Lnc-C/EBPβ Negatively Regulates the Suppressive Function of Myeloid-Derived Suppressor Cells

Myeloid-derived suppressor cells (MDSC) have emerged as regulators of immune responses in cancer and other pathologic conditions (1). In mice, MDSCs are identified as CD11b⁺Gr1⁺ cells. Gr1⁺ subsets may be more accurately identified based on Ly6C and Ly6G markers as CD11b⁺Ly6C^hiLy6G⁻monocytic MDSCs (M-MDSC) and CD11b⁺Ly6C^low/negLy6G⁺ polymorphonuclear MDSCs (PMN-MDSC; refs. 2, 3). Human MDSCs, including monocytic (CD14⁺) MDSCs and PMN (CD15⁺) MDSCs, are described as lineage-negative cells that coexpress CD11b and CD33 but lack HLA-DR (1). These MDSCs play a role in cancer progression and other associated diseases by suppressing both innate and adaptive immune responses. They have high levels of arginase-1 (Arg-1), nitric oxide synthase 2 (NOS2), NADPH oxidase 2 (NOX2), and COX2, which may result in the production of nitric oxide (NO) and reactive oxygen species (ROS; refs. 4, 5). NO reacts with multiple cellular compounds to produce many toxic and regulatory factors. ROS, including hydrogen peroxide (H₂O₂), and the hydroxyl radical, damages nucleic acids, proteins, and lipids (6). Arg-1 itself may cause elimination of key nutrition factors needed for T-cell proliferation by depleting the local environment of L-arginine (7), sequestering L-cysteine (8), and/or reducing local tryptophan levels (9, 10).

Our previous studies show that inflammatory or tumor-associated factors may affect development of MDSCs (11) and that the differentiation and function of MDSCs may be regulated by microRNAs and epigenetic modifying factors (12, 13). Studies show that epigenetic modification, such as lncRNAs (length greater than 200 nucleotides), also plays important roles in regulating differentiation and function of immune cells. Multiple lncRNAs have been described in myeloid-derived cells, such as lnc-DC in dendritic cells (14), long intergenic noncoding RNA (lincRNA)-Cox2, lincRNA-EPS, AS-IL-1a in macrophages (15, 16), and lncRNA Morrbid in myeloid cells (17). However, the effects of lncRNAs on the differentiation and function of MDSCs are not well understood. We here identified an lncRNA named lnc-C/EBPβ. We demonstrated that lnc-C/EBPβ regulated various transcripts in MDSCs to control immune-suppressive function and differentiation of MDSCs in inflammatory and tumor environments, which may provide a regulatory mechanism that can control MDSC differentiation and suppressive function.

C57BL/6 mice were purchased from the Beijing Animal Center (Beijing, China) and maintained in a specific pathogen-free facility. B6.129S6-Il-6tm1Kopf (IL6^−/−) and B6.SJL-CD45a(Ly5a; CD45.1⁺) mice were purchased from the Model Animal Research Center of Nanjing University (Nanjing, Jiangsu, China). OTI and OT-II OVA-TCR transgenic mice were provided by Dr. Linrong Lu of Zhejiang University. All animal experiments were carried out in accordance with the Nankai University Guide for the Care and Use of Laboratory Animals.

The peripheral blood and tissue samples from the 20 patients with colorectal adenocarcinoma, which were demonstrated according to pathologic criteria, were obtained after informed consent at People Union Hospital (Tianjin, China). Same-age and sex healthy human peripheral blood control samples were obtained after signed informed consents. The collection and use of all human samples (20 healthy individuals and 20 patients with colorectal adenocarcinoma) were approved by the Institute's Human Ethics Committee of Nankai University and in accordance with the Declaration of Helsinki. The samples were immediately analyzed or stored but not over one month.

Murine melanoma B16, Lewis lung carcinoma (LLC), breast cancer 4T1, human embryonic kidney HEK 293T, human breast cancer MCF-7 and MDA-MB, lymphoma U937, cervical cancer Hela, and human colon HT-29 cell lines were obtained from the American Type Culture Collection from 2013 to 2014. The murine ovarian tumor cell line 1D8 is from Dr. Richard Roden (Johns Hopkins University School of Medicine, a gift in 2010). B16-OVA was provided by Dr. Lieping Chen in 2016. These cell lines were not further authenticated in the past years. They were not contaminated by mycoplasma before and after experiments.

