Descours et al. reply

replying to L. Pérez et al. Nature 561, https://doi.org/10.1038/s41586-018-0493-4 (2018); C. E. Osuna et al. Nature 561, https://doi.org/10.1038/s41586-018-0495-2 (2018); L. N. Bertagnolli et al. Nature 561, https://doi.org/10.1038/s41586-018-0494-3 (2018)

In our previous work¹, we used an in vitro model of HIV-infected unstimulated CD4 T cells to identify CD32 as a candidate marker of HIV⁺ resting CD4 T cells in vitro, and a subset of HIV⁺ total CD4 T cells containing replication-competent viruses in individuals that underwent anti-retroviral therapy (ART). Of note, we did not explore the transcriptional status of hosted viruses (latent or active) ex vivo, nor the activation state of these cells (quiescent or activated)¹. In the accompanying Comments^2,3,4, colleagues attempted to reproduce these findings. They present experiments that support the following conclusions: (1) the isolation of the CD32⁺ CD4 T cell population results from artefacts caused by the flow cytometry sorting method^2,3, and (2) the sorted CD32 CD4 T cell population is not enriched in HIV nor in replication-competent proviral DNA^2,3,4. Here, we formulate two questions that mirror the major issues raised by these three Comments^2,3,4 and discuss their results in the context of our previous report¹ and more recently published studies.

Is there any evidence that a CD4 T cell can express CD32 in the context of HIV infection? This question is raised by both Osuna et al.² and Pérez et al.³. A recent report⁷, using in situ hybridization (which avoids the criticism of artefacts caused by flow cytometry sorting), showed that HIV-1 RNA co-localized with CD32A (also known as FCGR2A) RNA in 90% of examined cells in B cell follicles from four individuals. Because HIV primarily targets CD4 T cells, these data may support the ability of a CD4 T cell to upregulate CD32 mRNA transcription after infection in vivo. Three independent groups have identified CD32 as being expressed by latently or productively infected CD4 T cells in vitro^1,5,6,7. These models generated and analysed a substantial percentage of HIV-infected CD4 cells. Thus, any marker that is usually not expressed by CD4 T cells but that is detected at the surface of these cells after infection is unlikely to result from biased analyses of cellular doublets, as could be the case when working on rare events from ex vivo samples^2,3. Instead, these data suggest that transcriptional regulation leading to the expression of CD32 mRNA and protein can probably occur after in vitro and in vivo infection of a single CD4 T cell.

Does the CD32 CD4 T cell subset contribute to viral persistence under treatment? All three of the accompanying Comments^2,3,4 indicate that CD32 CD4 T cells are not enriched for HIV DNA in blood. Recent work suggests, however, that in some virally suppressed HIV-infected individuals, CD32 CD4 T cells were enriched in HIV DNA, although to a lesser extent than we reported⁸. Notably, this question has been recently addressed in tissues, and results seem to be less contrasted than in blood^7,9,10 . More importantly, they revealed functional properties of these reservoir cells that have not been previously explored^7,9,10. As discussed above, a recent report⁷ found that within the B cell follicles of virally suppressed HIV-infected individuals, most of the cells containing HIV RNA and persisting despite treatment were found to express CD32A RNA⁷. This result seems to be in line with other data¹⁰ that indicate that T follicular helper cells, primarily found in these territories, were enriched for HIV DNA and RNA when expressing CD32¹⁰, although at a lower extent than our previous findings¹. In non-lymphoid rectal tissue, CD4 T cells expressing CD32 were also enriched for both HIV DNA and RNA⁹. Notably, the co-expression of CD32 and HIV RNA reported in these two publications^9,10 suggests that CD32 marks transcriptionally active infected cells rather than latent cells. Together, these reports support the ability of CD32 to identify a subset of persistent HIV-infected CD4 T cells and suggest that they could contribute to viral persistence under ART in vivo.

In conclusion, we believe that rather than completely ruling out the relevance of CD32 for the identification of a subset of infected cells in vivo and their contribution to HIV persistence, the whole literature, including the three accompanying Comments^2,3,4, opens new technical challenges and questions that we should solve in the near future.

Benjamin Descours, Gael Petitjean and Monsef Benkirane are solely responsible for this Reply. The contributions of the remaining authors from the original Letter¹ were limited to recruiting patients or performing analysis on blinded samples, and thus only Descours, Petitjean and Benkirane have authored this Reply.