Reconstitution of Peripheral T Cells by Tissue-Derived CCR4+ Central Memory Cells Following HIV-1 Antiretroviral Therapy

Background: Highly active antiretroviral therapy induces clinical benefits to HIV-1 infected individuals, which can be striking in those with progressive disease. Improved survival and decreased incidence of opportunistic infections go hand in hand with a suppression of the plasma viral load, an increase in peripheral CD4+ T-cell counts, as well as a reduction in the activation status of both CD4+ and CD8+ T cells. Methods: We investigated T-cell dynamics during ART by polychromatic flow cytometry in total as well as in HIV-1-specific CD4+ and CD8+ T cells in patients with advanced disease. We also measured gene expression by single cell transcriptomics to assess functional state. Results: The cytokine pattern of HIV-specific CD8+ T cells was not altered after ART, though their magnitude decreased significantly as the plasma viral load was suppressed to undetectable levels. Importantly, while CD4+ T cell numbers increased substantially during the first year, the population did not normalize: the increases were largely due to expansion of mucosal-derived CCR4+ CD4+ TCM; transcriptomic analysis revealed that these are not classical Th2-type cells. Conclusion: The apparent long-term normalization of CD4+ T-cell numbers following ART does not comprise a normal balance of functionally distinct cells, but results in a dramatic Th2 shift of the reconstituting immune system.


INTRODUCTION
Highly active antiretroviral therapy (ART) for the treatment of HIV-1 effectively suppresses plasma viral load (PVL) in a vast majority of individuals, as well as gradually restoring CD4 + T-cell numbers and function. The reconstitution of the CD4 + T-cell compartment in peripheral blood is essentially biphasic [1][2][3]. An early, rapid increase during the first three weeks [3,4] may be due to redistribution of memory cells to the peripheral blood from sites of inflammation in the tissues; subsequently, a slower phase, evident after about three months of treatment, is at least in part due to de novo production of naïve CD4 + T cells from the thymus [3,5], as well as improved T-cell survival [6,7]. The frequency of proliferating (Ki67 + ) cells decreases in both the CD4 + and CD8 + T-cell compartments, with a transient increase after 6 months of therapy, mainly in CD4 + central memory (T CM ) cells [8]. More advanced patients are reported to have proportionately faster reconstitution rates [9], though the lower the CD4 + T-cell nadir, the longer it takes to normalize this population [10]. More advanced patients are reported to have proportionately faster reconstitution rates, though the lower the CD4 + T-cell nadir, the longer it takes to normalize this population.
Beyond these basic changes, less is known about the evolution of the T-cell compartment's composition during ART. The most profound change described within the CD4 + and CD8 + T-cell lineages is an overall reduction in activation, as evidenced by loss of cells expressing CD38 [1,9,11] and HLA-DR [1,11,12], and a decrease in the mean fluorescence intensity (MFI) of CD38 on CD8 + T-cells [11,13,14]. These changes represent a (partial) normalization of the T-cells' phenotype, towards that seen in healthy adults.
The HIV-specific T-cell response also changes dramatically following ART. Independent of the epitope, HIV-specific CD8 + T-cell responses exhibit an early, rapid decline, continued with slower kinetics once plasma viral loads have been suppressed to undetectable levels [15]. This reduction in magnitude is not accompanied by a change in the quality of the CD8 + T-cell response [16]; however, like the bulk T-cell compartment, the expression of CD38 and HLA-DR on HIV-1 Gag-specific T cells decreases during treatment [11].
Despite these apparent normalizations, treated subjects still have immune defects. Therefore, we set out to determine T-cell dynamics during ART in total, as well as in HIV-1 Gag-specific CD4 + and CD8 + T cells. We found an overall rebalancing in the differentiation of T cells, favoring less differentiated cells; in addition, molecules related to activation and functional suppression gradually decreased during treatment, trending towards levels observed in healthy individuals. In sharp contrast to these expected findings, the proportion of Th 2 -like CD4 + T CM increased for at least six months following ART initiation, in a direction away from frequencies typical for healthy adults; these cells have characteristics of mucosal-derived cells. Therefore, ART-induced immune reconstitution does not necessarily lead to a normalization of the immune system as a whole, and may, for at least a year, lead to a state that is Th 2 -biased in nature.

