CD32 expression is associated to T-cell activation and is not a marker of the HIV-1 reservoir

CD32 has been shown to be preferentially expressed in latently HIV-1-infected cells in an in vitro model of quiescent CD4 T cells. Here we show that stimulation of CD4+ T cells with IL-2, IL-7, PHA, and anti-CD3/CD28 antibodies induces T-cell proliferation, co-expression of CD32 and the activation of the markers HLA-DR and CD69. HIV-1 infection increases CD32 expression. 79.2% of the CD32+/CD4+ T cells from HIV+ individuals under antiretroviral treatment were HLA-DR+. Resting CD4+ T cells infected in vitro generally results in higher integration of provirus. We observe no difference in provirus integration or replication-competent inducible latent HIV-1 in CD32+ or CD32− CD4+ T cells from HIV+ individuals. Our results demonstrate that CD32 expression is a marker of CD4+ T cell activation in HIV+ individuals and raises questions regarding the immune resting status of CD32+ cells harboring HIV-1 proviruses.

1. There is a misconception of the results from Descours et al. Unlike what the authors state in the first line of the abstract, CD32 has not been shown to mark latently infected cells in vivo (although it was first identified in a latency model). Rather, CD32 identifies HIV-infected cells in ARTsuppressed individuals regardless of their activation and latency status. This difference is important in the context of the present study. The manuscript should be corrected accordingly and more details on the original study should be included. For instance, it would be important to mention that the differentially expressed genes (line 37) were identified in vitro. 2. Statistics are lacking in many figures (1a, 1c, 1e, 2a-c, 3c, 4b, 4d, 5a, 5d, 5e, 6a, 6c, 6d). Appropriate statistical tests should be used and justified and p values clearly indicated in the figures. 3. Line 70: 4 different stimulations are tested but only 3 are presented in Figure 1a. The IL-2+IL-7 results should be presented and IL-2 should be added to CD3/28 on the figure, for consistency with the text. The condition presented in Figure 2 (IL-7 alone?) is not described in line 70. I would recommend to merge Figure 1 and 2 and use similar ways to represent the data in all stimulation conditions. Figure 2 is poorly described and panel 2d (Ki67) not even mentioned in the main text. 4. In figure 1c, unstimulated cells (negative control) should be shown. 5. Results from Figure 1e-f suggest that CD32, HLA-DR and CD69 may be expressed through different signaling pathways since they do not overlap entirely. Whereas it looks like TCR engagement (with CD3 Ab) induces concomitant expression of HLA-DR and CD32, CD69 is rarely co-expressed by these cells. Knowing that the kinetic of expression of these markers differ after activation, how these results should be interpreted? Also, expression of HLA-DR and CD69 on nonstimulated CD32+ cells should be shown. 6. The Y axis of Figure 3c should be modified, as this figure does not represent the frequency of HLA-DR or CD69 cells expressing CD32, but rather the frequency of CD32+ cells expressing HLA-DR or CD69. Rather than correlations (line 69), these data show associations. 7. It is unclear if the assay used by the authors quantified total or integrated HIV DNA. The methods section (line 450) describes an assay that measures all genomes (integrated or not), which is in conflict with the title of the section, the figures and the main text. Proviral DNA should be used for integrated genomes only. This should be clarified. Of note, the original study by Descours et al. suggested that the integration step may be required for CD32 expression. Therefore, the use of an integrated assay in these in vitro experiments is critical. 8. Figure 6 is also problematic on multiple levels. Panels a and b show the number of HIV copies per cell (like in Descours et al), which was most probably calculated by making the ratio between the HIV DNA copies (again, it is unclear if these are integrated or not) and the number of cells. I do not understand the added value of panel c (absolute number of HIV DNA copies), since it is obvious that these numbers will depend on the number of cell used in this assay, which are unlikely to be similar between the subsets and the samples. Finally, panel d does not show a "contribution" (which should be represented as a percentage). Rather, the authors should represent the relative contributions of the CD32-and CD32+ subsets to the entire pool of infected cells (100%). 9. Line 187: I think there is a confusion between the frequency of infected cells (which is a ratio) and the contribution of a subset (which is a percentage). 10. As acknowledged by the authors, the manuscript does not include measurements of the replication competent reservoir, which limits the impact of the findings. 11. Several terms used in this paper should be replaced. Community representative prefer to no longer use the word "patient", particularly for HIV infected people on ART. Rather, "HIV infected individuals" or "participants" should be used. Similarly, "healthy donors" should be modified for control donors or uninfected controls.
