移行区富集的 WIF1+ 基底细胞亚型与良性前列腺增生相关A transition zone enriched WIF1+ basal cell subtype is associated with benign prostatic hyperplasia.

Abstract

Introduction

The human prostate is an androgen-regulated organ anatomically divided into distinct zones—peripheral zone (PZ), central zone (CZ), transition zone (TZ)—as well as the anterior fibromuscular stroma, which lacks epithelium. Each of the epithelial rich zones (PZ, CZ, and TZ) harbor at least some unique histological features, cellular composition, and disease susceptibilities. Notably, benign prostatic hyperplasia (BPH) arises almost exclusively in the TZ, whereas prostate cancer develops predominantly in the PZ [1]. This striking zonal selectivity underscores the need to understand spatial heterogeneity in prostate biology and pathology.

Recent advances in single-cell RNA sequencing (scRNA-seq) have enabled dissection of the prostate's complex architecture at unprecedented resolution [2-5]. These studies have uncovered substantial heterogeneity within both epithelial and stromal lineages, revealing zone-specific cell states, lineage trajectories, and molecular signaling pathways. For example, Henry et al identified two novel epithelial populations, termed club and hillock cells, which were more abundant in the TZ than in the PZ [2]. Yan et al further demonstrated age-associated changes, reporting a TFF3⁺ luminal subtype with elevated MYC and E2F activity enriched in the PZ, while the TZ exhibited stronger transcriptional signatures related to immunity and stemness [3]. Stromal heterogeneity has also been noted, with Joseph et al and Yan et al both describing proximal–distal fibroblast density differences [3, 4]. Despite these insights, basal cell heterogeneity has received limited attention and remains poorly understood, likely due to small sample sizes [5]. One recent study described two lineages of KRT5⁺ basal cells (KRT16⁺KRT17⁺ and KRT16⁻KRT17⁻), which were further subdivided into seven transcriptionally distinct subtypes [6]. The other study described multiple basal subpopulations, including KRT15⁺/c1 and DST⁺/c6 clusters representing a classical basal phenotype, a rare RARRES2⁺/c17 basal subtype, and S100A2⁺/c2 and PLCG2⁺/c3 subtypes that were enriched for basal and hillock cell signatures [7]. Most prior studies focused solely on normal prostate tissue, limiting their ability to connect cell states with disease contexts such as BPH.

BPH is characterized by non-malignant proliferation of epithelial and stromal cells, leading to prostatic enlargement that can lead to urethral obstruction, causing lower urinary tract symptoms (LUTS) that substantially reduce quality of life. Although the pathogenesis of BPH remains incompletely understood, dysregulated androgen signaling [8], chronic inflammation [9], and stromal–epithelial interactions [10] are considered key contributors. Additional studies have implicated cellular senescence, extracellular matrix remodeling, and hormonal imbalances in disease progression [11, 12].

In this study, we analyzed peri-cancerous prostate tissues from 10 patients spanning all three anatomical zones and identified a distinct basal cell population-WIF1⁺ basal cells. This subtype was significantly enriched in the TZ and in BPH tissues, pointing to a zonally restricted population potentially involved in BPH pathogenesis. These findings provide a potential novel cellular basis for the clinical observation that prostate cancer arises predominantly in the peripheral zone, whereas BPH develops almost exclusively in the transition zone.

Materials and methods

Ethics approval

The study was performed under the Johns Hopkins University School of Medicine Institutional Review Board (IRB protocols NA_00048544 and NA_00087094) approved protocol. Written informed consent was obtained from all participants and the study was conducted in accordance with the Declaration of Helsinki (https://www.wma.net/what-we-do/medical-ethics/declaration-of-helsinki/).

Human prostatectomy tissue punches

Prostate tissue specimens were collected from men diagnosed with primary prostate cancer undergoing radical prostatectomy at The Johns Hopkins Hospital. Prostatectomies were sectioned fresh from apex to base, and fresh tissue samples were collected using an 8 mm punch biopsy tool from each zone of the prostate. H&E staining was performed to confirm no cancer lesion was included in these punch areas. Freshly collected tissue punches were processed for scRNA-seq [13].

Dissociation of human prostate tissues

Dissected tissues were minced with razor blades and digested in 0.25% Trypsin–EDTA (Gibco, Thermo Fisher Scientific, Waltham, MA, USA; catalog no. 25200-072) for 10 min at 37°C, followed by incubation for 2.5  h at 37°C with gentle agitation in DMEM containing 10% FBS, 1 mg/ml Collagenase Type I (Gibco 17100-017), and 0.1 mg/ml of DNase I (Roche, Basel, Switzerland; catalog no. 10104159001). Digested tissues were centrifuged at 400×g for 5 min, washed with Hank's balanced salt solution (HBSS), and further incubated in 0.25% Trypsin–EDTA for 10 min at 37°C. Cells were suspended in DMEM containing 10% FBS and 0.4 mg/ml of DNase I and filtered through a 40 μm cell strainer.

