Myeloid-derived suppressor cells (MDSC) can suppress immunity and promote tumorigenesis, and their abundance is associated with poor prognosis. In this study, we show that SUMO1/sentrin-specific peptidase 1 (SENP1) regulates the development and function of MDSC. SENP1 deficiency in myeloid cells promoted MDSC expansion in bone marrow, spleen, and other organs. Senp1−/− MDSC showed stronger immunosuppressive activity than Senp1+/+ MDSC; we observed no defects in the differentiation of myeloid precursor cell in Senp1−/− mice. Mechanistically, SENP1-mediated regulation of MDSC was dependent on STAT3 signaling. We identified CD45 as a specific STAT3 phosphatase in MDSC. CD45 was SUMOylated in MDSC and SENP1 could deconjugate SUMOylated CD45. In Senp1−/− MDSC, CD45 was highly SUMOylated, which reduced its phosphatase activity toward STAT3, leading to STAT3-mediated MDSC development and function. These results reveal a suppressive function of SENP1 in modulating MDSC expansion and function via CD45–STAT3 signaling axis.
These findings show that increased SUMOylation of CD45 via loss of SENP1 suppresses CD45-mediated dephosphorylation of STAT3, which promotes MDSC development and function, leading to tumorigenesis.
Myeloid-derived suppressor cells (MDSC) are immature myeloid cells that substantially expand during various immunologic stresses. MDSC play an important role in immunosuppressive environments observed in tumor patients, which contributes to tumor progression and immunotherapy inefficiency. MDSC abundance in periphery is associated closely with poor prognosis in tumor patients. Upon tumor challenge, tumor-derived factors induce the expansion of MDSC and the pathologic activation (1). MDSC consist of 2 major subsets: PMN-MDSC and Mo-MDSC. PMN-MDSC are morphologically and phenotypically similar to neutrophil, and Mo-MDSC resemble monocytes. Functionally, PMN-MDSC show moderate suppressive activity via antigen-specific mechanisms, whereas Mo-MDSC have high suppressive activity in nonspecific manners (2).
Tumor-associated MDSC development is usually the result of perturbation of myeloid precursor cells. Several studies also report the conversion of normal MDSC to pathologic MDSC (3). Several transcription factors including STAT3, IRF8, and C/EBPβ have been implicated in MDSC development (4, 5). Among them, STAT3 is the most prominent factor in MDSC expansion, function, and differentiation to tumor-associated macrophages (TAM). Using a selective inhibitor of STAT3 or Stat3 gene ablation can reduce MDSC development and promote DC generation (6, 7). Several tumor-derived factors can activate STAT3 though canonical JAK–STAT pathway. The suppressor of cytokine signaling (SOCS) and phosphatase are critical negative regulators to prevent excessive activation of STAT3 (8). SOCS3 inhibits STAT3 by targeting JAK kinase. Deleting SOCS3 in myeloid cells promotes MDSC development and tumor progression in a STAT3-depended manner (9). Phosphatase PTP1B dephosphorylates JAK2 and STAT3 to inhibit STAT3 signaling and MDSC development (10). Recently, CD45 has been shown as a phosphatase to suppress STAT3 and downregulate MDSC expansion (11).
SUMOylation is a dynamic protein modification. SUMOylation is reversed by deSUMOylation protease SENPs (12). We have showed that SENP1 is essential for erythropoiesis and early lymphoid development. Deficiency of SENP1 leads to embryonic lethality (13). SENP1 deficiency in macrophages increases PTP1B SUMOylation to reduce its phosphatase activity toward JAK2–STAT3 signaling and eventually defects in IFNγ signaling and macrophage polarization (14). However, little is known about the role of SUMOylation or SENPs in myeloid cell development. In this study, we found SENP1 expressed at low level in MDSC as compared with other myeloid cell types. Deletion of SENP1 in myeloid cells resulted in the enormous expansion of MDSC and accelerated tumor progression through STAT3-dependent signaling. We further identified CD45, a phosphatase controlling STAT3 activation, as a SUMOylated protein, and SUMOylation attenuated CD45 phosphatase activity, thereby promoted STAT3 activation. These results reveal an essential role of SENP1 in modulating MDSC development and function via CD45–STAT3 signaling.
Materials and Methods
Senp1+/− mice were previously described (15, 16). Congenic CD45.1+ mice were purchased from The Jackson Laboratory. Mice were housed in a pathogen-free animal facility. All experiments with mice were approved by the Animal Care Committee of Shanghai Jiao Tong University School of Medicine. All experiments involving treatment used randomly assigned mice.
