We used three increasingly sensitive DNA sequencing methods – SGA, PASS and 454 – to look for genetic evidence of Nab selection on the evolving HIV-1 quasispecies. By three to six months post-seroconversion, SGA sequencing identified a set of candidate Nab escape mutations, which in every subject was discontinuous and could be distinguished from CTL escape mutations 
. Each of the candidate Nab escape mutations that we inferred from SGA sequencing was shown phenotypically to confer significant (2 to >70 fold) resistance to early Nabs (). Remarkably, at the time of initial detection of Nab titers, regardless of titer, the virus quasispecies in each subject demonstrated complete or near complete replacement of the T/F sequence by escape mutants at their respective Nab epitopes. This indicated a pre-existent Nab response. PASS analysis corroborated this finding by revealing genetic evidence of Nab escape significantly earlier at just 45 days and 32 days post-antibody seroconversion in subjects CH40 and CH77, respectively (Tables S1
). This was at a time point when Nab titers to each T/F virus were undetectable at a 1
20 plasma dilution in the TZM assay (). Nabs at this early time point were also below the level of phenotypic detection when tested in the sensitive A3R5 cell-based virus entry assay 
(D. C. M., unpublished). In subject CH40, where viral loads were highest and deeper sequencing could be done, 454 analysis identified a much larger number of variants in the V1 epitope region of the T/F virus sequence at a still earlier time point 16 days post-seroconversion as well as at 45 and 181 days post-seroconversion (). The 454 data further suggested a role for APOBEC mutations facilitating this escape, since at days 16 and 45 post-seroconversion the V1 Nab epitope region was enriched for mutations at APOBEC motifs (). APOBEC and Vif function have been implicated in virus escape from early CTL immune pressure 
, and our results suggest that APOBEC may play an analogous role in the dynamics of early Nab escape at certain epitopes. It is possible that the increased genetic diversity in V1 arising from APOBEC mediated polymorphisms facilitated more rapid escape in this region than in other regions of the Nab epitope. Overall, a combination of SGA, PASS and 454 pyrosequencing enabled the genetic detection of Nab escape variants significantly before Nabs rose to titers detectable in the TZM assay. This enhancement in detection amounted to 95 days in subject CH40, 70 days in subject CH77, and 109 days in subject CH58. Future studies in other subjects with more narrowly spaced sampling intervals may better define these windows.
In each subject we found an early monospecific Nab response directed toward a single conformational epitope that was unique to each T/F virus strain. This was demonstrated most clearly in subject CH40 where polyclonal plasma antibodies and autologous mAbs targeted essentially the same epitope at the base of V3 (depicted in ). Previous studies have reported epitopes in this region of gp120 to be immunogenic and a target of both broadly and narrowly reactive neutralizing mAbs 
. Interestingly, we observed that the binding of both AbCH83 and AbCH84 to autologous CH40 Env gp140 as assessed by Biocore analysis could be blocked by the potent and broadly neutralizing PGT 121 mAb, which in other contexts is dependent on N332 
(B.F.H., unpublished). The early Nab response in CH40, unlike responses in CH77 and CH58, targeted an epitope dependent on trimeric Env for structural integrity. Thus, structural modeling and empirical analyses suggested that in subject CH40, virus escaped Nab pressure indirectly by early mutations in V1 and directly through mutations in the putative V3 epitope, indicating a close association between V1 and V3 in the context of the native functional Env trimer. In all three subjects, we identified Nab epitopes involving unique sites on the Env glycoprotein, with continuous virus evolution at the respective epitopes, without evidence of broadening of the Nab response to additional sites on the Env trimer over the first year of infection (–). In CH77, escape occurred predominantly in V2, where the addition of PNLG site conferred Nab escape at a likely protein epitope. In CH58, modeling suggested that early Nabs targeted a single conformational epitope involving the Env outer domain, with escape arising through the loss of any of several component glycans. Thus, in each subject, virus employed glycan shifts as well as gain or loss of glycans to mediate escape from the sequential rounds of the Nab response. These findings, in conjunction with reports of monospecific early Nab responses in subtype C infection 
, suggest that individual immunodominant regions of Env, specific to the unique conformation of each T/F Env, are targeted by early Nab responses.
