In vivo Expansion of Naïve CD4+CD25high FOXP3+ Regulatory T Cells in Patients with Colorectal Carcinoma after IL-2 Administration

Regulatory T cells (Treg cells) are increased in context of malignancies and their expansion can be correlated with higher disease burden and decreased survival. Initially, interleukin 2 (IL-2) has been used as T-cell growth factor in clinical vaccination trials. In murine models, however, a role of IL-2 in development, differentiation, homeostasis, and function of Treg cells was established. In IL-2 treated cancer patients a further Treg-cell expansion was described, yet, the mechanism of expansion is still elusive. Here we report that functional Treg cells of a naïve phenotype - as determined by CCR7 and CD45RA expression - are significantly expanded in colorectal cancer patients. Treatment of 15 UICC stage IV colorectal cancer patients with IL-2 in a phase I/II peptide vaccination trial further enlarges the already increased naïve Treg-cell pool. Higher frequencies of T-cell receptor excision circles in naïve Treg cells indicate IL-2 dependent thymic generation of naïve Treg cells as a mechanism leading to increased frequencies of Treg cells post IL-2 treatment in cancer patients. This finding could be confirmed in naïve murine Treg cells after IL-2 administration. These results point to a more complex regulation of Treg cells in context of IL-2 administration. Future strategies therefore might aim at combining IL-2 therapy with novel strategies to circumvent expansion and differentiation of naïve Treg cells.


Introduction
Human regulatory T cells (T reg cells) have been characterized as CD4 + CD25 high T cells with inhibitory function [1]. They are crucial for the preservation of T-cell homeostasis and self-tolerance and regulate the immune responses to alloantigens, pathogens and tumors [2]. Both in humans and animal models activation of T reg cells results in exertion of their full suppressive function [3,4]. Natural T reg cells are generated in the thymus as a distinct lineage of anergic CD4 + T cells bearing self-reactive T-cell receptors, although cells with similar characteristics can also be generated in the periphery under appropriate conditions [5]. Typically, T reg cells express cytotoxic T-lymphocyte-associated-antigen 4 (CTLA-4) and glucocorticoid-induced tumor-necrosis-factor receptorrelated protein (GITR), although both molecules can also be expressed by activated T cells [6,7]. The transcription factor forkhead box P3 (FOXP3) has been demonstrated to be expressed exclusively on T reg cells in the mouse [2], while data concerning its expression in humans are not as clear-cut [8,9]. In numerous murine tumor models increased frequencies of CD4 + CD25 high T reg cells seem to be a hallmark of tumor progression and metastasis [10,11]. Moreover, efficient anti-tumor immune responses are induced by deletion of these cells resulting in complete tumor regression [12,13]. In humans, we and others have demonstrated that CD4 + CD25 high FOXP3 + T reg cells are also expanded in patients with solid tumors and hematologic malignancies and contribute to the overall immunosuppression in these patients [14,15]. Numerous animal models over the last years could demonstrate that increased numbers of T reg cells are beneficial for tumor growth while depletion of T reg cells can lead to tumor regression [16]. In humans, administration of an IL-2 immunotoxin to tumor patients results in decreased numbers of T reg cells and higher responses against simultaneously administered tumor peptides [17,18]. Several studies over the last years have addressed the question of T reg -cell frequencies in colorectal cancer patients [19,20,21]. These reports could demonstrate increased number of FOXP3 + T reg cells in the peripheral blood, tumor-draining lymph nodes and in close tumor proximity [19,20,21]. Still, the question why T reg cells are expanded in human tumors remains elusive. CD45RO expression has been primarily linked to T reg cells, which led to the assumption that T reg cells belong to the memory T-cell compartment [4,22]. Recently, a T reg -cell population with a naïve phenotype (CCR7 + CD45RA + ) was identified in healthy individuals [23,24]. These naïve T reg cells proliferated vigorously in response to auto-antigens suggesting that particularly this subpopulation was specific for self rather than foreign antigens [23]. Persistence of naïve T reg cells has been described throughout adult life [25,26], although it must be pointed out that this pool of naïve T reg cells is relatively small in peripheral blood of healthy individuals [27]. In patients with multiple myeloma (MM), however, we could demonstrate an expansion of naïve T reg cells [28]. We could validate this finding for a number of hematologic malignancies and patients with solid tumors and even show strong suppressive function for the naïve T reg -cell population [29]. Comparison of levels of T-cell receptor excision circles (TREC) in T reg cells of healthy individuals addressed the diversity and developmental stage of thymic emigrants as well as peripheral blood T reg cells [30,31]. As expected, TREC numbers were significantly higher in thymic emigrants compared to peripheral blood derived T reg cells, which supports thymic development of human CD4 + CD25 high T reg cells [31].
