Depletion of CD25+ regulatory cells results in suppression of melanoma growth and induction of autoreactivity in mice

Melanoma is a tumour that arises from the immortalisation of pigment cells called melanocytes and is associated with high mortality in human patients. Although the impact of the immune system on the growth and spread of naturally occurring melanoma is not well defined, T cell responses to antigens expressed by the tumour cells have been identified (1). These antigens can be divided into two broad subsets. The first subset describes cancer/testis antigens whose expression is almost exclusively confined to a variety of different tumour cells. Such antigens, for example those expressed by the MAGE gene family, are normally silent in somatic cells. The second subset of antigens consists of the melanocyte differentiation antigens gp100/pmel17, tyrosinase (Tyr), tyrosinase-related proteins-1 and -2 (Trp-1 and Trp-2) and Melan-A. These antigens are self-antigens that are expressed by both normal and malignant melanocytes. These studies clearly indicate that T cells that recognise either cancer/testis antigens or melanocyte antigens do escape thymic deletion and reach the periphery. The precise circumstances under which T cell responses to these antigens can be induced and sustained in the periphery are not well understood.

Accumulating evidence suggests that a population of CD4+ T cells that constitutively express the IL-2 receptor alpha-chain (IL-2R-alpha, CD25) may function as regulatory cells capable of down-regulating immune responses to self-antigens (2). These cells, which represent 5-10% of peripheral CD4+ T cells in naive mice, do not proliferate upon stimulation via their T cell receptor (TCR) but have been shown in vitro to suppress the activation of other T cells in an antigen-independent manner (3). Adoptive transfer of CD4+CD25+ T cells has been shown to prevent autoimmune diseases that develop in adult mice following neonatal thymectomy (4). In addition, adoptive transfer of splenocytes depleted of CD4+CD25+ T cells has been shown to lead to the development of colitis in SCID mice (5). Similarly, transfer of T cells from pre-diabetic NOD mice depleted of CD4+CD25+ T cells was shown to accelerate progression of insulin-dependent diabetes in NOD.SCID mice (6). It is therefore possible that the development of chronic T cell autoreactivity is, at least in part, a result of the inactivation of CD4+CD25+ regulatory T cells.

A recent study by Shimizu et al. showed that treating mice with antibodies specific for CD25 (PC61) prior to the injection of live melanoma cells resulted in a significant delay of tumour growth (7). The authors argued that, since antibody treatment reduced the number of CD4+CD25+ cells in the periphery approximately four-fold when compared to untreated mice, control of tumour growth was the result of loss of regulatory T cell activity and a concomitant rise in tumour immunity. The experiments did not, however, address the possibility that the injected CD25-specific mAbs influenced the behaviour not of CD25+ regulatory cells but of effector T cells that may express CD25 following activation. Evidence exists to suggest that autoreactive T cells are susceptible to IL-2-mediated activation induced cell death (AICD) in vivo (8). Since the CD25-specific mAbs used have previously been shown to inhibit binding of IL-2 to the IL-2R (9, 10) it is conceivable that the antibodies may also inhibit IL-2-mediated AICD of autoreactive T cells and thereby promote tumour immunity in vivo.

In this study we have further characterised the effect of treatment with anti-CD25 antibodies on growth of the melanoma cell line B16F10 (11), in C57BL/6 (B6) mice. We have specifically addressed whether or not the inhibitory impact of antibody treatment on primary tumour growth is 1) T cell dependent, 2) mediated through binding of antibodies to CD4+ cells that constitutively express CD25 rather than effector T cells that upregulate CD25 upon activation, 3) accompanied by the generation of protective immunity and 4) accompanied by the development of immune responses to melanocyte differentiation antigens.

Mice were injected with 1 mg of anti-CD25 mAb (PC61) or anti-beta-galactosidase mAb (GL113, isotype control) three and one days prior to injection with B16F10. Subsequent monitoring of tumour growth demonstrated that tumours grew more slowly, if at all, in mice injected with anti-CD25 mAb when compared to control mice that had received no antibody or the isotype control antibody (Figure 1a). In several similar independent experiments the proportion of anti-CD25 mAb-treated mice remaining tumour-free ranged from 30% to 60% whereas all control mice developed tumours. A total of 69 out of 111 anti-CD25 mAb-treated mice rejected B16F10 using the protocol described above. Using a logistic regression model the effect of treatment with anti-CD25 mAbs on tumour growth was strongly significant (P12), depigmentation was not observed in the majority of mice that had rejected tumours following treatment with anti-CD25 mAbs.

