Regulatory T cells (Tregs) are thought to facilitate tumor development by suppressing protective antitumor immune responses. However, recent clinical and laboratory studies show that Tregs are a favorable element against cancer. In this study, we provide evidence that Tregs have both promoting and inhibiting effects on tumors, depending on the stage of tumor development. By using 0.5 mg cyclophosphamide, we constructed a murine liver cancer model in which Tregs were continuously and selectively depleted. Under such conditions, we found that tumor growth was inhibited at early stages but accelerated later on. Analysis of the tumor microenvironment disclosed that long-term Treg depletion by 0.5 mg cyclophosphamide treatment induced Gr-1+CD11b+ myeloid-derived suppressor cells (MDSCs). Ablation of MDSCs by anti–Gr-1 Ab blocked Treg depletion-induced promotion of tumor growth. Furthermore, lipoxygenases 5 and 12, two enzymes participating in the biosynthesis of the lipid anti-inflammatory mediator lipoxin A4, were upregulated or downregulated by Treg depletion or adoptive transfer. Correspondingly, the levels of lipoxin A4 were increased or decreased. Lipoxin A4 thus regulated the induction of MDSCs in response to Treg depletion. These findings suggest that Tregs may play different roles at different stages of tumor growth: promoting early and inhibiting late tumor growth. Our study also suggests that the interplay among Tregs, MDSCs, and lipoxin A4 tunes the regulation of tumor-associated inflammation.
Despite major advances in understanding of the mechanisms leading to tumor immunity, the successful translation of these insights into effective tumor immunotherapy has not been achieved because of numerous immunosuppressive mechanisms (1, 2). Among them, myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs) play fundamental roles in blocking antitumor responses (3–9). It has been established that MDSCs and Tregs accumulate in the tumor-bearing host. Nevertheless, the manner in which the accumulation of these cells participates in tumor homeostasis remains elusive. Notably, MDSCs have been reported to induce tumor-specific Treg development, which further enhances tumor immunosuppression (10). To date, whether and how Tregs in turn regulate MDSCs is unclear.
The physiological significance of Tregs lies in their suppression of adverse immune responses, promoting peripheral self-tolerance and avoidance of autoimmunity (11, 12). Under pathological conditions, Tregs seem to prevent the formation of catastrophic inflammation, such as that observed in the tumor microenvironment (13–15). Inflammation is regarded as a pivotal element in tumor initiation, promotion, and progression; and inhibiting inflammation may negatively regulate tumor development. Therefore, if it is presumed that the recruitment of Tregs to tumors inhibits the destructive inflammation, Tregs probably have an intrinsic antitumor role by counteracting tumor inflammation. In support of this idea, studies from Erdman et al. (16) have shown that the transfer of Tregs prevents cancer development associated with inflammatory colitis. Thus, the roles of Tregs in tumors are probably complex and not just tumor promoting.
Besides immune regulatory cells, recent studies highlight an endogenous eicosanoid, called lipoxin A4, participating in the regulation of inflammation (17, 18). Lipoxin A4 is synthesized locally from arachidonic acid at inflammation sites in a transcellular biosynthesis manner (19). By binding its high-affinity G protein-coupled receptor, lipoxin A4 strongly inhibits the trafficking of leukocytes to the inflammatory site and stimulates the phagocytosis of apoptotic cells by macrophages (17, 18). Currently, lipoxin A4 and its analogues are considered as “braking signals” of inflammation. However, the regulation of tumor inflammation by lipoxin A4 remains unclear. In the current study, we show that the depletion of Tregs in large tumors resulted in the production of lipoxin A4 and the latter consequently induced the generation of MDSCs, thus promoting tumor growth. These findings suggest that Tregs may play different roles at different stages of tumor growth, promoting early and inhibiting late tumor growth, and also suggest that the body is capable of developing different strategies against tumor inflammation.
Materials and Methods
Tumor model and treatment protocol
BALB/c mice and BALB/c nude mice, 6 to 8 wk old, were purchased from Center of Medical Experimental Animals of Hubei Province (Wuhan, China) for studies approved by the Animal Care and Use Committee of Tongji Medical College. To establish a tumor model, a BALB/c background H22 hepatocarcinoma tumor cell line (2 × 105) was used to s.c. inoculate mice. To deplete Tregs, low doses of cyclophosphamide (CY) were i.p. injected into the mice next day. The injection was continued daily.