siRNAs were purchased from Riobio. siRNA sequences for lnc-C/EBPβ were listed in Supplementary Table S1. Two shRNA targets were chosen from the target sequences produced by BLOCK-iT RNAi Designer (Invitrogen) and/or by i-Score Designer 38. The shRNA constructs were made using the pGreenPuro shRNA Cloning and Expression Lentivector Kit (System Biosciences Inc.) according to the manual. The control shNC is a luciferase control shRNA from kit. For packaging of lentivirus particles, shRNA lentivector or lnc-C/EBPβ lentivector together with pMD2G and psPAX2 packaging plasmids were cotransfected into 293T cells. The full-length sequences of C/EBPβ LAP/LIP were amplified using PCR methods (primer pairs are described in Supplementary Table S1). The PCR products were directly cloned into the pcDNA3.1/V5-His TOPO TA plasmid (Invitrogen) using T4-conjugating enzymes, which was named C/EBPβ LAP/LIP/ pcDNA3.1/V5. The lentiviruses, siRNAs, or plasmid constructions were used to infect or transfect MDSCs.

Inflammatory factor-associated MDSCs were generated by culturing bone marrow cells (BMC) of C57BL/6 mice in 25 cm² flasks or 6-well plates for 4 days in the presence of 5% FBS RPMI medium containing GM-CSF(40 ng/mL) plus IL6 (40 ng/mL) or GM-CSF (40 ng/mL) plus TNFα (20 ng/mL; ref. 13). To prepare tumor cell supernatant-induced CD11b⁺Gr1⁺ MDSCs in vitro, 5 × 10⁴ 1D8, 4T1, or B16 tumor cells (upper chamber) were cocultured with 2 × 10⁶ BMCs (lower chamber) in a 24-transwell plate in the presence of GM-CSF (40 ng/mL) for 4 days. Human MDSC-like cells were generated according to previously reported methods (18). Human peripheral blood monocytes were isolated by centrifugation using Ficoll-Pague PLUS, and then cultured with human recombinant GM-CSF (40 ng/mL) and human IL6 (40 ng/mL) for 4 days. To prepare the tumor cell supernatant-induced human MDSC-like cells in vitro, 5 × 10⁴ MCF-7, Hela, HT29, or MDA-MB231 tumor cells (upper chamber) were cocultured with 2 × 10⁶ human peripheral monocytes (lower chamber) in a 24-transwell plate in the presence of GM-CSF (40 ng/mL) for 4 days.

BMCs were collected from C57BL/6 mice and cultured in 6-well plate. Human monocytes were collected and isolated from peripheral blood. The cells were infected with the lentiviral supernatants in the presence of 8 μg/mL polybrene (Millipore) by centrifugation and then cultured with 5% FBS RPMI medium for 24 hours. The cells were transfected with lnc-C/EBPβ siRNA, negative control siRNA or pcDNA3.1/lnc-C/EBPβ, and pcDNA3.1 control using Lipofectamine 3000 (Invitrogen) or HiPerFect transfection reagent (siRNA transfection; Qiagen) according to the manufacturer's instructions.

For JAK1 and STAT3 inhibitor-treated MDSCs, MDSCs were incubated with JAK1 inhibitor Filgotinib (20 nmol/L) or STAT3 inhibitor HO-3867 (100 nmol/L) for 24 hours.

Expression of lncRNAs, microRNAs, and coding mRNAs was analyzed by Shanghai OE Biotech., Ltd. Total RNA was extracted from MDSCs after treatment with GM-CSF with or without IL6. RNA was extracted using TRIzol (Life Technologies). Contaminating DNAs were removed using RNeasy spin columns (Qiagen). The quality of isolated RNA samples was evaluated with an Agilent Bioanalyzer 2100 (Agilent technologies) and the purified RNA was quantified using a NanoDrop ND-2000 spectrophotometer (Infinigen Biotech). The Agilent Gene Expression oligo microarrays and miRNA microarrays were analyzed using Agilent Gene Expression oligo microarrays Version 6.5, May 2010, and Agilent miRNA microarrays Version 2.3. The R software (v.2.13.0) platform was applied to analyze the microarray data, and the limma (linear regression model) package was used to statistically analyze differentially expressed genes. The expression levels of mRNAs at each time point were compared with control. Genes having a fold change > 2 or < −2 and an adjusted P < 0.05 were considered as differentially expressed. Microarray GEO accession numbers GSE104718 and GSE 104719. Microarray of coding mRNA in lnc-C/EBPβ knockdown and overexpressing MDSCs were constructed using the Affymetrix GeneChip mouse Genome 430 2.0 array by CapitalBio Technology Co., Ltd. Microarray GEO accession number is GSE104558.

