The present study integrated transcriptomic analyses, computational network modeling, in vitro functional assays and pharmacological perturbation experiments to characterize the DNHD1-HDAC6 axis in prostate cancer. Publicly available datasets were systematically analyzed to evaluate gene expression patterns, clinical associations, functional pathways and prognostic significance. Experimental validation was performed in prostate cancer cell lines with distinct AR and lineage states. A systems biology framework was employed to identify downstream effectors and infer regulatory hierarchies and mechanistic validation and therapeutic targeting were subsequently employed.

Gene expression profiles were obtained from The Cancer Genome Atlas (TCGA)-prostate adenocarcinoma (PRAD) cohort, Gene Expression Omnibus (GEO) (https://www.cancer.gov/ccg/research/genome-sequencing/tcga) and additional independent datasets (GSE21032, GSE35988, GSE48403 and GSE36133: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE21032; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE35988; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE48403; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE36133). Cross-platform pancancer datasets, including Cell 2015 (https://www.cbioportal.org/study/summary?id=prad_tcga_pub), Firehose Legacy (https://www.cbioportal.org/study/summary?id=prad_tcga) and PanCancer Atlas (https://www.cbioportal.org/study/summary?id=prad_tcga_pan_can_atlas_2018), were used for comparative oncogenomic analyses. Raw or processed expression matrixes were retrieved and harmonized across datasets by using platform-specific normalization strategies. Log2-transformed expression values were used for downstream analyses when available. Samples were stratified according to clinical annotations, including tumor stage, Gleason score, metastatic status and treatment history.

The differential gene expression between tumor and normal tissues, as well as between early- and late-stage tumors, was assessed using nonparametric statistical frameworks suitable for RNA-sequencing and microarray data distributions. Genes associated with tumor progression were identified through stage-stratified comparisons. Pancancer expression profiling was conducted across multiple epithelial malignancies to determine the tumor-type specificity of DNHD1 and HDAC6 dysregulation.

The present study evaluated associations between gene expression and clinicopathological variables, including pathological T and N stage, Gleason score, residual tumor status and treatment exposure, by using Spearman rank correlation analysis. Stratified subgroup analyses were performed to assess expression trends across disease severity gradients. A two-sided P<0.05 was considered to indicate significance.

Overall survival (OS) analyses were performed using the online tool Gene Expression Profiling Interactive Analysis (GEPIA2) (http://gepia2.cancer-pku.cn). For the analyses of disease-free interval (DFI) and progression-free interval (PFI) in the TCGA-PRAD cohort, as well as shorter disease-free survival (DFS) in the GSE21032 dataset, raw clinical data were extracted from the respective supplementary tables and analyzed using Kaplan-Meier survival models. Patients were stratified into high- and low-expression groups on the basis of median gene expression thresholds, unless otherwise specified. Hazard ratios (HR) and 95% confidence intervals (Cis) were estimated using Cox proportional hazards regression models. Combined gene signature analyses were performed using dual-gene stratification models to assess additive prognostic effects.

To construct DNHD1-associated regulatory networks, correlation analyses were performed across the TCGA-PRAD and multiple independent prostate cancer transcriptomic cohorts. Genes with Spearman's ρ≥0.3 or Spearman's ρ≤-0.3 relative to DNHD1 were defined as DNHD1-associated gene sets. To increase robustness and reduce cohort-specific bias, the present study generated integrated gene lists by using intersection analysis across multiple independent datasets. The resulting gene networks were subjected to downstream pathway enrichment analysis.

Functional enrichment analysis was performed using ingenuity pathway analysis (IPA). Canonical pathways, upstream regulatory networks and disease and function annotations were evaluated on the basis of enrichment significance scores (−log₁₀ P). Biological processes associated with cytoskeletal remodeling, vesicular trafficking, organelle organization, metabolic reprogramming and cilium assembly were specifically interrogated. In Gene Ontology classification, the present study focused on structural cellular organization pathways, particularly those related to microtubule dynamics and ciliary regulation.

Genes positively correlated with both DNHD1 and HDAC6 were identified through Venn-based integration across transcriptomic datasets. To ensure inclusion of only protein-coding genes, pseudogenes and long noncoding RNAs were filtered out. The resulting gene set, defined as the DNHD1-HDAC6 downstream transcriptional signature, was used for correlation analysis and survival modeling.

Human prostate cancer cell lines representing different AR and lineage states were used: LNCaP (cat. no. BCRC 60088), PC-3 (cat. no. BCRC 60122), 22RV1 (cat. no. BCRC 60545) and DU145 (cat. no. BCRC 60348) cells lines were purchased from the Bioresource Collection and Research Center. Cells were maintained under standard conditions (37°C; 5% CO₂) in appropriate culture media supplemented with fetal bovine serum and antibiotics (32,33). Cell line identity was consistent with American Type Culture Collection or validated repositories and cells were routinely tested for contamination.

