Pan-cancer profiling links C1orf50 to DNA repair and immune modulation in ovarian cancer

Abstract

Background

C1orf50 encodes a small, evolutionarily conserved protein, the function of which remains unclear. Its significance across various human cancers, particularly its specific role in ovarian cancer within an immunogenomic context, is not yet fully understood. Utilizing The Cancer Genome Atlas and single-cell RNA sequencing (scRNA-seq) public datasets, we conducted a comprehensive profiling of C1orf50 across multiple cancer types, with a particular focus on ovarian cancer, to investigate its associations with copy-number status, genomic instability, tumor programs, and the immune microenvironment.

Results

Across cancer types, copy-number gain or amplification of C1orf50 was most frequent in ovarian cancer and closely tracked with higher messenger RNA levels. Higher C1orf50 expression was associated with a greater tumor mutational burden and homologous recombination deficiency, as indicated by gene-set patterns that suggested heightened cell-cycle and cellular stress responses accompanied by reduced oxidative phosphorylation, enrichment of regulatory T cells, and depletion of resting memory CD4 T cells. In ovarian cancer, focal events at chromosome 1p34.2 were accompanied by stepwise increases in C1orf50 expression by clinical stage and were linked to higher tumor mutational burden, homologous recombination deficiency, and greater loss of heterozygosity, together with more frequent gene alterations in BRCA1 or BRCA2. Immune composition clustered into profiles consistent with an immunosuppressive context in tumors with higher C1orf50 expression. The scRNA-seq data further revealed that cancer cells enhanced immune-suppressive interactions with various immune cell populations and diminished antigen-presentation signals. Analyses of genomic instability in ovarian cancer suggested mutational processes compatible with base-substitution patterns associated with cytidine deaminase activity and with insertion-deletion patterns characteristic of homologous recombination failure, while transcript-level patterns pointed to a broad downshift of canonical DNA repair activity with apparent compensatory adjustments in related pathways rather than a uniform change in any single pathway.

Conclusions

The overexpression of C1orf50 characterizes an aggressive immunogenomic phenotype in ovarian cancer, distinguished by genomic instability, impaired DNA repair mechanisms, and extensive immunosuppression. These findings indicate that C1orf50 warrants consideration as a potential biomarker and a prospective target for therapeutic investigation. Furthermore, they advocate for the progression to prospective validation and functional studies to ascertain its clinical significance.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13048-025-01916-8.

Introduction

C1orf50 (Chromosome 1 open reading frame 50) is a gene located on chromosome 1p34.2 that encodes a small protein of ~ 26 kDa (DUF2452). C1orf50 is a highly conserved gene, with homologs present across a wide range of species, from Caenorhabditis elegans to mammals. This level of conservation usually indicates an essential biological function that has been preserved through evolution. However, despite this, the molecular functions of C1orf50 remain largely unknown, and the gene is still considered poorly characterized. Structural predictions from AlphaFold [1] provide initial clues, suggesting the protein folds into a compact, hairpin-like structure composed of two α-helices. Such a minimal fold suggests that C1orf50 may function as a small structural or regulatory protein rather than a typical enzyme. Transcriptomic analyses also reveal two annotated isoforms: one protein-coding and the other non-coding, indicating that C1orf50 may operate at multiple regulatory levels. Overall, these findings identify C1orf50 as an intriguing yet underexplored gene that likely plays an important role in cellular biology and merits deeper functional investigation.

We have previously identified C1orf50 as a prognostic biomarker in malignant melanoma [2], hepatocellular carcinoma [3], and Luminal A breast cancer [4]. In melanoma, high levels of C1orf50 are associated with poor clinical outcomes, as the gene appears to promote tumor progression by accelerating cell cycle activity, enhancing DNA damage repair, and supporting cancer stemness through YAP/TAZ signaling [2]. A similar trend is seen in Luminal A breast cancer, where increased C1orf50 expression is linked to a worse prognosis, aiding tumor cell proliferation, maintaining stem-like features, and increasing resistance to therapy. Besides its role in tumor growth, C1orf50 also influences immune-related transcriptional programs, including the upregulation of PD-L1, which helps tumors evade immune surveillance [4]. Importantly, these molecular and functional features have also been confirmed experimentally in our previous wet-lab studies. Using immunohistochemistry and Western blotting with a specific anti-C1orf50 antibody (Proteintech #20,957–1-AP, raised against the 1–199 aa domain of human C1orf50 [BC001711]), we validated that C1orf50 protein expression correlates with enhanced cell-cycle activity and cancer stemness in breast cancer, with stemness-related phenotypes in melanoma, and with stem-like features in hepatocellular carcinoma. These findings demonstrate that the oncogenic and stemness-promoting functions of C1orf50 are consistently supported by both transcriptomic analyses and direct experimental evidence.

These observations support a role for C1orf50 as an oncogenic factor that integrates tumor cell-intrinsic programs with modulation of the tumor immune microenvironment. However, its broader biological and clinical relevance across diverse human cancers has not been systematically evaluated.

