the PYxE insertion is significantly more prevalent in HIV-1C from Ethiopia and Eritrea than from South African and Indian strains needs a further evolutionary study to prove this hypothesis.

The Gag-p6 late domain mediates viral budding through the PTAP motif, the YPx_nL motif, and the PYxE motif. The PTAP motif mediates binding of Gag to ESCRT protein TSG101 and is present in all HIV and SIV variants. Therefore, this domain appears to be necessary for viral budding. Although one report shows that PTAP duplication only increases HIV-1 replication in the presence of a PI (23), we and others observed increased viral growth when HIV-1 contained a PTAP duplication (22). Although deletion of the PTAP motif has not been identified in clinical strains so far, artificial deletion or mutation of this PTAP motif severely abrogates release of HIV-1 from infected cells (11, 29, 30). However, the presence of either one of the two other motifs that both facilitate binding to host cell protein ALIX still allows for virion release (12, 14, 19, 31, 32). This indicates that both TSG101- and ALIX-mediated pathways for viral budding are not dependent on each other, and each pathway is sufficient to mediate viral budding. The first ALIX-binding motif is YPx_nL and is present in most HIV-1 subtypes but not in HIV-1C or HIV-2 (13, 14). HIV-2 Gag can mediate binding to ALIX through a distinct PYKEVTEDL motif, and mutations in this motif can abrogate ALIX binding and decrease virion release (14, 26). Thus, the PYxE motif in a subgroup of HIV-1C strains is similar but shorter than the ALIX-binding motif found in HIV-2. We showed that this insertion of PYxE in HIV-1C Gag at the natural deletion site of the YPx_nL domain could reconstitute binding of HIV-1C Gag to ALIX. Hence, together with our observed PYxE insertion in HIV-1C, all HIV-1 subtypes have an ALIX-binding motif next to their PTAP motif within the Gag late domain.

Our data suggest that ubiquitin plays a role in facilitating the binding of ALIX to the PYKE motif. This is not surprising, as ubiquitin has been shown to be involved in viral budding of HIV. Although dispensable for virus budding, ubiquitination of Gag and the ESCRT proteins is essential for efficient release of HIV-1 from infected cells (21, 33, 34). Ubiquitination of the lysine of the PYKE motif could increase the binding between ALIX and Gag and thereby mediate more efficient viral budding. Indeed, coimmunoprecipitation of HIV-1C Gag-PYKE was more efficient than that of the Gag-PYQE variant. As the PYKE motif insertion seems to be the original insertion within HIV-1C Gag, ubiquitination of its lysine might not be essential, since other variants of the PYKE motif, i.e., PYQE and PYRE, are identified among HIV-1C strains along with the evolution of another lysine residue in the late domain. Because we have only a few clinical strains with a PYxE insertion, we could not identify differences in viral growth between HIV-1C Gag-PYKE and other PYxE variants. Nonetheless, whereas the other three residues seem to be required for the binding of ALIX to Gag, the lysine residue might provide increased stability for this interaction.

Although the ALIX-mediated viral budding mechanism appeared to be lost in HIV-1C, it is surprising to observe that HIV-1C PYxE strains were prevalent among HIV-1C-infected individuals. This suggests that the interaction of Gag with ALIX is of biological importance for the virus. In our limited cohort of HIV-1C-infected patients, PYxE insertion was always copresent with the PTAP motif. This suggests that the PYxE motif is not redundant from the PTAP motif-mediated viral budding process through TSG101 or a compensatory mechanism for loss of the PTAP motif, and HIV-1 requires both motifs for efficient viral replication. Importantly, viruses with HIV-1C Gag PYKE or PYQE insertion showed enhanced viral growth compared to viruses with HIV-1C wild-type Gag (Fig. 5). This finding is in line with a recent report that showed that a PYRE insertion within Gag enhances HIV-1C viral growth (19).

Decreased sensitivity toward PI LPV or NRTI TAF was found in several HIV-1C strains with a PTAP duplication or PYxE insertion. For the PTAP duplication, it has previously been reported that it could enhance HIV-1 growth in the presence of the PIs lopinavir and ritonavir (23). For the PYxE insertion, our earlier studies have shown that the PYxE insertion is more frequent in HIV-1C therapy failure patients (15, 16). Nonetheless, since our clinical HIV-1C PYxE strains did not have any known drug resistance mutations, it seems likely that the correlation of these viruses with advanced viral growth and reduced sensitivity to the PIs LPV and TAF could increase the risk for therapy failure in HIV-1C PYxE-infected patients. Mechanistic studies show that at the time of reverse transcription of HIV-1C from the RNA template, the reverse transcription favors pausing at the nucleotides in the HIV-1C K65 position (AAG), and this correlates with an increased probability for the development of tenofovir (TDF)/TAF mutation K65R (35, 36). However, no studies have shown any corelation with Gag mutation on the susceptibility of TAF. We are presently performing research to understand the role of Gag mutations on sensitivity for TAF, as both TDF and TAF are used globally.