DNHD1, E2F6 and HDAC6 expression was modulated using a 2nd-generation lentiviral shRNA-mediated knockdown and cDNA overexpression systems (Table SI). Lentiviral particles were generated by co-transfecting HEK-293T cells (obtained from the Bioresource Collection and Research Center; cat. no. BCRC 60019) with pLKO.1-shRNA or pLenti6/V5-DEST-cDNA transfer plasmids, alongside the packaging plasmid pCMVΔR8.9 and envelope plasmid pMD.G at a mass ratio of 4:3:1 (total 10 µg plasmid per 10-cm dish: 5 µg transfer plasmid, 3.75 µg pCMVΔR8.9, and 1.25 µg pMD.G) using Lipofectamine^® 3000 (Thermo Fisher Scientific, Inc.).

Lentiviral supernatants were harvested at 48 and 72 h post-transfection, filtered through a 0.45-µm membrane filter, and stored at −80°C. To establish stable cell populations, target cells were transduced at a multiplicity of infection (MOI) of 5 in the presence of 8 µg/ml polybrene. Stable knockdown cell lines were selected with 2 µg/ml puromycin for 3 days and maintained in 0.5 µg/ml puromycin. Stable overexpression cell lines were selected with 5 µg/ml blasticidin for 4 days and maintained in 2 µg/ml blasticidin. Subsequent experiments were performed 48 h post-selection. Knockdown and overexpression efficiency were validated using reverse transcription-quantitative (RT-q) PCR and immunoblotting (32,33).

Target cells were seeded at a density of 1×10⁶ cells/well prior to RNA extraction. Total RNA extraction, cDNA synthesis, and qPCR assays were performed strictly according to the manufacturers' protocols. Total RNA was isolated and reverse transcription was performed to generate cDNA using the TOOLSQuant II Fast RT Kit (cat. no. KRT-BA06-2; BIOTOOLS Co., Ltd.). Gene expression levels were quantified using SYBR Green-based qPCR assays with PowerTrack SYBR Green Master Mix (cat. no. A46012; Applied Biosystems; Thermo Fisher Scientific, Inc.; Table SII). Relative expression was calculated using the 2^−ΔΔCq method (34), with housekeeping genes as internal normalization controls. For promoter binding analysis, potential E2F6-binding motifs within the HDAC6 promoter region were identified using a bioinformatic sequence prediction tool (JASPAR database). A putative E2F6-binding site (5′-GAGAGGGAAGA-3′, positions −781 to −770 bp relative to the transcription start site) was identified within the HDAC6 promoter. Chromatin immunoprecipitation was performed using the EZ-Magna ChIP A/G kit (MilliporeSigma) according to the manufacturer's instructions. In brief, target cells were harvested at a density of 1×107 cells per ChIP reaction and cross-linked with 1% formaldehyde (MilliporeSigma) to fix protein-DNA complexes. Chromatin was sheared by sonication using an Ultrasonic Disruptor UD-201 (TOMY Digital Biology Co., Ltd.; 15 cycles of 30 sec on/30 sec off at 4°C) to yield DNA fragments ranging from 200–1,000 bp, which were verified by agarose gel electrophoresis. A portion of the clarified lysate (10%) was saved as Input DNA for normalization. Cell lysates (100 µg per reaction) were immunoprecipitated with anti-E2F6 (2 µg/reaction; cat. no. 31464-1-AP; Proteintech Group, Inc.) or control IgG (2 µg/reaction; cat. no. 30000-0-AP; Proteintech Group, Inc.) overnight (12–16 h) at 4°C with magnetic A/G beads. Immunocomplexes were washed sequentially with Low Salt, High Salt, LiCl and TE Wash Buffers provided in the kit. Specific promoter DNA fragments encompassing the predicted E2F6 binding motif (positions −781 to −770 bp) were subsequently amplified and quantified through qPCR using PowerTrack SYBR Green Master Mix (cat. no. A46012; Applied Biosystems; Thermo Fisher Scientific, Inc.). The thermocycling conditions were: Initial denaturation at 95°C for 3 min; 35 cycles of denaturation at 95°C for 30 sec, annealing at 58°C for 30 sec and elongation at 72°C for 30 sec; followed by a final extension at 72°C for 5 min. The ChIP enrichment efficiency was determined using the Percent Input approach. Briefly, the Cq value of the 10% Input fraction was adjusted for dilution [ΔCq (Input)=Cq (Input)-log2(10)], and the percent input was calculated using the formula: % Input=100×2[^{ΔCq (Input)-Cq [ChIP)]}. Signal acquisition and quantitative processing were carried out using QuantStudio Design and Analysis Software (version 1.3; Applied Biosystems; Thermo Fisher Scientific, Inc.).

Whole-cell lysates were prepared using TOOLS RIPA Lysis Buffer (cat. no. TAAR-ZBZ5; BIOTOOLS Co., Ltd.) supplemented with protease and phosphatase inhibitors. Protein concentrations were determined using the Bio-Rad Protein Assay Kit II (cat. no. 5000002; Bio-Rad Laboratories, Inc.) based on the Bradford method.

