rAd vectors are widely used in the field of gene therapy and vaccination due to their efficiency for gene delivery and ability to stimulate cellular and humoral immune responses. However, the potential effects of pre-existing immunity to Ad on vector delivery and vaccine efficacy raise complications for clinical implementation. In the Phase IIB Step trial, the high rates of infection in the vaccine group, along with the lack of protective efficacy, led to termination of the study. These concerns highlight the need to understand the role of anti-vector immunity and its impact on efficacy. Such pre-existing immunity against the vector can blunt the efficiency of gene delivery and may cause side effects on disease outcomes. Despite substantial effort, no causal link has been established between Ad5 seropositivity, the MRKAd5 vaccination, and the risk of HIV acquisition, although retrospective analyses suggest a higher risk of infection after vaccination in Ad5 seropositive, uncircumcised participants 
Several hypotheses have been proposed to explain this trend. One is that the rAd5 vector may activate pre-existing Ad5-specific CD4 T cells to express HIV co-receptors in Ad5 seropositives 
. However, it was shown that there is no correlation between Ad5 serostatus and the presence of anti-Ad5 T-cell responses in humans 
. Furthermore, there was no elevated, long lasting reactivation of Ad5-specific CD4+ T-cell responses after rAd5 vaccination in seropositive populations 
. Therefore, it is unlikely that Ad5-specific CD4+ T cells reactivated by vaccination resulted in increased HIV targeting cells in the systemic compartment. Recent studies in monkeys and in Step participants have also suggested that there is no greater reactivation or trafficking of CD4+ T cells to mucosal compartments in Ad5 seropositives compared to Ad5 seronegatives post-vaccination 
. Another hypothesis is that anti-Ad5 neutralizing antibodies can form immune complexes with rAd5 vectors and dendritic cells which can enhance HIV infection in T cells. Although this possibility was demonstrated in vitro
, it is unknown how it correlates to the responses found in the high risk group of the Step trial.
In this study, we analyzed the specificity of anti-Ad5 Nab in Step trial participants and determined whether differences in specificity correlated with Ad5 seropositivity and risk for HIV infection. The seroprevalence reported here for Ad5 is higher than assayed by HVTN as 9 out of the 60 seronegatives were seropositive by our assay. This is likely due to the use of EDTA for harvesting cells rather than trypsin, which can cleave CAR and compromise physiological functions that require intact CAR protein 
. Trypsin may also cleave other cellular receptors, including integrins that affect virus uptake 
. Although the proteolysis of CAR by trypsin is limited to one cleavage site, trypsin treatment may affect the sensitivity of the assay to detect neutralizing antibodies to other rAd that use trypsin-sensitive cellular receptors 
. The results show that the MRKAd5 HIV vaccine efficiently converted all the Ad5 seronegative participants to seropositivity after a single vaccination. Thus, pre-vaccine serostatus was not maintained, even at four weeks after the first vaccination. Our data also suggest that vaccination induced Nab to the capsid of Ad5 more efficiently than to the fiber. This result is consistent with a previous study that showed differences in the specificity of Ad5 Nab from natural infection and vaccination with rAd5 vectors 
. Vaccination generated Nab to Ad5 capsid more efficiently than to fiber, while natural infection generated Nab to both Ad5 fiber and capsid. This may be due to the fact that the rAd vaccines are non-replicating and contain more capsid proteins than fiber proteins. In the Step trial, we show here that one vaccination was sufficient to convert all vaccinees to Ad5 capsid seropositivity, while three vaccinations converted them all to Ad5 fiber seropositivity. Therefore, pre-vaccination serostatus affected post-vaccination neutralizing antibody profiles and the quantity of anti-Ad5 fiber Nabs. There is no evidence to suggest that such quantitative differences in the titers of anti-Ad5 fiber Nab between these two groups correlated with HIV infection risk in the participants.