Single-cell RNA-sequencing and data pre-processing

Libraries for scRNA-seq were prepared using the 10x Genomics Chromium Single Cell 3′ Library and Gel bead Kit V2 (10x Genomics, Pleasanton, CA, USA; catalog no. CG00052_RevF) following the manufacturer's protocol for each human prostatectomy tissue punch. The cDNA libraries were sequenced (150 bp paired-end) on the Illumina HiSeqX platform (Illumina, San Diego, CA, USA). Each sequenced library was demultiplexed to FASTQ files using Cell Ranger (10x Genomics). Cell Ranger (version 3.2.0) count pipeline was used to align reads to the GRCh38 transcriptome and create a gene-by-cell count matrix. The resulting gene expression matrix was loaded into Seurat (v4.0.4) in R (https://satijalab.org/seurat/, last accessed 28 August 2026). Low-quality cells were filtered out if genes were below 500, as well as if mitochondrial genes were > 20%.

Data integration and cell type identification

We applied the ‘anchor-based’ integration method to assemble multiple samples into an integrated dataset, following the Seurat integration workflow [14]. After running principal component analysis (PCA), FindNeighbors and FindClusters functions, we performed nonlinear dimensionality reduction with the RunUMAP function to obtain a two-dimensional representation. Differential gene expression analysis of previously characterized cell type-specific genes was used to identify the cell type for each cluster. A small cluster of cells expressing biomarkers from more than one cell type (epithelial, stromal, and immune) were considered as doublets and removed from downstream analysis.

Differential gene expression analysis

The FindMarker function in the Seurat package was performed to identify differentially expressed genes (DEGs) among cell clusters or samples. Statistical significance was assessed using the Wilcoxon rank-sum test, and p values were adjusted for multiple testing using the Benjamini–Hochberg method. Genes with log₂(fold-change) > 0.5 and adjusted p < 0.05 were considered significant DEGs.

Pathway enrichment analysis

Gene sets from the Molecular Signatures Database (MSigDB), including the Hallmark and Gene Ontology Biological Process (GOBP) collections, were imported using the R package msigdbr [15]. Pathway enrichment analysis of significant DEGs was performed using clusterProfiler with the hypergeometric test [16]. Significantly enriched pathways were identified based on adjusted p values. For analyses with a limited number of DEGs, specifically WIF1⁺ basal genes upregulated in BPH and identified by intersecting two comparisons, pathway enrichment was performed using Gene Set Enrichment Analysis (GSEA) via the MSigDB online platform (https://www.gsea-msigdb.org/gsea/msigdb/index.jsp). The top 10 pathways with a false discovery rate (FDR) q value < 0.05 were selected for visualization.

Ingenuity pathway analysis (IPA)

Upstream regulators that are likely to mediate the observed gene expression differences across basal cell clusters were analyzed using ingenuity pathway analysis (IPA) [17]. Biological networks constructed from known interactions in the published literature were used to infer upstream molecular regulators and pathways. Using observed differential gene expression, a z-score was derived from predicted up-regulation or down-regulation of relevant genes in a pathway. Seurat was used to perform differential gene expression analyses comparing WIF1⁺ basal to other basal cell clusters. The subsequent IPA analysis was implemented with DEGs with at least abs(log2FC) > 1 and a FDR adjusted p value < 0.01. The predicted pathways and upstream regulators were plotted with their associated z-scores.

Chromogenic in situ hybridization (CISH)

To prepare formalin-fixed paraffin-embedded (FFPE) tissue for chromogenic in situ hybridization (CISH) staining, slides were incubated for 30 min at 60°C and deparaffinized by incubating slides at room temperature (RT) for 10 min in xylene twice, and then subsequently incubated in 100% ethanol twice and finally left to air dry. Hydrogen peroxide solution was added to the slides for 10 min at RT. Slides were steamed in 1× RNAscope Target retrieval reagent (Advanced Cell Diagnostics, Newark, CA, USA; catalog no. 322000) at 100°C for 18 min, followed by Protease Plus (Advanced Cell Diagnostics; catalog no. 322331) digestion for 30 min at 40°C to allow target accessibility. The following probes were used for CISH staining: Hs-TP63 (Advanced Cell Diagnostics; catalog no. 601891-C1), Hs-NRG1 (Advanced Cell Diagnostics; catalog no. 311181-C1), Hs-VCAN (Advanced Cell Diagnostics; catalog no. 430071-C1), and Hs-WIF1 (Advanced Cell Diagnostics; catalog no. 429391-C2). Probes were added to slides and incubated in the HybEZ Oven (Advanced Cell Diagnostics, model HybEZ II) for 2 h at 40°C, followed by signal amplification and detection assay following the manufacturer's protocol. The C1 probe signal was detected with green color and C2 probe signal was detected with red color. Slides were counter-stained with 50% Gill's Hematoxylin for 30 s, baked for 15 min at 60°C, coverslipped with VectaMount permanent mounting medium (Vector Laboratories, Newark, CA, USA; catalog no. H-5000), and then scanned using a VENTANA DP200 scanner (Roche).