Six-week-old female CD45.1+ mice were lethally irradiated with 950 rads to generate recipient mice. A total of 2 × 106 fetal liver cells obtained from Senp1+/+ or Senp1−/− embryo (E13.5) were injected intravenously into randomly assigned recipient mice. Six weeks after reconstitution, bone marrow (BM) engraftment was checked by staining of peripheral blood cells with CD45.1 and CD45.2 antibodies and subsequent flow cytometry (FCM) analysis.
To establish subcutaneous tumors, 105 B16F10 (murine melanoma cells) or 2 × 105 LLC (murine Lewis lung cancer cells) were injected subcutaneously into the flank of mice. Tumor growth was monitored every 2 days with a caliper, and tumor volumes were calculated using the formula (w2xl)/2.
For spontaneous metastasis, primary subcutaneous tumors were resected 3 weeks after inoculation under general anesthesia and mice were kept for another 2 weeks.
For experimental metastasis, 2 × 105 B16F10 or LLC were injected intravenously into recipient mice and kept for 4 weeks.
For MDSC/tumor cells admix assay, 105 splenic MDSC isolated from tumor-bearing mice were mixed with 105 B16F10 and injected subcutaneously into recipient mice.
For adoptive transfer assay, 106 splenic MDSC isolated from tumor-bearing mice were injected intravenously into mice. Three hours later, 2 × 105 LLC were injected intravenously into mice. Another 2 rounds of MDSC transfer on day 1 and 6 after tumor cells inoculation were applied.
For inhibitor treatment, JSI-124 (1 mg/kg, C4493; Sigma-Aldrich; every 2 days for 3 weeks until the mice were sacrificed) was intraperitoneally injected 7 days after tumor intravenous inoculation of recipient mice.
Cells were isolated and suspended in cold PBS with 2% heat-inactivated bovine serum and 2 mmol/L EDTA. The antibodies for CD45.2 (104), CD11b (M1/70), Ly-6G (1A8), Ly-6C (AL-21), CD11c (N418), CD4 (GK1.5), and Ki-67(B56) were purchased from BD Biosciences, lineage cocktail (17A2/RB6-8C5/RA3-6B2/Ter-119/M1/70), c-Kit (2B8), Sca1 (E13-161.7), FcγR (93), CD34 (RAM34), CD45.1 (A20), Gr-1 (RB6-8C5), F4/80 (BM8), and p-STAT3(Y705) (13A3-1) were purchased from Biolegend, SENP1 (ab108981) were purchased from Abcam. All preparations were preincubated with CD16/32 to reduce unspecific Fc receptor binding. For intracellular p-STAT3 staining, cells were fixed and permeabilized using the True-Phos Perm Buffer according to the manufacturer's instructions (Biolegend). For other intracellular staining, cells were fixed using the Cytofix/Cytoperm Staining Kit according to the manufacturer's instructions (BD Biosciences). For apoptosis analysis, cells were stained with surface marker followed by Annexin V staining in binding buffer (Biolegend). Cells were analyzed using flow cytometer (FACSVerse; BD Biosciences). The data were analyzed using Flowjo software (TreeStar).
Cell isolation and culture
BM cells were isolated by flushing the bones. Splenocytes were prepared by pressing the spleen against a 70-μm nylon mesh. To obtain single cell from tumors, tissues were minced and subjected to 1 hour of enzymatic digestion using 0.5 mg/mL collagenase IV (Sigma-Aldrich) and 0.01 mg/mL DNase I (Sigma-Aldrich) in RPMI1640 supplemented with 2% FBS at 37°C. Single cell suspensions from lungs and livers were obtained as described previously (17). Single cell suspensions were passed through a 70-μm mesh followed by erythrocytes removal using ammonium chloride lysis buffer.
The splenocytes of tumor-bearing mice (3 weeks after tumor cells inoculation) were used to isolate splenic MDSC. MDSC were labeled with Gr-1 and isolated by microbeads and MS column according to the manufacturer's instructions (Miltenyi). For CD11b+Gr-1− cells sorting, flow through in Gr-1 positive section labeled with Biotin-CD11b and selected by Biotin Positive Selection Kit (StemCell Technologies). Hematopoietic progenitor cells (HPC) were isolated from BM cells using the Mouse Hematopoietic Progenitor Cell Isolation Kit (StemCell Technologies). T cells were isolated from splenocytes using the Mouse T-Cell Positive Selection Kit (StemCell Technologies). Intratumor cells were flow sorted by high-speed sorter (FACSAria; BD Biosciences). All the purity and cell viability of sorting cells is over 90%. Cells were cultured in RPMI1640 supplemented with 10% FBS, 25 mmol/L HEPES, 50 μmol/L β-mercaptoethanol, and 1% antibiotics (Invitrogen).