Isolation of MDSCs from bone marrow and tumor tissue
Tumor-bearing mice were sacrificed, and their tibias and femurs were harvested. After lysis of RBCs, bone marrow cells were fractionated by centrifugation on a Percoll (Amersham Biosciences, Uppsala, Sweden) density gradient as described (15). Cells were collected from the gradient interfaces. Cell bands between 40 and 50% were labeled as fraction 1, between 50 and 60% as fraction 2, and between 60 and 70% as fraction 3.
As for tumor MDSCs, large tumors were digested with collagenase and hyaluronidase for 1 h at 37°C and homogenized with semifrosted slides. After lysis of RBCs, the dissociated cells were underlaid with 5 ml lymphocyte separation solution and centrifuged (2200 rpm for 20 min). Tumor-infiltrating lymphocytes including MDSCs were harvested from the interface for flow cytometric analysis.
Flow cytometric analysis
In vivo depletion of Tregs and MDSCs
RT-PCR and real-time RT-PCR
Western blot analysis
Tumor tissue homogenates (30 μg total protein) and prestained m.w. markers were separated by SDS-PAGE followed by transfer onto nitrocellulose membranes. The membranes were blocked in TBS with 0.5% of Triton X-100 containing 5% nonfat milk and were probed with anti-inducible NO synthase (iNOS) Ab (Cell Signaling, Danvers, MA) or anti-arginase 1 Ab (Abcam, Cambridge, MA). After incubation with the HRP-conjugated secondary Ab, membranes were extensively washed, and the immunoreactivity was visualized by ECL according to the manufacturer’s protocol (ECL kit; Santa Cruz Biotechnology, Santa Cruz, CA). Other Abs were purchased from Santa Cruz Biotechnology.
2 and lipoxin A4 detection, the mouse serum was assessed by ELISA kits (Lengton Bioscience, Shanghai, China).
Adoptive transfer experiments
Splenocytes were used for the isolation of Tregs with the Regulatory T Cell Isolation Kit (Miltenyi Biotec, Auburn, CA) according to the manufacturer’s instructions for adoptive transfer through tail vein injection (3 × 106 per mouse).
The suppressive activity of MDSCs was assessed in an anti-CD3 and anti-CD28–mediated proliferation assay of T cells. Briefly, splenocytes (1 × 105) from BALB/c mice were cultured in the presence of serial dilutions of irradiated MDSCs in 96-well microplates. [3H]Thymidine was added during the last 8 h of the 72-h culture.
Results were expressed as mean values ± SD and interpreted by repeated-measure ANOVA. Differences were considered to be statistically significant when the p value was <0.05.
Tregs are continuously depleted by low-dose CY injection in tumor-bearing mice
Low-dose CY treatment has shown efficacy in selective and transient depletion of Tregs in humans and mice, leading to therapeutic effect against tumors (20–22). However, whether long-term Treg depletion by low-dose CY treatment has a better antitumor effect remains unclear. In this study, we first addressed this question by establishing an animal model with long-term Treg depletion by low-dose CY treatment. One milligram CY has been used as low-dose CY in mice (22, 23). Based on this, we injected 0.25, 0.5, or 1 mg CY daily into H22 hepatocarcinoma cell line-inoculated mice for 14 or 25 d and compared the effect of different doses of CY on Treg depletion. As expected, all three dosages of CY effectively depleted Tregs in spleen and tumor tissue (Fig. 1A, 1B). The efficiency between 0.5 and 1 mg CY was similar and higher than that of 0.25 mg CY (Fig. 1A, 1B). However, we found that 1 mg CY but not 0.25 or 0.5 mg CY daily injection caused weight loss and sickness of mice (data not shown). Further analysis of liver and kidney function indicated that neither 0.25 nor 0.5 mg CY affected these two organs (Fig. 1C). In addition, we also analyzed the spleens and lymph nodes after 25-d 0.25 or 0.5 mg CY treatment and found that the relative total T cell number as well as CD4+ T cell number was not significantly affected by the injection of 0.25 or 0.5 mg CY (Fig. 1D). Taken together, these data suggested that Tregs can be selectively depleted by 0.25 or 0.5 mg CY daily injection and such injection has no effect on effector T cells in vivo.
