The pathogenetic role of natural regulatory T cells (Tregs) in HIV infection is still ill defined. Treg-induced suppression can be deleterious for the development of an efficient anti-HIV specific immune response, but it may be also helpful in preventing an inappropriate immune activation [1,2]. Indeed, the increase of Tregs in HIV⁺ patients has been linked to a reduced immune activation and low viral load [2] but the accumulation of Tregs in lymphoid tissues was also found to be associated with a high viremia [3], both in infected patients and simian immunodeficiency virus-infected macaques [4]. The data about the frequencies of circulating Tregs in HIV⁺ patients are conflicting [5-7], most likely because of the different biological samples (whole blood vs separated peripheral blood cells) or patients being analyzed. For example, one study demonstrated a lower Treg frequency in viremic patients compared with aviremic subjects [8], while another showed an increased Treg frequency only in the viremic HIV⁺ individuals with a late-stage disease and with a reduced CD4⁺ cell number [9]. Another explanation of these discrepancies relies on the different procedures employed for the characterization of Tregs: initially, they have been identified with a suboptimal CD4⁺CD25^high phenotyping [10,11]; then, using the more universally accepted CD4⁺CD25⁺FOXP3⁺ phenotype, which requires an intracellular staining [3,12]; more recently, Tregs have also been identified as CD4⁺CD25⁺CD127^low/− cells because of a low expression of the IL-7 receptor (CD127)and of a tight correlation with CD4⁺CD25⁺FOXP3⁺ cells [13-15].

The heterogeneity of the Treg subpopulations [16], which include the naïve- (CD45RA⁺) and memory-like (CD45RA⁻) subpopulations [13,17], should be also taken into account. Among naïve Tregs, the expression of the CD31 antigen, which identifies T cells recently released from the thymus [18], allows the identification of “recent thymic emigrants” (RTE)-Tregs, which are important in the context of Treg homeostasis [19, 20]. Memory Tregs, similarly to conventional central memory (T_CM) and effector memory (T_EM) T cells, can be further divided into central memory-like (Treg_CM) and effector memory-like (Treg_EM) on the basis of the chemokine receptor 7 (CCR7) expression [21, 22]. According to Menning et al. [23] the expression of the homing receptor CCR7 differentially regulates the in vivo function of Tregs: Treg_CM control the onset of immune responses in lymphoid organs like conventional CD4⁺ T_CM cells, while Treg_EM modulate the suppressive phase in peripheral tissues [22]. Finally, according to the variable expression of FOXP3 and CD25 in CD45RA⁺ or CD45RA⁻ cells, it is possible to identify two further subsets of Tregs endowed with different regulatory potential: resting Tregs (called rTregs [20] or naïve-like nTregs [24], depending on the study), which display the reduced suppressive capacity and properties of naïve cells, and activated Tregs (aTregs), which have an increased suppressive capacity and can be assimilated to memory-like Tregs [17, 20, 24].

To our knowledge, no studies have analyzed all these subsets (naïve Tregs, RTE-Tregs, Treg_CM, Treg_EM, rTregs, and aTregs) in HIV-infected individuals. Therefore, we quantified all Treg subpopulations in two groups of HIV-infected patients with similar CD4⁺ cell counts, one viremic but not requiring treatment because of a relatively conserved number of CD4⁺ cells and the other aviremic because of a long lasting (more than 6 years) combined antiretroviral therapy (cART), in order to investigate whether the modulation of the different Treg subsets could help explaining the preserved immunity observed in the first group of patients.

The percentage and number of the total Tregs, identified as CD4⁺CD25⁺CD127^low/− cells, did not differ in the untreated viremic and long-term treated aviremic patients from that of HD (7.8% vs 8.5% vs 8.6% and 47 cells/µL vs 45 cells/µL vs 55 cells/µL; p=NS). A similar result was found when an analysis of covariance was performed to adjust for the effects of age (not shown). Differently from the above, by analysing the Treg subsets we found a redistribution of these cells in the untreated viremic patients, with an increase in the proportion (but not in the number) of naïve Tregs at the detriment of memory Tregs and, in particular, of the Treg_CM cells. Indeed, these cells were significantly decreased both in terms of percentage (Fig. 1C) and total number (not shown), while only the total number, and not the percentage, of Treg_EM was reduced (data not shown).

A differential redistribution of Treg subsets was also found when these cells were identified as nTregs and aTregs (Fig. 1D), which display features of naive-like and memory-like Tregs, respectively [17, 20, 24]. We found, in fact, a significant increase of nTregs only in the untreated viremic patients and a strong positive correlation between the percentage of naive Tregs and nTregs, both when data were analyzed altogether (r=0.83, p<0.001) and in each patient group (Fig. 1E). Likewise, positive correlations were present between the total percentage of memory Tregs (Treg_{EM +} Treg_CM) and aTregs (Appendix Fig. 1). These changes in the Treg subset redistribution were not due to the known effects of age on new T-cell production because the analysis of covariance did not detect any significant group-specific age-mediated effect (p=0.31), but confirmed that the population of naïve Treg cells was more represented in the untreated than in cART-treated patients and HD (age-adjusted means: 45.8% vs 36.6% vs 32.2%, respectively; p<0.001).

