The CUL4B/AKT/β-Catenin Axis Restricts the Accumulation of Myeloid-Derived Suppressor Cells to Prohibit the Establishment of a Tumor-Permissive Microenvironment Free

Cancer initiation and progression require a favorable microenvironment that includes immune cells, endothelial cells, pericytes, mesenchymal cells and many soluble factors (1, 2). Myeloid-derived suppressor cells (MDSC) have emerged as one of the major players that enable cancer to escape immunosurveillance (1–4). In mice, MDSCs are characterized by the coexpression of CD11b and Gr-1 markers (5, 6). Their immunosuppressive activity is attributable to high activity of arginase 1 (ARG1) and/or inducible nitric oxide synthase (iNOS), production of nitric oxide (NO) and reactive oxygen species (ROS), as well as release of IL10 and TGFβ (3, 4, 7–9). Although tumor- and host-secreted factors, including VEGF, stem cell factor (SCF), granulocyte-macrophage colony-stimulating factor (GM-CSF), IL1β, IL6, and prostaglandin E2, have been reported to promote the expansion and activation of MDSCs during tumor development (3, 4, 8), little is known about the intracellular signaling pathways that regulate the expansion of MDSCs.

β-Catenin is critical for Wnt signaling, a pathway that governs many developmental processes, including specification of cell fate and the maintenance of stem cell pluripotency (10, 11). Many studies indicated that a tight regulation of Wnt signaling is required for normal hematopoiesis and lymphopoiesis (12–15). β-Catenin was recently shown to negatively regulate the functions of immunosuppressive cells, including regulatory T cell (Treg) and MDSCs (16–18). However, how Wnt signaling is regulated in those immunosuppressive cells remains to be elucidated.

Cullin 4B (CUL4B), as a scaffold protein in the Cullin 4B-RING E3 ligase complex (CRL4B), participates in the regulation of diverse physiologically and developmentally controlled processes (19–22). Mutations in human CUL4B cause X-linked mental retardation (23, 24). Recently, CUL4B has been shown to be substantially upregulated in various types of solid tumors and possess potent oncogenic properties. H2AK119 monoubiquitination catalyzed by CRL4B is critical for epigenetic inactivation of tumor suppressors by PRC2, SUV39H1/HP1/DNMT and SIN3A/histone deacetylase (HDAC) complexes (25–27). In this study, we demonstrated that CUL4B, AKT and β-catenin function in an axis that negatively regulates MDSCs. CUL4B sustains the activation of AKT/β-catenin pathway by repressing the antagonists of AKT. The CUL4B/AKT/β-catenin axis was downregulated in MDSCs and was further suppressed in tumor-bearing hosts. Our results establish a functional dichotomy of CUL4B in cancer cells and in tumor microenvironment.

Cul4b^flox/flox mice, as previously described (28), were crossed to Tek-Cre transgenic mice to produce Cul4b^flox/YTek-Cre^+/− (CKO) mice and Cul4b^flox/YTek-Cre^−/− (WT) littermate controls. Mice were housed in a specific pathogen-free animal facility at Shandong University. All procedures involving mice were approved by the Animal Care and Use Committee of Shandong University School of Medicine.

B16/F0 (C57BL/6 murine melanoma cells), 4T1 (BALB/c murine mammary tumor cells) and RAW264.7 (BALB/c murine macrophage leukemia) were cultured at 37°C in DMEM complete media containing 10% FBS. EL4 (C57BL/6 murine lymphoma cells) were cultured at 37°C in RPMI1640 containing 10% FBS. All cell lines were purchased from the authenticated ATCC repository.

Tissues from Cul4b^flox/YTek-Cre^+/−R26R mice were fixed in paraformaldehyde. Seven micron thick sections were stained with X-gal and counterstained with eosin. Photomicrographs were taken using an Olympus IX71 with DP71 camera.