Total protein (30–60 µg per lane) was separated through sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 12% polyacrylamide gels and transferred onto polyvinylidene difluoride membranes. These membranes were blocked with 5% non-fat milk in Tris-buffered saline with 0.1% Tween-20 for 1 h at room temperature. Membranes were then incubated overnight at 4°C with antibodies against DNHD1 (cat. no. PAU491Hu01; CLOUD-CLONE CORP), E2F6 (cat. no. 31464-1-AP; Proteintech Group, Inc.), HDAC6 (cat. no. 67250-1-lg; Proteintech Group, Inc.), α-tubulin (cat. no. 2144; Cell Signaling Technology, Inc.), acetylated α-tubulin (cat. no. 3971; Cell Signaling Technology, Inc.), AR (cat. no. 54653; Cell Signaling Technology, Inc.) and PSA (cat. no. 2475; Cell Signaling Technology, Inc.). Followed primary antibody incubation, membranes were washed and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies, including Goat Anti-Rabbit IgG (H+L) HRP (1:5,000; cat. no. GTX213110-01; GeneTex, Inc.) or Goat Anti-Mouse IgG HRP (1:5,000; cat. no. GTX213111-01; GeneTex, Inc.), for 1 h at room temperature. Protein signals were detected using chemiluminescence and quantified, as necessary. Protein bands were visualized using SuperKine West Femto Maximum Sensitivity Substrate (cat. no. BMU102-EN; Abbkine Scientific Co., Ltd.; distributed by Proteintech Group, Inc.) and captured using the Bio-Rad ChemiDoc MP Imaging System (Bio-Rad Laboratories, Inc.). Protein bands were visualized using SuperKine West Femto Maximum Sensitivity Substrate (cat. no. BMU102-EN.; distributed by Proteintech Group, Inc.) and captured using the Bio-Rad ChemiDoc MP Imaging System (Bio-Rad Laboratories, Inc.). Densitometric analysis was performed using Image Lab Software (version 6.1; Bio-Rad Laboratories, Inc.), with β-Actin (cat. no. GTX26276; GeneTex, Inc.) serving as the internal loading control for normalization.

For cell proliferation assays, target cells were seeded into 96-well plates at a density of 2×10³ cells/well and cultured at 37°C in a humidified incubator containing 5% CO₂. Cell proliferation was evaluated using AlamarBlue Cell Viability Reagent (cat. no. DAL1025; Invitrogen; Thermo Fisher Scientific, Inc.) strictly following the manufacturer's protocol. Briefly, AlamarBlue reagent was added to each well (10% of total volume), and cells were incubated for 2–4 h at 37°C. Fluorescence/absorbance was measured using a microplate reader at defined time points. For colony formation assays, cells were seeded into 6-well plates at a density of 2×10³ cells/well and cultured at 37°C in a 5% CO₂ atmosphere for 14 days to allow colony development. Following incubation, culture media were discarded, and cells were fixed with a methanol-acetic acid mixture (7:1, v/v) for 30 min at room temperature. The fixed cells were subsequently stained with 0.1% crystal violet solution (MilliporeSigma) for 30 min at room temperature and gently rinsed with distilled water. Stained plates were imaged, and colonies were captured under a stereomicroscope. Colony quantification was performed using ImageJ software (version 1.53t; National Institutes of Health) via the Analyze Particles plugin after thresholding.

Cell migration and invasive capacities were evaluated using a 48-well microchemotaxis Boyden chamber system (Neuro Probe, Inc.) to assess cell motility and invasive capacity. For the invasion assays, polycarbonate membranes with an 8-µm pore size (Neuro Probe, Inc.) were pre-coated with Matrigel (BD Biosciences; Corning, Inc.; diluted 1:8 in serum-free medium) at 37°C for 1 h to form a reconstituted basement membrane. Uncoated membranes were used for migration assays. Target cells were harvested, washed, and resuspended in serum-free culture medium. A total of 2×104 cells in 50 µl of serum-free medium (with or without specified drug treatments) were loaded into the upper chamber wells. The lower chamber wells were filled with 25–30 µl of culture medium supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, Inc.) as a chemoattractant. The chamber was assembled and incubated at 37°C in a humidified 5% CO₂ atmosphere for 24 h (migration) or 48 h (invasion). After incubation, non-migrated cells on the upper surface of the membrane were gently scraped off using a cotton swab. Cells that migrated or invaded cells were fixed with 4% paraformaldehyde (or methanol-acetic acid 7:1) for 15–30 min at room temperature and stained with Giemsa stain solution (cat. no. G5637; MilliporeSigma) for 20–30 min at room temperature. Stained membranes were mounted, and migratory/invasive cells were visualized and captured using an inverted light microscope (Olympus Corporation). Cell numbers were quantified across three randomly selected high-power microscopic fields per well using ImageJ software (version 1.53t; National Institutes of Health) (32,33).

To selectively inhibit HDAC6 catalytic activity, human prostate cancer LNCaP cells were seeded in multi-well plates and allowed to adhere overnight at 37°C in a humidified atmosphere containing 5% CO₂. Cells were then treated with Tubastatin A (cat. no. HY-13271A; MedChemExpress) at indicated concentrations (5 and 10 µM) or with dimethyl sulfoxide (DMSO; MilliporeSigma) as the vehicle control for 24 h at 37°C. The final concentration of DMSO in the culture medium was maintained below 0.1% (v/v) across all experimental groups. The efficacy and specificity of enzymatic HDAC6 inhibition were verified by evaluating the hyperacetylation levels of its downstream substrate, acetylated α-tubulin, via immunoblotting.