There were no significant differences in the titer of Ad5 neutralizing antibodies between HIV-infected and uninfected participants, and no significant differences among HIV-infected participants in the vaccine and placebo groups (Figure S1
). These data are consistent with a recent report showing that anti-Ad5 neutralizing antibody is not associated with risk of HIV infection in the absence of vaccination 
. On the other hand, further analysis of HIV-infected participants showed that their sera tend to contain lower anti-Ad5 capsid Nabs compared to that detected in uninfected participants. It seems that anti-vector immunity differed qualitatively in participants who acquired HIV infection compared to those who did not.
This finding highlights the fundamental differences in these participants in terms of how their immune systems recognize multiple components of viral antigens. Although our assay did not distinguish anti-hexon and anti-penton antibodies targeting the viral capsid, previous studies have demonstrated that anti-hexon and anti-fiber antibodies play a major role in mediating neutralization in Ad5 seropositive people and subjects vaccinated with rAd5 vectors 
. Anti-penton antibodies have also been detected in cancer patients receiving rAd5 therapies 
. Therefore, the anti-capsid antibodies documented here are likely composed of both anti-hexon and anti-penton antibodies, with anti-hexon antibodies as the major contributor to neutralization. As shown here, multiple neutralizing epitopes targeted by the immune system exist for adenoviruses. These epitopes can be serotype-specific or located in conserved regions among adenoviridae. It has been documented that Ad fiber-specific neutralizing antibodies are more serotype-specific and Ad capsid-targeted antibodies are group-specific with a broader spectrum 
. It seems that HIV-infected participants had lower baseline Ad5 capsid neutralizing antibodies compared to HIV-uninfected participants. It is interesting to note that 41% were Ad35 seropositive among the Ad5 seropositive HIV-uninfected participants compared to 19% Ad35 seropositive among the Ad5 seropositive HIV-infected subjects. This difference does not reflect the lack of exposure to other adenovirus serotypes in the 70 HIV-infected subjects as the seroprevalence for other serotypes including Ad14, Ad28 and Ad41 in the Ad5 seropositive HIV-infected subjects is not significantly different from that in the HIV-uninfected Ad5 seropositives. It is quite possible that the infection rates were comparable but the HIV-infected cases did not generate a strong, long-lasting humoral response to a weakly immunogenic pathogen like Ad35. The data on magnitude of Ad5 titers and immune responses after vaccination and the correlations with subsequent HIV infection suggest that differences in immune function may predispose to infections seen in the Step study.
We also document differential effects of baseline Ad5 serostatus on CD4 and CD8 immune responses stimulated by the MRKAd5 HIV vaccine. The blunting effects of pre-existing Ad5 immunity were more apparent on the magnitude of CD8 responses than on CD4 responses. In the case of specific responses to Gag and Pol, Ad5 serostatus only affected CD8 responses to these two antigens. On the other hand, Ad5 serostatus reduced both CD4 and CD8 responses to Nef. Such differential effects on immune responses to different antigens were observed in DNA/rAd5 HIV vaccine trials: Ad5 serostatus had no effect on immune responses to HIV Env but affected anti-Gag immune responses 
. Such blunting can affect vaccine immunogenicity in Ad5 seropositives and the boosting effect of vaccinations after the first in Ad5 seronegatives. Thus, a heterologous prime and boost regimen may further improve vaccine efficacy even in Ad5 seronegative subjects.
In summary, this study shows that the MRKAd5 HIV vaccine generated neutralizing antibodies to Ad5 hexon and penton capsid proteins more efficiently than to Ad5 fiber. Anti-vector immunity differed qualitatively in participants who acquired HIV infection compared to those who did not. This effect was independent of vaccination and may have arisen in part because of differences in the immune function of subjects enrolled from distinct geographic regions or HSV seropositivity or infection by other viruses. The increased HIV infection rates in Step were therefore confounded by these underlying increased risk factors and not clearly related to the vaccine. Pre-vaccination serostatus affected post-vaccination neutralizing antibody profiles and vaccine-induced immune responses to HIV antigens. These data also suggest that Nab to Ad5 fiber contributes to the reduction of vaccine potency. Thus, strategies to overcome pre-existing Nab must avoid the use of Ad5 fiber for vector construction.