ABSTRACT
A highly conserved threonine near the C terminus of gp120 of human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV) was investigated for its contributions to envelope protein function and virion infectivity. When this highly conserved Thr residue was substituted with anything other than serine (the other amino acid that can accept O-glycosylation), the resulting virus was noninfectious. We found that this Thr was critical for the association of gp120 with the virion and that amino acid substitution increased the amount of dissociated gp120 in the cell culture supernatant. When HIV virions were generated in cells overexpressing polypeptide N-acetylgalactosaminyltransferase 1 (GalNAcT1), viral infectivity was increased 2.5-fold compared to that of virus produced in wild-type HEK293T cells; infectivity was increased 8-fold when the Thr499Ser mutant was used. These infectivity enhancements were not observed when GalNAcT3 was used. Using HEK293T knockout cell lines totally devoid of the ability to perform O-linked glycosylation, we demonstrated production of normal levels of virions and normal levels of infectivity in the complete absence of O-linked carbohydrate. Our data indicate that O-glycosylation is not necessary for the natural replication cycle of HIV and SIV. Nonetheless, it remains theoretically possible that the repertoire of GalNAc transferase isoforms in natural target cells for HIV and SIV in vivo could result in O-glycosylation of the threonine residue in question and that this could boost the infectivity of virions beyond the levels seen in the absence of such O-glycosylation.
IMPORTANCE Approximately 50% of the mass of the gp120 envelope glycoprotein of both HIV and SIV is N-linked carbohydrate. One of the contributions of this N-linked carbohydrate is to shield conserved peptide sequences from recognition by humoral immunity. This N-linked glycosylation is one of the reasons that primary isolates of HIV and SIV are so heavily resistant to antibody-mediated neutralization. Much less studied is any potential contribution from O-linked glycosylation. The literature on this topic to date is somewhat confusing and ambiguous. Our studies described in this report demonstrate unambiguously that O-linked glycosylation is not necessary for the natural replication cycle of HIV and SIV. However, the door is not totally closed because of the diversity of numerous GalNAc transferase enzymes that initiate O-linked carbohydrate attachment and the theoretical possibility that natural target cells for HIV and SIV in vivo could potentially complete such O-linked carbohydrate attachment to further increase infectivity.
INTRODUCTION
The envelope glycoproteins of human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV) are initially synthesized as gp160 precursors, which become proteolytically cleaved into the gp120 surface (SU) component and the gp41 transmembrane (TM) component. Upon engagement of the CD4 receptor and subsequent engagement of a CCR5 or CXCR4 coreceptor on the surface of a target cell, the viral envelope proteins undergo conformational changes which allow gp41 to gain access to the cell membrane (1, 2). These conformational changes push the hydrophobic fusogenic domain on gp41 into the target cell membrane to initiate fusion and entry. The hydrophobic fusogenic domain of HIV and SIV is located at the amino terminus of gp41. For an enveloped virus to remain infectious, the fusogenic component must be maintained in a spring-loaded configuration prior to receptor engagement (1, 3).
In both HIV and SIV, gp120 and gp41 are held together by noncovalent interactions. There are five highly conserved regions of the gp120 surface component (4). Studies have defined sequences in conserved region 1 (gp120-C1) near the N terminus and in conserved region 5 (C5) near the C terminus of gp120 as regions that interact with the gp41 transmembrane component (4, 5). Mutations in these regions can result in the loss of gp120 association with gp41 (6). The amino acid that is the subject of this report is located in the C5 region.
Similar to many cellular proteins, HIV envelope (Env) protein undergoes a complex set of posttranslational modifications before finally being presented on the surface of a cell. When such membrane proteins travel through the endoplasmic reticulum and the Golgi compartment, they may be glycosylated with N-linked and/or O-linked carbohydrate. N-linked carbohydrate is attached to proteins at an asparagine with the amino acid sequence N-X-S or N-X-T, where X may be any amino acid other than proline (7, 8). Unlike N-linked glycosylation, there is no consensus sequence for the placement of O-linked carbohydrate. There are no precise recognition sequences that distinguish an O-glycosylated site. Mucin-type O-glycosylation occurs on a serine (Ser) or threonine (Thr) and is more likely to occur in areas of a protein where a Thr and/or Ser residue is more prevalent. The Env proteins of HIV and SIV are among the most heavily glycosylated proteins in the mammalian cell database, with over 50% of its mass composed of N-linked carbohydrate. These glycans play a critical role in Env protein folding as well as in helping Env avoid host antibody recognition. N-glycans coat the surface of a trimeric envelope spike and impart a “glycan shield” that allows the underlying protein structure to avoid recognition by antibody (9–12). When the amino acid sequence of Env is mutated to disrupt multiple sites of N-glycan attachment, viral fitness is decreased in animal models due to decreased replicative capacity and to increased antibody recognition and neutralization (9, 13–15). Due to the importance of Env glycosylation in HIV biology, much effort has been spent on classifying the glycosylation profile of HIV Env (16–23). Since Env is the target of antibodies that can neutralize viral infectivity, detailed knowledge of carbohydrate location and composition in the three-dimensional structure can help inform immunogen design for vaccine purposes.