Injection of mice with anti-CD25 mAbs using the protocol described in this study results in depletion of CD25+ cells for approximately 21 days after antibody administration (Figure 2a). Anti-CD25 mAbs are detectable in the serum of mice injected as described above for approximately 15-20 days after the first injection of antibody (Figure 2b). It is therefore possible that CD25-specific mAbs influence the behaviour not of CD25+ regulatory cells but of effector T cells that may express CD25 following activation. Although IL-2 is a growth factor for T cells following antigen stimulation (13), mice that are deficient for IL-2 accumulate autoreactive T cells and develop autoimmune disease. It has therefore been suggested that IL-2-mediated AICD contributes to the maintenance of peripheral T cell tolerance in normal mice (8). Since the anti-CD25 mAb PC61 has been shown to block IL-2-binding to CD25 (10), PC61 may promote tumour immunity in vivo by preventing IL-2-driven AICD of self-antigen-specific T cells. We determined whether or not antibody binding to effector T cells was indeed responsible for promoting tumour immunity in vivo. Twenty-one days after the first injection with anti-CD25 mAbs we inoculated a group of mice whose serum did not contain detectable levels of anti-CD25 mAbs with melanoma cells. These mice were selected as those whose serum did not stain splenic CD4+CD25+ T cells (Figure 2b). Using titrated amounts of purified anti-CD25 antibody we found that the lower limit of antibody concentration for staining splenic CD4+CD25+ cells for detection by flow cytometry corresponded to 3.2 µg per ml of serum. This amount of antibody is not biologically active since we had already found in previous experiments that administration of 0.1 mg rather than 1 mg of anti-CD25 antibodies did not result in tumour rejection even though serum concentrations of antibody were, in these mice, greater than 3.2 µg/ml (data not shown). The absence of anti-CD25 mAbs in the serum was also confirmed by ELISA (data not shown). Half of the anti-CD25 mAb-treated mice were able to reject melanoma whereas tumours grew in all mice that had been treated with the isotype control antibody (Figure 2c). This result confirms that anti-CD25 mAbs do not promote tumour immunity by binding to CD25+ effector T cells.

Anti-CD25 antibody-treated mice that remained tumour-free after challenge with B16F10 were re-challenged with a second inoculum of tumour cells eight weeks later. A significant proportion of these mice were able to resist a second tumour challenge. Control mice that had been treated with anti-CD25 mAbs approximately eight weeks previously but which had not been previously inoculated with tumour cells were not able to reject tumours (Table 1). This finding indicates that protective immunity is dependent upon immunisation with the primary tumour inoculum and is present at a time-point when CD25+ regulatory cells have returned to the periphery. Tumour-bearing mice that were not treated with anti-CD25 mAbs at the time of primary tumour inoculation do not reject second tumours injected 8 weeks later (data not shown).

In order to determine which immune subset was responsible for mediating long-term immunity to B16F10, we purified CD8+ and CD4+ T cells from the spleens of tumour-free mice injected with anti-CD25 mAbs and B16F10 cells three weeks previously. Purified T cell populations, the eluates consisting of CD4-depleted or CD8-depleted spleen cells and serum recovered from the same mice were injected at the indicated numbers into naive, syngeneic recipient mice. Recipient mice were injected with suspensions of B16F10 and tumour growth was monitored. As shown in Figure 3, tumours grew in mice receiving CD8+ T cells (3C), spleen cells depleted of CD4+ T cells (3E) or serum (3D). Tumour cells were however rejected in mice receiving CD4+ T cells (3B) or CD8-depleted spleen cells (3F). Overall the results obtained in this study indicate that whilst both CD4+ and CD8+ T cells are required for rejection of primary tumours following depletion of CD25+ regulatory cells, long-term protective immunity is mediated exclusively by CD4+ T cells. Similar results were obtained when purified T cell subsets were adoptively transferred into C57BL/6.RAG2-deficient (B6.Rag2-/-) mice (data not shown).