Cells were collected and rinsed twice with ice-cold PBS, incubated with FITC-, PE-, percy5.5-, or APC-labeled antibodies for 30 minutes in PBS with 1% FBS according to our previous method (19). After washed twice, cells were resuspended in PBS and analyzed using a FACScan flow cytometer (BD Biosciences). Dead cells were eliminated through 7-AAD staining. All antibodies used in this study were listed in Supplementary Table S1.

Total RNA was extracted from the cells, tissues, and organs using TRIzol reagent (Invitrogen). First-strand cDNA was generated from total RNA using oligo-dT primers and reverse transcriptase (Invitrogen). The PCR products were visualized on 1.0% (wt/vol) agarose gels. Quantitative real-time PCR (qRT-PCR) was conducted using QuantiTect SYBR Green PCR Master Mix (Qiagen) and specific primers in an ABI Prism 7000 analyzer (Applied Biosystems). GAPDH mRNA expression was detected in each experimental sample as an endogenous control. The fold changes were calculated using the ΔΔCt method according to the manufacturer's instructions (Applied Biosystems). All the reactions were run in triplicate. All primers used in this study were listed in Supplementary Table S1.

Cells were harvested at the indicated times and rinsed twice with ice-cold PBS. Cell extracts were prepared with lysis buffer and centrifuged at 13,000 × g for 10 minutes at 4°C. Protein samples were electrophored using 12% polyacrylamide gels and transferred to PVDF membranes. After the membranes were blocked with 5% skim milk powder for 1 hour at room temperature, they were incubated with first antibody in TBST overnight at 4°C. Secondary antibodies with horseradish peroxidase (HRP; 1:5,000) were labeled according to our method previously reported (19). The signals were checked by autoradiography film when HRP substrate was added to the membranes. The primary and secondary antibodies used in this study were listed in Supplementary Table S1.

MDSC immunosuppressive function was analyzed according to our previously reported methods (13). For arginase activity, cells were lysed for 30 minutes with 100 μL of 0.1% Triton X-100 at 4°C. Following lysis, 100 μL of 25 mmol/L Tris-HCl and 10 μL of 10 mmol/L MnCl₂ were added and the mixture was heated for 10 minutes at 56°C. Subsequently, the lysates were incubated with 100 μL of 0.5M L-arginine (pH 9.7) at 37°C for 120 minutes. The reaction was stopped with 900 μL of H₂SO₄ (96%)/H3PO4 (85%)/H₂O (1:3:7). Urea concentration was measured by absorbance at 540 nm after addition of 40 μL of 9% α-isonitrosopropiophenone, followed by heating at 95°C for 30 minutes. A standard curve was generated by serial dilution of 120 mg/mL urea. Arginase activity (unit) was defined by the amount enzyme that catalyzes the formation of 1 μg of urea per minute. For nitric oxide production, the total nitric oxide in the cell lysate was measured using the Nitrate/Nitrite Assay Kit (Kamiya). Equal volumes of cell lysate (60 μL), 2 mmol/L NADPH (5 μL), FAD(10 μL) and Nitrate Reductase (5 μL) were incubated at 37°C for 30 minutes, followed by addition of 10 μL of LDH buffer and LDH solution. After incubation for 30 minutes at 37°C. Then, addition of 50 μL of Griess Reagent I and Griess Reagent II, incubated at room temperature for 10 minutes, and absorbance at 540 nm was measured. Nitrite concentrations were quantified by comparing the absorbance values to a standard curve generated by serial dilution of 100 μmol/L sodium nitrite. For H₂O₂ production, production of H₂O₂ was evaluated using Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (Invitrogen). Briefly, 1 × 10⁴ cells were resuspended in Krebs-Ringer phosphate (contain 50 μmol/L Amplex Red reagent and 0.1 U/mL HRP). After addition of PMA (phorbolmyristate acetate, 30 ng/mL), the absorbance at 560 nm was measured using a microplate reader at 37°C. Absorbance values for the test samples were normalized to a standard curve generated by serial dilutions of 10 mmol/L H₂O₂. For ROS detection, oxidation-sensitive dye DCFDA was used to measure ROS production by MDSCs. Cells were incubated at 37°C in RPMI medium in the presence of 2.5 μmol/L DCFDA for 30 minutes. For PMA-induced activation, cells were simultaneously cultured with DCFDA and 30 ng/mL PMA and then flow-cytometric analysis was then performed.