SIVmac, SIVsmm, and HIV-2 isolates contain runs of serines and threonines in the V1 region of gp120, and O-linked glycosylation occurs at these sites (16, 24). Although O-glycosylation is not observed in the V1 region of HIV-1 gp120, a particular Thr residue near the C terminus of gp120 (consensus amino acid 499; 497 in the gp120 sequence of the NL4-3 strain) and of SIVmac (amino acid 510 in both the consensus and SIVmac239 gp120 sequences) is predicted by the NetOGlyc, version 4.0, software to be O-glycosylated (23, 25). When gp120 is made in monomeric form as the truncated, secreted product, this Thr residue is consistently O-glycosylated (19, 23). However, virions purified from productive infection of cultured cells lack O-glycosylation at this site (23). This was true for both HIV-1 and SIV virions. More recently, Behrens et al. have found that this site is also essentially devoid of O-glycosylation on authentic HIV-1 SOSIP trimers (∼0.2%) (20). Thus, any role for this Thr and its potential for glycosylation remain uncertain.
In this report, we describe a variety of approaches to investigate the contributions of this C-terminal Thr residue and its potential O-glycosylation. We demonstrate that this Thr residue is critical for the infectivity of both HIV-1 and SIVmac. Our results are consistent with a role for this Thr residue in maintaining a stable, prefusion configuration of the Env complex for both HIV-1 and SIV. However, production in cells that have been engineered to be totally devoid of the ability to O-glycosylate results in no loss in yield and no loss in inherent infectivity.
RESULTS
The highly conserved C-terminal Thr of HIV-1 gp120.A Thr residue near the C terminus of HIV-1 gp120 (consensus amino acid 449; amino acid 497 in the NL4-3 gp120 sequence) is predicted by the NetOGlyc, version 4.0, prediction software (http://www.cbs.dtu.dk/services/NetOGlyc/) to be a potential site of attachment for O-linked carbohydrate (23) (Fig. 1A). To determine the extent to which this Thr is conserved, we aligned the 4,556 available HIV-1/SIVcpz sequences in the Los Alamos HIV database to amino acid positions 491 to 505 of the NL4-3 gp120 sequence. The alignment was made from genome sequences available in 2017.
Predicted glycosylation site and conservation of NL4-3 Thr497. (A) Graphical representation of Thr499 site of O-glycosylation in relation to Env subunits. The colored structure attached to Thr499 represents a disialyated core 1 carbohydrate. The site of proteolytic cleavage of the Env gp160 precursor by furin protease into the gp41 transmembrane component and the gp120 surface component is indicated. (B) NL4-3 gp120 amino acid sequence was analyzed using the NetOGlyc, version 4.0, O-glycosylation prediction tool. The Thr highlighted in red is amino acid 497. This Thr corresponds to 499 in the consensus numbering system. Thr497 had a prediction confidence score of 0.5757. Of note, two other sites of O-glycosylation were predicted (Ser144 and Ser396). However, from mass spectrometry analysis, we know that these are not sites of O-glycosylation when gp120 is made as a monomeric secreted protein. (C) HIV-1/SIVcpz Env amino acid sequences from the Los Alamos HIV Sequence Database were aligned from amino acid 491 to 505 to analyze the conserved nature of Thr497. In total 4,556 amino acid sequences were compared. The larger the letter is, the more conserved is the amino acid. Amino acid usage for each location is displayed as a percentage in the table.
This C-terminal Thr was found to be highly conserved with 95.46% of sequences having Thr at the position corresponding to consensus position 499. Of the remaining 4.54% of other amino acids at this position, serine predominated. The high degree of amino acid conservation at this location, the lack of tolerance for amino acids other than Thr or Ser, and the prediction for O-glycosylation are all consistent with some potential role for O-glycosylation of this site.