Genomic profiling has identified recurrent copy number gains and focal amplifications at 1p34.2, which include C1orf50, across various malignancies, such as ovarian cancer, neuroblastoma, and lung adenocarcinoma [5–7]. Genomic instability is a hallmark of cancer, accelerating tumor evolution, driving therapy resistance, and predicting poor outcomes [8, 9]. By increasing tumor mutation burden and generating neoantigens, it can also reshape the immune landscape and influence responsiveness to immune checkpoint blockade [10]. C1orf50 may contribute to this process by supporting proliferative and DNA repair programs while simultaneously modulating immune pathways such as PD-L1 expression, linking genomic instability to both tumor progression and immune escape.

Based on our prior observations linking C1orf50 to cell-cycle regulation, stemness, and immune modulation, we hypothesized that C1orf50 functions at the intersection of genomic instability and immune regulation, driving cancer progression through dual effects on tumor biology and the tumor microenvironment. To investigate this hypothesis, we conducted a comprehensive pan-cancer analysis of The Cancer Genome Atlas (TCGA) to characterize C1orf50 expression, mutation spectrum, and copy number alterations across 33 cancer types, and to evaluate their associations with genomic instability metrics and immune cell infiltration. Given that C1orf50 amplification occurs most prominently in ovarian cancer, we concentrated our analysis on this setting. Through integration of transcriptomic, proteomic, and single-cell communication data, we identified the molecular pathways influenced by C1orf50, and we evaluated its potential clinical significance as both a biomarker and a therapeutic target.

Methods

Data sources and preprocessing

All analyses were conducted using R (v4.3.2; University of Auckland, Auckland, New Zealand) [11]. Genomic, transcriptomic, proteomic, and normal tissue expression data were obtained from The Cancer Genome Atlas (TCGA) and other publicly available resources. The TCGA Pan-Cancer dataset, accessed through the cBioPortal for Cancer Genomics [12, 13] provided four key metrics of genomic instability: the chromosomal instability (CIN) index, the fraction of genome altered (FGA), the microsatellite instability (MSI) sensor score, and the tumor mutation burden (TMB). Homologous recombination deficiency (HRD) scores and CIBERSORT-derived estimates of immune cell infiltration were obtained from the dataset of Thorsson et al. [14], and loss of heterozygosity (LOH) data were obtained from the dataset of Taylor et al. [15]. CIBERSORT data were available for 32 TCGA cancer types, excluding the acute myeloid leukemia (LAML) cohort, which lacked valid fraction values. All metrics were matched to the corresponding TCGA sample identifiers for downstream analyses. For the pan-cancer analysis, we utilized the "Batch normalized from Illumina HiSeq_RNASeqV2" mRNA expression dataset from the TCGA Pan-Cancer Atlas, accessed via cBioPortal. These data were generated using RSEM [16] for gene-level normalization to correct for sequencing depth and transcript length, followed by EB + + (a ComBat-based empirical Bayes harmonization method) [17] to adjust batch effects arising from different Illumina platforms, sequencing centers, and experimental plates. From this harmonized dataset available through cBioPortal, we obtained C1orf50 expression values across all tumor types, ensuring direct comparability of expression measurements throughout the TCGA cohort. For the cancer type-specific analyses, we used mRNA expression z-scores from cBioPortal. Samples were stratified into C1orf50-high and C1orf50-low groups based on the median expression level, using the maximum number of available samples for each analysis. In the ovarian cancer cohort, 300 cases with both RNA expression and copy-number alteration (CNA) data were used in this study. Among these, the most extensive available datasets with matched mutation data (n = 291) and clinical FIGO stage information (n = 296) were used for the respective analyses. Each analysis used all available samples for that data type, without arbitrary exclusions, to ensure transparency and reproducibility.

