In this work, we adapted two different HIV-1 isolates, HIV-1(NL4-3) and HIV-1(KB9), to replicate in cells using the common marmoset receptors CD4 and CXCR4. The analysis of the envelope glycoproteins of the adapted viruses revealed that in both cases a small number of changes that appeared during the adaptation allowed the viruses to use the marmoset receptors to gain entry into the cells. Probably because of differences in the phenotypes of the parental viruses, HIV-1(NL4-3) and HIV-1(KB9) adapted to the marmoset receptors by distinct mechanistic pathways. The genotypic and phenotypic properties of the adapted viruses are summarized in Table .
The changes that allow the HIV-1(NL4-3) virus to enter cells expressing marmoset CD4 and CXCR4 involve gp120 residues 281, 334, and 242. The changes that appeared in these residues during the adaptation increase the gp120 binding affinity for marmoset CD4 without a detectable effect on the affinity for human CD4. In X-ray crystal structures of the HIV-1 gp120-CD4 complex, alanine 281 makes direct contact with CD4 (Fig. ) (28
). Molecular modeling predicts that the larger side chain of valine that was associated with adaptation makes better contact with CD4 and increases the gp120 binding affinity for marmoset CD4 (Table ). The major differences between human and marmoset CD4 that impact gp120 binding involve residues 48 and 59 in CD4 domain 1 (30
). In human CD4, proline 48 contributes to a β-turn that results in an antiparallel orientation of the C" and D strands of CD4, each of which makes important contacts with gp120 (29
). The replacement of proline 48 with a glutamine residue in common marmoset CD4 might alter the β-turn and consequently affect the relationship of the C" and D strands. The S334N change in gp120 has been previously seen in an HIV-1 isolate adapted to replicate efficiently in cells expressing a human CD4 mutant (K46D) with a low affinity for gp120 (8
). Residue 334 is located near the base of the V3 loop and appears to be too far from the CD4 binding site to alter CD4 affinity directly; however, the resulting loss of a carbohydrate at asparagine 332 might change the conformation of the unliganded gp120 and thus indirectly increase CD4-binding ability (Table ). The V242I change alone increases the binding affinity of soluble gp120 for marmoset CD4 only minimally but does contribute to marmoset CD4 binding in the context of the A281V change. Valine 242 projects into a groove in gp120 that has been proposed to make contact with gp41 and regulate fusion efficiency (51
). It is possible that the adaptation-associated change in valine 242 amplifies the impact of CD4 binding on subsequent conformational changes required for virus entry. Consistent with this, the viruses containing the V242I change exhibited a slight increase in sensitivity to neutralization by different antibodies (Table ), possibly reflecting a more activation/deactivation-prone state. In summary, most of the HIV-1(NL4-3) adaptation-associated changes result in specific increases in binding to marmoset CD4, providing a natural explanation of the adaptation.
FIG. 6. Modeling the interaction of the adapted HIV-1(NL4-3) gp120 core with marmoset CD4. The modeled NL4-3 gp120 core (blue) bound to two-domain marmoset CD4 (magenta) is shown, with the gp120 residues that changed during the adaptation indicated. Also shown (more ...)
Modeling predictions of binding energies between HIV-1(NL4-3) gp120 core variants and two-domain marmoset CD4a
The changes that evolved in the HIV-1(KB9) envelope glycoproteins as a consequence of the adaptation to cells expressing marmoset receptors differ in location and mechanistic consequences from those observed in the case of HIV-1(NL4-3). One likely explanation for these differences lies in the already high level of efficiency with which the parental KB9 gp120 glycoprotein binds marmoset CD4. One obvious reason for this high binding affinity is that KB9 gp120 already has a valine at residue 281, which the above studies with the NL4-3 gp120 variants demonstrate contributes to marmoset CD4 binding. Of interest, the conversion of alanine 281 to valine occurred during the in vivo passages of SHIV-89.6 in rhesus monkeys that led to the generation of SHIV-KB9 (22
). Previous work, which we confirmed here, indicated that the KB9 gp120 glycoprotein binds human and rhesus monkey CD4 comparably to binding of the 89.6 gp120 glycoprotein; however, KB9 gp120 binds marmoset CD4 more efficiently than 89.6 gp120 (data not shown). Thus, the differences between the 89.6 and KB9 gp120 glycoproteins, including the alanine-valine substitution at residue 281, contribute to an increased binding affinity of KB9 gp120 for common marmoset CD4 but not for human or rhesus monkey CD4.
