The Cancer Genome Atlas (TCGA) datasets, which include prognostic data, clinical information, and gene expression profiles, were obtained from UCSC Xena (xenabrowser.net). These datasets include gene expression profiles from 34 types of cancer, covering tumor tissues (N=9807) and normal tissues (N=8295). Gene expression data were transformed using log2(x+1). Single-cell sequencing datasets were downloaded from the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/), including GSE189357, GSE242889, GSE263995, and GSE184198.

RNA expression differences of PDCL3 between multiple human tumor cases and corresponding non-tumor cases were analyzed and compared using the TCGA dataset. The differential expression of PDCL3 was further validated at the protein level using UALCAN (https://ualcan.path.uab.edu/) and the Human Protein Atlas (HPA) (http://www.proteinatlas.org/). Furthermore, RNA expression of PDCL3 was compared across tumor samples at different stages.

The expression level of PDCL3 in various tumors was used to differentiate tumor samples from normal samples. The R package ROCR was applied to generate the receiver operating characteristic (ROC) curve to evaluate the diagnostic efficacy of PDCL3.

The Kaplan–Meier (KM) method was employed to construct survival curves based on PDCL3 expression. The R packages survminer and survival were used to perform curve plotting. The optimal cut-off value was determined to divide tumor cases into two groups for survival analysis.

The cBioPortal database (http://www.cbioportal.org/) was used to examine the mutation characteristics of PDCL3 across pan-cancers.

The R package GSVA was used to quantify the level of immune infiltration in tumor cases, followed by an analysis of the correlation between PDCL3 expression and immune infiltration. Finally, the results of this correlation analysis were visualized using the R package ggplot2.

Spearman’s correlation was used for the correlation analysis. The relationship between PDCL3 expression and TMB or MSI was analyzed separately. Radar plots were then applied to visualize the results of the correlation analysis.

Tumor cases were divided into two groups based on the median PDCL3 expression level as the cut-off value. The R package limma was used to analyze the differences between the two groups, and the differentially expressed genes (DEGs) related to PDCL3 were identified. The threshold for filtering DEGs was set at an absolute log2 fold change ≥1.0. The DEGs were extracted, and the R package clusterProfiler was used to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses to explore the potential functional mechanisms of PDCL3 in tumors.

The GSE189357, GSE242889, GSE263995, and GSE184198 datasets were used to analyze the expression levels of PDCL3 in cell clusters. The R package Seurat was used for quality control of all single-cell datasets. Dimensionality reduction and clustering were performed using uniform manifold approximation and projection (UMAP). Based on the marker genes of each cell cluster, the clusters were annotated using the CellMarker 2.0 database (http://bio-bigdata.hrbmu.edu.cn/CellMarker/).

The human NSCLC cell lines A549 and H1299 were obtained from the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco, Grand Island, USA) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin under standard conditions (37°C, 5% CO₂). After 24 hours of incubation, siRNA transfection was performed using Lipofectamine 8000 (Beyotime Biotechnology, China) in serum-free medium. The siRNA sequences (designed and synthesized by Nanning Genesis Biotechnology Co., Ltd, China) were as follows:

Total RNA was reverse-transcribed into cDNA using the PrimeScript RT Master Mix (Takara Bio, Japan). RT-qPCR was subsequently conducted on a LightCycler 480 system (Roche, Switzerland) with 10 μg of cDNA template, 2X Q3 SYBR qPCR Master Mix (ToloBio, China), and gene-specific primers. Gene expression levels were quantified using the 2-ΔΔCt method. The PDCL3 primer sequences (synthesized by Sangon Biotech, China) were as follows:

Cell proliferation was assessed using the BeyoClick™ EdU-555 Cell Proliferation Detection Kit (Beyotime Biotechnology, China) and the Cell Counting Kit-8 (CCK-8; Beyotime Biotechnology). For EdU assays, cells were incubated with EdU solution 24 hours post-transfection. Nuclei were counterstained with Hoechst 33342, and fluorescence images were captured using an EVOS M7000 fluorescence microscope (Thermo Fisher Scientific, USA). For CCK-8 assays, 10% (v/v) CCK-8 reagent was added to the culture medium at the indicated time points (24, 48, and 72 hours). Absorbance was measured at 450 nm using a microplate reader (BioTek, USA).

