Selective Loss of Early Differentiated, Highly Functional PD1high CD4 T Cells with HIV Progression

The role of PD-1 expression on CD4 T cells during HIV infection is not well understood. Here, we describe the differential expression of PD-1 in CD127high CD4 T cells within the early/intermediate differentiated (EI) (CD27highCD45RAlow) T cell population among uninfected and HIV-infected subjects, with higher expression associated with decreased viral replication (HIV-1 viral load). A significant loss of circulating PD-1highCTLA-4low CD4 T cells was found specifically in the CD127highCD27highCD45RAlow compartment, while initiation of antiretroviral treatment, particularly in subjects with advanced disease, reversed these dynamics. Increased HIV-1 Gag DNA was also found in PD-1high compared to PD-1low ED CD4 T cells. In line with an increased susceptibility to HIV infection, PD-1 expression in this CD4 T cell subset was associated with increased activation and expression of the HIV co-receptor, CCR5. Rather than exhaustion, this population produced more IFN-g, MIP1-a, IL-4, IL-10, and IL-17a compared to PD-1low EI CD4 T cells. In line with our previous findings, PD-1high EI CD4 T cells were also characterized by a high expression of CCR7, CXCR5 and CCR6, a phenotype associated with increased in vitro B cell help. Our data show that expression of PD-1 on early-differentiated CD4 T cells may represent a population that is highly functional, more susceptible to HIV infection and selectively lost in chronic HIV infection.


Introduction
PD-1 is expressed on the surface of T-cells, macrophages, and B cells and functions as an inhibitory co-receptor in the B7:CD28 family, specifically in the regulation of immune activation, inflammation and tolerance [1,2]. Studies of chronic viral infection have demonstrated the importance of PD-1 in the regulation of immune exhaustion in CD8 T cells, and to a lesser extent, CD4 T cells. Exhausted T cells are defined by the gradual loss of effector function, typically by decreased secretion of IFN-g, TNF-a, IL-2 cytokines, and terminal differentiation, and have been described in chronic viral infections in mice, rhesus macaques, and humans [3][4][5][6]. Interfering or blocking the PD-1 pathway can improve or restore functional CD8 T cells during chronic LCMV or SIV infection [5,7]. Recently it was also shown that blocking the PD-1/ PD-L1 pathway resulted in clearance of parasitemia in a mouse model of blood-stage malaria with an increase in both CD4 T cell function and expansion of T follicular helper (T FH ) cells and plasmablasts, indicating that this interaction is important for the development of pathogen-specific adaptive immune responses [8].
Multiple lines of evidence suggest that T cells, even those with an exhausted phenotype, may retain some functional and proliferative capacity during a chronic viral infection [9][10][11]. Specifically, recent evidence from adoptive transfer studies in mice show that antigen-specific CD8 T cells retain proliferative capacity, though with reduced effector function, despite an exhausted phenotype [12,13]. Another study of PD-1 expression during chronic SIV infection in Rhesus macaques demonstrated that PD-1 expression on CD4 T cells is associated with retained proliferative capacity based on ex vivo Ki-67 expression [14]. Taken together, these studies suggest that PD-1 expression by itself may not solely be a phenotypic marker of immune exhaustion, but may regulate subsets of T cells with a specific differentiation state and effector function, thereby limiting the inflammatory response and tissue damage during chronic infection [15].
Here, we show that in the EI CD4 T cell population there is increased expression of PD-1 relative to CTLA-4 within the subset that is CD127 high¸a nd this population is initially increased in HIV-infected compared to uninfected individuals, but then decreases concomitant with the expansion of PD-1 high CTLA-4 high CD127 high EI CD4 T cells. HIV-infected subjects with higher plasma HIV RNA had a reduced frequency of PD-1 high CD127 high EI CD4 T cells along with increased cell-associated HIV gag DNA in this population. Further, we demonstrate that this population with increased PD-1 expression is also associated with increased in vitro cytokine production, suggesting PD-1 is expressed earlier in the differentiation of CD4 compared to CD8 T cells.