Importance of the highly conserved C-terminal Thr of gp120 for HIV and SIV infectivity.Substitution of the highly conserved C-terminal Thr497 of HIV-1 NL4-3 gp120 with alanine (A), glutamic acid (E), asparagine (N), valine (V), or tyrosine (Y) resulted in virus that was not detectably infectious in viral infectivity assays (Fig. 2A). Complete loss of infectivity was also observed for SIVmac239 T510A and T510V strains (Fig. 2B). Substitution of HIV-1 NL4-3 Thr497 with Ser resulted in virus that was 100-fold less infectious than the parental strain (Fig. 2A). In the case of SIVmac239 with a substitution of Ser for Thr, infectivity was similar to that of the parental strain (Fig. 2B).
Infectivity of HIV and SIV point mutants. Amino acids 497 of HIV NL4-3 and 510 of SIVmac239 were mutated in full-length proviral vectors to disrupt the potential site of O-glycosylation. (A) The highly conserved threonine at position 497 or 510 (HIV consensus 499 and SIV consensus 510 indicated by the red highlighting) was mutated as indicated on the legends of the graphs in panels B and C. Viral stocks were produced by transient transfection of HEK293T cells and normalized by p24 and p27 antigen capture assays. (B) HIV NL4-3 infectivity was assessed by TZM-bl assay starting at 12 ng of p24/ml followed by 2-fold serial dilutions. (C) SIVmac239 infectivity was assessed by SEAP assay starting at 12 ng of p27/ml followed by 2-fold serial dilutions.
The highly conserved C-terminal Thr is essential for gp120 association with the virion.To examine the Env content in virions of HIV-1 NL4-3, SIVmac239, and our viral variants, viral stocks were made in HEK293T cells. Cell culture supernatant was filtered, and virus was pelleted and normalized for p24 and p27 content on the basis of antigen capture assays prior to Western blot analyses. For the parental HIV-1 NL4-3, gp120 and gp41 were readily detected in the virus pellet (Fig. 3A). The content of gp120 and gp41 in pelleted virions was greatly diminished for NL4-3 T497S compared to that of the parental NL4-3 (Fig. 3A). NL4-3 T497A, NL4-3 T497E, NL4-3 T497N, and NL4-3 T497V lacked detectable gp120 in pelleted virions, and these variants also contained decreased amounts of gp41 compared to amounts in both the parental NL4-3 and to NL4-3 T497S strains (Fig. 3A). When equal volumes of cell culture supernatant from which virus was removed were analyzed, an increase of shed gp120 was detected compared to that of the parental NL4-3 (Fig. 3B). For SIVmac239, gp120 and gp41 were readily detected in the virus pellet (Fig. 3C). The SIVmac239 T510S variant incorporated similar levels of gp41 into pelleted virions while the levels of gp120 were considerably less than those of SIVmac239 (Fig. 3C). SIVmac239 T510A and SIVmac239 T510V did not have detectable gp120 in the virus pellet (Fig. 3C). For SIVmac239, SIVmac239 T510S, SIVmac239 T510A, and SIVmac239 T510V, the gp120 subunit of Env was readily detected in the virus-free cell culture supernatant (Fig. 3D). The amount of gp120 detected in the supernatant for SIVmac239 T510S, SIVmac239 T510A, and SIVmac239 T510V was greater than that detected for SIVmac239 (Fig. 3D).
The C-terminal Thr residues of HIV-1 gp120 (Thr499) and SIVmac239 (Thr510) are optimal for the association of gp120 with the virion. (A) Western blots of HIV-1 NL4-3 and variants from pelleted virions. (B) Western blot of cell culture supernatant gp120 of HIV-1 NL4-3 and variant viruses from which pelleted virions had been removed. (C) Western blots of SIVmac239 and variants from pelleted virions. (D) Western blots of cell culture supernatant gp120 of SIVmac239 and variant viruses. Proteins loaded onto SDS-PAGE gels in the experiments shown in panels A and C were normalized to the amount of p24 or p27 as determined by antigen capture. (E) Integrity of the C-terminal domain of HIV-1 NL4-3 gp120 was analyzed by Western blotting of parental and variant HIV-1 NL4-3 gp120 secreted into the cell supernatant. Protein was probed with a mouse (ms) pan-HIV-1 antibody to gp120 and two antibodies (1331A and 858D) that recognize the amino acids between the secondary and primary furin cleavage sites at the C terminus of HIV-1 gp120 (36, 37). ab, antibody.
Substitution for the highly conserved Thr close to the furin cleavage sites does not alter gp160 cleavage.The highly conserved Thr is 13 amino acids away from the primary furin cleavage site and 5 amino acids away from a secondary furin cleavage site of HIV-1 gp160. The proximity of the highly conserved Thr to these sites together with reports of an O-linked carbohydrate regulating protein cleavage (26, 27) led us to investigate whether the loss of infectivity of variant virus might be due to altered cleavage of gp160. We found in all cases that gp160 was processed into gp120 and gp41 since gp120 was readily detected in the cell culture supernatant (Fig. 3B, D, and E). Monoclonal antibodies 1331A and 858-D that bind to the C terminus of HIV-1 gp120 confirmed the presence of the appropriate C-terminal domain and the use of the primary furin cleavage site in all cases (Fig. 3E).
