USP15 Facilitates Colorectal Cancer Immune Evasion through SMYD3/CCL2-Dependent Myeloid-Derived Suppressor Cell Recruitment

The authors report USP15 as a key regulator of MDSC recruitment in colorectal cancer. USP15 deubiquitylates SMYD3, thereby enhancing CCL2 transcription and MDSC tumor infiltration. USP15 holds potential as a biomarker and therapeutic target in colorectal cancer.

Abstract

Colorectal cancer creates a suppressive tumor immune microenvironment that leads to tumor progression and resistance to immune checkpoint inhibitor therapy. Ubiquitin-specific protease 15 (USP15) broadly regulates immune responses and immune cell differentiation, but its involvement in shaping the tumor immune microenvironment of colorectal cancer remains unclear. This study demonstrated that USP15 is overexpressed in colorectal cancer and correlated with a poor prognosis. Employing colon orthotopic and metastatic tumor models, we performed loss- and gain-of-function assays for USP15 and revealed that overexpression of USP15 promotes tumor progression by increasing the abundance of myeloid-derived suppressor cells (MDSC) and decreasing the presence of CD8⁺ T cells in the tumor microenvironment. Through in vitro co-culture models and rescue experiments, we observed that tumoral USP15 decreased T-cell abundance by promoting MDSC recruitment rather than directly affecting T cells. Mechanistically, we found that USP15 deubiquitinated SMYD3, thereby activating H3K4me3-mediated transcription and the release of CCL2, which subsequently recruited MDSCs. Treatment with a USP15 inhibitor improved the efficacy of PD-1 blockade in colorectal cancer models. In a cohort of patients with colorectal cancer undergoing immunotherapy, we observed that those with high USP15 expression had a poor response to anti–PD-1 therapy. In summary, this research explored how USP15 facilitates the recruitment of MDSCs and identified it as a promising target for enhancing immunotherapy in colorectal cancer.

Introduction

Colorectal cancer is the third most prevalent cancer and the second leading cause of cancer-related deaths worldwide (1, 2). An immunosuppressive tumor microenvironment (TME) contributes to colorectal cancer recurrence, progression, and resistance to treatment, particularly immunotherapy (3). KRAS mutations, particularly G12D, are among the most common molecular drivers of colorectal cancer (4). These mutations can inhibit IRF2, resulting in the release of CXCL3, which recruits myeloid-derived suppressor cells (MDSC) and limits T-cell infiltration (5). VEGF-A is commonly overexpressed in colorectal cancer, in which it not only drives aberrant angiogenesis but also stimulates the expression of the transcription factor TOX in T cells, driving T-cell exhaustion (6). TGF-β is a central cytokine in the colorectal cancer microenvironment, capable of silencing antitumor immune responses by impairing T-cell function and fostering the differentiation of MDSCs and regulatory T cells (7). Targeting these key molecular events has shown limited efficacy in improving colorectal cancer immunotherapy, highlighting the necessity of fully understanding the regulatory mechanisms underlying immune resistance in colorectal cancer for potential clinical impact.

Ubiquitin-specific proteases (USP), the largest and most diverse deubiquitinating enzyme family, are key regulators of immune modulation in the TME (8–10). Therefore, investigating the role of USP family members in the tumor immune microenvironment (TIME) and identifying potential targets is crucial for advancing cancer immunotherapy. Studies have shown that USP15 plays a key role in tumor progression and broadly regulates immune responses and immune cells, making it a promising therapeutic target. In glioblastoma, USP15 interacts with the SMURF2 complex to deubiquitinate and stabilize the type I TGF-β receptor, boosting TGF-β signaling (11). TET2 modulates the IFN-JAK-STAT pathway, stimulating chemokine production and tumor infiltration by lymphocytes (12). In melanoma, the deletion of USP15, a deubiquitinase for TET2, promotes chemokine production and tumor infiltration by lymphocytes in a TET2-dependent manner (13). USP15 knockout (KO) promotes T-cell activation by stabilizing the E3 ubiquitin ligase MDM2, leading to a decrease in the levels of the transcription factor NFATc2 (14). Additionally, USP15 interacts with retinoid acid–related orphan receptor γt (RORγt) and enhances its activity by recruiting the coactivator SRC1, thereby influencing Th17 cell differentiation (15). These results emphasize the pivotal function of USP15 in modulating the tumor environment.

MDSCs facilitate tumor immune evasion and support resistance to immunotherapy. MDSCs suppress T-cell proliferation and activation by depleting critical amino acids, expressing indoleamine 2,3-dioxygenase 1, and secreting ADAM17, which hinders the migration of naïve T cells to tumors or lymph nodes (16–18). MDSCs also induce T-cell dysfunction by downregulating the ζ-chain of the T-cell receptor and generating reactive oxygen species and reactive nitrogen species (19). MDSCs promote regulatory T-cell activation and expansion by expressing PD-L1, binding to PD-1 on T cells, and secreting IL10 and TGF-β (20). Although USP15 is known to regulate various immune cells and signals, its specific role in the regulatory mechanisms of MDSCs is not completely defined.

In this study, we analyzed the expression and function of USP15 in colorectal cancer. We evaluated the expression and clinical significance of USP15 in patients with colorectal cancer using clinical cohorts. We established colon orthotopic and metastatic tumor models in both immunocompetent and immunodeficient mice to perform loss- and gain-of-function assays of USP15. Mass cytometry and immunofluorescence (IF) were employed to analyze the alterations in the immune microenvironment of colorectal cancer. Additionally, we validated the relationship between USP15 and the efficacy of immunotherapy in a clinical cohort of patients with colorectal cancer. This work highlights the crucial role of USP15 in immune regulation in colorectal cancer, particularly its function in modulating myeloid immune cells, especially MDSCs.

Materials and Methods

Cell lines

Mouse colorectal cancer cell lines (CT26 and CMT93) and human colorectal cancer cell lines (HT29, HCT116, HCT15, SW48, SW480, SW620, SW948, DLD1, and RKO) were sourced from the National Collection of Authenticated Cell Cultures in 2022. The number of passages of all cells before use is within five generations. Colorectal cancer cell lines HCT116, SW480, SW620, SW948, DLD1, CT26, and CMT93 were cultured in RPMI 1640 medium (Gibco, #C11875500BT). HCT15, HT29, SW48, and RKO were cultured in DMEM (Gibco, #C11995500BT). The culture medium was supplemented with 10% FBS (Gibco, #10099-141C) and 1% penicillin/streptomycin (Biyuntian Biotechnology, #C0222). Cell lines were passaged every 2 to 3 days. The identity of each cell line was verified by short tandem repeat DNA profiling, and Mycoplasma contamination was periodically tested.

Animal experiments

All animal experiments adhered to the Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of Harbin Medical University. We used 6-week-old female BALB/c (#210726240100158525), C57BL/6J (#210726240100158685), and BALB/c nu/nu mice (#210726240100364872), which were procured from Changsheng Bio-Technology. To examine tumor growth in vivo after USP15 KO or overexpression, we established subcutaneous xenograft tumors. CT26 or CMT93 cells (5 × 10⁵) were injected in 100 μL of PBS into the right flank of female BALB/c, C57BL/6J, or BALB/c nu/nu mice. Tumor size was measured using calipers every 2 to 3 days, beginning 6 to 10 days after injection. Tumor volume was calculated using the following formula: 0.5 × length × width² (in mm³). Mice exhibiting signs of distress, such as tumors larger than 2,000 mm³, ulcers exceeding 5 mm, weight loss greater than 20%, or abnormal posture, were humanely euthanized. We also established colorectal cancer orthotopic models. BALB/c or C57BL/6J mice were implanted with 1 × 10⁶ CT26 or CMT93 cells (carrying a luciferase reporter gene) into their cecal wall. The implantation procedure was performed under anesthesia to minimize distress. Tumor growth was monitored using the IVIS Spectrum imaging system (Caliper Life Sciences). For the lung metastasis models of colorectal cancer, BALB/c or C57BL/6J mice were intravenously injected via the tail vein with 1 × 10⁶ cells in 100 μL of PBS. Eighteen days after injection, lung metastasis progression was assessed by quantifying bioluminescent signals using the IVIS Spectrum imaging system.

Rescue experiments and treatment in mouse models

To evaluate the impact of inhibiting MDSC recruitment on USP15-regulated tumor growth in vivo, subcutaneous tumor models were established in BALB/c or C57BL/6J mice using CT26/CMT93 Usp15-overexpressing and wild-type (WT) cell lines. Mice were randomly assigned to receive either CXCR2 antagonists (SB225002, Selleck, #HY-16711; 10 mg/kg, intraperitoneally, three times per week) or no treatment once tumors reached approximately 100 mm³. Tumor growth was monitored as described above.

The evaluation of SMYD3 overexpression rescues tumor growth in Usp15 KO cells in vivo: Subcutaneous tumor models were established in BALB/c or C57BL/6J mice using four groups of CT26/CMT93 cell lines: WT, WT–overexpressing Smyd3, Usp15 KO, and Usp15 KO–overexpressing Smyd3. Each group received a subcutaneous injection of 5 × 10⁵ cells, and tumor growth was monitored by caliper measurements as described previously.

CCL2 blocking effects on USP15-regulated tumor growth

Subcutaneous xenograft tumor models were established in BALB/c or C57BL/6J mice using 5 × 10⁵ WT and Usp15-overexpressing CT26/CMT93 cells. When tumors reached around 100 mm³, mice were randomly assigned to receive either CCL2 blocking treatment or no treatment. Anti-mouse CCL2 (Selleck, #2H5) was administered intraperitoneally at 10 mg/kg, three times weekly for a total of seven doses. Tumor growth was monitored as described previously.