Study subjects
HIV uninfected peripheral blood mononuclear cells (PBMC) were obtained from individuals participating in the NIH research apheresis program. Cryopreserved, HIV-infected PBMCs were obtained from three different study populations. For untreated HIV infection, cells were obtained from volunteers who participated in a therapeutic vaccination trial (no efficacy was observed) prior to receiving anti-retroviral therapy [16], who had relatively preserved CD4 counts (median 525, interquartile range [IQR] 390-879). We also obtained PBMC from HIVinfected donors with more advanced HIV (median CD4 count 148 cells/μL, IQR 59-274) participating in AIDS Clinical Trials Group study A5142 prior to initiation of combination antiretroviral therapy (cART) and at 48 weeks of therapy [17,18]. The third study population consisted of donors obtained from a cohort used to identify individuals with HIV broadly neutralizing antibodies as previously described [19]. Characteristics of these populations are provided in Table 1. All studies involving human subjects were reviewed and approved by their respective institutional review boards to include the IRB's of the National Institute of Health, National Institute of Allergy and Infectious Diseases, the University of California San Diego, and the Walter Reed Army Institute of Research. Data and stored specimens were utilized from prior multi-center clinical studies under which written informed consent was obtained for all study volunteers to store samples for future use. The use of stored samples for this study was approved the Walter Reed Army Institute of Research Institutional Review Board and the NIH/NIAID. Regarding the ACTG samples, the use of the samples and the submitted manuscript was approved from the ACTG appropriate reviewing committee. ACTG, a multi-center network, samples were collected from different sites and analyzed based on their recovery and survival. This makes impossible to identify specific IRB protocol for the used ACTG samples.
Telomerase activity. From the same sorting experiment and populations used for in vitro cytokine production, 30,000 cells each were sorted and lysed using the Quantitative Telomerase Detection (QTD) lysis buffer for telomerase activity, which was performed using the QTD Kit according to manufacturer's instructions (Allied Biotech, Vallejo, CA).
HIV-1 Gag DNA PCR. Similarly, approximately 5000 cells were sorted from cryopreserved, HIV-infected PBMC's directly into lysis buffer and quantification of HIV gag DNA was performed by quantitative PCR (qPCR) by means of a 5 0 nuclease (TaqMan) assay with an ABI 7700 system (Perkin Elmer, Norwalk, CT) as previously described [20,21]. Standards were constructed for absolute quantification of gag and albumin copy number and were validated with sequential dilution of 8E5 cell lysates that contain one copy of gag per cell. Duplicate reactions were run and template copies calculated using ABI7700 software.
In vitro HIV infection. Sorted memory CD4 T cells from two healthy donors were subjected to in vitro infection with a R5-tropic EGFP HIV-1 AD8 at multiplicity of infection (MOI) of 0.01 for 5 days. Cells were harvested on day 5 and PD-1 levels in non-infected (GFP-) and cells harboring virus (EGFP+) were analyzed by flow cytometry.

Statistical analysis
Experimental variables were analyzed using nonparametric statistical tests of inference: Mann-Whitney U test, the Wilcoxon matched-pairs signed rank test, or the Kruskal-Wallis test with Dunn's multiple comparison post-test as appropriate. Correlation analysis was performed using the nonparametric Spearman test. The Generalized Estimating Equations regression analysis was utilized to model longitudinal measurements of HIV-1 viral RNA or CD4 count to account for the non-independence of intra-individual repeated measures. Statistical analyses were performed with GraphPad Prism (GraphPad Software, version 5.0) or Stata Statistical Software, Release 11 (StataCorp, College Station, TX).