In conclusion, our study showed that PYxE insertion in HIV-1C of patients originating from Ethiopia and Eritrea restored the interaction of Gag with ALIX, which is mediated in a ubiquitin-dependent or -independent way. Importantly, the insertion was positively correlated with the replication fitness that could affect the sensitivity against LPV in the absence of any PI drug resistance mutation. Based on the present study, our earlier clinical studies (15–17), and the study by Chaturbhuj et al. (19), we posit that a PYxE insertion in HIV-1 subtype C strains provides a replication advantage that can affect the susceptibility of the virus to certain antiretroviral drugs. As the PYxE insertion evolved within treatment failure patients infected with HIV-1C from both India and South Africa, HIV-1C_PYxEi also provides a replication advantage following treatment failure and following evolution of drug resistance mutations that may cost replication fitness. Most importantly, reduced sensitivity against TAF in the absence of any TAF mutation needs further studies to analyze any role of Gag mutations in reducing susceptibility to TAF. Altogether, it will be important to follow up on whether HIV-1C_PYxEi strains are emerging following the failure of antiretroviral therapy with LPV- and TAF-based regimens due to increased replication fitness. This information could provide important insights into the clinical significance of the PYxE insertion within HIV-1C Gag.

MATERIALS AND METHODS

Cell culture and plasmids.

Cells were cultured in 5% CO₂ at 37°C. HEK293T cells were maintained in Dulbecco’s modified Eagle medium (DMEM; Sigma, USA) supplemented with 10% fetal calf serum (Sigma, USA), 2 mM l-glutamine (Sigma, USA), 0.1 mM MEM nonessential amino acids (Gibco/Thermo Fisher Scientific, USA), and 10 U/ml penicillin combined with 10 μg/ml streptomycin (Sigma). TZM-bl reporter cells were also maintained in the medium mentioned above. This reporter cell line is a derivative of HeLa cells, but it stably expresses high levels of CD4 and CCR5 on the cell surface and has integrated copies of the luciferase gene under the control of the HIV-1 promoter that allows simple and quantitative analysis of HIV-1 infection. MT-4 cells were maintained in RPMI 1640 (Sigma, USA) medium supplemented with 10% fetal calf serum, 20 U/ml penicillin, and 20 μg/ml streptomycin. Cells were transfected using FuGene HD according to the manufacturer’s instructions (Promega, USA) in a 3:1 ratio with DNA.

The pEGFPN1 and mCherry plasmids were obtained from Addgene. The plasmid encoding hemagglutinin (HA)-tagged ubiquitin was a kind gift from Soham Gupta (Karolinska Institutet, Sweden). pCR3.1/HIV-Gag-mCherry (HIV-1B Gag) and pCR3.1-GFP-ALIX plasmids were kind gifts from Paul Bieniasz (The Rockefeller University). Codon-optimized HIV-1 Gag with a PYQE motif gene of East African HIV-1C viruses was cloned into the pCR3.1/HIV-Gag-mCherry plasmid that contained an mCherry protein C-terminal tag sequence. The Gag-PYQE-mCherry plasmid was modified by site-directed mutagenesis to make a base pair substitution mutation (C to A) that changes the PYQE motif to PYKE using the Q5 SDM kit (New England Biolabs, USA). Gag-PYQE-mCherry plasmid was linearized by PCR using 5′-AAGGAGCCTCTGACGAGCC-3′ as the forward primer (with the C-to-A base pair substitution underlined) and 5′-ATAGGGACCCTGGTCTTTCAGC-3′ as the reverse primer. Linear products were recircularized using DpnI, a polynucleotide kinase, and a ligase from the SDM kit by following the manufacturer’s protocol. DNA sequences were verified by Sanger sequencing.

Gag-p6 sequences and clinical specimens.

