Loss of PI5P4Kα Slows the Progression of a Pten Mutant Basal Cell Model of Prostate Cancer

Abstract

Although early prostate cancer depends on the androgen receptor signaling pathway, which is predominant in luminal cells, there is much to be understood about the contribution of epithelial basal cells in cancer progression. Herein, we observe cell type-specific differences in the importance of the metabolic enzyme phosphatidylinositol 5-phosphate 4-kinase alpha (PI5P4Kα; gene name PIP4K2A) in the prostate epithelium. We report the development of a basal cell-specific genetically engineered mouse model targeting Pip4k2a alone or in combination with the tumor suppressor phosphatase and tensin homolog (Pten). PI5P4Kα is enriched in basal cells, and no major histopathologic changes were detectable following gene deletion. Notably, the combined loss of Pip4k2a slowed the development of Pten mutant mouse prostatic intraepithelial neoplasia. Through the inclusion of a lineage tracing reporter, we utilize single-cell RNA sequencing to evaluate changes resulting from in vivo downregulation of Pip4k2a and characterize cell populations influenced in the established Probasin-Cre- and cytokeratin 5-Cre-driven genetically engineered mouse model. Transcriptomic pathway analysis points toward the disruption of lipid metabolism as a mechanism for reduced tumor progression. This was functionally supported by shifts of carnitine lipids in LNCaP prostate cancer cells treated with siPIP4K2A. Overall, these data nominate PI5P4Kα as a target for PTEN mutant prostate cancer. Implications: PI5P4Kα is enriched in prostate basal cells, and its targeted loss slows the progression of a model of advanced prostate cancer.

Figure 1.

Cell-type expression patterns of PI5P4Kα. A, A human adult prostate scRNA-seq dataset is annotated for basal epithelium (BE), luminal epithelium (LE), club, endothelia (Endo), fibroblasts (Fib), hillock, leukocytes (Leu), smooth muscle (SM), and neuroendocrine (NE) populations (bioRxiv 2021.04.05.438318; ref. ; B) populations are subset for epithelial clusters (BE, LE, club, and hillock) and 424 epithelial cells are positive detection of PIP4K2A transcript. Adult C57BL/6 mouse prostates stained for PI5P4Kα with IHC and found to express in three patterns: (C) focal and basal oriented , diffuse and punctate, and (D) basal epithelial layer (zoomed scale bars = 10 μm). E, Graphical summary of PI5P4Kα luminal cell patterns based on mouse prostate lobe localization. AP, DLP, and VP are included relative to the urethra (U). PI5P4Kα stain patterns are numerically summarized based on the prostate lobe for (F) luminal patterns and as (G) the number of positive basal cells per gland. t test values: n.s., not significant (P > 0.05); **, P < 0.01.

Figure 2.

PI5P4Kα is enriched in prostate basal cells. A, Graphical summary of basal cell lineage–marked phenotype of tamoxifen-inducible Krt5-Cre^ERT2; R26eYFP (CK5-eYFP) GEMM. Fluorescent eYFP protein is expressed in cells where Cre is activated. B, Z-stacked IF confirmation of eYFP activation in CK5⁺ basal prostate cells. C, 3D organoid cultures generated from the prostate AP lobe 2 months after eYFP activation in the CK5-eYFP model. D, PI5P4Kα and CK5 protein expression in mouse organoid cultures from two biological replicates enhanced following FACS of eYFP-labeled cells. Protein is quantified from three independent experiments relative to housekeeping controls (2.02-fold increase, t test, P value = 2.9E−03). t test values: n.s., not significant (P > 0.05); **, P < 0.01.

Figure 3.

scRNA-seq of lineage marked GEMMs. Mouse cells selected from digested in vivo prostate tissue were analyzed using the SORT-seq RNA-seq platform. A, Graphical summary of a PB-based luminal model and CK5-based basal model (CK5-eYFP; also see Fig. 2). Cre activation occurs by 8 weeks of age, and eYFP⁺ cells were collected with FACS 2 weeks later for both models. B, UMAP clustering of major cell types identified from GEMMs. B1, B2, and B3 are clusters from the CK5-eYFP basal model. L1–L6 are clusters with origin from the luminal model. C, Clusters have varying levels of canonical basal markers (Krt5 and Trp63) and luminal markers (Pbsn and Krt8) that have expression associated with expected lineage groups. D, Module scores from marker genes from Crowley and colleagues (see Supplementary Table S3 gene list) mouse single-cell atlas in comparison with targeted scRNA-seq clusters from this study. E, Summary of predicted cell types that are identified from lineage tracked basal and luminal GEMMs. UMAP, Uniform Manifold Approximation and Projection for Dimension Reduction; LumV, VP luminal; LumP, luminal proximal.

