Ibalizumab is a humanized anti-CD4 monoclonal antibody that potently and broadly blocks in vitro
infection by a large panel of HIV-1 isolates (4
). From phase 1 through phase 2b clinical trials in infected patients in need of salvage therapy, ibalizumab has demonstrated in vivo
antiviral activity by consistently lowering viral load by about 1 log, without causing significant adverse side effects (8
). Ibalizumab thus appears to be a promising agent for salvage therapy as well as for passive immunization against HIV-1 infection. Consequently, it seems important to define the epitope of this antibody in order to gain further insights into its mechanism of action as well as its safety profile.
Previous studies have shown that ibalizumab, like its murine progenitor (5A8), is able to bind to D2 of rhesus and human CD4, thereby preventing postbinding entry of the virus into CD4 T cells (5
). Here, extensive mutagenesis studies indicate that E77, S79, P121, P122, and Q163 are essential for ibalizumab binding to hCD4 (Fig. ). These findings do not exclude the possibility that exchanging the proline at positions 121 and 122, known to be important residues in overall protein architecture, for the mouse counterpart amino acids altered the conformation of the hCD4 protein and therefore its function. The results presented in Fig. are consistent with previous findings that amino acids 121 to 124 on hCD4 are necessary for 5A8 (ibalizumab precursor) binding (5
). Interestingly, amino acids 127 to 134, identified in the previous study as important for hCD4 binding (5
), seem to be less crucial in our analysis. This apparent discrepancy may be due to the different methods used for epitope mapping. Furthermore, the previous study used replacement of a stretch of amino acids whereas the current study utilized single amino acid substitutions.
It is noteworthy that residues E77 and S79, both located in D1 of hCD4, as well as Q163 also seem to be essential in for ibalizumab binding. These amino acids cluster together with P121 and P122 on one surface of the three-dimensional structure of hCD4 (Fig. ). If D1 were the head and neck and D2 were the torso, then it appears that ibalizumab would bind to the back of the “neck and upper torso,” whereas gp120 and MHC class II molecules bind to the “nose.” Thus, it is not surprising that ibalizumab does not interfere with gp120-CD4 binding (5
). Perhaps more importantly, the spatial separation of the ibalizumab epitope and the MHC class II binding site helps to explain why ibalizumab is well tolerated by patients without any evidence of adverse immunological sequelae noted to date.
In this study, we also studied another monoclonal antibody, M-T441, specific for D2 on human CD4, in order to compare and contrast its properties with ibalizumab. M-T441 was found to be substantially less potent in blocking HIV-1 infection in vitro (Fig. ). One possible explanation for this is that its epitope is lower down on D2 at the “back of the middle torso” and ~45° axially separated from that of ibalizumab (Fig. ), thereby resulting in less interference of some critical step in viral entry. The lower virus neutralization potency of M-T441, however, could also be explained by its lower binding affinity as measured in a Biacore assay (Fig. ). It is possible that both affinity and epitope differences contribute to the lower potency of M-T441 against HIV-1.
Armed with information on the epitope of ibalizumab, what could be said about the mechanism by which this antibody blocks HIV-1 infection of CD4 T cells? Specifically, which step of viral entry is inhibited? A previous study (5
) showed that gp120-CD4 binding is not inhibited by ibalizumab, but it is formally possible that virion binding to CD4 is blocked because monomeric gp120 does not accurately reflect the trimeric glycoprotein spike on the virion surface. This possibility is rendered less likely because the gp120 binding site in D1 is located on a surface opposite the ibalizumab epitope (Fig. ). Nevertheless, it would be prudent to conduct follow-up studies to examine whether ibalizumab could interfere with the binding of HIV-1 particles to CD4-positive cells.
It is known that after gp120 binds to CD4, both molecules undergo conformational changes, leading to the exposure of the so-called bridging sheet on gp120 that is important for engaging the coreceptor, either CCR5 or CXCR4 (10
). Thus, it is theoretically possible that ibalizumab attached to the “neck and upper torso” of CD4 could impede the formation of the bridging sheet. This possibility is, however, unlikely given that ibalizumab does not lower the affinity of CD4-gp120 binding.
Upon close inspection of the structure of gp120 core complexed with CD4 and a monoclonal antibody, 17b (which binds to the bridging sheet) (16
), it is clear that the ibalizumab epitope is in close proximity to the V5 loop of gp120 (Fig. , green elements) with its one or two N-linked glycans. This spatial relationship suggests that ibalizumab could easily impinge on this part of gp120 or the glycans thereon and thereby indirectly affect the structural alterations necessary in the subsequent steps during viral entry.
Upon exposure of the bridging sheet, gp120 must then move closer to the cell surface in order to engage either CCR5 or CXCR4. One can imagine this happening by having the CD4 molecule lie down on the cell surface (flexing between D4 and the transmembrane region) or jackknife from a rod-like structure into a ball (folding at the hinge between D2 and D3). Conceivably, ibalizumab could also cross-link two CD4 molecules and thus restrain them from lying down. The latter hypothesis is supported by the original observation made by Burkly et al. (5
) that Fab fragments of 5A8 do not have antiviral activity. However, our recent finding that Fab of ibalizumab is only slightly less active than the full antibody or F(ab′)2
(unpublished data) essentially rules out cross-linking of CD4 as a mode of inhibition. With ibalizumab attached to the back of the “neck and upper torso,” it is easy to envision CD4 having trouble lying down with its “face” up. But this does not explain how CD4 wouldn't be able to lie down with its “face” down. Two reports suggest that CD4 becomes a ball-like structure following binding to gp120 (1
), perhaps by jackknifing “backwards” with D1D2 anti-parallel to D3D4 (1
). With such a scenario, it would be easy to imagine how ibalizumab could prevent such a dramatic structural rearrangement of CD4. However, it must be pointed out that CD4 may not have such a high degree of flexibility at the hinge between D2 and D3 to allow jackknifing to occur (24
). A more subtle bending at this hinge may be possible though.
It remains possible that ibalizumab blocks an even later step in viral entry, such as coreceptor binding, unleashing of gp41, or the six-helix bundle formation associated with membrane fusion (6
). If so, it would suggest that CD4 stays involved with the envelope glycoprotein for much longer than current models of viral entry would indicate (9
In summary, by mapping the epitope of ibalizumab, we have gained a better appreciation for why this antibody has not had any adverse immunological consequences in infected patients to date. While the precise mechanism of its inhibition of HIV-1 entry remains elusive, it has become clear that CD4 does not merely serve as an anchor for the virus; instead, it is likely involved in a complicated dance with gp120/gp41 throughout a considerable part of the entry process. Follow-up structural and mechanistic studies are necessary to provide a better understanding of how ibalizumab blocks HIV-1 infection and how it could be further developed as a passive immunization agent to slow the spread of this pandemic.