On the one hand, IL-2 is both important for the development and expansion of effector T cells and also critical in the context of immune tolerance [32]. On the other hand, experiments performed in murine models established that IL-2 and its downstream effector molecules are essential for the generation, maintenance, and function of T reg cells [33,34,35,36]. Several studies have addressed the role of IL-2 administration on the frequency and function of human T reg cells in cancer patients and demonstrated that treatment with recombinant IL-2 induces an expansion of T reg cells in peripheral blood, thereby interfering with efficient anti-tumor immune responses [37,38,39,40,41,42,43]. These studies pointed to a peripheral expansion of T reg cells post IL-2 therapy [41]. It was further suggested that altered migratory behavior due to increased expression of CCR4 and CXCR4 on T reg cells might be associated with peripheral expansion and increased migration to the tumor site [41]. In two recent studies however, Correale et al. demonstrated that the combination of chemotherapy with GM-CSF and IL-2 administration leads to reduced T reg -cell numbers [44,45]. These findings suggested a more complex regulation of T reg -cell frequency, distribution and function than previously thought. Moreover, it remains unclear whether different mechanisms might be responsible for frequency changes of T reg cells in these different patient populations treated with IL-2.
In the current study, we investigated the impact of IL-2 administration on the frequency and function of CD4 + CD25 high -FOXP3 + T reg cells. We provide clear evidence for an increased expansion of naïve T reg cells particularly post IL-2 therapy as a major mechanism of overall T reg -cell expansion in these cancer patients. Our data clearly suggest an increase of newly generated T reg cells in cancer patients. These novel findings are of particular interest for strategies targeting T reg cells in cancer patients.

Results
Increased frequencies of CD4 + CD25 high FOXP3 + T reg cells in peripheral blood of patients with metastatic colorectal cancer Within a clinical phase I/II combined chemoimmunotherapy trial of patients with metastatic colorectal cancer we assessed frequencies of FOXP3-expressing CD4 + CD25 high T reg cells in peripheral blood before initiation of therapy in comparison to healthy controls (Fig. 1A). The frequency of T reg cells in healthy donors (n = 22, 2.9%61.2%) was comparable to previously published results (Fig. 1B) [46,47]. In contrast, individuals with colorectal cancer assessed before initiation of treatment (n = 15, 4.7%61.2%, p,0.001) showed significantly increased frequencies of T reg cells compared to healthy individuals ( Fig. 1A and 1B). We also assessed previously described surface receptors associated with T reg cells including CTLA4 and GITR on CD4 + CD25 high -FOXP3 + T cells in colorectal cancer patients as well as healthy individuals. As depicted in Figure 1C, we observed comparable expression of both molecules on T reg cells from healthy donors and colorectal cancer patients while percentages of CD4 + CD25 high -FOXP3 + T reg cells expressing either molecule were significantly increased in colorectal cancer patients (CTLA4: 3.7%61.2% vs. 1.5%60.6%, p,0.001; GITR: 1.7%60.8% vs. 0.5%60.2%, p,0.001) (Fig. 1D).
Expansion of CD4 + CD25 high FOXP3 + T reg cells after chemoimmunotherapy including low-dose IL-2 administration Next, we investigated changes in frequency of T reg cells in respect to IL-2 administration (Fig. 2). From the 15 patients included in the trial, 12 successfully completed the whole protocol, while 3 patients were rapidly progressing under therapy and deceased shortly thereafter. Blood samples from the 12 patients (P1-12) who completed the whole treatment were taken after completion of IL-2 treatment following vaccination. For the 3 rapidly progressing patients (P13-15) who were taken off study, T reg -cell frequencies were assessed at the last time point were material was available. As depicted for patient P11 in Figure 2A we observed an expansion of CD4 + CD25 high FOXP3 + T reg cells in the majority of patients. When assessing all patients, 11 colorectal cancer patients had higher frequencies post therapy while 2 (P3, P15) had lower frequencies at the second time point of analysis and two other patients (P4, P5) showed similar frequencies at both time points (Fig. 2B). No feature (laboratory test, treatment or clinical parameter) we have assessed so far showed an association with changes in T reg -cell frequency in these patients (data not shown). In addition, we did not observe any cytopenia greater grade 2 during therapy. Overall, in the majority of patients the frequency of T reg cells after combined chemoimmunotherapy was increased compared to the initial frequencies before treatment (5.8%61.7% vs. 4.7%61.2%, p,0.05) (Fig. 2C) as well as in comparison to healthy donors (5.8%61.7% vs. 2.9%61.2%, p,0.001). When assessing the absolute numbers of T reg cells we observed the same trend; total numbers of T reg cells were increased after chemoimmunotherapy (after: 29.2610 6 /l620.5610 6 /l vs. before: 21.36 10 6 /l617.1610 6 /l, p,0.005) and only one of the rapidly progressing patients showed reduced numbers of T reg cells (Fig. 2D).
