Recent studies have shown that ~20% of subjects chronically infected with HIV-1 mount broadly neutralizing antibody (bnAb) responses and that 1 to 3% of subjects have exceptional magnitude and breadth of neutralizing activity (level 3 neutralizers) (4
). It is unclear if such plasma neutralization breadth is mediated by many (17
) or only one or a few (23
) antibody specificities. A number of bnAbs have been isolated from HIV-1-infected individuals, and these include those directed to a variable loop 1 and 2 (V1V2) conformational (quaternary) epitope, to the CD4-binding site (CD4bs), and to outer domain glycans, all on the gp120 surface unit of the HIV-1 envelope glycoprotein (2
). Additional bnAbs target the membrane-proximal external region (MPER) of gp41 (13
). While evidence suggests that bnAbs to the gp41 MPER are limited by tolerance mechanisms, the V1V2 conformational and CD4bs antibodies are generally less polyreactive (2
). However, they both present unusual characteristics: the V1V2 conformational bnAbs have long heavy-chain complementarity-determining region 3 (HCDR3) (2
), whereas the CD4bs bnAbs display a high degree of somatic mutation (10
) and appear to derive from restricted VH
gene families (18
). Strategies that allow for highly specific serologic and/or neutralization assays to determine the epitopes of plasma bnAbs are now well established (1
), and recent studies suggest that a single HIV-1-infected subject can make bnAbs of multiple specificities (20
). If both V1V2 conformational and CD4bs antibodies could be isolated from the same individual and demonstrated to recapitulate the serum neutralizing activity, this would provide direct evidence in support of a polyvalent bnAb HIV-1 vaccine strategy.
For this study, we selected a chronically infected individual (CH0219) whose plasma displays extraordinary broad and potent neutralization and includes antibody specificities directed against MPER, CD4bs, CD4-induced (CD4i), gp120 core, and V1V2 conformational epitopes (20
). While responses against the CD4bs and a PG9-like V1V2 conformational epitope appeared to be responsible for most of the breadth, only a limited number of mapping reagents were available to confirm this observation (20
). Because polyvalent neutralizing antibody responses may be a key consideration for HIV-1 vaccine development, and considering the exceptional breath of serum neutralization in this subject, we interrogated the IgG+
memory B-cell repertoire of donor CH0219 to isolate and characterize the antibodies that recapitulated serum neutralization.
By using a clonal memory B-cell culture system (2
), we previously identified four bnAbs (CH01, CH02, CH03, and CH04), members of the same clonal lineage, binding to a V1V2 conformational epitope (2
). The CH01 through CH04 bnAbs neutralized 36% to 46% of 91 cross-clade HIV-1 isolates, which represented a subset of strains also neutralized by PG9 (2
). Another clone of five CD4bs-specific bnAbs (VRC-CH30, VRC-CH31, VRC-CH32, VRC-CH33, and VRC-CH34) was isolated from the same donor by antigen-specific B-cell sorting of individual IgG+
memory B cells reactive with RSC3 but not RSC3Δ371 (25
). VRC-CH30 through VRC-CH34 neutralized 75% to 95% of a multisubtype panel of viruses with breadth comparable to that of VRC01 (25
). shows the phylogenetic tree of the VRC-CH30 to -CH34 bnAb clonal lineage. Analysis of the V(D)J rearrangements of the CH01 to CH04 and VRC-CH30 to -CH34 sequences demonstrated that the two clones use distinct VH
3 and VH
1, respectively) (), that the frequency of somatic mutations of the VRC-CH30 to -CH34 VH
chains (23 to 24%) is approximately twice that of CH01 to CH04 (12 to 14%) (), and that, conversely, CH01 to CH04 have substantially longer HCDR3s (24 amino acids [aa] versus 13 aa, according to the Kabat numbering system [7
]) (). These data demonstrate that the two clones did not share a genetic background, relative to VH
genes, and suggest that they likely evolved independently.
Fig 1 Phylogenetic tree of the VRC-CH30 to VRC-CH34 monoclonal antibodies. The tree shows the evolutionary distances of the V(D)J nucleotide sequences of the VRC-CH30 to VRC-CH34 monoclonal antibodies and is rooted on the nucleotide sequence of the unmutated (more ...)
