Epigenetic silencing of retinoblastoma gene regulates pathologic differentiation of myeloid cells in cancer

Abstract

Two major populations of myeloid-derived suppressor cells (MDSCs), monocytic MDSCs (M-MDSCs) and polymorphonuclear MDSCs (PMN-MDSCs) regulate immune responses in cancer and other pathologic conditions. Under physiologic conditions, Ly6C(hi)Ly6G(-) inflammatory monocytes, which are the normal counterpart of M-MDSCs, differentiate into macrophages and dendritic cells. PMN-MDSCs are the predominant group of MDSCs that accumulates in cancer. Here we show that a large proportion of M-MDSCs in tumor-bearing mice acquired phenotypic, morphological and functional features of PMN-MDSCs. Acquisition of this phenotype, but not the functional attributes of PMN-MDSCs, was mediated by transcriptional silencing of the retinoblastoma gene through epigenetic modifications mediated by histone deacetylase 2 (HDAC-2). These data demonstrate a new regulatory mechanism of myeloid cells in cancer.

Figure 1. MDSC populations in tumor-bearing mice

(a) Flow cytometry analysis of Ly6C and Ly6G expression after gating on CD11b⁺ cells in MDSCs from spleen, bone marrow, blood and tumors from EL-4 tumor-bearing mice (left) and Wright-Giemsa staining of PMN-MDSC and M-MDSC sorted from BM of EL-4 tumor-bearing mice. Scale bar: 10 μm. Plots representative of 6 mice and staining representative of 4 mice. (b,c) Percentage (b) and total number (c) of PMN-MDSC and M-MDSC in the bone marrow, blood and spleen of EL-4 tumor-bearing mice followed during 4 weeks after s.c. inoculation of tumor (N – naive mice). Mean ± SD from 4 mice per groups. (d,e) Proliferation of M-MDSC, PMN-MDSC, monocytes (Mon) and PMN in the bone marrow (d) or spleens (e) of naïve and EL-4 tumor-bearing mice, measured by BrdU incorporation 5 or 24 hr after i.p. injection (2 mg/mouse) and gated as in (a) based on expression of CD11b, Ly6C and Ly6G. Percentage of BrdU⁺ cells is shown in the plots. Results are representative of 3 mice. (f) Total number of cells recovered after 3 and 5 days of culture with GM-CSF and tumor explant supernatants (TES) starting from 2.5×10⁵ PMN, PMN-MDSC, monocytes and M-MDSC sorted from naïve and EL-4 tumor-bearing mice. Mean ± SD from three experiments (each included cells pooled from 2–3 mice) are shown. Dashed lines show the initial number of cells in culture. * p<0.05. (g) Ly6C and Ly6G expression on Gr-1⁺CD11b⁺ MDSC cultured with GM-CSF and tumor explant supernatants at the indicated time points. Two experiments with the same results were performed.

Figure 2. Differentiation of PMN-MDSC from M-MDSC in vitro

(a) Differentiation of sorted PMN-MDSC or M-MDSC during 3 day culture with GM-CSF and tumor explant supernatants. Two experiments with the same results were performed. Each experiment included cells pooled from 2–3 mice. (b–e) Differentiation of sorted M-MDSC and monocytes in vitro. Cells were cultured for 5 days with GM-CSF (monocytes with and without tumor explant supernatants, M-MDSC with tumor explant supernatant). Each experiment included cells pooled from 2–3 mice. (b) The phenotype of gated CD11b⁺ cells. Three experiments with the same results were performed. (c) Wright-Giemsa staining of the cells. Scale bar: 10 μm. (d) Myeloperoxidase (MPO) activity of sorted M-MDSC and PMN-MDSC after 5-day culture of M-MDSC. Two experiments (each in triplicates) were performed. (**-p<0.01). Control PMN – CD11b⁺Ly6C^loLy6G⁺ cells were sorted from BM of naive mice (e) ROS level in cells, measured by staining with DCFDA. Three experiments with the same results were performed. (f) Staining of cells differentiated from M-MDSC with naphthol-AS-D chloroacetate esterase. Scale bar: 10 μm. Control Mon - monocytes sorted from BM of naïve mice, Control PMN – PMN sorted from peritoneum after mobilization with casein. (g) Immune suppressive activity of CD11b⁺Ly6C^loLy6G⁺ PMN-MDSC sorted after 3-day culture of M-MDSC with GM-CSF and tumor explant supernatant. Cells were stimulated with CD3/CD28 antibodies and the number of IFN-γ producing cells was evaluated in triplicates in ELISPOT ** -p<0.01. Two experiments with the same results were performed.