Loss of PD-1 high CTLA-4 low early-differentiated CD4 T cells in advanced HIV infection
First, we determined the expression patterns for two major co-inhibitory receptors for CD4 T cells from HIV uninfected (n = 15) and untreated HIV-infected subjects from a cohort with earlier HIV infection (Cohort 1, n = 31, median CD4 count 525 cells/μl) or with more advanced disease prior to treatment (Cohort 2, n = 14, median CD4 count 148 cells/ μl) ( Fig 1A and Table 1). Skewing of maturation subsets was evident by a significant lower frequency of CD27 high CD45RA high (naïve) and CD27 high CD45RA low CD4 T cells associated with an increased frequency of CD27 low CD45RA low (late differentiated, LD) CD4 T cells in Cohort 2 with lower median CD4 count, consistent with more advanced disease (S1A Fig). Although several markers have been used to define memory phenotypes, we found distinct patterns of PD-1 and CTLA-4 (two major regulators of T cell activation and function) expression in T cell populations dependent on CD127, the IL-7 receptor (Fig 1A and 1B). Overall, the relative frequency of cells expressing a PD-1 high CTLA-4 low phenotype was increased in HIV-infected compared to uninfected individuals, in all CD4 T cell memory subsets tested (Fig 1B). A strong association was found between the expression of PD-1 and differentiation of CD4 T cells in donors with less advanced disease (Cohort 1): the CD127 high EI compartment had the highest frequency of PD-1 high CD4 T cells (Fig 1B) (p < 0.0001, for PD-1 high CTLA-4 low in CD127 high vs. CD127 low EI and p = 0.268 for PD-1 high CTLA-4 low CD127 high vs. CD127 low in the LD compartments respectively). A significantly lower frequency of PD-1 high CTLA-4 low CD127 high EI CD4 T cells was found in the cohort with more advanced disease (Cohort 2), a pattern that was not seen in the other memory populations. On the other hand, the frequency of PD-1 high-CTLA-4 high cells was decreased, although not significantly, in both CD127 high and CD127 low EI CD4 T cells from HIV-infected individuals with more advanced disease ( Fig 1B).
The expression pattern of these receptors appeared to differ from CD8 T cells from HIVinfected and uninfected donors (S1C Fig), with differences in PD-1 high CTLA-4 low frequencies noted for more differentiated CD8 T cells (CD127 low ). Interestingly, the comparison between PD-1 high and PD-1 high CTLA-4 high expression profiles on CD27 high CD45RA low CD4 T cells from HIV negative individuals indicates that PD-1 is up-regulated prior to CTLA-4 during CD4 T cell differentiation. Minimal expression of other negative co-stimulatory molecules (2B4, CD160) on CD4 T as compared to CD8 T cells was found (data not shown). These observations indicate that the regulation of both PD-1 and CTLA-4 differs between CD4 and CD8 T cells, particularly in early-differentiated CD4 (CD8) T cell populations.
To further investigate the impact of HIV infection on the described CD4 T cell populations, we combined data for HIV-infected subjects from Cohorts 1 and 2, before treatment (n = 45). Consistent with previous studies, we found increased frequencies of PD-1 high cells in total (naïve and memory) CD4 and CD8 T cell compartments with higher viral load (Fig 2A). However, this association was stronger in the case of CD8 T cells (Spearman r = 0.45, p = 0.002) compared with CD4 T cells (r = 0.31, p = 0.044) (Fig 2A), where higher viral load was correlated with higher PD-1 expression on more differentiated, Late (CD127 low CD27 low CD45RA low ) CD4 T cells (Spearman r = 0.341, p = 0.042) (Fig 2A). However, PD1 high CD127 high EI subset was negatively associated with viral load, independent of the expression of CTLA4 (Fig 2A).
To extend these findings, we assessed the association of the EI CD4 T cell phenotype with longitudinal viral load measurements in a GEE regression analysis in which repeated measurements HIV-1 viral RNA were modeled as the dependent variable. We observed a slight decline in HIV-1 viral load over time associated with the PD-1 high CTLA-4 low CD127 high EI CD4 T cell phenotype (regression coefficient = -0.062, p = 0.023), which was the only statistically significant phenotypic association with longitudinal viral load (other data not shown). We further examined untreated subjects for HIV Gag-specific responses (S2 Fig) as HIV-specific CD4 T  cells have been shown to be preferentially infected [21]. HIV-specific CD4 T cells expressed an IFNg + IL-2profile (S2 Fig). Among 14 subjects with a detectable HIV Gag-specific CD4 response, we found that the majority of Gag-specific CD4 T cells (median 65.0%, range 18.4 to 94.4%) were more differentiated (CD127 low ) and co-expressed PD-1 and CTLA-4 (Fig 2B), which is consistent with prior studies [22,23].
No down-regulation of PD-1 was observed with in vitro HIV infection of sorted, memory CD4 T cells in infected compared to uninfected cells (S3A Fig). Our data indicate that PD-1 high CD127 high EI CD4 T cells may be preferentially lost during chronic HIV infection. Furthermore, the very low frequency of HIV-specific CD4 T cells challenges their importance for CD4 dynamics within this early-differentiated compartment.