Characteristics of the V-heavy and V-light chains of the monoclonal antibodies CH01 to CH04 and VRC-CH30 to VRC-CH34 isolated from memory B cells of donor CH0219
We have previously shown that the neutralization profiles of the individual bnAbs isolated within each of these two clones are comparable (2
). Therefore, we used CH01 and VRC-CH31 as representative of the respective clonal lineages to ask if these two bnAbs, either alone or in combination, could recapitulate the breadth of serum neutralization seen in donor CH0219. Neutralization assays using TZM-bl target cells were performed as previously described (2
). shows the 50% inhibitory concentrations (IC50
; μg/ml) of bnAbs CH01 and VRC-CH31 and their combination at a 1:1 ratio and the 50% inhibitory dose (ID50
; 1/dilution) titer of the donor CH0219 serum against a multiclade panel of 97 Env-pseudotyped viruses (pseudoviruses), including tier 1, tier 2, and transmitted/founder HIV-1 strains from clades A, B, C, AG, AE, G, and D. The serum neutralized 91/97 (94%) pseudoviruses. CH01 neutralized 43/96 (45%) of the tested pseudoviruses. VRC-CH31 had broader neutralization than CH01, with 81/97 (84%) of HIV-1 strains neutralized. Thirty-three pseudoviruses were neutralized by both monoclonal antibodies (MAbs), indicating that both epitopes were coexpressed on the virion surface in a conformation recognized by the cognate antibody. Conversely, 41 pseudoviruses were neutralized by VRC-CH31 but not by CH01, and 5 were neutralized by CH01 only. As suggested by previous data on plasma (20
), the spectra of neutralization of each of the two MAbs alone did not recapitulate the breadth of serum neutralization (). However, when CH01 and VRC-CH31 were combined at a 1:1 ratio, they neutralized 86/91 (95%) of HIV-1 strains neutralized by donor CH0219 serum (indicated with the symbol
in ). The five HIV-1 strains that were not neutralized by the combination of CH01 and VRC-CH31 bnAbs, AE.NP03, D.3016, G.X2088, C.Du422, and B.62357 (indicated with the symbol × in ), showed relatively weak serum neutralization (ID50
values of 294, 282, 61, 27, and 23, respectively; ). Indeed, the ability of donor CH0219 to generate additional antibody specificities with some degree of HIV-1 neutralizing activity was formally proven by the isolation from IgG+
memory B-cell cultures of anti-V3 loop (CH19) and anti-gp41 (CH11) MAbs with narrow spectra of neutralization: CH19 neutralized exclusively clade B tier 1 viruses, and CH11 neutralized only two tier 2 viruses out of 97 tested (data not shown). While the CH11 and CH19 antibodies neutralized pseudoviruses that were also neutralized by the combination of CH01 and VRC-CH31, it is plausible that other antibody specificities with restricted breadth of neutralization were made that account for the 5% of serum neutralization missed by the combination of CH01 and VRC-CH31. The combination of CH01 and VRC-CH31 weakly neutralized three pseudoviruses, B.6240, AE.703357, and C.ZM53 (IC50
s of 10, 18, and 7.7 μg/ml, respectively), that were not neutralized by the serum (ID50
, <20; indicated with a Δ in ), suggesting either that the CH01 and VRC-CH31 serum antibody concentrations were not sufficient to generate detectable levels of neutralization for these strains or that there was some type of interference in epitope recognition by other nonneutralizing antibodies.
Fig 2 Comparison of the neutralization profiles of the serum of donor CH0219 and MAbs CH01 and VRC-CH31 either alone or in combination. Neutralizing activity was tested on a panel of 97 Env-pseudotyped lentiviruses, which comprised tier 1 and tier 2 isolates (more ...)