Cell–cell communication network analysis

CellChat [18] was used to explore the cell–cell communication between cell types. First, the normalized gene expression data were converted into a CellChat object. A ligand–receptor interaction database was used to predict signaling networks based on gene expression patterns. Next, communication probabilities were computed at the signaling pathway level by aggregating the probabilities of all ligand–receptor interactions associated with each pathway. We filtered out communications if there were fewer than 10 cells in the cell group.

Results

Identification of major cell types in human prostate

We performed scRNA-seq on prostate tissues from 10 individuals who underwent radical prostatectomy for treatment of localized prostate cancer. From each subject, benign regions were sampled from the PZ, CZ, and TZ, and were evaluated by histology to assess the presence of malignant glands. No invasive adenocarcinoma was identified in the benign tissue punches (Figure 1A,B). After excluding low-quality samples due to technical failures (e.g., wetting failure), we retained seven PZ, nine CZ, and nine TZ samples, yielding a total of 129,887 high-quality single cells across 25 samples. Data from the PZ were previously reported in the context of assessing tumor specific alterations [13], and were re-analyzed with the data from the other two zones in this study.

Figure 1.

Uniform manifold approximation and projection (UMAP) based dimensionality reduction and unsupervised clustering revealed distinct cell populations that were grouped primarily by cell type in the UMAP projection (Figure 1C). Based on canonical markers, we annotated epithelial (KRT8, KRT18), endothelial (KDR, FLT1), pericyte (RGS5, NOTCH3), T cell (CD3D, CD3E), myeloid (CD68, C1QB), fibroblast (DCN, LUM), smooth muscle (ACTG2, DES), B cell (CD79A, MS4A1), and mast cell (TPSAB1, TPSB2) populations (supplementary material, Figure S1). Differential gene expression confirmed cell-type specificity and reinforced these annotations (Figure 1D).

To further explore the epithelial cells, we subclustered epithelial cell clusters and performed dimensionality reduction with PCA followed by UMAP. This revealed canonical luminal (KLK3, ACPP), basal (KRT14, DST), and rare neuroendocrine (CHGA, CHGB) subtypes (Figure 1E,F).

Strikingly, we identified a population of cells situated between basal and luminal clusters in the UMAP coordinate system. These cells often expressed LTF or KRT13 (Figure 1F) among a host of genes previously linked to prostate inflammatory atrophy (PIA) [13, 19-22] and club/hillock cells [2], suggesting a potential intermediate epithelial state. Differential expression analysis highlighted unique gene signatures for these intermediate cells, including LTF, LCN2, OLFM4, MMP7, and KRT13 (Figure 1G).

WIF1⁺ basal cells are enriched in human prostate transition zone

Recent studies have highlighted some zonal differences in basal cells in the human prostate [5, 23]. In this study, our analysis revealed substantial heterogeneity within the basal compartment to add to our understanding of prostate basal cell states/subtypes. In initial UMAP projections, basal cell populations were separated by intermediate epithelial clusters (Figure 1E), suggesting underlying transcriptional diversity and prompting further analysis.

To dissect this heterogeneity, we performed subclustering of only the basal cells and identified four distinct basal subpopulations (Figure 2A). The first, characterized by robust expression of canonical basal markers including KRT5, KRT15, and KRT23, was designated KRT23⁺ basal cells and exhibited a classical basal gene expression program (Figure 2B, supplementary material, Figure S2). The second, marked by DDIT3, TXNIP, HES1, and MAFB (Figure 2B, supplementary material, Figure S2), displayed a transcriptional program consistent with cellular stress responses and paligenosis [24, 25], likely corresponding to the previously reported PLCG2⁺/c3 basal population [7]. The third, a minor population defined by FOXI1 and vacuolar-type H⁺-ATPase subunits, resembled previously reported ionocyte-like basal cells (Figure 2B; supplementary material, Figure S2) [7].

Figure 2.

Notably, a fourth basal subpopulation exhibited high expression of WIF1, VCAN, PPP1R14C, NPPC, and SFRP1 (Figure 2B; supplementary material, Figure S2), and closely resembled the previously reported Basal-5-VCAN subtype [6]. To further assess the specificity of these genes, we examined their expression across all prostate cell types. While VCAN and SFRP1 were also expressed in fibroblasts, WIF1 expression was exclusive and significantly enriched in this basal subtype (Figure 2C). Based on the specificity, we designated this novel population as the WIF1⁺ basal cell subtype.