To obtain tumor explant supernatant (TES), single cells from subcutaneous tumors were seeded at 1 × 107/mL and incubated overnight. The supernatants were filtered to remove debris.
The HEK-293T, LLC, and B16F10 were obtained from the ATCC and cultured in DMEM (Gibco) plus 10% FBS (Gibco) supplemented with 1% penicillin–streptomycin (Gibco) at 37°C and 5% CO2. All cells were routinely validated for lack of Mycoplasma infection with LookOut Mycoplasma PCR Detection Kit (Sigma) and used within 10 passages.
T-cell suppression assay
A total of 2 × 105 splenic CD4+ T cells were labeled with 2 μmol/L CFSE (Thermo) at 37°C for 10 minutes and washed with prewarmed culture medium. CD4+ T cells were stimulated with precoated CD3 antibody (BD Biosciences) and 2 μg/mL CD28 (BD Biosciences) in U bottom 96-well plates for 72 hours. MDSC were cocultured with T cells at different ratios (1:1, 1:2, and 1:4). T-cell proliferation was analyzed by CFSE dilution. The supernatant were collected for assessing IFNγ by ELISA (Invitrogen).The data were represented as the percentage of proliferation of stimulated CD4+ CFSE+ T cells.
In vitro MDSC generation
For HPC-induced MDSC, HPC were cultured in 20 ng/mL recombinant GM-CSF (Peroptech) in culture medium or with 20% TES for 5 days. For BM-derived MDSC, BM cells were cultured with 40 ng/mL GM-CSF and 40 ng/mL IL6 (Peroptech) as described previously (5). 0.05 U/mL sialidase (10269611001; Roche), 10 μmol/L PTP1B inhibitor (539741; Merck), 50 μmol/L S3I-201(S1155; Selleck), or concentrated lentivirus were added during the culture period.
CD45 tyrosine phosphatase activity
Cells were washed with cold Tris-buffered saline (TBS) 3 times and lysed in lysis buffer (50 mmol/L HEPES, pH 7.4, containing 0.5% Triton X-100, 10% glycerol, and protease inhibitors) on ice for 30 minutes. The clarified supernatants were incubated with 0.2 mmol/L phosphor-tyrosine peptide corresponding to the sequence of the negative regulatory site of pp60c-src (T529; Abcam) at 37°C for 1 hour. The phosphate released was quantified using the Phosphate Assay Kit (Abcam). The activity was corrected by protein concentration and reactions without phosphor-tyrosine peptide.
Plasmids construction and mutagenesis
CD45 CDS was amplified from cDNA of murine BM cells and the intracellular domain (A.A 554-1293 corresponding to NP_001104786.2) was cloned into an expression vector. The fidelity of sequence was confirmed by sequencing. For site-directed mutagenesis, the mutant form was generated using the Mutagenesis Kit according to the manufacturer's instructions (Takara). To infect MDSC, CD45 was cloned into pCDH-copGFP vector (System Bioscience).
Immunoprecipitation and Western blot
Cells were washed with cold PBS twice and lysed with lysis buffer containing 50 mmol/L Tris-Cl pH7.4, 150 mmol/L NaCl, 1% Triton X-100, 2 mmol/L EDTA, 10 mmol/L NaF, 1 mmol/L PMSF, and protease inhibitors cocktail (Roche) on ice for 30 minutes. The lysates were pelleted at 14,000 × g for 12 minutes; the supernatants were diluted with sample buffer as loading control or subjected to immunoprecipitated. For immunoprecipitation, the clarified supernatants were incubated with unimmunized IgG or rabbit anti-SUMO1 antibody overnight and then 100 μg of magnetic protein A/G beads at room temperature for 1 hour (Thermo Fisher Scientific). The beads were washed 3 times with PBS-T and eluted with sample buffer at room temperature, then the supernatants were subjected to Western blot analysis.
For Ni-NTA pulldown, cells were lysed with 6M guanidine-HCl lysis buffer and His-SUMO1 conjugated proteins were precipitated with Ni-NTA agarose (Clontech). The beads were washed sequentially with washing buffer containing 6M guanidine-HCl or 8M urea. After extensive washing, the enriched proteins were eluted with elution buffer containing 0.2M imidazole and subjected to Western blot analysis.
For Western blot analysis, equal proteins were loaded on SDS-PAGE (6% or 10%) and transferred onto PVDF membranes. Membranes were blocked with 5% skimmed milk and then blotted with antibodies for STAT3 (9139; CST), p-STAT3 (9145; CST), SENP1 (ab108981; Abcam), actin (A1978; Sigma-Aldrich), Ha (sc-7392; Santa Cruz Biotechnology), Flag (F1804; Sigma-Aldrich), and SUMO1 (4930; CST). Protein blots were visualized using Enhanced Chemiluminescence (Millipore).