The observation that very low level Nab titers can impede virus entry and select for virus escape in vivo
is consistent with recent findings of selection for SIVmac251 and SIVsmE660 Nab escape mutations in early-chronic infection of rhesus macaques by low titers of Nabs 
, the association of low-titer Nabs with protection against SIVsmE660 challenge in the nonhuman primate (NHP) model 
, and results from low dose mucosal NHP challenge models in which concentrations of Nabs corresponding to modest in vitro
titers were able to effectively prevent SIV acquisition 
. To our knowledge, however, this is the first demonstration in human HIV-1 infection that very low titers of Nabs in the range of 1
50 to 1
20 or even lower can impede virus replication and select for virus escape. To explore quantitatively the in vivo
activity of early Nab responses, we employed a mathematical model to estimate the proportion of de novo
infection events blocked by Nabs, or the Nab efficacy (see Figures S3
and Dataset S1
). The results, which represent minimum estimates, ranged from a low of 19.6% to a high of 35.2% and represent a Nab response that is sufficiently potent to drive replacement of the T/F virus within several weeks (Figure S4
and unpublished data). Our conservative modeling likely underestimated true Nab efficacy because we utilized minimum estimates for biological parameters with uncertain quantities and did not account for potential fluctuations in Nab efficacy. Future studies where sampling time points are better structured for evaluating dynamic changes in Nab titers and viral quasispecies composition would allow for greater precision in estimations of Nab efficacy in vivo
and a better understanding of the kinetics of Nab development.
These caveats notwithstanding, the data raise the possibility that in the setting of sexual transmission, where the risk of infection per coital act is low and the number of transmitted viruses responsible for productive clinical infection is typically one, a vaccine that elicited Nabs of sufficient breadth but at titers as low as 1
50 to 1
20 or possibly even lower could have a demonstrable protective effect. It is possible that such a low titer neutralizing activity in vaccinees from the Thai RV144 trial could have contributed to the observed 31% protective effect of the vaccine 
The rates of Nab-driven T/F sequence replacement are more rapid than previously reported 
but substantially slower than rates of loss due to the initial CTL responses 
A unique aspect of the present study is that we could directly compare the rate of Nab escape with the rate of CTL escape in the same three subjects 
. Based on SGA analyses, we previously observed virtually complete replacement of the T/F virus population at defined CTL epitopes within 45 days (CH40), 14 days (CH77) and 45 days (CH58) of antibody seroconversion 
. This contrasts with 111, 102 and 154 day intervals shown in the present report for Nab escape. Similarly, in a 454 pyrosequencing analysis of CTL escape kinetics, we previously observed in subject CH40 a 1% replacement of T/F sequences just prior to antibody seroconversion (corresponding to day 0 in the present study), a 52% replacement by day 16, and a 99.4% replacement by day 45 
. Comparable numbers for Nab escape variant frequencies in subject CH40 in the present study were <1%, 2% and 3%, respectively, again highlighting the much faster rate of CTL escape compared with Nab escape. Furthermore, in the former study, we found that the average rate of HIV escape from CTL responses in acute infection to be 0.17 day−1
with a maximum of 0.42 day−1
. The average rate drops to 0.03 day−1
by 100 days post seroconversion 
. This slower rate of virus escape from chronic CTL responses is similar to that of contemporaneous Nab responses measured in the current study, suggesting that Nabs could contribute along with CTLs to virus containment during this later time period. In acute infection of unvaccinated subjects, however, Nab responses likely contribute negligibly to early virus containment.
The costs to replication fitness associated with virus escape from autologous Nab responses have not been well characterized but could contribute to partial virus containment at setpoint viremia. Fitness costs of Nab escape mutations have frequently been considered to be minimal 
, but Derdeyn and colleagues described a Nab escape mutation in V2, which when placed in the autologous T/F virus backbone, conferred a measurable fitness cost 
. Morris and colleagues 
similarly noted transient decrements in plasma virus load coincident with the development of strain-specific Nabs. We studied Nab escape mutations within the context of a 6 month consensus IMC so the effects of compensatory mutations could be accounted for and so mutations resulting from escape from Nabs could be distinguished from those resulting from escape from CTLs. Our analyses suggested that Nab escape mutations conferred reductions to replication fitness ranging from 0 to 24%. This corresponds to an estimated average impairment to virus entry due to early strain-specific Nabs of as much as 31.3% to 48.8%.
Finally, we note that the exquisite sensitivity and rapid adaption of HIV-1 Env to Nabs contrasts with recent observations for the HIV-2 Env, where high-titer Nabs register little effect on env
evolution or Env Nab escape 
. A biological explanation for these differences is not obvious. For HIV-1, the enhanced sensitivity and rapid adaptation to Nab pressure in vivo
provides an explanation for the HIV-1 Env's propensity to maintain a fully assembled glycan/conformational shield 
. Paradoxically, it is this enhanced sensitivity of HIV-1 to Nabs in vivo
that appears to be responsible for its vaunted ability to resist neutralization by all but the most broadly reactive and potent Nabs. Another provocative implication of the current study is that, in vivo
, Nabs impede HIV-1 spread whether this is occurring by ‘cell-free’ or ‘cell-to-cell’ mechanisms. This is at odds with the suggestion that ‘cell-to-cell’ spread of HIV-1 provides a mechanism for replicating virus to escape Nab or antiretroviral drug pressure 
. Further investigation is needed to resolve this question.