FAM135B Deficiency Inhibits Cytotoxic T-cell Activity in Triple-Negative Breast Cancer by Blocking the IFI16-Dependent STING Pathway

Immunotherapy shows limited efficacy against triple-negative breast cancer (TNBC). The authors identify FAM135B as a regulator of antitumor immunity in TNBC through its ability to modulate the IFI16-mediated STING pathway, offering new therapeutic opportunities.

Abstract

Immune checkpoint blockade (ICB) has improved outcomes for patients with several types of cancer. However, only a minority of patients with triple-negative breast cancer (TNBC) derive benefits, and the underlying mechanisms remain largely unknown. In this study, we identified family with sequence similarity 135 member B (FAM135B) as a regulator of antitumor immunity in TNBC. Single-cell sequencing data and functional assays demonstrated the critical role of FAM135B in activating cytotoxic T cells and improving the efficacy of ICB treatment by stimulating the STING pathway. Specifically, we found that FAM135B interacts with IFI16, inhibiting its ubiquitination and proteasomal degradation by competitively blocking its binding to the E3 ligase TRIM21. This initiated IFI16-dependent STING signaling, which ultimately led to increased cytotoxic T-cell activity. Deubiquitination of IFI16 at lysine 143 and lysine 561 was crucial for FAM135B-mediated activation of the STING pathway. These findings reveal that FAM135B regulates the IFI16-dependent STING pathway and subsequent immune activation. FAM135B may represent a potential predictor of ICB therapeutic responses for patients with TNBC.

Introduction

Triple-negative breast cancer (TNBC) is characterized by a lack of estrogen receptor and progesterone receptor expression, as well as the absence of HER2 amplification (1). This subtype of breast cancer is the most aggressive among the different subtypes and exhibits the worst prognosis (2). Because of limited therapeutic targets, chemotherapy is currently the standard treatment for TNBC (3). However, the efficacy of these cytotoxic agents in patients with TNBC is often suboptimal, highlighting the need to develop more effective treatment strategies. The emergence of immune checkpoint blockade (ICB) therapies targeting PD-L1 and PD-1 has revolutionized the treatment of many cancers, yielding substantial improvement in patient outcomes (4, 5). Nonetheless, only approximately 15% of patients with early-stage TNBC derive durable benefits from ICB in combination with other treatments (6). Consequently, elucidating the mechanisms underlying immune responses in TNBC is vital for identifying informative biomarkers for ICB response and developing effective combination therapy strategies.

Tumor-infiltrating lymphocytes are increasingly recognized as important prognostic indicators and determinants of responses to immunotherapy for patients with TNBC (7–9). Although tumor-infiltrating lymphocytes predominantly consist of T cells, single-cell RNA sequencing of primary breast tumors has revealed substantial heterogeneity within the infiltrating T-cell population (10). Notably, TNBC has been established to exhibit a higher degree of T-cell infiltration compared with other breast cancer subtypes, correlating with an enhanced efficacy of ICB treatment (9). However, a subset of patients with TNBC demonstrates resistance to immunotherapy because of diminished T-cell infiltration and increased T-cell dysfunction (11, 12). Therefore, identifying and investigating genes involved in the activation of cytotoxic T cells are crucial for improving the effectiveness of immunotherapy in TNBC.

Family with sequence similarity 135 member B (FAM135B) has emerged as an oncogene that is frequently mutated and correlates with poor prognosis in esophageal squamous cell carcinoma and colorectal cancer (13–16). Our previous work demonstrated that FAM135B induces DNA damage repair, resulting in chemotherapy insensitivity in colorectal cancer (15). Moreover, increasing evidence supports that cumulative DNA damage is an important determinant of tumor immunogenicity, thereby enhancing the antitumor immune response primarily by recruiting and activating tumor-specific CD8⁺ T cells (17, 18). In this study, we explored the correlation between FAM135B expression and immune cell infiltration in TNBC. We performed in vitro functional assays, along with in vivo animal models, to determine the impact of FAM135B on modulating antitumor immunity. Our work established a mechanism by which FAM135B remodeled the immune microenvironment in TNBC and led us to propose FAM135B as a potential predictor of responses to ICB therapy in TNBC.

Materials and Methods

Patients and clinical specimens

Consecutive sections of 62 tumor tissues from patients with TNBC and 12 tumor tissues from patients with TNBC undergoing immunotherapy were obtained from Nanfang Hospital of Southern Medical University and stored at 4°C before processing. Serum samples were obtained from healthy volunteers. The study was approved by the Nanfang Hospital Ethics Committee of Southern Medical University (NFEC-2025-165) and conducted in compliance with the Declaration of Helsinki. Written informed consent was obtained from all participants or their legal guardians before sample collection.

Cell lines and cell culture

Human breast cancer cell lines BT549 (RRID: CVCL_1092), MDA-MB-231 (RRID: CVCL_0062), HCC1806 (RRID: CVCL_1258), BT474 (RRID: CVCL_0179), MDA-MB-468 (RRID: CVCL_0419), and MCF7 (RRID: CVCL_0031); mouse breast cancer cell lines 4T1 (RRID: CVCL_0125) and EMT6 (RRID:CVCL_1923); and HEK293T (RRID: CVCL_0063) cells were purchased from the Cell Bank of the Chinese Academy of Sciences with short tandem repeat certifications. BT549, HCC1806, 4T1, and EMT6 cells were cultured in RPMI-1640 medium (KeyGen) supplemented with 10% FBS (Gibco). MDA-MB-231, MDA-MB-468, BT474, MCF7, and HEK293T cells were maintained in DMEM (KeyGen) with 10% FBS (Gibco). All cells were incubated at 37°C in a humidified atmosphere containing 5% CO₂ and used for experiments within five passages. Regular testing for Mycoplasma contamination was conducted using the GMyc-PCR Mycoplasma Test Kit (Yeasen). The FAM135B plasmid and short hairpin RNAs (shRNA) were constructed and packaged using a lentiviral expression system (GeneChem). BT549 and MDA-MB-231 cells were infected with the lentiviruses and selected using neomycin (2 mg/mL). The concentrations of the chemicals for cell treatment were 20 μmol/L for MG132 (MedChemExpress), 10 μmol/L for cycloheximide (MedChemExpress), and 5 μmol/L for H-151 (MedChemExpress).

Mice and syngeneic mouse models

BALB/c mice (RRID: MGI:2161072) and BALB/c nude mice (RRID: MGI:5649750; female, 6 weeks old) were purchased from the Southern Medical University Experimental Animal Center and the Guangdong Medical Laboratory Animal Center. Mice were housed under specific pathogen-free conditions with a standard 12-hour light–dark cycle and had free access to water and food. All animal experiments were performed in accordance with ethical and humane guidelines and authorized by the Institutional Animal Care and Use Committee of the Experimental Animal Center, Southern Medical University (IACUC-LAC-20241008-001). 4T1 or EMT6 cells (5 × 10⁵ cells per mouse) were injected into the right mammary fat pads of each BALB/c mouse. Six days after implantation, H-151 (750 nmol per mouse, MedChemExpress) was administered intraperitoneally every 2 days, cGAMP (5 mg/kg, MedChemExpress) was administered intraperitoneally every 3 days, or anti–mouse PD-1 (200 μg per mouse, Selleckchem, A2122) was administered intraperitoneally every 3 days. After 3 weeks, the mice were euthanized by cervical dislocation, and the xenograft tumors were quickly harvested for histologic analysis. Tumor weight was measured, and tumor volume was calculated using the formula volume (mm³) = width² (mm²) × length (mm)/2.