USP15-targeted mouse experiments

To assess the effects of the USP15 inhibitor (USP15-IN-1, MedChemExpress, #HY-148046) and anti-mouse PD-1 (CD279, Selleck, #A2122, RRID: AB_3644244), both individually and in combination, subcutaneous and orthotopic colorectal cancer models were established as described. In the subcutaneous model, treatment started when tumors reached 100 mm³, with mice divided into four groups. The regimens included USP15-IN-1 (5 mg/kg) and anti-mouse PD-1 (CD279, Selleck, #A2122, RRID: AB_3644244) administered at 200 μg/mouse three times weekly, or both combined via intraperitoneal injection. In the orthotopic colorectal cancer model, treatment began 1 week after implantation, with mice randomized into four groups. Treatment regimens were the same as in the subcutaneous model, with injections three times weekly for a total of six sessions. The mice were euthanized 14 days after the start of treatment or on the day of the last treatment. After humane euthanasia, tissues were collected for the evaluation of treatment effects.

USP15-targeted mouse experiments in immunodeficient mice

We utilized 6-week-old female BALB/c nu/nu mice to establish subcutaneous colorectal cancer xenografts. Treatment commenced when tumor volumes reached 100 mm³, with the mice assigned to two groups. The treatment regimens included USP15-IN-1 (5 mg/kg) or a control, administered via intraperitoneal injection three times per week. Tumor growth was monitored according to previously described protocols. Following humane euthanasia, tumor and tissue samples were harvested for subsequent analysis of treatment efficacy.

Patient samples and information collection

We collected pathologic tissue sections along with basic and prognostic information from 181 patients with colorectal cancer from Harbin Medical University Cancer Hospital (HMUCH), establishing two clinical cohorts. Cohort 1 (n = 78) is designated to evaluate the differential expression of USP15 between colorectal cancer tissues and adjacent normal tissues, as well as its association with prognosis in colorectal cancer (Supplementary Table S1). Patients in cohort 1 were diagnosed with colorectal cancer at HMUCH and underwent surgical resection of the primary tumor, with complete clinical follow-up data available. Cohort 2 (n = 93) is designed to evaluate the correlation between USP15 expression levels and the efficacy of immunotherapy (Supplementary Table S2). This cohort includes patients with advanced or metastatic colorectal cancer who have received treatment with PD-1/PD-L1 inhibitors. Eligible participants must have pretreatment tumor tissue obtained by surgical resection available for USP15 IHC analysis, measurable lesions as defined by Response Evaluation Criteria in Solid Tumors (RECIST) 1.1, and complete clinical efficacy data, including treatment response, progression-free survival (PFS), and overall survival (OS). Additionally, paired fresh tumor tissues and adjacent normal tissues were obtained from 25 patients with colorectal cancer, constituting an independent validation cohort specifically for assessing USP15 expression levels. The biopsy or surgically resected cancer tissues and adjacent normal tissues were collected prior to treatment, fixed in 4% formaldehyde (Sigma-Aldrich, #1.00496), and embedded in paraffin (Sigma-Aldrich, #8002-74-2).

Treatment responses were evaluated according to the RECIST version 1.1. Complete response (CR) is defined as the complete disappearance of all target lesions, with the short diameter of any pathologic lymph node reduced to less than 10 mm. Partial response (PR) is characterized by a reduction of at least 30% in the sum of the diameters of target lesions compared with baseline. Progressive disease is identified by an increase of at least 20% in the sum of target lesion diameters or the appearance of one or more new lesions. Stable disease (SD) refers to a state in which the tumor size does not meet the criteria for either PR or progressive disease. The objective response rate (ORR) is defined as the proportion of patients whose tumors exhibit a specified reduction in size, including those achieving either CR or PR. The disease control rate (DCR) refers to the proportion of patients whose tumors achieve CR, PR, or SD for a specified period. PFS is defined as the time from treatment initiation to either disease progression or death. OS refers to the duration from the start of treatment to death from any cause or to the conclusion of the study for patients who remain alive. All participants provided written informed consent, and the study adhered to ethical guidelines as outlined in the Declaration of Helsinki. This research received approval from the Ethics Committee of HMUCH (No. KY2022-37).

USP15 KO using CRISPR/Cas9

USP15 KO plasmids were designed based on predicted sequences obtained from the Eukaryotic Pathogen CRISPR Guide RNA/DNA Design Tool and inserted into the lentiCRISPR v2-GFP vector (Addgene, #52961) to generate USP15 single-guide RNA (sgRNA) plasmids. A total of 8 μg of USP15 sgRNA plasmids was transfected into HCT15 or CMT93 cells plated on a 100 mm plate using 500 μL of jetPRIME buffer and 20 μL of jetPRIME transfection reagent (Polyplus, #101000046). Forty-eight hours later, the single-cell suspension was plated into 96-well plates. When the cells reached confluence, they were passaged for expansion. Finally, Western blotting was conducted to confirm the absence of USP15 expression in individual clones. The USP15 sgRNA sequences are as follows: human-sgRNA-1:5′-CACCGGCGTCGCGATGTCAGACCGCGTTT-3′, human-sgRNA-2:5′-CACCGTGCTGATATCCAGAGAACTGGTTT-3′, mouse-sgRNA-1:5′-CACCGGGCGGAGCGGCGGACCTGGAGTTT-3, and mouse-sgRNA-2:5′-CACCGGACACCTGGTATCTAGTAGAGTTT-3′.

Lentivirus transduction

Following our previously described method (21), the target gene sequences (USP15, Usp15, and Smyd3) were retrieved from NCBI GenBank and subsequently cloned into the pCDH-EF1-MCS-CMV-copGFP-T2A-Puro vector (Synbio Technologies). The colorectal cancer cells were infected with the viral supernatant, followed by selection using puromycin (1 μg/mL, Sigma-Aldrich, #120-73-0) for a duration of 1 to 2 weeks. The overexpression of the gene was confirmed by qPCR and Western blot.

Luciferase-expressing cells were established via retroviral transduction. Lentiviral particles were generated using the pLenti-Puro-CMV-luciferase vector (Vigene Biosciences). Subsequently, the colorectal cancer cell line was transduced with these viral particles. Stable transformants were obtained by puromycin selection (1 μg/mL, Sigma-Aldrich, #120-73-0), differentiating between luciferase-overexpressing and mock-infected populations.

CCK-8 assay

Cells were seeded at a density of 1,500 cells per well in a 96-well plate. Cells were treated with 10% Cell Counting Kit-8 (CCK-8; GlpBio, #GK10001) solution at designated time points (1, 2, 3, and 4 days) for 2 hours. The absorbance at 450 nm was measured to assess cell proliferation.

EdU assay

For 5-ethynyl-2’-deoxyuridine (EdU) staining, cells were cultured overnight on slides and subsequently stained using the BeyoClick EdU Cell Proliferation Kit with Alexa Fluor 488 (Beyotime, #C0071S) following the manufacturer’s instructions.

Colony formation assay

Cells were seeded into six-well plates at a density of 700 to 1,000 cells per well. After incubating the plates for 14 days, colonies were fixed with methanol and stained with a 0.5% crystal violet solution. The density of the stained colonies was measured using ImageJ (RRID: SCR_003070).

Wound-healing assay

Colorectal cancer cells were seeded into six-well plates and incubated for 24 hours. After reaching full confluence, a scratch was made using a 200 μL pipette tip. Then, the scratched cells were cultured in serum-free medium for 24 hours and subsequently observed under an Olympus CKX53 inverted microscope (Olympus Corporation) equipped with a 4× objective lens at both 0 and 24 hours after the scratch. Images were captured using cellSens Entry software (version 4.2). The migration area was quantified using ImageJ 2.0.0.

Transwell migration assay

Cells were plated in the upper chambers of Transwell plates (Corning, #3422) and incubated for 24 hours. After incubation, cells were fixed with methanol and stained with crystal violet. Migrated cells on the lower membrane surface were visualized under an Olympus CKX53 inverted microscope (Olympus Corporation) equipped with a 10× objective lens. Images were captured using cellSens Entry software (version 4.2) and quantified using ImageJ software (version 2.0.0).

Multiplex IHC

Formalin-fixed, paraffin-embedded (FFPE) sections of subcutaneous xenograft tumor tissue were cut into 4 μm serial sections. Antigen retrieval was conducted following the IHC protocol, followed by incubation with primary antibodies. The primary antibodies used were CD11b (Abcam, #ab133357, RRID: AB_2650514) and CD8α (Abcam, #ab217344, RRID: AB_2890649). Subsequently, the primary antibodies were conjugated with tyramide signal amplification (TSA) fluorophores from the Multiplex IF kit (Absin, #abs50013). Fluorophores and 4',6-diamidino-2-phenylindole (DAPI) were prepared according to the manufacturer’s guidelines. Images were acquired using an Olympus BX53 microscope (Olympus Corporation) at a magnification of 6.3× under appropriate laser excitation conditions.

IHC staining

FFPE tissue sections of 4 μm thickness were dewaxed and rehydrated, followed by the inactivation of endogenous peroxidase and antigen retrieval. The sections were treated with antibodies overnight at 4°C: USP15 (Abcam, #ab97533, RRID: AB_10678830), SMYD3 (Proteintech, #12011-1-AP, RRID: AB_2193981), CCL2 (Proteintech, #26161-1-AP, RRID: AB_2918100), CD8α (Cell Signaling Technology, #85336, RRID: AB_2800052), CD11b (Abcam, #ab133357, RRID: AB_2650514), and Ki67 (Abcam, #ab15580, RRID: AB_443209). Next, the sections were treated with secondary antibodies (OriGene, #PV6001). The cell nucleus was stained with Mayer’s hematoxylin (Beyotime, #C0107). The immunoreactivity score was utilized to evaluate the protein expression in tumor and normal tissues. The method involves assessing the staining intensity (scored from 0 to 3: 0 = no staining, 1 = weak, 2 = moderate, and 3 = strong) and the proportion of positive cells (scored from 0 to 4: 0 = 0%, 1 = 1%–10%, 2 = 11%–50%, 3 = 51%–80%, and 4 = 81%–100%). The ultimate score falls within the range of 0 to 12. The percentage of positive cells in each slide was quantified.