To establish functional dependency, combinatorial gain-of-function and loss-of-function experiments were performed in prostate cancer cells (LNCaP, PC-3) using a lentiviral gene delivery system. For DNHD1 overexpression, prostate cancer cells were transduced with recombinant lentiviruses encoding full-length DNHD1 cDNA cloned into the pLenti6/V5-DEST Gateway vector (cat. no. V49610; Invitrogen; Thermo Fisher Scientific, Inc.) or an empty destination control vector. Transductions were performed at 37°C for 24 h in the presence of 8 µg/ml Polybrene (hexadimethrine bromide; MilliporeSigma). To establish stable DNHD1-overexpressing populations, cells were selected with Blasticidin (5.0 µg/ml; cat. no. A1113903; Gibco; Thermo Fisher Scientific, Inc.) for 5–7 days. Subsequently, DNHD1-overexpressing cells were transduced with lentiviral particles expressing specific short hairpin RNA targeting HDAC6 (sh-HDAC6; pLKO.1-shHDAC6) or a non-targeting scramble control shRNA (sh-NC) at an optimized multiplicity of infection (MOI=10-20) with 8 µg/ml Polybrene at 37°C for 24 h. Transduced cells were co-selected with Puromycin (2.0 µg/ml; MilliporeSigma) for 48–72 h. Overexpression and knockdown efficiencies were validated at 48–72 h post-transduction via RT-qPCR and immunoblotting (using an anti-V5 tag antibody for exogenous DNHD1 detection). Transduced cells were harvested and re-seeded into downstream functional assays to evaluate rescue outcomes, including cell proliferation (48 h readout), clonogenicity (14 days incubation), and Boyden chamber migration/invasion assays (24–48 h incubation) under standard conditions (37°C; 5% CO₂).

All statistical analyses were conducted using R (version 4.3.1; R Foundation for Statistical Computing, Vienna, Austria). Data were presented as means ± standard deviations, unless otherwise stated. Between-group comparisons were performed using unpaired Student's t-test, depending on relevant distribution assumptions. Correlation was assessed using Spearman rank correlation analysis. Survival significance was assessed using log-rank tests and Cox proportional hazards regression models. P<0.05 was considered to indicate a statistically significant difference.

To identify genes associated with advanced prostate cancer and to determine progression-associated transcriptional alterations, transcriptomic profiles from TCGA-PRAD) cohort were stratified by tumor stage (T2a vs. T4). Differential expression analysis revealed a gene subset markedly upregulated in higher-stage tumors; among these genes, DNHD1 was consistently enriched (Fig. 1A), suggesting that it is involved in prostate cancer progression rather than in tumor maintenance alone. In the TCGA-PRAD cohort, DNHD1 expression was markedly elevated in tumor tissues compared with adjacent nontumor tissues (P=0.0003; Fig. 1B; Table SIII), further corroborating its association with malignant transformation. Pancancer analysis demonstrated that DNHD1 upregulation occurs not only in prostate cancer but also across various epithelial malignancies, including cholangiocarcinoma, colon adenocarcinoma (COAD), head and neck squamous cell carcinoma, renal clear cell carcinoma (KIRC), hepatocellular carcinoma and rectal adenocarcinoma (Fig. 1C). However, prognostic stratification analysis revealed that DNHD1-associated high-risk clinical outcomes were most consistently observed in PRAD, KIRC and COAD (Fig. 1D), suggesting that DNHD1 has a context-dependent functional relevance in tumor progression rather than a universal oncogenic role. Notably, DNHD1 expression was significantly and positively correlated with Gleason score in prostate cancer tissues (P<0.0001; Fig. 1E). Moreover, it increased progressively across stratified Gleason grading groups (Fig. 1F; Table SIV), suggesting that DNHD1 expression is strongly associated with histopathological severity and tumor aggressiveness. To validate these findings across independent datasets, multiple external prostate cancer cohorts (GSE21032, GSE35988 and GSE48403) were analyzed (Fig. 1G; Table SV). Consistent with the observations in TCGA-PRAD, DNHD1 expression was markedly upregulated in GSE21032. Elevated DNHD1 expression was also observed in both localized prostate cancer and metastatic castration-resistant prostate cancer (mCRPC) tissues, suggesting its persistence throughout disease progression and metastatic evolution. Notably, ADT was associated with further upregulation of DNHD1 expression, indicating its potential involvement in therapy-adaptive transcriptional reprogramming. Survival analysis using TCGA-PRAD cohort data demonstrated that higher DNHD1 expression was significantly associated with poorer overall survival (OS; P=0.039, HR=3.8) and shorter disease-free survival (DFS; P=3.7×10⁻⁵; HR=2.4; Fig. 1H). These prognostic associations were independently validated in the GSE21032 cohort, where DNHD1 expression also correlated with shorter DFS (P=0.0034, HR=3.443, 95% CI=1.494–7.931; Fig. 1I; Table SVI). Taken together, these results suggest that in prostate cancer, DNHD1 is not only differentially expressed but also associated with a transcriptional program linked to progression, histopathological severity and adverse clinical outcomes. These consistent associations across independent cohorts further corroborate the potential role of DNHD1 as a progression-associated molecular indicator in prostate cancer.