Figure 3. Differentiation of PMN-MDSC from M-MDSC in vivo

(a) Differentiation of sorted CD45.1⁺ PMN-MDSC and M-MDSC in vivo in CD45.2⁺ recipients. 2×10⁷ PMN-MDSC or 5×10⁶ M-MDSC were injected and evaluated 2 days later. A typical example from two performed experiments is shown. Donor’s (CD45.1⁺) cells were analyzed. (b) Differentiation of sorted and CMFDA labeled PMN-MDSC (2×10⁷), monocytes (Mon) (5×10⁶) or M-MDSC (5×10⁶) two days after transfer to either EL-4 tumor-bearing (TB) or tumor-free (naïve) mice. Cumulative results (Mean ± SD) of 4 experiments are shown. Donor’s (CMFDA⁺) cells were analyzed. The differences between monocytes and all other groups were statistically significant (p<0.05).

Figure 4. Expression of Rb in MDSC

(a) The presence of Rb1 protein in splenocytes from naïve (N) or EL-4 tumor-bearing (TB) mice. (b) The expression of Rb1 and E2F1 in spleens of EL-4 tumor-bearing mice, during tumor progression at 0, 1, 2 or 3 weeks after tumor inoculation. (c) Rb1 expression in Gr1⁺ and Gr1⁻ cells isolated from the splenocytes of EL-4 tumor-bearing mice. (a–c) Proteins were analyzed by western blot and all experiments were repeated at least twice. (d) The relative expression of rb1 gene in Gr1⁺ and Gr1⁻ cells, F4/80 macrophages (MΦ), and CD11c⁺ DCs, isolated from spleen of EL-4 tumor-bearing mice, and measured by qRT-PCR. The values were normalized to β-actin. Each experiment was performed in triplicate and repeated twice. Cumulative mean ± SD are shown. The differences between Gr-1⁺ and all other populations were statistically significant with p<0.001. (e) Rb1 protein amount in the total population of Gr1⁺ MDSC, PMN-MDSC and M-MDSC isolated from spleens of EL-4 tumor-bearing mice. PMN were isolated from peritoneum after mobilization by casein; DCs, T-, and B-lymphocytes were isolated from spleens of tumor-free mice. Two experiments were performed. (f) Relative expression of rb1 in M-MDSC and PMN-MDSC isolated from spleens (SPL) or bone marrow (BM) of EL-4 tumor-bearing (TB) mice. Each experiment was performed in triplicate and repeated twice. The differences between M-MDSC and PMN-MDSC was statistically significant with p<0.01. (g) Rb1 expression in PMN, isolated from spleens of tumor-free mice, and PMN-MDSC, isolated from spleens of EL-4 tumor-bearing mice, measured by qRT-PCR. The difference between PMN and PMN-MDSC was statistically significant with p<0.01. (h) The amount of Rb1 protein in PMN, isolated from spleen of tumor-free mice, and PMN-MDSC, isolated from spleen of EL-4 TB mice. (i) Rb1 protein in PMN, isolated from BM of naive mice and PMN-MDSC, isolated from BM of EL-4 tumor-bearing mice, and cultured with GM-CSF for 2 days. Two experiments were performed. (j) Expression of rb1 in sorted M-MDSC and monocytes isolated from BM of tumor-bearing and tumor-free mice. Experiment was performed in triplicate and repeated once. The differences was statistically significant with p<0.05. (k) Rb1 protein level in sorted BM Mon and M-MDSC.

Figure 5. Association of Rb1 with subset of M-MDSC

(a) The staining of sorted BM M-MDSC and monocytes with Rb1-specifc antibody by immunofluorescence (red fluorescence). Nuclei are stained with DAPI (blue fluorescence). Scale bar: 10 μm. Bottom panel – fluorescence intensity from three samples. ** - the statistically significant (p<0.01) differences between groups. (b) The staining of sorted PMN-MDSC, Hoechst^high and Hoechst^low M-MDSC with Rb1 antibody and counterstained with DAPI. Scale bar: 10 μm. c. Rb1 staining of sorted monocytes from BM of naïve mice. (d,e). Rb1 staining of sorted Hoechst^low Mon or M-MDSC, 48 hr after the culture with GM-CSF. (d) A typical example of staining. Scale bar: 10 μm. (e) Fluorescence intensity calculated from three performed experiments * - the statistically significant (p<0.05) differences from Hoechst^high cells (positive control). (f) Differentiation of sorted Hoechst^high and Hoechst^low M-MDSC and monocytes cultured for 4 days with GM-CSF and tumor-explant supernatants (M-MDSC). Three experiments with similar results were performed. Each experiment include cells pooled from 4–5 mice.