Restoration of PD-1 high CTLA-4 low early-differentiated CD4 T cells after antiretroviral therapy
Our data imply that the progressive loss of PD-1 high CD4 T cells in an early differentiation state could be mediated, at least in part, by increased infection by HIV. To confirm this, we evaluated sorted CD4 T cell populations ( Fig 2C) and observed an increase in the frequency of HIV-1 gag DNA (n = 5 donors) in the PD-1 high compared to PD-1 low populations. Since cell-associated DNA content will vary with plasma viral RNA, we compared paired differences in HIV-1 gag DNA across differentiation subsets which were statistically significant (p = 0.031, Friedman test), consistent with the interpretation that early differentiation is associated with increased HIV infection. Only the difference between the PD-1 high CD127 high EI and naïve compartments was statistically significant after Dunn's multiple comparisons correction (p < 0.05). Interestingly, a comparable frequency of HIV-1 gag DNA copies was observed between the PD-1 high CD127 high EI and LD CD4 T cell compartments (Fig 2C).
In the HIV treatment cohort (Cohort 2, Table 1) in which PBMC were obtained before and 48 weeks after initiation of combination antiretroviral therapy (cART), we observed an increase of the relative frequency of CD4 T cells expressing high levels of CD127 with therapy as previously described [24] (Fig 2D). Consistent with their increased loss during advanced disease ( Fig  1C), a significant expansion (p = 0.0273, Wilcoxon matched-pairs test) of the PD-1 high CTLA-4 low CD127 high EI CD4 T cell subset was found in subjects with lower CD4 counts (<200 cells/ μL) at treatment initiation (Fig 2D). In contrast, cART led to a decreased frequency of PD-1 high-CTLA-4 high cells, especially in theCD127 high LD CD4 T cell compartment (p = 0.013), as well as within the respective CD127 low populations (data not shown). Although the data shown are relative frequencies and not absolute counts, there is the possibility of redistribution of memory subsets following cART initiation. Taken together, our data indicate that the dynamics of earlydifferentiated CD4 T cells could be regulated by infection-depletion along with other mechanisms that could promote their differentiation towards more mature CD4 T cell phenotypes.  figure). The PD-1 high CTLA-4 low CD127 high group is analyzed separately for subjects who started cART with a CD4 count less than 200.

PD-1 up-regulation in early-differentiated CD4 T cells is associated with increased activation and expression of the HIV co-receptor CCR5
Next, we investigated whether expression of PD-1 in CD127 high EI T cells was associated with differential expression of chemokine receptors, particularly CCR5 or activation markers, parameters directly associated to HIV infectivity. We performed further CD4 T cell phenotypic analysis for the chemokine receptors CCR4, CCR5, CCR6, and CCR7 as well as the activation markers CD38, HLA-DR, and BTLA, a co-inhibitory receptor with functional characteristics similar to PD-1 and CTLA-4, which is increased among less differentiated T cells [25]. PD-1 high CD127 high EI CD4 T cells had significantly higher CCR5 expression Compared to PD-1 low CD127 high EI CD4 T cells (Fig 3A), but did not show any other significant differences in chemokine receptor expression (Fig 3A). Increased expression of activation markers per cell (judged by Mean Fluorescence Intensity-MFI) was observed with increased differentiation. However, both HLA-DR and CD38 expression was significantly up-regulated in the PD-1 high cells in the early differentiated CD27 high CD45RA low compartment and CD27 low CD45RA low CD4 T cells (Fig 3B). We observed similar patterns of chemokine and activation marker expression in the CD127 low compartment with increasing differentiation (data not shown) where there were significant differences in expression of CCR5, HLA-DR, and BTLA between PD-1 high and low in the EI CD4 T cell population. To assess whether this represents actual T cell activation rather than up-regulation of PD-1 by other bystander mechanisms [26], we characterized several sorted populations for telomerase activity. We found an overall increased telomerase activity with differentiation (p = 0.02), but the PD-1 high CD127 high EI population was the only differentiation phenotype significantly different from the naïve population (p < 0.05 after Dunn's multiple comparisons correction, Fig 3C). This suggests that PD-1 high CD127 high EI CD4 T cells represent an early state of CD4 T cell differentiation that have received activation signals by TCR engagement [27,28]. Our data suggest that increased expression of CCR5 on an activated background could result in the increased susceptibility of PD-1 high CD127 high EI CD4 T cells to HIV infection.