Among the 15 colorectal cancer patients, 4 patients had progressive disease at the time of analysis while 11 showed a response or stable disease. Comparing these two patient cohorts revealed no significant difference in the proportion of T reg cells, whereas both patient cohorts showed higher frequencies of T reg cells compared to healthy controls (data not shown). Next, we assessed if T reg -cell frequencies were indicative of longer freedom from treatment failure or overall survival or if these 2 parameters were linked to the expansion of T reg cells, yet no statistically significant correlation was detected (data not shown). Our data indicate that an expansion of CD4 + CD25 high FOXP3 + T reg cells occurred in the majority of colorectal cancer patients after IL-2 administration as part of combined chemoimmunotherapy.

Inhibitory function of CD4 + CD25 high T reg cells from colorectal cancer patients
In 4 colorectal cancer patients sufficient numbers of highly purified CD4 + CD25 high T reg cells ( Fig. 3A and B) were isolated from peripheral blood by flow cytometric cell sorting before and after vaccination to analyze their inhibitory function in comparison to T reg cells from healthy controls (n = 4). Proliferation of allogeneic conventional CD4 + CD25 2 T cells stimulated by beads coated with CD3 and CD28 mAbs was used as the read out to assess inhibitory function of CD4 + CD25 high T reg cells [28]. Proliferation of allogeneic conventional CD4 + CD25 2 T cells was significantly inhibited when highly purified CD4 + CD25 high T cells from healthy donors were added at a 1:1 ratio (white bar, Fig. 3C, p,0.001). On a cell-to-cell basis highly purified CD4 + CD25 high T reg cells from colorectal cancer patients (before initiation of therapy) showed an equally strong inhibitory function on conventional CD4 + CD25 2 T-cell proliferation (dark grey bar, Fig. 3C, p,0.001). Titration experiments demonstrated that the inhibitory function of T reg cells from healthy individuals and colorectal cancer patients was comparable also at lower T reg :T conv ratios (data not shown). Moreover, after IL-2 treatment of colorectal cancer patients (after treatment), T reg cells had equal suppressive function on conventional CD4 + CD25 2 T-cell proliferation when compared to T reg cells isolated before start of therapy (light grey bar, Fig. 3C, p,0.001). Taken together, these data suggest, that T reg cells from colorectal cancer patients have normal suppressive function.   Expansion of fully functional naïve T reg cells in colorectal cancer patients after chemoimmunotherapy It has been previously reported by us and others that differential expression of the cell surface receptors CD45RA and CCR7 can be used to differentiate between naïve, central and effector memory T reg cells (Fig. 4A) [23,28,48]. This approach can be used to address the question, if naïve, central or effector memory T reg cells contribute to the overall expansion of T reg cells in patients with solid tumors and how treatment, particularly administration of IL-2 influences the frequency and function of naïve versus memory CD4 + CD25 high FOXP3 + T reg cells. In healthy individuals, naïve CCR7 + CD45RA + T reg cells were hardly detectable (Fig. 4B). T reg cells were almost exclusively of a memory phenotype ( Fig. 4C and D) with a higher frequency of T EM cells (Fig. 4D). In contrast, in colorectal cancer patients we observed a significantly higher number of T reg cells with a CCR7 + CD45RA + naïve phenotype (Fig. 4B). This was further accompanied by an increase of T reg cells with a central memory phenotype in colorectal cancer patients prior to therapy (Fig. 4C) while effector memory T reg cell levels were comparable in colorectal cancer patients (prior to therapy) and healthy donors (Fig. 4D). However, most surprising, after IL-2 treatment, expansion of T reg cells almost exclusively occurred within the naïve T reg -cell population ( Fig. 4B) while frequencies of central and effector memory T reg cells remained unchanged ( Fig. 4C and D). To further characterize the increased subset of naïve CD4 + CD25 high FOXP3 + T reg cells, we assessed the expression of intracellular CTLA4 or GITR expression. Comparable to the data obtained for the total T reg -cell population both molecules were expressed at similar levels on a per cell basis on naïve T reg -cells from healthy individuals and cancer patients irrespective of IL-2 treatment (data not shown) while percentages of CD4 + CD25 high FOXP3 + T reg cells expressing either molecule were significantly increased in colorectal cancer patients (CTLA4: 0.31%60.23% vs. 0.05%60.01%, p,0.05; GITR: 0.10%6 0.07% vs. 0.02%60.01%, p,0.05) with a further increase after IL-2 administration (CTLA4: 0.78%60.56% vs. 0.31%60.23%, p,0.001; GITR: 0.24%60.19% vs. 0.10%60.07%, p,0.05) (Fig. S1).