Plasmids, shRNAs, and transfection

Flag-FAM135B (P20267), HA-IFN-inducible protein 16 (IFI16; P70437), Flag-Tripartite motif protein 21 (TRIM21; P63528), Myc-Ubi (P59265), and GST-IFI16 (P68624), along with mutants of Flag-FAM135B, HA-IFI16, and Myc-Ubi plasmids, were obtained from MiaoLing Biology. These plasmids were amplified and purified using the EndoFree Mini Plasmid Kit (TIANGEN). shRNAs targeting FAM135B were constructed, cloned into the pLV vector, and provided by Hanyi Biotech. The sequences of shRNA expression plasmids are available in Supplementary Table S1. All plasmids and shRNAs were transfected into cells with Lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions.

IHC staining

Paraffin-embedded tissue sections from patients with TNBC and tumor-bearing mice were deparaffinized and rehydrated. Sections were subsequently subjected to antigen retrieval with Tris/EDTA buffer (pH 9) or citrate buffer (pH 6) at 95°C for 20 minutes. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 15 minutes at room temperature. Sections were then incubated with 5% BSA to block nonspecific binding and incubated overnight at 4°C with the specified primary antibodies: anti-FAM135B (28061-1-AP, Proteintech), anti–human CD8 (ZA-0508, ZSGB-Bio), anti–mouse Ki67 (ab279653, Abcam), anti–mouse CD8α (98941, Cell Signaling Technology), and anti–mouse granzyme B (46890, Cell Signaling Technology). Following this, the sections were incubated with horseradish peroxidase–conjugated secondary antibody (PV-6000, ZSGB-Bio) at room temperature for 1 hour, stained with DAB substrates, subjected to hematoxylin staining, dehydrated, and finally mounted for microscopic analysis. High-resolution whole-slide imaging was performed using the K-Viewer slide scanner. The density of CD8 staining was quantified using ImageJ software by the number of CD8⁺ cells per square millimeter for each sample. In parallel, the expression level of FAM135B was scored based on staining intensity (0 for negative, 1 for weak, 2 for moderate, and 3 for strong) and the percentage of stained cells (0 if no cell had staining, 1 if 0% to 25%, 2 if 25% to 50%, 3 if 50% to 75%, and 4 if more than 75% of cells had staining). The final score was calculated by multiplying the intensity and percentage scores, resulting in a range from 0 to 12, with a cutoff for high FAM135B expression established at 6. The staining was scored by two pathologists blinded to the study results. A list of primary antibodies and detailed information are available in Supplementary Table S2.

Immunofluorescence analysis

BT549 and MDA-MB-231 cells seeded in glass-bottom dishes were fixed in 4% formaldehyde for 15 minutes and then permeabilized with 0.5% Triton X-100 for 20 minutes. For tissue sections from patients with TNBC, the sections were deparaffinized, rehydrated, and then subjected to antigen retrieval using Tris/EDTA buffer (pH 9) or citrate buffer (pH 6) at 95°C for 20 minutes. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 15 minutes at room temperature. After blocking with 1% BSA for 1 hour, the cells or sections were incubated overnight at 4°C with the primary antibodies targeting specific proteins: anti-FAM135B (28061-1-AP, Proteintech), anti-IFI16 (sc-8023, Santa Cruz Biotechnology), and anti–IFN regulatory factor 3 (IRF3; 11312-1-AP, Proteintech). Fluorescent-conjugated secondary antibodies (SA00013-2 or SA00013-3, Proteintech) were subsequently used for staining the cells, and the nuclei were costained with DAPI. Images were captured using a confocal microscope (FV3000, Olympus) and processed using Imaris software (version 9.0.1). A list of primary antibodies and detailed information are available in Supplementary Table S2.

Co-immunoprecipitation and mass spectrometry

Cells were lysed and then immunoprecipitated using an antibody against FAM135B or IFI16 overnight at 4°C. Following this, protein A/G magnetic beads (Selleckchem) were added, and the mixture was incubated for 4 hours at 4°C. Next, the immunoprecipitants were washed with lysis buffer and then eluted by boiling in loading buffer containing SDS for 5 minutes at 100°C. The samples were subsequently subjected to mass spectrometry (MS) analysis (Wininnovate Bio). The lyophilized peptide fractions were re-suspended in ddH₂O containing 0.1% formic acid, and 2 μL aliquots of these were loaded into a NanoViper C18 trap column (Acclaim PepMap 100, 75 μm × 2 cm). Online chromatography separation was performed on the Easy nLC 1200 system (Thermo Fisher Scientific). The trapping and desalting procedures used a volume of 20 μL of 100% solvent A (0.1% formic acid). Next, an elution gradient of 5% to 38% solvent B (80% acetonitrile and 0.1% formic acid) over 60 minutes was applied on an analytic column (Acclaim PepMap RSLC, 75 μm × 25 cm, C18, 2 μm, and 100 Å). Data-dependent acquisition MS was used to acquire tandem MS data on a Thermo Fisher Scientific Q Exactive mass spectrometer equipped with a Nano Flex ion source. For a full MS survey scan, the target value was 3 × 10⁶, with a scan range from 350 to 2,000 m/z, a resolution of 70,000, and a maximum injection time of 100 ms. For the MS2 scan, only spectra with a charge state of two to five were selected for fragmentation by higher-energy collision dissociation with a normalized collision energy of 28. The MS2 spectra were acquired in the ion trap in rapid mode, with an AGC target of 8,000 and a maximum injection time of 50 ms. Dynamic exclusion was set for 25 seconds. MS/MS spectra were searched against the UniProt Homo Sapiens Reference Proteome dataset using Mascot v2.5.1 (Matrix Science).

Transcriptome sequencing and qRT-PCR

Total RNA was extracted from MDA-MB-231 cells that were either transfected with a control vector or the FAM135B plasmid, using VeZol reagent (Vazyme). The RNA concentration was quantified at an optical density of 260 nm (NanoDrop Technology). RNA quality was assessed with the RNA 6000 LabChip kit (Agilent Technologies). Purified mRNA was fragmented into small pieces using a fragmentation buffer at the appropriate temperature. Next, first-strand cDNA was generated through random hexamer-primed reverse transcription, followed by the synthesis of second-strand cDNA. RNA Index Adapters and an A-Tailing Mix were then added, and the mixture was incubated for end repair. The resulting cDNA fragments were amplified using PCR, and the PCR products were purified with AMPure XP Beads before being dissolved in an EB solution. The quality of the products was validated using an Agilent Technologies 2100 bioanalyzer. The double-stranded PCR products obtained from the previous steps were denatured by heating and subsequently circularized using the splint oligo sequence to create the final library. This RNA library was further amplified using phi29 to generate DNA nanoballs. The DNA nanoballs were then loaded into a patterned nano-array, in which paired-end reads of 150 bp were generated on the T7 platform (Benagen Technology). The raw sequencing data were initially processed with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to filter out adapters and low-quality sequences. The cleaned reads were then mapped to the human genome (hg38) using HISAT. Gene expression levels for each sample were calculated using the RSEM package and expressed as fragments per kilobase of transcript per million mapped fragments. To identify differentially expressed genes (DEG), the gene expression levels in TNBC cells overexpressing FAM135B were compared with those in control cells. DEGs were determined based on the criteria of a gene expression change greater than 1.2-fold and a P value less than 0.05, utilizing three biological replicates.

For qRT-PCR analysis, total RNA was reverse transcribed into cDNA using HiScript IV RT Super Mix (Vazyme). qRT-PCR was conducted on the Applied Biosystems 7500 Sequence Detection system using SYBR Green PCR Master Mix (Vazyme). The PCR conditions were as follows: an initial denaturation step at 95°C for 30 seconds, followed by 45 cycles of 95°C for 10 seconds, and 60°C for 30 seconds. Relative mRNA levels of the targeted genes were calculated via the 2^−ΔΔCT method and normalized to those of GAPDH. The primer sequences used for qRT-PCR are available in Supplementary Table S3.