The First Choice RNA-ligation mediated RACE (RLM-RACE) kit (Ambion) was used to obtain full sequence of lnc-C/EBPβ. For the 5′ end, total RNA from mouse MDSCs was treated with calf intestinal phosphatase to remove 5′ phosphate from noncapped transcripts, resulting in 5′ capped transcripts and RNAs with 5′ hydroxyl ends. Then, the 5′ 7-methylguansine cap structure was removed by tobacco acid pyrophosphatase, resulting in 5′monophosphate transcript exclusively from intact 5′ transcripts. An RNA adaptor with 5′ and 3′ hydroxyl groups was then ligated to the 5′ monophosphate RNAs. For the 3′RACE, cDNA was synthesized using a 3′RACE adapter. RT-PCR using a lnc-C/EBPβ specific primer and a primer binding to the ligated RNA adaptor was performed to amplify the ligated lnc-C/EBPβ followed by TOPO TA cloning and sequencing to determine the 5′ and 3′ end sequences of the lncRNA.

Lnc-C/EBPβ levels were measured by northern blot based on the reported method with modification (20). Briefly, Total RNAs harvested with TRIzol reagent were run on 1% agarose-formaldehyde gel. RNA was transferred to a Hybond nylon membrane using the Trans-Blot SD semidry electrophorectic transfer (Bio-Rad). Digoxin-labeled antisense lnc-C/EBPβ was made using T7 RNA polymerase by IVT with the Digoxin labeling Kit (Roche). The membrane was prehybridized for 1 hour at 42°C and incubated with the probe overnight at the same temperature. After washing, the membrane was blocked and incubated with digoxin antibody conjugated with HRP. The signals were checked by autoradiography film when HRP substrate was added to the membranes.

ChIP-PCR was performed using EZ-ChIP Chromatin Immunoprecipitation Kit (Millipore) according to the reported method (21). MDSCs were crosslinked with 1% paraformaldehyde and incubated with rotation at room temperature. Crosslinking was stopped after 10 minutes with glycine to a final concentration of 0.125 mol/L and incubated 5 minutes further with rotation. Cells were washed with ice-cold PBS (containing 1% PMSF) 3 times and immediately resuspended in SDS lysis buffer (containing 1% PMSF). Cell lysates were sonicated for 40 cycles of 30 seconds ON and 30 seconds OFF in 10 cycle increments using a Biorupter (Diagenode) on ice. After pelleting debris, protein G agarose was added and incubated for 1 hour at 4°C with rotation for preclearing. For immunoprecipitation, precleared cell lysates were incubated with the indicated antibodies (1 μg per 2 million cells) overnight with the rotation at 4°C and protein G agarose was added for the final 2 hours of incubation. Beads were washed with low salt, high salt, LiCl wash buffer, and chromatin immunocomplex was eluted using elution buffer through incubating at room temperature for 15 minutes. Reverse crosslinks of protein/DNA complexes to free DNA were realized through adding 5 mol/L NaCl and incubating at 65°C overnight. qPCR was performed on DNA purified after treatment with RNase (30 minutes, 37°C) and proteinase K (2 hours, 55°C) after reversal of crosslinks.

Immunostaining and RNA fluorescence in situ hybridization (RNA-FISH) were performed according to the reported protocol (22). Cells were first slicked on sterile and 0.01% poly-lysine-treated slides in the bottom of 6-well tissue culture dish. After that, the slides were processed sequentially with ice-cold CSK buffer, CSK + 0.4% Triton X-100 buffer and CSK buffer for 30 seconds for cell membrane perforation. The slides were then treated with 4% PFA for 10 minutes and cold 70% ethanol three times for cells fixation. After being rinsed three times with ice-cold PBS, the slides were blocked in prewarmed 5% goat serum for 30 minutes at 37°C. Then, the slides were incubated with primary antibody at 37°C for 1 hour, washed three times with 1 × PBS/0.2% Tween-20 for 3 minutes on a rocker, and then incubated with secondary antibody at 37°C for 30 minutes. After washing three times with 1 × PBS/0.2% Tween-20, the slides were fixed with 2% PFA at room temperature for 10 minutes. The slides were dehydrated by moving them through a room temperature ethanol series (85%, 95%, and 100% ethanol) for 2 minutes each, and air dried at room temperature for 15 minutes and hybridized using the indicated probes overnight at 37°C in a humid chamber. After washing with 2 × SSC/50% formamide, 2 × SSC, and 1 × SSC each for three times, DAPI dye was added. Finally, the slides were sealed and then observed using a confocal microscope.