PD-1 high early-differentiated CD4 T cells are characterized by increased functionality
Given the role of PD-1 in CD8 T cell exhaustion during chronic viral infections [29] we sought to investigate whether PD-1 high CD127 high EI CD4 T cells were characterized by impaired functionality, a hallmark of the exhaustion phenotype. To this end, we used sorted CD4 (S3B Fig) T cell populations from uninfected individuals and examined the effect of in vitro TCR stimulation using a functional plate-bound anti-CD3 antibody and measuring cytokine production in the supernatants. We observed increased cytokine production (IFN-g, MIP1-a, IL-4, IL-10, and IL-17) from naïve to PD-1 low , PD-1 high CD127 high EI and LD CD4 T cells (p = 0.0026, Kruskal-Wallis test, Fig 4A). Interestingly, a considerable production of IL-17 between PD-1 high CD127 high EI and LD CD4 T cells was found (Fig 4A) underlining the possible impact of the loss of this particular early-differentiated CD4 T cell population in the overall compromised of IL-17-mediated defense mechanisms in chronic HIV [30]. In addition, there was evidence of proliferative capacity with increased Ki-67 expression in the CD127high and CD127low EI populations with expression of PD-1 and CTLA-4 ( Fig 4B, S3C Fig). We then used staphylococcal enterotoxin B (SEB) for polyclonal stimulation to assess, using a flow cytometry assay, ex vivo production of IFN-g and IL-17 in untreated HIV-infected individuals (n = 5) (Fig 4B, S3C Fig). We observed a similar pattern of cytokine secretion compared with anti-CD3 stimulation (Fig 4B). These results indicate the possible impact of the loss of EI CD4 T cell population in the overall compromised defense mechanisms of chronic HIV patients [30].
We have previously shown that CD4 T cells expressing a CCR7 high CXCR5 high CCR6 high PD-1 high phenotype can provide increased in vitro B cell help and is decreased in HIV infection [31]. Here, we further examined this "peripheral T FH " phenotype and found that within the PD-1 high CD127 high CCR7 high EI CD4 T compartment, a subset of CD4 T cells with increased HIV gag DNA content (Fig 2C), there is a significant loss of CXCR5 high CCR6 high cells (Fig 4C,  S3D Fig). Taken together, these data indicate that PD-1 expression in early-differentiated CD4 T cells may not be associated with functional defects and exhaustion of this less differentiated population.

Discussion
PD-1 is considered to play an important role in the regulation of the CD-4 T cell response due to its increased expression in virus-specific CD4 T cells [22]. A large body of evidence has established that PD-1 is a critical mediator of CD8 T cell exhaustion. Up-regulation of PD-1 in antigen-specific CD8 T cells results in decreased proliferative and effector capacities in chronic viral infection [4,5,32] and cancer [33]. However, the role of PD-1 as a regulator of CD4 T cell function is not well understood. Here, we demonstrate that the pattern of PD-1 expression differs between bulk CD4 and CD8 T cells for both HIV-infected and uninfected individuals indicating a differential regulation of the receptor on CD8 T cells compared to CD4 T cells either at the transcription level or due to the sensitivity of extracellular/intracellular signals regulating the surface expression of PD-1. Analysis of the co-expression of PD-1 and CTLA-4 in early-differentiated (CD127 high CD27 high CD45RA low ) CD4 T cells from uninfected subjects revealed that PD-1 is expressed earlier than CTLA-4 during the differentiation of CD4 T cells (p < 0.0001). In agreement with previous studies [22,23], the majority of HIV Gag-specific cells express a more differentiated phenotype (CD127 low ) skewed towards PD-1 high CTLA-4 high , especially in the latedifferentiated "effector memory" compartment (CD27 low CD45RA low ) where a minority of bulk CD4 T cells is characterized by a PD-1 high CTLA-4 low phenotype. However, the inclusion of CTLA-4 in our analysis revealed an altered dynamic between PD-1 high CTLA-4 low and PD-1 high CTLA-4 high early during the differentiation of CD4 T cells and with respect to HIV progression. We further analyzed the phenotype and function of PD-1 high CTLA-4 low CD127 high EI CD4 T cells and found that PD-1 expression was associated with increased activation, measured