Rapid Progression to Simian AIDS Can Be Accompanied by Selection of CD4-Independent gp120 Variants with Impaired Ability To Bind CD4

Abstract

Aspartate 368 on human immunodeficiency virus type 1 (HIV-1) gp120 forms multiple contacts with CD4; in mutagenesis studies, its replacement by asparagine and corresponding changes in simian immunodeficiency virus SIVmac (D385N) reduced binding with CD4. Nevertheless, simian immunodeficiency virus envelopes with D385N were prevalent in several studies. Extending these observations, we also found D385N to be dominant among env clones from two rhesus macaques that progressed rapidly to simian AIDS. These envelopes showed a CD4-independent phenotype as well as reduced affinity to CD4. Moreover, an adjacent change, G383R, which was frequently coselected with D385N, further decreased binding. An optical biosensor study demonstrated that the SIVmac239 gp120 bound to CD4 with kinetics similar to those of HIV-1. However, the gp120s with D385N and G383R showed a 40-fold reduction in affinity, with a drastic increase in dissociation rate, indicating an inherently unstable complex. This finding showed that rapid progression to simian AIDS may be accompanied by the selection of CD4-independent gp120 variants with impaired CD4 binding ability.

The attachment of human immunodeficiency virus (HIV) and its simian counterpart, simian immunodeficiency virus (SIV), to susceptible cells involves complex interactions between the viral surface glycoprotein, gp120, and two cell membrane proteins, CD4 (6, 16, 27, 29) and one of a family of chemokine receptors (5, 24, 28, 47). Binding between gp120 and CD4 is thought to induce conformational changes in gp120 leading to the formation or exposure of a chemokine receptor binding site, which is either hidden or disordered prior to CD4 contact (35, 38, 39, 42). Binding between gp120 and the chemokine receptor in turn results in additional changes in the gp120/gp41 structure, with consequent release of the fusion peptide and fusion between the viral and cell membranes (3, 44, 45).

Most HIV and SIV strains utilize this dual-receptor strategy for cell entry. Nevertheless, several primary HIV type 2 (HIV-2) and SIV strains (4, 7, 11, 34) as well as some laboratory-selected HIV-1 strains (14, 20) can infect cells efficiently even in the absence of CD4, presumably because of spontaneous complete or partial exposure of the coreceptor binding site (8, 14, 49). Notably, despite their potential to infect CD4-negative cells, CD4-independent strains retain the ability to bind CD4 with an affinity that in some cases can exceed that of CD4-dependent viruses (14, 40).

A model based on the crystal structure of the core of the HIV-1 gp120 molecule indicates that distal parts of gp120 including the bridging sheet and the inner and outer domains are brought together to form the CD4 contact area (21, 22). Twenty-two amino acid residues on CD4 and 26 residues on gp120 form the direct interatomic contacts. Among these, the most important interactions are centered on Phe43 and Arg59 of CD4 and Asp368, Glu370, and Trp427 of gp120. Asp368 in particular is involved in multiple interatomic contacts with CD4, and it forms salt bridges with Arg59, Lys44, and Phe43 (21). The critical role of Asp368 and the corresponding SIV residue Asp385 in this interaction with CD4 was confirmed with mutagenesis studies (31, 33). As with other gp120 residues essential for CD4 binding, the Asp at this location is highly conserved among the primate immunodeficiency viruses.

Replacement of Asp385 of the SIV gp120 with Asn (D385N) impairs the fundamental ability of gp120 to bind CD4 (31). Nevertheless, in two independent studies this substitution was found to be dominant among env sequences obtained from different tissues of SIVmac239-infected macaques (1, 17). It was also demonstrated elsewhere that one of these envelopes, 316BSS, was CD4 independent and, in contrast with previously described CD4-independent SIVmac strains, also demonstrated a diminished ability to bind CD4 (40).