To identify the distribution of PYQE, PYRE, and PYKE motifs, we reanalyzed the HIV-1C Gag-p6 sequences that were collected from three cohorts, the Swedish InfCare Cohort (n = 140), the German Cohort (n = 127), and the Ethiopian Cohort (n = 73), as reported previously, with respect to naturally occurring polymorphisms in the PYxE motif (15). Additional HIV-1B (n = 2,754) and HIV-1C (n = 1,432) Gag-p6 sequences were collected from the Los Alamos HIV-1 Database. Stored plasma samples from therapy-naive patients infected with HIV-1C (n = 10) were randomly selected based on Gag-p6 sequences from the HIV cohort at Karolinska University Hospital, Stockholm, Sweden.

Recombinant virus production with patient-derived gag-pol.

The recombinant viruses were produced as described by us recently (37). Briefly, the gag-pol fragment (HXB2:0702-5798) was cloned into the pNL4-3 plasmid following digestion with BssHII and SalI (New England Biolab, USA) and ligation using T4 DNA ligase (New England Biolabs, USA). The chimeric viruses were produced by transient transfection of the plasmids into the 293T cell line using FuGene HD and harvested 72 h later by a collection of the cell-free supernatant cleared by centrifugation and stored in aliquots at −80°C.

In silico analysis: molecular modeling, docking, and ubiquitin binding motif prediction.

Homology modeling techniques generated the structure of HIV-1C Gag late domain with the PYKE insertion. The published crystal structures of the Gag late domain in complex with ALIX (PDB entries 2XS1 [14] and 2R02 [38]) were used as template molecules to model the HIV-1C Gag late domain. For this purpose, the Prime utility of Schrödinger Suite (Schrödinger Inc., USA) was used. The structure was subjected to restricted minimization using OPLS_2005 force field. The resulting structure of the HIV-1C Gag late domain structure was docked into the crystal structure of ALIX (PDB entry 2XS1) using the PIPER protein-protein docking program of BioLuminate Suite (Schrödinger Inc., USA) after deleting the SIV_mac239 PYKEVTEDL late domain structure. The resulting structure was further minimized for 1,000 iterations to remove steric clashes. A similar protocol was used to form the complex among ALIX, Gag late domain, and ubiquitin. The crystal structure of ubiquitin in complex with the TSG101-binding PTAP domain (PDB entry 1S1Q) was used after deleting TSG101 coordinates (39). The entire complex was then subjected to molecular dynamics simulations for 1,000,000 steps with a step size of 50 ps using an OPLS3 force field. The most energetically stable model of the complex was used for further analysis. To obtain energetically favored polar interactions, the conformational search of polar side chains at the interface of proteins was also conducted.

The prediction of ubiquitin binding with the HIV-1C Gag late domain with the PYKE motif P6 was performed by adding the amino acid sequence in the prediction tools presented at the website http://www.ubpred.org/.

Immunofluorescence.

HEK293T cells were cultured on poly-l-lysine-coated glass coverslips and cotransfected with GFP-tagged ALIX and mCherry-tagged codon-optimized Gag variants. At 24 h posttransfection, cells were washed twice with phosphate-buffered saline (PBS), fixed in 10% formalin (Sigma, USA) for 20 min at room temperature, washed three times with PBS, and stored in PBS at 4°C. Later, nuclear counterstaining with 4′,6-diamidino-2-phenylindole (DAPI) was performed, followed by mounting of the glass coverslips on glass slides. When HA-ubiquitin was also cotransfected, cells were fixed in ice-cold methanol for 15 min at −20°C, washed twice with PBS, and stored in PBS at 4°C. For ubiquitin detection, fixed cells were incubated with the anti-HA antibody (mouse monoclonal, clone 12CA5; Sigma, USA) for 1 h at room temperature, followed by goat-anti-mouse-Alexa Fluor 647-Plus-conjugated secondary antibody (Invitrogen, USA) for 1 h at room temperature before nuclear counterstaining with DAPI. Fluorescence was analyzed by confocal laser scanning microscopy using a Nikon single-point scanning confocal microscope with 60×/1.4-numeric-aperture oil objective (Nikon, Japan). Fluorescence intensity was measured (along drawn lines) using Fiji/ImageJ software (40). The highest fluorescence intensity value for each fluorophore along each line was set to 100%, and values were plotted as relative fluorescence intensity in percentage against distance in microns using GraphPad Prism v6 (GraphPad, Inc., USA).

MST.