B16 (2 × 10⁵), EL4 (2 × 10⁵) or 4T1 (5 × 10⁶) tumor cells were suspended in PBS and inoculated subcutaneously in the flank of mice. Tumor measurements were performed every 2 or 3 days with a vernier caliper, and volumes were calculated using the following formula: V = ½{length (mm) × [width (mm)]²}. For the myeloid/tumor admix experiments, 2 × 10⁵ MDSCs were mixed with 2 × 10⁵ EL4 cells and implanted. For adoptive transfer assay, 3 × 10⁶ MDSCs were injected intravenously into CD45.1 WT tumor-bearing mice on days 3 and 5 after tumor challenge. For lithium chloride (LiCl) treatment experiment, 5 days after EL4 transplantation, the WT or Cul4b CKO recipients were randomly divided into two groups, and treated daily with saline (0.9% NaCl/100 μL) or with LiCl (0.2 mmol/100 μL, GIBCO-BRL) intraperitoneally. In the tumor metastasis model, mice administered with B16/F0 cells (2 × 10⁵) in 200 μL PBS via tail vein were sacrificed at day 21 and evaluated for tumor metastasis into lungs.

MDSCs were purified by immunomagnetic separation according to the manufacturer's protocol (Miltenyi Biotec). Cell purity (>95%) was determined by flow cytometric analysis using anti-CD11b and Gr-1 antibodies. MDSCs isolated from tumor-bearing mice were induced to differentiate in the presence or absence of 10 ng/mL GM-CSF (Sigma-Aldrich) for 3 and 5 days.

Lin⁻ cells were purified using a biotinylated lineage cell depletion cocktail (Miltenyi Biotec). The differentiation of Lin⁻ cells was as described (18).

Human peripheral blood was collected before treatment from cancer patients aged between 45 and 65. Control samples were collected from age- and gender-matched healthy controls. MDSCs were isolated from freshly peripheral blood mononuclear cells by using HLA-DR–negative selection and CD33-positive selection magnetic beads (Miltenyi Biotec). Purity was confirmed by flow cytometry using anti-HLA-DR and CD33 antibodies (BioLegend; >90%). Use of human specimens was approved by the Institutional Review Board at Shandong University School of Medicine. Informed consent was obtained from all blood donors.

Freshly isolated splenocytes (5 × 10⁶ cells/mL) from C57BL/6 WT mice or OT-1 mice were depleted of red cells and labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE; 1 μmol/L; Invitrogen) for 10 minutes at 37°C and washed with fresh culture media. Splenocytes from WT mice were stimulated with coated anti-CD3 antibody (10 μg/mL; BioLegend) and soluble anti-CD28 antibody (6 μg/mL; BioLegend). Splenocytes from OT-1 mice were stimulated with SIINFEKL (10 pmol/L; Sigma). MDSCs (1 × 10⁵) purified from spleen were cocultured with 2 × 10⁵ CFSE-labeled splenocytes for 72 hours in RPMI1640 (10% FBS) in U-bottom 96-well plates. Cells were stained for anti-CD4 or anti-CD8 antibody and analyzed for CFSE dilution using FACSCantoII (BD Biosciences).

Single-cell suspensions were freshly prepared from mouse tissues. Cells were stained as described (17) with the antibodies listed in Supplementary Table S1. Data were analyzed with FlowJo 7.6.5 (Treestar) software.

The measurements of NO and ROS were as previously described (17). DCFDA (Invitrogen) was used to measure ROS production by purified G-MDSCs. Nitrite was quantified by a standard Greiss reaction (Invitrogen).

Lentivirus shCul4b, PLOC-CUL4B, PLOC-CA-β-catenin, shPhlpp2, and shPpp2cb as well as their controls were prepared in HEK293T cells, as described previously (29). For the differentiation assay of mouse MDSCs and Lin⁻ cells, 24 hours after infected with viral particles, the cells were cultured in fresh medium containing GM-CSF or GM-CSF and IL4 for 5 days.

Quantitative real-time PCR, Western blotting, and chromatin immunoprecipitation (ChIP) assays were performed as previously described (20, 25). Primers and antibodies are listed in Supplementary Tables S1 and S2.

Statistical analysis was performed using unpaired Student t test to calculate a two-tailed P value between two groups. Differences were considered significant at P < 0.05. The data are recorded as the mean ± SEM of the indicated number of mice or biologic replicates and as the mean ± SD of the indicated number of technical replicates.