Single-cell RNA sequencing data preprocessing and annotation

Single-cell RNA sequencing (scRNA-seq) data were obtained from a publicly available dataset (GSE158722), which includes 10 × Genomics profiles from 20 ovarian tumor samples. We included both pre-treatment samples and post-treatment samples to maximize statistical power. Raw gene-cell count matrices were downloaded and processed with the Seurat package (v5.2.1) [18]. Each tumor sample was initially imported as a separate Seurat object before merging for integrated analysis. To maintain data quality, we applied strict quality control, excluding cells with fewer than 200 genes, more than 5,000 genes, or over 10% mitochondrial gene transcripts, as these often indicate stressed or dying cells. For gene-level filtering, a gene was considered expressed if it had non-zero counts in ≥ 3 cells across the merged object (Seurat default). We then normalized counts in Seurat using log-normalization (scale factor = 10,000), identified highly variable genes with the ‘vst’ method (2,000 features), followed by scaling and PCA for downstream analyses. Batch effects were corrected with Harmony (v1.2.3) [19] using sample ID as the batch variable. We used sample ID as the Harmony batch variable because libraries were generated at the sample level, which introduces technical variation across patients. We verified that integration improved sample mixing within cell types, supporting that batch effects were reduced without over-correction (Supplemental Fig. 1). The resulting batch-corrected embeddings were used for unsupervised clustering and visualization via Uniform Manifold Approximation and Projection (UMAP). Clustering employed the top 30 Harmony-corrected PCs, with a moderate resolution to capture both broad cell types and finer subpopulations. Initial cell type annotations were assigned based on the GSE158722 annotation file. To refine immune cell classification, macrophages and CD4⁺ T cells were further subtyped. Macrophages were classified into M1-like (pro-inflammatory) or M2-like (anti-inflammatory) phenotypes based on gene expression scores derived from established marker sets. M1-like macrophages were characterized by high expression of NOS2, IL1B, TNF, IL6, and IL12B, while M2-like macrophages were characterized by high expression of MRC1, CD163, TGFB1, IL10, and ARG1. Cells with indeterminate scores were labeled as "Intermediate" and excluded from downstream analysis. Regulatory T cells (Tregs) within the CD4⁺ T cell population were identified by high expression of FOXP3, IL2RA, and CTLA4. We focused on key immune and tumor cell types, including B cells, CD4⁺ and CD8⁺ T cells, Tregs, NK cells, dendritic cells, monocytes, M1 and M2 macrophages, and malignant epithelial cells. The final dataset comprised 24,267 cells. This annotated and batch-corrected dataset was employed for all subsequent analyses.

Fig. 1.

Genomic alterations and mRNA expression of C1orf50 across 33 TCGA cancer types. A C1orf50 mRNA expression levels (nTPM) in normal human tissues based on the Human Protein Atlas Consensus dataset. B Lollipop plot showing the distribution of somatic mutations in C1orf50 (transcript variant 1, NM_024097) across TCGA pan-cancer samples. Green dots represent missense mutations, and red dots represent nonsense mutations. The vertical axis indicates the number of samples harboring a mutation at the same amino acid position. C Proportion of mutated cases by mutation type across 33 TCGA cancer types. Missense (VUS) and truncating (VUS) mutations are shown separately. D Relative composition of copy-number alteration categories (amplification, gain, diploid, shallow deletion, deep deletion) in C1orf50 across TCGA cancer types. E Violin plots showing C1orf50 mRNA expression levels by cancer type, mutation status, and copy-number category. Circles represent individual tumor samples; colors indicate mutation type and copy-number status as defined in the legend

Differential expression and genomic signature analysis

Differential expression analysis was performed using the limma package (v3.58.1). Mutational signatures were calculated using COSMIC Mutational Signatures v3.4. Protein expression profiles were obtained from the RPPA500 dataset downloaded from the TCPA portal (https://tcpaportal.org/rppa500/download.html) [20, 21]. Normal tissue RNA expression values (nTPM) were obtained from the consensus dataset of the Human Protein Atlas [22].

Gene set enrichment analysis (GSEA)

GSEA was performed in pre-ranked mode using fgsea (v1.28.0), ranking genes by limma moderated t-statistics for the C1orf50-high versus C1orf50-low comparison within each cancer type. The analyses were based on gene expression z-score data from the TCGA Pan-Cancer Atlas obtained via cBioPortal. MSigDB Hallmark (v2023.2.Hs) gene sets were used as the reference, and pathways with a false discovery rate (FDR-adjusted p value) < 0.05 were considered significant.

Copy-number analysis using GISTIC2.0

Segmented copy-number data for the TCGA ovarian cancer cohort were analyzed using GISTIC2.0 (v2.0.23) with Human_hg19.mat as the reference. Key parameters were set as follows: amplification/deletion thresholds = ± 0.1, focal length cutoff = 0.70, confidence level = 0.90, q-value cutoff = 0.01, and join segment size = 4. Both gene-level and broad-level analyses were enabled ("gene GISTIC" = yes; "run broad analysis" = yes) using "wide peak" boundaries. Segments were median-centered, capped at ± 2.0, and the X chromosome was retained. Gene-level CNA values were integrated with mRNA expression using matched TCGA sample IDs.

Unsupervised clustering of tumor immune profiles

CIBERSORT-derived immune cell fractions were initially pre-processed to ensure data quality. Samples with missing values were removed, and the remaining fractions were standardized using z-score transformation across all cell types to facilitate direct comparison. To identify patterns of immune composition, we used unsupervised k-means clustering with k = 3, which divided tumors into three distinct immune subtypes. These clusters, labeled C1 through C3, displayed consistent differences in immune cell infiltration profiles, providing a basis for further analysis of tumor-immune interactions. To evaluate the robustness of this clustering solution, we conducted a bootstrap stability analysis using 80% subsampling, repeated 1000 times. Agreement between subsampled clusters and the full dataset was quantified using both the Adjusted Rand Index (ARI) and Normalized Mutual Information (NMI). The results demonstrated strong reproducibility of the three-cluster solution (median ARI = 0.889, interquartile range 0.757–0.941; median NMI = 0.832, interquartile range 0.700–0.906).

