In this study, we tested the ability of antibodies present in superinfected and singly infected women to neutralize a spectrum of circulating HIV-1 variants and thus, discern whether antigenic stimulation by two viruses compared to one has an effect on the subsequent NAb response. Our results demonstrate that the NAb response of superinfected women is significantly broader and more potent than that of singly infected women when compared at matched time points. These data suggest that SI elicits a substantial enhancement of the NAb response with regards to cross-reactivity in the years following reinfection. This conclusion is supported by our analysis controlling for NAb breadth/potency prior to SI and clinical measures that are associated with NAb breadth such as CD4+ T cell count and viral load 
. Therefore, the study of superinfected individuals may yield additional insights into the development of broad and potent NAb responses to diverse HIV-1 antigens.
The majority of superinfected individuals exhibited breadth and potency scores that were greater than the average of their matched controls post-SI. Among the SI cases, the three women with the most potent responses (QB850, QA013, QC885) experienced intersubtype superinfections that were characterized by persistence of both the initial and superinfecting viruses. While we attempted to parse out which factors relating to SI may be influencing the generation of a broad response, we were unable to do so conclusively due to a limited sampling of 12 cases (data not shown). However, the characteristics of these three women suggests that continuous stimulation by two distinct viruses from different subtypes may be critical to the induction of broad NAbs. Still, other factors, including the antigenic nature and replication potential of the infecting viruses, may also be important. Conversely, we found a minority of superinfected individuals (QB045, QC858, QD022) exhibited breadth and potency scores that were lower than that of the average from their matched controls. As all three had a CD4+ T cell count >400 post-SI, it is unlikely that lower breadth and potency simply reflected a lack of T cell help that could arise at terminal stages of infection. Some potentially informative features of these cases are that two of the three (QB045 and QD022) are the only individuals in the cohort who became superinfected late after initial infection (>4 years post-initial infection), suggesting that perhaps the timing or duration of either or both infections may be critical to the development of NAb breadth. Moreover, in QD022 and QC858 the superinfecting variant became dominant as the initial virus could no longer be detected post-SI, at least at levels captured by Sanger sequencing, suggesting that continued antigenic stimulation by both infecting viruses may be important in augmenting the NAb response. Additional studies that include deep sequencing of viruses in tissue and plasma would be one starting point to further explore this hypothesis.
In one individual, QA413, multiple Tier 2 viruses were neutralized prior to SI, while the Tier 1 viruses (SF162 and Q461d1) were not. This may suggest that there are unique epitopes common to the two Tier 1 viruses that are not present in the Tier 2 viruses or vice versa. In the case of Q461d1, the envelope is highly sensitive to neutralization because of conformational changes that expose multiple epitopes 
. However, we have found that Q461d1 is relatively insensitive to neutralization by the PGT-type antibodies that recognize a quaternary structure that includes an Asn at amino acid position 160 (unpublished). SF162 is also not recognized by PG9 and certain PGT antibodies because it encodes a Lys at position 160 
. Interestingly, the virus that initially infected QA413 encodes an Asn at position 160, while the superinfecting virus encodes a Lys, perhaps suggesting that PGT-like antibodies could contribute to the antibody response in this woman. Mapping studies to define the epitope specificity will be needed to understand the unusual pattern of virus neutralization observed in QA413 pre- and post-SI.
In this study, we examined breadth and potency after SI against a panel of heterologous viruses. In a prior study, the responses to autologous viruses were examined in five of these individuals 
. Interestingly, one of the elite neutralizers, QA013, developed very high titers against her superinfecting virus, with IC50 values >1,800 (range: 1,000–25,000+) to all four SI variants cloned at ~5 years post-SI (6.3 years post-initial infection). The other four superinfected individuals, who did not develop elite responses, had autologous responses that ranged from <100 to 10,000. There was insufficient data to determine if there was an association between autologous and heterologous responses.