Next, we assessed the suppressive function of naïve versus memory T reg -cell populations before and after therapy (Fig. 4E). We observed a clear inhibition of conventional CD4 + CD25 2 Tcell proliferation induced by stimulation with CD3/CD28-coated beads by adding sorted T reg cells to the culture irrespective of the subtype (naïve vs. memory) used or the status of therapy (no therapy vs. chemoimmunotherapy) demonstrating that the different subtypes of T reg cells have full suppressive activity independent of IL-2 therapy. Taken together, these data demonstrate a significant expansion of naïve T reg cells post IL-2 treatment. Furthermore, these cells are similarly effective in suppressing conventional T-cell activation when analyzed on a cell-to-cell basis, suggesting an overall higher suppressive effect of these cells in colorectal cancer patients.
Thymic expansion of naïve CD4 + CD25 high FOXP3 + T reg cells in patients with colorectal cancer after IL-2 administration As T reg cells with a naïve phenotype were increased in patients with colorectal cancer, particularly post IL-2 treatment, we were interested to assess whether the increase of T reg cells resulted from peripheral expansion or possibly thymic generation of CD4 + CD25 high FOXP3 + T reg cells. We previously reported that the level of TREC can be used as a marker to estimate the developmental vicinity of T reg cells to the thymus and their division history [28] and therefore applied TREC analysis to address this issue in relation to IL-2 treatment. PB derived CD4 + T cells from two colorectal cancer patients and two age-matched healthy individuals were sorted according to their CD25, CCR7 and CD45RA expression into the appropriate CD4 + CD25 high T reg -cell subsets, namely T naïve , T CM , and T EM (Fig. 5A and B). TREC values for the highly purified T reg -cell subsets were assayed by real-time PCR. As described before, CD4 + CD25 high T reg cells showed relatively low TREC contents in the naïve T-cell population and TREC levels were below detection threshold in the T CM and T EM subsets (Fig. 5C) [28]. These observations are in line with the concept of antigen-driven peripheral expansion of T reg cells in healthy individuals rather than recent thymic emigration. In contrast, the TREC content on the single cell level in naïve CD4 + CD25 high T reg cells in colorectal cancer patients was more than two-fold higher in average compared to healthy individuals before initiation of chemoimmunotherapy and even more increased after administration of IL-2 (.4-fold in average, Fig. 5C). These results strongly suggest that the expansion of T reg cells in colorectal cancer patients was nurtured by the generation of new T reg cells in the thymus and the further increase of T reg -cell frequencies post IL-2 administration was strongly associated with an increase of thymic emigrants of naïve T reg cells.
Administration of IL-2 leads to an expansion of a population of murine ''naïve'' CD4 + CD25 high FOXP3 + T reg cells To demonstrate that IL-2 administration indeed leads to an increase of naïve T reg cells, we administered IL-2 or PBS i.p. in C57BL/6 mice for 10 days and assessed the frequency of CD4 + CD25 high FOXP3 + T reg cells in these animals after the treatment period. In comparison to PBS-treated animals a significant expansion of CD4 + CD25 high FOXP3 + T reg cells occurred after IL-2 administration in spleen, peripheral as well as mesenteric lymph nodes, peripheral blood, thymus, and liver (Fig. 6A). Of particular interest was the increase of T reg cells in the thymus, which could be indicative of an increased thymic output. As a first approximation for vicinity of T cells to thymic output expression of CD45RB on CD4 + CD25 high FOXP3 + T reg cells was assessed since up to now no true marker for murine naïve T reg cells has been established. We observed a significant increase of CD45RB high CD4 + CD25 high FOXP3 + T reg cells in spleen, peripheral as well as mesenteric lymph nodes, peripheral blood, thymus, and liver of IL-2 treated mice (Fig. 6B) clearly indicating that an increase of naïve T reg cells contributed to the overall expansion of T reg cells after IL-2 administration and that the overall T reg -cell expansion was related to elevated frequencies of naïve T reg cells in the thymus.