To determine whether basal subtypes displayed zonal preferences, we quantified their relative abundance across all samples. Strikingly, the WIF1⁺ basal subtype showed strong enrichment in the TZ (Figure 2A,D).

To validate this striking zonal selectivity of the WIF1⁺ basal cell subtype, we performed CISH for TP63, a canonical basal marker, and WIF1 on prostatectomy specimens from multiple zones. Co-localization of TP63 and WIF1 confirmed that WIF1 expression was largely restricted to TP63⁺ basal epithelial cells in the TZ, supporting the classification of WIF1⁺ cells as a distinct basal subtype. Notably, TP63⁺WIF1⁺ basal cells were markedly enriched in the TZ, with nearly absent WIF1 expression observed in basal cells of the PZ or CZ. Rare WIF1-expressing cells located in the stromal compartment were also observed, mostly in the PZ and CZ (Figure 2E; supplementary material, Figure S3).

To further validate the molecular features of the WIF1⁺ basal cell subtype, we performed CISH for VCAN and WIF1 on the same prostatectomy specimens. Consistent with the scRNA-seq analysis and the expression patterns shown in Figure 2C, co-expression of VCAN and WIF1 was observed in basal cells within the TZ glands. In addition, VCAN expression was also detected in stromal fibroblast-like cells in both the TZ and PZ (supplementary material, Figure S4). These findings further support the existence of this distinct WIF1⁺VCAN⁺ basal cell population enriched in the TZ while demonstrating that VCAN expression alone is not specific to this epithelial subtype.

We next examined whether this basal subtype exists in the mouse prostate. Intriguingly, Wif1 was not expressed in any mouse basal epithelial cells. Instead, high expression was detected in a subset of fibroblasts (supplementary material, Figure S5), corresponding to the ductal fibroblast population previously described by Joseph et al [4] and subglandular fibroblast by our previous study, characterized by expression of Rorb [26]. These findings indicate that the WIF1⁺ basal cell subtype is absent in mouse prostate.

EMT, stromal remodeling, and immune modulation pathways are upregulated in WIF1⁺ basal cells

Differential gene expression among the four basal subtypes revealed 104 genes uniquely upregulated in WIF1⁺ basal cells compared with all other basal cell subtypes (Figure 3A; supplementary material, Table S1). To explore the biological functions of this basal subtype, we performed pathway analysis using Hallmark and Gene Ontology Biological Process (GOBP) terms from MSigDB. WIF1⁺ basal cells were most significantly enriched for the Hallmark epithelial-mesenchymal transition (EMT) pathway, along with angiogenesis, androgen response, coagulation, and apical junction gene sets, suggesting a specialized role in tissue remodeling and zonal dynamics (Figure 3B). Gene ontology (GO) analysis further implicated this subtype in prostate gland development and morphogenesis (Figure 3C).

Figure 3.

To complement these findings, we computationally inferred the activation and inhibition states of pathways and upstream regulators in WIF1⁺ basal cells compared with all other basal subtypes using IPA. The top activated pathways included extracellular matrix organization, collagen biosynthesis and modifying enzymes, collagen chain trimerization (Figure 3D; supplementary material, Table S2), suggesting a potential role for WIF1⁺ basal cells in extracellular matrix and basement membrane remodeling and glandular tissue architecture support. Based on these observations, we speculated that these WIF1⁺ basal cells may have activated tissue-structural programs.

In contrast, immune-associated pathways such as phagosome formation, neutrophil degranulation, and S100 family signaling were predicted to be inhibited in WIF1⁺ basal cells. Likewise, the top inhibited upstream regulators of WIF1⁺ basal cells included IFNG, IL1B, NFKB1, and TNF, while the top activated regulators included several anti-inflammatory cytokines, such as IL10RA, IL4R (Figure 3E; supplementary material, Table S3). Collectively, these analyses suggested an immunomodulatory role for WIF1⁺ basal cells within the transition zone microenvironment. Based on these observations, we speculated that these WIF1⁺ basal cells may have inhibited immunogenic programs.

To investigate potential intercellular communication, we applied ligand–receptor interaction analysis across all cell types in the human prostate. Notably, NRG (neuregulin) signaling was inferred to occur exclusively from WIF1⁺ basal cells to other epithelial populations (Figure 3F). A chord diagram visualized the directionality and strength of the interaction, revealing that WIF1⁺ basal cells act as a unique ligand source, while luminal, intermediate, and FOXI1⁺ basal cells serve as target populations. Among these, luminal cells exhibited the strongest inferred interaction.