Cul4b heterozygous mice exhibit impaired vascularization in placentas (28). An elevation in monocytes was also detected in patients with CUL4B mutation (24). To test the function of CUL4B in angiogenesis and hematopoiesis, we generated Cul4b^flox/YTek-Cre^+/− mice, referred to as Cul4b CKO hereinafter. CUL4B was effectively ablated in endothelial cells and hematopoietic tissues such as blood, spleen and bone marrow in Cul4b CKO mice (Fig. 1A and B). Cul4b CKO mice did not exhibit overt histologic defects (data not shown), and the percentages of lymphocytes, dendritic cells, macrophages and natural killer cells in spleen and the various hematopoietic progenitors in bone marrow were similar between two groups (Supplementary Tables S3 and S4). But the CD11b⁺Ly6G⁺ neutrophils and CD11b⁺Ly6C⁺ monocytes, the two subsets of CD11b⁺Gr-1⁺ immature myeloid cells (30), were increased in the spleen of Cul4b CKO mice. Consistently, flow cytometry analyses revealed a significant increase in the population of CD11b⁺Gr-1⁺ cells in peripheral blood, spleen, and bone marrow of Cul4b CKO mice (Fig. 1C).

To determine whether CD11b⁺Gr-1⁺ cells in Cul4b CKO mice possess immunosuppressive activity, we examined their inhibitory effects on T-cell proliferation in vitro. As shown in Fig. 1D, Gr-1⁺ cells from Cul4b CKO mice, but not those from Cul4b WT mice, significantly inhibited CD8⁺ T-cell proliferation under both antigen- and mitogen-driven stimulatory conditions. These results suggest that the CD11b⁺Gr-1⁺ cells that are highly accumulated in Cul4b CKO mice probably represent MDSCs.

Because MDSCs are generally expanded in the presence of tumor and favor tumor establishment and progression, we next determined whether the increased abundance of CD11b⁺Gr-1⁺ cells in Cul4b CKO mice would promote tumor growth. The tumor masses formed by injected tumor cells, EL4 lymphoma cells, or B16/F0 melanoma cells, were significantly larger in Cul4b CKO mice than in WT mice (Fig. 2A and Supplementary Fig. S1A). In accordance with accelerated tumor growth in Cul4b CKO mice, MDSCs were expanded more drastically (Fig. 2B and Supplementary Fig. S1B). We next determined whether lack of CUL4B has distinct effects on the two MDSC subsets and found that both granulocytic (G-MDSCs) and monocytic (M-MDSCs) MDSCs were significantly elevated in Cul4b CKO mice compared with WT mice (Fig. 2C), suggesting that CUL4B has a negative effect on both subpopulations.

To determine whether the increased abundance of MDSCs is responsible for enhanced tumor growth in Cul4b CKO mice, we tested the effect of MDSCs admixed with EL4 cells. MDSCs from Cul4b CKO mice were significantly more potent in promoting tumor growth than those from WT mice (Supplementary Fig. S1C). Consistently, when adoptively transferred into tumor-bearing CD45.1⁺ WT recipients, CD45.2⁺ Cul4b-deficient MDSCs were more potent than their WT counterparts in promoting tumor growth (Fig. 2D). The fact that MDSCs from CKO mice possessed a stronger tumor-promoting property, whereas their retention in recipients was indistinct from that of their WT counterparts (Fig. 2E) suggests that Cul4b-deficient MDSCs individually have stronger tumor-promoting property. Thus, the MDSCs in CKO mice could have contributed to a tumor-permissive environment by their increased abundance and by the enhanced activity of the individual cells. The function of CD8⁺ T cells and natural killer cells, on the other hand, appeared to be unimpaired by lack of CUL4B (Supplementary Fig. S2). Strikingly, even BALB/c 4T1 mammary cancer cells were able to form tumor grafts and grow in C57BL/6 Cul4b CKO (Fig. 2F), suggesting that Cul4b-deficient mice are greatly immunocompromised. We also determined the effect of CUL4B depletion on metastasis of intravenously injected B16/F0 melanoma cells to lung. As shown in Fig. 2G, both the number and the volume of micrometastatic nodules were markedly increased in the Cul4b CKO mice.