The oncogenic role of DNHD1 in prostate cancer remains largely undefined. To investigate its potential functional relevance, the present study first examined DNHD1 expression across prostate cancer cell lines by using the CCLE dataset GSE36133 (Table SVII). Notably, DNHD1 expression demonstrated a lineage-associated distribution, with lower expression in AR-positive luminal-like cell lines (LNCaP and 22RV1) but higher expression in AR-negative, more aggressive and mesenchymal-like cell lines (PC-3 and DU145; Fig. 2A). Thus, DNHD1 may be associated with lineage plasticity and a more dedifferentiated tumor state in prostate cancer. Protein-level validation through immunoblotting across prostate cancer cell lines further confirmed this differential expression pattern (Fig. 2B), corroborating the transcriptional and translational consistency of DNHD1 deregulation in distinct prostate cancer cellular contexts. To investigate its functional role, the present study generated stable DNHD1-knockdown and -overexpression models by using lentiviral short hairpin RNA (shRNA) and cDNA systems. On the basis of the endogenous expression profiles, LNCaP and PC-3 cells were used for gain-of-function and loss-of-function experiments, respectively. Efficient modulation of DNHD1 expression was validated by RT-qPCR and immunoblotting (Fig. 2C and D).

Functional assay results demonstrated a potent regulatory role of DNHD1 in prostate cancer cell function. DNHD1 knockdown markedly impaired proliferative capacity in PC-3 cells, whereas ectopic DNHD1 expression considerably enhanced proliferation in LNCaP cells, as indicated by the Alamar Blue assay results (Fig. 2E). These findings were further corroborated by colony formation assay results: DNHD1 knockdown suppressed long-term clonogenic survival, whereas DNHD1 overexpression promoted robust colony expansion (Fig. 2F), indicating a critical role for DNHD1 in sustaining the proliferative and survival capacities of prostate cancer cells. Considering the clinical relevance of metastatic dissemination in prostate cancer progression, the present study subsequently evaluated the regulatory role of DNHD1 in prostate cancer cell motility and invasive behavior. Fibronectin- and Matrigel-based Boyden chamber assays revealed that DNHD1 knockdown markedly reduced both migratory and invasive potential in PC-3 cells; by contrast, DNHD1 overexpression led to considerable improvements in LNCaP cell motility and invasion (Fig. 2G and H). These findings indicate that DNHD1 functionally enhances both proliferative and metastatic phenotypes in prostate cancer cells, consequently establishing DNHD1 as a functional regulator of prostate cancer cell growth and motility. Notably, the associations observed between DNHD1 expression and AR status indicate a potential link between DNHD1 and prostate cancer lineage plasticity: In prostate cancer cells, DNHD1 upregulation may promote a shift toward more aggressive, AR-independent states with enhanced metastatic competence.