by HLA-DR and CD38 expression and increased susceptibility to HIV infection, based on the expression of the HIV co-receptor CCR5. These cells are also characterized by increased telomerase activity (Fig 3C), suggesting that this population represents an early state of CD4 T cell differentiation that have been preferentially activated by TCR engagement [27,28]. Furthermore, PD-1 high expression marks higher sensitivity to in vitro spontaneous and CD95/Fasinduced apoptosis in less differentiated CD4 T cells, although in significantly lower levels compared to PD-1 high "effector" CD4 T cells (data not shown). Accordingly, a decreased frequency of PD-1 high CTLA-4 low CD127 high EI CD4 T cells was associated with increased HIV-1 viral load over time. This observation and the demonstration of increased HIV gag DNA in this compartment suggest increased susceptibility to HIV infection which is in line with previous studies, where HIV infection was analyzed in CD127 high CD4 T cells [23]. We should emphasize that in vitro infection of CD4 T cells was not associated with any down-regulation of PD-1 in infected compared to uninfected cells. Hence, we hypothesize that increased infection and depletion could affect the dynamics of early-differentiated CD4 T cells along with other mechanisms that promote their differentiation towards mature CD4 T cell phenotypes. More importantly, the contribution of PD-1 high CTLA-4 low CD127 high EI CD4 T cells to the establishment of a latent HIV-1 reservoir compared to highly differentiated CD4 T cell populations (for example, effector PD-1 high cells) which are more susceptible to cell death should be investigated further. Collectively, our data suggest that these early-differentiated CD4 T cells may be more susceptible to HIV infection due to increased activation and increased co-expression of CCR5 and PD-1. We propose that the dynamics of CD4 T cells may be altered by their susceptibility to HIV infection (PD-1 high CD127 high CD4 T cells) and the skewed maturation of HIV-specific CD4 T cells (PD-1 high CD127 low ), which are preferentially infected and highly sensitive to viral load changes and TCR stimulation.
Interestingly, we found that PD-1 high early-differentiated T cells were capable of producing a wide range of cytokines with overall cytokine production higher in PD-1 high compared to PD-1 low cells from the CD127 high EI CD4 T cell compartment among HIV-uninfected donors. This is consistent with the observation in the Rhesus macaque SIV model in which PD-1 expression on CD4 T cells, although not defined by differentiation phenotype, had retained proliferative capacity [14]. Hence, PD-1 signaling in the CD4 T cell compartment does not necessarily appear to confer an "exhaustion" status [14]. A loss of CD4 T cells producing IL-17 in HIV infected individuals has been previously described [34,35]. Our data indicate that the decline of PD-1 high CTLA-4 low CD127 high EI CD4 T cells, mediated, at least in part, by increased susceptibility to HIV infection, could contribute to the loss of IL-17+ CD4 T cells even at a very early step of CD4 T cell differentiation. Similarly, we observed that this phenotype overlaps with a circulating "T FH " phenotype, which was decreased in HIV-infected subjects, consistent with our previous study [31]. Previous studies have shown that CD4 T FH cells within lymph nodes may be the major reservoir for HIV infection and replication [36]. Whether increased HIV gag DNA in circulating PD-1 high CTLA-4 low CD127 high EI CD4 T cells reflects increased infection of a particular follicular CD4 T cell population within the lymph node needs further investigation. Together, our data show an accelerated expression of PD-1 in the early differentiation of CD4 T cells that is associated with increased cytokine production as opposed to an expected decrease in cytokine response observed with PD-1 expression [4].
Overall, our data indicate that PD-1 and CTLA-4 could serve as a very early marker of differentiation of CD4 T cells during HIV infection marking cells with increased sensitivity to infection. In contrast to CD8 T cells, our data suggest that a functional restoration of CD4 T cells in HIV possibly requires the manipulation of PD-1 as well as other co-inhibitory receptors, like CTLA-4.