Treg depletion inhibits early but promotes late tumor growth
Next, we asked 1) whether long-term Treg depletion suppressed tumor growth and 2) whether 0.5 mg CY dose resulted in a better outcome than 0.25 mg CY, considering that 0.5 mg CY had a stronger Treg depletion efficiency (Fig. 1A). Mice were s.c. inoculated with H22 tumor cells and then daily i.p. injected with 0.25 or 0.5 mg CY from day 1 to day 25. We found that both doses of CY treatment inhibited tumor growth (Fig. 2A). However, although having a better Treg depletion efficiency, 0.5 mg CY treatment did not result in a better suppression of tumor growth. Surprisingly, if we compared the two groups of 0.5 and 0.25 mg CY, we found that during the first 15 d, 0.5 mg CY showed a more inhibitory effect; thereafter, 0.25 mg CY showed a better antitumor effect (Fig. 2A). This phenomenon could not be explained by the direct pharmacological activity of CY. CY is a prodrug subject to a series of biological activation steps necessary for its cytotoxic effect. Thus, we took the blood plasma from CY-treated tumor-bearing mice at different time points to culture H22 cells for 48 h. We found that plasma from 0.5 mg CY-treated mice had no effect on H22 cell growth (Fig. 2B) or cell death by trypan blue staining (data not shown). Differently, plasma from high-dose CY (2 mg)-treated mice inhibited H22 cell proliferation (Fig. 2B) and induced tumor cell death as examined by trypan blue staining (data not shown).
To ask why a more efficient Treg depletion at 0.5 mg CY treatment did not result in a better tumor suppression as expected, we daily injected mice with 0.25 or 0.5 mg CY for a shorter period of 16 d. We chose this time point because compared with 0.25 mg CY, 0.5 mg CY treatment accelerated tumor growth after 16 d (Fig. 2A). Surprisingly, we found that further CY injection after 16 d promoted instead of inhibited tumor growth, which was more evident in the 0.5 mg CY group (Fig. 2C). Because 0.5 mg CY treatment had no effect on effector T cells and no pharmacological activity on tumor cells (Fig. 1D and Fig. 2B), this tumor-promoting effect might have been due to Treg depletion by CY treatment. Namely, Tregs might have an inhibitory effect on tumor growth. We then used nude mice to validate this idea, as Treg cells are deficient in those mice. We inoculated H22 tumor cells to BALB/c nude mice and treated the mice with 0.5 mg CY from day 1 to day 25 and found that such 0.5 mg CY treatment had no effect on tumor growth in nude mice (Fig. 2D). However, we also used anti-CD25 Ab to deplete Tregs in vivo, as the IL-2R α-chain molecule CD25 is expressed on Tregs and has been used as a target for Treg depletion (24). Consistently, by using anti-CD25 Ab to further deplete Tregs after 16 consecutive days of 0.5 mg CY injection, we observed the accelerated tumor growth (Fig. 2E). Again, these data together indicate that Tregs may have an inhibitory effect on tumor growth, and as a result, Treg depletion promotes tumor development in late tumor.
Treg depletion induces MDSCs in large tumor
The Treg depletion-mediated tumor promotion led us to screen in the tumor microenvironment for those molecules and cells that could be affected by Treg depletion and favor tumor growth. We first tested a panel of inflammation-associated factors, including IFN-γ, IL-6, IL-10, CCL2, TGF-β, iNOS, and arginase 1. Our rationales were 1) Tregs participate in the regulation of inflammation; 2) inflammation is a hallmark of tumorigenesis, which promotes tumor development (13–15); and 3) these factors are involved in tumor inflammation. By using real-time RT-PCR, we found that compared with 16-d depletion, 25-d Treg depletion downregulated the expression of IFN-γ but upregulated IL-6, IL-10, TGF-β, CCL2, iNOS, and arginase 1 (Fig. 3A). The protein levels of these factors, evaluated by ELISA or Western blot (Fig. 3B, 3C), were concordant with the RT-PCR results.