RIP was performed according to the previously reported protocol (22). Briefly, the cells were harvested, washed, added ice-cold IP lysis buffer (Thermo Scientific Pierce) containing 0.5% ribonuclease inhibitor (Invitrogen), and incubated on ice for 5 minutes with periodic mixing. Then, the lysates were transferred into a microcentrifuge tube and centrifuged at 13,000 × g for 10 minutes to pellet the cell debris at 4°C, and the supernatants were transferred into a new tube, and protein G agarose was added and incubated for 1 hour at 4°C with rotation for preclearing. The immunoprecipitating antibody was added and incubated overnight at 4°C with rotation. Protein G agarose was pelleted by brief centrifugation (3,000 × g for 1 minute) and then washed sequentially with IP lysis buffer (containing 0.5% ribonuclease inhibitor). Finally, RNA was extracted from protein/RNA complexes on the beads using TRIzol reagent and dissolved in DEPC water and quantified by quantitative PCR (qPCR).

RNA–protein pulldown analyses were performed using the Pierce Magnetic RNA-Protein Pull-Down Kit. MDSCs were harvested and cell lysates were prepared using IP lysis buffers (Thermo Scientific Pierce; ref. 23). Lnc-C/EBPβ was transcribed (NEB, manual HiScribe T7 in vitro transcription Kit) and labeled using RNA 3′ Desthiobiotinylation Kit (Thermo Scientific Pierce) in vitro. Fifty microliters beads and 50 pmol/L of labeled RNA were added into RNA capture buffer and incubated for 30 minutes at room temperature with agitation to binding of labeled lnc-C/EBPβ to streptavidin magnetic beads. After washing beads with an equal volume of 20 mmol/L Tris (pH 7.5), 100 μL of 1 × protein–RNA binding buffer was added into the beads and mixed well. Master mix (100 μL) of RNA–protein binding reaction was added to the RNA-bound beads, mixed by pipetting, and then incubated 60 minutes at 4°C with rotation to binding of RNA-binding proteins to RNA. After washing beads with 100 μL wash buffer twice, 50 μL of elution buffer was added and incubated 30 minutes at 37°C with agitation. The samples were analyzed on a gel.

Statistical analyses were performed using two-tailed Student t test and GraphPad Prism 5 software (GraphPad Software). Tumor growth kinetics were assessed by two-way ANOVA test. The Mann–Whitney U test was used to determine significant differences between healthy individuals and patients. A 95% confidence interval was considered significant and was defined as *, P < 0.05; **, P < 0.01; ***, P < 0.001.

MDSCs derived from different environments (Fig. 1A) have distinct immunosuppressive functions (24). To understand the molecular mechanisms responsible for MDSC differentiation and function, we compared gene expression between MDSCs induced with GM-CSF plus IL6 and MDSCs induced with GM-CSF alone. We used lncRNA, miRNA, and protein-coding mRNA microarrays (Supplementary Fig. S1A–S1C). Data showed differences not only in the expression of miRNAs and mRNAs but also in lncRNAs (Fig. 1B; Supplementary Fig. S1B; GSE104718 and GSE104719). Out of 16,251 candidate probe sets that represent 7,811 lncRNA transcripts, we found 1,139 probe sets (representing 303 lncRNA transcripts) to be significantly up- or downregulated in GM-CSF with IL6-mediated MDSCs (GSE104718). Most of these lncRNAs were intergenic and are thus lincRNAs (71.95%). Among all, 169 probe sets (129 lncRNA transcripts) had a fold change difference more than 2, with the majority (85 of 101) of lncRNAs being downregulated (GSE104718). To validate our findings, we further analyzed expression of the top 15 probe sets. Most upregulated lncRNAs could be further confirmed by qRT-PCR and/or RT-PCR based on the predicted sequence published in UCSC (ref. 25; Fig. 1C; Supplementary Fig. S1C). Because E130102H24Rik is expressed in GM-CSF plus IL6-mediated MDSCs (Supplementary Fig. S1C), we further characterized it and named it lnc-C/EBPβ. Lnc-C/EBPβ is an intergenic lncRNA encoded on chromosome 4 (Supplementary Fig. S2A and S2B). This lncRNA was conserved among mouse and human, as well as other species (Fig. 1D). The 5′ and 3′ RACE (rapid amplification of complementary DNA ends) analysis showed that mouse lnc-C/EBPβ was about 750 bases in length, a size confirmed by northern blotting (Fig. 1E). Human lnc-C/EBPβ was longer (around 1,000 bp) and is located on chromosome 1 (Fig. 1F). Lnc-C/EBPβ was predominately localized to the nucleus and lacked coding capacity in mouse and human (Fig. 1G; Supplementary Fig. S2C–S2E). Human and mouse lnc-C/EBPβ sequences shared about 45% homology (Supplementary Fig. S3A–S3C). Lnc-C/EBPβ was not only expressed in myeloid-derived cells but also induced in in vitro and in vivo tumor and inflammatory environments (Fig. 1H–M; Supplementary Fig. S4). Inhibitors of both STAT3 and JAK1, factors in the IL6-mediated signaling pathway, affected IL6-mediated expression of Lnc-C/EBPβ (Fig. 1K; Supplementary Fig. S5). Isolated MDSCs from IL6-deficient mice bearing B16 tumors had less lnc-C/EBPβ (Fig. 1L). Lnc-C/EBPβ was not detected in lymphoid-derived cells, such as CD4⁺ and CD8⁺ T cells and B cells (Fig. 1M). Taken together, our data demonstrate that inflammatory and tumor-associated factors such as IL6 may induce lnc-C/EBPβ expression in MDSCs.