To elucidate the downstream regulatory network associated with DNHD1 in prostate cancer, the present study performed a correlation-based system-level analysis using transcriptomic datasets from TCGA-PRAD, Cell 2015, Firehose Legacy and PanCancer Atlas cohorts (35–38). Genes with a robust positive correlation (Spearman's ρ≥0.3) or negative correlation (Spearman's ρ≤-0.3) with DNHD1 expression were integrated across datasets to construct a high-confidence DNHD1-associated gene signature (Fig. 3A; Table SVIII). Functional enrichment analysis of 2,145 positively correlated and 463 negatively correlated genes conducted using IPA revealed that DNHD1-associated transcriptional programs are strongly enriched in pathways regulating cytoskeletal organization, vesicular trafficking, organelle biogenesis and cellular metabolic rewiring (Fig. 3B; Table SIX). Notably, among the top-ranked canonical pathways, cilium assembly emerged as a consistently enriched biological process, highlighting a potential association between DNHD1 and structural signaling regulation in prostate cancer. Within this cilium-associated regulatory network, HDAC6 was identified as a biologically and clinically relevant candidate downstream effector. HDAC6 is a well-established cytoplasmic deacetylase, regulating microtubule dynamics through α-tubulin deacetylation. It plays a key role in primary cilium disassembly, protein homeostasis and oncogenic stress adaptation. Notably, although HDAC6 is implicated in prostate cancer progression and AR signaling regulation (39), its position within upstream regulatory networks remains unclear. In the TCGA-PRAD cohort, HDAC6 expression was positively correlated with DNHD1 expression (R=0.4; P=3.1×10⁻²⁰; Fig. 3C). The result of the network-based analysis therefore suggested that HDAC6 represents a functional convergence node, linking DNHD1-associated transcriptional programs to cytoskeletal and ciliary remodeling pathways in prostate cancer. To validate the clinical relevance of this association, the present study subsequently examined HDAC6 expression in TCGA-PRAD and independent prostate cancer cohorts. HDAC6 expression was significantly upregulated in primary tumor tissues compared with in the adjacent nontumor tissues (P=0.0230; Fig. 3D; Table SX). Notably, HDAC6 expression was strongly and positively correlated with DNHD1 expression across multiple independent datasets, including TCGA-PRAD (Spearman's ρ=0.4283; P=0.0092) and validation cohorts (Fig. 3E; Table SX), confirming the robustness of this regulatory association. In clinical specimens, higher HDAC6 expression was positively correlated with a higher Gleason score, suggesting a role in prostate cancer progression and aggressiveness (Fig. 3F; Table SIV). Moreover, comparative analysis across these specimens revealed a significant positive correlation between DNHD1 and HDAC6 expression (Spearman's ρ=0.3436; P<0.0001; Fig. 3G; Table SIV), confirming a coordinated transcriptional relationship between the two molecules. Consistently, elevated HDAC6 expression was observed across localized prostate cancer (GSE21032) and mCRPC (GSE35988) cohorts, further indicating its association with advanced prostate cancer. Notably, ADT led to further upregulation of HDAC6 expression (GSE48403), suggesting that HDAC6 is involved in the progression to ADT-adaptive tumor states (Fig. 3H; Table SV). Survival analysis revealed that high HDAC6 expression was significantly associated with poor OS (P=0.047; HR=4.4) and DFS (P=0.039; HR=1.5) in the TCGA-PRAD cohort (Fig. 3I). The relationship between HDAC6 expression and DFS was independently validated in the external cohort GSE21032 (P=0.0453; HR=2.332; 95% CI=1.025–5.306). Notably, combined stratification based on DNHD1 and HDAC6 expression revealed a high-risk patient subgroup with the poorest clinical outcomes across all cohorts (P<0.0001), suggesting additive or synergistic prognostic value for both genes (Fig. 3J; Table SVI). Together, these results indicate that HDAC6 may be a key downstream effector within the DNHD1-associated transcriptional network, affording a model in which DNHD1-HDAC6 signaling constitutes a coordinated regulatory axis governing cytoskeletal remodeling, ciliary dynamics and prostate cancer progression. In prostate cancer, this axis may not only be correlated with tumor aggressiveness and therapy adaptation but also define a clinically relevant high-risk molecular subclass of the condition.

To further delineate the mechanistic relationship between DNHD1 and HDAC6 in prostate cancer, the present study first evaluated HDAC6 expression across prostate cancer cell lines by using CCLE transcriptomic datasets (GSE36133). It was observed that HDAC6 expression exhibited a lineage-associated distribution, with lower expression in AR-positive luminal-like prostate cancer cells than in AR-negative prostate cancer cell lines (Fig. 4A; Table SVII). This distribution closely parallels DNHD1 expression patterns, further suggesting a potential functional coupling between DNHD1 and HDAC6 within various prostate cancer cell states. Additionally, integrative correlation analysis revealed a strong positive association between DNHD1 and HDAC6 expression across prostate cancer datasets (Spearman's ρ=0.90; Fig. 4B; Table SVII), corroborating a strong transcriptional association between the two genes. To determine whether DNHD1 functionally regulates HDAC6 expression, the present study performed gain-of-function and loss-of-function experiments in prostate cancer cell models. qPCR and immunoblotting analyses revealed that DNHD1 overexpression markedly upregulated HDAC6 mRNA and protein expression in LNCaP cells, whereas DNHD1 knockdown considerably downregulated HDAC6 mRNA and protein expression in PC-3 cells (Fig. 4C and D). These findings suggested that DNHD1 positively regulates HDAC6 expression at both the transcriptional and the translational level. Notably, reciprocal perturbation experiments revealed that modulating HDAC6 expression did not affect DNHD1 expression (Fig. 4E), indicating a unidirectional regulatory hierarchy in which DNHD1 acts upstream of HDAC6. These results establish HDAC6 as a downstream effector rather than a feedback regulator within the DNHD1 signaling axis. To assess the functional contribution of HDAC6 in prostate cancer progression, the present study established HDAC6 gain-of-function and loss-of-function models in prostate cancer cell lines (Fig. 4F and G). Consistent with a previous report (29), HDAC6 knockdown markedly suppressed proliferation and clonogenic growth in PC-3 cells, whereas HDAC6 overexpression enhanced proliferative capacity in LNCaP cells (Fig. 4H and I). Moreover, Boyden chamber assays revealed that in both cell lines, HDAC6 knockdown considerably reduced migratory and invasive potential, whereas ectopic HDAC6 expression promoted aggressive motility (Fig. 4J and K). These results confirmed that HDAC6 expression is essential and sufficient to drive prostate cancer cell growth and metastatic behavior. To establish whether HDAC6 functionally mediates DNHD1-driven oncogenic phenotypes, the present study performed combinatorial epistasis experiments. DNHD1 overexpression substantially enhanced proliferation, colony formation, migration and invasion in LNCaP cells. Notably, simultaneous silencing of HDAC6 and DNHD1 markedly attenuated all DNHD1-induced oncogenic phenotypes (Fig. 4L-O), indicating that HDAC6 is required for the complete execution of DNHD1-mediated tumor-promoting effects.