H&E staining assay

Hematoxylin and eosin (H&E) staining was performed to evaluate the safety of USP15-IN-1. FFPE tissue sections were deparaffinized in xylene, rehydrated through graded ethanol, and stained with Mayer’s hematoxylin (Beyotime, #C0107) for 2 minutes to visualize the nuclei. Counterstaining with eosin Y (BASO, #BA4024) was performed for 1 minute to stain the cytoplasmic components. The sections were then dehydrated, cleared in xylene, and mounted for microscopic examination.

IF assay

Cells were cultured overnight on slides, fixed in 4% paraformaldehyde, and permeabilized with 0.3% Triton X-100. After blocking, the cells were incubated overnight with the primary antibody at 4°C. The antibodies used were as follows: USP15 (Santa Cruz Biotechnology, #sc-100629, RRID: AB_2214725) and SMYD3 (Proteintech, #12011-1-AP, RRID: AB_2193981). Subsequently, Goat Anti-Mouse IgG H&L (Alexa Fluor 488; Abcam, #ab150117, RRID: AB_2688012) or Goat Anti-Rabbit IgG H&L (Alexa Fluor 594; Abcam, #ab150084, RRID: AB_2734147) was added for 1 hour, and antifade mounting medium with DAPI was used for sealing (Beyotime, #P0131). IF images were obtained on a ZEISS LSM 900 confocal microscope, and confocal image processing was performed using ZEISS software.

Western blotting

Cultured cells and tissues were lysed using a protease inhibitor (Beyotime, #P1005) containing RIPA (Thermo Fisher Scientific, # 89001) on ice for 30 minutes. The supernatant was obtained by centrifuging at 12,000 rpm for 10 minutes and quantified using a bicinchoninic acid assay (Thermo Fisher Scientific, #23228). SDS loading buffer (Beyotime, #P0015F) was added to the lysates. Proteins underwent electrophoresis, membrane transfer, and blocking, followed by the application of antibodies overnight at 4°C. The antibodies were USP15 (Abcam, #ab97533, RRID: AB_10678830), SMYD3 (Proteintech, #12011-1-AP, RRID: AB_2193981), GAPDH (used as the loading control; Proteintech, #10494-1-AP, RRID: AB_2263076), ubiquitin (Abcam, #ab134953, RRID: AB_2801561), H3K4me3 (PTMab, #PTM-613), histone H3 (PTMab, #PTM-6600), P53 (Proteintech, #21891-1-AP, RRID: AB_10896826), MDM2 (Proteintech, #27883-1-AP, RRID: AB_2881003), and Goat Anti-Rabbit IgG H&L (HRP; Abcam, # ab205718, RRID: AB_2819160). Proteins were visualized using an enhanced chemiluminescence reagent (Epizyme, #SQ202L) on an e-BLOT imaging system (Touch Imager).

Co-immunoprecipitation

The cells were lysed on ice for 30 minutes using RIPA (Thermo Fisher Scientific, #89001) containing protease inhibitors (Beyotime, #P1005). The lysate was centrifuged at 12,000 × g for 10 minutes. Immunoprecipitation was performed using antibodies against USP15 (Abcam, #ab97533, RRID: AB_10678830), SMYD3 (Proteintech, #12011-1-AP, RRID: AB_2193981), or IgG (Abcam, #ab171870, RRID: AB_2687657), followed by incubation with protein A/G agarose beads (Santa Cruz Biotechnology, #sc2003) overnight at 4°C with gentle rotation. After washing three times with lysis buffer, the eluted proteins were analyzed by SDS-PAGE and Western blot.

Ubiquitination assay

Cells were transfected with the indicated plasmids. After 48 hours, the cells were treated with 10 μmol/L MG132 (Selleck, #S2619) for 4 hours and then harvested. The lysate was incubated overnight at 4°C with antibodies against SMYD3 (Proteintech, #12011-1-AP, RRID: AB_2193981) or HA tag (Cell Signaling Technology, #3724, RRID: AB_1549585). Immune complexes were captured by incubation with 40 µL of protein A/G agarose beads (Santa Cruz Biotechnology, #sc2003) for 6 hours at 4°C. The complexes were then centrifuged at 2,500 rpm for 5 minutes at 4°C and analyzed by Western blot.

Protein half-life assay

Cells were treated with 10 μmol/L cycloheximide (MedChemExpress, #HY-12320) for various durations (0, 3, 6, and 9 hours) to inhibit protein synthesis. The cell lysates were prepared, and protein expression was analyzed by Western blot.

RT-qPCR

Total RNA from three replicates was isolated using TRIzol reagent (Thermo Fisher Scientific, #15596026) according to the manufacturer’s instructions. A total of 1 μg of purified RNA was used for RT-PCR using the PrimeScript FAST RT Reagent Kit with gDNA Eraser (Takara, #RR092A) and run on a T100 Thermal Cycler (Bio-Rad, #621BR65372). qPCR was performed using the SYBR Green kit (TIANGEN, #B0211A) and run on the Real-Time PCR System (Thermo Fisher Scientific, #7500). Subsequently, RNA was reverse transcribed into cDNA for fluorescence quantification. The changes in target gene expression in the control group and experimental group were evaluated using the 2^−ΔΔCt method. Below are the primer sequences utilized for qRT-PCR:

Human-USP15-F: 5′-CGA​CGC​TGC​TCA​AAA​CCT​C-3′

Human-USP15-R: 5′-TCC​CAT​CTG​GTA​TTT​GTC​CCA​A-3′

Mouse-Usp15-F: 5′-CCG​TGG​ATG​AAA​ACC​TGA​GTA​G-3′

Mouse-Usp15-R: 5′-TTC​TCT​TAG​GCA​GAC​AGG​GAT​AA-3′

Human-SMYD3-F: 5′-CGC​GTC​GCC​AAA​TAC​TGT​AGT-3′

Human-SMYD3-R: 5′-CAA​GAA​GTC​GAA​CGG​AGT​CTG-3′

Mouse-Smyd3-F: 5′-CCG​ACC​CCT​TGG​CTT​ACA​C-3′

Mouse-Smyd3-R:5′-CGGCATTGAGAACAACGCATC-3′

Human-CCL2-F: 5′-CAG​CCA​GAT​GCA​ATC​AAT​GCC-3′

Human-CCL2-R: 5′-TGG​AAT​CCT​GAA​CCC​ACT​TCT-3′

Mouse-Ccl2-F: 5′-TTA​AAA​ACC​TGG​ATC​GGA​ACC​AA-3′

Mouse-Ccl2-R: 5′-GCA​TTA​GCT​TCA​GAT​TTA​CGG​GT-3′

Mouse-Cxcl9-F: 5′-TCC​TTT​TGG​GCA​TCA​TCT​TCC-3′

Mouse-Cxcl9-R: 5′-TTT​GTA​GTG​GAT​CGT​GCC​TCG-3′

Mouse-Cxcl10-F: 5′-CCA​AGT​GCT​GCC​GTC​ATT​TTC-3′

Mouse-Cxcl10-R: 5′-GGC​TCG​CAG​GGA​TGA​TTT​CAA-3′

Human-TP53-F: 5′-CAG​CAC​ATG​ACG​GAG​GTT​GT-3′

Human-TP53-R: 5′-TCA​TCC​AAA​TAC​TCC​ACA​CGC-3′

Mouse-Tp53-F: 5′-CTC​TCC​CCC​GCA​AAA​GAA​AAA-3′

Mouse-Tp53-R: 5′-CGG​AAC​ATC​TCG​AAG​CGT​TTA-3′

Human-MDM2-F: 5′-GAA​TCA​TCG​GAC​TCA​GGT​ACA​TC-3′

Human-MDM2-R: 5′-TCT​GTC​TCA​CTA​ATT​GCT​CTC​CT-3′

Mouse-Mdm2-F: 5′-TGT​CTG​TGT​CTA​CCG​AGG​GTG-3′

Mouse-Mdm2-R: 5′-TCC​AAC​GGA​CTT​TAA​CAA​CTT​CA-3′

Human-ACTB-F: 5′-CAT​GTA​CGT​TGC​TAT​CCA​GGC-3′

Human-ACTB-R: 5′-CTC​CTT​AAT​GTC​ACG​CAC​GAT-3′

Mouse-Actb-F: 5′-GGC​TGT​ATT​CCC​CTC​CAT​CG-3′

Mouse-Actb-R: 5′-CCA​GTT​GGT​AAC​AAT​GCC​ATG​T-3′

Human ACTB and Mouse Actb were used as internal controls to normalize the expression in human and mouse cells, respectively.

ChIP-qPCR

The protocol for chromatin immunoprecipitation (ChIP)-qPCR was as outlined in our previous publication (22). Cells were cross-linked by treating them with 1% formaldehyde (Sigma-Aldrich, #F8775) at room temperature for 10 minutes. The reaction was then quenched using 0.125 mol/L glycine (Sigma-Aldrich, #G8790). The cells were lysed ultrasonically; chromatin was incubated overnight with antibodies: SMYD3 (Proteintech, #12011-1-AP, RRID: AB_2193981), H3K4me3 (PTMab, #PTM-613), and IgG (Abcam, #ab171870, RRID: AB_2687657). Agarose beads (Santa Cruz Biotechnology, #sc2003) were added for 4 hours to facilitate immunoprecipitation, followed by chromatin elution.

Phenol/chloroform was used to purify DNA, and qPCR was performed with the following primers: CCL2-F: 5′-CCT​GCT​TCC​CTT​TCC​TAC-3′ and CCL2-R: 5′-TTC​CTC​TGG​CTG​CTG​TCT-3′.