We next determined whether tumor-associated MDSCs are functionally more potent in Cul4b CKO mice than in WT. Remarkably, G- and M-MDSCs from Cul4b CKO tumor-bearing mice were more suppressive on CD8⁺ T-cell proliferation than those from WT mice in both antigen- and mitogen-driven T-cell stimulatory conditions (Fig. 3A). The frequencies of CD4⁺ and CD8⁺ T cells in tumors from Cul4b CKO mice were significantly lower than those in WT mice (Fig. 3B). Consistently, the frequencies of CD4⁺ and CD8⁺ T cells in spleen were significantly decreased in tumor-bearing Cul4b CKO mice (Fig. 3B). We also examined the frequencies of Tregs in spleen, thymus and tumor. While the numbers of Tregs remained largely unchanged in thymus and spleen of Cul4b CKO mice, an increased abundance of Tregs was detected in tumor grafts of Cul4b CKO mice (Fig. 3C). To determine whether increased Tregs in tumor were secondary to the increased accumulation and activity of MDSCs, we examined the effect of MDSCs on Treg induction. As shown in Fig. 3D, MDSCs isolated from Cul4b CKO mice were more efficient in inducing Tregs than those from WT mice.

MDSCs have been shown to suppress antitumor immunity through the production of NO and ROS by M-MDSCs and G-MDSCs, respectively (8). Compared with those isolated from WT mice, MDSCs from Cul4b CKO mice produced ROS and NO at significantly higher levels (Fig. 3E–G). Furthermore, Cul4b CKO MDSCs exhibited higher levels of Arg1, Nox2, Nos2, Cox2, and IL-10 transcripts (Fig. 3H). Taken together, these results indicate that lack of CUL4B does not only permit a greater accumulation of MDSCs but also enhances the immunosuppressive activity per cell, thus conferring a more permissive microenvironment for tumor progression.

We evaluated the proliferation and apoptosis of MDSCs in bone marrow in vivo in EL4 tumor-bearing Cul4b CKO and WT mice, but found no difference (Fig. 4A and B). Similarly, no significant differences in the proliferation or apoptosis of CD11b⁺Gr-1⁺ cells were observed in naive WT and CKO mice (Supplementary Fig. S3A and S3B). These results indicate that the enhanced accumulation of MDSCs in Cul4b CKO mice may not be caused by abnormal proliferation or apoptosis.

We next examined whether MDSCs are more favorably generated from hematopoietic progenitors in the absence of CUL4B. Lineage-negative (Lin⁻) hematopoietic progenitor cells isolated from bone marrow were cultured for 5 days in the presence of IL4 and GM-CSF to induce myeloid differentiation. There was a significantly increased generation of MDSCs, marked by the expression CD11b and Gr-1, from Cul4b-deficient progenitors compared with WT progenitors (Fig. 4C).

Because MDSCs could differentiate into dendritic cells (DC; CD11b⁺CD11c⁺) and macrophages (CD11b⁺F4/80⁺) upon appropriate stimulations (31), we tested whether the increased accumulation of MDSCs in Cul4b CKO mice could also be caused by impaired differentiation of MDSCs into dendritic cells or macrophages. Indeed, when treated with GM-CSF for 3 or 5 days, Cul4b-deficient MDSCs gave rise to lower percentages of CD11b⁺CD11c⁺ and CD11b⁺F4/80⁺ cells than WT MDSCs (Fig. 4D and E and Supplementary Fig. S3C). S100A8 and S100A9, which are known to inhibit MDSC maturation (32), were significantly upregulated in MDSCs isolated from Cul4b CKO mice (Fig. 4F).

In accordance with the impaired differentiation in Cul4b CKO mice, immunoblotting showed that the level of CUL4B was dramatically reduced when Lin⁻ hematopoietic progenitor cells were induced to undergo myeloid differentiation. Conversely, CUL4B expression was increased when MDSCs were induced to undergo maturation (Fig. 4G). Taken together, these results indicate that the increased production of MDSCs from hematopoietic progenitors and their impaired differentiation into mature myeloid cells could be both responsible for the enhanced accumulation of MDSCs in Cul4b CKO mice.