Looking into the ligand–receptor pair in NRG signaling, NRG1 produced by WIF1⁺ basal cells was predicted to interact with ERBB2/ERBB3 receptors on luminal cells (Figure 3G). To validate this observation spatially, we performed CISH on transition zone tissue from the same prostatectomy specimens. Co-localization of NRG1 and WIF1 confirmed that NRG1 is expressed within WIF1⁺ basal cells (Figure 3H). Given that NRG–ERBB signaling plays a key role in epithelial morphogenesis and homeostasis [27], these findings suggest that WIF1⁺ basal cells may serve as the regulator of epithelial maintenance and differentiation within the transition zone niche.

In addition, WIF1⁺ basal cells were inferred to secrete multiple extracellular matrix components, including LAMC1, LAMC2, LAMB3, and TNC, which were predicted ligands interacting with integrin receptors expressed on both epithelial and stromal cells (supplementary material, Figure S6). These predicted interactions further support a potential role for WIF1⁺ basal cells in epithelial-stromal crosstalk and microenvironmental remodeling.

WIF1⁺ basal cells are enriched in BPH

Given the enrichment of WIF1⁺ basal cells in the transition zone and involvement in pathways such as EMT [12], androgen response [8], and stromal remodeling [10], we hypothesized that this basal subtype may be involved in the pathogenesis of benign prostatic hyperplasia (BPH).

To test this, we analyzed a previously published single-cell RNA-seq dataset [4], which included three BPH glandular samples (GSM5252126, GSM5252128, GSM5252130), and six normal prostate donor samples (GSM5252457, GSM5252459, GSM5252461, GSM5252458, GSM5252460, GSM5252462). Dimensionality reduction and clustering of this dataset revealed clear segregation by canonical cell types (Figure 4A). Upon subsetting basal epithelial cells, we identified the same four transcriptionally distinct basal subtypes described in our dataset, marked by KRT23, DDIT3, WIF1, and FOXI1, respectively (Figure 4B,C). The WIF1⁺ basal cells from this dataset exhibited a highly similar gene signature to that in our prostatectomy samples, including expression of WIF1, VCAN, NPPC, and PPP1R14C (supplementary material, Figure S7), strongly supporting the identity of this population as the same cell type.

Figure 4.

We next quantified the relative abundance of basal subtypes across BPH tissues and those from normal TZ samples from organ donors. The proportion of WIF1⁺ basal cells was significantly elevated in BPH tissues compared to organ donor TZ (Figure 4D,E).

Next, we analyzed three BPH stromal nodule samples (GSM5252127, GSM5252129, GSM5252131) from the same study [4]. Using canonical marker genes, we identified major prostate cell types within the stromal nodules. In this dataset, epithelial cells represented only a small fraction compared to stromal populations such as fibroblasts and smooth muscle cells. Notably, within the epithelial compartment, a significant fraction of basal cells expressed WIF1 (supplementary material, Figure S8).

To confirm this observation at the tissue level, we performed CISH on BPH samples obtained via transurethral resection of the prostate (TURP; n = 2 subjects, ages 66 and 71 years), alongside TZ tissues from young organ donors from JHU (n = 2, ages 21 and 25 years). Co-staining for TP63 and WIF1 confirmed the identity of WIF1⁺ basal cells, and these cells were significantly abundant in BPH specimens (Figure 4F,G; supplementary material, Figure S9).

Notably, we observed spatial variation in WIF1 expression relative to inflammatory status. Basal cells within BPH nodules lacking prominent inflammatory infiltrates exhibited more consistent and stronger WIF1 expression. In contrast, glands located in regions with marked chronic inflammation or adjacent inflammatory infiltrates contained fewer WIF1⁺ basal cells, suggesting a potential association between WIF1⁺ basal cell and inflammatory modulation (supplementary material, Figure S10).

To further validate this enrichment of WIF1⁺ basal cells in BPH, we analyzed previously published bulk RNA-seq data from BPH and normal prostate tissues [28]. Several WIF1⁺ basal cell-associated genes, including WIF1, NPPC, and VCAN, were among the most significantly upregulated genes reported in that study when comparing BPH with normal prostate (supplementary material, Figure S11). To further assess the potential contribution of age as a confounding factor, we performed differential expression analysis using DESeq2 with and without age adjustment and observed highly consistent results across both analyses (supplementary material, Figure S12). Using our scRNA-seq dataset as a reference, robust cell type decomposition (RCTD) analysis further confirmed a significant enrichment of WIF1⁺ basal cells in BPH samples relative to both normal prostate and BPH stromal nodule samples (Figure 4H; supplementary material, Figure S11).