MATERIALS AND METHODS
Ethics statement. HIV-1 + subjects were enrolled and provided written informed consent at the Clinical Center of the National Institute of Allergy and Infectious Diseases, NIH, under a protocol approved by the NIAID Institutional Review Board. These studies were registered at www. clinicaltrials.gov as #NCT00557570 and #NCT00286767. Samples were coded; all analyses were performed blinded to identity.
Human subjects and sample collection. The patient cohort has been described elsewhere [17]. Briefly, all patients (1) were ART-naïve (n = 56) or had interrupted treatment for at least one year (n = 4, plus n = 2 who had previously received brief mono-or dual therapy) with a viral rebound of > 10,000 copies/ml; (2) had ≤ 200 CD4 + T cells/μl at baseline; (3) suppressed their HIV-1 viral load to <500 copies/ml within one year of ART; and (4) had available peripheral blood mononuclear cell (PBMC) samples taken pre-ART as well as after 1, 3, 6, and 12 months of ART. Seventeen patients developed episodes of immune reconstitution inflammatory syndrome (IRIS; defined according to the AIDS Clinical Trials Group criteria, <https://actgnetwork.org/IRIS_ Case_Definitions>) following commencement of ART, while 39 underwent uneventful immune reconstitution. PBMC from 12 healthy donors served as controls (Table 1).
Pre-amplified cDNA, and TaqMan TM primer/probes were loaded onto a microfluidic chip (Fluidigm), and multi-parametric quantitative RT-PCR was performed using a Biomark TM cycler (Fluidigm) as previously described [19].
Data analysis. Flow cytometry data were analyzed using FlowJo (FlowJo, LLC), Pestle (NIAID, NIH; by M. Roederer), and SPICE 5.1 [20]. The gating scheme is identical to that used in our previous publications [17,18]. All cytokine measurements were background subtracted, taking into account the frequency of cells producing cytokines in the absence of antigenic stimulation (mock control). For the phenotypic analysis of Ag-specific cells, only those samples with >10 cytokine-positive events and response magnitudes > 3x that of the corresponding mock control were considered. The mean fluorescence intensity (MFI) of CD38 + cells was calibrated using the experiment-matched internal control sample.
RT-PCR data were analyzed using JMP 11 (SAS), R 3.1, and Bioconductor [21]. Because varying cell numbers were sorted for RT-PCR of individual T-cell subsets, all samples were normalized to 50 cells. Relative gene expression levels or "expression threshold" (Et) are proportional to log 2 RNA abundance and were calculated using the "cycle threshold" (Ct) obtained, where Et = 28-Ct [19]. The following genes, expressed by less than 10% of samples analyzed, were excluded from the analysis, as this could due to inefficient amplification: CXCL11; CXCR1; CXCR2; GPR44; IL5; IL9; and TGFB2.
"Th-ness" and Differentiation Index (DI). Th-ness was defined as the posterior probability [22,23] of class membership given by a support vector machine (SVM) [24,25] trained to differentiate between all sorted healthy donor Th 1 -and Th 2 -like cells with radial basis kernel. All sorted CCR4 + T CM , CCR4 + non-T CM , CCR4 − T CM , and CCR4 − non-T CM samples were then assigned a Th-ness value according to their gene expression pattern, indicating whether their phenotype was more similar to Th 1 -or Th 2 -like cells. Accuracy of the SVM was 90% for three-fold cross validation.

Statistical analysis.
Nonparametric tests were used for all analyses (SAS version 9.2); matched comparisons were performed where applicable. Changes from baseline (paired differences) were evaluated using the Sign test. Statistical comparisons of pie charts were performed in SPICE 5.1 software using 10,000 permutations [20]. Given the exploratory nature of this study, there was no adjustment for multiple comparisons; in most analyses, only P-values less than 0.01 are reported.
Differential expression analysis of genes assayed by RT-PCR was performed using Limma [26][27][28]. Results for CD103 were obtained via robust regression [29,30]. All P-values from differential expression analyses were then pooled for control of false discovery rate [31]. Significance was then defined as an adjusted P-value less than 0.01.