We wished to determine whether or not T cell responses to melanocyte differentiation antigens were induced in mice that had been treated with anti-CD25 mAbs prior to the inoculation of tumour cells. Since very little information exists regarding the nature of immune responses to self-antigens following the depletion of regulatory T cells we wished to address the question outlined above without bias towards a particular type of effector cell. Measurement of immunity to recombinant vaccinia viruses (rVV) was our chosen approach since immunity to vaccinia viruses can be achieved by CD4+ T cells, CD8+ T cells or antibodies. We therefore measured protective immunity to rVVs expressing the melanocyte differentiation antigens gp100 (rVVgp100), Mart-1 (rVVMart-1), Trp-1 (rVVTrp-1), Trp-2 (rVVTrp-2) and Tyr (rVVTyr) in mice that had rejected tumours after anti-CD25 mAb treatment. Fourteen days after the inoculation of tumour cells the mice were challenged with each rVV and a rVV expressing the irrelevant influenza nucleoprotein (rVVUb-R-NP) antigen and virus titres were measured in the ovaries 5 days later. A subsequent determination of virus titres indicated that 5/6 anti-CD25 mAb-treated tumour-free mice, but not naive mice (0/3) or control mAb-treated, tumour-bearing mice (0/3), were partially protected against infection with rVVTyr (>90% reduction in virus titre) but not rVV's expressing gp100, Trp-1, Trp-2, Melan-A or Ub-R-NP (Figure 4 and data not shown). All 9 mice that had rejected melanoma approximately eight months previously and which exhibited signs of depigmentation were also protected against infection with rVVTyr. These results indicate that inoculation of mice depleted of CD25+ regulatory cells with melanoma cells can result in the induction of tyrosinase-specific autoreactivity.

The observations described in this study demonstrate that CD25+ regulatory cells suppress immune responses that are capable of mediating rejection of melanoma in mice. These results extend earlier findings by showing that co-administration of anti-CD25 mAb with either depleting anti-CD4 mAb or anti-CD8 mAb abrogates the anti-tumour effect of anti-CD25 mAb administration alone, thereby indicating that both T cell subsets are important for the rejection of tumours inoculated into anti-CD25 mAb-treated naive B6 mice. Long term tumour immunity however was conferred by CD4+ T cells alone since these cells isolated from anti-CD25 mAb-treated tumour immune mice were able to reject tumour cells injected into B6.RAG2-/- mice in the absence of CD8+ T cells. Overall, these data suggest that the CD8+ T cells required for rejection of primary tumours in anti-CD25 mAb-treated naive B6 mice may be antigen-nonspecific and therefore unable to provide long-term tumour immunity. The mechanism through which CD4+ T cells mediate immunity to melanoma following depletion of CD25+ regulatory cells is not yet known although several studies have shown that CD4+ T cells can eliminate tumour cells in vivo through production of cytokines such as IFN-gamma (14, 15) and more recently IL-10 (16).

It is noteworthy that in the adoptive transfer experiments described in this study B6 recipients of CD4+ T cells recovered from tumour-immune mice contained normal numbers of CD25+ regulatory T cells. These data indicate that although CD4+ T cell-mediated tumour immunity is generated only in the absence of CD25+ regulatory cells, effector cells can promote tumour immunity in vivo even in the presence of CD25+ regulatory cells. There are at least three potential explanations for this finding. Firstly, whilst CD25+ regulatory cells inhibit priming of tumour-specific naive T cells, they may not inhibit tumour-specific effector T cells in vivo. Secondly, the frequency of tumour-specific effector T cells may be too high for control by CD25+ regulatory cells. Thirdly CD25+ regulatory cells may require activation by tumour cells in order to mediate their immunosuppressive effects. Since CD25+ T cells have been shown to require activation through their T cell receptors before they mediate suppression (3) it is possible that the CD25+ regulatory cells present in naive mice at the time of adoptive transfer are yet to be activated by exposure to tumour cells. This would provide a window of opportunity whereby effector T cells could mediate their anti-tumour effect.

Since the induction of tumour immunity and autoimmunity are both subject to regulation by CD25+ cells, it is possible that in the absence of CD25+ regulatory cells immunity to melanoma is mediated by immune responses to melanocyte differentiation antigens. Immune responses to melanocyte differentiation antigens can be accompanied by depigmentation and although in this study a proportion of treated mice did show signs of depigmentation, the overall number was low (et al. demonstrated a CD8+ T cell response to the melanocyte antigen Trp-2 in mice treated with antibodies specific for CD25 and CTLA4 and inoculated with B16 engineered to express GM-CSF (B16-GM-CSF) (17). Although treated mice were tumour immune a definitive link between tumour immunity and Trp-2-specific CD8+ T cells was not established. Following depletion of CD25+ regulatory cells and inoculation of mice with live B16F10 cells, we did not find evidence of Trp-2-specific immunity either in cytotoxicity assays or after staining spleen cells with MHC class I tetramers (data not shown). It appears likely therefore that the experimental protocols described above induce a different spectrum of immune responses to melanocyte antigens. This possibility is supported by the observation that whilst unmodified B16F10 induces, in the absence of CD25+ cells, protective tumour immunity that is mediated exclusively by CD4+ cells, immunity generated by B16-GM-CSF after treatment of mice with CD25- and CTLA4-specific antibodies appears to be dependent upon both CD4+ and CD8+ T cells (17).