It is difficult to directly compare the NAb responses of the superinfected women to broad neutralizers identified in other studies because there is no standard for quantifying NAb breadth against HIV-1. We addressed this issue by using a diverse panel of viruses weighted towards Tier 2 variants and several alternate breadth scoring methods, which showed that no single virus drove the association observed between SI and NAb breadth and that our conclusion is the same irrespective of the scoring method used. Simek et al. previously defined elite activity as “the ability to neutralize, on average, more than one pseudovirus at an IC50 titer of 300 within a clade group and across at least four clade groups 
.” While we were unable to completely satisfy these criteria since our 8-virus panel included only single variants of subtypes B and D, we still found that the two superinfected individuals (QA013 and QB850) with the broadest responses could neutralize at least two viruses in subtypes A and C, as well as both single viruses tested from subtypes B and D at an IC50 titer greater than 300, supporting the characterization of these individuals as elite neutralizers. Notably, plasma antibodies from QB850 neutralized viruses from all four subtypes tested at IC50 values greater than 600, more than 2-fold higher than the bar set for elite neutralizers 
. Furthermore, the responses of these two women were greater than those found in a similar screen of 70 singly infected women in this cohort at ~5 years post-initial infection 
, with some observed IC50 titers against Tier 2 viruses 6-fold more potent than those of the top 10% of the singly infected women (data not shown). Although we had limited opportunities to directly compare the responses of the superinfected individuals studied here and individuals identified as elite neutralizers by Simek et al. in the IAVI cohort, we did find that the SI cases had IC50 values that were either comparable or above the cohort GM IC50 for the three viruses tested in common between the studies. Together, these data suggest that 2 of the 12 individuals from our SI cohort developed NAbs that exhibit elite activity. This is a remarkable fraction (17%) of individuals with elite activity, although because of differences between screening methods, it is difficult to compare this to the 1% of presumably singly infected elite neutralizers reported previously by Simek et al.
In a prior study that examined four cases of SI, a greater increase in NAb breadth post-SI among superinfected individuals compared to singly infected individuals was also observed 
. However, it is hard to compare their data with our own, as the previous study used randomly chosen primary isolates for measuring breadth in a cohort with unknown seroconversion dates. It is also interesting to note that this previous study observed a decrease in viral load in three of the four superinfected cases studied at a time point post-SI, two of which had undetectable viral loads. Such a significant drop in viral load has not been previously reported for cases of SI 
. In contrast, we observed an increase or no change in viral load in the majority of superinfected women examined. This is consistent with the observation that viral load is highly correlated with NAb breadth 
. However, after adjusting our breadth score analysis for contemporaneous viral load, the estimate of 1.68 was unchanged, while adjusting for this variable caused an increase in our estimate for differences in potency. This would imply that breadth in these superinfected cases alone cannot be explained entirely by an increase in viral load following SI, but rather suggests that stimulation from antigenically distinct viruses may contribute to the development of potency.
Finally, we found evidence of cross-subtype breadth, including detectable neutralization of viruses from four different subtypes in the two women with elite NAb responses within 1 year of their SI. Because both of these individuals were superinfected soon after their initial infection, these cross-subtype responses arose relatively soon after HIV-1 seroconversion. Indeed, by 2 years after their initial infection, both women had antibodies capable of neutralizing 7 of the 8 primarily Tier 2 viruses tested. Recent studies suggest that cross subtype breadth is rare before 2 years post-infection in individuals who are presumably singly infected 
. Our findings raise the interesting possibility that some of the individuals identified as having broad responses in prior screens may have been superinfected.
A few caveats to these findings must be considered. First, there may have been potential for misclassification of singly infected women, due to our limit of detection or if recombination occurred between the initial and superinfecting strains in HIV-1 genomic regions outside of gag
. However, this misclassification would be expected to decrease our ability to detect differences between superinfected and singly infected women, making it more likely that the true association between SI and NAb breadth is stronger than what we observed. Also, we cannot exclude the possibility that there are other factors involved with the development of the broad responses in some of the superinfected women.
This study reveals an unexplored source of naturally-occurring broadly NAbs, and represents a highly relevant approach to inform vaccine strategies in three key ways. First, studies on this and other SI cohorts may provide additional support for immunizing with particular combinations of different HIV-1 strains could be an effective vaccine approach 
. Given that the greatest breadth was observed in cases of intersubtype SI, the use of Env immunogens from different subtypes may be optimal. Second, the NAbs and viruses isolated from members of this cohort may hold important clues to antigenic determinants capable of eliciting cross-reactive antibodies that can protect against multiple subtypes of HIV-1. Third, longitudinal studies of superinfected individuals who develop broad and potent NAb responses, such as those identified here, may foster an understanding of the mechanism leading to the elicitation of breadth. Of particular interest is the observation that two elite neutralizers developed broad and potent responses soon after infection by a second HIV-1 strain, suggesting that the processes that lead to the development of broad HIV-specific NAbs are accelerated by a second infection. If SI ultimately leads to the rapid capacity of the overall NAb response to recognize diverse circulating HIV-1 variants, a successful vaccination strategy that mimics natural SI may lead to the development of broad NAb in immunized individuals.