Overexpression of the polypeptide GalNAcT1 increases viral infectivity and virion-associated gp120.Polypeptide N-acetylgalactosaminyltransferase (GalNAcT) enzymes modify target Thr or Ser by attaching N-acetylgalactosamine (GalNAc) in the initial step of the attachment of a mucin-type O-linked carbohydrate. Twenty distinct polypeptide N-acetylgalactosaminyltransferases have been reported in humans (28). We overexpressed either the polypeptide GalNAc transferase 1 (GalNAcT1) or the polypeptide GalNAc transferase 3 (GalNAcT3) by cotransfection during viral stock production. In a TZM-bl infectivity assay, the p24-normalized infectivity of virus produced from cells overexpressing GalNAcT1 was 2.5-fold greater than that of the HIV-1 NL4-3 produced in parallel in the absence of GalNAcT1 provided in trans (Fig. 4A). For the HIV-1 NL4-3 T497S variant, infectivity of virus produced from cells overexpressing GalNAcT1 was increased 8-fold in a GalNAcT1 dose-dependent fashion compared to that of virus produced in parallel in the absence of GalNAcT1 provided in trans (Fig. 4B). The NL4-3 T497A variant remained noninfectious even when virus was produced from cells overexpressing GalNAcT1 (Fig. 4B). The increased infectivity observed by overexpression of GalNAcT1 was specific for this transferase. When virus was produced from HEK293T cells overexpressing GalNAcT3, no increase in infectivity of NL4-3, NL4-3 T497S, and NL4-3 T497A was observed (Fig. 4C and D).
Infectivity of virus from cells overexpressing GalNAcT1 and GalNAcT3 enzymes. (A) Infectivity of HIV-1 NL4-3 produced from HEK293T cells and HEK293T cells overexpressing GalNAcT1. Virus stocks were produced by transient transfection of HEK293T cells. Cells were transfected with 5 μg of proviral DNA plus 10 μg of empty pCMV control vector (NL4-3), 5 μg of proviral DNA plus 5 μg of GalNAcT1 in the pCMV vector (NL4-3 + GalNAcT1 5 μg), or 5 μg of proviral DNA plus 10 μg GalNAcT1 in the pCMV vector (NL4-3 + GalNAcT1 10 μg). Virus containing normalized amounts of p24 was used to infect TZM-bl cells, which contain a stably integrated Tat-inducible luciferase reporter gene. Viral infectivity is directly correlated to the amount of luciferase produced within the cell. (B) Infectivity of HIV-1 NL4-3 T497S and HIV-1 NL4-3 T497A produced from HEK293T cells and HEK293T cells overexpressing GalNAcT1. (C) Infectivity of HIV-1 NL4-3 produced from HEK293T cells and HEK293T cells overexpressing GalNAcT3. (D) Infectivity of HIV-1 NL4-3 T497S and HIV-1 NL4-3 T497A produced from HEK293T cells and HEK293T cells overexpressing GalNAcT3. (E) Virion gp120 and p24 of HIV-1 NL4-3 produced from HEK293T cells, HEK293T cells overexpressing GalNAcT1, and HEK293T cells overexpressing GalNAcT3. Virus was pelleted from cell culture supernatant. Proteins loaded on the SDS-PAGE gel were normalized to the amount of p24 as determined by antigen capture assay. HIV-1 gp120 was probed with the mouse anti-HIV-1MN gp120 (0085-P3F5-D5-F8) hybridoma supernatant. HIV-1 p24 was probed with a commercially available antibody. (F) Virion gp120 and p24 of an HIV-1 NL4-3 T497S variant produced from HEK293T cells, HEK293T cells overexpressing GalNAcT1, and HEK293T cells overexpressing GalNAcT3.
To examine the Env content of HIV-1 NL4-3 and NL4-3 T497S produced from HEK293T cells, HEK293T cells overexpressing GalNAcT1, and HEK293T cells overexpressing GalNAcT3, viral stocks were made by transient transfection of proviral DNA. Seventy-two hours after transfection, cell culture supernatant was filtered, and virus was pelleted. Pelleted virions were normalized to HIV-1 p24 for Western blot analyses. Virion-associated gp120 increased for HIV-1 NL4-3 produced from cells overexpressing GalNAcT1 compared to that of virus produced from HEK293T cells and HEK293T cells overexpressing GalNAcT3 (Fig. 4E). gp120 for HIV-1 T497S was markedly increased when virus was produced from cells overexpressing GalNAcT1 compared to the level in virus from HEK293T cells or HEK293T cells overexpressing GalNAcT3 (Fig. 4F).