The adaptation-associated changes in the HIV-1(KB9) envelope glycoproteins involve the gp120 V1 and V2 variable loops and the gp41 HR1 region. Presumably because the KB9 gp120 glycoprotein starts off with a relatively high affinity for marmoset CD4, the adaptation-associated changes in the HIV-1(KB9) envelope glycoproteins only minimally affect marmoset CD4 binding. Only the V2 loop change, E172K, is associated with a slight but specific increase in the affinity of gp120 for marmoset CD4. How then do the observed adaptation-associated changes exert their dramatic impact on the ability of the HIV-1(KB9) envelope glycoproteins to utilize marmoset receptors? One possibility that we considered is that the adapted KB9 gp120 glycoprotein binds marmoset CXCR4 more efficiently. Although the gp120 V1/V2 and gp41 changes are not expected to affect the CXCR4-binding region of gp120 directly, they could hypothetically exert indirect effects. An increase in affinity for CXCR4 predicts a decrease in sensitivity to the CXCR4 ligand, AMD3100, which we did not observe. In fact, the adaptation-associated changes resulted in a slightly increased sensitivity to AMD3100 for infection of cells bearing human or marmoset CXCR4. Thus, an increase in affinity for CXCR4 is an unlikely explanation for the HIV-1(KB9) adaptation to marmoset receptors.
An alternative explanation that is more consistent with the observations recognizes the high degree of resistance to neutralizing antibodies that the HIV-1(KB9) envelope glycoproteins achieved as a result of in vivo SHIV propagation in monkeys. Resistance to neutralization by antibodies is often accompanied by an increase in HIV-1 dependence on CD4 for entry (27
). Thus, as a result of in vivo requirements to avoid neutralizing antibodies, the HIV-1(KB9) envelope glycoproteins may less efficiently undergo the requisite conformational changes for virus entry in response to nonoptimal CD4. Even though the KB9 envelope glycoproteins bind marmoset CD4 efficiently, minor differences in the interaction of CD4 and the gp120 variable loops, for example, could impact the efficiency of entry. CD4 binding has been shown to reposition the gp120 V1/V2 loops, promoting subsequent events in the virus entry process (46
). These events may include chemokine receptor binding and gp41 conformational rearrangements. Although the structural details of V1/V2 loop interaction with CD4 are unknown, the surface of CD4 domains 1 and 2, which contains an abundance of charged residues, is likely involved. In this light, it is intriguing that both of the KB9 V1/V2 alterations that contribute to marmoset CD4 adaptation involve a change in charge, from acidic glutamic acid residues to basic lysine residues. Because at least 10 surface-exposed residues potentially able to contact the V1/V2 gp120 loops differ in charge between human and marmoset CD4 domains 1 and 2 (Fig. ), the charge changes in the adapted KB9 gp120 V1/V2 loops may allow attractive or repulsive electrostatic forces to drive the appropriate loop movements in a marmoset CD4-specific fashion. A salt bridge between lysine 172 and an acidic residue on the marmoset CD4 surface might additionally contribute to the slight increase in binding affinity observed for this variant. The decreased ability of KB9 envelope glycoproteins with some of V1/V2 loop changes to support entry into cells expressing human CD4 is consistent with the above model. The adaptation-associated V1/V2 loop changes, particularly E172K, also appear to disrupt interactions between elements on the unliganded KB9 trimer that govern neutralization resistance, since the adapted virus envelope glycoproteins are much more susceptible to neutralization by antibodies and sCD4.
FIG. 7. Differences between human and common marmoset CD4. The primary amino acid sequences of the amino-terminal two domains of human (Hu) and common marmoset (Cj) CD4 are aligned. Identical residues are shaded. Amino acid residues that exhibit differences in (more ...)
The A561T change in the gp41 HR1 region contributes to a two- to threefold improvement in the efficiency of infection of cells expressing the marmoset receptors by HIV-1(KB9) viruses. This change did not alter the sensitivity of the virus to antibodies or sCD4. The mechanism by which the A561T change contributes to adaptation to marmoset receptors presumably involves promotion of entry-related conformational events in the setting of nonoptimal receptor interactions. Of interest, changes in the adjacent residue 560 have been associated with SIV macrophage tropism (34
), which involves the ability to use the low levels of CD4 on the macrophage surface more efficiently (2
). Both residues 560 and 561 are located in the b and c positions, respectively, within the gp41 heptad repeat; the predicted surface exposure of these b and c residues would allow interactions with other envelope glycoprotein elements either prior to or after formation of the gp41 six-helix bundle thought to represent the fusogenic conformation.
The derivation of HIV-1 envelope glycoprotein variants that can efficiently use the common marmoset receptors CD4 and CXCR4 to enter cells advances efforts to overcome the major early block to HIV-1 in these New World monkeys. The use of these viruses should allow exploration of any additional blocks to HIV-1 in these monkeys. Understanding and circumventing these blocks may lead to novel animal models for study of HIV-1 pathogenesis.