Upstream regulator analysis of DNHD1-simulated networks revealed E2F6 as a potential transcription factor involved in the transactivation of HDAC6 under DNHD1 regulation (P=0.0398; Fig. S1A). Transcriptomic profiling of TCGA-PRAD cohort data revealed that E2F6 expression was positively correlated with the expression of both DNHD1 (R=0.22; P=1.2×10⁻⁶) and HDAC6 (R=0.29; P=1.0×10⁻¹⁰). Clinical outcome analysis further revealed that in patients with prostate cancer, E2F6 upregulation led to significantly worse OS than low E2F6 expression did (P=0.046; HR=3.8; Fig. S1B). To investigate whether E2F6 functionally modulates HDAC6 expression, the present study performed gain-of-function and loss-of-function experiments. E2F6 overexpression considerably upregulated HDAC6 expression in LNCaP cells, whereas E2F6 knockdown substantially downregulated HDAC6 mRNA and protein expression, as validated through qPCR and immunoblotting (Fig. S1C and D). To evaluate whether E2F6 directly transactivates HDAC6, the present study performed promoter sequence analysis and identified a putative E2F6-binding motif (5′-GAGAGGGAAGA-3′) located between positions −781 and −770 bp within the HDAC6 promoter (Fig. S1E). Chromatin immunoprecipitation coupled with qPCR demonstrated an enriched recruitment of E2F6 to the HDAC6 promoter relative to the immunoglobulin G (IgG) control; this effect was further enhanced after ectopic E2F6 expression in LNCaP cells (Fig. S1F). Thus, E2F6 drives HDAC6 transactivation by directly binding to the HDAC6 promoter. Given the regulatory hierarchy established in the aforementioned text, the present study further explored the functional role of E2F6 in the DNHD1-HDAC6 axis. Immunoblotting revealed that DNHD1 overexpression stimulated E2F6 expression, whereas E2F6 knockdown suppressed DNHD1 expression (Fig. S1G), suggesting a reciprocal regulatory relationship between DNHD1 and E2F6. Functional assays revealed that E2F6 knockdown markedly attenuated DNHD1-induced proliferation, migration and invasion in LNCaP cells (Fig. S1H-J). By contrast, although E2F6 transactivated HDAC6, HDAC6 knockdown slightly downregulated E2F6 expression (Fig. S1K). The results of epistatic rescue experiments demonstrated that HDAC6 knockdown substantially reversed the enhancements in colony-forming, migratory and invasive capacities conferred by E2F6 overexpression (Fig. S1L-N), indicating that HDAC6 is an indispensable downstream mediator of E2F6-driven malignant phenotypes.

These results further confirmed the presence of a DNHD1-HDAC6 axis in prostate cancer cells, in which DNHD1 is an upstream regulator that activates HDAC6 through E2F6 to promote proliferation and motility in tumor cells in vitro. Notably, HDAC6 is a critical epistatic mediator of DNHD1-driven malignancy, indicating that the DNHD1-HDAC6 axis is a functionally coherent regulatory module determining prostate cancer aggressiveness.

Given the vital role of HDAC6 in regulating microtubule dynamics, proteostasis and oncogenic signaling networks, its selective inhibition has emerged as a promising cancer treatment strategy (24–26). To evaluate the translational relevance of the DNHD1-HDAC6 axis in prostate cancer, the present study employed tubastatin A, a well-characterized selective HDAC6 inhibitor. In LNCaP cells, tubastatin A treatment resulted in a dose-dependent increase in acetylated α-tubulin levels (Fig. 5A), confirming that the enzymatic activity of HDAC6 was effectively inhibited. Time-course analysis further revealed sustained accumulation of acetylated α-tubulin after tubastatin A exposure, indicating strong suppression of HDAC6-dependent deacetylation activity (Fig. 5B). Notably, HDAC6 inhibition was accompanied by suppression of downstream AR signaling output, as indicated by decreased AR and PSA levels (Fig. 5C). Thus, the enzymatic activity of HDAC6 may functionally contribute to maintenance of AR-driven transcriptional programs in prostate cancer cells. Functionally, tubastatin A markedly impaired prostate cancer cell viability, clonogenic capacity, migratory behavior and invasive potential, demonstrating that pharmacological inhibition of HDAC6 effectively abrogates key malignant phenotypes associated with cancer progression (Fig. 5D-G). These results establish HDAC6 as a critical regulator of prostate cancer growth and metastatic competence. Notably, pharmacological inhibition of HDAC6 also attenuated DNHD1-driven oncogenic phenotypes. In DNHD1-overexpressing LNCaP cells, tubastatin A treatment effectively reversed DNHD1-induced upregulation of HDAC6 expression and suppressed associated AR and PSA expression (Fig. 5H). These findings indicate that DNHD1-mediated signaling is functionally dependent on the enzymatic activity of HDAC6. Combinatorial functional assays further revealed that HDAC6 inhibition considerably abrogated DNHD1-induced enhancement of proliferation, colony formation, migration and invasion (Fig. 5I-L). This epistatic evidence confirms that HDAC6 is required for the execution of DNHD1-driven cancer-promoting programs.