Data visualization

Heatmaps were generated using the ComplexHeatmap package (v2.18.0) [23] for GSEA results and other comparative analyses. Statistical significance was indicated by asterisks where applicable.

Cell–cell communication analysis based on C1orf50 expression

To examine the impact of variation in tumor-intrinsic C1orf50 expression on intercellular communication, we conducted a comparative CellChat [24] analysis between malignant epithelial cells with high and low C1orf50 expression. This analysis utilized a previously annotated and batch-corrected Seurat object comprising 20 tumor samples (GSE158722), which included major immune and tumor cell types as previously described. Malignant epithelial cells were divided into C1orf50-high and C1orf50-low groups based on the median expression level of C1orf50. All non-epithelial cell populations were included in both groups to allow a direct comparison of outgoing signaling from tumor cells to the surrounding immune microenvironment. Two distinct CellChat objects were constructed, one for each group, employing the human ligand-receptor database. Overexpressed genes and interactions were identified, and cell–cell communication probabilities were calculated. Interactions with low abundance were filtered by requiring a minimum of 10 cells per cell type. Communication probabilities were subsequently aggregated at the signaling pathway level, and centrality scores were computed for each group. To evaluate the copy number differential communication between the two groups, the two CellChat objects were merged and compared. Interaction counts and total communication strength were summarized across cell types. Specifically, outgoing signaling from malignant epithelial cells to immune cells was analyzed to assess changes in potential immune-regulatory interactions associated with C1orf50 expression.

Results

Genomic alterations and mRNA expression of C1orf50 across 33 human cancer types

Normal tissue analysis of the Human Protein Atlas Consensus dataset showed that C1orf50 RNA expression is highest in the liver, nervous system, testis, kidney, and ovary, while it is comparatively low in tissues such as the tonsil and gallbladder (Fig. 1A). Examination of somatic mutations in TCGA pan-cancer cohorts revealed that, within the 199 amino acids encoded by transcript variant 1 (NM_024097), missense mutations (green) were the most common. In contrast, nonsense mutations (red) were infrequent and scattered, with no evidence of recurrent hotspots (Fig. 1B). Mutation rates were generally low across all tumor types, reaching their highest levels in uterine corpus endometrial carcinoma (UCEC, 1.36%) and skin cutaneous melanoma (SKCM, 1.14%) (Fig. 1C).

Copy-number analysis identified frequent gains or amplifications in ovarian cancer (OV, 48.3%), uterine carcinosarcoma (UCS, 40.4%), and cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC, 31.6%). In contrast, shallow or deep deletions were most common in pheochromocytoma and paraganglioma (PCPG, 57.9%) and chromophobe renal cell carcinoma (KICH, 80.0%) (Fig. 1D). When copy-number data were integrated with transcriptomic profiles, tumors with C1orf50 amplification or gain—most notably ovarian cancers—consistently exhibited elevated mRNA expression (Fig. 1E). These findings indicate that copy-number amplification may play a key role in C1orf50 overexpression in certain malignancies.

Association of C1orf50 expression with genomic instability, pathway activity, and immune microenvironment in pan-cancer analysis

We then examined the relationship between C1orf50 expression and various measures of genomic instability across 33 TCGA cancer types (Fig. 2A). Using Spearman correlation analysis, we found strong positive relationships in OV and adrenocortical carcinoma (ACC), where higher C1orf50 levels correlated with increased tumor mutational burden (TMB) and homologous recombination deficiency (HRD) scores. These results suggest that C1orf50 may be associated with genomic instability in specific tumor types, potentially contributing to the accumulation of mutations and faulty DNA repair. Conversely, several cancers, including mesothelioma (MESO), pheochromocytoma and paraganglioma (PCPG), and rectum adenocarcinoma (READ), showed notable negative correlations, especially with chromosomal instability (CIN) and the fraction of genome altered (FGA). This suggests that the relationship between C1orf50 and genomic instability is inconsistent, varying by cancer type. Such variation highlights the possibility that C1orf50 may have context-dependent roles in genome maintenance, promoting or reducing instability depending on the cellular and molecular environment.

Fig. 2.

Association of C1orf50 expression with genomic instability, pathway activity, and immune cell composition across TCGA cancers. A Spearman correlations between C1orf50 mRNA expression and key genomic instability metrics, including chromosomal instability (CIN) index, fraction of genome altered (FGA), microsatellite instability (MSI) score, tumor mutational burden (TMB), and homologous recombination deficiency (HRD) scores. Color scale denotes correlation coefficient (rho). Asterisks indicate statistical significance (* FDR < 0.05, ** FDR < 0.01, *** FDR < 0.001; false discovery rate adjusted using the Benjamini–Hochberg method). B Heatmap of normalized enrichment scores (NES) from Gene Set Enrichment Analysis (GSEA) using Hallmark gene sets, comparing C1orf50-high versus C1orf50-low tumors (median split) within each of the 33 TCGA cancer types. Significance is based on Benjamini–Hochberg adjusted FDR values (* FDR < 0.05, ** FDR < 0.01, *** FDR < 0.001). C Spearman correlations between C1orf50 expression and CIBERSORT-estimated immune cell fractions across TCGA cancer types. Positive values indicate higher abundance, and negative values indicate lower abundance of the corresponding immune cell type in C1orf50-high tumors. Significance is based on Benjamini–Hochberg adjusted FDR values (* FDR < 0.05, ** FDR < 0.01, *** FDR < 0.001)