It was reported that loss of β-catenin in myeloid cells resulted in accelerated tumor growth and a greater accumulation of MDSCs, whereas expression of a constitutively active β-catenin mutant in myeloid cells inhibited tumor growth and reduced accumulation of MDSCs (17). Our recent study showed that CUL4B could activate Wnt/β-catenin signaling in human hepatocellular carcinoma (29). Thus, we next examined the effect of Cul4b ablation on the level of β-catenin in MDSCs. Immunoblotting showed that lack of CUL4B resulted in a significantly decreased level of β-catenin in naive Gr-1⁺ cells as well as in two subsets of MDSCs from tumor-bearing mice (Fig. 5A). Because phosphorylation of β-catenin by GSK3β is crucial for its degradation and the phosphorylation of GSK3β itself reflects its inactivation, we next examined the phosphorylation status of GSK3β in MDSCs from Cul4b CKO and WT mice. As shown in Fig. 5A, the levels of GSK3β phosphorylated at Ser9, an inactive form of GSK3β, were significantly reduced in Cul4b-deficient MDSCs, indicating that increased kinase activity of GSK3β, which phosphorylates β-catenin for its degradation, could be responsible for the increased accumulation of MDSCs caused by lack of CUL4B. Indeed, when proteasome inhibitor MG132 was applied to Cul4b-deficient MDSCs, the reduction of β-catenin level was greatly attenuated (Supplementary Fig. S4A). Furthermore, two inhibitors of GSK3β, SB216763 and lithium chloride (LiCl), could efficiently block the reduction of β-catenin level caused by Cul4b ablation (Fig. 5B). Consistently, increased generation of MDSCs from Lin⁻ progenitors and the impaired differentiation of MDSCs into CD11b⁺CD11c⁺ and CD11b⁺F4/80⁺ cells caused by lack of CUL4B were rescued by either of the two inhibitors as well as by ectopic expression of constitutively active β-catenin (Supplementary Fig. S4B and Fig. 5C). The ability of LiCl to decrease the accumulation of MDSCs was also demonstrated in CKO mice in vivo (Fig. 5D). Strikingly, LiCl greatly impeded the tumor growth in Cul4b CKO mice (Fig. 5E) as well as tumor-induced expansion of MDSCs (Fig. 5F). Taken together, these results indicate that increased GSK3β activity, and consequently β-catenin downregulation, may mediate the enhanced accumulation of MDSCs in Cul4b CKO mice.

As Ser9 of GSK3β is phosphorylated by AKT (33), we next determined the level of AKT phosphorylated at Thr308 and Ser473, which marks maximal AKT activity (34, 35). We found that the levels of phosphorylated AKT were markedly decreased in naive Gr-1⁺ cells and in both subsets of MDSCs in CKO mice, whereas the amount of total AKT was not altered (Fig. 6A), indicating that the decreased GSK3β phosphorylation at Ser9 is due to a decreased AKT kinase activity. Decreased level of AKT phosphorylation could be attributable to decreased kinase activities upstream of AKT or increased activities of phosphatases that dephosphorylate AKT. However, no significant difference in the levels of phosphorylated PDK1 (the kinase of AKT at the site of Thr308) and PTEN (a negative regulator of AKT) was found between Cul4b-deificient and control MDSCs (Supplementary Fig. S5). Interestingly, when AKT phosphatases were examined, the expression levels of PP2A (AKT phosphatase for Thr308), PHLPP1 and PHLPP2 (AKT phosphatases for Ser473; refs. 35–37), were found to be significantly elevated in Cul4b-deficient MDSCs (Fig. 6B). Consistently, the phosphorylation level of PKCβII, another substrate of PHLPP1/2 (38), was significantly decreased in Cul4b-deficient MDSCs (Supplementary Fig. S6).