RESULTS
Overall ART-Responsiveness Fifty-six HIV-1 + patients commenced ART when their CD4 + T-cell count was ≤ 200/μl. Phenotype and HIV-1 Gag reactivity of their PBMC-derived T cells were characterized before ART, as well as at 1, 3, 6, and 12 months after ART-initiation (Table 1). All patients rapidly responded to ART, evidenced by a 3-log suppression of the PVL within a month and to undetectable levels within 3 months (Supplementary Figure 1A). The CD4 + T-cell counts gradually increased during the time of follow-up. Though the increase was significant within 1 month of treatment, T-cell counts still remained largely below those observed in healthy adults at 1 year (Supplementary Figure 1B). CD8 + T-cell counts, which started in the range of levels typically observed in healthy adults, increased only during the first month of treatment (Supplementary Figure 1C) [18]. points. All time-points were compared to corresponding pre-ART measurements: *P <0 .01, **P < 0.001, ***P < 0.0001.

Longitudinal Analysis of HIV-Specific T-Cell Responses During ART
Even though the magnitude of the HIV-specific CD4 + T-cell response did not change within the first year of ART ( Figure 1A), these cells became more polyfunctional (i.e., producing two or three cytokines) over time, achieving statistical significance at late sampling time-points (6-12 months of ART, Figure 1B). This change in cytokine pattern was mainly due to increased IL-2 production ( Figures 1B, C). The subset distribution within HIV-responsive CD4 + T cells was also affected by ART: as early as 3 months after commencing ART, less differentiated cells (T CM , T TM ) increased, with a concomitant reduction in T EM cells ( Figure 1D). However, no significant change in the differentiation index (DI; see supplementary methods section) of HIV-1 Gag-specific CD4 + T cells was observed. Furthermore, the ART-induced reversal of other differentiation and inhibitory receptors' expression was much less dramatic for HIV-specific cells ( Figure 1E) than that for total CD4 + T cells (see below). Though there were trends mirroring total CD4 + T cells, no statistically significant changes were observed in the phenotype of HIV-specific CD4 + T cells. Taken together, these data indicate that while the overall magnitude of HIV-specific CD4 + T cells remained unaffected by ART, these cells became mildly more enriched for less differentiated cells without changes in activation state.
In contrast, HIV-specific CD8 + T cells reacted very differently to ART than their CD4 + counterparts. As previously published [11,16], there was a significant decrease in the magnitude of the CD8 + T-cell response to HIV-1 ( Figure 1F). However, the cytokine pattern remained virtually unchanged for at least one year of treatment ( Figures 1G, H). The subset distribution of HIV-responsive CD8 + T cells remained unchanged ( Figure 1I), and their DI did not change significantly over the course of study. The ART-induced reversal of other differentiation and inhibitory receptors' expression was also less dramatic ( Figure 1J) than that observed in total CD8 + T cells (Supplementary Figure 2).

Longitudinal Analysis of CD4 + and CD8 + T-Cell Differentiation During ART
We examined the evolution of T-cell differentiation over the course of ART; differentiation stage was defined by classifying cells (in rough order of maturation) as naïve (T NV ), central memory (T CM and T CM* ), transitional memory (T TM* and T TM ), effector memory (T EM ), or terminal effector (T TE* and T TE ) [32]. There were significant changes in the CD4 + T-cell subset distribution after starting ART (Figure 2A), with increasing proportions of less differentiated subsets (T NV , T CM , T TM* ) over the course of treatment and a concomitant reduction in the proportion of highly differentiated cells (T TE ). As previously reported, the frequency of T CM was increased at mo1 ( Figure  2B), prior to that of T NV , which became significant only at 1 year ( Figure 2D). This is a reflection of the initial redistribution of memory cells [4] followed by a delayed de novo production and improved survival of cells [3,[5][6][7]. While the relative frequency of T NV initially decreased due to the preferential release of memory cells from secondary lymphoid tissues, their absolute numbers increased upon introduction of ART ( Figure 2E), together with that of T CM ( Figure 2C). The proportion of CD4 + T cells in late differentiation stages (T EM , T TE* , and T TE ) steadily decreased after ART initiation ( Figure 2F). Consequently, the DI of the total CD4 + T-cell compartment progressively decreased over the course of ART ( Figure 2A). Interestingly, patients with higher pre-ART levels of late differentiation (T EM−TE ) CD4 + T cells demonstrated lower pre-ART PVL ( Figure 2G), but also a less dramatic recovery of CD8 + T-cell counts. time-points were compared to corresponding pre-ART measurements: *P < 0.01, **P < 0.001, ***P < 0.0001.