We next determined whether lnc-C/EBPβ was involved in the regulation of MDSC-mediated immune suppression. We used in vitro gain-of-function and loss-of-function studies to investigate effects of lnc-C/EBPβ on MDSCs (Supplementary Fig. S6A and S6B). Microarray analysis showed differences in the gene expression in lnc-C/EBPβ knockdown or exogenous lnc-C/EBPβ overexpressed MDSCs as compared with their controls (Fig. 2A; Supplementary Fig. S6C and S6D and GSE104558). Expression of Arg-1, CYBB (NOX2), NOS2, and ptgs2(COX2), which play a role in regulating immunosuppressive functions of MDSCs (26, 27), was significantly upregulated in lnc-C/EBPβ knockdown MDSCs but downregulated in the exogenous lnc-C/EBPβ treated MDSCs. QRT-PCR and immunoblot further confirmed these alterations (Fig. 2B and C). Their respective metabolic products Arg-1, NO, H₂O₂, and ROS were more abundant in lnc-C/EBPβ knockdown MDSCs but less abundant after gain-of-function treatment, in which lnc-C/EBPβ inhibited production of Arg-1, NO, H₂O₂, and ROS (Fig. 2D–G). Consistent with a previous report (10), we observed that M-MDSCs can produce high amounts of Arg-1, whereas PMN-MDSCs mainly depend on H₂O₂ (Fig. 2D–G). Because the metabolic products Arg-1, NO, H₂O₂, and ROS may affect CD4⁺ and CD8⁺ T-cell proliferation and function (10, 27, 28), we performed in vitro experiments to observe the effects of MDSC on antigen-specific T cells. Although lnc-C/EBPβ knockdown MDSCs were added to OT-I CD8⁺ or OT-II CD4⁺ T cells, which may respond to MHCI- or MHCII-restricted ovalbumin (OVA) peptides, these MDSCs could significantly reduce IFNγ production in CD4⁺ and CD8⁺ T cells in the presence of CD4- or CD8-specific OVA peptides as compared with control MDSCs. On the other hand, MDSCs lentivirally transduced with lnc-C/EBPβ promoted IFNγ production (Fig. 2H and I). Thus, lnc-C/EBPβ may reduce immunosuppressive function of both M-MDSCs and PMN-MDSCs.