For the qPCR analysis, a total of 1 μL of the eluted chromatin sample was used as the template for each qPCR. Each experiment was conducted with three biological replicates, and each biological replicate had three technical replicates. The qPCR was run on the Real-Time PCR System (Thermo Fisher Scientific, #7500). The input DNA was used as an internal reference for normalization.

ChIP-seq analysis

To analyze the relationship between H3K4me3 and CCL2, the target gene CCL2 was selected, and the H3K4ME3 ChIP sequencing (ChIP-seq) datasets of Caco2 (https://www.ncbi.nlm.nih.gov/geo/, #GSM945162; ref. 23), HCT116 (#GSM2701778; ref. 24), LS174T (#GSM1365901; ref. 25), and CRYPT3 (#GSM883692; ref. 26) cell lines were screened and downloaded from the Cistrome Data Browser database. Subsequently, the UCSC Genome Browser was used for visualization processing.

CCL2 ELISA detection

We evaluated the endogenous level of CCL2 using an ELISA according to the manufacturer’s instructions (Meimian, #MM-0723 M1). Smyd3 was overexpressed in both WT and Usp15 KO colorectal cancer cells. Supernatants from these cells, as well as from their corresponding subcutaneous tumor tissues, were collected and centrifuged to remove cellular debris. The supernatants, along with other reaction reagents, were added to the wells of a 96-well ELISA plate. After incubation, the absorbance was measured at 450 nm on the Tecan microplate reader (Tecan, #30086376) and calculated using ELISA Calc to quantify the CCL2 levels.

MDSC isolation and in vitro migration assay

Bone marrow (BM)–MDSCs were cultured following the established protocols (27). Tibias and femurs from C57BL/6J and BALB/c mice were harvested using sterile techniques, and BM was collected through flushing. To remove red blood cells, ammonium chloride was used for lysis. For the differentiation of BM-derived MDSCs, 2.5 × 106 BM cells were plated in 100 mm dishes with 10 mL of medium, supplemented with 40 ng/mL GM-CSF (PeproTech, #250-05) and 40 ng/mL IL6 (PeproTech, #216-16). The cells were incubated at 37°C in a 5% CO₂-humidified atmosphere for 3 days to facilitate co-cultivation experiments (27). Filtered conditioned medium (CM) from different colorectal cancer cells was added to the lower chambers as a chemoattractant. After 4 hours of incubation, the number of migrated cells in the bottom chamber was quantified.

T-cell suppression assay

The BeaverBeads Mouse CD8⁺ Cell Isolation Kit (BEAVER, #170905-100) was used to isolate CD8⁺ T cells from the spleens of BALB/c or C57BL/6J mice. The isolated T cells were then seeded into plates coated with anti-CD3/CD28 (BioLegend, #100340, RRID: AB_11149115; BioLegend, #102116, RRID: AB_11147170). On the day of co-culture, T cells were collected, labeled with 5,6-carboxyfluorescein diacetate succinimidyl ester(Beyotime, #C1031), and co-cultured with tumor cells at a 1:1 ratio, with or without Usp15 overexpression in tumor cells. After 72 hours of co-culture, T-cell proliferation was assessed by flow cytometry. Proliferation peaks were identified using the proliferation modeling tool in FlowJo 10.8.1 software (RRID: SCR_008520).

Flow cytometry analysis

To analyze immune cells infiltrating the tumor tissue, the tumor was placed in a medium containing collagenase IV and DNase and digested on a shaking platform at 37°C for 1 hour. The resulting single-cell suspension was used for flow cytometry and stained with the following antibodies: mCD45-PerCP/Cyanine5.5 (BioLegend, #103132, RRID: AB_893340), mCD3-FITC (BioLegend, #100204, RRID: AB_312661), mCD8-APC (BioLegend, #100712, RRID: AB_312751), mCD11b-FITC (BioLegend, #101206, RRID: AB_312789), and m-Ly-6G/Ly-6C (Gr-1)-PE/Cyanine7 (BioLegend, #108416, RRID: AB_313381). Stained cells were analyzed using a BD FACSMelody flow cytometer (BD Biosciences), and data analysis was performed using FlowJo 10.8.1 software (RRID: SCR_008520).

CyTOF

Mouse subcutaneous tumor tissues were collected and cut into small pieces before undergoing enzymatic digestion. Following digestion, the tissue fragments were processed into a single-cell suspension using 0.5 mmol/L cisplatin in 1 mL of PBS (without Ca²⁺ and Mg²⁺) at room temperature for 2 minutes. To stop the reaction, 2 mL of cell staining buffer (Fluidigm, #201068) was added, and the mixture was centrifuged at 500 × g for 5 minutes. After blocking Fc receptors (BioLegend, #422302), the cells were incubated at room temperature for 30 minutes with a panel of 41 metal-labeled antibodies (Supplementary Table S3). The cells were then washed with cell-staining buffer and subsequently incubated with 125 mmol/L Intercalator-Ir (Fluidigm, #201192A) in a fixation and osmotic buffer (Fluidigm, #201067) at room temperature for 1 hour. Following this, the cells were washed twice with Maxpar Water (Fluidigm, #201069), mixed with EQ Beads (Fluidigm, #201078), and analyzed using a cytometry by time-of-flight (CyTOF) instrument (Fluidigm, #Helios).

LC/MS-MS analysis

Quantitative proteomic analysis was performed by Jingjie PTM BioLabs (Hangzhou, China) using LC/MS-MS on TMT-labeled proteomes from HEK293T cells transfected with either a blank pENTER or USP15-flag vector. Forty-eight hours after transfection, cells were lysed in 8 mol/L urea lysis buffer (Sigma-Aldrich, #57-13-6) with 1% protease inhibitors (Merck Millipore, #P8340) and 50 μmol/L PR-619 (Selleck, #S7130) and sonicated on ice. After centrifugation at 12,000 × g for 10 minutes at 4°C, protein concentrations were measured using a BCA kit (Beyotime, #P0009). Proteins were reduced with 5 mmol/L DTT (Sigma-Aldrich, #3483-12-3) at 56°C for 30 minutes and alkylated with 11 mmol/L iodoacetamide (Sigma-Aldrich, #DNTP-RO) at room temperature. Trypsin (Promega, #V5111) digestion was carried out overnight at a 1:50 enzyme-to-protein ratio, followed by a second 4-hour digestion at 1:100. Peptides were purified using a Strata X C18 SPE column and labeled with a TMT/iTRAQ kit (Thermo Fisher Scientific, #90309). Fractionation was performed via high pH reverse-phase high-performance liquid chromatography, and LC/MS-MS was conducted on a Q Exactive mass spectrometer (Thermo Fisher Scientific). The peptides were dissolved in liquid chromatography mobile phase A (0.1% formic acid and 2% acetonitrile in water) and separated using a nanoElute ultrahigh-performance liquid chromatography system (Bruker). Mobile phase B consisted of 0.1% formic acid in 100% acetonitrile. The gradient elution was programmed as follows: 7% to 24% B from 0 to 42 minutes, 24% to 32% B from 42 to 54 minutes, 32% to 80% B from 54 to 57 minutes, and 80% B from 57 to 60 minutes, with a constant flow rate of 450 nL/minute. The separated peptides were then introduced into a CaptiveSpray ion source for ionization prior to analysis on a timsTOF Pro mass spectrometer (Bruker). The ion source was operated at 1.65 kV, and both precursor ions and their fragment ions were detected and analyzed using high-resolution time-of-flight detection. The mass spectrometry (MS)/MS scan range was set to 100 to 1,700 m/z, and data acquisition employed parallel accumulation serial fragmentation mode. For each full-scan MS spectrum, 10 parallel accumulation serial fragmentation–mode MS/MS spectra were acquired for precursor ions with charge states ranging from 0 to 5, and dynamic exclusion was set to 30 seconds to prevent redundant scanning of the same precursor ions. Differentially expressed peptides were identified with a log fold change >1.2 and P value < 0.05.

Bulk RNA sequencing

Tumor tissues were isolated from CT26-WT and CT26-Usp15 KO subcutaneous tumor models. Each group consisted of four mice. The tumor tissues were dissociated into single-cell suspensions. Total RNA was extracted using the RNeasy Mini Kit (QIAGEN) and quantified with a NanoDrop 2000 (Thermo Fisher Scientific). The integrity of the total RNA was evaluated using the Agilent 2100 Bioanalyzer (Agilent Technologies Inc.), and only samples with an RNA integrity number above 7.0 were subjected to sequencing. For RNA sample preparation, 1 μg of RNA was used as the input material.

After the total RNA samples pass the quality control test, eukaryotic mRNA is enriched by using Oligo(dT) magnetic beads. Then, fragmentation buffer is added to fragment the mRNA. Next, using mRNA as a template, the first- and second-strand cDNAs are synthesized, followed by cDNA purification. After that, the purified double-stranded cDNA undergoes end-repair, addition of base A, addition of sequencing adapters, and fragment screening to recover cDNA of about 350 bp. Finally, PCR enrichment is carried out to obtain a cDNA library. Subsequently, the double-stranded cDNA is denatured into single-stranded DNA by high-temperature treatment, and a circularization primer is added to form a single-stranded circular library (a platform-specific step for the DNBSEQ-T7 platform). After the constructed library passes the quality inspection by Qubit 3.0 and Agilent 2100, it is sequenced using a high-throughput sequencing platform with a sequencing strategy of PE 150.

Raw sequencing data frequently harbor adapter sequences and low-quality reads. To ensure high-quality data for subsequent analyses, these are removed to yield clean reads. The specific filtering procedures are as follows: adapter trimming is carried out to eliminate reads with more than 5 bp of adapter contamination, and in the case of paired-end sequencing, either end being contaminated leads to discard; low-quality removal involves getting rid of reads in which more than 50% of the bases have a quality score (Q) of 19 or lower, and for paired-end data, if either end meets this criterion, the read is discarded; ambiguity filtering is performed to exclude reads containing more than 5% ambiguous bases (“N”), and in paired-end sequencing, if either end surpasses the 5% threshold, the read is removed. Clean reads were aligned to the reference genome using HISAT2 software. The reference genome used is Mus_musculus.GRCm38.89. Differential gene expression was analyzed with the criteria of |log₂Fold Change| ≥ 1 and FDR <0.05.