Nineteen of the 56 patients developed Immune Reconstitution Inflammatory Syndrome (IRIS) following ART initiation, which we showed alters CD4 + T-cell reconstitution kinetics, mainly by delaying T NV recovery and the concomitant reduction of T EM , which are most evident at mo6 [18]. As a result, the inclusion of patients experiencing IRIS after commencing ART delayed the overall observed decrease of T EM−TE CD4 + T cells (mo3 vs. pre-ART: P = .0465 IRIS, P = .0005 non-IRIS).
Changes in CD8 + T-cell subset distribution, though similar to those observed in CD4 + T cells, were much more subtle and only became statistically significant after many months of treatment, though some individual subsets exhibited significant changes early on (T CM* , T CM ; Supplementary  Figure 2A, B).

Expression of Differentiation, Activation and Inhibitory Markers
Alterations in T-cell activation phenotypes that might occur as a result of ART were comprehensively evaluated using a large range of cellular markers of T-cell differentiation, activation, and negative regulation. An ART-induced normalization of CD4 + T-cell differentiation was evidenced by a gradual increase in the frequency of cells expressing CD28 and CD127, paralleled by a down-regulation of the senescence marker CD57, as well as an increase in CD31 + cells and recent thymic emigrants (CD31 + CD45RO − CCR7 + ) after 1 year of therapy (Figure 3(a)). A concomitant decrease in CD4 + T-cell activation was indicated by decreasing frequencies of CCR5 + , CD38 + , GrB + , and Ki67 + cells (Figure 3(b)). There was also a decrease in the MFI of CD38 expression (Figure 3(c)); elevated expression of CD38 has been closely linked to poor prognosis in HIV-1 infection [33]. The proportion of CD4 + T cells expressing the inhibitory receptors CTLA-4, LAG-3, or TIM-3 also declined during this time (Figure 3(d)). Compared to healthy adults, markers of CD4 + T-cell differentiation, activation, and expression of negative regulators normalized (or trended in that direction) over 1 year of ART.
In contrast, the frequency of cells expressing CCR4 ( Figure 3A), HLA-DR, ICOS ( Figure 3B), and PD-1 ( Figure 3D) increased to levels significantly more disparate from those observed in healthy adults. This trend was transient for HLA-DR, ICOS, and PD-1, while for CCR4 it continued for at least one year. Note that the CCR4-expressing cells must be predominantly against specificities other than HIV, as their numbers are substantially greater than HIV-specific CD4 T cells ( Figure  1).
CD8 + T-cell differentiation and activation normalized during treatment, but much less dramatically than that of CD4 + T cells, as indicated by increased CCR4, CD28, and CD127 and decreased CD57 (Supplementary Figure 2C). The inhibitory receptors CTLA-4, LAG-3, PD-1, and TIM-3, the activation markers CCR5, CD38, FoxP3, GrB, HLA-DR, ICOS, and Ki67, as well as the MFI of CD38, all declined towards normal levels (Supplementary Figure 2D-F). Interestingly, while the phenotype of CD4 + T cells appeared to become more similar between patients over time (reduced range of activation marker expression), this was not the case for CD8 + T cells.
Co-Expression of CCR4, HLA-DR, ICOS, and PD-1 In contrast to all the other measured parameters, the expression of CCR4, HLA-DR, ICOS, and PD-1 on CD4 + T cells increased upon ART initiation, becoming more disparate from levels observed in healthy donors. Thus, we investigated co-expression of these molecules pre-ART and at mo1 of ART in HIV-1 + individuals, as well as in healthy donors. Even in healthy individuals, a large proportion of HLA-DR + , ICOS + , and PD-1 + CD4 + T cells expressed CCR4. In HIV-1 + patients, the CCR4 + fraction of ICOS + and PD-1 + cells significantly increased shortly after commencing ART ( Figure 4A). In contrast, even though the frequency of activated CCR4 + CD4 + T cells was more elevated in HIV-1 + patients compared to healthy donors, as measured by the expression of HLA-DR, ICOS, and PD-1, the introduction of ART did not significantly alter the proportion of activated cells ( Figure 4B) or the co-expression pattern of these three activation markers (unpublished data). Taken together, these results show that the increase of HLA-DR + , ICOS + , and PD-1 + CD4 + T cells is due to the increase in activated CCR4 + cells.