To identify associated biological programs, we performed GSEA by comparing C1orf50-high and C1orf50-low tumors, defined by the median expression within each cancer type (Fig. 2B). Consistent with prior observations [2–4], gene sets related to the mitotic spindle and downregulated genes in response to ultraviolet (UV) radiation were significantly enriched in the high-expression group across most cancers. In contrast, oxidative phosphorylation-related genes were broadly downregulated. These trends suggest that C1orf50 overexpression is linked to proliferative and stress-response pathways, alongside suppression of mitochondrial oxidative metabolism.

Analysis of CIBERSORT-derived immune cell fractions showed that C1orf50-high tumors were significantly enriched in regulatory T cells (Tregs) across multiple cancer types, including thymoma (THYM), testicular germ cell tumors (TGCT), thyroid carcinoma (THCA), kidney renal clear cell carcinoma (KIRC), breast carcinoma (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), sarcoma (SARC), and kidney renal papillary carcinoma (KIRP) (Fig. 2C). These findings suggest that C1orf50 overexpression contributes to creating an immunosuppressive tumor microenvironment. Conversely, resting memory CD4⁺ T cells were depleted in numerous cancer types, including THYM, KIRP, chromophobe renal cell carcinoma (KICH), and others potentially reflecting a shift toward activated or exhausted states. Collectively, these results indicate that C1orf50 overexpression is associated with genomic instability in selected cancers, activation of cell cycle and stress pathways, and an immune landscape characterized by elevated regulatory T cell levels and reduced memory CD4⁺ T cells, features consistent with immune evasion.

C1orf50 amplification leads to increased mRNA levels and immune evasion signatures in ovarian cancer

We next turned our analysis to ovarian cancer, where C1orf50 showed frequent copy number gains and amplifications. Using GISTIC2.0 on the TCGA ovarian cancer cohort, we identified recurrent amplification at chromosome 1p34.2, the locus containing C1orf50 (Fig. 3A). Tumors with copy-number gain or high-level amplification exhibited significantly higher C1orf50 mRNA expression compared with diploid or deletion cases (p = 4.95 × 10⁻¹³; Fig. 3B). When stratified by FIGO stage, C1orf50 expression showed a stepwise increase from stage II to stage IV disease, with a significant difference observed between stage II and stage IV tumors (p = 0.027; Fig. 3C). Together, these findings suggest that C1orf50 amplification is a major contributor to its transcriptional upregulation, and may be linked to tumor progression and disease aggressiveness in ovarian cancer.

Fig. 3.

C1orf50 amplification is associated with increased expression and an immunosuppressive tumor microenvironment in ovarian cancer. A GISTIC2.0 analysis of the TCGA ovarian cancer cohort showing recurrent copy-number alterations across the genome. The 1p34.2 locus containing C1orf50 is indicated. B C1orf50 mRNA expression stratified by copy-number status (Other = diploid or deletion; Amp/Gain = amplification or gain). C C1orf50 mRNA expression by FIGO stage (II, III, IV). D Top five significantly upregulated and downregulated proteins in C1orf50-high versus C1orf50-low tumors, identified by reverse-phase protein array (RPPA) analysis. Bars represent log₂ fold change; p-values are indicated. E Unsupervised clustering of CIBERSORT-estimated immune cell fractions in ovarian cancer, identifying three immune clusters (C1-C3). Heatmaps show the immune cell composition (left) and the mean z-scores per cluster (right). Boxplots depict Treg and M2 macrophage abundance, and C1orf50 mRNA expression, by cluster. Significance levels: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001

Proteomic profiling using RPPA identified multiple proteins whose abundance differed significantly between C1orf50-high and C1orf50-low tumors (Fig. 3D). The immune checkpoint molecule B7-H3 (CD276) showed the strongest upregulation in the high-expression group, alongside SGK3 and PRDX1, whereas phospho-PKCα (T638/T641), RHEB, and IGFBP3 were significantly reduced.

Unsupervised clustering of CIBERSORT-estimated immune cell fractions identified three characteristic immune clusters (C1-C3) (Fig. 3E). The C3 cluster, which had the highest C1orf50 expression, was enriched in immunosuppressive cell types, including regulatory T cells (Tregs) and M2 macrophages. Collectively, these findings indicate that C1orf50 amplification is associated with increased transcript expression and an immunosuppressive tumor microenvironment in ovarian cancer. To confirm the robustness of this immune clustering, we performed a bootstrap stability analysis (80% subsampling, B = 1000). The clustering structure was highly reproducible (median ARI = 0.889, interquartile range [IQR]: 0.757–0.941; median NMI = 0.832, IQR: 0.700–0.906), confirming the stability of the k = 3 solution.