HIV-1 NL4-3 remains fully infectious in the absence of O-linked glycosylation.To determine whether O-glycosylation is critical for HIV infectivity, we produced HIV-1 NL4-3 viral stocks devoid of all O-glycans. This was accomplished using HEK293T cell lines previously generated by our laboratory that are genetically modified by CRISPR/Cas9 to knock out enzymes of the Leloir pathway of galactose metabolism (29). These enzymes are needed to generate the UDP-sugar precursors necessary for N- and O-linked glycosylation. In the Leloir pathway, UDP-galactose-4-epimerase (GALE) reversibly converts UDP-glucose and UDP-N-acetylglucosamine (UDP-GlcNAc) into UDP-galactose and UDP-N-acetylgalactosamine (UDP-GalNAc), respectively (30, 31). These interconversions are the primary way in which a cell generates UDP-galactose and UDP-GalNAc for glycosylation. However, these UDP precursors are also available via salvage pathways. Salvaged galactose and GalNAc are converted to UDP-sugar precursors through the addition of multiple phosphate groups, the first phosphate addition being catalyzed by galactokinase-1 (GALK1) and galactokinase-2 (GALK2), respectively. By generating knockouts (KOs) in the GALE, GALE plus GALK1, and GALE plus GALK2 genes, we eliminated the cell lines' ability to O-glycosylate. Cellular proteins in these HEK293T knockout cell lines were found to be devoid of all O-linked carbohydrates by mass spectrometry (29). Also, gp120 made as the monomeric secreted protein in these knockout cell lines was found to be devoid of O-linked carbohydrate (29). To determine the importance of O-glycosylation for NL4-3, virions were packaged in these cell lines to yield virus devoid of all O-linked glycans.
NL4-3 virions were packaged in a GALE knockout (GALE KO) cell line. Because these cells are defective in GALE, they cannot synthesize UDP-galactose or UDP-GalNAc. However, the GALE KO cell line still has functional salvage pathways. By supplementing the culture medium with galactose and GalNAc (indicated on the figures as sugars), the defect observed in O-glycosylation in this cell line can be reversed. To control for potential salvage of glycoproteins and sugar precursors from the fetal bovine serum (FBS) of the cell culture medium, cells were also grown in lipoprotein-depleted fetal bovine serum (LDFBS), which is depleted of salvageable glycoproteins. Virus containing normalized amounts of p24 produced from HEK293T cells and GALE KO cells was used to infect TZM-bl cells. Virus produced under all conditions was equally infectious (Fig. 5A).
NL4-3 retains infectivity in the absence of O-glycosylation. HIV NL4-3 viral stocks were produced by transient transfection of proviral DNA into HEK293T cells or glycosylation-defective cell lines as indicated in the legends on the graphs. Viral stocks were normalized by p24 antigen capture assay. Infectivity was assessed by TZM-bl assay starting from 10 to 40 ng of p24/ml followed by 2-fold serial dilutions. (A) NL4-3 was produced by transient transfection of HEK293T and GALE KO cells. Cells were grown in either 10% FBS or 3% LDFBS or supplemented with galactose and GalNAc (+Sugars), as indicated. (B) NL4-3 was produced by transient transfection of HEK293T, GALK1 KO, and GALE+GALK1 KO cells. All cells were grown with the addition of 10% FBS to cell culture medium. (C) NL4-3 was produced by transient transfection of HEK293T, GALK2 KO, and GALE+GALK2 KO cells. All cells were grown in cell culture medium containing 10% FBS. Where indicated, cell culture medium was supplemented with galactose and GalNAc (+Sugars) or with UDP-galactose and UDP-GalNAc (+UDP).
To further control for the potential salvage from serum glycoproteins, NL4-3 viral stocks were produced in two additional glycosylation-defective cell lines with strategic knockouts in enzymes of the salvage pathway. The GALE+GALK1 KO cell line is defective in the synthesis of UDP-galactose and UDP-GalNAc as well as the salvage of galactose. The GALE+GALK2 KO cell line is defective in the synthesis of UDP-galactose and UDP-GalNAc as well as the salvage of GalNAc. Both of these cell lines have been well characterized and shown to be completely devoid of O-glycosylation regardless of the medium conditions utilized (29).