Immunoblotting analysis

Cells were lysed in RIPA lysis buffer (Beyotime Biotechnology) supplemented with protease inhibitor cocktail (GK10014, GlpBio) and phenylmethylsulfonylfluoride (BB-3341, BestBio). The lysates were then boiled in SDS sample loading buffer. Equal amounts of protein were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore). The membranes were blocked in PBS buffer supplemented with 0.1% Tween containing 5% nonfat milk for 1 hour at room temperature, followed by overnight incubation with the specified primary antibodies at 4°C: anti-FAM135B (28061-1-AP, Proteintech), anti-IFI16 (YT2274, Immunoway), anti-TRIM21 (12108-1-AP, Proteintech), anti-Flag (20543-1-AP, Proteintech), anti-HA (51064-2-AP, Proteintech), anti-MYC (16286-1-AP, Proteintech), anti–β-tubulin (66240-1-Ig, Proteintech), anti-γH2AX (YP0128, Immunoway), anti-Ubiquitin (43124, Cell Signaling Technology), anti–TANK-binding kinase 1 (67211-1-Ig, Proteintech), anti-stimulator of interferon genes (19851-1-AP, Proteintech), anti-IRF3 (11312-1-AP, Proteintech), anti–p-TBK1 (AF8190, Affinity), anti–p-STING (AF7416, Affinity), and anti–p-IRF3 (80519-2-RR, Proteintech). After washing with PBS buffer supplemented with 0.1% Tween, the membranes were incubated with horseradish peroxidase–conjugated antibodies for 1 hour at room temperature. The immunoblots were visualized using the enhanced chemiluminescence substrate (Epizyme) with a 4200SF detector (Tanon). A list of primary antibodies and detailed information are available in Supplementary Table S2.

Flow cytometric analysis

Human peripheral blood mononuclear cells (PBMC) were isolated from serum samples obtained from healthy volunteers using human lymphocyte separation medium (P8610, Solarbio). PBMCs were maintained in RPMI-1640 medium supplemented with 10% FBS and 10 ng/mL human IL2 (MedChemExpress) for 3 days before coculture with tumor cells. TNBC cells were seeded in 12-well plates with complete medium, and the PBMCs were added to coculture with the TNBC cells at a 3:1 ratio for 48 hours. Six hours prior to cell collection, 5 ng/mL Brefeldin A (MedChemExpress) was added to inhibit cytokine secretion. The T cells were collected and washed twice with ice-cold PBS. Mouse tumor tissues were minced and digested at 37°C for 30 minutes in RPMI-1640 medium supplemented with 1 mg/mL collagenase IV (BioFroxx), 1 mg/mL DNase I (BioFroxx), and 0.5 mg/mL hyaluronidase (BioSharp). Single-cell suspensions were ground and filtered through 70-μm cell strainers, followed by two additional washes with ice-cold PBS. For surface staining, cells were incubated with fluorochrome-conjugated antibodies for 30 minutes at 4°C: Alexa Fluor 700 anti–human CD45 (368514, BioLegend), PerCP/Cy5.5 anti–human CD3 (317336, BioLegend), APC/Cy7 anti–human CD8 (344714, BioLegend), APC anti–mouse CD45 (103112, BioLegend), FITC anti–mouse CD3 (100204, BioLegend), and APC/Cy7 anti–mouse CD8 (100714, BioLegend). For intracellular staining, cells were fixed and permeabilized using the Fixation/Permeabilization Kit (BD Biosciences) and subsequently stained with Brilliant Violet 421 anti–granzyme B (GZMB; 396414, BioLegend). The stained cells were analyzed using a flow cytometry system (BD Biosciences), and the data were analyzed using FlowJo software (version 10.0). A list of primary antibodies and detailed information are available in Supplementary Table S2.

GST pull-down

The GST-IFI16 plasmids were transformed into Escherichia coli strain BL21 using the heat-shock method (42°C for 45 seconds). When the cell density reached the optical density at 600 nm value of 0.6, the expression of GST or GST-fusion proteins was induced with IPTG (Yeasen) at 16°C overnight. Cells were collected and lysed using ultrasonic disruption in sterile PBS supplemented with a protease inhibitor cocktail and phenylmethylsulfonylfluoride, all while maintaining an ice-water bath. Subsequently, the mixture was centrifuged at 12,000 rpm for 30 minutes at 4°C, and the clear lysates were incubated with glutathione magnetic agarose beads at 4°C for 1 hour. Immunoprecipitants were washed with PBS and then incubated with cell lysates from HEK293T cells transfected with Flag-FAM135B at 4°C overnight. The protein complexes bound to the beads were washed five times, eluted by boiling in SDS sample loading buffer at 95°C for 5 minutes, and then subjected to immunoblotting analysis.

Molecular docking

The protein sequences for FAM135B and IFI16 were obtained from the UniProt database and submitted to the AlphaFold3 server to predict their binding model. Interatomic interactions between FAM135B and IFI16 are visualized using PyMOL software.

Analysis of somatic mutation

The somatic mutation data from The Cancer Genome Atlas (TCGA) PanCancer Atlas were downloaded (https://gdc.cancer.gov/about-data/publications/pancanatlas). TNBC samples were defined as those with negative expression of estrogen receptor, progesterone receptor, and HER2, and 107 samples were selected for genomic profiling. Next, the maftools package in R (version 4.4.2) was used for somatic mutation analysis, and the oncoplot function was used to visualize the mutation rate and mutation type of significantly mutated genes in TNBC.

Analysis of bulk RNA sequencing data

TCGA breast cancer (BRCA) patients in the UCSC Xena database, METABRIC dataset in the cBioPortal database, and breast cancer datasets in the Gene Expression Omnibus database were used to analyze the expression of FAM135B in TNBC. The immune cell composition analysis was performed using the R package IOBR. Gene Ontology (GO) and Reactome pathway enrichment analyses were executed using the DAVID database (https://davidbioinformatics.nih.gov/). Gene set enrichment analysis (GSEA) pathway analysis was conducted using GSEA software.

Analysis of single-cell RNA sequencing data

Single-cell RNA sequencing (RNA-seq) data and paired bulk RNA-seq data of eight TNBC samples were obtained from the public dataset GSE176078 in the Gene Expression Omnibus database and analyzed using the Seurat package in R (version 4.4.2). These eight TNBC samples were divided into two subgroups (CID4495, CID44971, CID4515, and CID44041 in the FAM135B-high group, and CID4513, CID4465, CID4523, and CID3946 in the FAM135B-low group) based on FAM135B expression as detected by bulk RNA-seq. Cells were filtered based on the following criteria: a minimum of 200 detected genes and no more than 20% mitochondrial reads per cell. Cells with excessively high numbers of reads or detected genes were also filtered out to reduce the likelihood of doublets. Genes expressed in fewer than three cells across individual samples were excluded. Multiple single-cell samples were integrated, and the batch effect was corrected using the Harmony algorithm. We employed the Uniform Manifold Approximation and Projection algorithm for dimensionality reduction and visualization. The “FindAllMarkers” function was used to detect the cluster-specific expressed genes, which were then annotated for cell type classification. The annotation of diverse T-cell subclusters referenced a single-cell–resolved pan-cancer study of tumor-infiltrating T cells conducted by Zheng and colleagues (19). The gene signatures associated with T-cell proliferation (GO:0042098), activation (GO:0042110), and cytotoxicity (GO:0001913) were sourced from the GO subset using MsigDB. Single-cell signature scores were calculated using the “AddModuleScore” function in the Seurat package.