Extending these observations, we also found D385N to be dominant among env clones from two rhesus macaques that progressed rapidly to simian AIDS (SAIDS). These envelopes showed a CD4-independent phenotype and reduced binding to CD4. Moreover, the adjacent change, G383R, which was frequently coselected with D385N, decreased binding even further. An optical biosensor study demonstrated that the SIVmac239 gp120 bound to CD4 with kinetics similar to that known for HIV-1, whereas the gp120 carrying D385N and G383R showed a 40-fold reduction in affinity with a drastic increase in dissociation rate. This suggested that this complex is inherently unstable in the presence of these substitutions. Our results demonstrated that the rapid progression to SAIDS can be associated with the selection for CD4-independent gp120 variants with severely impaired ability to bind CD4.

Multiple primary full-length env genes were cloned from the brain and spleen tissues collected at necropsy from two rhesus macaques infected intravenously with molecularly cloned SIVmac239. Both animals, Mm179-94 and Mm269-95, were rapid progressors, dying from SAIDS within 3 to 4 months, an unusually short period of time for SIVmac239 infection. By the time of death both animals showed poor health, weight loss, and diarrhea. Postmortem examinations revealed opportunistic infections in the lungs and intestines and, in the case of macaque Mm269-95, mild inflammation in the brain. The CD4 counts in peripheral blood were not low at the time of death, which can be explained by the recent observation that SIV selectively targets and destroys CD4-positive cells of the intestine and less prominently in peripheral blood (for a review see reference 43).

Analysis of the env nucleotide sequences showed that Asp385 was replaced in a majority of the clones from each animal: D385N was found in 11 of 13 clones (6 from brain and 7 from spleen) from Mm179-94 and in 8 of 10 clones (6 from brain and 4 from spleen) from Mm269-95. Approximately 70% of these also had an additional mutation at position 383 (G→R). To ensure that the D385N change had occurred in vivo, an aliquot of the inoculum was obtained and its env sequences were analyzed. All of the clones tested (around 100) had an Asp at position 385. Since the D385N change affects critical contacts with CD4, we speculated that, if functional, these envelopes would be CD4 independent with an attenuated affinity to CD4. To test this hypothesis, we selected the functional env clones and determined their dependence on CD4 by a well-described cell-to-cell fusion assay (36).

Effector cells (293T) were infected with a vTF1.1 recombinant vaccinia virus expressing the T7 RNA polymerase followed by CaCl₂-mediated transfection of a plasmid encoding the full-length SIV env to be tested. QT6 target cells were cotransfected with plasmids expressing CCR5 and/or CD4 and a T7 luciferase plasmid expressing the luciferase gene under the control of the T7 promoter. One day after transfection, the effector cells were overlaid on the target cells and incubated at 37°C for 7 h to allow the formation of syncytia. To quantify cell-to-cell fusion, the cells were lysed with 0.5% Triton X-100 in phosphate-buffered saline (PBS), mixed with luciferase substrate (Promega), and immediately counted in a luminometer (Packard LumiCount).

Figure 1 depicts the results of a representative fusion assay performed with several envelope clones where the CD4 and CCR5 receptors were present either together or independently, as indicated in the legend. Several clones demonstrated functional activity, defined as at least a 25-fold increase in luciferase activity over background (SIVmac239 or SIVmac251 with either CCR5 or CD4 only). A subset of these, including F4 and F5 (cloned from the brain of Mm179-94), S41 (spleen, Mm269-95), and B19 (brain, Mm269-95), showed consistent and robust functional activity and, as expected, a CD4-independent phenotype—they could direct CCR5-dependent fusion in the absence of CD4 as efficiently as in its presence. We selected F4 and F5 for further analysis based on their high fusogenic activity and differences in sequences.

FIG. 1.

Functional analysis of the env clones in the fusion assay. Effector cells expressing T7 RNA polymerase and the indicated env gene were overlaid on target cells transfected with a T7-luciferase plasmid and plasmid expressing rhesus CCR5 and/or rhesus CD4 as indicated. Seven hours later the cells were lysed and fusion was quantified by measurement of luciferase activity. Representative results of fusion assays are shown for brain (B) and spleen (S) env clones from animal Mm269-95 (A) and for brain (F) and spleen (S) clones from animal Mm179-94 (B). Please note that the scales for the two panels are different. For the 239 clone in panel A, the CCR5-only results are within the background level (121 RLU [relative light units]).