RNA was extracted from cells using TRIzol according to the manufacturer's instructions (Invitrogen). The cDNA was synthesized using random hexamers and the cDNA Synthesis Kit according to the manufacturer's instructions (Takara). The qPCR was performed in triplicates using 10 μL of SYBR Master Mixture (Applied Biosystems). Specific primers were as follows: Gp91phox, 5′- TGTGGTTGGGGCTGAATGTC-3′ and 5′- CTGAGAAAGGAGAGCAGATTTCG-3′; Arginase1, 5′- AAGAAAAGGCCGATTCACCT-3′ and 5′- CACCTCCTCTGCTGTCTTCC-3′; S100a8, 5′- AAATCACCATGCCCTCTACAAG-3′ and 5′- CCCACTTTTATCACCATCGCAA-3′; Actb, 5′- ATACTCTAGGAAGGAAGGACAC-3′ and 5′- TCCATGATGTCATTTATGAGGGC-3′. Relative quantification of gene expression was calculated using the ΔΔCt method. Expression levels of the genes were normalized by actin.
Human stage III colon cancer samples were acquired from the Renji Hospital, Shanghai (affiliated to Shanghai Jiao Tong University School of Medicine). Written informed consent was obtained from all patients. All procedures related to human samples were approved by the Ethics Committee of Shanghai Jiao Tong University School of Medicine. Tumor tissues were fixed with 4% PFA and embedded by OCT. Cryosections of 8 μm were blocked with 3% BSA, antibodies for CD45 (sc-59071, Santa Cruzs), CD11b (550374, BD Pharmingen), CD14 (ab63319, Abcam), CD163 (ab17051, Abcam) and SENP1 (ab108981, Abcam) were then incubated overnight at 4°C. The secondary antibodies conjugated with Alexa Fluor 488 or 647 (Life Technologies) were incubated for 60 min at 37°C. The nucleus was stained using DAPI (Life Technologies). Images were obtained using a Nikon Eclipse 80i microscope.
Unless otherwise stated, statistical analyses were performed using 2-tailed Student t test and GraphPad Prism 7 software (GraphPad Software). All the data are presented as mean ± SEM unless otherwise mentioned and a P value less than 0.05 was considered significant.
SENP1 expresses at low level in MDSC
As we previously showed a role of SENP1 in macrophage activation responding to inflammatory stimuli (14), we reasoned that SENP1 might play a role in macrophage-involved inflammation in tumorigenesis. Therefore, we checked SENP1 expression in macrophages in human colon cancer samples. By costaining SENP1 and leukocytes marker CD45, we detected a large population of SENP1 expression in CD45+ leukocytes in tumor area (Supplementary Fig. S1). As BM-derived myeloid cells progressively accumulated in tumors, we thus analyzed SENP1 expression in CD11b+ (a marker for myeloid cells and natural killer cells), CD14+ (a marker for Mo-MDSC and monocytes), or CD163+ (a marker for TAM) positive cells in these samples by immunofluorescence staining. These staining results showed that SENP1 was highly expressed in CD11b+ cells, and partially expressed in CD163+ TAM, but rarely expressed in CD14+ cells. These date indicate that SENP1 may express with different level in the different kind of myeloid-derived cells.
To validate the expression pattern of SENP1 in mouse tumor models, we subcutaneously inoculated LLC or B16F10 tumor cells into the flank of mice. We observed 2 peaks of SENP1 expression in CD45+ leukocytes from tumor tissues by FACS analysis. SENP1 expression in MDSC (CD45+CD11b+Gr-1+) was much less than that in others myeloid cells (CD45+CD11b+Gr-1−). We also detected a lower expression of SENP1 in MDSC in B16F10 tumor model (Fig. 1A). However, SENP1 expressed at high level in TAM (CD11b+F4/80+) and DC (CD11b+CD11c+) as compared with other CD11b+CD11c− (or F4/80−) cells (Fig. 1B). We also confirmed the SENP1 expression pattern in the sorted intratumor myeloid cells from LLC tumors by Western blot analysis (Fig. 1C).
Additionally, we analyzed SENP1 expression in MDSC in mice with or without tumors. We isolated splenic MDSC from mice at different time after inoculating tumor cells. SENP1 in MDSC was shown decrease after tumor load (Fig. 1D). Interestingly, SENP1 expression was much more decrease in splenic MDSC with lung metastasis than that without metastasis, suggesting that reducing SENP1 expression in MDSC is not only related to tumor but also to tumor progression (Fig. 1E). We used HPC differentiation into MDSC in vitro to show that tumor cultural supernatant could reduce SENP1 expression in MDSC development (Fig. 1F–G). The decreased SENP1 also be recapitulated in GM-CSF/IL6-induced MDSC, noteworthy a simultaneous accumulated SUMO-conjugated proteins during the culture (Fig. 1H and I).