Because CUL4B is a transcriptional corepressor (25), we next examined the mRNA levels of the genes encoding these phosphatases. Quantitative real-time PCR (qRT-PCR) revealed that the mRNA levels of genes encoding different isoforms of PP2A, Ppp2r1a, Ppp2r1b, Ppp2ca and Ppp2cb, as well as those of Phlpp1 and Phlpp2 were much higher in Cul4b-deficient MDSCs than in WT MDSCs (Fig. 6C). Moreover, ChIP assay showed that CUL4B could directly bind to the promoters of the phosphatase genes in MDSCs (Supplementary Fig. S7A). To determine whether the enhanced accumulation of MDSCs in Cul4b CKO mice is due to the upregulation of AKT phosphatases, we examined the generation of MDSCs from Phlpp2- or Ppp2cb- depleted Lin⁻ cells. Importantly, depletion of either Phlpp2 or Ppp2cb significantly attenuated the generation of MDSCs from Cul4b-deficient Lin⁻ cells (Supplementary Fig. S7B and Fig. 6D), suggesting that downregulation of AKT mediates the enhanced production of MDSCs.

The negative regulation of PP2A and PHLPP1/2 by CUL4B was also confirmed in Cul4b knockdown RAW264.7 cells as well as in cells overexpressing human CUL4B (Fig. 6E). Quantitative ChIP assay showed that depletion of Cul4b in RAW264.7 cells resulted in a significantly reduced recruitment of CUL4B, HDAC and EZH2 to the promoters (Fig. 6F and Supplementary Fig. S7C). Consistently, the levels of H3K27me3 and H2AK119ub1 were also markedly decreased, whereas those of acetylated H3 and H4 were increased, providing further evidence that CUL4B is critical for the transcriptional repression of these phosphatases. Furthermore, while overexpression of CUL4B could lead to a remarkable decrease in the expression of Phlpp1, Phlpp2, Ppp2ca, and Ppp2cb, treatment with HDACs inhibitor TSA, EZH2 inhibitor DNZep or both could efficiently attenuate the reduction in the expression of these genes (Fig. 6G). Taken together, these results indicate that CRL4B/PRC2/HDAC complexes can repress AKT phosphatases PP2A and PHLPP1/2 by promoting H2AK119 monoubiquitination, H3K27 trimethylation, as well as H3 and H4 deacetylation, en route to activate AKT/GSK3β/β-catenin signaling pathway.

The data shown above clearly demonstrated a critical role of CUL4B in activating the AKT/β-catenin cascade that limits the accumulation of MDSCs. To determine whether this finding bears general implications, we measured the levels of CUL4B in MDSCs and other hematopoietic cells in wild-type mice. As shown in Fig. 7A, Cul4b was remarkably downregulated in bone marrow and spleen MDSCs when compared with other hematopoietic cells. Similarly, CUL4B in MDSCs isolated from human peripheral blood, based on the absence of HLA-DR and expression of CD33, was also significantly downregulated when compared with other leukocytes (Fig. 7B). We speculated that the CUL4B/AKT/β-catenin pathway may generally restrain the population of MDSCs and that a downregulation of CUL4B is probably required for the generation of MDSCs from their progenitors and for the maintenance of their immunosuppressive ability. Indeed, there was a significant reduction in the levels of CUL4B, β-catenin, p-GSK3β, and p-AKT in MDSCs isolated from tumor-bearing mice when compared with those from tumor-free mice (Fig. 7C). Consistently, the levels of AKT phosphatases PP2A and PHLPP1/2 were significantly increased (Fig. 7C).

We next determined whether CUL4B/AKT/β-catenin signaling pathway is altered in MDSCs of patients with cancer. Consistent with previous reports (2–4), the percentages of MDSCs were significantly higher in 22 patients with cancer than in age-matched healthy donors (Fig. 7D). Furthermore, MDSCs from cancer patients were more suppressive on CD4⁺ and CD8⁺ T-cell proliferation than those from healthy controls (Fig. 7E). Importantly, the expression levels of CUL4B, measured by qRT-PCR, were negatively correlated with the percentages of MDSCs (Fig. 7F). Western blot analysis showed a significant reduction in the levels of CUL4B, β-catenin, p-GSK3β (Ser9) in MDSCs of cancer patients compared with those of healthy donors (Fig. 7G). Correspondingly, PP2A was significantly upregulated (Fig. 7G). Taken together, these results indicate that CUL4B/AKT/β-catenin signaling pathway is generally downregulated in MDSCs during tumor progression.