TCGA analysis

We analyzed the relationship between USP15 expression and TP53 expression or status in colorectal cancer using The Cancer Genome Atlas (TCGA; https://portal.gdc.cancer.gov/) database. The colorectal cancer dataset included 546 cases from TCGA. When grouping by TP53 expression status, there were 326 cases in the TP53-mutant group and 220 cases in the TP53-WT group. When stratifying by USP15 expression levels, there were 273 cases in the USP15-high group and 273 cases in the USP15-low group. Batch correction was performed using the log₂(TPM+1)-transformed data. Differential expression analysis and visualization were conducted in R (version 4.2.3) using the ggplot2 package, with the Wilcoxon test selected for statistical comparison. Genes with a P value < 0.05 were considered significantly differentially expressed between normal and tumor groups.

Gene Expression Omnibus analysis

We used the Gene Expression Omnibus (RRID: SCR_005012) database to analyze the relationship between USP15 expression and prognosis. GSE75500 included 114 cases. The Kaplan–Meier method was used to generate survival curves and to analyze OS and disease-free survival (DFS). Optimal cutoff values for USP15 expression were determined using the “surv_cutpoint” function from the “survminer” package in R. Survival distributions among experimental groups were compared using the log-rank test.

Statistical analysis

Statistical analyses were conducted using GraphPad Prism version 10.0 software (RRID: SCR_002798). Quantitative data are presented as mean ± SEM. Fisher exact test, one-way ANOVA, and Student t test were employed to compare differences among groups. Spearman correlation analysis was used to explore the associations between two variables. The Kaplan–Meier method was used to analyze PFS, OS, and DFS. Optimal USP15 cutoffs were determined via the “surv_cutpoint” function in R’s “survminer” package, and survival differences were assessed by the log-rank test. Statistical significance is indicated as follows: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

Data availability

The data generated in this study are publicly available in the Genome Sequence Archive at the China National Center for Bioinformation under accession numbers PRJCA029001 (https://ngdc.cncb.ac.cn/bioproject/browse/PRJCA029001) and CRA023135 (https://ngdc.cncb.ac.cn/gsa/).

Results

USP15 is upregulated and correlates with poor prognosis in colorectal cancer

USP15 protein expression was first assessed in tumor and adjacent normal tissues from 78 patients with colorectal cancer using IHC. High expression (immunoreactivity score ≥6) was observed in 57.7% of tumor samples (Fig. 1A). Western blot analysis of eight paired colorectal cancer and adjacent normal tissues, along with RT-qPCR analysis of 17 paired samples, confirmed significantly higher USP15 levels in tumor tissues compared with adjacent normal tissues (Fig. 1B and C). Additionally, both USP15 protein and mRNA expression were markedly elevated in nine colorectal cancer cell lines compared with three normal colon tissue samples (Fig. 1D and E). Kaplan–Meier survival analysis demonstrated a strong correlation between high USP15 expression and poor prognosis (Fig. 1F and G). To investigate the functional role of USP15, Usp15 was knocked out in CMT93 and HCT15 cells, whereas its expression was overexpressed in CT26 and HCT116 cells (Supplementary Fig. S1A and S1B). Cell proliferation was evaluated using CCK-8, EdU, and colony formation assays (Fig. 1H–J; Supplementary Fig. S1C–S1F). Migration ability was assessed through wound-healing and Transwell migration assays (Fig. 1K and L; Supplementary Fig. S1G and S1H). USP15 modulation had no significant effect on colorectal cancer cell proliferation or migration. Collectively, these findings suggest that USP15 is a prognostic marker associated with adverse outcomes in patients with colorectal cancer.

Figure 1.

USP15 is upregulated and correlates with poor prognosis in colorectal cancer. A, Colorectal cancer tissues (n = 78) and paired normal tissues (n = 78) from HMUCH were stained for USP15, and the IHC score was assessed by two independent pathologists. Representative images showing the different levels of USP15 expression are presented (scale bars, 60 μm). B, Western blot analysis showed the expression of USP15 in colorectal cancer tissues (n = 8) and paired normal tissues (n = 8) from HMUCH. C, RT-qPCR was used to detect the expression of USP15 in colorectal cancer tissues (n = 17) and paired normal tissues (n = 17) from HMUCH. D and E, The levels of USP15 protein and mRNA in normal colon tissues (n = 3) and cancerous cell lines (n = 9) were assessed. F, The PFS and OS rates in 78 patients with colorectal cancer (cohort 1 from HMUCH) with low (n = 33) and high (n = 45) USP15 levels. G, The DFS and OS rates were analyzed between 114 patients with colorectal cancer with low (n = 101) or high (n = 13) USP15 expression from GSE75500. H–J, Cell proliferation was analyzed using CCK-8 reagents, EdU IF staining (scale bar, 200 μm), and colony formation assays in HCT15 WT cells and USP15 KO cells. K and L, Cell migration was analyzed by wound-healing assays (scale bar, 20 μm) and Transwell analysis in CMT93 WT or CMT93 SgUsp15 cells (scale bar, 150 μm). Data are expressed as the mean ± SEM. The indicated sample size (n) represents biological replicates. Three independent replicates were performed. P values determined by Fisher exact test (A), paired Student t test (C), ANOVA (E), log-rank test (F and G), or unpaired Student t test (others). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. Ctrl, control; N, normal tissue; OD, optical density; T, tumor tissue.

USP15 promotes tumor growth in colorectal cancer immunocompetent mouse models

Initially, multiple colorectal cancer animal models were established to assess the effect of USP15 expression on tumor progression. Subcutaneous tumors were generated using CMT93 WT and Usp15 KO cells and CT26 LV-control and LV-Usp15 cells (Fig. 2A; Supplementary Fig. S2A). Compared with controls, the tumor volume and weight of the Usp15 overexpression group were substantially increased, whereas the Usp15 KO tumor volume and weight were decreased (Fig. 2B–D; Supplementary Fig. S2B–S2D). In addition, we further assessed the expression level of USP15 in the subcutaneous tumors using IHC, which confirmed the KO and overexpression of Usp15 (Supplementary Fig. S2E and S2F). Orthotopic colorectal cancer models further demonstrated that Usp15 overexpression promoted tumor growth and increased tumor weight, whereas Usp15 KO led to diminished tumor growth and weight (Fig. 2E–G; Supplementary Fig. S2G–S2J). Lung metastasis models showed that Usp15 overexpression significantly enhanced bioluminescence intensity and resulted in more metastatic nodules and higher tumor weight as opposed to the CT26 LV-control group (Fig. 2H–L). Conversely, Usp15 KO suppressed lung metastasis (Supplementary Fig. S2K–S2N). In mice with T-lymphocyte dysfunction colorectal cancer models, the groups between WT and SgUsp15 cell lines showed no significant differences in tumor growth (Fig. 2M–P). These data suggested that USP15 facilitates tumor progression through mechanisms that are related to T cell–mediated antitumor immunity. Collectively, USP15 promotes tumor growth in colorectal cancer immunocompetent mouse models.

Figure 2.

USP15 promotes tumor growth in colorectal cancer (CRC) immunocompetent mouse models. A, Schematic diagram of constructing CT26 syngeneic subcutaneous colorectal cancer models. BALB/c mice (n = 5 per group) were subcutaneously injected with 5 × 10⁵ CT26-LV-Ctrl or CT26-LV-Usp15 cells. B–D, Display of tumor volume, tumor weight, and tumor growth of subcutaneous tumors in CT26 cells. E, Conceptual diagram of constructing CT26 cell orthotopic colorectal cancer models. CT26-LV-Ctrl-Luc and CT26-LV-Usp15-Luc cells (1 × 10⁶) were injected into the cecal wall of BALB/c mice (n = 3 per group). F, Bioluminescence images were captured, and the corresponding signals from orthotopic colorectal cancer models were quantified. G, The orthotopic colorectal cancer models were displayed. H, Pattern diagram of constructing CT26 cell lung metastasis colorectal cancer models. CT26-LV-Ctrl-Luc and CT26-LV-Usp15-Luc cells (1 × 10⁶) were intravenously injected via the tail vein in BALB/c mice (n = 5 per group). I, Bioluminescence images of lung metastasis in colorectal cancer models were quantified. J, Lung metastases derived from CT26-LV-Ctrl-Luc and CT26-LV-Usp15-Luc cells were excised for H&E staining, and the number of lung metastatic nodules was quantitated (scale bar, 1,000 μm). K and L, Lung weight of mice in lung metastasis colorectal cancer models. M, Schematic of constructing CT26 cells subcutaneous colorectal cancer models in immune-deficient mice (n = 5 per group). BALB/c nu/nu mice (n = 5 per group) were subcutaneously injected with 5 × 10⁵ CT26-LV-Ctrl or CT26-LV-Usp15 cells. N–P, Tumor volume, tumor weight, and tumor growth of mice in immune-deficient colorectal cancer models. Data with error bars are shown as mean ± SEM. Three independent replicates were performed. P values determined by unpaired Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. Ctrl, control.