Impact of C1orf50 on cancer cell-immune cell communication networks

To explore how C1orf50 expression influences cell–cell communication in the tumor immune microenvironment, we stratified cancer cells into C1orf50-high and C1orf50-low groups based on the median expression level using single-cell RNA-seq data from the ovarian cancer cohort [25]. Cell–cell communication networks were inferred using CellChat across 10 major cell types, including cancer cells, regulatory and effector T cells (Tregs), NK cells, macrophages, monocytes, and other immune populations. The UMAP plot showed the distribution of cell types used in the analysis, with cancer cells forming the majority, while immune cells were resolved (Fig. 4A). Comparison of intercellular communication revealed that the C1orf50-low group exhibited a greater number of ligand-receptor interactions (1,976) than the C1orf50-high group (1,626), along with slightly stronger overall interaction strength (57.2 vs. 56.4). These results suggest a modest attenuation of global cell–cell communication in tumors with elevated C1orf50 expression (Fig. 4B). In the C1orf50-high group, cancer cells showed increased strength of cell–cell communication to all cell types, while non-cancer cells showed generally reduced signaling number and communication strength (Fig. 4C). We then explored the ligand-receptor pair connection between each cell type (Fig. 4D). In C1orf50-high cancer cells, we observed enhanced outgoing communication from cancer cells toward various immune subsets. Immunosuppressive interactions that include NECTIN2-TIGIT [26], MIF-(CD74 + CXCR4) [27, 28], PGE2-PTGER2 [29], and HLA-E-NKG2A [30] were upregulated in CD4⁺ T cells, CD8⁺ T cells, and NK cells, consistent with reduced anti-tumor immune activity. In addition, regulatory T cells (Tregs) received stronger signals through NECTIN2/3-TIGIT and LAMC2-CD44 [31], suggesting an enhanced immunosuppressive function of C1orf50. M1 and M2 macrophages showed increased engagement via LAM family-CD44 [31], MDK-NCL [32], APP-CD74 [33], and ICAM1-integrins [34], reflecting a tumor-supportive environment. B cells, DCs, and monocytes also received intensified adhesion- and growth-related signals such as ICAM1-integrins or MDK-NCL. Conversely, antigen presentation via HLA class II molecules from cancer cells to immune cells was reduced, along with attenuated signals to T/NK cells and macrophages via the CD99 axis [35], further supporting immune evasion in C1orf50-high tumors. To characterize how C1orf50 expression influences immune microenvironment communication, we assessed pathway-level signaling from cancer cells to immune subsets (Fig. 4E). C1orf50-high cancer cells showed elevated signaling toward CD8 + T cells through multiple pathways, including prostaglandin, APP, and LAMININ signaling. However, MIF signaling toward CD8 + T cells was reduced, potentially reflecting suppressed inflammatory cues. In relation to Treg and NK cells, C1orf50-high cancer cells exhibited elevated signaling through NECTIN, LAMININ, prostaglandin, MK (Midkine), and cholesterol-related pathways, all of which have been implicated in immune checkpoint activity. The NECTIN-TIGIT axis was reinforced across Treg, NK, and M2 macrophage targets. For M1 and M2 macrophages, the C1orf50-high group showed increased communication through the LAMININ, APP, and complement pathways. At the same time, MHC class II signaling was reduced in M1 macrophages, indicating a shift toward a tolerogenic phenotype (fewer signals from M1 macrophages to T cells). These pathway-level changes suggest that C1orf50-high cancer cells create an immunosuppressive microenvironment by upregulating checkpoint and matrix-related pathways, while downregulating pro-inflammatory signals.

Fig. 4.

C1orf50-associated modulation of tumor-immune cell communication networks in ovarian cancer single-cell transcriptomes. A UMAP plot of annotated cell types used for CellChat analysis. B Bar plots showing the total number of inferred ligand-receptor interactions and overall interaction strength in C1orf50-low and C1orf50-high groups. C Heatmaps depicting differences in the number (left) and strength (right) of interactions between source (y-axis) and receiver (x-axis) cell types. Red indicates higher signaling in the C1orf50-high group, blue indicates higher signaling in the C1orf50-low group. Top bar plots represent cumulative incoming signals, and side bar plots indicate outgoing signals. D Bubble plots of differentially enriched ligand-receptor pairs between C1orf50-high and C1orf50-low cancer cells. Dot size denotes statistical significance, and color indicates communication probability. Immune checkpoint- and suppression-related interactions (e.g., NECTIN-TIGIT, PGE2-PTGER, MIF-CD74) are enriched in C1orf50-high tumors. E Bar plots of pathway-level information flow from cancer cells to selected immune cell subsets, comparing C1orf50-high (red) and C1orf50-low (blue) tumors. Abbreviations: DC, dendritic cell; M1, M1 macrophage; M2, M2 macrophage. Information flow represents the overall communication strength of a signaling pathway