Viral stocks were produced in HEK293T cells and GALE+GALK1 KO cells and were normalized by p24 content before infection of TZM-bl cells. Similar to the findings in the GALE KO cells, virus produced in GALE+GALK1 cells was equally infectious as the control HEK293T viral stocks (Fig. 5B).
Similar to the findings with the GALE+GALK1 KO cell lines, when viral stocks were produced in GALE+GALK2 KO cell lines, all viral stocks exhibited similar infectivities (Fig. 5C).
SIVmac239 remains fully infectious in the absence of O-linked glycosylation.Since SIVmac239 infectivity is also sensitive to mutation of the C-terminal threonine of gp120, we wanted to determine if the ability to maintain infectivity in the absence of O-glycosylation is unique to HIV-1 NL4-3. When SIVmac239 viral stocks were produced in HEK293T, GALE KO, and GALE+GALK2 cells, we observed findings similar to our NL4-3 data. SIVmac239 viral stocks were equally infectious when produced in both the GALE KO and GALE+GALK2 cell lines compared to infectivity in HEK293T viral stocks (Fig. 6A). When SIVmac239 viral stocks were produced in GALK1 KO and GALE+GALK1 KO cell lines, these viral stocks were also equally infectious as virus produced in HEK293T cells (Fig. 6B). These data support the conclusion that both the HIV-1 NL4-3 and SIVmac239 virus production from HEK293T cells and their infectivities are independent of any and all O-glycosylation.
SIVmac239 retains infectivity in the absence of O-glycosylation. HIV SIVmac239 viral stocks were produced by transient transfection of proviral DNA in HEK293T cells or glycosylation-defective cell lines as indicated in the legends on the graphs. Viral stocks were normalized by p27 antigen capture assay. Infectivity was assessed by TZM-bl assay starting with 40 ng of p24/ml followed by 2-fold serial dilutions. (A) SIVmac239 was produced by transient transfection of HEK293T, GALE KO, GALK2 KO, and GALE+GALK2 KO cells. Cell cultures were supplemented with galactose and GalNAc (+Sugars) where indicated. All cells were grown in the presence of 10% FBS. (B) SIVmac239 was produced by transient transfection of HEK293T, GALK1 KO, and GALE+GALK1 KO cells. All cells were grown with the addition of 10% FBS to cell culture medium.
Full infectivity in the absence of complex N-glycans.HIV-1 NL4-3 and SIVmac239 viral stocks were produced in GALE+GALK1 cells to analyze the impact of N-glycans on HIV/SIV infectivity. The GALE+GALK1 cell line is distinct from the other glycosylation-defective cell lines due to its inability to make complex N-glycans as well as O-glycans. This cell line is documented to produce only high-mannose N-glycans (29). Because GALE+GALK1 cells cannot make complex N-linked glycans, virions produced in these cells will contain only high-mannose N-glycans. NL4-3 and SIVmac239 viral stocks were equally infectious when produced in GALE+GALK1 cell lines as viral stocks produced in HEK293T cells (Fig. 5B and 6B). These data suggest that both NL4-3 and SIVmac239 are fully infectious with only high-mannose N-glycans. This finding is consistent with previous publications in which the formation of complex N-linked glycans on HIV virions was inhibited by other means (32). These data suggest that both HIV-1 NL4-3 and SIVmac239 do not require complex N-glycans for viral production or infectivity.
DISCUSSION
Thr at Env position 499 in the consensus numbering system of HIV-1 is highly conserved across all clades of HIV-1. Our results demonstrate the functional importance of this highly conserved Thr for the stabilization of gp120 on the surface of HIV-1 and SIVmac239 virions and for viral infectivity. Variants of HIV-1 NL4-3 and SIVmac239 were constructed in which this highly conserved Thr was substituted with a variety of other amino acids. With the exception of a Thr-to-Ser mutation, other amino acids were not tolerated at the position in question; gp120 with mutations at this position failed to remain properly associated with gp41 and was shed into the culture supernatant. It has been suggested that failure of gp120 to remain associated with gp41 can result in triggering of the spring-loaded complex (1, 3) and increased endocytosis and turnover of gp41 (4). Not only was the amount of mutant gp120 on virions dramatically reduced, but a reduced amount of gp41 was also observed on virions with an altered C-terminal threonine (Fig. 3). Our findings demonstrate a critical role for this C-terminal threonine residue.