The microscale thermophoresis (MST) experiments were conducted with a Monolith NT0.115 instrument (NanoTemper Technologies) at 40% MST and LED power at 25°C. Hexahistidine-tagged ALIX protein was labeled by nitrilotriacetic (NTA) dye (NanoTemper Technologies) using the manufacturer’s protocol. The peptides were synthesized at the Molecular Interaction Core (University of Missouri). Thermophoresis was induced by mixing peptides (0.001 to 1 μM) and fixed NTA-labeled ALIX (50 nM) in a buffer of 50 mM Tris-Cl, pH 7.8, 100 mM NaCl, and 0.1% pluronic-F127. The binding isotherms were obtained by plotting the difference in normalized fluorescence against increasing peptide concentration. The binding affinities were determined to fit the data points to a quadratic equation (equation 1) using nonlinear regression with MO Affinity software (NanoTemper Technologies), Prism (version 6.0; GraphPad, Inc.), or OriginLab (version 18; OriginLab Corp., Northampton, MA, USA).

$$F_b=\frac{\left(K_{d1}+\left[{\text{PEP}}_0\right]+\left[{\text{ALIX}}_0\right]\right)-\sqrt{{\left(K_{d1}+\left[{\text{PEP}}_0\right]+\left[{\text{ALIX}}_0\right]\right)}^2-4\left[{\text{ALIX}}_0\right]\left[{\text{PEP}}_0\right]}}{2\left[{\text{ALIX}}_0\right]}$$ (1)

Where F_b is the fraction of ALIX/peptide complex, K_d = [ALIX][PEP]/[ALIX-PEP], [ALIX] is the concentration of free ALIX, [ALIX₀] is the concentration of total ALIX, [PEP] is the concentration of free peptide, and [PEP₀] is the total concentration of peptide.

Coimmunoprecipitation assay.

HEK293T cells were cotransfected with GFP-tagged ALIX and codon-optimized Gag variants from HIV-1B or HIV-1C. At 24 h posttransfection, cells were collected and washed twice with ice-cold PBS, and immunoprecipitation of GFP-tagged proteins was performed using a GFP-Trap_A kit (Chromotek, Germany) according to the manufacturer’s protocol. Briefly, cells were lysed in ice-cold lysis buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.5% NP-40, supplemented with a protease cocktail inhibitor from Roche [Switzerland]) for 30 min on ice. Lysates were diluted in ice-cold wash buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 0.5 mM EDTA) supplemented with a protease cocktail inhibitor from Roche, input samples (10% of total) were saved for immunoblot analysis, and the remaining (90%) part of the diluted lysates was incubated with equilibrated GFP-Trap_A beads end over end for 1 h at 4°C. Beads were washed three times with ice-cold wash buffer before resuspending in 2× Leammli buffer (4× buffer diluted 1:1 with PBS; Invitrogen) and heated for 10 min at 95°C. The protein concentration of input samples was determined by DC protein assay (Bio-Rad, USA) according to the manufacturer’s microplate assay protocol. Equal amounts of protein for input samples and equal volumes of pulldown samples were subjected to immunoblotting using primary antibodies against GFP (rabbit monoclonal, clone EPR14104; Abcam, UK), Gag (rabbit polyclonal to HIV1 p55 + p24 + p17; ab63917; Abcam), and β-actin (rabbit polyclonal; ab8227; Abcam) and secondary horseradish peroxidase-conjugated antibodies (polyclonal goat anti-rabbit; Dako/Agilent, USA). Immunoblotted proteins were detected using the enhanced chemiluminescence detection system (Pierce/Thermo Fisher Scientific, USA) and Hyperfilms (Amersham/GE Healthcare Life Sciences, UK).

Viral growth kinetics assay in MT-4 cells.

First, the 50% tissue culture infective dose (TCID₅₀) for each propagated recombinant viral clone was determined in TZM-bl reporter cells. For each recombinant viral clone, 10-fold serial dilutions were prepared to range from 10² to 10⁷ in dilution factor from a single aliquot in medium containing DEAE (20 μg/μl). TZM-bl cells were plated in 96-well plates, and six replicates were incubated with the various virus dilutions for 48 h at 37°C in a 5% CO₂ humidified incubator. Virus infectivity was quantified by measuring Renilla luciferase activity (relative light units [RLU]) using a Bright-Glo Luciferase assay system (Promega, USA) on a Tecan microplate reader (Tecan Infinite 200 Pro; Tecan Group Ltd., Switzerland). The Spearman-Karber method was used to calculate the TCID₅₀ for each recombinant viral clone. MT-4 cells (2 × 10⁵) then were seeded in a 12-well plate and infected with each recombinant virus at a multiplicity of infection (MOI) of 0.05 PFU per cell in triplicates for each condition. Supernatants were collected 0, 3, 5, and 7 days postinfection, and HIV-1 p24 levels in supernatants were measured on the HIV COBAS 8000 platform (Roche Diagnostics, Switzerland). Viral growth kinetics (VGK) were analyzed as a relationship between the signal-to-cutoff ratio (i.e., electrochemiluminescence signal of the sample relative to calibrated negative samples for the HIV combi PT kit on the COBAS 8000 system) for each time point and replicate. Graphical representation of the result was made using GraphPad Prism, v6 (GraphPad Inc., USA).