Cells were seeded into six-well plates and cultured until reaching 80-90% confluency. A uniform wound was created in the monolayer using a sterile 200 μL pipette tip. After washing with PBS to remove detached cells, images of the wound area were acquired at 0 hours (baseline) and 24 hours post-wounding using a microscope (Nikon, Japan).

Cell migration and invasion assays were performed using Transwell chambers (8 μm pore size; LABSELECT, China). Briefly, cells were harvested 24 hours post-transfection, resuspended in serum-free medium, and seeded into the upper chamber with 250 μL. The lower chamber was filled with 700 μL of complete medium containing 10% FBS. The chambers were incubated for 36 hours.

The t-test was applied to analyze the differential expression between two groups. Spearman’s correlation method was used for the correlation analysis. A p-value of less than 0.05 indicated that the results were statistically significant (ns, p > 0.05; *, p < 0.05; **, p < 0.01; ***, p < 0.001).

The pan-cancer analysis results indicated that PDCL3 was upregulated in most cancers ( Figure 1A ). Data from the UALCAN and HPA databases further confirmed that PDCL3 expression was upregulated in breast invasive carcinoma (BRCA), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), ovarian cancer (OC), and pancreatic adenocarcinoma (PAAD) at the protein level ( Figures 1B–M ).

The area under the ROC curve (AUC) for PDCL3 in BRCA, COAD, GBM, HNSC, LIHC, LUAD, LUSC, STAD and THYM were 0.882, 0.768, 0.886, 0.770, 0.948, 0.768, 0.916, 0.759 and 0.775, respectively ( Figures 2A–I ). All AUC values were greater than 0.7, indicating that PDCL3 exhibited good diagnostic performance for these tumors.

PDCL3 expression was upregulated in ACC, BRCA, LIHC, and LUAD samples with high T stage ( Figures 3A, C, F, H ). Additionally, PDCL3 tended to be highly expressed in ACC, HNSC, and LUAD samples with lymph node metastasis ( Figures 3B, D, I ), and was also highly expressed in KIRP cases with distant metastasis ( Figure 3E ). A significant difference in PDCL3 expression was observed among LIHC or LUAD samples across different pathological stages ( Figures 3G, J ).

Survival analyses from the TCGA database revealed a worse outcome for tumor samples with upregulated PDCL3 expression ( Figures 4A–L ).

The cBioPortal database was used to analyze genetic alterations of PDCL3 across different tumor types. Our analysis revealed that PDCL3 alterations in cancer were primarily mutations and amplifications ( Figure 5A ), with the most common alteration type being missense mutations ( Figure 5B ). Genetic alterations in PDCL3 were associated with poor overall survival and disease-specific survival in STAD, though no significant difference was observed in disease-free survival ( Figure 5C ).