O-glycosylation of the C-terminal threonine on HIV-1 gp120 when made as the truncated secreted product has been noted by several groups, including our own (19, 23, 33). Due to these findings and the fact that the C-terminal threonine is predicted to be O-glycosylated by the NetOGlyc, version 4.0, software, we chose to explore the possible contributions of O-glycosylation of the C-terminal threonine to the function of Env and infectivity of virions. When either HIV-1 NL4-3 proviral DNA or the NL4-3 serine mutant was transfected together with a GalNAc transferase 1 expression cassette, the infectivity of virions and the gp120 content of virions were increased substantially (Fig. 4). However, in contrast to these findings, when HIV-1 Env from in vitro-produced virions was examined, the C-terminal threonine from these virions was devoid of O-glycosylation (23). To help discriminate any role of O-glycosylation from the necessity of having a Thr at position 499, we turned to glycosylation-defective cell lines developed from HEK293T cells (29). When HIV and SIV virions were packaged in O-glycosylation-defective cell lines, we observed similar infectivities from virions produced in all cell lines (Fig. 5 and 6). Our findings indicate that O-glycosylation of the gp120 C-terminal threonine, as well as that of all other viral and cellular proteins, is not critical for virion production, for HIV/SIV infectivity, or for the association of gp120 with the trimer spike. Thus, the critical determinant is the presence of a threonine at position 499 and not that of any O-glycans. This idea is supported by the highly conserved nature of this Thr (Fig. 1B). Also, due to the inability of the GALE+GALK1 cell lines to make complex N-linked glycans (29), our data also demonstrate that NL4-3 is equally infectious with only high-mannose N-glycans. The latter finding is consistent with previous publications in which the formation of complex N-linked glycans on HIV virions was inhibited by other means (32).
Although the results presented here demonstrate that O-glycosylation is not critical for HIV/SIV infectivity, it remains theoretically possible that the repertoire of GalNAc transferase isoforms in natural target cells for HIV and SIV in vivo could result in O-glycosylation of the threonine residue in question and that this could boost the infectivity of virions beyond the levels seen in the absence of such O-glycosylation. From our data, it is apparent that overexpression of GalNAcT1 does result in higher infectivity and greater association of gp120 with gp41 on virions. But any role that this or another glycosyltransferase may be playing remains unclear. What is clear from the results presented here is that there is a remarkable conservation of C-terminally located Thr across both HIV and SIV. For both HIV and SIV, this highly conserved Thr is critical for maintaining stable association of gp120 with gp41 and for maintaining the stable prefusion configuration of the viral envelope complex prior to receptor engagement. Importantly, high levels of infectivity can be maintained in the complete absence of O-glycosylation. However, more work will be needed to determine whether O-glycosylation of this Thr residue may occur in natural target cells and whether such O-glycosylation can serve to boost infectivity to even higher levels.
MATERIALS AND METHODS
Plasmids and cell lines.HEK293T cells were obtained from the ATCC. TZM-bl cells were obtained from the NIH AIDS Reagent Program and were propagated as recommended. The generation and validation the GALE KO, GALK1 KO, GALK2 KO, GALE+GALK1 KO, and GALE+GALK2 KO cell lines have been recently described (29). The parental infectious HIV-1 pNL4-3 clone was originally obtained from the NIH AIDS Reagent Program. The parental infectious SIVmac239 construct has been previously described (34). Endotoxin-free transfection-grade plasmid was purified using a ZymoPURE Plasmid Maxiprep kit (Zymo Research).
Construction of variant proviral constructs.For proviral NL4-3 and SIVmac239, standard PCR methods were utilized to change the codon that encodes Thr to GCC (alanine), GAG (glutamic acid), GAT (aspartic acid), AAT (asparagine), TCC (serine), GTC (valine), or TAC (tyrosine) in the variant constructs. Following mutagenesis, proviral plasmid DNA was sequenced in full to verify the absence of off-site changes.
Viral stocks.Viral stocks were produced in HEK293T, GALE KO, GALK1 KO, GALK2 KO, GALE+GALK1 KO, and GALE+GALK2 KO cell lines as indicated in the text, figures, and figure legends. There was no differential effect on yield for the viral stocks produced in different cell lines as measured by antigen capture (HIV-1 capsid p24/SIV capsid p27). For virus stocks produced with overexpression of GalNAc transferases, 5 μg of proviral DNA plus 10 μg of the empty vector, 5 μg of proviral DNA plus 5 μg of GalNAcT plasmid, or 5 μg of proviral DNA plus 10 μg of GalNAcT plasmid was transfected into HEK293T cells. Culture medium was changed 24 h prior to harvest of virus.
For viral stock preparation in glycosylation knockout cell lines, cells were grown in 10% FBS, 3% lipoprotein-depleted fetal bovine serum (LDFBS) (Kalen Biomedical), or 3% LDFBS with 10 μM galactose and 100 μM GalNAc (indicated on the figures as sugars) for 4 days prior to transfection with proviral DNA. Cells were maintained under the culture conditions indicated in the figures and figure legends for the duration of viral stock preparation.
