In this study, we identified a high-affinity variant of Fab Z13 through targeted random mutagenesis of CDR L3 and affinity selection against recombinant gp41 and an MPER peptide. The enhanced variant Z13e1 binds to the selecting antigens with affinity ~100-fold greater than that of parental Fab, which is respectable and not atypical for in vitro affinity maturation experiments (40
). In theory, it should be possible to further optimize Z13e1 affinity, although further improvements in neutralization potency may require mutation of other Z13 CDRs and more technically challenging selections using specific antigens that present the MPER of gp41 as in a native trimeric and/or membrane context. There may also be a limitation in neutralization potency by Z13 variants due to steric restrictions to the angle of approach of the Fab arm to its MPER epitope or due to other factors. Nevertheless, the apparent affinity of Z13e1 to optimized MPER peptides (IC50
, ~10 nM) is within the range of binding affinities typically observed for MAbs generated in vivo, such as 2F5 and 4E10 (2
). We note that whereas other variants besides Z13e1 were selected, Z13e1 was largely representative of these, though perhaps it had slightly higher affinity. Thus, no strong-binding clones that were nonneutralizing were found, and conversely, no low-affinity clones that had strong neutralizing activity were found. Hence, for the clones we tested at least, improved affinity to MPER antigens was predictive of improved neutralization potency, although Z13e1 neutralizes the sensitive HIV-1MN
strain over an order of magnitude more potently than it neutralizes the MPER-sequence-matched primary virus, HIV-1JR-FL
The mapping studies all suggest strongly that Z13e1 binds to an epitope between those of 2F5 and 4E10. In the peptide truncation analysis, the shortest peptide epitope for Z13e1 for which Ab affinity was near optimum was the 12-mer motif (indicated in boldface) within the peptide 178-3 (WASLWNWFDITN
, ~30 nM). The first six and the last seven residues of this 12-mer motif have been implicated in 2F5 and 4E10 recognition, respectively (2
). In addition, Z13e1 is clearly able to block the binding of both 2F5 and 4E10 to recombinant gp41. The Ala substitutions N671A and D674A both knock out Z13e1 peptide binding and neutralization, suggesting that the Z13e1 epitope may overlap more with that of 4E10, the crucial residues of which are W672 and F673 (10
). Since 4E10 recognizes a largely helical epitope, it has been suggested that this helical region bears distinct “neutralizing” and “nonneutralizing” faces, since access to the MPER would presumably be restricted to an exposed face on the functional trimer (6
). If the MPER of gp41 were strictly helical throughout the Z13e1 epitope, that would place the residues most critical to Z13e1 recognition, N671 and D674, on the putative “nonneutralizing” face of the helix, away from W672, the residue that is most buried in the 4E10 paratope (Fig. ). Thus, an explanation for the relatively modest neutralization potency of Z13e1 may be that its helical epitope impinges on the nonneutralizing face of the MPER, which is occluded by neighboring MPERs. However, a nonhelical epitope for Z13e1 cannot be ruled out. Indeed, the 4E10 core epitope loses a-helical character toward its N-terminal residues (6
), and the 2F5-bound peptide adopts an extended conformation (33
), suggesting either that there is significant flexibility in the MPER of gp41 or that a distinct structural transition occurs in the segment between the 2F5 and 4E10 epitopes (60
). In either case, the high-affinity epitope conformation(s) that is available to Z13e1 on recombinant gp41 and MPER peptides does not appear to be persistently present on the cognate epitope found on the native trimer (as evidenced by the low neutralization potency of Z13e1). A structure of Z13e1 in complex with a peptide that overlaps with those used in the structure determinations of 2F5 and 4E10 (10
) should help to explain the limited ability of Z13e1 to bind to the native Env trimer and refine the existing structural models of the MPER (59
FIG. 9. Hypothetical model of the MPER of HIV-1 gp41 as an ideal alpha helix, in side (0°) and end-on (90°) views, places 4E10 and Z13e1 crucial binding residues on opposite faces of the helix. The MPER residues that are crucial in the neutralizing (more ...)
It is perhaps not surprising that Fab Z13e1 was able to bind both monomeric and trimeric Env species by use of the BN-PAGE gel shift assay because Z13e1 recognizes a continuous epitope. Given the established relationship between neutralization and the ability of Ab to bind to trimeric envelope glycoprotein (18
), the neutralizing activity of Fab Z13e1 most likely derives from its ability to bind to the trimeric envelope, even though monomer binding affinity is strong and is improved to an extent similar to that seen for trimer binding affinity.
The Fab and IgG formats of Z13e1 appeared to be roughly equivalent in neutralization potency on a molar scale, differing by only threefold or less. Such a result may reflect the inability of two Fab arms of the IgG to bind simultaneously to a functional Env trimer(s), although other steric factors affecting the accessibility to receptor-bound envelope trimers may also influence neutralization by IgG (27
). We also found that neutralization of HIV-1JR-FL
by 4E10 was not greatly improved in going from the Fab to the IgG format. This result is in apparent conflict with a previously published result in which IgG 4E10 was found to be severalfold more potent than Fab against HIV-1JR-FL
). The reason for this discrepancy is not clear, although a different assay format and target cells were used in the previous study.