Reducing SENP1 expression promotes MDSC expansion
Above data demonstrated that SENP1 expressed at a low level in MDSC but not in other myeloid cells, suggesting that reducing SENP1 expression in myeloid cells might promote MDSC development. As Senp1 gene knockout mice died at embryonic day 14 (17), we generated a Senp1−/− fetal liver transplantation model to test the role of SENP1 in myeloid cell development. We isolated Senp1+/+ or Senp1−/− fetal liver cells as donor following transplanting into lethal-radiated recipient mice to reconstitute BM (designed as Senp1+/+-BMT or Senp1−/−-BMT). Both mice did not show any gross defects and had similar numbers of BM cells and splenocytes (Fig. 2A). The lymphocytes in spleen showed a minor decrease in Senp1−/−-BMT mice (Fig. 2B). However, the percentage of CD11b+ myeloid cells were significantly elevated in Senp1−/−-BMT mice compared with Senp1+/+-BMT controls (Fig. 2C). We further analyzed the population of major myeloid cells in CD11b+ gated cells. Both CD11b+F4/80+ cells and CD11b+Gr-1+ cells but not CD11b+CD11c+ cells increased in Senp1−/−-BMT mice (Fig. 2D). Because CD11b+Gr-1+ cells in tumor-free mice show similar features as MDSC in tumor-bearing mice, we catalogued these cells as MDSC. These data suggest that reducing SENP1 expression would lead to myeloid disorder characterized by robust expansion of CD11b+Gr-1+ myeloid cells.
MDSC are rich in BM, liver, and lung (18, 19). We found that Senp1−/−-BMT mice had much more MDSC in the BM, lung, or liver as compared with Senp1+/+-BMT mice (Fig. 2E). We also analyzed that the 2 major subsets PMN-MDSC and Mo-MDSC in tumor-free mice. PMN-MDSC were more in BM, spleen, liver, and lung of Senp1−/−-BMT mice than in that of Senp1+/+-BMT mice (Fig. 2F). Although the similar changes as PMN-MDSC, however, Mo-MDSC were much less than PMN-MDSC in both mice (Fig. 2F). These results demonstrate that reducing SENP1 expression would increase MDSC in the BM and other organs.
Reducing SENP1 expression promotes tumorigenesis
Given that MDSC suppress tumor immunology, we expected that SENP1 deficiency in myeloid cells would promote tumorigenesis. To test it, we subcutaneously injected LLC or B16F10 tumor cells into Senp1+/+-BMT or Senp1−/−-BMT mice. Both tumors in Senp1−/−-BMT mice grew faster than those in Senp1+/+-BMT mice (Fig. 3A and B). Furthermore, we evaluated the effect of myeloid cells SENP1 deficiency on tumor metastasis. To do it, LLC tumor cells were first planted in subcutaneous of Senp1+/+-BMT or Senp1−/−-BMT mice. After resecting the primary tumors, the metastatic tumors appeared in the lung or liver of the same mice later (Fig. 3C). We applied intravenous inoculation and observed that LLC metastatic tumors in the livers were more in Senp1−/−-BMT mice than that in Senp1+/+-BMT mice (Supplementary Fig. S2), and that B16F10 metastatic tumors were more in both lung and liver of Senp1−/−-BMT mice than in that of Senp1+/+-BMT mice (Fig. 3D).
Meanwhile, we analyzed MDSC in these tumor-bearing mouse models. We detected more MDSC in BM, spleen, and tumor tissues in the LLC or B16F10 tumor-bearing Senp1−/−-BMT mice as compared with Senp1+/+-BMT control mice (Fig. 3E–G). Although tumor challenge–induced MDSC expansion was observed in both mouse models, PMN-MDSC and Mo-MDSC were more in Senp1−/−-BMT mice than in Senp1+/+-BMT mice (Supplementary Fig. S3A and S3B). Interestingly, MDSC but not TAM or DC from tumor tissue were more in Senp1−/−-BMT mice than in Senp1+/+-BMT mice (Supplementary Fig. S3C). We observed more MDSC in liver and lung in Senp1−/−-BMT mice when the primary tumor was similar with Senp1+/+-BMT control mice (Supplementary Fig. S4A and S4B). We also showed more MDSC but not TAM accumulation in tumors (Supplementary Fig. S4C and S4D).