CUL4B has recently been shown to be highly expressed in a variety of human malignancy and is positively correlated with tumor progression (25, 26, 29, 39, 40). In the current study, we demonstrated that CUL4B functions to limit the expansion and activity of MDSCs that are an integral part of the immunosuppressive tumor microenvironment. The increased accumulation of MDSCs in mice that lack CUL4B in hematopoietic cells greatly facilitated tumor growth and metastasis, and even tolerated allogeneic tumor growth. Thus, while CUL4B may drive the malignancy of cancer cells, it negatively regulates a cancer-supporting microenvironment in hematopoietic system. Thus, the function of CUL4B in cancer development is dichotomous.

Our finding that the negative regulation of MDSCs by CUL4B is mediated by β-catenin is consistent with two recent reports of β-catenin being a negative regulator of MDSCs (17, 18). We showed that pharmacologic inhibition of GSK3β drastically reduced the growth of tumor grafts in Cul4b CKO mice. Importantly, our findings provided an insight into the mechanism by which β-catenin is regulated. We demonstrated that AKT, which phosphorylates and inactivates GSK3β (33), is positively regulated by CUL4B. CUL4B functions to sustain AKT activity by transcriptionally repressing AKT phosphatases, such as PP2A and PHLPP1/2, via H2AK119 monoubiquitination. The fact that the repression of those phosphatases by CUL4B can be compromised by HDAC and EZH2 inhibitors suggests that the functions of PRC2 and HDACs, which are physically associated with CRL4B, are also essential for the epigenetic silencing of those phosphatases. Interestingly, unlike in solid tumors in which EZH2 acts as an oncogene, inactivating somatic mutations in EZH2 cause myelodysplastic/myeloproliferative disorders, implicating EZH2 as a tumor suppressor in myeloid malignancy (41, 42). Thus, EZH2 appears to behave similarly in this regard (41–43). Because EZH2 is essential for the function of PRC2, it is possible that an EZH2/AKT/β-catenin axis may operate in suppressing myeloid malignancy. The negative regulation of MDSCs by AKT described here bears some analogy to the reduction of AKT activity required for the development and function of Tregs, another kind of immunosuppressive cells contributing to tumor immune evasion (44, 45), although dephosphorylation of AKT occurs only at Ser473, but not at Thr308, in activated Tregs. These findings suggest that the AKT signaling is regulated differently in Treg and MDSCs. Indeed, the numbers of Tregs remained largely unchanged in thymus and spleen of Cul4b CKO mice. Although there was an increased abundance of Tregs in tumor grafts in Cul4b CKO mice, it could be secondary to the increased accumulation and activity of MDSCs. PTEN, a primary negative regulator of AKT, was shown to be repressed by CUL4B in cancer cells (25). However, the level of PTEN was not found to be elevated in Cul4b-deficient hematopoietic cells, including MDSCs. Thus, the CUL4B/AKT/β-catenin axis described here could be cell type specific.

Our studies showed that whether CUL4B functions as a promoter or as a suppressor of tumorigenesis is context dependent. Because AKT/β-catenin signaling cascades in carcinomas are often sought as therapeutic targets, systemic downregulation of those pathways may sabotage the antitumor immune defense due to the expansion and activation of immunosuppressive cells such as MDSCs and Tregs. On the other hand, the aberrant accumulation of MDSCs, and its promotion of tumor growth, in Cul4b CKO mice could be managed by the use of GSK3β inhibitors. Lithium, commonly used for treatment of bipolar disorders, may help restore cancer-targeting immunosurveillance.

No potential conflicts of interest were disclosed.

Conception and design: Y. Qian, C. Shao, Y. Gong

Development of methodology: Y. Qian, H. Hu, Q. Yang, B. Jiang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Qian, Q. Yang, J. Li, B. Jiang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Qian, J. Yuan, C. Shao

Writing, review, and/or revision of the manuscript: Y. Qian, C. Shao, Y. Gong

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Yuan, S. Zhang

Study supervision: C. Shao, Y. Gong

This work was supported by National Basic Research Program of China (973 Program) grants (2011CB966200 to C. Shao; 2013CB910900 to Y. Gong; 81330050 and 81571523 to Y. Gong; 81101522 to B. Jiang) from National Natural Science Foundation of China.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.