Reviewer #2 (Remarks to the Author): The authors explore the validity of Descours et al. on the ability of CD32A to mark the latent HIV reservoir. Their findings generally contradict those previously published, and they speculate on some of the possible reasons for their different observations. It is important to share this information.
The work is convincing, as far as it goes, and although the question of the frequency of replicationcompetent HIV in the CD32 + and negative populations is unaddressed, it was not really sufficiently addressed in the Descours work as well.
While the discussion was readable, some of the writing style was awkward and its meaning might have been clarified. Light editing for style might be helpful.
Reviewer #3 (Remarks to the Author): The manuscript by Badia et al seeks to validate the use of CD32 expression on CD4+ T cells as a marker of cells that carry latent HIV originally originally published as an observation by Descours et al. As the author's discuss a cell surface expressed protein that can be readily used to identify the small number of resting CD4+ T cells that comprise the HIV latent reservoir would extremely useful for the study and targeting of latently infected cells.
Their data show that a subset of resting CD4+ T cells express CD32 following stimulation using a number of polyclonal activation stimuli, in comparison to HLA-DR and or CD69 expression this is a minor population cells.
The authors also looked for CD32 expression on CD4+ T cells from HIV infected individuals that had good virus control <50 copies/ml), this data showed that CD32 expression was significantly higher in HIV+ individuals and also associated with HLA-DR expression.
Activated CD4 T cells infected with HIV induced CD32 expression which was partially blocked by a HIV RT inhibitor. However CD32-HLA-DR+ cells were also infected. Utilizing a similar approach to Descours et al a system that allowed HIV-1 infection without activating CD4+ T cells was used and showed that CD32 was induced upon infection in agreement with published observation. These experiments also showed that in most donors used the proviral DNA was predominantly in CD32-T cells.
10 HIV+ individuals were used to derive CD32 + and -CD4 T cells and qPCR used to measure proviral load, this recapitulates the experiment from Descours et al fig4b. The result from this analysis showed no statistical difference in proviral load between CD32-and CD32+ and only 2/10 individuals had a higher load in CD32+ T cells. They conclude that the majority of infected CD4 T cells are CD32-.
This group were unable to perform viral out growth assays to be able to compare if this proviral population could give rise to reactivatable replication competent HIV.
This group conclude that their data suggest that CD32 expression identify activated CD4 T cells, but is not a marker of the latent reservoir. They discuss that their results show in 6/10 HIV+ donors tested that proviral copies/ cell were higher in CD32-T cells and that this was comparable to Descours et al raw data in supplemental data (5/9 HIV+ donors), however it is unclear what specific reinterpretation of raw data that is being performed here? This requires clarification.
The discussion is not clear, as to why the results from this publication are different from Descours et al., this clearly needs to be resolved. The ability of the latent reservoir to reactivate and produce replication competent virus is clearly an important attribute over and above just proviral load. Performing these experiments would be highly desirable in this situation and these experiments should be performed, it needs to be addressed if CD32+ T cells contribute the majority of replication competent HIV or if again the CD32-T cell population is equally important in contributing to the reactivatable replication competent reservoir ?
Minor points the English in the paper could benefit from editing to help with the flow and readability.
Response to reviewers. Reviewer #1: Badia et al. evaluated the impact of T cell activation on CD32 expression. They observed that different stimuli (TCR triggering, IL--2 and IL--7) increase IL--32 expression in CD4+ T cells and that these cells frequently co--expressed HLA--DR. Importantly, higher frequencies of CD32+ CD4+ T cells were measured in HIV--infected individuals compared to uninfected controls, and these cells frequently expressed HLA--DR but not CD69. In vitro infection showed that only a small fraction of productively infected cells expressed CD32. In addition, HIV DNA contents were similar in the CD32-- and CD32+ subsets. Using samples from suppressed individuals, the authors show that CD32 does not enrich in CD4+ T cells harboring HIV DNA and that the majority of infected cells do not express CD32. Although the results of this paper are potentially interesting, there are several important issues that preclude its publication. In Figures 1 and 2, negative controls are lacking and several panels are not described in the main text. Statistics should be entirely revised throughout the manuscript (many panels do not include p values). Response: Figures 1 and 2