In ACC, a positive correlation was observed between PDCL3 expression and T helper 2 (Th2) cell infiltration; In contrast, PDCL3 expression was negatively correlated with several immune cell types, including Mast cells, Cytotoxic cells, NK CD56bright cells, Th1 cells, Central memory T cells (Tcm), TFH, B cells, T cells, Neutrophils, NK CD56dim cells, CD8 T cells, plasmacytoid dendritic cells (pDCs), and Macrophages ( Figure 6A ). In BRCA and GBM, PDCL3 expression positively correlated with Th2 cell infiltration, but negatively correlated with pDC infiltration ( Figures 6B, C ). In LIHC,PDCL3 expression was positively associated with Th2 cells, Macrophages, and NK CD56bright cells, but negatively associated with Th17 cells and Eosinophils ( Figure 6E ). In LUAD, PDCL3 expression was positively correlated with Th2 cell infiltration, but inversely correlated with infiltration of seven other immune cell types, including T cells, Mast cells, Cytotoxic cells, B cells, Tcm, Macrophages, and pDC ( Figure 6F ). In STAD, PDCL3 expression showed a positive relationship with Th2 cells, but a negative correlation with Mast cells ( Figure 6G ). In uterine corpus endometrial carcinoma (UCEC), PDCL3 expression positively correlated with Th2 cells but was inversely correlated with NK CD56bright cells, NK cells, pDC, and Effector memory T cells (Tem) ( Figure 6H ). In uveal melanoma (UVM), PDCL3 expression was positively correlated with five immune cell types (Th cells, Tcm, Th2 cells, Tem, and T gamma delta cells [Tgd]) but negatively correlated with five others (Th17 cells, pDC, NK cells, NK CD56bright cells, and Mast cells) ( Figure 6I ). Taken together, the immune infiltration analysis indicates that PDCL3 expression consistently showed a positive correlation with Th2 cell infiltration across all tumor types, while it was inversely correlated with pDC infiltration, except in LIHC.

PDCL3 expression showed a positive correlation with TMB in several cancers, including glioblastoma multiforme (GBMLGG, r =  0.4278), low grade glioma, LGG) (r  =  0.3031), LUAD (r  =  0.2533), BRCA (r  =  0.2250), stomach adenocarcinoma (STES, r  =  0.1644), sarcoma (SARC, r  =  0.1373), kidney pan-cancer (KIPAN, r  =  0.1762), stomach adenocarcinoma (STAD, r  =  0.2679), prostate adenocarcinoma (PRAD, r  =  0.0895), head and neck squamous cell carcinoma (HNSC, r  =  0.0992), kidney renal clear cell carcinoma (KIRC, r  =  0.1712), bladder urothelial carcinoma (BLCA, r  =  0.1802), and adrenocortical carcinoma (ACC, r  =  0.3764). In contrast, a negative correlation was observed between PDCL3 expression and TMB in thyroid carcinoma (THCA, r  =  -0.1158) ( Figure 7A ).

Regarding MSI, PDCL3 expression was positively correlated with MSI in STES (r  =  0.0867), STAD (r  =  0.0998), UCEC (r  =  0.1463), HNSC (r  =  0.1702), KIRC (r  =  0.1432), LUSC (r  =  0.2474), and UVM (r  =  0.3135). On the other hand, PDCL3 expression was negatively correlated with MSI in GBMLGG (r  =  -0.3073), LGG (r  =  -0.0920), COAD (r  =  -0.1861), colon adenocarcinoma and rectum adenocarcinoma (COADREAD, r  =  -0.1806), and KIPAN (r  =  -0.1246) ( Figure 7B ).