MDSCs are involved in the occurrence and development of tumor (10, 26). Because lnc-C/EBPβ may regulate differentiation and the immune-suppressive function of MDSCs, we next used two mouse transplantable tumor models to investigate the effects of lnc-C/EBPβ on tumor growth, one bearing ovalbumin (OVA)-expressing mouse B16 melanoma and the other bearing mouse LLC. CD45.1⁺ mouse BMCs from homogeneous mice were infected with lnc-C/EBPβ shRNA/lentivirus or lnc-C/EBPβ/lentivirus and injected into B16 tumor-bearing mice via the tail vein at the indicated time, and then tumor growth kinetics were monitored (Fig. 3A). Promotion of tumor growth was observed in mice injected with lnc-C/EBPβ knockdown cells compared with controls, and slower tumor growth was observed in mice injected with exogenous lnc-C/EBPβ overexpressed cells compared with mice injected with control lentivirus-treated MDSCs (Fig. 3B–D). Similar results were observed in mice bearing LLC (Fig. 3E–G). The proportion of CD4⁺ and CD8⁺ T cells decreased in the tumor tissues of mice injected by lnc-C/EBPβ knockdown cells. Increased percentages of CD4⁺ and CD8⁺ T cells were observed in mice injected by lnc-C/EBPβ overexpressed MDSCs (Fig. 3H and I). More IFNγ could also be detected in CD4⁺ and CD8⁺ T cells isolated from the tumor tissues of mice injected by exogenous lnc-C/EBPβ modified MDSCs as compared with control (Fig. 3J and K). These results indicated that lnc-C/EBPβ may reduce immunosuppressive function of MDSCs.

The decreased proportion of Ly6G^lowLy6C⁺ M-MDSCs was also observed in the mice injected by lnc-C/EBPβ-transduced cells, whereas increased percentages of Ly6G⁺Ly6C⁺ cells appeared in the mice injected with lnc-C/EBPβ knockdown cells (Fig. 3L and M), suggesting that lnc-C/EBPβ may also affect differentiation of M-MDSCs (Supplementary Figs. S7 and S8). Indeed, lnc-C/EBPβ impeded the differentiation of MDSCs into M-MDSCs. CD45.1⁺ cells could be confirmed in the spleen and tumor tissues of mice injected by lnc-C/EBPβ-infected cells (Fig. 3N), indicating successful establishment of animal models. Taken together, our data demonstrated that lnc-C/EBPβ could reduce the immunosuppressive function of MDSCs.

We next sought to determine the underlying molecular mechanisms by which lnc-C/EBPβ regulates MDSC immunosuppressive functions. Because lnc-C/EBPβ was mostly localized to chromatin, it may regulate expression of target genes by recruiting accessory factors, similar to other characterized lncRNAs (15, 29, 30). Multiple transcription factors, such as C/EBPβ, STAT3, and CHOP, play a role in the differentiation and function of MDSCs (1, 24). We, thus, hypothesized that lnc-C/EBPβ might combine with these transcription factors to regulate MDSC function and differentiation. To test this, we performed RIP and pulldown analyses. We found that lnc-C/EBPβ only bound with the C/EBPβ isoform LIP (Fig. 4A–C; Supplementary Fig. S9). This binding limited the separation of C/EBPβ LIP and LAP isoforms (Fig. 4D and E). The interaction of C/EBPβ with lnc-C/EBPβ was confirmed by FISH and immunostaining (Fig. 4F).

C/EBPβ is required for the immunosuppressive program in both tumor-induced and bone marrow–derived MDSCs (31, 32). C/EBPβ has three isoforms: two liver-enriched activator proteins (LAP* and LAP) and a liver-enriched inhibitory protein (LIP). LAP* and LAP function as transcriptional activators. LIP lacks DNA activation domains, but it may heterodimerize with other family members to control gene expression (33). Thus, the interaction of lnc-C/EBPβ with C/EBPβ LIP may affect binding of C/EBPβ LAP to gene promoters (Fig. 4G). To test this, we analyzed the promoter region of lnc-C/EBPβ-regulated genes Arg-1, NOS2, NOX2, and COX and found potential binding sites for C/EBPβ (Fig. 4G). We next used ChIP-PCR analysis to determine the effects of binding of lnc-C/EBPβ to C/EBPβ LIP on the C/EBPβ activity on the promoter region of these genes. ChIP-PCR showed that lnc-C/EBPβ knockdown promoted enrichment of C/EBPβ, whereas exogenous lnc-C/EBPβ reduced enrichment of C/EBPβ to these promoter regions (Fig. 4H), indicating that binding of lnc-C/EBPβ with LIP inhibits activity of the C/EBPβ isoform LAP. Because IL6 may promote the expression of lnc-C/EBPβ in MDSCs, we also examined the correlation between Arg-1, NOS2, NOX2, COX2, and lnc-C/EBPβ expression during IL6-mediated differentiation. The expression of Arg-1, NOS2, NOX2, and COX2 was induced after exposure to IL6 but decreased after 24 hours (Fig. 4I). The pattern of lnc-C/EBPβ expression was different from the expression patterns of Arg-1, NOS2, NOX2, and COX2, where lnc-C/EBPβ expression gradually increased, even 4 days after exposed to IL6 (Fig. 4I), indicating that negative feedback role of lnc-C/EBPβ on Arg-1, NOS2, NOX2, and COX2 genes. The protein levels of LIP and LAP did not significantly change after 24 hours (Fig. 4J), indicating that the alteration of Arg-1, NOS2, NOX2, and COX2 gene expression is not dependent on LIP and LAP expression. Taken together, lnc-C/EBPβ may interact with the C/EBPβ isoform LIP to impede the activation of the C/EBPβ gene.