Both F4 and F5 contain the D385N change previously described as a key determinant of CD4 binding activity (Fig. 2). To determine whether these envelopes could bind CD4, we produced recombinant gp120s from these clones as well as from the parental SIVmac239, by using a baculovirus expression system (41). To facilitate purification of soluble glycoproteins, a six-histidine tag followed by a stop codon was introduced into the gp120 coding sequence immediately before the gp120-gp41 cleavage site. The secreted proteins were then purified from baculovirus-infected Sf9 insect cell culture supernatants by Ni-nitriloacetic acid affinity chromatography (Qiagen) according to the manufacturer's recommendations.

FIG. 2.

Amino acid variations detected in F4 and F5 gp120 sequences in comparison with 239 gp120 from the parental strain SIVmac239. The sequences are numbered according to the work of Kodama et al. (17). Shaded areas represent variable regions 1 through 5 (labeled V1 through V5).

To assess the antigenic integrity of the recombinant gp120s, we tested their reactivity with a panel of monoclonal antibodies raised to both linear and conformational epitopes of the gp120 of CP-MAC, an SIVmac strain closely related to SIVmac239 (9). Each of the three proteins was recognized by all of the monoclonal antibodies in enzyme-linked immunosorbent assay (ELISA), indicating the exposure of antigenic determinants and suggesting the correct folding of the recombinant proteins (data not shown).

The recombinant gp120 proteins were tested for their ability to interact with CD4 by an ELISA, as described previously (46). All experiments used the commercially available soluble human CD4 (Progenics). Human CD4 was shown to be interchangeable with rhesus monkey CD4 in terms of SIV infection (40) and should therefore provide an accurate model for gp120 binding. Ninety-six-well ELISA plates (Costar) were coated in duplicate with the gp120s at a saturating concentration of 5 μg/ml and washed with washing buffer (0.2% Tween 20 in PBS, pH 7.4) and then blocked with blocking buffer (0.2% Tween 20, 5% nonfat milk in PBS, pH 7.4). The plates were subsequently incubated with serial dilutions of soluble CD4 (Progenics), followed by rabbit anti-CD4 serum (T4-4; National Institutes of Health AIDS Research and Reference Reagent Program) and goat anti-rabbit immunoglobulin G coupled with horseradish peroxidase (Amersham), which gave a colorimetric reaction in the presence of ABTS reagent (Moss, Inc.). Optical density at 405 nm was measured with a Perkin-Elmer HTS 7000 Bioassay Reader.

The CD4 binding properties of F4 and F5 gp120s are illustrated in Fig. 3. In comparison with the wild-type SIVmac239 gp120, F4 demonstrated a significant reduction in CD4 binding; the reduction in binding for F5 was even more dramatic. While the decreased reactivity of the F4 gp120 could be explained at least in part by the D385N mutation, the drastic drop in binding seen with F5 required additional explanation. F4 and F5 amino acid sequences differ in only two residues, G383 and D516 (Fig. 2). Between these two, the G383R change that was coselected with D385N in 70% of the clones was of particular interest. Although G383 has not been implicated in CD4 binding in previous studies, we speculated that, because of its proximity to the CD4 binding site, the G383R mutation could have an adverse effect on the gp120-CD4 interaction by causing a conformational distortion and electrostatic interference with R59 on CD4 (discussed below).

FIG. 3.

Analysis of CD4-gp120 binding in ELISA. Soluble gp120 proteins 239, F4, mutated F4 with G383 replaced by R (F4/R), F5, and mutated F5 with R383 replaced by G (F5/G) were tested for their ability to bind CD4 in an ELISA performed as described in the text. NP, no protein. OD₄₀₅, optical density at 405 nm.

We therefore produced the reciprocal mutants: R383G and G383R were introduced into the F5 and F4 gp120s, respectively (Fig. 2), by using the QuikChange XL site-directed mutagenesis kit (Stratagene). Mutated DNAs were generated with PCR using PfuTurbo DNA polymerase and primers containing the desired mutations. After 18 thermal cycles, the DNA templates were digested with DpnI endonuclease, and the DNA with the introduced mutations was transformed into XL10-Gold ultracompetent cells (Stratagene). Recombinant clones with correct sequences were selected by sequencing. Mutant proteins designated F4/R and F5/G were produced with the same baculovirus expression system and tested in the CD4 binding ELISA (Fig. 3).