Overall the data described above indicate that immune responses to melanocyte antigens are generated in mice depleted of CD25+ regulatory cells and inoculated with melanoma cells. The data do not however prove that T cells specific for melanocyte antigens mediate tumour immunity. The inhibitory impact of anti-CD25 mAb treatment on tumour growth may result from uncovering immune responses to self-antigens although these antigens may not necessarily be melanocyte differentiation antigens (Figure 5A). It is also possible however that tumour rejection is mediated by immune responses to neo-antigens or cancer/testis antigens that could normally be inhibited by CD25+ regulatory cells. In this case self-antigen-specific immune responses would arise indirectly as a consequence of tumour destruction (Figure 5B). Both possibilities are currently under investigation.

B6 mice were bred under specific pathogen-free conditions at Biomedical Services, Oxford. Female mice aged 5-8 wks were used in all experiments. During experimental procedures mice were housed in conventional facilities.

Cultures of the melanoma cell line, B16F10 (H-2^b) were maintained in RPMI (Sigma) supplemented with 10% FCS (Gibco), L-glutamine, penicillin-streptomycin (Sigma), non-essential amino acids (Gibco), and sodium glutamate (Gibco). Sublines of B16F10 were obtained from Professors R. Zinkernagel (Zurich, Switzerland), D. Pardoll (Johns Hopkins Medical School, Baltimore, USA) and I. Hart (London, UK). Unless otherwise indicated the subline obtained from Zurich was used in all experiments. In the experiment shown in Table 1, cells were incubated in the serum-free medium (OPTIMEM, Gibco) for 48 hours prior to inoculation. In all experiments mice were injected s.c. with 2 x 10⁴ tumour cells in PBS. Tumour volume was calculated using the formula V=pi x abc/6 where a, b and c are the orthogonal diameters.

Hybridomas secreting CD25- (PC61, rat IgG1), CD4- (YTS 191.1.2, YTA 3.1.2, both rat IgG2b), CD8- (YTS 169.4.2.1, YTS 156.7.7, both rat IgG2b), and rat anti-E. coli beta-galactosidase- (GL113, rat IgG1) specific mAbs have been described previously (18, 19). Hybridomas were grown in culture and mAbs were purified by precipitation in saturated ammonium sulphate. Mice received 1 mg of antibodies GL113 and PC61 twice, with a one-day interval. For depletion of CD4+ or CD8+ T cells, mice were injected twice with either 100 µg of both anti-CD4 antibodies or 100 µg of both anti-CD8 antibodies, with a one-day interval. Antibodies, injected i.p. in volumes of 100-200 µl, were administered 3 and 1 day prior to the injection of tumour cells.

In order to identify anti-CD25 in the serum of mice injected with anti-CD25 antibodies, spleen cells from untreated mice were incubated with 5 µl of undiluted serum and PE-conjugated anti-CD4 antibodies (rat IgG2a, Pharmingen) for 45 minutes on ice. After washing the cells twice in wash buffer (PBS, 2% FCS, 2mM EDTA), the cells were incubated with FITC-conjugated anti-rat IgG1 antibodies for 30 minutes at 4°C. The cells were washed twice and resuspended in Facs Fix (wash buffer containing 2% formalin). CD25-positive T cells were subsequently identified by flow cytometry using Cellquest software.

Single cell suspensions were prepared from the spleens of donor tumour-immune mice. Donor mice were injected i.p. 24 and 22 days previously with 1 mg of anti-CD25 mAbs and s.c. with 2 x 10⁴ B16F10 cells 21 days previously. The cells were labelled with MACS antibodies for CD4+ and CD8+ T cells and purified using MACS VS⁺ separation columns according to the manufacturer's instructions (Miltenyi). Flow cytometry confirmed greater than 90% purity of the cells. Recipient mice were injected i.v. with 10⁷ purified T cells or with 5 x 10⁷ CD4-depleted or CD8-depleted spleen cells. Mice were injected s.c. on the same day with 2 x 10⁴ B16F10 cells.

Mice were infected with 2 x 10⁵ PFU of rVVs. Five days later ovaries were removed and analysed for rVV titres as described previously (20, 21). rVVs expressing gp100, Tyr, Melan-A, Trp-1, Trp-2 have been described previously (22, 23).

The authors wish to acknowledge Dr. William Reece for help with the statistical analyses, Professor Nick Restifo for the recombinant vaccinia viruses and Dr. Andy Sewell for being very helpful in general. This work was funded by The Wellcome Trust (grant no. GR056527MA).