The expression profile of the above inflammation-associated factors fits the promotion and progression phenotype of tumor (25). However, our attention was drawn to the increased expression of CCL2, iNOS, and arginase 1. It is well known that iNOS and arginase 1 can be expressed by MDSCs, which are accumulated in bone marrow and chemoattracted to tumors through CCL2/CCR2 pathway (3, 26). Moreover, the induction of Tregs by MDSCs (10) also implies a possibility of Tregs acting on MDSCs. Therefore, we analyzed Gr-1+CD11b+ MDSCs by flow cytometry and found that 25-d Treg depletion by 0.5 mg CY injection increased the proportion of MDSCs in both bone marrow and tumor tissue (Fig. 4A). To clarify whether such increased MDSCs were functional, we conducted an in vitro suppression assay. We found that T cell proliferation was strongly inhibited by the induced MDSCs in response to the stimulation of anti-CD3 and anti-CD28 Abs (Fig. 4B), suggesting that those MDSCs are functional. In line with this, we found tumor-infiltrating T cells from the 25-d Treg depletion group showed a hyporesponsive phenotype by much lower expression of IFN-γ and IL-2 and higher expression of IL-4, IL-10, and TGF-β compared with those from the 16-d Treg depletion group (Fig. 4C). In addition, by comparing the proportion of MDSCs at different time points during Treg depletion, we found that 16-d Treg depletion did not significantly increase MDSCs (Fig. 4D, left). Thus, the induction of MDSCs by long-term Treg depletion seemed to occur at the late stage rather than the early stage of tumor development.
Then, we asked if the induction of MDSCs was due to the absence of Tregs rather than other effects of low-dose CY. For this purpose, we replaced the injection of 0.5 mg CY after day 16 with the injection of anti-CD25 depleting Ab (PC-61) for the continuous Treg depletion. We then analyzed MDSCs on day 25. Consistently, the injection of PC-61 also increased the proportion of MDSCs compared with that of the isotype control (Fig. 4D, right). To confirm this result, we inoculated H22 tumor cells to nude mice and treated the mice with 0.5 mg CY for 24 d. We analyzed MDSCs in bone marrow and tumor tissue and found that such low-dose CY treatment had no effect on the number of MDSCs under the condition of T cell deficiency in nude mice (Fig. 4E). These data suggested that low-dose CY-induced MDSCs are attributable to depletion of Tregs by CY.
We also clarified whether the tumor-promoting effect of Treg depletion was mediated by MDSCs. Although MDSCs are heterogeneous, Gr-1 is the common marker. Recently, the depletion of MDSCs by anti-mouse Gr-1 Ab (clone RB6-8C5) has been shown by different laboratories, including ours (27–29). Using this approach, we started to i.p. inject anti–Gr-1 Ab into CY-treated, Treg-ablated mice on day 14 then once every 3 d. We found that the depletion of MDSCs by Gr-1 Ab reversed the tumor-promoting effect by Treg depletion (Fig. 4F). Therefore, the long-term Treg depletion results in the induction of MDSCs, which promotes tumor growth.
Treg depletion increases lipoxin A4 production for MDSC induction
Next, we wondered how MDSCs were induced by Treg depletion. Previous studies reported that tumor cells produce stem cell factor (SCF), which is required for MDSC accumulation in the tumor-bearing mice (30). However, we found that Treg depletion had no effect on SCF expression in tumor (Fig. 5A, top). Besides SCF, PGE2, derived from arachidonic acid through cyclooxygenase-catalyzed pathway, may also act as an inducer of MDSCs (31). We thus determined the levels of PGE2 in peripheral blood and found that 25-d Treg depletion did not alter peripheral PGE2 levels (Fig. 5A, bottom), suggesting that PGE2 is not involved in the induction of MDSCs in our model. We then detected another arachidonic acid-derived lipid mediator, lipoxin A4, which is synthesized through lipoxygenase-catalyzed pathway and highlighted as a key regulator of inflammation (17–19). Notably, the levels of lipoxin A4 were gradually increased by Treg depletion (Fig. 5B, top), indicating that lipoxin A4 might be involved in the induction of MDSCs. To test this, we cultured bone marrow cells with the supernatant from a freeze–thaw lysate of tumor cells in the presence or absence of lipoxin A4 or its receptor agonist BML-111. Seventy-two hours later, we collected cells for flow cytometric analysis. The result showed that lipoxin A4 and BML-111 both increased the proportion of Gr-1+CD11b+ MDSCs (Fig. 5C, top). To validate this in vitro data in vivo, we injected lipoxin A4 or BML-111 to naive or tumor-bearing mice daily for 7 d. Consistently, the administration of lipoxin A4 or BML-111 could induce MDSCs in either naive or tumor-bearing mice (Fig. 5C, bottom), suggesting that Treg depletion may induce MDSCs by increasing lipoxin A4 production.