Replacement of Gly with Arg at position 383 of F4 (F4/R) significantly decreased its binding to CD4. In turn, the reverse mutation R383G introduced in F5 (F5/G) resulted in binding with CD4 to a level comparable to that of F4 (Fig. 3). G383 is conserved among the majority of SIV strains, and its analogous residue is present in virtually all HIV-1 strains. These findings suggested a possible role for G383 in the maintenance of the optimal structure for interaction with CD4.

We then analyzed the binding of the soluble gp120s to CD4 by surface plasmon resonance optical biosensor assay (BIACORE 3000 optical biosensor; Biacore Inc., Uppsala, Sweden). The experimental and reference surfaces were prepared in duplicate by immobilization of soluble CD4 (Progenics) or control bovine serum albumin to CM sensor chip flow cells. All four flow cells were monitored simultaneously. The final density was 640 resonance units (RU) for CD4 surface and 900 RU for bovine serum albumin surface. Ligand coupling was performed according to procedures described elsewhere (50). Binding kinetics data were collected by injection of the gp120s at various concentrations in PBS buffer (pH 7.4) containing 0.005% Tween 20 and 0.1% dextran, over the experimental and reference chip surfaces at a flow rate of 100 μl/min for 2 min (association) followed by a surface wash with the same buffer for 4 min (dissociation). The binding data were globally fitted to a one-to-one interaction model (Fig. 4) to determine kinetic constants. The biosensor assay was performed three times with similar results, and kinetic constants for the best fit are summarized in Table 1. The monomeric gp120 of SIVmac239 bound to CD4 with an affinity similar to that previously reported for HIV-1 gp120 (50). In contrast, the F4 gp120, and to an even greater extent the F5 gp120, demonstrated a significant reduction in affinity for CD4. Overall, the equilibrium dissociation constants (K_D) for the F5, F4, and 239 gp120s were 1.19 × 10⁻⁶ M, 2.37 × 10⁻⁷ M, and 3.01 × 10⁻⁸ M, respectively, indicating an approximately 40-fold difference in affinity between the F5 gp120 and that of the parental virus. The difference between F4 and F5 (over sixfold) is likely attributed to the G→R change at position 383. The observed decrease in affinity for CD4 was due primarily to a drastic increase in the dissociation rate constant (k_off), indicating an inherently unstable gp120-CD4 complex.

FIG. 4.

Kinetic analysis of CD4-gp120 binding. The soluble gp120s were injected over a CD4 surface (surface density, 640 RU) at increasing concentrations for 2 min at the flow rate of 100 μl/min. (A to C) The collected kinetic data for 239 (A), F4 (B), and F5 (C) gp120 proteins were globally fitted to a 1:1 binding model with BIAevalution 3.0 software. The lines represent the experimental data, and the symbols represent the fitting results. (D) Sensorgram overlay for normalized binding response over CD4 for 239, F4, and F5 gp120s at a concentration of 1.5 μM.

TABLE 1.

Kinetic constants from global analysis of the biosensor binding of the 239, F4, and F5 soluble gp120 proteins to the surface-immobilized CD4

Protein	kon (M−1 s−1)	koff (s−1)	rmax (RU)	KD (M)	χ2a
239	1.5 × 104	4.53 × 10−4	385	3.01 × 10−8	1.04
F4	7.33 × 103	1.74 × 10−3	217	2.37 × 10−7	1.84
F5	3.53 × 103	4.21 × 10−3	106	1.19 × 10−6	0.53

^aχ² represents a standard statistical measure of the closeness of the fit. Values below 10 are acceptable.

