Fig 2: Elevated ERG and AR correlates with higher DLX1 levels representing advanced-stage aggressive disease.a Heatmap showing TCGA-PRAD RNA-Seq data for ERG and DLX1 expression in primary PCa specimens (n = 498). Shades of red and blue represent the log2 (norm_count+1) value. b Representative core of PCa tissue microarray (TMA) showing RNA in situ hybridization (RNA-ISH) scoring pattern for DLX1 in 144 PCa patient specimens, score 0 represents DLX1 negative, score 1 signifies low DLX1, score 2, and score 3 represents medium and high DLX1 expression, respectively. Scale bar, 50 µm. c Bar plot showing percentage of patients negative (DLX1–) and positive (DLX1+) for DLX1 expression based on the scoring pattern (P < 0.0001). d Same as b except Immunohistochemistry (IHC) for ERG (top panel) and RNA-ISH for DLX1 (bottom) in 144 PCa patient tissue specimens. e Bar plot showing percentage of patients with varying DLX1 expression in ERG-positive (ERG+) and -negative (ERG–) PCa cases (P < 0.0001). f Same as e except an association between ERG and DLX1 expression status and Gleason scores of PCa patients (P = 0.0003). g Same as b except representative tumor cores showing IHC for AR, ERG, and RNA-ISH for DLX1 representing AR+/ERG+/DLX1+ status in 144 PCa patient tissue specimens. h Same as g except for representative AR + /ERG-/DLX1+ patient in TMA containing 144 PCa specimens. i Bar plot depicting percentage of patients with positive and negative DLX1 expression in AR + /- and ERG + /- respective background. Patients showing low, medium, and high DLX1 expression categorized as DLX1-positive (P < 0.0001). j Contingency table for the AR and DLX1 status in TMA patient specimens. P-value denotes Fisher’s exact test (P < 0.0001). k Bar plot showing association between DLX1 expression and Gleason scores of tumor specimens (P < 0.0001). l Same as k, except association of DLX1 expression with tumor stage (P = 0.0041). m Correlation plot of AR and DLX1 using Stand Up To Cancer (SU2C) dataset by categorizing patients as TMPRSS2-ERG positive (top panel) and negative samples (bottom panel). P-value was calculated using two-tailed test with 95% confidence interval (P < 0.0001). n Heatmap showing DLX1 levels in tumor specimens representing distant metastatic sites of metastatic CRPC patients. o Same as n except for RNA-ISH for DLX1 expression in TMA containing 121 mCRPC biospecimens collected from various metastatic sites. Scale bar, 25 µm. p Bar plot showing DLX1 expression in percent metastatic sites from CRPC patients same as n (P < 0.0001). For panels d, g, and h scale bars are represented as 300 µm for the entire core and 50 µm for the inset image. For panels, c, e, f, i, k, l, and p P-value were calculated using Chi-Square test. Source data are provided as a Source Data file.

Fig 3: BET inhibitor alone or in combination with Enzalutamide downregulates DLX1 expression and mitigates its oncogenic properties.a ChIP-Seq data (GSE55064) showing ERG enrichment on DLX1 promoter in VCaP cells treated with JQ1 or vehicle control for 24 h. H3K27Ac represents active promoter in untreated cells. PLAU used as a positive control. b ChIP-qPCR data showing relative ERG enrichment on DLX1 promoter in VCaP cells treated with JQ1 (0.5 µM) for 48 h (P < 0.0001). c Q-PCR data showing relative expression of target genes in VCaP cells treated with Enza (10 µM), JQ1 (0.5 µM) alone or in combination for 48 h. KLK3 used as a positive control for JQ1 treatment. d Same as c except immunoblot. ß-actin was used as loading control. e Same as c except 22RV1 cells. f Same as e except immunoblot (P < 0.0001). g Cell proliferation assay in VCaP cells treated with drug conditions as mentioned in c. h Same as g except 22RV1 cells (P < 0.0001). i Same as g except 42D ENZR cells (P < 0.0001). j Boyden Chamber Matrigel migration assay in VCaP cells using same treatment conditions as c. Inset shows representative image of the migrated cells (scale bar 30 µm). k Same as j except 22RV1 cells (scale bar 30 µm, P < 0.0001). l Same as j except Enza-resistant, 42D ENZR cells (scale bar 30 µm, P < 0.0001). m Fluorescence intensity of the catalyzed ALDH substrate in VCaP cells under same treatment conditions as c. Marked windows show ALDH1+ percent cell population. n Same as m except 22RV1 cells. o Same as m except 42D ENZR cells. p Bar plot showing number of spheres formed in prostatosphere assay using VCaP under same treatment conditions as c. Inset shows representative image of the spheres formed (scale bar 100 µm). q Q-PCR data for the DLX1 expression using RNA isolated from VCaP prostatospheres. Data shown from three biologically independent samples (n = 3). Data represent mean ± SEM. For panels b, c, e two-way ANOVA, Tukey’s multiple comparison test; g, h, i One-way ANOVA, Dunnett’s multiple comparison test; j–l, p, q One-way ANOVA, Tukey’s multiple comparison was applied. Source data are provided as a Source Data file.