Statistical analyses

All experiments were performed in three independent replicates, excluding the xenograft mouse model. The biological replicates for the animal experiments varied between three and six. Animals were randomly assigned to experimental groups utilizing blinding methods. Statistical analyses were conducted using GraphPad Prism 9.0 software, with data presented as means ± SD or SEM. For comparisons between two independent groups, the unpaired two-tailed Student t test was utilized. To compare multiple groups, the one-way ANOVA was employed. P value < 0.05 was considered statistically significant, with specific P values denoted in the figures (****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, P > 0.05).

Results

FAM135B expression positively correlates with T-cell infiltration and cytotoxic activity in TNBC

Considering the significance of the FAM135B mutation in tumor progression, we performed genome analyses on 107 TNBC samples from the TCGA cohort. We found that, in addition to six notably mutated genes (TP53, PIK3CA, PTEN, BRCA1, RB1, and BRCA2), the FAM135B mutation was present in 4% of the cases, suggesting its potential involvement in TNBC tumorigenesis (Supplementary Fig. S1A). We then examined the expression of FAM135B in TNBC samples from the TCGA breast cancer cohort, METABRIC dataset, and GSE65194 dataset. We observed that the mRNA levels of FAM135B were substantially reduced in TNBC tissues compared with normal tissues and non-TNBC tumor tissues (Fig. 1A and B). Immune cell infiltration analysis using bulk sequencing data from the GSE137356 dataset indicated a positive correlation between the infiltration of CD8⁺ T cells and NK cells and the expression of FAM135B in TNBC. In contrast, M2 macrophages were more abundant in patients with low FAM135B expression (Fig. 1C; Supplementary Fig. S1B and S1C). To further elucidate the association between FAM135B expression and immune cell infiltration in TNBC, we analyzed eight TNBC samples that were simultaneously detected by single-cell and bulk RNA-seq. These samples were divided into two groups, with four samples in each group, based on FAM135B expression as determined by bulk RNA-seq (Fig. 1D). We performed unsupervised clustering analysis to identify the major cell clusters in FAM135B^high and FAM135B^low tumors (Fig. 1E). In the FAM135B^high tumors, we observed a significant increase in the populations of CD4⁺ and CD8⁺ T cells, B cells, and plasmablasts compared with the FAM135B^low tumors. In contrast, the proportion of myeloid cells was markedly decreased (Fig. 1F). Subsequent reclustering of the T-cell populations identified four distinct subpopulations: effector memory T cells (Tem), exhausted T cells, naïve T cells, and tissue-resident memory T cells (Fig. 1G). The proportion of Tem cells was increased in FAM135B^high tumors, whereas the percentage of exhausted T cells was reduced (Fig. 1H). Functional analysis of Tem cells and naïve T cells showed that genes involved in T-cell proliferation, activation, and cytotoxicity were enriched in FAM135B^high tumors (Fig. 1I). Similarly, the protein levels of FAM135B were positively associated with the infiltration of CD8⁺ T cells in human TNBC tissues (Fig. 1J).

Figure 1.

FAM135B expression is positively correlated with T-cell infiltration and cytotoxic activity in TNBC. A, FAM135B mRNA expression in TNBC and normal tissues. B, mRNA levels of FAM135B in TNBC and non-TNBC tissues. C, Box plots showing the CD8⁺ T-cell infiltration score indicated by CIBERSORT in each sample, grouped by FAM135B expression in the GSE137356 dataset. D, Eight TNBC samples from the GSE176078 dataset were divided into two subgroups based on FAM135B expression as detected by bulk RNA-seq. E, Uniform Manifold Approximation and Projection plot illustrating patients with high FAM135B expression (left, n = 4 samples) vs. low expression (right, n = 4 samples) from the GSE176078 dataset. F, Bar plot of proportional differences in annotated cell types between the FAM135B^high and FAM135B^low groups. G, Reclustering of CD8⁺ T lymphocytes with Uniform Manifold Approximation and Projection visualization of their subtypes. H, Bar plot depicting proportional differences in CD8⁺ T lymphocytes between the FAM135B^high and FAM135B^low groups. I, Bubble plot of functional analysis of CD8⁺ Tem and CD8⁺ naïve T cells. J, Consecutive sections of 62 human TNBC tissues were immunostained with anti-FAM135B and anti-CD8. Representative images are shown (left; scale bar, 20 μm). The density of CD8 staining was quantified by ImageJ (right, n = 44 patients for the FAM135B^high group and n = 18 patients for the FAM135B^low group). *, P < 0.05; ***, P < 0.001; ****, P < 0.0001. Tex, exhausted T cell; Tn, naïve T cell; Trm, tissue-resident memory T cell; UMAP, Uniform Manifold Approximation and Projection.

FAM135B overexpression inhibits tumor growth by inducing activation of cytotoxic T cells

To determine the biological function of FAM135B in TNBC, we detected its expression in TNBC cell lines. We selected the TNBC cell line BT549 with relatively high FAM135B expression and the MDA-MB-231 cell line with relatively low FAM135B expression for subsequent analyses (Supplementary Fig. S2A). Next, we transfected these cells with either an empty vector, the FAM135B plasmid, or FAM135B shRNAs and assessed transfection efficiency through immunoblotting analyses (Supplementary Fig. S2B and S2C). To confirm the promoting effect of FAM135B on the infiltration and activation of CD8⁺ T cells in TNBC, the BT549 and MDA-MB-231 cells overexpressing FAM135B were cocultured with activated T cells. Flow cytometry analysis revealed an increase in the population of CD8⁺ T cells among CD3⁺ T cells after coculturing with TNBC cells that overexpressed FAM135B (Fig. 2A; Supplementary Fig. S3A). Additionally, we observed an elevated proportion of GZMB-expressing CD8⁺ T cells following coculture with FAM135B-overexpressing TNBC cells (Fig. 2B; Supplementary Fig. S3A). However, the populations of CD8⁺ T cells and CD8⁺ GZMB⁺ T cells within CD3⁺ T cells remained largely unchanged after coculturing with non-TNBC cell lines, specifically BT474 and MCF7 (Supplementary Fig. S3B and S3C). This suggests a specific stimulatory effect of FAM135B on the activation of cytotoxic T cells in TNBC. To further examine the impact of FAM135B overexpression on TNBC tumorigenesis, we injected 4T1 and EMT6 cells expressing FAM135B into the mammary fat pads of female immunocompetent BALB/c mice and immunodeficient nude mice. We found that the overexpression of FAM135B substantially suppressed tumor growth in BALB/c mice, whereas no obvious effect was observed in nude mice (Fig. 2C and D; Supplementary Fig. S4A and S4B), indicating the critical involvement of T cells in tumor suppression. To determine the impact of FAM135B overexpression on antitumor immunity, we analyzed orthotopic tumor tissues from BALB/c mice using flow cytometry. We found that FAM135B overexpression increased the infiltration of CD8⁺ T cells and induced the activation of cytotoxic CD8⁺ T cells (Fig. 2E and F; Supplementary Fig. S4C). Accordingly, immunostaining of CD8α and GZMB in mouse TNBC tissues demonstrated the stimulatory effect of FAM135B overexpression on the infiltration and cytotoxic activity of CD8⁺ T cells. The inhibitory effect of FAM135B overexpression on tumor growth was further confirmed by immunostaining of Ki67 in mouse TNBC tissues (Fig. 2G; Supplementary Fig. S4D).

Figure 2.