In this study we analyzed the CD4 binding properties of SIV gp120s carrying a mutated aspartic acid at position 385 (D385N), which has been shown by others to play an important role in gp120's interaction with this receptor. These envelope sequences were dominant (i.e., present in a majority of clones) in the brains and spleens of two rhesus macaques that developed fulminating SAIDS within 3 to 4 months after infection. Interestingly, in a previous study D385 was found unaltered in numerous env clones from similar sites of animals that were infected with the same virus inoculum but had a more prolonged (2 to 3 years) disease course (37). Nevertheless, D385N was described previously for several SAIDS cases with fast disease progression (17). Taken together, these observations suggest a possible association between a rapid disease phenotype and emergence of the D385N change.

Our findings confirm that acquisition of CD4 independence in vivo may be accompanied by a significant reduction in the binding to CD4 and, furthermore, that in addition to the previously described D385N, an adjacent G383R change is another determinant of this phenotype. Gly is very conserved during protein evolution regardless of its proximity to functionally active sites because, as the smallest amino acid, its replacement invariably introduces a bulkier molecule into the polypeptide chain, resulting in structural distortion at least in the immediately surrounding domains (13, 32). Although crystal models do not indicate that G383 forms a molecular bond with CD4, its replacement with arginine, one of the largest amino acids, may induce conformational distortion proximal to the CD4 binding site as well as electrostatic interference with Arg59 on CD4, thus explaining its adverse effect on CD4 binding.

The D385N and G383R changes are frequently found together. In this study both substitutions were found in 70% of the env clones from spleen or brain of two animals with rapid progression to SAIDS. The rapid fixation of these mutations and their spread to two sites imply a strong positive selection for this phenotype. Most of the CD4-independent strains studied so far infect CD4-negative cells somewhat less efficiently than CD4-positive cells (7, 14, 26), and yet this phenotype may confer an evolutionary advantage by expanding the spectrum of cells susceptible to infection. For example, the CD4-independent SIVmac strain SIV17E/Br is capable of replicating in brain capillary endothelial cells (8), perhaps allowing easier penetration of this organ. How these changes—thought to result in increased sensitivity to neutralizing antibodies (10, 18, 19, 25)—could arise is more perplexing. The chemokine receptors were probably the sole ancestral receptors for SIV (34, 48) and other lentiviruses (such as feline immunodeficiency virus). The use of another receptor such as CD4 was acquired subsequently during evolution, perhaps in order for the virus to target key cells of the immune system and to sequester the highly conserved chemokine receptor binding site from the host immune response. The selection of envelopes with reduced affinity to CD4 indicates a tendency to evolve towards a primordial phenotype in certain situations. The “crash-and-burn” progression to disease with a rapid death within several months seen in the animals in this study is often associated with a failure to seroconvert (15), perhaps providing conditions that are favorable for the selection of this potentially more sensitive phenotype. Unfortunately, archival serum samples with which to test this hypothesis directly were not available for us to directly assay for neutralization.

The ability of some isolates to use chemokine receptors without CD4 implies the spontaneous complete or partial exposure of a chemokine receptor binding site, and indeed those strains that have been studied so far demonstrate (i) enhanced binding to the coreceptors, (ii) exposure of epitopes overlapping the coreceptor binding site, and (iii) increased sensitivity to neutralization with monoclonal antibodies mapped to these epitopes (10, 18, 19, 25). The molecular determinants responsible for this structural transition are different for each example and include the loss of N-linked glycosylation sites at the base of the V1/V2 loop in some strains (20) or single-amino-acid substitutions in the V3 and C4 regions of gp120 and/or in the cytoplasmic portion of gp41 in others (10, 12, 23). The CD4-independent envelopes from this study acquired surprisingly few amino changes compared with the parental virus. In addition to D385N, F4 and F5 have two common changes in gp120, I40V (C1 domain) and D180N (V1/V2 loop), and one in gp41 (R753Q). Whether any of them assist the CD4-independent interaction with coreceptor remains to be answered.

Most SIV strains that are CD4 independent also replicate well in macrophages (7), and it is possible that selection for robust replication in these cells is associated with this phenotype (2, 30). We have yet to determine whether the changes noted in our clones will fall into this grouping once they have been engineered back into full proviruses, and this will be the subject of further studies.