The data presented herein confirm and extend our previous conclusion [29
] that the neutralizing potency of sCD4-17b derives from the ability of both moieties on a single bifunctional molecule to associate simultaneously with their corresponding binding sites on a single gp120 subunit. Variants of the protein with sufficiently long L1 linkers, i.e. the original sCD4-35
-17b and the newly described sCD4-40
-17b displayed potency much greater than either sCD4 alone, or in equimolar amounts with unlinked 17b SCFV. Preliminary results indicated that a construct containing an L1 linker of 55 amino acids (11 G4
S repeats) was comparably effective (data not shown). By contrast, sCD4-5
-17b, which contains an L1 linker too short to allow simultaneous binding of the sCD4 and 17b moieties, had a much weaker potency. However when tested against strains that were highly sensitive to sCD4-17b, sCD4-5
-17b proved significantly more effective than the mixture of unlinked sCD4 plus 17b SCFv. A plausible explanation is that while sCD4-5
-17b was incapable of mediating simultaneous binding of both components to the same gp120 subunit, reversible binding of molecules via the sCD4 portion effectively increased the local concentration of 17b SCFv in the vicinity of Env, where it could bind to gp120 subunits that had been induced by sCD4 moieties on separate molecules of the chimeric protein. Thus bifunctional binding molecules can potentially display enhanced activities even when the two binding moieties are incapable of interacting simultaneously with a single target molecule.
The exceptional breadth displayed by sCD4-17b (neutralization of 100% of isolates tested from genetically diverse HIV-1 subtypes) confirms our original expectation, based on the requirement for CD4 binding amongst all natural HIV variants coupled with the high conservation of the bridging sheet due to its critical role in coreceptor binding of HIV-1 (and HIV-2 as well [16
]). The breadth exceeded by a considerable margin those of the well-characterized broadly neutralizing MAbs IgG b12, 2G12, 2F5 and 4E10 tested against the same HIV-1 isolates [39
] (also V. Polonis and S. Tovanabutra, personal communication). Of interest in this regard is the recent report of 2 new human gp120-targeted MAbs, PG9 and PG16, that neutralize a large fraction (79% and 73%, respectively) of Env pseudotypes from a genetically diverse panel of primary HIV-1 isolates (IC50
values ranging from 0.001 to 50 μg/ml); of the 162 pseudotypes examined, only 32 were resistant to both MAbs (IC50
>50 μg/ml) [44
]. In the present study, 1 of these PG9/PG16-resistant isolates was tested against sCD4-17b and was found to be highly sensitive (QH0692.42, Clade B, IC50
= 0.66 μg/ml, Fig. ). It will be most interesting to test the sCD4-17b sensitivities of other strains resistant to these newly described MAbs.
Our previous observation of limited breadth in the MAGI-CCR5 system was not due to the requirement of a longer L1 linker for some strains, as we previously speculated. Nor was it associated with the use of infectious virus in that system, since the TZM-bl assay demonstrated sCD4-17b sensitivity for infectious virus from several isolates (Fig. , Fig. ); where tested, the potency was comparable to that observed with the corresponding Env pseudotyped particles (Fig. ). Retesting one of the previously described resistant isolates, namely 91US054, in the MAGI-CCR5 assay suggested that the limitation of neutralization breadth in our earlier study might have resulted from insufficient concentration of sCD4-17b (unpurified or partially purified), coupled with the higher virus input volume required in the MAGI-CCR5 assay compared to the TZM-bl assay (20 μl versus 5 μl, respectively). We believe the present findings of extremely broad neutralization activity of sCD4-17b in the commonly used TZM-bl assay override the limitations noted in our previous report, which were most likely due to the specific conditions associated with those experiments rather than to inherent properties of the sCD4-17b protein or the Env glycoproteins that it targets.