have been revised. All panels are now clearly mentioned in the text and statistics are provided. We have merged old Figures 1 and 2 into the new Figure 1, included data of cells without any stimulus as control. Results and conclusions have not changed: T cell activation induced CD32 expression in CD4+ T cells.
The nature of the PCR assay (total or integrated) is critical and should be better explained. Several data are hard to understand, since they are presented in an unconventional manner.

Response: The PCR assay measures integrated DNA and is now clearly explained in Methods. We have revised all figures and figure legends to clarify the presentation of data.
1. There is a misconception of the results from Descours et al. Unlike what the authors state in the first line of the abstract, CD32 has not been shown to mark latently infected cells in vivo (although it was first identified in a latency model). Rather, CD32 identifies HIV--infected cells in ART--suppressed individuals regardless of their activation and latency status. This difference is important in the context of the present study. The manuscript should be corrected accordingly and more details on the original study should be included. For instance, it would be important to mention that the differentially expressed genes (line 37) were identified in vitro.

Response: We disagree. As mentioned by the reviewer, Descours et al first identify CD32 in vitro in a latency model and only when referring to HIV+ individuals they describe CD32 as "marker of CD4 T--cell HIV reservoir…" Indeed, the HIV reservoir definition, i.e. cells (CD4 T cells or others) in which a replication--competent form of
expression. Activated CD4 T cells infected with HIV induced CD32 expression which was partially blocked by a HIV RT inhibitor. However CD32--HLA--DR+ cells were also infected. Utilizing a similar approach to Descours et al a system that allowed HIV--1 infection without activating CD4+ T cells was used and showed that CD32 was induced upon infection in agreement with published observation. These experiments also showed that in most donors used the proviral DNA was predominantly in CD32--T cells. 10 HIV+ individuals were used to derive CD32 + and -CD4 T cells and qPCR used to measure proviral load, this recapitulates the experiment from Descours et al fig4b. The result from this analysis showed no statistical difference in proviral load between CD32--and CD32+ and only 2/10 individuals had a higher load in CD32+ T cells. They conclude that the majority of infected CD4 T cells are CD32--. This group were unable to perform viral out growth assays to be able to compare if this proviral population could give rise to reactivatable replication competent HIV.

Response: We have now evaluated replication competent HIV in the new Table 2.
This group conclude that their data suggest that CD32 expression identify activated CD4 T cells, but is not a marker of the latent reservoir. They discuss that their results show in 6/10 HIV+ donors tested that proviral copies/ cell were higher in CD32--T cells and that this was comparable to Descours et al raw data in supplemental data (5/9 HIV+ donors), however it is unclear what specific reinterpretation of raw data that is being performed here? This requires clarification.

Response: We referred to the raw data presented as supplemental material in the Descours original article. This is now clearly stated in the revised manuscript.
The discussion is not clear, as to why the results from this publication are different from Descours et al., this clearly needs to be resolved.