FAM135B overexpression inhibits tumor growth by inducing the activation of cytotoxic T cells. A and B, BT549 and MDA-MB-231 cells expressing an empty vector or FAM135B were cocultured with activated T cells for 48 hours, and the T cells were then subjected to flow cytometry analysis. Representative flow cytometry plots and statistical quantification of CD8⁺ cells (A) and GZMB⁺ CD8⁺ cells (B) in CD3⁺ tumor-infiltrating lymphocytes are shown (mean ± SD, n = 3). C and D, 4T1 cells (5 × 10⁵ cells per mouse) expressing an empty vector or FAM135B were injected into the mammary fat pads of female BALB/c mice (C) or BALB/c nude mice (D). Three weeks after injection, the mice were humanely euthanized. Tumor weight was recorded, and tumor volume was calculated (mean ± SD, left, n = 6 BALB/c mice and n = 4 BALB/c nude mice). Resected tumors from each group are shown (scale bar, 1 cm). E and F, Flow cytometry analysis revealed the population of CD8⁺ cells (E) and GZMB⁺ CD8⁺ cells (F) in CD3⁺ tumor-infiltrating lymphocytes from mouse TNBC tissues. Representative plots and quantification of flow cytometry are shown (mean ± SD, n = 3). G, Immunostaining of Ki67, CD8ɑ, and GZMB in mouse orthotopic tumor tissues. Representative images are shown (Scale bar, 20 μm). *, P < 0.05; **, P < 0.01; ****, P < 0.0001; ns, P > 0.05 (not significant). Nu, nude mice; SSC, side scatter.

FAM135B activates the STING–type I IFN pathway

To understand the underlying mechanism by which FAM135B promotes the cytotoxic activity of T cells, we performed RNA-seq analysis on MDA-MB-231 cells after FAM135B overexpression. DEGs were identified (Supplementary Fig. S5A). Reactome and GO analysis showed that these DEGs were significantly enriched in IFNα/β signaling and type I IFN production pathways (Fig. 3A and B). GSEA of TNBC cells and the GSE21653 dataset also revealed a positive correlation between high expression of FAM135B and STING signaling, as well as type I IFN signaling, implying the role of FAM135B in the STING–type I IFN pathway activity (Fig. 3C; Supplementary Fig. S5B and S5C). We next demonstrated increased mRNA levels of target genes involved in the STING–type I IFN pathway in TNBC cells after overexpressing FAM135B (Fig. 3D). Immunoblotting analysis also confirmed elevated phosphorylation levels of proteins associated with the STING pathway in BT549 and MDA-MB-231 cells after FAM135B overexpression (Fig. 3E). Consequently, depletion of FAM135B inhibited STING–type I IFN signaling in TNBC cells (Fig. 3F and G). Immunostaining analysis further revealed that FAM135B overexpression facilitated the nuclear translocation of IRF3 (Fig. 3H), a crucial transcription factor involved in cytokine and chemokine production in TNBC cells (22). These cytokines and chemokines, including IFNβ, C–C motif chemokine ligand 5 (CCL5), and C–X–C motif chemokine ligand 10 (CXCL10), are vital for the infiltration and activation of CD8⁺ T lymphocytes within the tumor microenvironment.

Figure 3.

FAM135B activates the STING–type I IFN pathway. A and B, MDA-MB-231 cells expressing an empty vector or FAM135B were collected for RNA transcriptional sequencing (three replicates versus three replicates). The bubble plot presents the Reactome pathway analysis (A) and the GO analysis (B) of the DEGs in FAM135B-overexpressing versus control MDA-MB-231 cells. C, GSEA reveals the correlation between FAM135B expression and the STING signaling pathway in the GSE21653 dataset. D, The mRNA levels of the indicated genes were detected by qRT-PCR in BT549 and MDA-MB-231 cells overexpressing FAM135B (fold change of overexpression level was about 30 times; mean ± SEM, n = 3). E, The expression of the indicated proteins was detected by immunoblotting in BT549 and MDA-MB-231 cells after FAM135B overexpression. F, qRT-PCR analysis of the indicated genes in BT549 and MDA-MB-231 cells expressing FAM135B shRNAs (mean ± SEM, n = 3). G, Immunoblotting analysis of the indicated proteins in BT549 and MDA-MB-231 cells after FAM135B depletion. H, BT549 (left) and MDA-MB-231 (right) cells were immunostained with an anti-IRF3 antibody. Confocal microscopy images exhibited the subcellular localization of IRF3 (scale bar, 10 μm). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. NES, normalized enrichment score; p-IRF, phosphorylated IRF; p-STING, phosphorylated STING; p-TBK, phosphorylated TBK.

FAM135B interacts with the PYRIN domain of IFI16

Considering that the activation of the STING pathway relies on the recognition of intracellular DNA damage by DNA sensors, we investigated whether FAM135B affected DNA damage response in TNBC. Immunofluorescence staining with γH2AX foci revealed that the overexpression of FAM135B had no obvious effect on DNA repair in TNBC cells after ionizing radiation (Supplementary Fig. S6A), indicating that there might be an alternative mechanism by which FAM135B modulates the STING pathway. To clarify the molecular mechanism by which FAM135B regulates the STING pathway activity, we investigated what proteins were interacting with FAM135B. We conducted immunoprecipitation on MDA-MB-231 cell lysates using an anti-FAM135B antibody, and then the immunoprecipitants were analyzed by MS. Consistent with our previous results, functional analysis revealed that proteins correlated with FAM135B were enriched in the IRF3-mediated induction of type I IFN and the STING-mediated induction of host immune response (Fig. 4A). Thus, IFI16, an activator of STING-dependent type I IFN production, was recognized as a potential interacting protein of FAM135B (Fig. 4B; Supplementary Fig. S6B; refs. 22, 23). To validate this interaction, reciprocal immunoprecipitation assays were conducted, confirming the FAM135B–IFI16 interaction in BT549 and MDA-MB-231 cells (Fig. 4C). We also transfected Flag-FAM135B and HA-IFI16 plasmids into 293T cells and observed the exogenous interaction between these proteins (Fig. 4D). Double staining of FAM135B and IFI16 revealed their co-localization in TNBC cells (Fig. 4E). Moreover, the purified GST-tagged IFI16 protein was incubated with lysates from 293T cells after transfection with Flag-FAM135B. Subsequent immunoblotting analysis further supported the direct binding of FAM135B to IFI16 (Fig. 4F).

Figure 4.

FAM135B interacts with the PYRIN domain of IFI16. A, Bubble plot depicting the functional analysis of the protein list identified by MS. B, The mass spectra of IFI16. C, BT549 and MDA-MB-231 cell lysates were immunoprecipitated using anti-FAM135B or anti-IFI16 antibodies and then were analyzed by immunoblotting. D, 293T cells were co-transfected with Flag-FAM135B and HA-IFI16 plasmids. Cell lysates were immunoprecipitated with either anti-Flag or anti-HA antibodies and then were subjected to immunoblotting analysis using the indicated antibodies. E, BT549 and MDA-MB-231 cells were double-stained with anti-FAM135B and anti-IFI16 antibodies. Confocal microscopy images showed the subcellular localization of FAM135B and IFI16 (scale bar, 10 μm). F, The GST-IFI16 fusion protein was purified and incubated with cell lysates of 293T cells expressing Flag-FAM135B and then analyzed by immunoblotting. Purified GST proteins served as negative controls. G, Flag-tagged full-length FAM135B or truncation mutants were co-transfected with HA-IFI16 into 293T cells. Cell lysates were immunoprecipitated using an anti-Flag antibody and then analyzed by immunoblotting. H, HA-tagged full-length or deletion mutants of IFI16 were co-transfected with Flag-FAM135B into 293T cells. Cell lysates were immunoprecipitated with an anti-HA antibody and were subjected to immunoblotting analysis. FL, full length; IP, immunoprecipitated.