C1orf50 overexpression is associated with genomic instability and altered DNA repair pathway regulation in ovarian cancer

To evaluate the impact of C1orf50 expression on DNA repair capacity in ovarian cancer, we compared genomic instability metrics, mutational signatures, and DNA repair gene expression between tumors with high and low C1orf50 expression. As shown in Figs. 5A-C, C1orf50-high tumors exhibited significantly higher TMB (p = 0.0044), HRD scores (p = 0.00031), and LOH fractions (p = 0.0037) than the low-expression group. BRCA1/2 gene alterations were also more frequent (p = 0.012), with a modest predominance of BRCA2 events (Fig. 5D). Mutational signature analysis identified enrichment of single-base substitution signature 13 (SBS13) in C1orf50-high tumors (22.5% vs 9.7%, p = 0.014; Fig. 5E) and a greater prevalence of insertion-deletion signature 8 (ID8), which is linked to HRD (56.4% vs 44.6%, p = 0.12; Fig. 5F). Gene expression analysis (Fig. 5G) revealed a broad decrease in the expression of key DNA repair genes within the homologous recombination (HR), Fanconi anemia, and mismatch repair (MMR) pathways. This included genes such as ATM, ATR, BARD1, BRCA1/2, FANCC, FANCD2, FANCI, MSH3, PMS2, PRKDC, and TP53BP1. In contrast, several other repair-related genes, including NHEJ1, PAXX, EPCAM, PCNA, RAD51B, and RAD51C, showed significantly increased expression. In summary, C1orf50 overexpression is strongly associated with extensive genomic instability in ovarian cancer, characterized by elevated TMB, HRD, LOH, distinct mutational and copy-number signatures, and extensive suppression of DNA repair pathways. The selective upregulation of specific repair components may reflect compensatory remodeling of the repair network supporting, tumor survival under high genomic stress.

Fig. 5.

C1orf50 overexpression is associated with genomic instability and altered regulation of the DNA repair pathway in ovarian cancer. A-C Genomic instability metrics comparing C1orf50-high and C1orf50-low ovarian tumors, including tumor mutation burden (TMB; p = 0.0044), homologous recombination deficiency (HRD) scores (p = 0.00031), and loss of heterozygosity (LOH) fraction (p = 0.0037) (Wilcoxon rank-sum test). D Stacked alteration frequencies of BRCA1 and BRCA2, showing a higher prevalence of BRCA1/2 alterations in the C1orf50-high group (p = 0.012), with a modest predominance of BRCA2 events (Fisher’s exact test). E–F Mutational signature analysis demonstrating enrichment of the APOBEC-associated SBS13 (22.5% vs. 9.7%, p = 0.014; Fisher’s exact test), and a higher prevalence of the HR-deficiency-associated ID8 (56.4% vs. 44.6%, p = 0.12; Fisher’s exact test) in C1orf50-high tumors. G Expression of DNA repair genes across HRR, NHEJ, MMR, and Fanconi anemia pathways. Most core repair genes (ATM, ATR, BARD1, BRCA1/2, FANCC/D2/I, MSH3, PMS2, PRKDC, TP53BP1) were broadly downregulated, while a subset (NHEJ1, PAXX, EPCAM, PCNA, RAD51B/C) showed significant upregulation (Wilcoxon rank-sum test)

Discussion

In this study, we conducted a comprehensive investigation of C1orf50 using a pan-cancer cohort followed by an in-depth analysis focused on ovarian cancer. C1orf50 overexpression is consistently associated with genomic instability and immune suppression, suggesting that this gene may be involved in pathways linking these two tumor hallmarks.

C1orf50 copy number gain and overexpression were detected across multiple tumor types, with the strongest signal observed in ovarian cancer (Figs. 1 and 2). Pan-cancer analysis revealed that high C1orf50 expression was consistently associated with greater chromosomal instability and impaired DNA repair capacity, as evidenced by its positive correlation with TMB and HRD. C1orf50-high tumors also exhibited an immunosuppressive phenotype, characterized by the enrichment of regulatory T cells (Tregs) and the depletion of resting memory CD4⁺ T cells, suggesting a role in promoting immune evasion.

Focusing on ovarian cancer (Figs. 3, 4, 5), we found that C1orf50 amplification was a major contributor to its overexpression. C1orf50-high tumors displayed significantly elevated TMB, higher HRD scores, and increased frequencies of BRCA1/2 alterations, consistent with defective homologous recombination repair. Pathway-level analysis revealed broad suppression of DNA repair processes, including the non-homologous end joining (NHEJ), the Fanconi anemia, and mismatch repair (MMR) pathways. At the same time, microsatellite instability-related gene sets were enriched in the high-expression group. At the gene level, most canonical repair genes (ATM, ATR, BRCA1, BRCA2, FANCC, FANCD2, MSH3, PMS2) were downregulated. In contrast, a subset of repair components (NHEJ1, PAXX, EPCAM, PCNA, RAD51B, RAD51C) was selectively upregulated, indicating a compensatory activation of alternative repair mechanisms that may help sustain tumor survival under genomic stress.