USP15 induces high MDSC infiltration and low CD8⁺ T-cell infiltration in the colorectal cancer microenvironment

We hypothesized that USP15 promotes tumor progression by modulating the TIME. To test this, we processed three fresh tumor tissues from subcutaneous colorectal cancer models bearing either CT26 WT or SgUsp15 cells in BALB/c mice. We conducted CyTOF to investigate the role of USP15 in shaping the TME (Fig. 3A). We identified and annotated nine distinct clusters according to the varying expression levels of cell type–specific markers (Fig. 3B). A t-distributed stochastic neighbor embedding (t-SNE) plot was used to represent the distributions of these populations (Fig. 3C). As shown in the t-SNE plot, the infiltration of MDSCs and macrophages decreased in the SgUsp15 group, with an increase in CD8⁺ T-cell infiltration (Fig. 3D and E). Furthermore, variant markers were visualized in t-SNE plots (Supplementary Fig. S3A–S3I). Most notably, the expression of the MDSC markers Ly6G and CD11b decreased, whereas CD8α expression increased in samples with SgUsp15 (Supplementary Fig. S3A–S3C). However, the M2/M1 macrophage ratio remained unchanged between the two groups (Supplementary Fig. S3J). Flow cytometry confirmed that Usp15 KO subcutaneous tumors contained fewer MDSCs and more CD8⁺ T cells than WT tumors (Fig. 3F and G; Supplementary Fig. S3K and S3L). Our multiplex IHC results showed increased CD8α⁺ cells and decreased CD11b⁺ cells in subcutaneous transplant tumors of CT26 Usp15 KO or CMT93 Usp15 KO (Fig. 3H; Supplementary Fig. S3M). We performed RNA sequencing on CT26-WT and CT26-Usp15 KO subcutaneous tumors. We compared immune cell types in subcutaneous tumors, consistent with CyTOF results (Supplementary Fig. S4A). Additionally, there were no significant differences in macrophage proportion or M2/M1 ratio between the two groups (Supplementary Fig. S4A and S4B). Signature genes related to M1 and M2 macrophages were identified, but no significant expression differences were found between the groups (Supplementary Fig. S4C and S4D). Gene set enrichment analysis also showed no changes in M1 polarization and M2 polarization (Supplementary Fig. S4E and S4F). These results indicate that USP15 facilitates tumor progression in syngeneic colorectal cancer models by remodeling the suppressive TIME.

Figure 3.

USP15 induces high MDSC infiltration and low CD8⁺T-cell infiltration in the colorectal cancer microenvironment. A, Schematic illustration of CyTOF data acquisition. It is divided into five steps: tumor tissue collection, single-cell dissociation, cell staining, mass cytometry, and cell clustering and dimensionality reduction. B, Heatmap of the median marker intensities of the 41 markers in the nine cell populations. C, Nine cell populations obtained by manual merging in a t-SNE plot. D, t-SNE plot of cell infiltration in the USP15 WT (n = 3) and SgUsp15 (n = 3) groups. E, Histogram of the nine cell populations in CD45⁺ cells in the WT and SgUsp15 groups, respectively. F and G, The composition of MDSCs and CD8⁺ T cells in CT26 WT or CT26 SgUsp15 tumors was evaluated by flow cytometry. H, Detection of infiltration by CD8α-positive cells and CD11b-positive cells in subcutaneous tumors was observed (n = 5 per group, scale bar: 100 μm). Data are expressed as the mean ± SEM. Three independent replicates were performed. P values determined by ANOVA (E) or unpaired Student t test (F, G, and H). *, P < 0.05; ****, P < 0.0001; ns, not significant. DC, dendritic cell; DNT, double-negative T cell; gdT, gamma delta T cell.

MDSCs are essential effectors for USP15-induced suppressive TIME

To identify the immune cells crucial for Usp15 KO–mediated tumor suppression in colorectal cancer models, we first focused on the role of USP15 in MDSC regulation. Referencing previous research methods (27), we generated MDSCs from mouse BM cells by treating them with GM-CSF and IL6 (Fig. 4A). Following this, we evaluated the effect of USP15 on MDSC chemotaxis through an in vitro migration assay (Fig. 4B). Usp15 overexpression significantly increased MDSC migration toward CM from CT26 cells (Fig. 4C). In contrast, CM from CMT93 Usp15 KO cells inhibited MDSC migration (Fig. 4D). We then investigated the effects of USP15 on T cells. Using co-culture models, we confirmed that interfering with USP15 in colorectal cancer cells did not affect T-cell proliferation (Fig. 4E and F; Supplementary Fig. S5A). RT-qPCR experiments showed that CXCL9 and CXCL10 levels did not differ in colorectal cancer cells with Usp15 overexpression or KO (Fig. 4G; Supplementary Fig. S5B). Interfering with USP15 had no direct impact on T cells in vitro, contrasting with findings from in vivo models. These findings suggest that tumor cell–derived USP15 influences T-cell abundance by promoting MDSC recruitment rather than by directly affecting T-cell function. This was confirmed by inhibiting MDSC recruitment with the CXCR2 antagonist SB225002 in CT26 and CMT93 subcutaneous tumor models. When MDSC recruitment was blocked, Usp15 overexpression failed to significantly promote tumor growth (Fig. 4H–J; Supplementary Fig. S5C–S5F). Flow cytometry analysis supported the inhibitory effect of SB225002 on MDSC recruitment (Fig. 4K; Supplementary Fig. S5G). These results indicate that USP15 primarily promotes tumor growth through MDSC-mediated immune suppression.

Figure 4.

MDSCs are essential effectors for remodeling USP15-induced suppressive TIME. A, Workflow for inducing differentiation of mouse BM cells into MDSCs. B, Flowchart of the in vitro MDSC migration assay. C and D, Representative images of MDSCs that migrated to the lower chamber and quantification of MDSC migration toward CM (scale bar, 100 μm). E, Flowchart of the T-cell suppression assay. F, CD8⁺ T-cell proliferation rate was determined by flow cytometry. G, RT-qPCR analysis of CXCL9 and CXCL10 in CT26 LV-Ctrl and CT26 LV-Usp15 cells. H–J, Tumor volume, tumor weight, and tumor growth of CT26 LV-Ctrl and CT26 LV-Usp15 subcutaneous tumor models treated with either the control (Ctrl) or SB225002 (n = 5 per group). K, Overexpression of Usp15 did not increase MDSC levels after the application of SB225002 in subcutaneous tumors. Data are expressed as the mean ± SEM. Three independent replicates were performed. P values determined by ANOVA (C, D, and I–K) or unpaired Student t test (F and G). **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. BM, bone marrow; CFSE, carboxyfluorescein diacetate succinimidyl ester; CRC, colorectal cancer.

USP15 facilitates the stability of the SMYD3 protein by deubiquitination

Previous results showed that USP15 promotes MDSC accumulation in colorectal cancer progression. Studies also indicated that USP15 stabilizes MDM2 and regulates TP53 (14). Given the link between TP53 and MDSCs (28), we examined the relationship between USP15 and TP53/MDM2. Western blot and RT-qPCR revealed no effect of Usp15 KO or overexpression on Tp53/Mdm2 levels in colorectal cancer cells (Supplementary Fig. S6A–S6E). TCGA data also showed no correlation between USP15 expression and TP53 expression or status (Supplementary Fig. S6F–S6H). These findings suggest that USP15’s impact on the TME is independent of TP53. We hypothesized that USP15 regulates MDSC recruitment through a deubiquitinated protein and explored this using LC/MS-MS for USP15 interaction candidates (Supplementary Fig. S7A). Among the 13 potential USP15 interaction candidates identified by LC/MS-MS, Spearman correlation analysis identified SMYD3 as the top candidate associated with the MDSC signature (Fig. 5A; Supplementary Fig. S7B). This suggests that USP15 may modulate SMYD3 stability, contributing to MDSC accumulation. Co-immunoprecipitation assays confirmed the interaction between endogenously expressed USP15 and SMYD3 in HCT15 and CMT93 cells (Fig. 5B and C), with IF staining showing predominant cytoplasmic co-localization (Fig. 5D). To explore the clinical relevance, IHC analysis in a cohort of 93 patients with colorectal cancer demonstrated a strong positive correlation between USP15 and SMYD3 expression levels (Supplementary Fig. S7C). Mechanistically, we found that SMYD3 protein levels decreased upon USP15 KO and increased with USP15 overexpression, whereas SMYD3/Smyd3 mRNA levels remained unchanged (Fig. 5E and F). The reduction in SMYD3 protein expression due to USP15 KO was reversed by the proteasome inhibitor MG132 (Fig. 5G), and the protein half-life of SMYD3 was significantly shortened following USP15 KO (Fig. 5H; Supplementary Fig. S7D). Ubiquitination assays further demonstrated that USP15 overexpression inhibited SMYD3 ubiquitination, a process reversed by USP15 KO (Fig. 5I). In vivo rescue experiments in syngeneic subcutaneous colorectal cancer models validated that Smyd3 overexpression restored tumor growth in Usp15 KO cells. Specifically, Smyd3 overexpression in Usp15 KO cells significantly increased tumor volume compared with the Usp15 KO group (Fig. 5J–L). Flow cytometry revealed a higher frequency of MDSCs in tumors with Smyd3 overexpression cells, underscoring SMYD3’s role in enhancing MDSC recruitment (Fig. 5M; Supplementary Fig. S7E). Consistent findings were observed in the syngeneic subcutaneous colorectal cancer model using CMT93 cell lines (Supplementary Fig. S7F–S7I). Together, these results reveal that USP15 stabilizes SMYD3 through deubiquitination, providing a molecular mechanism by which USP15 drives MDSC accumulation and promotes colorectal cancer progression.

Figure 5.

USP15 facilitates the stability of the SMYD3 protein by deubiquitination. A, Spearman correlation analysis among 13 genes with MDSC signature. B and C, Co-immunoprecipitation (Co-IP) assays conducted on HCT15 and CMT93 cells. D, Co-localization analysis of USP15 and SMYD3 proteins in CMT93 and HCT15 cells. E and F, Expression of SMYD3 in HCT15 and CMT93 cell lines with USP15 KO or in HCT116 and CT26 cell lines overexpressing USP15. G, MG132 stabilized SMYD3 in the WT group and rescued the reduction of SMYD3 induced by SgUSP15. H, USP15 affected the half-life of the SMYD3 protein. I, Western blot detected the ubiquitination (Ub) level of SMYD3. J–L, Tumor volume, tumor weight, and tumor growth in the four groups of CT26 cell subcutaneous transplantation tumors (n = 5 per group): LV-Ctrl, LV-Smyd3, SgUsp15, and SgUsp15 + LV-Smyd3. M, Statistical chart of flow cytometry analysis of MDSCs. Data are expressed as the mean ± SEM. Three independent replicates were performed. P values determined by ANOVA (E, K, L, and M) or unpaired Student t test (F and H). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. CHX, cycloheximide; CRC, colorectal cancer; Ctrl, control; IB,  immunoblotting.