To explore the impact of FAM135B on its interaction with IFI16, we generated HA-tagged deletion mutants and Flag-tagged truncation mutants to examine the specific protein domains mediating this interaction. We observed that the PGAP1 and DUF3657 domains of FAM135B were involved in the binding to IFI16, suggesting that these domains are essential for the FAM135B–IFI16 binding (Fig. 4G). Additionally, deletion of the PYRIN domain of IFI16 almost abrogated its interaction with FAM135B, implying an indispensable role of the PYRIN domain in this interaction (Fig. 4H). Next, we employed AlphaFold3 to predict the protein–protein binding model and identified several residues within the PYRIN domain, including Try20, Leu28, Asp71, and Ile72, which were involved in the interaction with FAM135B (Supplementary Fig. S6C). To determine whether the interaction between FAM135B and IFI16 was mediated by DNA, we treated BT549 and MDA-MB-231 cells with ionizing radiation to induce DNA damage. Our findings indicated no significant difference in the binding between FAM135B and IFI16, thereby implying that FAM135B binds directly to the PYRIN domain of IFI16, independent of DNA (Supplementary Fig. S6D).

FAM135B competes with TRIM21 for binding to IFI16

The PYRIN domain of IFI16 is responsible for its interaction with the E3 ubiquitin ligase, prompting us to examine the impact of FAM135B on IFI16 expression (24). We observed that overexpression of FAM135B led to an increase in IFI16 protein levels in BT549 and MDA-MB-231 cells. In contrast, depletion of FAM135B resulted in a decrease in IFI16 expression (Supplementary Fig. S7A). Additionally, overexpression of FAM135B in TNBC cells inhibited the degradation of IFI16 under treatment with cycloheximide, indicating the role of FAM135B in regulating IFI16 degradation to maintain its protein stability (Fig. 5A and B). Moreover, treatment with MG132, a proteasome inhibitor, restored the reduced IFI16 levels in BT549 and MDA-MB-231 cells after the depletion of FAM135B, highlighting the effect of FAM135B on suppressing the proteasomal degradation of IFI16 (Supplementary Fig. S7B). Immunoblotting analysis further demonstrated that overexpression of FAM135B decreased the ubiquitination of IFI16 in BT549 and MDA-MB-231 cells (Fig. 5C). A consistent result was observed in 293T cells, suggesting that FAM135B might compete with an E3 ubiquitin ligase for binding to IFI16, thereby inhibiting its ubiquitination and subsequent proteasomal degradation (Supplementary Fig. S7C).

Figure 5.

FAM135B competes with TRIM21 for binding to IFI16. A and B, BT549 and MDA-MB-231 cells expressing an empty vector or FAM135B were treated with cycloheximide (CHX; 10 μmol/L) for the indicated time intervals, and cell lysates were subjected to immunoblotting using the indicated antibodies (A). The line graph shows the quantification of relative IFI16 levels (B). C, TNBC cells expressing an empty vector or FAM135B were treated with MG132 (20 μmol/L) for 6 hours. Cell lysates were immunoprecipitated using an anti-IFI16 antibody and then analyzed by immunoblotting. D, HA-tagged full-length (FL) or deletion mutants of IFI16 were co-transfected with Flag-TRIM21 into 293T cells. Cell lysates were immunoprecipitated using an anti-HA antibody and were subjected to immunoblotting analysis. E, BT549 and MDA-MB-231 cells expressing an empty vector or FAM135B were lysed, immunoprecipitated using an anti-TRIM21 antibody, and then analyzed by immunoblotting. F, Overlapping ubiquitination sites of IFI16 were identified by MS and estimated by the PhosphoSitePlus database. G, BT549 cells expressing an empty vector or TRIM21 were transfected with HA-IFI16 or its mutants (K143R or K561R). Cells were treated with MG132 (20 μmol/L) for 6 hours before harvest, and cell lysates were immunoprecipitated using an anti-HA antibody and analyzed by immunoblotting. H, TNBC cells expressing FAM135B shRNA were transfected with HA-IFI16 or its mutants (K143R or K561R). Cell lysates were analyzed by immunoblotting using the indicated antibodies. I, 293T cells were transfected with Flag-TRIM21, HA-IFI16, and Myc-Ubi or its mutants (K33R or K48R). Cells were treated with MG132 (10 μmol/L) for 6 hours before harvest, and cell lysates were immunoprecipitated using an anti-HA antibody, followed by immunoblotting analysis. IP, immunoprecipitated; p-IRF, phosphorylated IRF; p-STING, phosphorylated STING; p-TBK, phosphorylated TBK; WT, wild type.

To identify the E3 ubiquitin ligase for IFI16 in TNBC cells, we purified immunoprecipitants from BT549 cells using an anti-IFI16 and analyzed them by MS. TRIM21 was identified as the potential E3 ubiquitin ligase for IFI16 as it exhibited the highest confidence score in the MS data (Supplementary Fig. S8A). The UbiBrowser database also indicated that TRIM21 possessed the highest confidence prediction among the potential E3 ubiquitin ligases for IFI16 (Supplementary Fig. S8B). Reciprocal immunoprecipitation assays demonstrated the binding between IFI16 and TRIM21 in BT549 and MDA-MB-231 cells (Supplementary Fig. S8C). We also transfected 293T cells with HA-IFI16 and Flag-TRIM21 plasmids and observed their exogenous interaction (Supplementary Fig. S8D). Moreover, we examined the specific domains of the IFI16 protein relevant to its binding with TRIM21 and found that the PYRIN domain was required for the TRIM21–IFI16 binding (Fig. 5D). Immunoblotting analysis further confirmed that the overexpression of FAM135B in TNBC cells suppressed the interaction between IFI16 and TRIM21, accounting for the decreased ubiquitination of IFI16 and the increased protein stability (Fig. 5E).

To determine the specific ubiquitination sites on IFI16 by the E3 ubiquitin ligase TRIM21, we performed an MS analysis of ubiquitination on purified IFI16 protein. We identified lysine 143 (K143) and lysine 561 (K561) as candidate ubiquitination sites on IFI16, which have also been noted as potential ubiquitination sites in the PhosphoSitePlus database (Fig. 5F; Supplementary Fig. S8E). To examine whether K143 and K561 are the ubiquitination sites on IFI16 by the E3 ubiquitin ligase TRIM21, we constructed ubiquitination-defective mutants of IFI16, denoted as K143R and K561R, in which the lysine residues at positions 143 and 561 were substituted with arginine. We found that overexpression of TRIM21 increased the ubiquitination of wild-type IFI16 while showing no significant impact on the IFI16 mutants (Fig. 5G). Additionally, we observed decreased ubiquitination and reduced proteasomal degradation of the IFI16^K143R and IFI16^K561R mutants in TRIM21-overexpressing cells compared with wild-type IFI16, indicating that TRIM21 ubiquitinates IFI16 at residues K143 and K561 (Fig. 5G; Supplementary Fig. S8F). To investigate whether FAM135B activates the STING pathway by regulating the ubiquitination of IFI16 at the residues K143 and K561, we measured the phosphorylation levels of proteins associated with the STING pathway in TNBC cells after FAM135B depletion. We found that the expression of IFI16^K143R and IFI16^K561R mutants led to increased phosphorylation levels and substantially reversed the inhibitory effect of FAM135B depletion on the STING pathway activity, suggesting that the deubiquitination of IFI16 at residues K143 and K561 is crucial for the activation of the STING pathway mediated by FAM135B (Fig. 5H). Moreover, we transfected wild-type and mutant ubiquitins (K33R and K48R) into 293T cells and found that the ubiquitin-K33R mutant significantly suppressed IFI16 ubiquitination by the E3 ubiquitin ligase TRIM21, whereas the K48R mutant had a marginal effect. These results imply that TRIM21 primarily ubiquitinates IFI16 through K33-linked ubiquitin chains (Fig. 5I).