Overall, the numbers of HIV-1 strains neutralized by the serum and the combination of CH01 and VRC-CH31 were strongly associated (P
= 0.007; Fisher's exact test) (A). Also, the neutralization titers of the serum and the combination of CH01 and VRC-CH31 strongly correlated (R
= 0.7, P
< 0.0001; linear correlation) (B). We previously demonstrated the substantial contribution of V1V2 conformational epitope and CD4bs antibodies to CH0219 serum neutralizing activity (20
). To evaluate the levels of serum antibodies with CH01-like and VRC-CH31-like specificities, we measured the ability of CH0219 serum to compete binding of biotin-labeled CH01 and VRC-CH31 MAbs to the E.A244 gp120 envelope glycoprotein (16
) and calculated their concentrations based on standard curves measuring the blocking obtained with cold CH01 and VRC-CH31 MAbs. At a 1:12.5 dilution, CH0219 serum blocked 74.3% and 50.1% of CH01 and VRC-CH31 MAb binding, respectively (data not shown). The serum concentrations of CH01-like and VRC-CH31-like antibodies were 70.4 ± 8.1 (mean ± standard deviation) CH01 μg/ml equivalents and 42 ± 10.5 VRC-CH31 μg/ml equivalents, respectively. In both cases, the serum levels of CH01-like and VRC-CH31-like antibodies estimated from standard curves of CH01 and VRC-CH31 MAbs are higher than the concentration needed to mediate neutralization in the TZM-bl assay (). These findings suggest that, within the context of the polyclonal anti-envelope response, the combination of CH01 and VRC-CH31 is responsible for the majority of the neutralizing activity observed in the serum.
Fig 3 Correlation analysis of neutralization breadths and titers of donor CH0219 serum and the combination of CH01 and VRC-CH31 MAbs. (A) Contingency table showing the numbers of isolates neutralized by serum of donor CH0219 and/or the combination of CH01 and (more ...)
Remarkably, the combination of CH01 and VRC-CH31 bnAbs achieved a near panneutralization (89/97; 92%) of the panel of Env-pseudotyped viruses, which is broader than the neutralization achieved by the two bnAbs alone (A and ). The combination of CH01 and VRC-CH31 bnAbs, tested at a 1:1 ratio, did not display any appreciable additive or synergistic effect on the magnitude of neutralization: in fact, 87 of the 89 viruses (97.8%) were neutralized by the combination of CH01 and VRC-CH31 bnAbs with a titer comparable to the most potent of the two bnAbs (). However, for 22 viruses, the neutralization titer of a single bnAb was already at or below the lower limit of detection (IC50 = 0.02 μg/ml), and in these cases, a synergistic or additive effect could not be ruled out. In only one case was the IC50 of the bnAb combination >1 order of magnitude higher than that of the most potent bnAb (for A.191955, the CH01 IC50 was <0.02 μg/ml and the CH01 plus VRC-CH31 IC50 was 0.1 μg/ml; ). Therefore, we conclude that the combination of CH01 and VRC-CH31 bnAbs does not have detrimental effects on the magnitude of neutralization of each of the individual antibodies and that the combination broadens the overall spectrum of in vitro neutralization.
The findings presented here are relevant for vaccine development design in three aspects: first, the clones of the two bnAbs produced by subject CH0219 came from different VH genes and did not present similarities in terms of accumulation of somatic mutations or HCDR3 length; second, the combination of two bnAbs from each of the two clones is sufficient to confer near-pan-HIV-1 neutralization; and, third, our data suggest that vaccine formulations capable of inducing bnAbs of both specificities will have an advantage over monovalent formulations in inducing broader neutralization.
In conclusion, we present the first isolation of two clonal lineages of bnAbs with distinct specificities from memory B cells of a single individual. Two bnAbs from these clones largely recapitulate the breadth of the donor's serum neutralization and achieve near pan-HIV-1 neutralization, implying that a vaccine capable of inducing bnAbs against both the CD4bs and the V1V2 conformational epitope could achieve broader HIV-1 neutralization than a vaccine inducing only one of the two specificities. These observations demonstrate that the concurrent development of bnAbs targeting diverse regions of the gp120 envelope glycoprotein in the memory B-cell compartment of a single subject is possible. Such in vivo cooperative regulation of two different bnAbs provides proof of concept supporting the design of polyvalent vaccines.