The increased naive Tregs were not likely to result from an increased output of naive Tregs from the thymus because RTE-Tregs, which represent the subset of Tregs recently released from the thymus [19, 20], were not different in the three groups (4% vs 4% vs 4%, and 4.7 cells/µL vs 8.5 cells/µL vs 6 cells/µL; p=NS), even if data were age-adjusted (not shown). Thus, the present findings may be explained by a slow differentiation rate of naïve Tregs into Treg_CM and/or a preferential infection and killing of Treg_CM [25]. Memory Tregs are in fact highly proliferative compared to their naïve counterparts, and in this situation, which favours viral replication [26], Treg_CM could represent a population more prone to apoptosis [20]. Accordingly, we found that a higher viral load correlated with a lower percentage of Treg_CM (r=−0.51; p=0.016). Thus, an accelerated turnover of Treg_CM in conditions of high viral replication could drive a homeostatic expansion of naïve Tregs, whose proportion, in turn, was found to be positively correlated with the viremia (r=0.51; p=0.017). This interpretation implies the assumption that homeostatically expanded naive cells accumulated because unable to differentiate into memory cells, in contrast to what would be normally expected [27]. Indeed, whereas this process is normally dependent on IL-7 stimulation [27], in naive Tregs of viremic patients the downstream parts of IL-7 signalling machinery are known to be disrupted [28].

Alternatively, the reduced proportion of circulating Treg_CM in the blood of untreated patients could merely reflect an enhanced retention of this population in lymphoid organs. In fact, while Treg_CM and Treg_EM should not differ in terms of regulatory capacity [22], the differential expression of CCR7 empowers them to migrate to the lymphoid and the peripheral tissues, respectively, to perform their suppressive functions [22, 23]. In the lymphatic tissues, Treg_CM might underlie an increased immune suppression that would be responsible, on one hand, of a less efficient control of viral replication (shown by the residual viremia), and, on the other hand, of a lower degree of chronic immune activation, which, in turn, could explain the conserved CD4⁺ cell number present in the viremic patients even if untreated. Interestingly, this interpretation fits with observations made in African green monkeys infected by simian immunodeficiency virus [4].

Differently from the untreated, in the cART-treated patients the percentage of naïve and Treg_CM were similar to that of HD (in whom Treg_CM is usually the most represented subset) [29]. In addition, only in treated patients and not in the untreated (data not shown), a highly significant, strong positive correlation between the percentage of naïve Tregs and the CD4⁺ cell count, and an inverse correlation between Treg_CM and CD4⁺ cells were found (Fig. 1F). As already suggested by Simonetta et al. [17], this indicated that the relationship between naïve Tregs and the CD4⁺ cell count is restored by the long lasting cART. On the other hand, our data differ from those of the same authors reporting that the naïve-like Treg number was preserved in HIV-infected patients regardless of treatment [17]. This discrepancy could be explained either by the different methodological approaches, or by the different HIV plasma levels in the patients of the two studies. The strong correlation found between the percentage of naive Tregs and the viral load (r=0.51, p=0.016) supports the second hypothesis. Therefore, although performed with different methodological approaches and type of patients, our study, at least in part, complements that of Simonetta et al. [17] by demonstrating that a long lasting effective cART restores Treg homeostasis, most likely by hampering HIV replication in the periphery and into the lymphoid sanctuaries [30, 31].

It is also not straightforward to compare our results with those by Zhou et al. [24], who found that nTregs were greatly reduced while aTregs were increased in chronically infected, treatment-naive viremic patients (but not in those acutely-infected). Indeed, in the mentioned study patients were therapy-naive at the time of recruitment, and then started treatment because of disease progression, whereas our untreated patients were stable, chronically infected subjects never undergoing treatment throughout the study period. On the other hand, even if we could not perform a Treg quantification before treatment beginning, our findings in the long-term cART-treated patients are in agreement with those of Zhou et al. [24] in demonstrating that the extent of post-therapy nTreg recovery relied on the amount of CD4⁺ cell number before therapy. Indeed, by stratifying our patients in two groups on the basis of the routinely-tested pre-therapy CD4⁺ cell number, we found that the higher median percentage of both naive Tregs (33% vs 23%, p=0.046) and nTregs (2.10% vs 1.23%, p= 0.010) was measured in the group of patients with a baseline CD4⁺ cell count over 200 cells/µL.

In the context of our study, one concern may arise from the method employed to identify Treg cells, because the down-regulation of CD127 due to the HIV-induced chronic immune activation [32, 33] may lead to an overestimation of Tregs in viremic patients, when these cells are identified only according to the lack of CD127 and not to the presence of FOXP3 [8]. However, also the more commonly used Treg marker FOXP3 is transiently up-modulated after cell activation in a fraction of effector, non-regulatory cells [34], and FOXP3⁺CD25⁺CD45RA⁻ cells have also been reported to be non-regulatory [20]. Furthermore, even in viremic patients, the overwhelming majority of CD25^highCD127^low cells have been shown to coincide with the FOXP3^high population [7]. To definitively rule out the possibility that our results could be technically biased, we analyzed the expression of the CD127 marker within the total naive and memory T-cell subsets. In contrast to the selective increase of CD127^low/− cells within the naive Treg subset and to their decrease within the Treg_CM or Treg_EM subsets in the untreated viremic patients (Fig. 1C), an increased frequency of CD127^low/− cells was found within the T_CM and T_EM subsets, but not within the naive one in the same patients (Fig. 2). This result indicates that the present findings are not an artefact mirroring the chronic immune cell activation typical of viremic patients, but are likely to represent a modulation of bona fide Treg cell subsets.

In conclusion, we propose that the redistribution of Treg subsets contributes to explain the partially preserved CD4⁺ cell counts despite the ongoing viremia seen in untreated patients. Therefore, an analysis of the fine modulation of the Treg subsets, rather than a study of the total Treg population, should be performed for a better understanding of the Treg role in HIV infection and therapy. However, the heterogeneity of HIV patients and methods advocates the need of further studies, aimed at identifying the most appropriate Treg definition in the different subsets of HIV patients.