A weak Foxp3 hypomorph enhances spontaneous and therapeutic immune surveillance of cancer in mice

It is well established that therapeutic impairment of Foxp3+ regulatory T cells (Treg) in mice and humans favors immune rejection of solid tumors. Less explored are the genetic associations between Foxp3 allelic variants and tumor incidence, only sporadically reported in human studies. In this work, we tested and demonstrate that Foxp3fGFP, an allele classified as hypomorphic in Th1 inflammatory contexts but not affecting health at steady state, confers increased anti-tumor immunity. Our conclusions stem out of the analysis of three tumor models of different tissue origin, in two murine genetic backgrounds. When compared to wild type animals, mice carrying the Foxp3fGFP allele spontaneously delay, reduce or prevent primary tumor growth, decrease metastasis growth and potentiate the response to anti-CTLA4 monotherapy. These findings suggest that allelic variance at the Foxp3 locus may have significant impact on cancer incidence and/or the success of cancer-immunotherapies in humans.


Regulatory T cells (Treg), a subset of CD4 + cells, express the transcription factor
Foxp3 that defines a transcriptional profile essential for their differentiation and function (1). By controlling the activation of conventional T cells, Treg guarantee the establishment and maintenance of immune tolerance to self-components (2). It is also well established that depletion or inhibition of Treg in mice and humans, favors immune rejection of solid tumors (3,4). Several allelic variants in humans have been associated with various autoimmune diseases (5) and loss of function mutations in the Foxp3 gene are responsible for the fatal IPEX syndrome (6). Foxp3 allelic variants were also associated with increased susceptibility to colorectal and non-small cell lung cancer, and progression of breast cancer (5). However, Foxp3 expression is not restricted to Treg and acts as a cell intrinsic tumor suppressor that represses the oncogenes SKP2, HER2 (7,8) or cMYC (9) in solid tumors. Thus, it remains unclear whether allelic variants of the Foxp3 gene can affect immune surveillance of cancer.
In turn, it is conceivable that protective Foxp3 alleles may also enhance the effectiveness of immune-therapies for cancer.
The development of Foxp3 reporters in mice fortuitously generated Foxp3 alleles that are functionally impaired, to various degrees (10)(11)(12)(13). The commonly used Foxp3 fGFP knock-in allele, that encodes a Foxp3 protein fused at its N-terminus to the enhanced green fluorescence protein (eGFP), moderately alters the transcriptional signature and phenotype of Treg, with functional impacts in models of spontaneous autoimmunity and infection (11,12,14). As the Foxp3 fGFP allele does not affect health at steady state in the reference C57Bl/6 (B6) genetic background (12), it is an ideal model to test for specific effect on tumor progression. To dissociate tumorigenesis from anti-cancer immunity, we used three transplantable tumor models. We evidence reduced primary tumor growth associated with increased immune responses, reduced metastatic progression and enhanced response to anti-CTLA4 monotherapy in Foxp3-GFP mice when compared to wild-type (WT) controls. This preclinical analysis supports the notion that allelic variance at the Foxp3 locus may serve as predictive indicators for personalised therapy and prognostics, to the benefit of cancer patients.