There were 917, 1810, 844, and 444 DEGs related to PDCL3 in LUAD, LIHC, STAD, and BRCA, respectively ( Figures 8A, D, G, J ). Functional enrichment analysis was conducted for the four groups of DEGs. The biological process (BP) was presented with 5 terms in GO analysis. DEGs of LUAD were concentrated in “microtubule-based movement”, “cilium movement”, “cilium or flagellum-dependent cell motility”, “cilium-dependent cell motility”, and “axoneme assembly” ( Figure 8B ). DEGs of LIHC were concentrated in “signal release”, “positive regulation of secretion”, “positive regulation of secretion by cell”, “hormone metabolic process”, and “cellular response to xenobiotic stimulus” ( Figure 8E ). DEGs of STAD were concentrated in “regulation of membrane potential”, “muscle system process”, “synapse organization”, “muscle contraction”, and “synapse assembly” ( Figure 8H ). DEGs of BRCA were concentrated in “skin development”, “keratinocyte differentiation”, “epidermal cell differentiation”, “keratinization”, and “membrane depolarization” ( Figure 8K ). Ten pathways annotated by DEGs were presented in KEGG analysis. DEGs of LUAD were clustered in the following terms: “Neuroactive ligand-receptor interaction”, “cAMP signaling pathway”, “Metabolism of xenobiotics by cytochrome P450”, “Drug metabolism-cytochrome P450”, “Complement and coagulation cascades”, “Retinol metabolism”, “Steroid hormone biosynthesis”, “Chemical carcinogenesis-DNA adducts”, “Pentose and glucuronate interconversions”, and “Ascorbate and aldarate metabolism” ( Figure 8C ). DEGs of LIHC were clustered in the following terms: “Neuroactive ligand-receptor interaction”, “Calcium signaling pathway”, “Bile secretion”, “Retinol metabolism”, “Metabolism of xenobiotics by cytochrome P450”, “Drug metabolism-cytochrome P450”, “Chemical carcinogenesis-DNA adducts”, “Drug metabolism-other enzymes”, “Steroid hormone biosynthesis”, and “Glycosphingolipid biosynthesis-lacto and neolacto series” ( Figure 8F ). DEGs of STAD were clustered in the following terms: “Neuroactive ligand-receptor interaction”, “Calcium signaling pathway”, “cAMP signaling pathway”, “Cytoskeleton in muscle cells”, “Cell adhesion molecules”, “Protein digestion and absorption”, “Pancreatic secretion”, “Insulin secretion”, “Circadian entrainment”, and “Gastric acid secretion” ( Figure 8I ). DEGs of BRCA were clustered in the following terms: “Neuroactive ligand-receptor interaction”, “Chemical carcinogenesis-receptor activation”, “Drug metabolism-cytochrome P450”, “Metabolism of xenobiotics by cytochrome P450”, “Retinol metabolism”, “Chemical carcinogenesis-DNA adducts”, “Drug metabolism-other enzymes”, “Nicotine addiction”, “Pentose and glucuronate interconversions”, and “Proximal tubule bicarbonate reclamation” ( Figure 8L ).

After dimensionality reduction and clustering of single-cell sequencing data, each cell cluster was annotated according to marker genes ( Figures 9A–D ). By analyzing the expression of PDCL3 in LUAD cell clusters, it was found that PDCL3 expression was mainly in cancer cells, lung cells, M1 macrophages, M2 macrophages, mast cells, natural killer cells, plasma cells, and T cells ( Figure 9E ). In LIHC, PDCL3 was predominantly expressed in cancer cells, dendritic cells, exhausted T (Tex) cells, hepatic progenitor cells, hepatic stellate cells, hepatocytes, and macrophages ( Figure 9F ). Analysis of PDCL3 expression in BRCA cell clusters revealed that PDCL3 expression was primarily in cancer cells, macrophages, and pericytes ( Figure 9G ). In STAD, the expression of PDCL3 was most prominent in cancer cells, chief cells, endocrine cells, endothelial cells, fibroblasts, myeloid cells, and T cells ( Figure 9H ). In the differential analysis of single-cell data from LUAD, PDCL3 was upregulated in IAC compared to MIA and AIS, suggesting that PDCL3 may promote the invasive growth of LUAD ( Figure 9I ). Differential analysis was performed in 13 cell clusters from LUAD, and PDCL3 was upregulated in T cells compared to other cell clusters, indicating that the role of PDCL3 in LUAD may be related to T cells ( Figure 9J ).

To investigate the functional role of PDCL3 in NSCLC, si-PDCL3 was applied to A549 and H1299 cell lines. RT-qPCR analysis confirmed a significant reduction in PDCL3 mRNA levels following transfection with si-PDCL3-2, validating the silencing efficiency ( Figure 10A ). Functional assays were subsequently performed to assess the phenotypic consequences of PDCL3 depletion. CCK-8 and EdU proliferation assays revealed that PDCL3 knockdown significantly suppressed NSCLC cell growth, as evidenced by decreased absorbance values and EdU-positive cell ratios ( Figures 10B, C ). Furthermore, Transwell migration/invasion assays demonstrated that PDCL3 silencing attenuated both the migratory and invasive capacities of NSCLC cells ( Figure 10D ), and impaired migration was observed in the scratch wound closure assay ( Figure 10E ).