Because human lnc-C/EBPβ is homologue to mouse lnc-C/EBPβ, we also examined whether human lnc-C/EBPβ is similar in function to murine lnc-C/EBPβ. Human lnc-C/EBPβ siRNA knockdown reduced expression of Arg-1, NOS2, NOX2, and COX2. Similar results were seen in lnc-C/EBPβ shRNA knockdown in human MDSC-like cells (Fig. 5A and B). Finally, we detected expression of lnc-C/EBPβ in the samples of patients who underwent surgical resection of the colon. RNA in situ hybridization and immunostaining of colitis/carcinoma tissues demonstrated expression of lnc-C/EBPβ in CD11b⁺ monocytes/macrophages (Fig. 5C and D). MDSCs with the phenotype CD3⁻HLA-DR⁻CD33⁺CD11b⁺ have been isolated from the blood of patients with colon cancer (34). More lnc-C/EBPβ could be detected in CD3⁻HLA-DR⁻CD33⁺CD11b⁺ MDSCs and their subsets (Fig. 5E–G). RNA in situ hybridization and immunostaining also demonstrated expression of lnc-C/EBPβ in isolated CD3⁻HLA-DR⁻CD33⁺CD11b⁺ MDSCs (Fig. 5H). Because human lnc-C/EBPβ may regulate expression of MDSC-associated genes in vitro, overexpression of lnc-C/EBPβ may play a role in the differentiation and function of MDSCs in patients with colorectal cancer. Taken together, human lnc-C/EBPβ is also involved in the differentiation and function of myeloid-derived immune cells.

Previous data have shown that the transcription factor C/EBPβ plays a central role in regulating the immunosuppressive function of MDSCs (1, 24). We, here, demonstrated that the binding of lnc-C/EBPβ with C/EBPβ affected the activity of C/EBPβ, interrupting the immune-suppressive function of MDSCs. We also found that lnc-C/EBPβ may be induced in inflammatory and tumor environments. Thus, our data suggest that negative feedback by lnc-C/EBPβ limits the immunosuppressive function of MDSCs in inflammatory and tumor environments. The identification of lnc-C/EBPβ revealed a regulatory locus that allowed the immunosuppressive function of MDSCs to be controlled in response to extracellular inflammatory and tumor-associated signals. Because lnc-C/EBPβ is present in human monocyte-derived cells, it may be useful as a potential therapeutic target for inflammatory and tumor-associated diseases.

The C/EBPβ isoform LAP is necessary for the immunosuppressive program in both tumor and inflammatory induced MDSCs (24) and plays a critical role in regulating the expression of immune-suppressive genes (31, 32). We demonstrated that the binding of lnc-C/EBPβ to LIP limited LAP from inactivation of C/EBPβ LAP. Other studies have also found that when heterodimerized with other family members, LIP may inhibit the transcriptional activation activity of its partner (33). Although lnc-C/EBPβ is expressed by MDSCs, it negatively regulates MDSC activity, suggesting a negative feedback role of lnc-C/EBPβ to avoid oversuppression of MDSCs on immune responses.

No potential conflicts of interest were disclosed.

Conception and design: Y. Gao, R. Yang

Development of methodology: R. Yang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Zhang, T. Wang, X. Zhang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Gao, W. Sun, W. Shang, Y. Li, R. Yang

Writing, review, and/or revision of the manuscript: W. Sun, R. Yang

Study supervision: Y. Zhang, R. Yang

Other (offered assistance for the animal experiments): T. Wang

This research was supported by NSFC grants 91029736, 9162910, 81600436, and 91442111; the Joint NSFC-ISF Research Program, which is jointly funded by the National Natural Science Foundation of China and the Israel Science Foundation (ISF-NSFC program 31461143010); the National Key Research and Development Program of China (2016YFC1303604); and the State Key Laboratory of Medicinal Chemical Biology.

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.