These results indicated that the DNHD1-HDAC6 axis is a pharmacologically tractable signaling module in prostate cancer. In addition to holding prognostic relevance, this pathway represents a functional therapeutic vulnerability, whereby HDAC6 inhibition can efficiently suppress DNHD1-driven tumor growth and metastatic behavior. Thus, targeting the DNHD1-HDAC6 axis is a potential therapeutic strategy for advanced prostate cancer.

To further delineate the clinical relevance of the DNHD1-HDAC6 axis, the present study elucidated its association with established pathological parameters and disease progression metrics in the TCGA-PRAD cohort. At the gene level, DNHD1 and HDAC6 expression demonstrated a consistent positive correlation across multiple pathological strata, including the N stage (Spearman's ρ=0.3308; P<0.0001; Fig. 6A) and T stage (Spearman's ρ=0.3203; P<0.0001; Fig. 6B), suggesting coordinated upregulation during prostate cancer progression. Notably, this coregulatory pattern persisted in residual tumor specimens, where DNHD1 expression remained positively correlated with HDAC6 expression (Spearman's ρ=0.3290; P<0.0001; Fig. 6C), indicating that the DNHD1-HDAC6 axis is maintained even in posttreatment disease contexts. Clinically, both DNHD1 and HDAC6 expression individually stratified patients with significantly worse disease-free interval (DFI) and progression-free interval (PFI). Higher DNHD1 expression was associated with a considerably poorer DFI (HR=5.733; P<0.0001); moreover, higher HDAC6 expression was associated with a considerably worse PFI (HR=2.157; P=0.0309; Fig. 6D and E; Table SXI). Notably, integrated analysis of the combined DNHD1-HDAC6 signature further enhanced prognostic discrimination, identifying a subgroup of patients with the poorest DFI (P<0.0001; Fig. 6F; Table SXI). A similar trend was observed for PFI: High-risk patients with markedly decreased survival probabilities demonstrated high expression of both DNHD1 and HDAC6 (Fig. 6G-I; Table SXII). Together, these findings indicated that the DNHD1-HDAC6 axis provides more additive prognostic information than single-gene models do, supporting its role as a coordinated molecular module rather than a set of independent biomarkers.

To further characterize the transcriptional architecture associated with the DNHD1-HDAC6 axis, the present study identified a set of 625 positively correlated genes through Venn-based integration of DNHD1- and HDAC6-associated expression profiles (Fig. 6J; Table SXIII). Pathway enrichment analysis of this gene set revealed that despite partial divergence from single-gene pathway outputs, cilium assembly remained the most consistently enriched biological process. Thus, in prostate cancer, ciliary regulation may represent a convergent functional output of DNHD1-HDAC6 signaling. After excluding non-protein-coding transcripts, the present study defined a refined 570-gene signature as the downstream transcriptional program of the DNHD1-HDAC6 axis (Fig. 6K; Table SXIV). This signature was strongly and positively correlated with both DNHD1 (Spearman's ρ=0.75; P<7.5×10⁻⁸⁹) and HDAC6 (Spearman's ρ=0.58, P<1.2×10⁻⁴⁵; Fig. 6L; Table SXV), indicating robust transcriptional coupling across independent patient samples. Notably, this axis-associated gene program also demonstrated significant prognostic value, stratifying patients with regard to OS (P=0.028; HR=4.3) and DFS (P=0.0062; HR=1.8; Fig. 6M).

To further assess whether the DNHD1-HDAC6 axis is an independent prognostic indicator in prostate cancer, the present study performed a multivariate Cox proportional hazards regression analysis including DNHD1, HDAC6 and KLK3 as covariates across multiple clinical endpoints (Table I). In the GSE21032 validation cohort (n=113), the independent risk factors for shortened DFS were identified to be DNHD1 (HR=1.015; 95% CI=1.001–1.028; P=0.0334) and HDAC6 (HR=1.012; 95% CI=1.003–1.020; P=0.0073). Consistent with these findings, evaluation within the larger TCGA-PRAD cohort revealed that the overall Cox models for both DFI and PFI were highly significant. In particular, shortened DFI was predicted by elevated expression of DNHD1 (HR=1.804; 95% CI=1.233–2.640; P=0.0024) and HDAC6 (HR=3.883; 95% CI=1.059–14.233; P=0.0407), independent of KLK3 expression. Moreover, both genes demonstrated strong independent prognostic value for PFI, with a 1.54-fold and 2.24-fold increase in progression risk conferred by DNHD1 (HR=1.541; 95% CI=1.238–1.919; P=0.0001) and HDAC6 (HR=2.238; 95% CI=1.139–4.395; P=0.0194), respectively. These multivariate analysis results confirmed that the DNHD1-HDAC6 axis affords a robust, independent molecular prognostic framework for predicting disease recurrence and progression in prostate cancer.

Taken together, these results support a systems-level model in which DNHD1 and HDAC6 operate as a coordinated regulatory axis that establishes a stable transcriptional state associated with tumor progression, ciliary dysregulation and adverse clinical outcomes in prostate cancer. Rather than acting as isolated biomarkers, DNHD1 and HDAC6 appear to converge on a shared downstream gene network that denotes a biologically and clinically coherent program underlying prostate cancer aggressiveness.