Th Subsets
Though the Th 1 /Th 2 dichotomy, and wider Th-subsetting, of CD4 + T cells is less applicable in humans than in mice where it was first described, this system allows the identification of cellular subsets that are associated with specific functions. Hence, we here make use of the phenotypically and functionally described Th-subsets, referring to them as "Th X -like" where possible.
In accordance with CCR4 being preferentially expressed on Th 2 -like cells [40], the majority of CCR4 + CD4 + T cells did not express any of the other chemokine receptors analyzed. In HIV-1 + patients, most CCR4 + cells were thus defined as Th 2 -like cells, and their composition was not affected by ART ( Figure 4C). However, the fraction of non-naïve CD4 + T cells expressing a Th 2like phenotype increased significantly after the induction of ART, while the proportion of Th 1 -like cells was relatively low in HIV-1 + patients at both time-points ( Figure 4D).

Th 2 -Like Cells and the T CM Phenotype
Because a hallmark of ART-induced immune reconstitution is the early rise in the number and frequency of CD4 + T CM [41], we investigated how the Th subsets, in particular Th 2 -like cells, correlated with this phenotype. As seen with total CD4 + T cells, the differentiation pattern of CCR4 + CD4 + T cells was biased towards a more differentiated population in HIV-1+ individuals. After 1 month of ART, this pattern, as well as the DI, remained unchanged, with T CM representing close to 50% of CCR4 + cells ( Figure 4E). While in healthy individuals CD4 + T CM cells harbored balanced proportions of Th1-, Th 2 -, Th 17 -, and Th 1 Th 17 -like cells, in HIV-1 + patients this population was biased towards a Th 2 -like phenotype. Unlike most other changes we observed within the CD4 + T-cell compartment, this altered representation was dramatically exacerbated by ART initiation ( Figure 4F). The co-expression pattern of HLA-DR, ICOS, and PD-1 on CD4 + T CM closely mimicked that of CD4 + CCR4 + cells and was not affected by ART. This indicates that the observed early increases of CD4 + CCR4 + T cells and CD4 + T CM largely identify the same population. Indeed, CCR4 + T CM increased in frequency upon ART initiation, while CCR4−T CM did not ( Figure 4G). Notably, proliferation, as measured by the expression of Ki67, did not appear to be the primary mechanism for this perceived expansion ( Figure 4H).

CCR4 + T CM are Not Genotypically Identical to Th 2 -Like Cells
To investigate whether CCR4 + T CM are bona fide Th 2 -like cells, we compared the transcriptional profile of CCR4 + T CM , Th 2 -and Th 1 -like SEB-reactive CD4 + T cells from healthy donors. CCR4 + T CM indeed closely resembled Th 2 -like cells in respect to the expression of Th 2 -( Figure 5A) and Th 1 -associated genes (Supplementary Figure 5), as well as most of the other cytokine genes investigated ( Figure 5B). The only genes differentially expressed in these two cellular populations were IL21 and IL22 (both positive in CCR4 + T CM ), the latter of which is typically produced by Th 17 -and Th 22 -like cells [42,43]. CCR4 + T CM are unlikely to be Th 22 -like, however, as their levels of IL4 and IFNg transcripts are similar to those of Th 2 -and Th 1 -like cells, respectively, neither of which is produced by Th 22 [44]. Unlike Th 2 -or Th 1 -like cells, CCR4 + T CM demonstrated some IL17a expression. This, together with expression of IL21 and IL22, suggests that these cells might contain an important fraction of cells with Th 17 -like functionality, even though chemokine receptor expression reveals only a small fraction of cells with a Th 17 -like phenotype (CCR4 + T CM have a chemokine receptor expression pattern comparable to total CCR4 + cells ( Figure 4C). CCR4 + T CM cells isolated from HIV-1 infected individuals one month after ART initiation had a transcriptional profile similar to that of healthy donors with respect to Th 2 -( Figure 5C) and Th 1 -associated genes (Supplementary Figure 5), as well as most other cytokine genes investigated ( Figure 5D). However, there was evidence of reduced Th 2 -type cytokine transcript levels, as well as of IL22. This cytokine deficiency was not restricted to CCR4 + T CM cells or Th 2 -associated cytokines. We found that total non-naïve cells from ART-naïve HIV-1 patients expressed lower levels of CSF2, IFNg, IL-2, IL-4, IL-10, IL-13, and IL-22 mRNA transcripts ( Figure 5E).