Fig 4: Genetic ablation of DLX1 inhibits oncogenic properties.a Q-PCR data showing expression of EMT markers in 22RV1-DLX1-KO and control SCR cells. b Immunoblots showing vimentin and E-cadherin using same cells as a. ß-actin was used as loading control. c Same as b except for cleaved PARP, cleaved Caspase-3 and Bcl-xL. d Flow cytometry-based apoptosis assay using 22RV1-DLX1-KO and control cells (top panel) and DLX1-silenced VCaP cells (bottom panel). e Flow cytometry data for cell cycle distribution using same cells as in d. f Q-PCR data for stem cell markers using same cells as d. g Flow cytometry data depicting ABCG2 (top panel) and CD44 (bottom panel) expression in DLX1 overexpressing RWPE1 cells, 22RV1-DLX1-KO, and DLX1-silenced VCaP cells. h Fluorescence intensity of catalyzed ALDH substrate in 22RV1-DLX1-KO and control cells. Marked windows show ALDH1 + percent cell population. i Same as h, except for DLX1 silenced VCaP cells. j Q-PCR data showing expression of target genes in 22RV1-DLX1-KO and control cells(P < 0.0001). k Schema showing the chromosomal location of DLX1 binding motif (DBM1/2) at the ALDLH1A1 (top) and HNF1A (bottom) promoters. l ChIP-qPCR data of DLX1 (top panel) and H3K9Ac (bottom panel) on ALDH1A1 in 22RV1-DLX1-KO and SCR control cells (P < 0.0001). m Same as in l except for the HNF1A promoter. Data shown from three biologically independent samples (n = 3). Data represent mean ± SEM. For panels, a, f, and j Two-way ANOVA, Dunnett’s multiple comparison test; l, m Two-way ANOVA Turkey’s multiple comparison test was applied. Source data are provided as a Source Data file.

Fig 5: High DLX1 expression associates with poor prostate cancer prognosis and promotes disease progression.a Dot plot showing DLX1 expression in PCa patients (n = 498) and matched normal (n = 52) in TCGA-PRAD RNA-Seq dataset, data represents log2 (norm_count+1), center depicts mean ± (standard deviation) SD (P < 0.0001). b Dot plot of DLX1 expression using microarray profiling data (GSE35988) comprising benign (n = 28), localized (n = 59), and mCRPC (n = 35) patient specimens, data represents log2 (norm_mRNA), center depicts mean ± SD (P < 0.0001). c Kaplan–Meier plot showing survival probability in TCGA-PRAD (n = 498) dataset categorized in high DLX1 (DLX1Hi) and low DLX1 (DLX1Lo) expression. d RNA-Seq data showing DLX1 transcript read counts in publicly available datasets (GSE128399 and GSE118206). e Cell proliferation assay using isogenic RWPE1 cells overexpressing DLX1 at indicated time-points (P < 0.0001). f Foci formation assay using same cells as e (P < 0.0001). g Boyden Chamber Matrigel migration assay using same cells as e (P < 0.0001). Representative images for panels f (scale bar 500 µm) and g (scale bar 100 µm) are shown as inset. h Cell proliferation assay using 22RV1-DLX1-KO (C-1, C-2, and C-3 are independent clones) and control cells at indicated time-points (P < 0.0001). i Boyden Chamber Matrigel migration assay using same cells as h (P < 0.0001). j Anchorage-independent soft agar assay using same cells as h (P < 0.0001). Representative images for panels i and j are shown as inset (scale bar 100 µm). k DAVID analysis showing upregulated (right) and downregulated (left) biological processes in 22RV1-DLX1-KO against control cells. Bars represent the -log10 (P-value) and the frequency polygon (black line) denotes number of genes. l Same as k, except gene set enrichment analysis (GSEA) plots representing deregulated pathways. m Heatmap displaying downregulated genes involved in cancer stemness and EMT in 22RV1-DLX1-KO cells compared to control. Data shown from three biologically independent samples (n = 3). Data represent mean ± SEM unless specified. For panels, a Unpaired Student’s two-tailed t-test was applied; b One-way ANOVA with Dunnett’s multiple comparison test was applied; e Two-way ANOVA Sidak’s multiple comparison test; f, g Unpaired two-tailed Welch’s t-test; h Two-way ANOVA, Dunnett’s multiple comparisons test; i, j One-way ANOVA, Dunnett’s multiple comparisons test was applied. Source data are provided as a Source Data file.