USP15 stabilizes SMYD3 via deubiquitination to activate CCL2 transcription through H3K4 trimethylation

Based on our findings that USP15 regulates MDSC recruitment via SMYD3, we next investigated the key immune mediators involved. SMYD3, as a histone methyltransferase, catalyzes the trimethylation of histone H3 at lysine 4 (H3K4me3; ref. 29). Previous studies have implicated H3K4me3 in regulating chemokines, including CCL2 (30), which are critical for MDSC recruitment via the CCL2–CXCR2 axis. Given these findings, we hypothesized that SMYD3 might epigenetically regulate CCL2 transcription. We analyzed publicly available ChIP-seq datasets, revealing significantly higher H3K4me3 enrichment at the CCL2 promoter in colorectal cancer cells compared with normal colonic crypt cells (Fig. 6A). SMYD3 binds to specific DNA sequences (5′-CCCTCC-3′or 5′-GGAGGG-3′) to regulate gene transcription (29). We identified potential SMYD3 binding sites within 2,000 bp of the CCL2 transcription start site, suggesting direct regulation of CCL2 by SMYD3. ChIP-PCR analysis confirmed reduced H3K4me3 enrichment at the CCL2 promoter in USP15 KO cells (Fig. 6B and C). Correspondingly, Western blot and RT-qPCR analyses showed decreased H3K4me3 levels and reduced CCL2 expression following USP15/Usp15 KO, whereas the opposite was observed in USP15/Usp15-overexpressing colorectal cancer cells (Fig. 6D and E). ELISA and RT-qPCR revealed reduced CCL2 levels in the supernatants of Usp15 KO CT26 and CMT93 cells, and Smyd3 overexpression effectively rescued the reduced CCL2 levels in Usp15 KO cells (Fig. 6F; Supplementary Fig. S8A). Similar results were observed in the supernatants of their corresponding subcutaneous tumors in mice (Fig. 6G; Supplementary Fig. S8B). In vitro, anti-CCL2 eliminated the differences in MDSC migration between control and Usp15-overexpressing culture supernatants (Fig. 6H and I; Supplementary Fig. S8C and S8D). In vivo, anti-CCL2 effectively suppressed USP15-driven colorectal cancer tumor growth and MDSC recruitment (Fig. 6J–M; Supplementary Fig. S8E–S8H). These findings demonstrate that SMYD3 promotes H3K4 trimethylation to activate CCL2 transcription, facilitating MDSC recruitment.

Figure 6.

SMYD3 promotes CCL2 expression via trimethylation of H3K4 in colorectal cancer. A, ChIP-seq analysis of H3K4me3 binding to the CCL2 promoter region in various cells as predicted by the Cistrome Data Browser and UCSC Genome Browser. B and C, ChIP-qPCR analysis of the binding between H3K4me3 and the CCL2 promoter in colorectal cancer WT or USP15 KO cells. D and E, Effect of the USP15/SMYD3/H3K4me3 axis on CCL2 expression in colorectal cancer cells knocked out or overexpressed in USP15. F, ELISA analysis of CCL2 release in the supernatant of four colorectal cancer cell line groups: LV-Ctrl, LV-Smyd3, SgUsp15, and SgUsp15 + LV-Smyd3. G, Tumor CCL2 concentration in CT26 and CMT93 syngeneic mouse models. H and I, Representative images of MDSCs that migrated to the lower chamber and quantification of MDSC migration toward CM (scale bar, 100 μm). J–L, Tumor volume, tumor weight, and tumor growth of CT26 LV-Ctrl and CT26 LV-Usp15 subcutaneous transplant tumors treated with anti-CCL2 or PBS (n = 5 per group). M, After anti-CCL2 treatment, MDSCs in subcutaneous transplant tumors were examined by flow cytometry. Data are expressed as the mean ± SEM. Three independent replicates were performed. P values determined by unpaired Student t test (B–E) or ANOVA (others). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. Ctrl, control.

Blocking USP15 augments anti–PD-1 therapy in colorectal cancer models through MDSC reprogramming

Considering the findings that USP15 stabilizes SMYD3 to promote MDSC recruitment via CCL2 transcription, thereby reducing CD8⁺ T-cell infiltration, we proposed that combining USP15 inhibition with immune checkpoint inhibitors (ICI) would yield superior antitumor activity compared with either monotherapy alone in colorectal cancer. Consequently, we investigated the antitumor effects of USP15-IN-1 in mice bearing subcutaneous CT26 or CMT93 WT tumors, and explored its therapeutic efficacy when combined with ICI treatment (Fig. 7A; Supplementary Fig. S9A). The combination therapy of USP15-IN-1 and ICI demonstrated significantly greater efficacy in reducing tumor volume and weight compared with individual treatment (Fig. 7B–D; Supplementary Fig. S9B–S9D). Consistently, the combination therapy reduced MDSC infiltration while enhancing CD8⁺ T-cell infiltration (Fig. 7E and F; Supplementary Fig. S9E–S9H). Next, we assessed the therapeutic efficacy of USP15 inhibition combined with ICI in orthotopic colorectal cancer mouse models. Similarly, quantitative fluorescence imaging and gross tumor examination demonstrated that the combination of USP15-IN-1 and anti–PD-1 therapy resulted in enhanced tumor control in orthotopic colorectal cancer in CT26 and CMT93 mouse models (Fig. 7G–J; Supplementary Fig. S9I and S9J). H&E staining of major organs (heart, liver, spleen, lungs, and kidneys) after USP15-IN-1 administration, as well as after the combination treatment of USP15-IN-1 with anti–PD-1, showed no signs of drug toxicity, indicating satisfactory biosafety (Supplementary Fig. S9K).

Figure 7.

Blocking USP15 augments the anti–PD-1 therapy in colorectal cancer models through MDSC reprogramming. A, CT26 syngeneic subcutaneous colorectal cancer models (n = 7 per group) were treated intraperitoneally with PBS, anti–PD-1, USP15-IN-1, or USP15-IN-1 + anti–PD-1. Treatment started on the seventh day after subcutaneous tumor inoculation and ended on the 21st day. B–D, Tumor volume, tumor weight, and tumor growth of CT26-challenged mice after treatment. E and F, The proportions of CD11b⁺/Gr1⁺ cells and CD8⁺/CD45⁺ cells in tumors were determined by flow cytometry. G and H, The mouse orthotopic colorectal cancer models were treated intraperitoneally with PBS, anti–PD-1, USP15-IN-1, or USP15-IN-1 + anti–PD-1 (n = 3 per group). Treatment started on the seventh day after subcutaneous tumor inoculation and ended on the 21st day. J, Tumor weight of CT26-challenged orthotopic colorectal cancer models. I, The overall appearance of the CT26-challenged orthotopic colorectal cancer models is displayed. K, Representative images of IHC staining for USP15, SMYD3, CCL2, CD11b, and CD8α in clinical colorectal cancer specimens (scale bar, 60 μm). L, Waterfall plot of patients who received immune checkpoint blockade therapy. M and N, Kaplan‒Meier survival analysis of PFS and OS in 93 patients with colorectal cancer. Data are expressed as the mean ± SEM. Three independent replicates were performed. P values determined by Fisher exact test (L), log-rank test (M and N), or ANOVA (others). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant.

Next, we assessed the impact of USP15-IN-1 on colorectal cancer cell proliferation using EdU (Supplementary Fig. S10A and S10B), colony formation assays (Supplementary Fig. S10C–S10F), and Ki67 IHC staining (Supplementary Fig. S10G and S10H), all of which showed no significant effect. In the immune-deficient mouse subcutaneous CT26 tumor model, USP15-IN-1 treatment alone did not significantly inhibit tumor growth (Supplementary Fig. S10I–S10L). Ultimately, our findings suggest that USP15 inhibition reprograms MDSCs and enhances CD8⁺ T-cell infiltration, thereby potentiating anti–PD-1 therapy in colorectal cancer models.

Our preclinical data strongly suggest that USP15 inhibition reprograms MDSCs and enhances CD8⁺ T-cell infiltration, thereby potentiating ICI in colorectal cancer models. To further validate the clinical relevance of these findings, we analyzed tissue samples from patients with colorectal cancer. Specifically, we performed IHC to analyze the levels of USP15, SMYD3, CCL2, CD8α, and CD11b in a clinical cohort of 78 patients with colorectal cancer. We observed positive correlations between the expressions of USP15 and SMYD3, as well as CCL2 and CD11b. We also noted a negative correlation between the expression levels of USP15 and CD8α (Fig. 7K). In our clinical cohort of 93 patients with colorectal cancer undergoing ICI therapy, most patients (42/48, 87.5%) in the USP15-low group achieved SD or PR (Fig. 7L). In the USP15-low group, the ORR was 39.6% (19/48), and the DCR was 87.5% (42/48), both of which were higher than those in the USP15-high group (ORR, 8.9%, 4/45; DCR, 57.8%, 26/45; Fig. 7L). Moreover, Kaplan–Meier survival analysis indicated that patients exhibiting high USP15 expression had significantly reduced DFS and OS (Fig. 7M and N). Collectively, our data indicate that USP15 is a promising therapeutic target, as its inhibition restores antitumor immunity and enhances the efficacy of ICI therapy in colorectal cancer (Supplementary Fig. S11).