The HIV-1 IMC experiments demonstrated a marked contrast between sCD4-17b and neutralizing antibodies with respect to the influence of the cellular source from which the virus particles were derived. The potency of sCD4-40
-17b was equivalent against virions isolated directly from the transfected producer cell line and progeny virions obtained after single passage through PBMCs (Fig. ). By contrast, the PBMC-passaged virions were significantly less sensitive to the broadly neutralizing MAbs, as previously shown by the group that provided the IMCs for our experiments; importantly, they also demonstrated the absence of any sequence change in Env after PBMC passage [30
]. In that earlier report, several possible explanations were offered for the reduced antibody susceptibility of PBMC-passaged viruses compared to their cell line-derived counterparts, including higher levels of Env on virions from PBMCs, differential incorporation of host cell factors such as adhesion proteins during virus assembly in PBMCs, and producer cell-dependent biochemical differences in the gp160, such as differential glycosylation. In our opinion, the first two mechanisms might be expected to result in neutralization differences that are similar for various classes of Env-blocking agents, and thus would not readily explain the lack of effect of sensitivity to sCD4-17b and sCD4 despite the strong effects on neutralizing antibody sensitivity. Differences in glycosylation seem an appealing possibility for several reasons. First, the decreased antibody sensitivity upon PBMC passage was most dramatic for the 2G12 MAb (Fig. , also [30
]), whose epitope on gp120 consists of large mannose-rich carbohydrate clusters that can be recognized by an unusual domain-swapped structure of the antibody [45
]. Second, gp120 is noted for its extensive "glycan shield' that continually evolves in the infected host to sterically block neutralizing antibodies directed against non-carbohydrate epitopes without affecting receptor binding [46
], since the CD4 binding site and the bridging sheet are devoid of carbohydrate [10
]. Indeed there is precedence for variations in glycosylation patterns of isogenic HIV-1 Envs dependent on the producer cell type [47
]. The potent activity of sCD4-17b against PBMC-produced virions is critical, since these presumably reflect the properties of in vivo
particles more closely than do the cell line-derived virions.
The broad, potent antiviral activity of sCD4-17b suggests several possible applications. Passive immunotherapy with MAbs has been examined both in nonhuman primate models and human clinical trials [48
], and sCD4-17b could be considered as an additional component to a mixture of broadly neutralizing MAbs; however the practical limitations of passive immunotherapy for treating chronic HIV infection greatly reduces enthusiasm for this mode of use. A related alternative would involve gene therapy strategies using either viral vectors [49
] or engineered hematopoietic stem cells [50
] to continually produce sCD4-17b in the body, for treatment or protection against HIV infection. For such applications, it is likely that the molecule would need to be modified for enhanced plasma half-life by linking it to immunoglobulin constant regions, as has been done for sCD4 [51
]; indeed such a modification might provide the additional advantage of increased potency due to multivalent binding. Perhaps a more likely antiviral application of sCD4-17b would be as a topical microbicide to prevent sexual transmission. Several classes of proteins and peptides targeting either the virus or receptors on the host cell are being actively studied for this purpose [52
], encouraged by technologies to manufacture candidate proteins on an economically viable scale [54
]. A particularly intriguing approach involves genetic modification of commensal bacteria native to the healthy vaginal or rectal mucosa to produce the anti-HIV proteins in situ
]. For example, vaginal strains of Lactobacillus
producing CD4 in either secreted [56
] or surface-bound [57
] forms have been described; this "live microbicide" concept is being investigated with various potent anti-HIV proteins and peptides [58
], and preliminary efforts have been undertaken for sCD4-17b (L. Lagenaur and E. Berger, unpublished). For any of the applications suggested above, sCD4-17b has significant advantages compared to some other candidate proteins in that it is highly specific for HIV, and is composed of entirely human-derived sequences (except for the linkers). Thus problems associated with immunogenicity and induction of inflammatory responses are predicted to be relatively minor. We propose that sCD4-17b warrants continued investigation in the ongoing efforts to develop new antiviral strategies to combat the HIV/AIDS pandemic.