WIF1⁺ basal-associated genes are upregulated in BPH

To determine whether WIF1⁺ basal cells undergo transcriptional changes in BPH, we compared their transcriptomes in BPH and normal TZ from organ donors using the previously published single cell sequencing dataset [4]. A large number of ribosomal genes, including RPS10, RPS29, RPL39, RPL12, were significantly upregulated in WIF1⁺ basal cells from BPH compared to those from normal TZ (Figure 5A; supplementary material, Table S4). Gene ontology (GO) analysis of the DEGs revealed strong enrichment of pathways involved in ribosomal biogenesis (Figure 5B), suggesting elevated biosynthetic activity in these cells during BPH development. Oxidative stress response pathways and cell adhesion pathways were also upregulated.

Figure 5.

To pinpoint genes associated with both cell identity and disease, we intersected DEGs from two comparisons: (1) WIF1⁺ basal versus other basal subtypes (current study), and (2) BPH versus donor TZ from WIF1⁺ basal cell (derived from Joseph et al [4]). This yielded 45 overlapping upregulated genes, that we refer to as ‘WIF1⁺ basal-related BPH up-regulated genes’ (Figure 5C; supplementary material, Table S5). Among these, the most significantly elevated genes included WIF1, VCAN, PPP1R14C, NRG1, ITGA2, PRSS23, GLIPR1, GPR87, UCHL3, and PAWR (Figure 5D).

Several of these genes are key components of the WIF1⁺ basal cell signature and were also upregulated in BPH, supporting their potential relevance in disease pathology. For example, upregulation of WIF1, a known WNT signaling inhibitor [29], and VCAN, a major extracellular matrix (ECM) proteoglycan, suggested a role in modulating epithelial-stromal signaling and extracellular matrix remodeling [30]. PPP1R14C, a regulatory inhibitor of protein phosphatase 1 may contribute to cytoskeletal or contractile regulation [31]. Notably, NRG1, which was predicted to signal from WIF1⁺ basal cells to ERBB2/ERBB3 receptors on neighboring epithelial cells, was also elevated in BPH, suggesting a potential paracrine signaling role.

We next performed CISH for VCAN and NRG1 on BPH specimens. Both genes exhibited spatial expression patterns consistent with WIF1⁺ basal cells, with clear localization to the basal layer across these tissues (Figure 5E,F), supporting their involvement within the WIF1⁺ basal cell program in BPH.

Other genes, such as PRSS23 (a serine protease that cleaves peptide bonds in proteins) [32], GPR87 (encodes a G protein-coupled receptor), found to contribute to fibrosis [33], and UCHL3 (ubiquitin C-terminal hydrolase L3), a protein that plays a crucial role in cellular processes by removing ubiquitin from other proteins [34], further suggested involvement in cell proliferation and epithelial-stromal remodeling. Notably, these genes displayed a graded expression pattern across conditions, with the highest levels observed in BPH, the lowest in the donor TZ, and intermediate levels in TZ samples from prostatectomy (Figure 5D), many of which showed evidence of nodular hyperplasia consistent with BPH (supplementary material, Figure S3).

To further validate the association of WIF1⁺ basal cells with BPH, we analyzed an independent publicly available single-cell RNA-seq dataset (GSE226237, Cao et al [35]), comprising BPH samples stratified by prostate gland volume. Reanalysis of this cohort identified WIF1⁺ basal cell as a major basal cell subpopulation across samples. Notably, expression levels of WIF1 and its associated genes were elevated in basal cells from large-volume BPH compared with small-volume BPH (supplementary material, Figure S13).

Pathway enrichment analysis of WIF1⁺ basal-related BPH up-regulated genes revealed activation of ‘epithelial-mesenchymal transition (EMT)’, along with ‘estrogen response’, ‘apical junction’, and ‘angiogenesis’ pathways (Figure 5G), indicating potential epithelial-mesenchymal transition, hormonal sensitivity, and stromal activity. GOBP terms such as cell proliferation, cell differentiation, and cell locomotion were also significantly enriched (Figure 5H). Taken together, we speculated that TZ enriched WIF1⁺ basal cells may play an active role in BPH pathogenesis by participating in remodeling of the prostate epithelium and stroma.

Discussion

Human prostate epithelium comprises four principal cell types: luminal, basal, intermediate, and a small population of neuroendocrine epithelial cells. Among these, subtypes within the basal cell compartment have been rarely characterized in the literature. In this study, we identified WIF1⁺ basal cells as a transcriptionally and anatomically distinct subpopulation compared to other basal epithelial cells. These cells display a unique gene expression profile enriched for WIF1, VCAN, PPP1R14C, and NRG1, consistent with the Basal-VCAN cell type described by Hu et al [6]. While Hu et al reported that this cell type is predominantly located in both the peripheral and transition zones [6], our scRNA-seq data and CISH staining demonstrated that WIF1⁺ basal cells are markedly enriched in the TZ of the human prostate.