Finally, we clarified the regulation of lipoxin A4 by Treg depletion. After 25-d Treg depletion in tumor-bearing mice, we determined the expression of lipoxygenases 5 and 12, two critical enzymes for lipoxin biosynthesis (19). In line with the increased levels of lipoxin A4, we found that these two enzymes were upregulated in bone marrow, spleen, and blood (Fig. 5D, top), suggesting the potential regulation of lipoxin production by Treg depletion. We then used a comparable approach to confirm this. We isolated Tregs and adoptively transferred these cells to large tumor-bearing mice. Twenty-four hours later, we measured the levels of lipoxin A4. The result showed that the levels of lipoxin A4 were decreased by the adoptive transfer of Treg cells (Fig. 5B, bottom), concomitant with the downregulated expression of lipoxygenases 5 and 12 (Fig. 5D, bottom). Taken together, these data suggested that Treg depletion induces MDSCs by regulating lipoxin A4 production.
Several implications may be drawn from the current study. 1) Tregs might act as a double-edged sword in tumor: the protumor effect by inhibiting antitumor immunity and the antitumor effect by preventing tumor inflammation. 2) The relationship of Tregs and MDSCs might be coordinative or restrictive in tumor inflammation, depending on the microenvironment. 3) Endogenous lipoxin A4 might be involved in the regulation of the balance between Tregs and MDSCs.
To study the pathophysiological role of Tregs in vivo, several strategies are adapted to deplete or inactivate Tregs, including genetic mutation of Foxp3 gene, anti-CD25, anti-GITR, or anti–CTLA-4 Ab treatment (32–35). However, the mutation of the foxp3 gene profoundly affects the host immune system from its early development, and CD25, CTLA-4, and GITR are also expressed by other immune cells. Therefore, the selective depletion of Tregs is highly desirable. Recently, low-dose treatment with the chemotherapeutic drug CY has been highlighted for Treg depletion (20–23). In this study, we confirmed and expanded this observation by establishing a repeating Treg depletion model. We show that Tregs are selectively depleted by the continuous injection of 0.5 or 0.25 mg CY. These doses are safe because neither tumor cells nor other cell types were affected, and the functions of liver and kidney were not changed in mice tested in this study. Other advantages of this method include the relatively simple technical protocol and low cost. Therefore, we establish an ideal approach for long-term Treg depletion in a murine model.
Tregs accumulate in tumor and peripheral lymphoid organs, but the teleological reason for this accumulation remains unclear. Due to their potent immunosuppression, Tregs are currently considered as a major hurdle for developing antitumor immunotherapy. Many studies show that the ablation of Tregs results in the repression of tumors (36–38), and high frequency of Tregs is associated with a worse prognosis (39–41). However, the inconsistent findings are mounting. In inflammation-associated intestinal tumors, Tregs could prevent cancer development (16); and in gastric and colorectal cancers and certain lymphomas, Treg numbers are an indicator of a favorable prognosis (42–44). In the current study, we found that although they promote early tumor growth, Tregs exert an inhibitory effect on tumors later on. Consistently, Curtin et al. (45) reported that although Treg depletion inhibited or even eliminated small glioblastoma, no improvement in survival was observed when Tregs were depleted 24 d after tumor implantation. Therefore, Tregs might have two opposite effects on tumors at different developmental stages of tumor, and these effects also depend on the tumor microenvironment. From a teleological view, we speculate in this study that the body mobilizes Tregs to control the dangerous inflammation by tumor, however the body has to make a compromise by sacrificing the antitumor immunity, thus leading to both protumor and antitumor effects of Tregs.