Reducing SENP1 expression promotes MDSC suppressive activity
We next determined whether SENP1 deficiency would alter MDSC suppressive activity. We first evaluated MDSC suppressive activity by coincubating splenic MDSC from Senp1+/+-BMT or Senp1−/−-BMT mice with T cells. In the presence of CD3/CD28, CD4+ T cells undergo robust proliferation, and addition of MDSC would delay the proliferation. Senp1−/− MDSC demonstrated a stronger suppression on CD3/CD28-induced CD4+ T-cell proliferation than Senp1+/+ MDSC did (Fig. 4A). We also detected that Senp1−/− MDSC from tumors had more suppressive activity on T cells to produce IFNγ than Senp1+/+ MDSC did (Supplementary Fig. S4E and S4F). As the expressions of Arginase1, Nox2, and S100a8 are associated with MDSC suppressive activity, we further compared their expression in splenic MDSC isolated from Senp1+/+-BMT or Senp1−/−-BMT mice. As shown in Fig. 4B, the expressions of Arginase1, Nox2 (Gp91phox), and S100a8 increased much more in Senp1−/− MDSC than in Senp1+/+ MDSC. We also showed the elevated level of the key immune suppressive cytokine IL10 in Senp1−/−-BMT serum as compared with that in Senp1+/+-BMT controls but not in naïve mice (Fig. 4C).
We then analyzed MDSC functions on tumor growth by carrying on admix assay, in which the same number of splenic MDSC and tumor cells were mixed following subcutaneous injection into mice. As shown in Fig. 4D, Senp1−/− MDSC markedly promoted the tumor growth as compared with Senp1+/+ MDSC did. We noticed that the MDSC amounts in the BM were similar in these mice (Supplementary Fig. S4G). However, we detected more MDSC but similar TAM in tumors of MDSC-mixed group as compared with the saline group (Supplementary Fig. S4G and S4H). We further analyzed MDSC functions in tumor metastasis by using MDSC adoptive transfer tumor models. After MDSC transfer, we inoculated LLC cells intravenously into the mice. We found that Senp1−/− MDSC markedly increased tumors burdens in livers as compared with the saline control or Senp1+/+ MDSC groups (Fig. 4E and F). These data suggest that reducing SENP1 expression not only increases MDSC expansion but also promotes MDSC suppressive activity.
Reducing SENP1 expression enhances STAT3 activity in MDSC
SENP1 deficiency–related changes of MDSC could be resulted from the alteration in the differentiation of hematopoietic precursors, or in the survival or proliferation of MDSC. We first analyzed the hematopoietic progenitor cells in BM and found that both of Senp1+/+-BMT and Senp1−/−-BMT mice developed the similar percentages of Lin−c-Kit+ cells (LK), common myeloid progenitors (CMP), granulocyte/macrophage progenitors (GMP), and megakaryocyte-erythroid progenitors (MEP; Supplementary Fig. S5A and S5B). Consistently, Senp1−/− hematopoietic precursor cells (HPC) also produced similar percentages of myeloid cells as Senp1+/+-BMT mice did. On the other hand, in presence of TES Senp1−/− HPC produced more MDSC (Supplementary Fig. S5C and S5D), especially more PMN-MDSC but not Mo-MDSC (Supplementary Fig. S5E and S5F), than Senp1+/+ HPC did. Furthermore, we generated similar macrophages and DC using BM cells (Supplementary Fig. S5G and S5H). These results suggest that reducing SENP1 expression in myeloid cells does not affect myeloid precursor cell development.
MDSC are prone to apoptosis and alleviated by tumor-derived factors (20). We found the prosurvival capacity but not proliferation was higher in Senp1−/− MDSC (Supplementary Fig. S5I, S5J, and S5K). STAT3 is a transcription factor to regulate MDSC apoptosis and expansion (21). We previously found that SENP1 deficiency promotes STAT3 activity in macrophages. We thus speculated that SENP1 deficiency activates STAT3, leading to MDSC expansion and activation. Indeed, we detected higher phosphorylation of STAT3 in Senp1−/− BM cells (Supplementary Fig. S6A). We further purified Gr-1+ and Gr-1− cells from BM cells of Senp1+/+-BMT or Senp1−/−-BMT mice to analyze STAT3 activation. As expected, Senp1−/− Gr-1+ cells but not Gr-1− cells exhibited higher phosphorylated STAT3 than Senp1+/+ control cells did (Fig. 5A). We also carried out an analysis on intracellular p-STAT3 staining by flow cytometry and confirmed the higher phosphorylated-STAT3 in Senp1−/− MDSC (Fig. 5B).