As with an earlier study involving 2F5 and 4E10 (21
), we found that Z13e1 bound weakly and somewhat nonspecifically to different antigens in an ELISA setup in which detergent was limiting, although its reactivity was typically less than that of 4E10 (nonspecific binding with 2F5 was the least of the three). It is presently unclear what this low-level and nonspecific binding means in terms of vaccine design involving the MPER of gp41. However, if there were tolerance mechanisms that would limit the elicitation of neutralizing anti-MPER Abs, they would less likely be a result of Ab cross-reactivity with a specific host antigen than be attributable to a more general phenomenon associated with their specificity for a membrane-proximal epitope. A casual inspection of the primary sequences of the heavy- and light-chain CDRs of Z13e1 reveals a relatively high content of hydrophobic relative to charged residues in the heavy-chain CDRs (16 hydrophobic versus 4 charged residues), particularly in CDRs H1 and H3 (64
). The subtle “stickiness” of Z13e1 may relate to these hydrophobic CDRs, which could facilitate recognition of a hydrophobic epitope very close to a membrane. Nevertheless, Z13e1 and the other anti-MPER MAbs bind with very high (low-nanomolar) affinity to their MPER peptide epitopes. A more detailed analysis of the interaction of 2F5 and 4E10 with CL and other lipid antigens is in progress (Erin Scherer, MBZ, DRB). Isolating and studying more Abs against the MPER of gp41 will also help to determine if such Abs are subject to a strict requirement for the type of weak and nonspecific reactivity to unrelated antigens that we observed with the existing three MPER MAbs.
The concentration of neutralizing Ab required to protect against challenge with HIV has been suggested to be on the order of 100-fold greater than that required to completely neutralize the virus in a classical peripheral blood mononuclear cell-based assay (35
). Furthermore, neutralizing Ab alone is likely to have limited control over an established HIV-1 infection (39
). In fact, the potential of 2F5 and 4E10 to contribute to protection against HIV-1 has been questioned due to a clinical study in which a passively administered cocktail of 2G12, 2F5, and 4E10 showed a limited ability to control viral rebound in several HIV-1-infected patients (50
). Since the potency of Z13 was so readily improved in this study, a similar approach might enable improvements to the potencies of 2F5 or 4E10, although the affinities of 2F5 and 4E10 for gp41 peptides are already quite high (Kd
s in the low nanomolar range), so improvements to these MAbs may be more challenging. In order to determine the in vitro neutralization titers that will have to be elicited by immunogens based on the MPER, it will be important to establish the individual doses of 2F5 and 4E10 required for protection in vivo.
Using HIV-2 chimeric viruses that bear HIV-1 MPER sequences, MPER-specific neutralizing Abs have recently been detected in the sera of certain HIV-1 infected individuals; these Abs do not recognize the 2F5 or 4E10 epitopes (G. F. Shaw, Bibollet-Ruche, J. Decker, H. Li, P. Goepfert, M. Peeters, S. Allen, E. Hunter, J. Robinson, and P. Kwong, presented at the 13th Conference on Retroviruses and Opportunistic Infections, Colorado Conference Center, Denver, CO, 2006). Perhaps these Abs, like Z13e1, recognize epitopes between those of 2F5 and 4E10. If Abs are more easily elicited to the segment between the 2F5 and 4E10 epitopes than to these two epitopes themselves, it will be important to determine their neutralization potencies. We speculate that the segment of the MPER to which Z13e1 binds may elicit Abs that are more potent than Z13e1. Such Abs would have to bind with affinities greater than that of Z13e1 to the functional Env trimer, regardless of their affinities for MPER peptides or recombinant gp41 (60
). MPER immunogens designed to elicit potently neutralizing Abs will need to reproduce the structural features that limit access to the Z13e1 epitope but preserve access to the 2F5 and 4E10 epitopes, as found on the functional Env trimer.
An HIV-1 envelope mimetic against which in vitro affinity maturation of a neutralizing MAb leads to diminished neutralization potency is probably a poor vaccine candidate. This concept was appreciated in a prior study in which the anti-gp120 neutralizing Ab, b12, was affinity optimized against monomeric gp120, eventually leading to decreased neutralization potency (24
). In the present study, affinity maturation of Z13 against recombinant gp41 and an MPER peptide did translate to improved neutralization potency, but the potency of Z13e1 is still less than those of 2F5 and 4E10, which bind to these antigens in vitro with affinities comparable to that of Z13e1. The observed modest neutralization potency of IgG Z13e1, despite its very strong binding affinity for MPER peptides and recombinant gp41, strongly suggests that there are limitations in exposure of the Z13e1 epitope, and possibly others like it, on the native envelope trimer. Firmer conclusions as to whether these limitations are due to steric occlusion of a static conformation of the MPER or to short-lived access to a transient epitope on a flexible MPER await crystallographic studies with a Z13e1-peptide complex. Z13e1 Fab may also be useful as a probe in cryo-electron microscopy studies to determine whether the MPERs within the native Env trimer are strictly separated as in a “tripod,” are bundled together as in a single “trunk,” or perhaps participate in a more dynamic equilibrium between these two states. Finely optimizing the presentation of the “neutralizing face” and native conformation of the MPER may be crucial in order to elicit Abs with neutralization potencies equal to or exceeding those of 2F5 and 4E10.