Isolation of CD4⁺ T cells.

Primary CD4⁺ T cells were isolated from donor peripheral blood mononuclear cells (PBMCs). Briefly, 4.3 × 10⁸ PBMCs were obtained from 50 ml buffy coat using the standard method of gradient separation using Ficoll Histopaque reagent followed by centrifugation. Isolated PBMCs were used to isolate CD4⁺ T cells by negative selection using the EasySep human CD4⁺ T-cell isolation kit (Stem Cell Technologies, Canada). Briefly, PBMCs were pooled in 5 ml polystyrene round-bottom tubes at a concentration of 5 × 10⁷ cells and incubated with 50 μl of an antibody cocktail for 5 min at room temperature before mixing with RapidSphere beads followed by magnetic separation to yield pure CD4⁺ T cells.

Viral growth kinetics assay in CD4⁺ T cells.

CD4⁺ T cells were cultured using RPMI medium supplemented with 10% fetal calf serum and 1% penicillin and streptomycin. The cells were stimulated with phytohemagglutinin (20 μg/ml final concentration) for 3 days before starting the assay. To start the viral growth kinetics assay, 1 × 10⁶ CD4⁺ T cells were seeded in each well in a V-bottom 96-well culture plate and infected with viruses at an MOI of 0.05 PFU/cell in the presence of 20 μg/ml of DEAE. The plate was spinoculated at 37°C for 2 h at 800 rpm in a temperature-controlled centrifuge. After that, the plate was further incubated in a 5% CO₂ incubator for 8 h. Afterward, infected cells were transferred to a 48-well culture plate and washed six times before harvesting the initial supernatant from each corresponding well and designating it day 0. The experiment was then monitored for 7 days, and the supernatant was harvested on days 3, 5, and 7. HIV p24 was then measured in the supernatants using an HIV alliance p24 enzyme-linked immunosorbent assay kit (Pekin Elmer, USA), and results were analyzed using GraphPad Prism, v6 (GraphPad, Inc., USA).

DSA.

The following drugs were purchased from Selleckchem, USA: atazanavir sulfate (ATV), darunavir ethanoate (DRV), lopinavir (LPV), azidothymidine (AZT), tenofovir alafenamide (TAF), efavirenz (EFV), rilpivirine (RPV), etravirine (ETR), raltegravir (RAL), elvitegravir (EVG), dolutegravir (DTG), and cabotegravir (CAB). The drug sensitivity assay (DSA) was performed by determining the extent to which the antiretroviral drug inhibited the replication of the reference virus (pNL4-3) and wild-type, PTAPP-duplicated, and PyxE-derived recombinant viruses. One round was required for all drugs, as previously reported (37), except for the PI drugs, which required two rounds of infection. Briefly, for the two-round infection using PI DSA, TZM-bl cells (1 × 10⁴) were seeded in 96-well plates and cultured for 24 h. Drugs were serially diluted in culture media (ranging from 0.1 μM to 0.01 pM) and added in triplicate to the cells. The following day, viruses were added to each well at an MOI of 0.05 PFU/cell in the presence of DEAE (10 μg/ml). After 48 h, the viral supernatant from respective 96-wells plates with each drug was transferred to newly TZM-bl preseeded 96-wells plates without adding new drugs. After that, virus replication was quantified by measuring Renilla luciferase activity (in RLU) using the Bright-Glo Luciferase assay system (Promega, USA) 48 h after reinfection. Drug concentrations required for inhibiting virus replication by 50% (EC₅₀) were calculated by a dose-response curve using nonlinear regression analysis (GraphPad Prism, version 6.07; GraphPad Software, USA). The DSA experiments were performed with three technical replicates for each virus with the specified dynamic concentration range of the drug, and at least two independent analyses were performed. The reproducibility of the DSA was assessed by the 95% confidence interval obtained for the drug EC₅₀ and the degree of correlation between technical replicates. The output for the drug EC₅₀ results was used to compute the fold change value (FCV) for each virus relative to pNL4-3.

Ethical consideration.

The study was approved by regional ethics committees of Stockholm (2014/928–31 and 2005/1167–31/3). The patient information was anonymized and de-identified prior to analysis.