Figure 6. PMN-MDSC and M-MDSC in cancer patients

(a) The Percentages of CD14⁻CD11b⁺CD33⁺CD15⁺ granulocytic and CD14⁻CD11b⁺CD33⁺CD15⁻ monocytic cells in the fraction of mononuclear cells from healthy donors or from patients with different types of cancer. *** - p<0.001; ** - p<0.01; * - p<0.05. (b) Differentiation of sorted BM CD14⁺HLA-DR^−/low cells from healthy donors and multiple myeloma (MM) patients cultured for 5 days with GM-CSF. Top panel - typical examples of staining. Bottom panel – the result of all performed experiments. * - p<0.05 (c) Wright-Giemsa staining of the cells after 5-day culture of BM M-MDSC from a MM patient. Scale = 10 μm. (d). Proliferation of M-MDSC. BM mononuclear cells from a MM patient were labeled with 10 mM BrdU for 18 hr and stained for surface markers. Two experiments with similar results were performed. (e) RB1 expression (determined by qRT-PCR) in sorted PMN-MDSC and M-MDSC, in three healthy volunteers and 5 renal cell cancer patients,. Each sample was analyzed in triplicate. (f) RB1 expression in PMN-MDSC and PMN from the same renal cell cancer patients determined by qRT-PCR. ** - p<0.01. (g) RB1 expression in sorted BM CD14⁺HLA-DR^hi and CD14⁺HLA-DR^−/low cells from MM patients.

Figure 7. Rb and regulation of myeloid differentiation in cancer

(a) Rb1 protein in splenocytes of wild type (Rb1-WT) Rb^fl/flMx1-Cre^−/− and knockout (Rb1-KO) Rb^fl/flMx1-Cre^+/− mice, treated with 3 polyI:C injections. (b). The phenotypes of splenocytes analyzed 8 weeks after polyI:C injections. Four mice were analyzed. (c) Differentiation of monocytes, sorted from BM of Rb1-WT or Rb1-KO mice cultured for 5 days with GM-CSF. The proportion of CD11b⁺Ly6C^loLy6G⁺ granulocytic cells was analyzed. Each point represents an individual mouse. ** - p<0.01. (d). Wright-Giemsa stain of cells differentiated from monocytes on day 5. Scale bar: 10 μm. (e). Growth of EG7 tumor in Rb1-KO mice and their wild-type (Rb1-WT) littermates after inoculation of 3×10⁶ cells. Tumor growth of individual mice is shown. (f) Differentiation of Hoechst^low M-MDSC after infection with 25 MOI of control Ad-GFP or Ad-Rb1-GFP viruses and 4-day culture with GM-CSF and tumor explant supernatants. Two experiments with similar results were performed. (g) The phenotype of gated GFP negative cells from the experiment described in Fig. 7f.

Figure 8. The role of HDAC-2 in silencing of Rb1 in MDSC

(a). Expression of rb1 in PMN-MDSC isolated from spleen of EL-4 tumor-bearing mice and cultured in presence of TES, with or without 1 mM VPA, 6 μM SAHA (1–2 days) or 10 nM LBH-589 (1 day). The relative expression of rb1 was analyzed in triplicate by qRT-PCR. Three experiments were performed. The differences between untreated and treated cells were statistically significant (p<0.05). (b). The amount of Rb1 protein in splenic Gr-1⁺ and Gr-1⁻ cells isolated from spleens of tumor-bearing mice (left panel) or in Gr-1⁺ cells cultured for 24 hr with or without 1 mM VPA in the presence of tumor explant supernatants (TES). (c) Differentiation of Hoechst^low M-MDSC from BM of tumor-bearing mice cultured, with or without 1 mM VPA in the presence of GM-CSF and tumor explant supernatants for 24 hr. After that time, the VPA was removed and the cells were cultured for an additional 4 days. Two experiments with the same results were performed. (d). Differentiation of monocytes from BM of Rb KO mice, in the presence of GM-CSF and tumor explant supernatants, with or without 1 mM VPA for 24 hr. After that time, the VPA was removed and cells were cultured for additional 4 days. Two experiments with the same results were performed. (e) ChIP of rb1 promoter with acetylated histone H3 or acetylated histone H4 antibodies in DC and splenic PMN-MDSC cultured overnight, with or without 10 nM LBH 589. The results are expressed as DNA enrichment, normalized to corresponding input values. Each experiment was performed in triplicate 3 times. The differences in PMN-MDSC between LBH 589 treated and not-treated groups were significant (p<0.01) (f). ChIP of rb1 promoter with HDAC-1, -2, -3 or -6 specific antibodies in DC and PMN-MDSC. Each experiment was performed in triplicate 5 times. The differences between DC and PMN-MDSC with HDAC-2 antibody group were significant (p<0.01) (g). ChIP of rb1 promoter performed similar to that described in Fig. 8f with PMN mobilized to peritoneum after casein injection used as a control. Each experiment was performed in triplicate twice. p<0.01 between the groups. (h) Rb1 expression in splenic PMN-MDSC cultured with GM-CSF. Each experiment was performed in triplicates three times. (i). ChIP of rb1 promoter with HDAC-2 antibody in splenic PMN-MDSC cultured for 48 hr with GM-CSF, in the presence of tumor explant supernatants. The mature DCs were used as a control. The differences between PMN-MDSC (0 hr) and all other groups were significant (p<0.01) (j). Rb1 expression in splenic PMN-MDSC transfected with scramble siRNA, or siRNA specific for HDAC-1 or HDAC-2 and cultured for 48 hr. Each experiment was performed in triplicate and repeated twice. * - significant differences from values in scramble siRNA samples (p<0.05).