Infectivity assays.The secreted alkaline phosphatase (SEAP) infectivity assay for measurement of SIV infectivity has been previously described (35). TZM-bl cells were also used to measure SIV and HIV infectivity where indicated in the text and figure legends. A total of 10,000 TZM-bl cells were seeded per well in a 96-well plate. Cells were infected with normalized amounts of p24 or p27 as determined by antigen capture assay (Advanced Bioscience Laboratories, Inc., Kensington, MD). Three days after infection, luciferase activity was measured using a Britelite Plus Reporter Gene Assay System (PerkinElmer) according to the manufacturer's recommendation.
Western blotting.HEK293T cells were transfected with proviral DNA constructs using Fugene 6 (Roche) according to the manufacturer's recommendations. Forty-eight hours after transfection, cell lysates and cell culture supernatant were collected. To collect cell lysates, the cell monolayer was washed with 10 ml of cold (4°C) phosphate-buffered saline (PBS) and then disrupted with NP-40 lysis buffer (50 mM Tris HCl, 150 mM NaCl, and 1% NP-40 at pH 7.4 with complete protease inhibitor cocktail [Roche]) on ice for 10 min. Nuclear debris was removed by centrifugation for 5 min at 8,000 × g. Cell culture supernatant was removed and filtered through a 0.45-μm-pore-size polyethersulfone membrane (Nalgene). The cell-free culture supernatant was further processed into virion pellet and soluble supernatant fraction by centrifugation at maximum g force in a Heraeus Biofuge 13 for 1 h in a 4°C cold room in a biocontainment cabinet. Following centrifugation, the supernatant was collected as the soluble fraction. Pelleted virions were washed by gentle resuspension in 1 ml of PBS at 4°C and then spun at maximum g force in a Heraeus Biofuge 13 for 1 h in a 4°C cold room in a biocontainment cabinet. Pellets were washed twice and then resuspended in PBS. Antigen capture assays were conducted on the pelleted fraction (Advanced Biosciences Laboratory). Proteins normalized to p24/p27 content determined by antigen capture were separated by SDS-PAGE and transferred to polyvinylidene fluoride membranes (Whatman). Membranes were blocked for 1 h at room temperature with membrane blocking solution (Invitrogen). Membranes were probed with the following primary antibodies: mouse anti-gp120 (pan-HIV-1; Immune Technology) or mouse anti-HIV-1MN gp120 hybridoma supernatant (1:100 dilution; AIDS and Cancer Virus Program, Frederick National Laboratory), human anti-gp41 (HIV-1 clone 7B2; James Robinson), mouse anti-p24 (HIV-1 clone 3G8; Immune Technology), mouse anti-gp120 (SIVmac239 clone 36-E2; Immune Technology), mouse anti-gp41 (SIVmac239 clone KK41; NIH AIDS Repository), and the appropriate horseradish peroxidase (HRP)-conjugated species-specific secondary antibodies. Specific signals were detected by an enhanced chemiluminescence system using a SuperSignal West Pico chemiluminescent substrate (Pierce). All images were acquired using a Fujifilm LAS-3000 instrument.
ACKNOWLEDGMENTS
We thank Leydi Guzman for all her administrative assistance during this project. We also thank Susan Zolla-Pazner for providing monoclonal antibodies (1331A and 858-D) that recognize the C terminus of HIV-1 gp120. James Robinson kindly provided the HIV-1 gp41 monoclonal antibody (7B2). TZM-bl cells and pNL4-3 proviral DNA were obtained from the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH.
J.M.T., E.S.C., and R.C.D. were responsible for experimental design. J.M.T., E.S.C., and Z.A.S. generated all experimental data. J.M.T., E.S.C., and R.C.D. were in charge of manuscript preparation. A.D. and S.M.H. assisted in data interpretation and manuscript preparation.
This work was supported by the National Institutes of Health R01-AI025328 (E.S.C. and R.C.D.), R01-AI104523 (E.S.C. and R.C.D.), P30-AI060354 (E.S.C.), and P51-RR000168 (E.S.C. and R.C.D.), State of Florida contract CODMR (J.M.T.), Biotechnology and Biological Sciences Research Council grant BBF0083091 and grant BBK0161641 (A.D. and S.M.H.), and a Wellcome Trust grant (082098 to A.D.).
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
FOOTNOTES
- Received 17 July 2017.
- Accepted 18 July 2017.
- Accepted manuscript posted online 26 July 2017.
- Copyright © 2017 American Society for Microbiology.