One surprising finding in this study is that Treg depletion results in the induction of MDSCs. As early as the 1980s, a correlation between accumulation of immature myeloid cells and immune suppression was recognized in both tumor-bearing mice and cancer patients (46–48). Now, MDSCs are considered to be important in tumor progression (49). An important pathway mediating the immunosuppression function of MDSCs is through educating tumor-specific T cells to become Tregs (10). However, whether and how Tregs act on MDSCs remains unclear. In this study, we find that Tregs in large tumor-bearing mice actually suppress the induction of MDSCs, and Treg depletion strongly induces MDSCs. Although both Tregs and MDSCs primarily mediate immunosuppression, MDSCs belong to immature myeloid cells, which may lose the suppressive function and become inflammatory cells such as macrophages and others. Moreover, tumor-infiltrating MDSCs may employ a nonimmune pathway to promote tumor growth (50). Therefore, it may not be surprising that Treg depletion promotes large tumor growth by inducing MDSCs.
It has been reported that the lipid mediator PGE2 induces MDSCs in tumor-bearing mice (31). In this study, we found that PGE2 had little or no role in Treg depletion-induced MDSCs. Instead, lipoxin A4, another lipid mediator, is involved in the induction of MDSCs after Treg depletion. Lipoxin A4 is an eicosanoid locally and endogenously synthesized from arachidonic acid by 5-lipoxygenase (5-LO) and 12-LO or 15-LO. Upon Treg depletion, the levels of lipoxin A4 in peripheral blood were gradually increased in our murine tumor model. Consistently, the expression of 5-LO and 12-LO was also upregulated in peripheral blood, bone marrow, and spleen. More convincingly, the injection of lipoxin A4 into tumor-bearing or untreated mice strikingly induced the accumulation of MDSCs. Therefore, Treg depletion may lead to MDSC induction by increasing production of lipoxin A4. Even though a detailed mechanism remains to be elucidated, we hypothesize that the altered inflammatory signals by Treg depletion modulate activities of some transcription factors, leading to the upregulation of lipoxygenases for lipoxin A4 production.
We should indicate the possible limitations of the approaches used to deplete Tregs in this study. We adapted the low-dose CY approach to deplete Tregs and obtained the experimental results. However, it is still possible that low-dose CY might affect cells other than Tregs, thus influencing the interpretation of the results. However, we also used anti-CD25 Ab PC-61 for Treg depletion in this study. We stained splenocytes from large tumor-bearing mice (21 d) and found that in CD4+CD25+ T cell subset, 4.8% cells were Foxp3 negative. After the treatment with PC-61, such CD25+ effector T cell population decreased to 0.82% (data not shown). Thus, although the proportion of CD25+ effector T cells is very low in large tumor-bearing mice, these cells are still affected by PC-61. This might also influence our interpretation of the results more or less. In addition, it is reported that diphtheria toxin receptor may be introduced into the 3′ untranslated region of mouse Foxp3, resulting in the specific depletion of Tregs by the administration of diphtheria toxin (51). However, this approach is not very suitable for our experiments, because the consecutive injection of diphtheria toxin might cause additional effects on immune cells (52) and on H22 hepatocarcinoma tumor cells (53, 54). Therefore, the really ideal approach for Treg depletion in murine tumor models seems unavailable to date.
In summary, our data show that Tregs play opposite roles in tumor development: promoting early and inhibiting late tumor growth. This might be explained by the observation that Treg depletion increases lipoxin A4 levels, which leads to the induction of MDSCs. On the basis of our findings, we suggest that the interplay among Tregs, MDSCs, and lipoxin A4 tunes the regulation of tumor inflammation and maintenance of a relative homeostasis in the tumor-bearing host.
We thank Dr. Lee Leserman (Centre d'Immunologie de Marseille-Luminy) and Dr. Yan Su (University of Maryland) for helpful discussion and assistance in editing this article.
Disclosures The authors have no financial conflicts of interest.
This work was supported by the Scientific Research Foundation of Wuhan City Human Resource for Returned Scholars, the Program for New Century Excellent Talents in University (NCET-08-0219), the Funds for International Cooperation and Exchange of the National Natural Science Foundation of China (30911120482), the Special Research Foundation for Universities affiliated with China Ministry of Education (Z2009005), and the Important National Science & Technology Specific Projects (2009ZX09301-014).
Abbreviations used in this paper:
- glutamic-pyruvate transaminase
- inducible NO synthase
- myeloid-derived suppressor cell
- stem cell factor
- tumor-infiltrating lymphocytes
- regulatory T cell.
- Received June 7, 2010.
- Accepted October 6, 2010.