Results and Discussion
The Foxp3 fGFP allele enhances spontaneous immune-surveillance of primary tumors.
The Foxp3 fGFP allele on a B6 background has been reported to affect Treg phenotype and function but not health at steady state (12). We generated BALB/c (Ba) Foxp3-GFP mice which, compared to gender and age matched WT animals, have slightly underrepresented Treg that express increased Foxp3 protein (as reported for mice on the B6 background (11)), bear normal numbers of activated or interferon-g (IFN-g) producing T cells (Fig. S1) and do not show sign of disease.
To test whether the Foxp3 fGFP allele provides for enhanced anti-tumor immunity, we chose to monitor the CT26 colorectal carcinoma cell line, derived from a BALB/c mouse. While CT26 engraftment and growth is successful in WT animals, its intrinsic immunogenicity is readily revealed upon depletion of Treg by administration of diphtheria toxin (DT) in DEREG mice, either one (15) or two weeks (Fig. 1A) 1H). Strikingly, while all WT animals allowed engraftment and regular tumor size progression, CT26 growth was either delayed or fully prevented in more than a third of the Foxp3-GFP mice. To ascertain that immune responses were enhanced in Foxp3-GFP mice, tumor infiltrating lymphocytes (TIL) were analyzed 15 days postimplantation ( Fig. 1C-E and S2). The tumor weight was reduced (Fig. 1C), the frequency of CD8 IFN-g producing cells was increased (Fig. 1D) and the ratio Treg to CD8 was inverted (Fig.1E) in Foxp3-GFP compared to WT mice. In the tumor draining lymph node (DLN), where Treg were not overrepresented, these differences were not found in the draining lymph nodes (Fig. S2). Overall, these data demonstrate that the Foxp3 fGFP allele, a fusion of eGFP at the N-terminal part of Foxp3, potentiates spontaneous anti-tumor immunity without affecting health in the BALB/c reference strain. These results resonate with recent findings from the Mathis-Benoist' lab (17).
By introducing distributed alanine replacement mutations in the mouse Foxp3 gene, they identified two mutant alleles that conferred B6 mice with reduced growth of the MC38 primary tumor. One of these alleles does not affect health at steady state and is located in the N-terminal proline-rich region of Foxp3, where no specific motifs have been identified.
We next tested whether the Foxp3 fGFP allele confers a disadvantage to Treg in the CT26 context. Owing to the fact that Foxp3 is located on the X chromosome, heterozygous Foxp3 fGFP/WT females are mosaic and mimic a cell competition assay.
Analysis of TIL 15 days post-implantation of CT26 into Foxp3 fGFP/WT females revealed that a large majority of Treg expressed the WT allele (Fig. 1F), indicating that Treg expressing the Foxp3 fGFP allele are less fit. A similar trend was found when analyzing the dLN (Fig. S2). Although Bettini et al did not address the impact of the Foxp3 fGFP allele on tumor growth, they reported a similar disadvantage for Treg expressing the Foxp3 fGFP allele when analyzing B6.Foxp3 fGFP/WT females implanted with the B16 melanoma (12). Cancer immunosuppression is a complex process, for which both thymic derived (tTreg) (18) and peripherally differentiated Treg (pTreg) (19) have been incriminated, and both cell subsets were shown to be affected by the Foxp3 fGFP allele (12,13).

The Foxp3 fGFP allele enhances the effectiveness of cancer immunotherapies.
We next tested whether the Foxp3 fGFP allele potentiates therapeutic response to anti-CTLA4 mAb (aCTLA4) treatment. We administered aCTLA4 i.p. during the first week following tumor implantation in WT and Foxp3-GFP mice, and followed subsequent tumor progression (Fig. 1G-K). As previously reported (20), WT mice implanted with CT26 partially responded to aCTLA4 monotherapy, with only a fraction of them rejecting or delaying tumor growth. In contrast, all treated Foxp3-GFP mice prevented tumor growth, with most animals maintaining a tumor free state 1-month postimplantation (Fig. 1G). These dramatic results encouraged us to test B6.Foxp3-GFP mice implanted with the poorly immunogenic B16 melanoma for which aCTLA4 monotherapy has been reported ineffective (20). Strikingly, although the Foxp3 fGFP allele by itself did not affect B16 engraftment or progression (Fig. 1I), aCTLA4 treatment delayed tumor growth in the majority of Foxp3-GFP mice but not in WT controls (Fig. 1J, K). Finally, enhanced therapeutic response afforded by the Foxp3 fGFP allele was not accompanied by overt systemic effects as indicated by constant body weight (Fig. S2).
Our finding that the Foxp3 fGFP allele potentiates aCTLA4 therapy, together with the evidence that aCTLA4 kills Treg (21,22), echoes with a previous report indicating that the same allele amplifies the effectiveness of DT treatment in DEREG mice in an infection setting (14). Moreover, the improved efficacy of aCTLA4 treatment that we evidence in Foxp3-GFP mice, both in BALB/c and in B6 backgrounds, suggests that allelic variance at the Foxp3 or other Treg signature genes may discriminate responder from non-responder patients submitted to this therapy.