To further delineate if IL-2 treatment induces generation of naïve T reg cells in the thymus we sorted CD45RB + CD44 low CD62L + naïve T conv and T reg cells from IL-2 as well as PBS treated animals (Fig. 6C) and assessed TREC levels in the naïve T conv and T reg -cell population. We observed significantly higher levels of TREC in T conv and T reg -cell populations after IL-2 administration (Fig. 6C) suggesting that IL-2 treatment indeed results in a higher thymic output of naïve T reg cells as observed in IL-2 treated human colorectal cancer patients.

Discussion
In the current study we addressed the question whether IL-2 treatment leads to an expansion of fully functional T reg cells in colorectal cancer patients and whether an increase of T reg cells in colorectal cancer patients is due to peripheral or thymic expansion. Using phenotypic, functional and molecular approaches we demonstrate an increase of CD4 + CD25 high FOXP3 + T reg cells in colorectal cancer patients already before initiation of an IL-2 containing chemoimmunotherapy regimen. The increase of T reg cells in these patients was further augmented after completion of treatment, clearly demonstrating that IL-2 treatment is associated with an increase in T reg -cell numbers an alarming effect which could potentially influence and dampen the immune response in an immunotherapy protocol. T reg cells expressed phenotypic markers associated with T reg cells (e.g. CTLA-4 and GITR) and suppressed the proliferation of CD4 + CD25 2 T conv cells. Furthermore, we can clearly show that T reg cells in colorectal cancer patients have normal suppressive function.
In addition, we demonstrate that the expansion of T reg cells in colorectal cancer patients was particularly prominent within the naïve CCR7 + CD45RA + CD4 + CD25 high FOXP3 + T reg cell population and to a lesser extent in central memory T reg cells, while there was no difference in T reg cells with an effector memory phenotype. Most surprising, expansion of T reg cells post IL-2 treatment was almost exclusively due to an increase of naïve T reg cells. Expansion of naïve T reg cells was further supported by the increase of TREC numbers that were significantly higher in naïve T reg cells from untreated colorectal cancer patients when compared to healthy controls and further increased post IL-2 treatment. To demonstrate that the observed results in humans are dependent on the IL-2 administration, we treated mice with IL-2 and could detect increased levels of T reg cells in these animals. In line with our results in humans, we observed thymic generation of naïve T reg cells after IL-2 treatment as shown by high TREC levels in the naïve T reg -cell population. These findings strongly suggest that T reg -cell expansion in colorectal cancer patients is due to higher thymic output which is further increased by IL-2 treatment.
Over the last years several murine studies showed the importance of IL-2 for T reg cells. IL-2 is a critical growth factor of CD4 + CD25 high FOXP3 + T reg cells as exemplified for a representative healthy donor (left) and a representative colorectal cancer patient (right). Frequencies of (B) CCR7 + CD45RA + naïve T reg cells (T naive ), (C) CCR7 + CD45RA 2 central memory T reg cells (T CM ), and (D) CCR7 2 CD45RA 2 effector memory T reg cells (T EM ) were assessed in peripheral blood of colorectal cancer patients (n = 15) before (light grey bars) and after therapy (dark grey bars) as well as healthy individuals (white bars, n = 22). Significant differences (p,0.05, Student's t test) between healthy donors and colorectal cancer patients before and after chemoimmunotherapy are marked by an asterisk (*). Error bars represent SD. (E) Assessment of regulatory function of naïve and memory CD4 + CD25 high T reg cells sorted according to their CD45RA expression from colorectal cancer patients. Reduction of proliferation of CD4 + CD25 2 T conv cells stimulated with beads coated with CD3 and CD28 mAbs by highly purified naïve and memory CD4 + CD25 high T reg cells from colorectal cancer patients before and after therapy. doi:10.1371/journal.pone.0030422.g004 for murine T reg cells, it is responsible for maintenance and regulation of T reg cells in the periphery [33,34,35]. IL-2 seems to be involved in the generation of T reg cells during antigen-specific immune responses [49] and has been suggested to be involved in the suppressive function of T reg cells [50,51]. In human T reg -cell biology, IL-2 is supposed to induce a peripheral expansion of CD4 + CD25 high FOXP3 + T reg cells as suggested by Wei et al. [41] while it has no effect on FOXP3 expression in conventional T cells. This has been further confirmed in a second study which reported that IL-2 induced STAT-dependent mechanisms are responsible for the selective expression of FOXP3 in T reg cells and in vivo expansion following IL-2 administration indicating a unique programming of CD4 + CD25 high T reg cells within the IL-2 signaling pathway [52].
