Construct a single-cell transcriptomic landscape of normal mouse bladder

We gathered three single-cell transcriptome datasets from literature and public databases, focusing on the normal mouse bladder. Following rigorous quality control measures and the elimination of duplicate cells, a comprehensive set of 24,930 cells was assembled (Fig. 1A). These datasets comprised 15,118, 4845, and 4967 cells, respectively. By employing the Harmony algorithm, we effectively mitigated batch effects and executed the initial dimensionality reduction clustering, culminating in the identification of 14 distinct clusters (Fig. 1B). According to the markers used in the literature, we accurately classified these clusters into epithelial (22,335 cells, Epcam, Krt19), stromal (2211 cells, Col1a2, Acta2), and immune (384 cells, Ptprc) populations, each demonstrating unique transcriptional signatures [40, 41] (Fig. 1C, D).

Fig. 1

Construction of single-cell transcriptome profiles of the normal mouse bladder. A UMAP plot of all cells grouped by original datasets. B UMAP plot of all cells grouped by seurat_clusters. C UMAP plot of all cells grouped by major cell types. D Feature plot of markers used to annotate major cell types. E Heatmap of DEGs of every cell types and GO enriched pathways. DEGs differentially expressed genes, GO gene ontology

Subsequent gene ontology (GO) enrichment analyses unveiled profound insights into the functional characteristics of these cell types (Fig. 1E). Specifically, epithelial cells were enriched in pathways associated with cytoplasmic transport, while stromal cells displayed enrichment in pathways pertaining to extracellular matrix organization and extracellular structure organization. Immune cells, in contrast, showcased enrichment in pathways linked to lymphocyte-mediated immune response and adaptive immune response, consistent with their respective cellular functions and previous studies [30]. These findings not only confirm the accuracy of our cell type annotation but also provide valuable insights into the functional diversity within the normal mouse bladder transcriptome.

Cluster epithelial cells into multiple subtypes

Following the isolation of epithelial cells, we further categorized them into distinct subtypes based on traditional markers, namely basal cells (characterized by Krt5, Trp63), umbrella cells (marked by Upk2, Upk3a), and intermediate cells occupying transitional states [29, 30] (Fig. 2A). This segmentation resulted in 13 discrete epithelial subtypes, comprising 6 basal cell subtypes, 4 umbrella cell subtypes, and 3 intermediate cell subtypes, each exhibiting a distinct transcriptional landscape (Fig. 2B).

Fig. 2

Cluster epithelial cells into multiple subtypes. A UMAP plot of epithelial cell grouped by subtypes. B Dot plot of markers used to annotate subtypes. C Heatmap of DEGs of subtypes and GO enriched pathways. D Feature plot of top markers expressed by subtypes. DEGs differentially expressed genes, GO gene ontology

GO enrichment analysis revealed functional differences among these subtypes (Fig. 2C). Notably, Basal_I, Basal_III, and Basal_IV were enriched in pathways related to RNA splicing, mRNA processing, and DNA replication, highlighting their involvement in fundamental cellular processes. Conversely, Basal_II and Basal_V exhibited enrichment in cytoplasmic transport pathways, while Basal_VI featured marker genes associated with cellular respiration. Umbrella cells emerged as pivotal players in oxidative phosphorylation and cell–cell junction organization, underscoring their role in bladder epithelial function.

Of particular interest, the Basal_I subtype exhibited heightened expression of the Chka gene, responsible for encoding choline kinase α—an enzyme pivotal in catalyzing the initial step of phosphatidylcholine synthesis, critical for membrane biogenesis [42] (Fig. 2D). Meanwhile, the Basal_III subtype showcased elevated expression of the Lig1 gene, encoding DNA ligase I, which is crucial for DNA repair and replication [43]. The Basal_IV subtype displayed enhanced expression of the Cenpf gene, responsible for encoding centromere protein F, vital for cell mitosis and centromere structure maintenance [44]. Collectively, these findings suggest the proliferative and stem cell-like properties inherent in the Basal_I, Basal_III, and Basal_IV subpopulations, shedding light on their potential roles in bladder epithelial homeostasis and regeneration.

Pseudotemporal and SCENIC analysis identified Basal_I as transitional basal cell

Our pseudo-temporal analysis of all epithelial cell subtypes using Monocle3 revealed a developmental trajectory from basal cells to intermediate cells and ultimately to umbrella cells, consistent with previous studies (Fig. 3A, B). Specifically, Basal_IV subtype occupies the initial stage of this developmental trajectory, which subsequently transitions through Basal_I, Basal_II, basal_III, and Basal_V subtypes, followed by three states of intermediate cells, before finally differentiates into four umbrella cell subtypes. Thus, Basal_I serves as an intermediate transition state in the differentiation of basal cells into intermediate cells and umbrella cells. The 3D pseudo-temporal analysis corroborated these findings (Fig. 3C). In terms of cell density, Basal_IV exhibited the highest density at the beginning, and the density of Basal_I, Basal_II, Basal_III and Basal_V subtypes gradually increased and then decreased. The intermediate cells exhibited a similar pattern, while the umbrella cells showed a gradual increase in density toward the end (Fig. 3D).