Discussion

USP15, a well-characterized deubiquitinase, has captured substantial attention lately for its potential therapeutic value in the management of cancer. Evidence suggests that USP15 is essential for the regulation of DNA damage repair, redox reactions, and antitumor immune responses (31, 32). USP15 KO in mice results in genomic instability because of its role in promoting the localization of the BARD1/BRCA1 complex at DNA double-strand breaks for DNA repair (33). Additionally, USP15 engages with PARP1 and facilitates its deubiquitination, thereby boosting its stability and facilitating DNA repair, genomic stability, and the proliferation of triple-negative breast cancer cells (34). In human and mouse acute myeloid leukemia models, USP15 deletion greatly disrupts leukemic progenitor function and viability while also activating an antioxidant response via the KEAP1–NRF2 axis (35). Furthermore, we report that the ACVRL1/USP15 complex ubiquitinates GPX2, enhancing the redox capacity of tumor cells and contributing to colorectal cancer resistance to multitarget kinase inhibitors (36). USP15 also has a critical influence on immune pathways (e.g., TGF-β and IFNγ signaling) and immune cell differentiation (e.g., Th17, CD8⁺ T cells, and Th1 cells; refs. 11, 15). Collectively, these studies emphasize the promise of targeting USP15 for precision cancer therapy.

Reports suggest that USP15 has both promotive and suppressive effects on the malignant progression of various cancers. USP15 is upregulated in breast cancer, multiple myeloma, and glioblastoma, in which it promotes cell proliferation and inhibits apoptosis, indicating oncogenic functions and emerging as a potential therapeutic target (37–39). In gastric cancer, USP15 also drives malignant progression by remodeling glucose metabolism, further supporting its role as an oncogene (40). Conversely, other studies suggest that in cancers such as osteosarcoma and gastric cancer, USP15 inhibits cell proliferation and is associated with a better prognosis (41, 42). This suggests that USP15 may have varying roles in different cancers. Our previous research demonstrated that USP15 interacts with ACVRL1 to deubiquitinate GPX2 at the K187 site, leading to GPX2 accumulation and enhanced resistance of colorectal cancer cells to multitargeted tyrosine kinase inhibitors such as regorafenib and sorafenib (32). USP15 plays a crucial oncogenic role in colorectal cancer. It stabilizes PLOD2 to activate the AKT/mTOR pathway (43), promotes stemness and epithelial–mesenchymal transition, and facilitates liver metastasis (44). Additionally, USP15 regulates the ubiquitination of adenomatous polyposis coli (APC), modulating Wnt/β-catenin signaling (45), and collaborates with the COP9 signalosome to maintain APC stability, ensuring Wnt/β-catenin signaling pathway homeostasis (46). Dysregulation of USP15 disrupts these processes, driving tumorigenesis and metastasis. These findings underscore the pivotal role of USP15 in promoting tumor progression and drug resistance through key signaling pathways and protein interactions. However, its involvement in immune regulation within colorectal cancer remains unexplored. Through this research, we identified that USP15 expression was elevated in tumor tissues and linked to poor prognosis in patients with colorectal cancer. However, USP15 did not exhibit oncogenic functions or promote cell proliferation or migration in vitro or in nude mice. These findings suggest that USP15’s effects on colorectal cancer progression and prognosis are dependent on the immune microenvironment.

Research on USP15 in immune cells has primarily focused on T-cell differentiation and functional maintenance. In methylcholanthrene-induced fibrosarcoma, a lack of USP15 within T cells results in excessive IFNγ production, creating an immunosuppressive TME (47). USP15 interacts with RORγt, removes ubiquitin from K446, and enhances RORγt activity by promoting coactivator SRC1 recruitment. This suggests that USP15-mediated deubiquitination of RORγt positively affects Th17 differentiation (15). However, the role of USP15 in regulating myeloid cells within the TME remains unclear. Using an in vivo model combined with CyTOF analysis, we demonstrated that USP15 KO resulted in an increased abundance of tumor-associated macrophages and MDSCs and a decreased density of CD8⁺ T cells. Our results demonstrate that USP15 promotes the recruitment of MDSCs, which, in turn, contribute to tumor progression. However, in the subcutaneous xenograft model in nude mice, we did not observe a direct effect of USP15 on tumor growth. Through co-culture models and rescue experiments, we showed that USP15 drives malignant progression by recruiting MDSCs, which indirectly influence T-cell function. USP15 contributed to malignant progression and resistance to immunotherapy in colorectal cancer by recruiting MDSCs.

Mechanistically, LC/MS-MS analysis revealed that SMYD3 had the strongest association with MDSCs. SMYD3, a “histone code reader,” interprets H3K4me3 marks to promote gene activation (48, 49). The CCL2–CXCR2 axis is the primary signal for MDSC recruitment, and CCL2 transcription is highly dependent on H3K4me3 (50). We hypothesized that SMYD3 binds to specific DNA sequences on the CCL2 gene and enhances its transcription through H3K4me3 modification, thereby recruiting MDSCs. We demonstrated that USP15 interacts with SMYD3 and upregulates its expression through deubiquitination. Additionally, we identified the specific SMYD3 binding sequence located near the transcription start site of CCL2 and observed a reduction in H3K4me3 at the CCL2 gene promoter following USP15 deletion. Therefore, we propose that USP15 promotes MDSC recruitment by activating the SMYD3/H3K4me3/CCL2 signaling axis.

Previous studies reported that short hairpin RNA–mediated USP15 knockdown in HCT116 cells reduced MDM2 expression and activated the TP53 pathway (14). However, we did not observe similar results, possibly due to differences in gene silencing methods. Unlike previous studies, we employed CRISPR/Cas9-mediated permanent KO, which may trigger compensatory mechanisms restoring MDM2 and TP53 levels over time. Consistently, our Western blot, RT-qPCR, and TCGA data showed no significant correlation between USP15 and TP53, suggesting that USP15’s regulation of MDM2/TP53 may be context- or cell type–dependent.

ICI therapy has revolutionized the treatment of colorectal cancer and offers hope for long-term survival in metastatic patients with colorectal cancer (51–54). However, its effectiveness remains limited. Although many studies have highlighted the roles of molecules such as VEGF-A, TGF-β, and MAPK in promoting resistance to colorectal cancer immunotherapy, clinical results have been unsatisfactory (55, 56). To evaluate the potential of targeting USP15 to enhance immunotherapy efficacy for colorectal cancer, we applied USP15-IN-1 alone or alongside PD-1 mAb in murine models, demonstrating a remarkable therapeutic effect of the combination strategy. Using colorectal cancer tissue samples, we validated the stability of the USP15/SMYD3/CCL2 signaling axis and its relevance to the suppressive immune microenvironment. Finally, in a clinical cohort of patients with colorectal cancer undergoing immunotherapy, we confirmed that high USP15 expression is linked to a poor response to immunotherapy, including lower ORRs, shorter PFS, and shorter OS. These results affirm the potential value of USP15 in enhancing the efficacy of immunotherapy for patients with colorectal cancer.

Conclusion

Targeting protein ubiquitination represents a promising strategy for antitumor therapy. As a prominent deubiquitinase in cancer treatment, USP15 has the capacity to improve the sensitivity of immunotherapy for colorectal cancer. Our study demonstrated that USP15 is overexpressed in colorectal cancer and contributes to tumor progression and resistance to immunotherapy by promoting MDSC recruitment. Mechanistically, USP15 deubiquitinates SMYD3, activates the SMYD3/H3K4me3/CCL2 signaling axis, and recruits MDSCs. In summary, this study provides a comprehensive analysis of USP15 expression in colorectal cancer and its immunomodulatory role in promoting MDSC recruitment, offering a new potential target for enhancing immunotherapy.

Supplementary Material

Supplementary Table 1

Clinical information of Cohort 1 comprising 78 patients.

Supplementary Table 2

Clinical information of Cohort 2 comprising 93 patients.

Supplementary Table 3

CyTOF Antibodies Table.

Supplementary Figure 1

USP15 is upregulated and correlates with poor prognosis in CRC.

Supplementary Figure 2

USP15 promotes tumor growth in CRC immunocompetent mice models.

Supplementary Figure 3

USP15 induces high MDSC infiltration and low CD8+T cells infiltration in the CRC microenvironment.

Supplementary Figure 4

Immune cell composition and macrophage polarization in CT26-WT and CT26-Usp15 KO subcutaneous tumors.

Supplementary Figure 5

MDSC is indispensable for remodeling USP15-induced suppressive TIME.

Supplementary Figure 6

Validation of the relationship between USP15 and TP53/MDM2 expression in CRC cells and tumors.

Supplementary Figure 7

USP15 facilitates the stability of the SMYD3 protein by deubiquitination.

Supplementary Figure 8

SMYD3 promotes CCL2 expression via trimethylation of H3K4 in CRC.

Supplementary Figure 9

Blocking USP15 augments the anti-PD-1 therapy in CRC Models through MDSCs reprogramming.

Supplementary Figure 10

The effect of USP15 inhibitor on the proliferation of colorectal cancer cells.

Supplementary Figure 11

Graphical abstract.

Authors’ Disclosures

No disclosures were reported.

Authors’ Contributions

S. Han: Conceptualization, data curation, software, formal analysis, writing–original draft, writing–review and editing. L. Cui: Data curation, software, visualization, project administration. B. Wang: Resources, formal analysis, supervision, investigation. Y. Ruan: Data curation, investigation, visualization, methodology. M. Shi: Software, formal analysis, investigation, visualization. C. Hong: Investigation, visualization, methodology, writing–original draft. X. Guan: Data curation, formal analysis, validation. Z. Chen: Methodology, writing–original draft. Y. Li: Data curation, formal analysis, supervision. Y. Liao: Visualization, methodology. M. Ma: Supervision. X. Lu: Funding acquisition, investigation, methodology. H. Wang: Validation, investigation. Y. Zhang: Conceptualization, supervision, funding acquisition, project administration. C. Liu: Conceptualization, resources, funding acquisition, writing–original draft, project administration.