Most recently increased frequencies of CD4 + CD25 high FOXP3 + T reg cells were reported for patients with renal cell carcinoma, malignant melanoma or ovarian cancer patients after IL-2 monotherapy [37,39,40,41] and administration of IL-2 during immune reconstitution after chemotherapy in pediatric sarcomas led to a preferential expansion of T reg cells after cytoreductive chemptherapy [38]. In contrast, patients with metastatic colorectal cancer treated with a combined chemoimmunotherapy containing gemcitabine and FOL-FOX-4 (oxaliplatin, fluorouracil, and folinic acid) polychemotherapy followed by the subcutaneous administration of GM-CSF and lowdose IL-2 showed clinical objective responses in the majority of patients associated with a significant reduction in CD4 + CD25 high FOXP3 + T reg cells as reported by Correale et al. [44,45].
Our data however support an alternative outcome of combined chemoimmunotherapy as low-dose IL-2 in combination with a peptide-vaccination resulted in increased frequencies of CD4 + CD25 high FOXP3 + T reg cells, particularly naïve T reg cells.
These rather opposite results might be explained by the differences in the vaccination and chemotherapy protocols, e.g. high-dose vs. low-dose 5-FU, irinotecan vs. oxaliplatin, forgoing of gemcitabine, dosage and schedule of GM-CSF administration, or addition of CAP-1-peptide, in the time course of administration, in the patient cohort under study (metastatic colorectal cancer vs. unselected colorectal cancer), number of patients analyzed, and additional, yet unknown, confounding factors. Further conflicting might be technical issues as it has been previously stated by Baecher-Allan et al. that the assessment of human T reg cells is still difficult and the use of different assays sometimes makes it difficult to compare different studies [47]. Indeed, a recent study in patients with renal cell carcinoma or malignant melanoma suggested an enormously high proportion of naïve T reg cells [39]. However, the number of naïve T reg cells in the healthy control group was also reported to exceed 50% of all T reg cells, a frequency that could never be confirmed by us and others [24,27].
Irrespective of these experimental differences in earlier studies, we clearly unraveled the expansion of naïve T reg cells to be one of the mechanisms leading to an overall expansion of fully functional T reg cells in colorectal cancer patients which was further augmented by IL-2 therapy. Over the last years differentiation of T reg cells into naïve, central and effector memory T reg cells according to their expression of CCR7 and CD45RA has been established for healthy individuals [23,24,25,31,53]. We have incorporated this strategy for the analysis of T reg cells in cancer patients and demonstrated an expansion of naïve T reg cells in human multiple myeloma patients and B-CLL patients [28]. Remarkably, in B cell malignancies, increase of T reg cells was associated with peripheral expansion of naïve T reg cells while in patients with colorectal cancer the expansion seems to be thymus  . IL-2 administration leads to an expansion of ''naïve'' CD4 + CD25 high FOXP3 + T reg cells in C57BL/6 mice. (A) Flow cytometric analysis of CD4 and FOXP3 expression in CD4 + T cells from untreated as well as IL-2-treated animals in spleen, peripheral and mesenteric lymph nodes, peripheral blood, thymus, and liver. Significant differences (p,0.05, Student's t test) between untreated and IL-2 treated animals are marked by an asterisk (*). (B) Analysis of ''naïve'' CD45RB high CD4 + CD25 high FOXP3 + T reg cells in spleen, peripheral and mesenteric lymph nodes, peripheral blood, thymus, and liver. Significant differences (p,0.05, Student's t test) between untreated and IL-2 treated animals are marked by an asterisk (*). Similar results were obtained in two independent experiments. (C) CD45RB + CD44 low CD62L + naïve CD4 + CD25 2 T conv and CD4 + CD25 high T reg cells were isolated by flow cytometric cell sorting and assessed for TREC content. Genomic DNA of sorted subsets was isolated, and the number of TREC was determined by quantitative real-time PCR (n = 3, p,0.05, Student's t test). Error bars represent SD. Similar results were obtained in three independent experiments. doi:10.1371/journal.pone.0030422.g006 dependent as determined by analysis of TREC as an approximation of adjacency of T reg cells to the thymus. Moreover, administration of IL-2 further augmented this increase. Two recent publications underlined the importance of IL-2 for the development of T reg cells in the thymus [54,55], while other results support a more restricted role for IL-2 of regulating T reg cells in the periphery without alteration of the thymic output [34]. The latter proposition has been further strengthened by data obtained by Wei et al. who could demonstrate expansion of memory T reg cells in vitro by IL-2 treatment [41]. To address whether effects induced by IL-2 were limited to peripheral expansion of mainly memory T reg cells, we treated C57BL/6 mice with IL-2 and assessed the frequencies of T reg cells with a special focus on naïve T reg cells in the periphery and in the thymus. We observed an overall expansion of T reg cells in these animals and detected a particular enrichment of naïve T reg cells in the thymus and peripheral lymph nodes. By assessment of TREC in the naïve T regcell population we could demonstrate thymic generation of naïve T reg cells as a result of IL-2 adminstration, clearly suggesting that increased thymic output is an important mechanism leading to increased frequencies of naïve T reg cells post IL-2 treatment and these cells subsequently can differentiate into peripheral T reg cells with a memory phenotype. Whether peripheral expansion might also occur under these conditions might be further studied by administration of IL-2 to thymectomized animals.