Proteomic profiling provided deeper insight into the molecular effects of C1orf50 overexpression. RPPA analysis showed that C1orf50-high tumors had increased levels of the immune checkpoint molecule B7-H3 (CD276) and the antioxidant protein PRDX1, while displaying decreased phosphorylation of PKCα (T638/T641). The induction of B7-H3 aligns with the strong immunosuppressive signals observed at the transcriptomic and single-cell levels, supporting the idea that C1orf50 may help tumors evade host immune responses. Upregulation of PRDX1 indicates these tumors are better able to neutralize reactive oxygen species [36, 37], which could give them a survival advantage under oxidative stress conditions common in the tumor environment. Conversely, the reduction in PKCα phosphorylation, often linked to adhesion and tumor-suppressive signaling [38, 39], suggests a loss of regulation that may further promote tumor growth, invasion, and immune evasion. Overall, these proteomic changes demonstrate how C1orf50 amplification coordinates both internal survival mechanisms and external immune-modulating processes, highlighting its potential as a therapeutic target in ovarian cancer.

Immune cell profiling confirmed that C1orf50-high ovarian tumors were enriched for Tregs and M2 macrophages, with elevated expression of immunosuppressive molecules. Cell–cell communication analysis revealed strengthened signaling through the NECTIN-TIGIT axis and other pathways supporting immune suppression and tumor tolerance. These features indicate that C1orf50 shapes a microenvironment that may hinder anti-tumor immunity and reduce responsiveness to immunotherapies.

In summary, these results highlight C1orf50 as a potential oncogenic factor that promotes genomic instability while simultaneously fostering immune evasion in ovarian cancer. The combination of DNA repair defects and selective upregulation of alternative repair components suggests that C1orf50-high tumors may be particularly vulnerable to DNA-damaging agents, such as platinum compounds or PARP inhibitors. In parallel, the enrichment of immune checkpoint molecules, particularly B7-H3, together with increased infiltration of immunosuppressive cell types, suggests that these tumors may be amenable to immune checkpoint blockade, particularly in combination with DNA damage-targeted therapies. Such a dual approach could exploit both the impaired DNA repair machinery and the immunosuppressive signaling environment characteristic of these tumors. In addition, the observed downregulation of phosphorylated PKCα points to broader signaling alterations, as PKCα is involved in pathways regulating cell proliferation, survival, and immune modulation. Its reduced activity may therefore influence not only tumor cell-intrinsic behavior but also the dynamics of tumor-immune interactions, further shaping the therapeutic vulnerabilities of C1orf50-high tumors.

In conclusion, C1orf50 overexpression is strongly associated with genomic instability, DNA repair suppression, and an immunosuppressive tumor microenvironment in ovarian cancer. This study provides hypothesis-generating evidence that C1orf50 may serve as a molecular link between genomic instability and immune modulation, warranting further experimental validation.

Abbreviations

CIN

Chromosomal instability

CNA

Copy-number alteration

COSMIC

Catalogue of somatic mutations in cancer

EMT

Epithelial-mesenchymal transition

FGA

Fraction of genome altered

FIGO

International federation of gynecology and obstetrics

GEO

Gene expression omnibus

GSEA

Gene set enrichment analysis

HRD

Homologous recombination deficiency

HRR

Homologous recombination repair

LOH

Loss of heterozygosity

MMR

Mismatch repair

MSI

Microsatellite instability

MSigDB

Molecular signatures database

NES

Normalized enrichment score

NHEJ

Non-homologous end joining

PCA

Principal component analysis

RPPA

Reverse-phase protein array

SBS

Single base substitution (mutational signature)

scRNA-seq

Single-cell RNA sequencing

TCGA

The cancer genome Atlas

TCPA

The cancer proteome Atlas

TMB

Tumor mutational burden

TPM

Transcripts per million

Treg(s)

Regulatory T cell(s)

UMAP

Uniform manifold approximation and projection

Authors’ contributions

A.R. and Y.O. contributed equally to this work: performed data analysis, visualization, original draft preparation, review, and editing. A.O., V.D.C., S.A.: performed data analysis, writing, review, and editing. T.P.: performed data analysis, visualization. S.T.: project administration, writing, review, and editing. A.F.: contributed to conceptualization, funding acquisition, investigation, supervision, writing of the original draft, review, and editing. A.T.: contributed to conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, writing of the original draft, review, and editing. All authors discussed the results and contributed to the final manuscript.

Data availability

We thank the TCGA Research Network (https://www.cancer.gov/tcga) for generating the genomic datasets and the cBioPortal for Cancer Genomics (https://www.cbioportal.org) for providing processed TCGA data. Single-cell RNA-seq data used in this study were obtained from the Gene Expression Omnibus (GEO) under accession number GSE158722.

Declarations

Ethical approval and consent to participate

This study used publicly available, de-identified data from The Cancer Genome Atlas (TCGA) and previously published studies. Therefore, Institutional Review Board (IRB) approval and informed consent were not required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.