To determine whether STAT3 was responsible for the MDSC expansion and function in Senp1−/−-BMT mice, we used a selective STAT3 inhibitor S3I-201 to treat GM-CSF/IL6-converted MDSC. We found that inhibition of STAT3 could completely diminish the difference of MDSC but not DC population between Senp1+/+-BMT and Senp1−/−-BMT mice (Fig. 5C and D). S3I-201 abolished suppressive functions of Senp1−/− GM-CSF/IL6-converted MDSC (Fig. 5E and F). We further test whether inhibition of STAT3 could reverse tumor progression observed in Senp1−/−-BMT tumor model. To do it, we had treated Senp1−/−-BMT mice with STAT3 inhibitor JSI-124 via intraperitoneally after LLC injection via intravenously. We found that inhibition of STAT3 markedly decreased LLC tumor burdens in the liver of Senp1−/−-BMT mice as compared with DMSO treated mice (Fig. 5G). Taken together, these data reveal that reducing SENP1 expression in myeloid cells induces STAT3 activation, leading to MDSC expansion and activation.
SENP1 deSUMOylates CD45 and regulates CD45–STAT3 signaling in MDSC
We previously showed that SENP1 deSUMOylates PTP1B to reduce STAT3 phosphorylation in macrophages. To test whether the same regulation present in MDSC of Senp1−/−-BMT mice, we used PTP1B inhibitor (CAS 765317-72-4) to determine the effect of PTP1B on STAT3 activation in MDSC. We found that PTP1B inhibitor had minor effects on STAT3 activation as well as the population of MDSC and DC in HPC-generated cells (Supplementary Fig. S6B, S6C, and S6D). These data suggest that SENP1 deficiency-induced STAT3 activation and expansion is independent of PTP1B in MDSC.
It has been reported that CD45 as a phosphatase de-phosphorylates STAT3 (22). Therefore, we determined whether CD45 is response for SENP1 inhibition of STAT3 activation in MDSC cells. We first used CD45 inhibitor sialidase and found this inhibitor but not PTP1B inhibitor could significantly increase STAT3 phosphorylation in HPC-derived MDSC (Supplementary Fig. S6B). We also observed that inhibition of CD45 would markedly increase MDSC population in Senp1+/+ BM cells, which was up to the similar level as SENP1 deficiency–induced MDSC expansion, indicating a crucial role of CD45 in MDSC population (Fig. 5D and E). These data suggest that SENP1 deficiency reduces CD45 activity, which activates STAT3 in Senp1−/− MDSC. Moreover, we demonstrated that CD45 phosphatase activity was lower in Senp1−/− MDSC than in Senp1+/+ MDSC (Fig. 6A). These results indicate that CD45 phosphatase might be targeted by SENP1 in regulation of STAT3 activation in MDSC.
Based on the above observations, we hypothesized that CD45 could be SUMOylated and be regulated by SENP1 through de-SUMOylation in MDSC. To test it, we first showed that SENP1 bound to sCD45 (short CD45 containing the catalytic domain D1 and noncatalytic domain D2; Fig. 6B). We further demonstrated SUMO1 conjugation of sCD45 indicated by the upshifted band in the cells with cotransfection of sCD45 and SUMO1. Importantly, we determined that SENP1 but not SENP1 mutant could dismiss this SUMO1-conjugated band (Fig. 6C). We also showed the endogenous SUMOylation of CD45 in BM cells and showed the accumulated SUMO1-conjugated CD45 in the peritoneal macrophage obtained from Senp1−/−-BMT mice as compared with Senp1+/+-BMT controls (Fig. 6D and E). We predicated CD45 at lysine 867 and lysine 977 as 2 potential SUMOylation sites. The analysis on the mutation of the 2 lysine residues to arginine on CD45 (CD45 K867R or CD45 K977R) showed that CD45K867R but not CD45K977R was a sole SUMOylation site (Fig. 6F).
We further determined whether CD45 SUMOylation regulates MDSC. BM cells were infected by the lentivirus of CD45 or CD45K867R. We compared their ability to develop into MDSC by GM-CSF and IL6. Overexpression of CD45 inhibited MDSC but not DC development. However, CD45K867R showed more inhibition on MDSC development than CD45 WT did (Fig. 6G). These data indicate that SENP1 deSUMOylates CD45 and promotes its phosphatase activity toward STAT3 in myeloid progenitor cells, leading MDSC expansion and activation.