The assessment of the source of expanded T reg cells in cancer patients is of particular importance as the mechanism of expansion, an augmented production of T reg cells in the thymus, expansion of T reg cells in the periphery, preferential migration of T reg cells to the tumor site, increased conversion of conventional T cells into T reg cells as well as diminished apoptosis and cell death of peripheral T reg cells, might influence the strategy to therapeutically target T reg cells to increase anti-tumor immunity. E.g. peripherally expanded T reg cells might be deleted by short-term use of cytotoxic agents such as denileukin diftitox while continuing thymic expansion would require either long-term treatment or deletion of the cause of thymic expansion of these inhibitory cells to induce a long-lasting reduction of T reg cells. Other mechanisms such as conversion of differentiated conventional T cells into T reg cells might also contribute to the overall expansion of T reg cells in cancer patients as well the preferential egress of activated Treg cell from the tumor microenvironment into the peripheral blood and will have to be addressed when targeting T reg cells [5].
Taken together, we demonstrate an in vivo expansion of fully functional CD4 + CD25 high FOXP3 + T reg cells in colorectal cancer patients due to an increase in naïve T reg cells with an increased TREC content. Moreover, naïve T reg cells with a further increase of TREC are expanded post IL-2 treatment clearly pointing to an increased thymic output of naïve T reg cells after IL-2 therapy, a mechanism also observed in mice in vivo post IL-2 treatment. This expansion of T reg cells post administration of IL-2 can potentially hinder an immune response towards co-administered anti-tumor reagents and should therefore be avoided when planning new cancer immunotherapy protocols. The existence of different mechanisms of expansion of T reg cells (thymic vs. peripheral expansion and conversion) highlights the complexity of regulation of these cells and cautions the use of simple strategies targeting these highly regulated cells in future cancer immunotherapy approaches.

Patients and clinical parameters
15 HLA-A2 + patients with primary metastatic colorectal cancer were enrolled in this Phase I/II combined chemoimmunotherapy with a HLA-A2 peptide derived from the carcinoembryonic antigen CEA (CAP-1) and irinotecan, 5-fluorouracil, and leucovorin after approval by the institutional review committee at the University of Cologne and the German Drug Administration [56]. All patients signed informed consent. Mean age was 56.2613.1 years; 8 were male and 7 were female; 12 had colon while 3 had rectal cancer. Mean age for the corresponding healthy controls was 49.569.3 years with no significant differences in gender and age. Inclusion criteria required an age of between 18 and 75 years, positive HLA-A2 status, elevated serum CEA (.5 mg/l) and/or CEA-positive tumor, untreated metastatic disease, chemotherapyfree interval after adjuvant treatment of at least 6 months, Karnofsky index .70%, life expectancy of at least 3 months, sufficient bone marrow and liver function, HIV and hepatitis B and C negativity, absence of central nervous system metastases, no immunosuppressant medication, and negative pregnancy test. All patients received low-dose IL-2 (1610 6 IU) post vaccination. For vaccination, several approaches were compared in this study, namely a vaccine containing only the CAP-1-peptide, or the CAP-1-peptide together with 50 mg GM-CSF (Novartis) or the CAP-1peptide together with an oligonucleotide adjuvant (dSLIM, Mologen); as a fourth option, patients obtained autologous CAP-1-pulsed dendritic cells as a cellular vaccine. For the analysis of T reg -cell frequency and function post IL-2 treatment, no statistically significant influence of the different types of vaccinations could be established (data not shown). Two cycles of vaccination and IL-2 treatment were given 2 and 1 week before the first cycle of chemotherapy consisting of 80 mg/m 2 irinotecan, 2,000 mg/m 2 high-dose 5-FU, and 500 mg/m 2 leucovorin (six weekly administrations). Alternating, two cycles of IL-2 and vaccination were combined with one cycle of chemotherapy and repeated three times. After the third cycle, patients were treated with IL-2 and CAP-1-peptide vaccination on a weekly schedule until progressive disease. At baseline and after three cycles of chemoimmunotherapy, patients underwent a leukapheresis to obtain PBMC for diagnostic and therapeutic purposes. PBMC were isolated using Ficoll/Hypaque (Amersham, Uppsala, Sweden) density centrifugation. Staging was performed according to the UICC classification for colorectal cancer. All patients were UICC stage IV. The treatment schedule is provided as Figure S2. Characteristics of the patients studied are summarized in Table 1.