When considering all measured genes, CCR4 + cell populations (both T CM and non-T CM ) were mixed in terms of having a more Th 1 -or Th 2 -like gene expression profile, while CCR4 − samples were mostly Th 1 -like ( Figure 5F). , CCR4 − T CM , and CCR4 − T CM − cells was compared to that of Th 1 -or Th 2 -like cells sorted from healthy donors. Their calculated "Th-ness" is expressed as a point between those two extremes. Bar graphs show interquartile ranges, median bars, as well as individual data points. Statistically significant differences are indicated: *P < 0.01, **P < 0.001, ***P < 0.0001.

Emerging CCR4 + T CM Upon ART Induction Originate From Peripheral Tissue Sites
Strikingly, in PBMC from HIV-1 infected patients having received ART for one month, the integrin αE chain (CD103) was only found to be expressed in CCR4 + T CM ( Figure 6A). None of the other cell migration markers analyzed were found to differ between these cell populations. This is intriguing, as CD103 is part of the mucosa-homing receptor α E β 7 that is widely expressed on intra-epithelial lymphocytes and lamina propria T cells, as well as on skin-resident T cells. We further investigated this phenomenon by interrogating PBMC samples obtained pre-ART, at mo1 or mo12 of ART, or from healthy donors ( Figure 6B). Pre-ART, none of the T-cell subsets investigated showed any CD103 expression, whereas after 1 year of treatment expression was found in both CCR4 + T CM and Th 1 -like cells, similar to expression detected in healthy donors. This suggests that, in healthy donors, some CCR4 + T CM and Th 1 -like cells recirculate through peripheral tissue sites, while classical Th 2 -like cells do not. Figure 6. CCR4 + T CM appear to be released from peripheral tissue sites upon ART initiation. PBMC from healthy donors, as well as cells isolated before or after 1 or 12 months of ART from HIV-1-infected adults were stained with the "sorting" panel (Supplementary Table 1). Subsets of CD4+ T cells were sorted as indicated in Supplementary Figure 4 and their gene expression profiles determined by multi-parametric quantitative RT-PCR. (A) CD103 expression on Th 2 -like, CCR4 + T CM , and Th 1 -like cells isolated after 1 month of antiretroviral therapy. (B) Expression of CD103 in CCR4 + T CM , Th 2 -, and Th 1 -like cells from healthy donor PBMC (n = 9), and longitudinal samples from HIV-1 patients (n = 12). Bar graphs show interquartile ranges, median bars, as well as individual data points. Statistically significant differences are indicated: *P < 0.01, **P < 0.001, ***P < 0.0001.
We investigated the cell surface expression of several migration molecules on CD4 + T cells and found varied expression patterns between patients. HIV-1 patients appeared to have somewhat lower levels of CD103 + cells than healthy donors, while exhibiting higher levels of CCR9 + and integrin b7 + cells (Supplementary Figure 6). This was the case both pre-treatment and after 1 month of ART. Within CD4 + T cells, CD103 seemed to be preferentially expressed on CCR4 + cells.
There was a discordance in mRNA and protein expression of CD103, which might explain why this aberrantly expanding cellular subset was not identified previously. Down-regulation of CD103 (and other homing markers) typically occurs in a larger fraction of tissue-resident CD4 + T CM , likely heralding their release into peripheral blood (where we detected them). Maintenance of CD103 mRNA would allow for a rapid re-expression of CD103 proteins and a subsequent return to peripheral tissues.