Response: We clearly show that CD32 is a marker of activated cells, which may not be considered an HIV reservoir. Our data indicate also that CD32 positive cells do not significantly differ from CD32 negative cells in the number of viral DNA copies per cell and in the presence of replication competent viruses, being both evidences different from that of Descours et al. We have clarified the discussion in order to better reflect our findings. However, although we feel that there are a number of significant shortcomings in the work of Descours et al., it is not our aim to disqualify their work. We expect that further research by others and ourselves will provide adequate explanations to the role of CD32 in HIV--1 infection.
The ability of the latent reservoir to reactivate and produce replication competent virus is clearly an important attribute over and above just proviral load. Performing these experiments would be highly desirable in this situation and these experiments should be performed, it needs to be addressed if CD32+ T cells contribute the majority of replication competent HIV or if again the CD32--T cell population is equally important in contributing to the reactivatable replication competent reservoir? Table 2. Minor points the English in the paper could benefit from editing to help with the flow and readability.

Response: The manuscript has been revised by the Springer Nature Editing service.
I appreciate the author's answers to my comments. However, there are still several aspects in this manuscript that would need to be modified or clarified. The experiments presented in this manuscript clearly show a link between CD32 expression on CD4+ T cells and T cell activation. However, the viorological data are much less convincing and the role of HIV infection in that process needs to be clarified. The new IUPM data are difficult to interpret and the IUPM values reported here are 3 log higher than in published studies. In addition, the integrated HIV DNA values are also surprisingly high (1 to 10% of infection in CD4+ T cells). The authors should consider the following remarks: 1. I still disagree on the first point, which is in my view, critical. Stating that the HIV reservoir identified by Decours et al. is "latent" is an incorrect interpretation of the original study by the authors (and apparently by other scientists who commented on this article in "news and views" as mentioned in the author's response.) In Descours et al., it was never said that the CD32+ T cells isolated from people on ART were latently infected. Accordingly, I disagree with the following sentence form the authors: "the HIV reservoir definition, i.e. cells (CD4 T cells or others) in which a replication-competent form of HIV persists, entails the existence of viral latency". Latency and HIV persistence are 2 different concepts. For instance, residual viral replication can sustain a persistent viral reservoir without requiring latency. Therefore, I think the manuscript should be modified to clearly distinguish these concepts. In the last sentence of the abstract ("These results raise questions regarding the immune resting status of CD32+ cells harboring HIV-1 proviruses"), the authors infer that Descours et al. investigated the "immune status" of the CD32+ cells which is not correct. The only data that directly contradicts the Descours findings are those showing no enrichment in HIV DNA (or replication competent HIV) in CD32+ cells, which is independent form the activation status of these cells. The first sentence of the abstract should also be corrected and all the manuscript should be modified to acknowledge that Descours et al. did not identify CD32+ CD4+ T cells as a "latent" reservoir. Indeed, the recent work from Abdel-Mohsen et al (Science Translational Medicine, 2018) confirms that CD32+ identifies a transcriptionally active reservoir and not a latent reservoir for HIV, further reinforcing the importance of distinguishing these 2 concepts. 2. Line 109: "HIV-1 infection induced CD32 expression in PHA/IL-2 activated CD4+ T cells (Fig.  3a)." When looking at the third dot plot, it looks like the majority of the CD32+ cells are found within the GFP negative population. Although efavirenz somewhat reduces CD32 expression, it is hard to distinguish the relative effect of HIV sensing and HIV infection in these experiments. 3. Line 128: "This finding indicates that CD32 expression is a marker of T cell activation." This conclusion applies to Figure 1, not to  10%). This is at least a log higher than expected. 5. The new experiment aimed at measuring replication competent HIV in CD32-and CD32+ populations is hard to interpret. The authors report a mean IUPM value of 39886 in Table 1. IUPM are usually in the range of 0.1 to 10. I don't think the numbers in Table 1 represent IUPM. Also, the 95%CI range do not overlap with the IUPM values. This should be clarified. 6. Finally, VOA sensitivity depends on the number of cells used to perform the assay. These should be indicated (particularly for the CD32+ fraction). It is surprising that the authors found positive culture in almost all CD32+ sorted populations given their very low frequency, and according to the authors, their similar frequency of infection compared to total CD4+ T cells. The number of cells in each well (500-20,000 cells) seems extremely low as well to measure replication competent HIV.