Fig. 3

Pseudotemporal and SCENIC analysis identified Basal_I as transitional basal cell. A UMAP plot of epithelial cell differentiation colored by subtypes. B UMAP plot of epithelial cell differentiation colored by pseudotime. C 3D plot of epithelial cell differentiation colored by subtypes. D Density plot of epithelial cell along with pseudotime. E Heatmap of expression of gene profiles along with pseudotime. F Heatmap of expression of TFs along with pseudotime. G GO enrichment of cluster 2 (left) and 3 (right) in F. H Feature plot of six TFs expressed by subtypes. I Expression of six TFs along with pseudotime. J Heatmap of SCENIC outcomes. DEGs differentially expressed genes, GO gene ontology, TFs transcription factors

We then extracted the genes contributing to the epithelial pseudotime series analysis for cluster analysis, resulting in four distinct gene expression clusters (Fig. 3E). The expression of genes in cluster 1 gradually increased with pseudotime, the expression of genes in cluster 2 gradually decreased with pseudotime, and the expression of genes in cluster 3 first increased with pseudotime and then gradually decreased. The genes in cluster 4 were initially highly expressed, then decreased, before gradually returning to baseline levels.

Further analysis of transcription factors (TFs) revealed three distinct clusters (Fig. 3F). TFs in cluster 1 were highly expressed at the beginning of the pseudotime, then decreased, and finally increased. TFs in cluster 2 were gradually increased and TFs in cluster 3 were gradually decreased at pseudo time. Performing GO enrichment analysis of the three clusters of transcription factors, we found that cluster 3, whose expression gradually decreased with pseudotime, was mainly related to pathways related to mRNA metabolism regulation, regulation of DNA-binding transcription factor activity, miRNA transcription, chromatin remodeling and maintenance of cell population (Fig. 3G). In contrast, cluster 2, whose expression gradually increased over pseudo time, was mainly associated with pathways related to muscle cell differentiation and gland development. Notably, six transcription factors in cluster 3—Gata6, Nr3c1, Klf9, Egr1, Jun and Nfix—were highly expressed in Basal_I, with their expression gradually declining as pseudotime progressed (Fig. 3H, I). SCENIC analysis of epithelial cell subsets was also performed, which revealed higher levels of activity of Gata6, Klf9, and Nr3c1 in Basal_I compared to other subtypes (Fig. 3J). In conclusion, Basal_I represents an intermediate transition state in the differentiation of basal cells into intermediate cells and umbrella cells.

Basal_I subtype dominated the interaction with other subtypes

In our comprehensive analysis, cell communication assessment using CellChat across all epithelial cell subsets highlighted Basal_I as exhibiting the highest number and intensity of intercellular interactions among all subtypes (Fig. 4A, B). Basal_I demonstrated intricate communication with other epithelial cell subsets through multiple ligand-receptor pairs, including Wnt5a-Fzd6, THBS1-SDC4/Sdc1/Cd47, Lamb3-Dag1/Cd44, Jag2-Notch1, F11-F11r, and Agrn-Dag1 (Fig. 4C). In various malignancies such as melanoma and gastric cancer, upregulation of WNT5a has been associated with heightened cancer cell migration and metastasis, mediated by the activation of protein kinase C (PKC) and WNT/calcium pathways. Similarly, Fzd6 activation has been linked to WNT/calcium pathways and PKC activation, frequently observed in glioblastomas, suggesting a potential synergy in driving tumor migration and invasion [45]. THBS1, a multifunctional glycoprotein, regulates extracellular matrix structure, cell–cell interactions, and physiological processes including vascular regulation, vasoconstriction, and tissue repair. Furthermore, THBS1 plays a significant role in immune regulation and tumor development [46]. Notably, elevated THBS1 expression has been associated with advanced stages of bladder cancer and poorer prognosis (Supplementary Fig. 1). LAMB3 has also been implicated in the invasiveness and metastatic potential of various cancers, including colon, pancreatic, lung, cervical, gastric, and prostate cancers, underscoring its significance in cancer progression [47]. Interestingly, interactions involving Basal_I predominated in pathways associated with THBS, LAMININ, AGRN, and non-canonical WNT signaling (Fig. 4D), emphasizing the potential impact of Basal_I-mediated communication on tumor microenvironment dynamics and disease progression.