DISCUSSION
We characterized the phenotypic and functional T-cell dynamics in peripheral blood of severely immuno-compromised HIV-1 + individuals following ART. Our data confirm previous findings of an early increase in T CM cells, as well as a gross reduction in overall activation levels. Almost all markers investigated, whether involved in T-cell differentiation, activation, or negative regulation, started normalizing early after ART initiation. These changes were most dramatic in CD4 + T cells, but mirrored by similar changes in CD8 + T cells, which retained a larger range of individual marker expression than CD4 + T cells.
In contrast to the general trend towards normalization, CD4 + T-cell activation (HLA-DR, ICOS, PD-1) initially increased in peripheral blood before gradually decreasing, too. CCR4 + cells demonstrated a more sustained increase, reflecting a skewing towards a Th 2 -like environment early after ART initiation, and a further deregulation away from the phenotype of healthy donors.
A shift from cells with Th 1 -to those with Th 2 -like functionality has previously been suggested to occur during HIV-1 infection [45][46][47]. It has been reported that long-term non-progressors exhibit a Th 1 -like cytokine profile, while progressors exhibit a Th 2 -like cytokine profile [47], and that increasing viral loads correlate with lower cytoplasmic levels of IL-2 and IFN-γ and concomitant increases in IL-4 and IL-10 levels after stimulation with PMA/ionomycin [46]. Also, certain IL-4Rα SNPs linked to IL-4 hypo-responsiveness may be associated with slower HIV-1 disease progression [48]. The Th 1 -to Th 2 -like switch observed during disease progression was suggested to be at least partially due to an initial selective loss of CCR5 + Th 1 -like cells [49]. Our data confirm the presence of an overwhelming predominance of phenotypically Th 2 -like cells during very advanced (< 200 CD4 + T cells/μl) HIV-1 infection.
Early studies have shown that CD4 + T CM are released from tissues into the bloodstream early after commencing ART [4], leading to the initial boost in CD4 counts. The present data suggest that these CD4 + T CM are primarily CCR4 + . Thus, the CCR4 + T CM cells appearing in the PBMC upon ART do not reflect a phenotypic change of cells preexisting in the blood stream-that is, an ART-induced change in the Th environment-but rather the appearance of a cell type previously sequestered in the tissues [50,51]. This is supported by our findings that, after one month of therapy, CD103 transcripts were specifically expressed by CCR4 + T CM , while no expression was detected in any of the subsets investigated prior to ART. The α E integrin chain (CD103) is typically found on intra-epithelial lymphocytes, allowing them to home to and circulate through mucosal sites [52]. It represents a rare transcript in peripheral blood CD4 + T cells, and indicates that ART induces the release of CCR4 + T CM from tissues.
Early after ART commencement, while PVL levels were still in the decline, these CCR4 + T CM were highly activated, expressing ICOS, HLA-DR, and PD-1. As the PVL were suppressed to undetectable levels at mo3 of ART, the activation of CCR4 + T CM also leveled off. However, there was no normalization of the Th 2 -like phenotype (% CCR4 + CD4 + T cells), even 1 year after ART.
Gene expression analyses confirmed that the CCR4 + T CM cells are largely comparable to classic Th 2 cells in their Th 1 -and Th 2 -associated transcriptome. However, the fact that CD103 mRNA was detected in CCR4 + T CM but not Th 2 -like cells indicated that the cells released from tissues are not classical Th 2 -like cells. Expression of IL21 and IL22 (and to a lesser degree IL17a) transcripts, cytokines not typically associated with Th 2 -like cells, suggests that the CCR4 + T CM contains a fraction of cells with a Th 17 -like functionality. Such cells are important in controlling bacterial infections at mucosal surfaces such as the gut and lungs [53]; the measured Th 17 -like functionality correlates well with expression of CD103 mRNA, the protein product of which has been implicated with homing to gut and skin [54].