FAM135B promotes cytotoxic T-cell activity by stimulating the STING–type I IFN pathway

Next, we examined the role of the STING–type I IFN pathway in mediating FAM135B-induced cytotoxic CD8⁺ T-cell infiltration and tumor suppression. BT549 and MDA-MB-231 cells overexpressing FAM135B were treated with H-151, a STING antagonist, and subsequently cocultured with activated T cells. Flow cytometry analysis revealed that FAM135B-overexpressing TNBC cells enhanced the cytotoxic activity of CD8⁺ T cells, and this effect could be reversed by inhibiting STING signaling (Fig. 6A; Supplementary Fig. S9A). Accordingly, the inhibition of STING signaling restored the effects of FAM135B on the activation of the STING–type I IFN pathway, along with IRF3 nuclear translocation and increased expression of IFNB, CCL5, and CXCL10 (Fig. 6B and C; Supplementary Fig. S9B and S9C).

Figure 6.

FAM135B promotes cytotoxic T-cell activity by stimulating the STING–type I IFN pathway. A, BT549 and MDA-MB-231 cells expressing an empty vector or FAM135B were treated with H-151 (5 μmol/L) and cocultured with activated T cells for 48 hours. The T cells were then analyzed using flow cytometry. Representative flow cytometry plots and statistical quantification of GZMB⁺ CD8⁺ cells in CD3⁺ tumor-infiltrating lymphocytes are shown (mean ± SD, n = 3). B, Immunoblotting analysis of the indicated proteins in TNBC cells expressing an empty vector or FAM135B under H-151 treatment. C, BT549 cells expressing an empty vector or FAM135B were treated with H-151 (5 μmol/L) and were immunostained with an anti-IRF3 antibody. Confocal microscopy images exhibited the subcellular localization of IRF3 (scale bar, 10 μm). D, Schematic representation of mouse treatment schedules. 4T1 cells (5 × 10⁵ cells per mouse) expressing an empty vector or FAM135B were injected into the mammary fat pads of female BALB/c mice. Mice were injected intraperitoneally with H-151 (750 nmol per mouse) or PBS every 2 days. Three weeks after injection, the mice were humanely euthanized. E, Resected tumors from each group are shown (scale bar, 1 cm). Tumor weight was recorded, and tumor volume was calculated (mean ± SD, n = 5). F, Flow cytometry analysis revealed the population of GZMB⁺ CD8⁺ cells within CD3⁺ tumor-infiltrating lymphocytes of mouse TNBC tissues. Representative plots and flow cytometry quantification are shown (mean ± SD, n = 3). G, The orthotopic tumor tissues were immunostained with CD8ɑ. Representative images are shown (scale bar, 20 μm). *, P < 0.05; ***, P < 0.001; ****, P < 0.0001; ns, P > 0.05 (not significant). p-IRF, phosphorylated IRF; p-STING, phosphorylated STING; p-TBK, phosphorylated TBK.

To further assess the role of FAM135B-mediated activation of STING–type I IFN pathway in TNBC tumorigenesis, we injected 4T1 cells into the mammary fat pads of female immunocompetent mice (Fig. 6D). We found that the suppressive effect of FAM135B on TNBC tumor growth was diminished when STING signaling was blocked (Fig. 6E). Additionally, we observed that inhibiting STING signaling reversed the impact of FAM135B on activating the STING–type I IFN pathway in mouse TNBC tissues (Supplementary Fig. S9E). Flow cytometry analysis of orthotopic tumor tissues further demonstrated that the overexpression of FAM135B resulted in increased cytotoxic CD8⁺ T-cell activity, and this effect could be restored by blocking STING signaling (Fig. 6F; Supplementary Fig. S9D). Moreover, immunostaining of CD8α, GZMB, and Ki67 in mouse TNBC tissues revealed that inhibiting STING signaling rescued the stimulation of cytotoxic T cells and the tumor growth suppression induced by FAM135B overexpression (Fig. 6G; Supplementary Fig. S9F). We also evaluated the role of FAM135B in STING agonist treatment using an immunocompetent mouse model of orthotopic TNBC (Supplementary Fig. S10A). We observed that FAM135B overexpression in combination with STING agonist notably suppressed tumor growth (Supplementary Fig. S10B). Immunostaining analysis demonstrated that the combination of FAM135B overexpression and the STING agonist significantly enhanced cytotoxic T-cell activity compared with FAM135B overexpression alone (Supplementary Fig. S10C). Accordingly, we assessed the phosphorylation levels of STING in mouse TNBC tissues and found that these levels were elevated in tumors with FAM135B overexpression. This elevation was significantly greater following treatment with the STING agonist and decreased after STING blockade treatment (Supplementary Fig. S10D).

FAM135B augments the efficacy of anti–PD-1 immunotherapy

We assessed the impact of FAM135B on the response to ICB in TNBC treatment. Using an immunocompetent mouse model of orthotopic TNBC (Fig. 7A), we observed that PD-1 blockade therapy alone moderately inhibited TNBC growth. However, the combination of anti–PD-1 and the overexpression of FAM135B led to almost complete inhibition of tumor growth (Fig. 7B). In mouse TNBC tissues, PD-1 blockade therapy in combination with FAM135B overexpression contributed to an increase in the infiltration and cytotoxic activity of CD8⁺ T cells, suggesting that FAM135B overexpression augments the antitumor immune response mediated by CD8⁺ T cells and enhances the sensitivity of TNBC cells to anti–PD-1 immunotherapy (Fig. 7C). Moreover, we collected TNBC specimens from patients undergoing PD-1 blockade therapy to investigate the correlation between the therapeutic response and the expression levels of FAM135B. To evaluate the therapeutic effects of ICB treatment, we conducted comparative analyses of tumor sizes using CT images obtained before and after immunotherapy. Immunostaining of FAM135B in TNBC tissues demonstrated that elevated expression levels of FAM135B were predictive of an improved response to anti–PD-1 treatment (Fig. 7D).

Figure 7.

FAM135B augments the efficacy of anti–PD-1 immunotherapy. A, Schematic diagram of mouse treatment schedules. 4T1 cells (5 × 10⁵ cells per mouse) expressing either an empty vector or FAM135B were injected into the mammary fat pads of female BALB/c mice. Mice received intraperitoneal injection of anti–PD-1 (200 μg per mouse) or IgG every 3 days. Three weeks after injection, the mice were humanely euthanized. B, Resected tumors from each group are shown (scale bar, 1 cm). Tumor weight was measured, and tumor volume was calculated (mean ± SD, n = 4). C, Immunostaining of CD8ɑ and GZMB in mouse orthotopic tumor tissues. Representative images are shown (scale bar, 20 μm). D, CT images obtained before and after immunotherapy display the mediastinal lymph node metastasis in patients with TNBC (left). Consecutive sections of 12 human TNBC tissues were immunostained with anti-FAM135B antibodies. Representative images are shown (middle; scale bar, 40 μm; left; scale bar, 20 μm; right). The staining intensity was assessed by IHC scoring. Bars in the right panel indicate the percentages of high and low FAM135B expression in TNBC tissues from patients who responded to anti–PD-1 treatment compared to those who did not. E, Illustration of the FAM135B–IFI16–STING axis in regulating CD8⁺ T cell–mediated antitumor immunity in TNBC. FAM135B stabilizes IFI16 by competing with TRIM21 for binding to IFI16, thereby inhibiting the K33-linked ubiquitination of IFI16 at K143 and K561. FAM135B suppresses the proteasome degradation of IFI16, activates the IFI16-mediated STING–type I IFN pathway to induce the cytotoxic CD8⁺ T-cell infiltration, and thus inhibits TNBC tumorigenesis. H&E, hematoxylin and eosin; P, phosphorylation modification; Ub, ubiquitin molecules. [E, Created in BioRender. Yang, T. (2026) https://BioRender.com/biejamk]

Discussion

Immunotherapy represents a promising treatment for patients with TNBC; however, the immunosuppressive microenvironment characteristic of TNBC largely diminishes the therapeutic effectiveness. In this study, we show that FAM135B induces the infiltration and activation of cytotoxic T cells, thereby improving responses to ICB treatment and suppressing tumor growth. We found that FAM135B activates the STING–type I IFN pathway by inhibiting the ubiquitination and proteasomal degradation of IFI16 at residues K143 and K561. Our findings consistently support the essential involvement of the FAM135B–IFI16–STING axis in CD8⁺ T cell–mediated antitumor immunity in TNBC, indicating that FAM135B might be a potential biomarker for predicting responses to immunotherapy in patients with TNBC.