Samples were acquired on a FACSCanto and analyzed with FlowJo software (TreeStar Inc.). Frequencies of CD4 + CD25 high -FOXP3 + T cells are shown as percent values of CD4 + T cells.
Isolation of CD4 + CD25 high and CD4 + CD25 2 T cells For functional analysis, CD4 + CD25 high T cells were purified from PBMC. Briefly, CD4 MACS Beads (Miltenyi Biotec) were used for isolation of CD4 + T cells [28,46]. After staining with CD25-PE, and CD4-APC (BDPharMingen) according to the manufacturer's recommendations, CD4 + CD25 high T cells were purified using a FACSDiVa Cell Sorter (BDBiosciences) and used for functional characterization. The CD4 + CD25 2 T cells isolated from healthy individuals were used as effectors to assess T reg cell function independently of potential defects of conventional CD4 + T cells from colorectal cancer patients [28,46]. The cells were reanalyzed for FOXP3 expression after sorting and routinely showed .95% purity.
Isolation of CD4 + CD25 high T-cell subpopulations for assessment of T-cell receptor excision circles and functional characterization Briefly, CD4 MACS Beads were used for isolation of CD4 + T cells [46]. After staining with CCR7-FITC, CD25-PE, CD45RA-PE-Cy-5, and CD4-APC, CD4 + CD25 high T cells and the respective T cell subsets, CCR7 + CD45RA + T naïve , CCR7 + CD45RA 2 T CM , and CCR7 2 CD45RA 2 T EM cells were purified using a FACSDiVa Cell Sorter and used for either functional characterization or assessment of TREC levels.
DNA was isolated from purified CD4 + CD25 2 , CD4 + CD25 low and CD4 + CD25 high T naïve , T CM , and T EM cells respectively using a DNA Isolation Kit (Roche Diagnostics) following the manufacturer's instructions.
Relative TREC levels were determined using real-time PCR with a LightCycler (Roche Diagnostics) based on specific primers and general fluorescence detection with SYBR Green. All PCR were performed using LightCycler-FastStart DNA Master SYBR Green I kit (Roche Diagnostics). All samples were studied in duplicate reactions using the human TREC primer kit (Search-LC). The number of TREC molecules in the sample was calculated as number of copies per 10 4 cells (detection limit $10 molecules).

Statistical analysis
Comparison between paired or unpaired groups was performed using the appropriate Student's t-test. A p-value,0.05 was defined as statistically significant. All statistical analyses were performed using the SPSS statistical software package (SPSS 19, SPSS Inc.). Figures were created using SigmaPlot 12.0 (Systat Software Inc). Figure S1 CTLA4 and GITR expression in naïve T reg cells. Frequency of CTLA4 (left) and GITR (right) expressing naïve CCR7 + CD45RA + CD4 + CD25 high FOXP3 + T reg cells in healthy donors (white) and colorectal cancer patients before (light grey) and after IL-2 administration (dark grey) before treatment. Shown here are median, 25 th and 75 th percentile (box), 10 th and 90 th percentile (whiskers) and outliers (dots), (*, p,0.05, Student's t test). (TIF) Figure S2 Therapy schedule. Patients were first randomized to receive CAP-1 and IL-2 with different adjuvants (dSLIM, GM-CSF, or none). Subsequently, they were randomized to receive their first vaccination with or without pulsed autologous dendritic cells. Vaccinations (V) and chemotherapy (Chemo) were given in an alternating schedule, starting with two vaccinations. (TIF)