The recovery kinetics of naïve and memory CD4 + T cells on ART have been shown to differ depending on the extent of a patient's CD4 + T-cell loss at the time of ART initiation [9]. Therefore, the present findings might not apply to all HIV-1 + individuals on ART, as we focused our study on severely immuno-compromised patients (< 200 CD4 + T cells/μl). In fact, we found an inverse correlation between CD4 + (P = 0.0008) or CD8 + (P = 0.0037) T-cell recovery and the CD4 + T-cell count pre-ART, reflecting a greater change in T-cell numbers experienced by those patients starting with a more severe T-cell depletion. Furthermore, individuals with lower pre-ART PVL were found to start with more differentiated CD4 + T cells and to exhibit a more rapid drop in late differentiation cells.
However, even though a substantial immune reconstitution occurs in the peripheral blood, it has been demonstrated that the mucosal immune system is more recalcitrant [55,56], with CD4 + T cells expressing CCR5 and/or CXCR4 remaining preferentially depleted [56].
Overall we found that even in these very advanced HIV-1 + patients the CD4 + and CD8 + T-cell compartments in peripheral blood slowly revert towards a more "healthy" phenotype, with an overall reduction in the expression of activation markers and molecules associated with inhibition of cellular functions, as well as an upregulation of factors associated with T-cell homeostasis and a more balanced immune system. For the most part, these changes commence early after ART initiation. An exception was an ART-induced increase in the frequency of initially highly activated peripheral CCR4 + T CM cells. This reflects a redistribution of these cells from peripheral tissue sites to peripheral blood, caused by a reduction in the viral burden, as highlighted by their expression of mRNA transcripts of the gut-and skin-homing integrin CD103. However, even 1 year post-therapy, the immune system maintains an unusually Th 2 -biased composition, potentially underlying continued immunodeficiency in the presence of higher CD4 + T-cell counts.  Supplementary Figure 4. T-cell subsets sorted for gene expression analysis. PBMC were stimulated with SEB for 3 hours, before sorting seven CD4 + T-cell subsets from SEB-reactive cells as indicated in this gating scheme. After identifying live CD3 + cells within singlet, aggregate-negative lymphocytes, non-naive CD4 + T cells were selected by excluding CD45RO − CCR7 + cells. Within these, T cells expressing TCR-Vβ12, -Vβ14 or -Vβ17 are known to react to SEB, so FITC-conjugated Abs were used for these three TCR-Vβ chains. Within such SEB-reactive cells, CCR4 + and CCR4− cells were selected and individually gated for T CM (CD45RO + CCR7 + CD27 + ) or T CM − subsets. Further, total SEB-reactive CD4 + T cells, as well as stringently gated Th 1 -(CCR4 − , CCR6 − , CCR10 − , CXCR3 + ) and Th 2 -like cells (CCR4 + CCR6 − CCR10 − CXCR3 − ) were also sorted. Figure 5. CCR4 + T CM resemble Th 2 -like cells in respect to their expression of Th 1 associated genes. PBMC from healthy donors, as well as cells isolated before or after 1 or 12 months of ART from HIV-1-infected adults were stained with the "sorting" panel (Supplementary Table 1). Subsets of CD4 + T cells were sorted as indicated in Supplementary Figure 4 and their gene expression profiles determined by multi-parametric quantitative RT-PCR. (A) Th 1 -associated genes were compared in Th 2like, CCR4 + T CM , and Th 1 -like cells isolated from healthy donors. (B) CCR4 + T CM from HIV-1 + patients after 1 month of ART were compared to their counterparts from healthy donors in respect to expression of Th 1 -associated genes. Bar graphs show interquartile ranges, median bars, as well as individual data points. Statistically significant differences are indicated: * P < 0.01, ** P < 0.001, *** P < 0.0001. Figure 6. Migration marker expression on CD4 + T cells. The expression of migration markers on the cell surface of CD4 + T cells was investigated by flow cytometry. (A) Dot plots show the expression of CD103, CCR9, integrin β7, CD49d, and CD11a on CD4 + T-cells from two different patients after 1 month of ART. (B) CD103 and CCR9 expression pre-ART and after 1 month of ART compared to healthy donors. (C) CD49d, CD11a, and integrin β7 expression pre-ART and after 1 month of ART compared to healthy donors. Bar graphs show interquartile ranges, median bars, as well as individual data points. Due to the small sample size, no statistical comparison was performed.