In this study, we find that SENP1 plays a vital role in controlling MDSC development and function. SENP1 exerts its regulation role by deconjugation of SUMOylated CD45. CD45 is identified as a SUMOylation substrate for the first time. SENP1 deficiency promotes CD45 SUMOylation and lead to decrease in phosphatase activity against STAT3, which is a crucial regulator for MDSC expansion and function. Furthermore, Senp1−/− MDSC demonstrate to accelerate tumor progression in mouse tumor models.
MDSC are heterogeneous immature myeloid cells. In tumor, MDSC accumulation gains immunosuppression to favor tumor progression. MDSC, DC, and TAM are major myeloid cells in tumor. It has reported that tumor-induced abnormal differentiation of DC causes MDSC expansion (6). Reports also show that MDSC can convert to TAM in tumor tissue (11). MDSC have been shown as precursors of CD103+ DC with the potent antigen-presentation ability (23). PMN-MDSC and TAM are balanced via carcinoma-associated fibroblasts (CAF) depended on the chemokine network (24). We show that SENP1 levels in subsets of tumor myeloid cells are different, which encouraged us to seek whether manipulation of SENP1 level in myeloid cells would affect tumor progression. By using different tumor models and adoptive transfer experiments, we show that SENP1 deficiency in myeloid cells significantly promotes tumor progression via MDSC expansion and increased immunosuppression. By analyzing SENP1 deficiency mice and in vitro myeloid cells differentiation, we find that SENP1 deficiency in myeloid cells resulted in more MDSC (especially PMN-MDSC), less DC. However, the alteration of MDSC and DC balance does not result from abnormal hematopoietic progress in Senp1−/−-BMT mice (Supplementary Fig. S5F). We still do not know how SENP1 regulates this balance.
Many transcriptional factors and inflammatory mediators involving the regulation of myeloid cell differentiation process (25, 26). STAT3 is crucial for MDSC in steady status and pathogenic processes (6, 27, 28). In our previous report, we elucidated PTP1B as a major phosphatase for STAT3 during macrophage polarization. SENP1 regulates macrophage polarization through PTP1B deSUMOylation (14). PTP1B has been also reported to regulate MDSC expansion and activity in the colitis model (10). In this study, we identify CD45 as a SENP1 target to mediate the regulation of STAT3 activation in SENP1 deficiency MDSC. It also is interesting to figure out how to choose phosphatase for targeting STAT3 in different cell type or cell location. In addition, our data reveal that SUMOylation regulates CD45 function in MDSC expansion and activity, but not in MDSC-TAM transition in tumor (11).
CD45 is wildly used as a leukocytes marker due to its abundance on most leukocytes (22, 29). As a transmembrane phosphatase, its activity can be negatively regulated by homodimerization mediated mainly by extracellular sialylation (30). Besides, CD45 can be regulated by phosphorylation and govern cell motility (31). Recent advances imply an “inhibitory wedge” model, but whether additional regulatory exist is still unknown (32). In this work, we revealed a previously unappreciated mechanism by which SUMOylation favored the formation of homodimerization and blunted the phosphatase activity. The underlying molecular mechanism remains to be investigated, but a potential SUMO interaction motif (SIM) located at the second PTPase domain of CD45 (A.A 1182-1185 corresponding to NP_001104786.2) may facilitate the dimer formation (33). The evolutionally conserved SUMOylation site located on the molecular surface together with SIM may provide molecular bridges to stabilize the dimer-mediated autorepression structure of CD45 (34). Additional question will be whether SENP1–CD45–STAT3 signaling exists in other CD45+ leukocytes.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: Y. Zuo, J. Cheng
Development of methodology: X. Huang, Y. Zuo
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Huang, Y. Zuo, X. Wang, H. Tan, Q. Fan, B. Dong
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Huang, Y. Zuo, H. Tan, W. Xue, G.-Q. Chen, J. Cheng
Writing, review, and/or revision of the manuscript: X. Huang, Y. Zuo, X. Wu, J. Cheng
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zuo, J. Cheng
Study supervision: Y. Zuo, J. Cheng
We sincerely thank Lei Shen (Shanghai Institute of Immunology, Shanghai Jiao Tong University School of Medicine, Shanghai, China) for technical support. We also appreciate Shengdian Wang (Institute of Biophysics, Chinese Academy of Sciences, Beijing, China) for his support on study design. This work was supported by National Natural Science Foundation of China (91229202 and 81430069 to J. Cheng; 81672880 to Y. Zuo; 81721004 to G. Chen), Natural Science Foundation of Shanghai (14ZR1423500 to Y. Zuo), Shanghai Committee of Science and Technology (15140904300 to Y. Zuo), Shanghai Municipal Commission of Health, and Family Planning (20174Y0092 to Q.J. Fan), and Shanghai Municipal Education Commmission (2017-01-07-00-01-E00050 to J. Cheng).
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.