Among the upregulated genes, WIF1 emerged as the most prominent marker of this population. Pathway and Gene Ontology analyses revealed enrichment of ‘androgen response’ (HALLMARK pathway), and ‘gland development and morphogenesis’ (GO pathway), consistent with a prior study showing that WIF1 is androgen-regulated and contributes to prostate gland development [36]. Functionally, WIF1 is known to act as a secreted antagonist of Wnt signaling, a pathway that plays a central role in cell proliferation and carcinogenesis [37-47]. In addition to WIF1, two other classes of secreted Wnt antagonists have been described: the secreted Frizzled-related protein (sFRP) family [48] and the Dickkopf (Dkk) family [49]. Notably, in our dataset, both SFRP1 and DKK3 were also highly expressed in WIF1⁺ basal cells (supplementary material, Table S1). The combined high expression of WIF1, SFRP1, and DKK3 in the basal layer of the TZ may establish a localized Wnt signaling gradient, limiting the ability of stromal Wnt ligands to activate epithelial receptors [50].

Furthermore, IPA analyses revealed that inflammation-related pathways were suppressed in WIF1⁺ basal cells relative to other basal populations. Together, these findings lead us to speculate that WIF1⁺ basal cells in the TZ help create an epithelial microenvironment characterized by attenuated WNT activation and reduced inflammation, potentially preserving epithelial homeostasis and contributing to the relatively low incidence of prostate cancer initiation in the TZ. This hypothesis warrants further mechanistic investigation.

BPH is a highly prevalent age-associated condition characterized by epithelial and stromal hyperplasia, predominantly in the TZ of the prostate [51]. Analysis of publicly available datasets, combined with CISH staining of BPH tissues from TURP, revealed an enrichment of WIF1⁺ basal cells in BPH. Furthermore, the gene signature of WIF1⁺ basal cells was distinctly altered in BPH compared with donor TZ, suggesting that this TZ-enriched population may contribute to BPH pathogenesis.

Among the top differentially expressed genes in WIF1⁺ basal cells, VCAN, encoding the extracellular matrix (ECM) component versican, points to active involvement in ECM remodeling. Its expression in WIF1⁺ basal cells was also confirmed by CISH. Versican has been shown to promote fibroblast activation and myofibroblast-like phenotypes [52, 53], potentially driving the fibromuscular expansion characteristic of BPH. Pathway enrichment analysis identified EMT as the most significant hallmark pathway in WIF1⁺ basal cells. EMT, a process by which epithelial cells acquire mesenchymal features, underlies tissue plasticity, regeneration, and fibrosis [54]. Its activation in WIF1⁺ basal cells aligns with previous evidence implicating EMT in BPH etiology [12].

NRG1, another gene differentially expressed in WIF1⁺ basal cells and enriched in BPH, is known to activate ERBB2/3 receptors on neighboring epithelial cells, thereby stimulating proliferation and survival [55]. Its role in promoting epithelial regeneration has been demonstrated in the intestinal stem cell niche [56]. Our cell–cell communication analysis inferred that NRG signaling occurs exclusively between WIF1⁺ basal cells and other epithelial cells, particularly luminal cells. CISH further confirmed NRG1 expression within WIF1⁺ basal cells in both prostatectomy and BPH TURP specimens. We propose that WIF1⁺ basal cells may secrete NRG1 to regulate adjacent luminal epithelium, establishing a paracrine loop that promotes epithelial development in BPH. Interestingly, previous studies have reported fibroblast secretion of NRG1, which then interacts with EGFR and ERBB3 receptors on epithelial cells [4, 6].

Interestingly, our findings suggest that WIF1⁺ basal cells may exhibit context-dependent functions within the TZ microenvironment. These cells express multiple Wnt antagonists, including WIF1, SFRP1, and DKK3, and display relative suppression of inflammatory pathways, supporting a potential role in epithelial homeostasis. Conversely, WIF1⁺ basal cells also express factors associated with tissue remodeling and proliferation, including NRG1 and VCAN, and are enriched for EMT, extracellular matrix organization, angiogenesis, and proliferative pathways in BPH. CISH further confirmed co-expression of WIF1, NRG1, and VCAN within the same basal cell population. Together, these findings suggest that WIF1⁺ basal cells may coordinate both homeostatic and remodeling programs depending on physiological or disease context. Similar context-dependent behavior has been described for androgen receptor signaling in the prostate [57]. Further functional studies will be required to define the relative contribution and temporal dynamics of these pathways in BPH pathogenesis.