FAM135B is mutated and is associated with a poor prognosis in esophageal squamous cell carcinoma and colorectal cancer (13, 15). Additionally, FAM135B induces DNA damage response, contributing to insensitivity to chemotherapy and radiotherapy (14, 15). Recent evidence suggests that the accumulation of DNA damage enhances tumor immunogenicity, thereby stimulating the activation of cytotoxic T cells and promoting antitumor immunity (17, 18). We found that FAM135B mutations were present in 4% of cases within the TNBC cohort, supporting its role in the tumorigenesis of TNBC. Using bulk and single-cell RNA-seq data, we observed that FAM135B was correlated with the infiltration and activation of cytotoxic T cells in TNBC. Furthermore, we demonstrated that the overexpression of FAM135B markedly inhibited tumor growth by enhancing the activity of cytotoxic T cells in TNBC cells. However, this stimulatory effect of FAM135B on T-cell activation was not observed in non-TNBC cells. The specific role of FAM135B in different subtypes of breast cancer may be attributed to the high immunogenicity of TNBC, which leads to increased neoantigen generation and provides more targets for T-cell recognition and activation (25).

The STING pathway constitutes an important component of the T cell–dependent antitumor immune response (26–28). Activation of STING and its associated downstream signaling cascades results in the production of type I IFN and chemokines, which collectively promote the infiltration and activation of cytotoxic T cells (22). Numerous natural and synthetic agonists targeting the STING pathway have been developed (29). However, the clinical application of these STING agonists remains constrained by their potent activation of innate immunity, leading to cytokine storms and systemic toxicity (30). Therefore, it is crucial to identify novel regulators of the STING pathway for safer therapeutic strategies. In this study, we demonstrated that the overexpression of FAM135B induces cytotoxic T-cell infiltration by activating the STING pathway, suggesting that FAM135B may function as a regulator of this pathway. Moreover, PD-1 blockade therapy in combination with FAM135B overexpression resulted in nearly complete inhibition of tumor growth in the TNBC mouse model. Our findings imply that targeting FAM135B to modulate STING pathway activity could enhance the efficacy of ICB therapy without relying on direct STING agonists. Moreover, immunostaining analyses of TNBC specimens from patients receiving PD-1 blockade therapy revealed that elevated levels of FAM135B were predictive of an improved treatment response. Thus, FAM135B expression might serve as a potential predictor of ICB efficacy, enabling better patient stratification for TNBC. Patients with high FAM135B expression, indicative of a preexisting immunostimulatory microenvironment, may be more likely to benefit from immunotherapy.

IFI16 is an innate immune sensor for intracellular DNA that activates the STING pathway, inducing the production of type I IFN and chemokines (22, 23). Particularly, IFI16 induces STING-mediated type I IFN production and enhances antitumor effects in TNBC during chemotherapy (31). The downregulation of IFI16-mediated STING signaling has been implicated in immune evasion in HER2-positive breast cancer (32). Similarly, our study confirmed that the overexpression of IFI16 induced activation of the STING pathway in TNBC cells. Moreover, FAM135B inhibited the ubiquitination of IFI16, thereby increasing its stability and facilitating the initiation of IFI16-dependent STING signaling. The posttranslational modification of IFI16 is crucial for modulating its stability and subcellular localization, as well as determining the outcomes of the immune response (24, 33, 34). Previous studies have revealed that TRIM21, an E3 ubiquitin ligase, is recruited to mediate the ubiquitination and degradation of IFI16, thereby promoting the evasion of innate immune surveillance (24). Consistent with this, we identified TRIM21 as the E3 ubiquitin ligase for IFI16 in TNBC cells and demonstrated that FAM135B inhibited the ubiquitination and proteasomal degradation of IFI16 by competitively blocking its binding to TRIM21. Furthermore, we identified two ubiquitination sites at lysine 143 and lysine 561 on IFI16, which were responsible for its proteasomal degradation. Mutations of these two sites significantly abolished TRIM21-mediated ubiquitination of IFI16, thereby stabilizing it. Our study confirmed that either the overexpression of FAM135B or the targeted mutation of these ubiquitination sites led to the stabilization of IFI16, subsequently activating the STING pathway and stimulating antitumor immunity.

In conclusion, we have demonstrated that FAM135B stabilizes IFI16 by inhibiting the interaction between IFI16 and the E3 ubiquitin ligase TRIM21, thereby initiating the IFI16-dependent STING–type I IFN signaling and contributing to enhanced cytotoxic T-cell activity. Activating the FAM135B–IFI16–STING signaling pathway induces the infiltration and activation of cytotoxic T cells, suppressing tumor growth and sensitizing tumor cells to PD-1 blockade therapy. Our findings present a potential predictor for the effectiveness of anti–PD-1 treatment in TNBC.

Supplementary Material

Figure S1

FAM135B is mutated and associated with T cell infiltration in TNBC.

Figure S2

FAM135B expression in TNBC cell lines.

Figure S3

FAM135B overexpression in non-TNBC cells has no significant effect on T cell activation.

Figure S4

FAM135B inhibits TNBC growth and activates T cells.

Figure S5

FAM135B overexpression is correlated with type I interferon signaling and cytokine-mediated signaling.

Figure S6

FAM135B interacts with IFI16 independent of DNA.

Figure S7

FAM135B stabilizes IFI16 by inhibiting its ubiquitination and proteasomal degradation.

Figure S8

TRIM21 ubiquitinates IFI16 at K143 and K561.

Figure S9

Blocking the STING pathway restores the promoting effect of FAM135B overexpression on T cell activation.

Figure S10

FAM135B overexpression in combination with cGAMP significantly suppresses TNBC growth and activates T cells.

Table S1

Sequences of shRNAs.

Table S2

Sequences of RT-qPCR primers.

Table S3

Antibody information used in this study.

Data Availability

The bulk RNA-seq data generated in this study are publicly accessible in the Sequence Read Archive database under the accession number PRJNA1243561. Additionally, the MS proteomics data have been deposited to the ProteomeXchange Consortium via the iProX partner repository (20, 21) with the dataset identifier PXD068216. All other data generated in this study are available upon request from the corresponding author.

Authors’ Disclosures

No disclosures were reported.

Authors’ Contributions

W. Lin: Investigation, visualization, methodology, writing–original draft. J. Li: Investigation. J. Wu: Investigation. B. Huang: Investigation. J. Liang: Investigation. Y. Zhang: Investigation. L. Zhao: Conceptualization, data curation, supervision, writing–review and editing. G. Yao: Resources, supervision, funding acquisition.