RNA-Seq of single prostate CTCs implicates noncanonical Wnt signaling in antiandrogen resistance

Abstract

Prostate cancer is initially responsive to androgen deprivation, but the effectiveness of androgen receptor (AR) inhibitors in recurrent disease is variable. Biopsy of bone metastases is challenging; hence, sampling circulating tumor cells (CTCs) may reveal drug-resistance mechanisms. We established single-cell RNA-sequencing (RNA-Seq) profiles of 77 intact CTCs isolated from 13 patients (mean six CTCs per patient), by using microfluidic enrichment. Single CTCs from each individual display considerable heterogeneity, including expression of AR gene mutations and splicing variants. Retrospective analysis of CTCs from patients progressing under treatment with an AR inhibitor, compared with untreated cases, indicates activation of noncanonical Wnt signaling (P = 0.0064). Ectopic expression of Wnt5a in prostate cancer cells attenuates the antiproliferative effect of AR inhibition, whereas its suppression in drug-resistant cells restores partial sensitivity, a correlation also evident in an established mouse model. Thus, single-cell analysis of prostate CTCs reveals heterogeneity in signaling pathways that could contribute to treatment failure.

Fig. 1

Single cell RNA-sequencing of prostate CTCs. (A) Heat map of unsupervised hierarchical clustering analysis of RNA-seq data from 77 single lineage-confirmed prostate CTCs, 12 primary tumor samples, and 30 single cells from four prostate cancer cell lines. (B) Heterogeneity, measured by mean correlation coefficient within individual samples with 3 or more cells available for analysis. (C) Heterogeneity analysis showing mean correlation coefficients from expression data for CTCs between and within patients (0.0013838 vs. 0.10055, Holm corrected P=2.0×10⁻¹¹), and for prostate cancer cell lines between and within lines (0.11568 vs. 0.43534, Holm corrected P=5.42×10⁻¹⁴).

Fig. 2

Gene signatures and signaling pathways in prostate CTCs. (A) High resolution heat map showing expression of selected gene panels in single prostate CTCs, primary tumor samples, and prostate cancer cell lines. GS, Gleason score; CSPC, castration-sensitive prostate cancer; CRPC, castration-resistant prostate cancer; LNCaP.R, LNCaP cells treated with R1881; LNCaP.D, LNCaP cells treated with dimethyl sulfoxide (DMSO) as a vehicle control. (B) Volcano plot showing genes differentially expressed between prostate CTCs and primary prostate tumors (FDR<0.1 and fold-change >2). (C) Venn diagram showing Pathway Interaction Database (PID) molecular pathways (14) enriched in CTCs compared to primary tumors and in metastases compared to primary tumors (based on analysis of multiple data sets; see Fig. S5 and Table S4). (D) Signaling pathways enriched in prostate CTCs. Molecular pathways from the Pathway Interaction Database (PID) upregulated in CTCs versus primary tumors (excluding those enriched in metastases compared to primary tumors), organized by PID categorization (14) (Fig. S5).

Fig. 3

Heterogeneity of treatment resistance mechanisms in prostate CTCs. (A) Heat map depicting androgen receptor (AR) abnormalities, selected signaling pathway signatures, and genes in radical prostatectomy specimens, prostate CTCs from enzalutamide-naïve patients (Group A), and prostate CTCs from patients who had radiographic or biochemical progression of disease while receiving treatment with enzalutamide (Group B). Non-canonical Wnt signature is from reference (14), glucocorticoid receptor (GR) signature is from reference (28), and AR signature is from reference (30) (Table S7). Numbers at top of heatmap represent ID numbers (Pr numbers) for patients from which each CTC is derived. (B) Top, Gene Set Enrichment Analysis (GSEA) plots showing enrichment of non-canonical (nc) Wnt pathway in CTCs from Group B (patients with cancer progression on enzalutamide) compared to Group A (enzalutamide-naïve patients). Bottom, GSEA plots showing enrichment of ncWnt pathway in CTCs from Group B compared to Group A, stratified by GR gene expression. (C) Left: Representative micrograph (40X) of RNA-in situ hybridization assay in metastatic prostate tumors, probing for WNT5A and KRT8/18, scale bar = 50 μm. Inset, high magnification, arrow points to WNT5A signal (red dot), arrowhead points to KRT8/18 signal (blue dot), scale bar = 10 μm. Adjacent tissue sections were probed for WNT7B, and quantification of RNA-ISH data are displayed in the table. Of 9 primary tumors examined, 5 had >1% WNT5A expression in KRT+ cells (range 0.3%–42%) and 7 had >1% WNT7B expression (range 0.5%–33.6%). Of 24 metastatic tumors examined, 16 had >1% WNT5A expression (range 0%–50.5%) and 15 had >1% WNT7B expression (range 0%–26%). Right: Representative fluorescence micrographs of RNA-in situ hybridization in prostate CTCs, probing for WNT5A/7B (yellow dots), and prostate CTC-specific markers (EPCAM, KLK3, FOLH1, KRT8/18/19) (red dots). DNA is stained with DAPI (blue). Scale bar = 10 μm.

Fig. 4

ncWnt signaling and enzalutamide resistance. (A) Ectopic expression of ncWnt in LNCaP cells increases cell survival in the presence of enzalutamide (3 μM). (B) Enzalutamide treatment induces WNT5A mRNA expression in LNCaP cells, and WNT5A suppression in enzalutamide-treated LNCaP cells results in decreased proliferation. (C) Stable enzalutamide-resistant cells (LN_EnzR) derived from LNCaP cells express increased levels of WNT5A, and WNT5A siRNA results in reduced proliferation in the presence of enzalutamide. Each experiment (A–C) was performed ≥3 times. Data are presented as mean +/− SD. (D) GSEA plot showing enrichment of ncWnt pathway in mouse xenografts derived from enzalutamide-resistant LREX cells (28) compared to control LNCaP cells (data from GEO GSE52169). Lower GSEA plots show ncWnt pathway enrichment in enzalutamide-resistant xenografts, when stratified by GR gene expression.