Figure 4.

GEMM targeting Pip4k2a in CK5⁺ basal population. A, Graphical summary of Krt5-Cre^ERT2; R26eYFP; Pip4k2a^fl/fl (CK5-Pip) GEMM. Homozygote genetic deletion of Pip4k2a and expression of fluorescent eYFP protein occurs where Cre is activated. Multiple timepoints and tissues were evaluated up to 1-year timepoint. No significant differences were measured in (B) animal body weight or (C) prostate lobe cellularity (anterior lobe represented) compared with CK5-eYFP controls. D, Representative IF-IHC detection of CK5 (red) and eYFP Cre (green) reporter with DAPI (blue) nuclear stain in 8-month prostate of CK5-Pip. Zoom regions represent quantified examples of CK5+ basal cells without (gray box) and with (yellow arrow) eYFP expression. Additionally, eYFP⁺ luminal cell morphology (lower right, yellow box; scale bar = 20 μm). Multi timepoint cohorts of CK5-eYFP and CK5-Pip prostate IF-IHC stain quantification for (E) CK5+ cells (P value = n.s), (F) percent eYFP-positive cells (6-month, 2.85-fold increase, t test, P value = 0.05), and (G) eYFP-positive with luminal morphology (P value = n.s). t test values: n.s., not significant (P > 0.05); *, P < 0.05. TAM, tamoxifen.

Figure 5.

Loss of Pip4k2a slows Pten mutant tumorigenesis. A, Graphical summary of Krt5-Cre^ERT2; R26eYFP; Pten^fl/fl (CK5-Pten) and Krt5-Cre^ERT2; R26eYFP; Pten^fl/fl; Pip4k2a^fl/fl (CK5-PipPten) GEMMs. Homozygote genetic deletion of floxed genes and expression of fluorescent eYFP protein occurs where Cre is activated. Timepoints and tissues were evaluated up to 3-month timepoint. B, Representative images of mouse prostate lobes and CK5-Pten mouse prostate sampled with lobes containing dense tumor tissue. C, AP lobes were quantified for the area of mPIN at 4 and 12 weeks following Cre activation. CK5-PtenPip cases have reduced mPIN annotated tissue compared with CK5-Pten (4 weeks, −1-fold, P value = 0.02; 12 weeks, 0.35-fold, P value = 0.01). Representative H&E and CK5 IHC images of (D) CK5-Pten (scale bar = 20 and 50 μm) and (E) CK5-PipPten (scale bar = 20 and 100 μm) AP lobes. Representative H&E and CK5 IHC images of (F) CK5-Pten DLP lobe and (G) CK5-PipPten DLP (left) alongside tumor-free ventral lobes (VP; right). H, DLP lobes were quantified for the area of mPIN at 4 weeks (0.67-fold, P value = 0.67) and 12 weeks (1.0-fold, P value = 0.97) following Cre activation. t test values: n.s., not significant (P > 0.05); *, P < 0.05. TAM, tamoxifen.

Figure 6.

Targeting Pip4k2a triggers a shift in metabolic pathways. The differences in transcript expression of CK5-Pip eYFP⁺ prostate cells were compared with CK5-eYFP controls sampled in vivo at a 2-month timepoint using SORT-seq scRNA sequencing. A, GSEA of most altered Hallmark pathways are shown with P value < 0.01 cut-off. The LNCaP human prostate cancer cell line was treated in vitro with siRNA targeting PIP4K2A or a NT control in conditions of hormone-stripped medium and evaluated using targeted metabolomics. B, Key metabolites linked to lipid regulation are highlighted, and the extended list is shown in Supplementary Fig. S8 and Supplementary Table S5. C, Most noteworthy were changes to carnitine metabolites associated with lipid β-oxidation. Enrichment of carnitines with lipid moieties in medium and long (C6−C18) acyl chain length observed following siPIP4K2A knockdown and depletion of C1 and C3 short acylcarnitines. t test values: n.s., not significant (P > 0.05); *, P < 0.05. GSEA, gene set expression analysis.