Our scRNA-seq data further demonstrated that Wif1 is not expressed by basal cells in the mouse prostate but instead marks a subset of fibroblasts, termed Rorb⁺ subglandular fibroblasts [26]. This raises the possibility that WIF1⁺ basal cells exhibit mesenchymal-like characteristics. In a previous study, Alonso-Magdalena et al concluded that BPH stroma can arise from epithelium through EMT, a process in which epithelial cells lose their polarity and basement membrane attachment while acquiring mesenchymal features [12]. Whether WIF1⁺ basal cells enriched in TZ and BPH represent an intermediate state between basal epithelial and mesenchymal cells warrants further investigation.

The identification of WIF1⁺ basal cells may also have implications for zone-specific susceptibility. We observed that WIF1⁺ basal cells were consistently enriched in the TZ across both organ donor samples and BPH TURP specimens, despite differences in patient age between these cohorts. To further address potential age-related confounding, we reanalyzed the bulk RNA-seq dataset from Middleton et al using DESeq2 with and without age adjustment and observed highly concordant results between both analyses. Although we cannot fully exclude the contribution of age-related effects, and it remains possible that the functional activity of WIF1⁺ basal cells may increase with aging, their preferential localization to the TZ appears to be a robust feature.

Functionally, WIF1⁺ basal cells in the TZ may contribute to a microenvironment that supports EMT, cellular proliferation, extracellular matrix organization, and angiogenesis. In contrast, the PZ and CZ may lack this WIF1⁺ basal cell-mediated niche. Together, these findings suggest that the enrichment of WIF1⁺ basal cells within the TZ may represent a key cellular determinant underlying the preferential development of BPH within this anatomical region.

A key limitation of our study is the absence of direct functional experiments to validate the inferred roles of WIF1⁺ basal cells in BPH pathogenesis. Although our single-cell transcriptomic analysis, pathway enrichment results, and histological validation collectively support their involvement in EMT activation, extracellular matrix remodeling, cell proliferation, and modulation of WNT signaling, these conclusions remain correlative. Without functional assays, such as lineage tracing, in vitro co-culture systems, or in vivo animal models, it remains uncertain whether WIF1⁺ basal cells are causal drivers or secondary responders in the observed tissue remodeling. Future studies integrating targeted manipulation of WIF1 expression or depletion of this basal cell population will be essential to establish causality and delineate their precise mechanistic contributions to BPH development. Spatial transcriptomic approaches will also be crucial for mapping their interactions with stromal, immune, and vascular components in situ.

Author contributions statement

RW, AMD and SY were responsible for conceptualization. RW, QZ, AMD and SY designed and performed the methodology. QZ, AMD and SY performed and analyzed validation experiments. RW, AMD and SY performed data analysis. WGN, AMD and SY provided resources and acquired funding; RW, AG, YZ, KS, JM, AS, DH and SY carried out data curation. RW and SY wrote the original manuscript draft; AMD and SY supervised the study. All authors contributed to editing and approval of the manuscript.

Data availability statement

We additionally processed scRNA-seq data of human normal and BPH tissues by Joseph DB et al [4]. Three datasets of BPH glandular samples (GSM5252126, GSM5252128, GSM5252130), BPH stromal samples (GSM5252127, GSM5252129, GSM5252131), and six datasets of young organ donor samples (GSM5252457, GSM5252459, GSM5252461, GSM5252458, GSM5252460, GSM5252462) were obtained from the GEO database under accession code GSE172357 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE172357). We processed the data using a similar strategy described above to filter low-quality cells, integrate and cluster cells. The cell type was then identified, and differential gene expression analysis was performed among cell types.

We additionally analyzed an independent publicly available single cell RNA-seq dataset of BPH samples stratified by prostate gland volume from Cao et al [35]. Three large BPH samples (GSM7068696, GSM7068697, GSM7068698), and three small BPH samples (GSM7068699, GSM7068700, GSM7068701) were obtained from the GEO database under accession code GSE226237 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE226237). We processed the data using a similar strategy described above to filter low-quality cells, integrate and cluster cells. Cell types were subsequently annotated, followed by differential gene expression analysis across cell types and between the large and small BPH groups.

Bulk RNA-seq dataset of BPH, BPH stromal nodules, and normal prostate from Middleton et al [28] were obtained from the dbGAP database (https://dbgap.ncbi.nlm.nih.gov/home/), with accession number phs001698.v1.p1. Controlled-access data were downloaded following approved dbGaP data access authorization and reanalyzed in this study. DESeq2 with and without age adjustment were applied to identify DEGs between BPH and normal prostate. RCTD [58] was applied to deconvolute cell types in the bulk RNA-seq data using our human prostatectomy single cell RNA-seq dataset as reference.

The mouse prostate data were from Graham MK et al [26], and were deposited in the NCBI GEO database under accession code GSE228945 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE228945).

The human prostatectomy data generated in this study have been deposited in the NCBI dbGaP database under accession code phs003480.v2.p1 (https://dbgap.ncbi.nlm.nih.gov/beta/search/?OBJ=study&TERM=phs003480).