In vitro studies suggest that CD4+ T cells can exhibit different susceptibilities to HIV infection depending on their differentiation or activation states, but relatively little is known about the features of CD4+ T cells preferentially targeted by HIV in vivo. By applying bioinformatics analysis on CyTOF-phenotyped blood specimens from 11 viremic individuals matched to ART-suppressed specimens from the same subjects, we provide an atlas of the phenotypic features of in vivo HIV-susceptible cells. The properties of in vivo susceptible cells were compared to those of in vitro infected cells to assess which properties of in vivo susceptible cells are recapitulated in vitro. Manual gating identified Th17 cells and memory CD4+ T cells expressing high levels of a4b1 as preferentially targeted for in vivo HIV infection, while memory CD4+ T cells of the Tem, Ttm, Th1, and Th1/Th17 subsets were preferentially targeted in vitro. PP-SLIDE bioinformatics analysis of the datasets revealed that multiple proteins antigens – particularly those encoding chemokine and cytokine receptors –were remodeled by HIV during in vivo infection, and this was mostly recapitulated in vitro. To identify features of HIV-susceptible cells shared between the in vivo and in vitro specimens, we implemented unbiased clustering. We identified from both specimen types a subset of highly-susceptible Tem-like cells, and validated the high susceptibility of this cellular subset. These studies provide an in-depth look at the features of HIV-susceptible cells in viremic individuals and demonstrate that some – but not all – subsets of HIV-susceptible CD4+ T cells characterized in vitro can effectively model HIV susceptibility in vivo.

FCS files for in vivo and in vitro datasets  for each donor. Datasets are provided for uninfected CD4+ T cells, uninfected memory CD4+ T cells, bystander CD4+ T cells, productively-infected cells, and predicted precursor (PRE) cells. Please refer to parent publication for details.