We hypothesized that naïve T cells entering the tumor microenvironment would be exposed to the signals and factors needed for induction of Foxp3 and the acquisition of a regulatory phenotype. Some of these signals have been defined in vitro and include TCR engagement and exposure to TGFβ. We used TCR transgenic mice - one devoid of CD4 nTreg and the other with < 1% Foxp3 CD8s - whose tumor-infiltrating T cells and perhaps those in lymphoid organs of tumor-bearing mice would be engaged by cognate antigen, as well as being exposed to tumor-produced TGFβ. We demonstrated that naïve Foxp3-negative CD4 and CD8 populations, both transgenic and nontransgenic, could be induced to express Foxp3 in vitro by αCD3/αCD28 and TGFβ treatment (see Table 1). These in vitro findings confirm that Foxp3 expression can be induced in both T cells subtypes, and to comparable degrees, using similar sets of signals. We confirmed a significant enrichment of Foxp3-positive CD4 TIL and splenocytes in wild-type, B16 tumor-bearing C57BL/6 Foxp3EGFP mice. However, neither CD8s in these mice, nor TCR transgenic T cells in either Foxp3EGFP cross, seemed to have encountered the necessary intratumoral signals to induce Foxp3 expression. Administration of systemic IL-2, which we have shown in vitro to act in concert, but not alone, with TGFβ to induce Foxp3, did not support induction of iTreg in these transgenic mice. These findings indirectly support the view that natural, thymus-derived Treg preferentially accumulate, or proliferate, in the tumor microenvironment.
Two subsets of Tregs are recognized - adaptive or induced (iTreg) and natural (nTreg) - which together are responsible for maintaining tolerance to self-antigen and preventing autoimmune diseases or inappropriate immune responses involved in allergic diseases through the suppression of auto-reactive T cells [30, 31]. Both require TGF-β and IL-2 for maintenance; express similar phenotypic markers such as CTLA4, GITR, CD25, and CCR4; and require the expression of Foxp3 to carry out a contact-dependent mechanism of action . Where these two developmentally distinct populations differ is in their antigen specificities, strength of TCR stimulation and co-stimulatory signals required for their generation, and their stability of suppressive action. nTreg are generated in the thymus in a CD28-dependent manner, constitutively express CD25, express TCRs specific for self-antigen, demonstrate a more stable expression of Foxp3, and exert suppressive function [30–32]. Conversely, iTregs are generated in the peripheral lymphoid organs through the de novo conversion of CD4+CD25-Foxp3- T cells in a TGF-β- and IL-2-dependent-manner, have TCRs specific for foreign antigens presented by professional antigen-presenting cells, require weaker TCR stimulation (CD28 stimulation not required), and demonstrate a less stable expression of Foxp3 [30–34].
Possible mechanisms for Treg enrichment within tumors include: preferential accumulation of nTreg, proliferation of nTreg within tumors, or peripheral conversion of naïve T cells to iTreg. Each mechanism is dependent upon tumor-derived signals (chemokines, cytokines, TCR engagement). One hypothesis is that resident nTreg proliferate in the tumor microenvironment ; another is that nTregs may weakly perceive tumor related signals and be selectively recruited where they may exert their suppressive function . Treg can migrate more efficiently into major non-lymphoid tissue sites, such as tumors, due to their up-regulation of non-lymphoid tissue-specific homing receptors . In human ovarian carcinoma, tumor cells and tumor-associated macrophages were shown to produce a Treg-specific chemokine CCL22, which resulted in the specific recruitment of Treg to the tumor microenvironment via their CCR4 receptor . Interestingly, the CCR4 receptor is expressed at higher levels in Treg than in effector T cells in leukemia studies, suggesting its up-regulation may be Treg-specific . In this same ovarian cancer model, IL-2 treatment was shown to up-regulate CCR4 as well as CXCR4, which further enhanced the ability of Treg to migrate to the tumor microenvironment based on its elevated levels of the Treg ligands CCL22 and CXCL12 [38, 39]. Similar results were seen in a gastric cancer model, in which elevated levels of CCL17 and CCL26 in the tumor microenvironment demonstrated a positive correlation with the frequency of Foxp3 Treg, which can bind both these ligands with its CCR4 receptor . Furthermore, Treg had a higher affinity for CCL17 and CCL22 than effector T cells in vitro as determined by a migration assay.
Expression of the regulatory cytokine TGFβ is abundant in many tumors, particularly in advanced stages. TGFβ can promote proliferation of Treg in vivo; the addition of IL-2 to TGFβ can induce Treg both in vivo and in vitro . A number of additional chemokines and cytokines are implicated in Treg proliferation or induction, including IFNγ, IL-6, IL-23, IL-21, Cox2, and indolamine 2,3 dioxygenase [22, 23, 42–45]. Convincing in vivo evidence that iTreg develop from conventional CD4+CD25- T cells was shown a non-obese diabetic (NOD) model, which normally exhibits extensive autoimmune manifestations around 12-16 weeks of age. When CD28-/- NOD mice were treated with αCD3, which has been shown to increase CD4+CD25+ T cells and result in long-term remission of disease in regular NOD, CD4+CD25+ T cells were generated de novo and shown to be suppressive in vitro . Neutralizing anti-TGF-β given to these CD28-/- NOD mice alongside αCD3 prevented the aforementioned disease remission, demonstrating the role of TGF-β in the generation of iTreg and further supporting previous in vitro results.
Linehan [47, 48] demonstrated that naïve CD4 T cells were converted into Foxp3-positive Treg when administered to Rag1-/- mice bearing TGFβ-producing (Pan 02) tumors. This conversion could be inhibited by TGFβ neutralizing antibody, was abrogated if naïve T cells were obtained from mice whose T cells were insensitive to TGFβ signaling, and did not occur in mice bearing tumors that did not produce TGFβ (ESO 2). The high levels of TGFβ expression by this murine pancreatic tumor cell line, and the longer interval (seven weeks) of in vivo co-residence of naïve CD4 T cells and tumor, may explain their clearcut but different findings.