The Foxp3 fGFP allele delays metastasis dissemination
The last phase of tumor progression is metastatic dissemination, a process that is well modelled by the 4T1 breast carcinoma derived from a BALB/c mouse. The 4T1 tumor grows slower than CT26 or B16 at the site of implantation, resists many immune interventions and is highly metastatic with an average mean survival of the host of 50 days post-implantation (23). A moderate role for Treg in facilitating 4T1 primary tumor growth (15) and precipitating death (24) has been reported. Monitoring WT and Foxp3-GFP animals implanted s.c. with 4T1 cells revealed similar growth of the primary tumor while survival was significantly prolonged in the latter group ( Fig. 2A, B and S3).
Prolonged survival was also observed in DEREG animals administrated with DT, as mean to induce a transient depletion of Treg, early after 4T1 implantation (Fig 2C, D and S3). We next ascertained that the prolonged survival in Foxp3-GFP mice bearing 4T1 tumors associated with reduced metastasis dissemination. We first confirmed that resection of the primary tumor during the second week post-implantation, but not later, greatly enhances survival (Fig. S3), an intervention shown by others to prevent the metastatic process (23). This result suggested that analysis at 3 weeks post implantation would be suitable to quantify metastasis dissemination in WT and Foxp3-GFP mice. Lungs were harvested and a histological assessment of metastasis number and size was performed (Fig. 2E). Although metastatic foci were similar in number in both groups of mice, large nodules were only found in WT animals. Together, these findings indicate that the Foxp3 fGFP allele restrains the dissemination stage of 4T1 tumors and suggest that identification of allelic variants at the Foxp3 or downstream genes may serve to guide prognostic estimates for cancer patients.

Concluding remarks
Of relevance for experimental biology at large, our work provides further evidence that experimental variations can be related to reporter alleles, now in the context of cancer immune-surveillance. In the frame of immune tolerance, the evidence that Treg are essential components of immune regulation produced the notion that affecting Treg function for therapeutic reasons, such as for unleashing tumor immunity, will also unleash pathologic autoimmune reactivities. The adverse events tightly associated with tumor immunotherapies, and most notably with aCTLA4 therapy, are indeed autoimmune and inflammatory manifestations, often jeopardizing treatment continuation. In apparent contradiction with this notion, the Foxp3 fGFP allele on a B6 or BALB/c genetic background, does not compromise health at steady state and yet favors anti-tumor immunity. However, when introduced on a genetic background carrying several susceptibility alleles that together promote autoimmune Type 1 Diabetes, the Foxp3 fGFP allele precipitates disease (11,12). While steady state mice are unlikely to be faithful models of humans continuously exposed to inflammatory triggers, it is conceivable that in the range of variations allowing for return to homeostasis, weak Foxp3 hypomorphs on an otherwise healthy genetic background would be beneficial to fight chronic infections and cancer, without tilting the balance toward autoimmunity. It therefore remains possible that weak but beneficial Foxp3 alleles are present in human population, more easily identifiable in males, and that genetic analysis would provide biomarkers to guide cancer prognostics and therapeutic strategies.   Table 1. Samples were processed on a Cyan ADP instrument and analyzed with the FlowJo software. Statistics of cellular analysis and iTCI were performed using nonparametric Mann-Whitney test. Tumor growth analysis were also performed using two-way ANOVA.

Mice
Logrank tests were used for survival curves and two-way ANOVA for body weight kinetics. Correlation analyses were performed using Pearson correlation coefficients.