Fig. 4

Basal_I subtype dominated the interaction with other subtypes. A The number of cell communications between epithelial cell subpopulations. B The strength of cell communications between epithelial cell subpopulations. C Dotplot of ligand-receptor between epithelial cell subpopulations. D Chord plot of enriched pathways between epithelial cell subpopulations

Tumor cells showed higher expression of specific transcription factors of Basal_I

The Monocle3 analysis revealed heightened expression of six transcription factors—Gata6, Nr3c1, Klf9, Egr1, Jun, and Nfix—in Basal_I. Statistical analysis confirmed that the expression levels of these transcription factors in Basal_I were significantly higher compared to other epithelial cell subtypes (Fig. 5A, B).

Fig. 5

Tumor cells showed higher expression of specific transcription factors of Basal_I. A UMAP plot of normal mouse bladder cells grouped by basal_I and others. B Box plot of expression of six TFs grouped by basal_I and others. C UMAP plot of human bladder cancer and adjacent normal cells grouped by major cell types. D Box plot of expression of six TFs grouped by human bladder cancer and adjacent normal cells. E UMAP plot of mouse bladder cancer cells grouped by major cell types. F Box plot of expression of six TFs grouped by mouse bladder cancer and normal cells

To further investigate, we analyzed mouse and human bladder cancer single-cell transcriptome data to discern differences in the expression of these transcription factors between bladder cancer cells and normal epithelial cells. In the human bladder cancer single-cell dataset, the expression levels of Gata6, Nr3c1, Klf9, Egr1, Jun, and Nfix were notably elevated in tumor cells compared to normal epithelial cells (Fig. 5C, D). Similarly, in the mouse bladder cancer single-cell dataset, tumor cells exhibited significantly higher expression levels of Egr1, Jun, and Nr3c1 compared to their normal epithelial counterparts (Fig. 5D, E). These findings suggest that these transcription factors are not only highly expressed in the transitional state of basal cells but also exhibit further elevation in expression levels in tumor cells. This underscores their potential significance in both normal epithelial cell differentiation processes and the pathogenesis of bladder cancer.

Subsequently, we conducted immunohistochemical staining on samples from both mouse bladder carcinoma in situ and human bladder cancer, juxtaposed with their respective normal bladder counterparts. The results revealed elevated Immunoreactive Assessment (IRA) scores for the aforementioned transcription factors in bladder cancer samples compared to normal bladder tissues, across both human and mouse specimens (Fig. 6). This consistent elevation in IRA scores further substantiates the role of these transcription factors in bladder cancer pathogenesis and progression.

Fig. 6

IHC staining of key TFs in bladder cancer and normal samples. A IHC staining of EGR1 in human bladder cancer and normal samples. B IHC staining of EGR1 in mouse bladder cancer and normal samples. C IHC staining of GATA6 in human bladder cancer and normal samples. D IHC staining of GATA6 in mouse bladder cancer and normal samples. E IHC staining of JUN in human bladder cancer and normal samples. F IHC staining of JUN in mouse bladder cancer and normal samples. G IHC staining of NR3C1 in human bladder cancer and normal samples. H IHC staining of NR3C1 in mouse bladder cancer and normal samples. I IHC staining of NFIX in human bladder cancer and normal samples. J IHC staining of NFIX in mouse bladder cancer and normal samples

Key transcription factors are associated with the stage and prognosis of bladder cancer

In our exploration, we delved into the TCGA database to probe the relationship between the expression levels of the aforementioned six transcription factors and the stage and prognosis of bladder cancer. With the exception of NFIX, the expression levels of the remaining five transcription factors demonstrated a significant increase in patients with stage III and IV compared to those with stage II (Fig. 7A–F, left). Furthermore, patients exhibiting heightened expression of these transcription factors experienced shorter survival durations (Fig. 7A–F, right). These findings underscore a direct correlation between the expression of these transcription factors and the progression of bladder cancer, with higher expression levels associated with advanced disease stages and poorer prognosis.

Fig. 7

Key transcription factors are associated with the stage and prognosis of bladder cancer. A Box plot of expression of JUN with stages and survival curve of patients with differentially expressed JUN. B Box plot of expression of KLF9 with stages and survival curve of patients with differentially expressed KLF9. C Box plot of expression of GATA6 with stages and survival curve of patients with differentially expressed GATA6. D Box plot of expression of EGR1 with stages and survival curve of patients with differentially expressed EGR1. E Box plot of expression of NR3C1 with stages and survival curve of patients with differentially expressed NR3C1. F Box plot of expression of NFIX with stages and survival curve of patients with differentially expressed NFIX