1. Supplementary Table 1 – Details of clinical prostate samples used for miRNA microarray analysis. TURP: Transurethral resection of prostate, LRP: laparoscopic resection of prostate, ORP: open resection of prostate, and chTURP: channel TURP. N/A: Not Applicable.

2. High in SCunchangedHigh in CB BPH50817310 PCa5245306 CRPC48740308 Common to all 429-239 Supplementary Table 2 – Shared miRNA expression in BPH, PCa, and CRPC fractionated populations. Values indicate the total number of miRNAs in each category (Total miRNAs: 835). Note the generally higher number of miRNAs overexpressed in SCs.

3. Supplementary Table 3 – Composite miRNA signatures of PCa relative to BPH (top panel) and CRPC relative to PCa (bottom panel)

4. Supplementary Fig. 1 – Validation of the microarray expression pattern with qRT-PCR, performed on the same samples used for microarray profiling (n = 5 BPH and 5 PCa, each sample in triplicate).

5. Integration of miRNA-mRNA microarray data derived from patient-derived prostate samples revealed that 60 potential miR-548c-3p target genes are repressed in SCs even at RNA level (manuscript in preparation). Literature analyses showed that higher expression of miR-548c-3p is associated with poor survival (1) Analysis of our published work (2) revealed that stimulation/overexpression of several transcription factors (e.g. RXR, VDR, GR, TAZ, SRF, HSF1) can promote prostate stem cell differentiation. Bioinformatic analysis revealed that miR-548c can inhibit expression of all these transcription factors. We hypothesised that higher expression of miR-548c-3p should be necessary for SC maintenance in prostate epithelium Supplementary Fig. 2 – Rationale for selecting miR-548c-3p

6. Gene expression changes, as assessed by qRT-PCR, of differentiation-associated genes after miR-548c-3p transfection into CB cells for 3 days, relative to control transfection. (n = 2 BPH and 3 PCa, each sample in triplicate). Data are expressed as mean ± s.d. *P < 0.05, **P < 0.01 (Student's t-test). Expression of miR-548c-3p in prostate epithelial sub-populations. (Newly cultured patient samples distinct from those used in Fig 1 or Figure S1a. n = 5 BPH and 5 PCa, each sample in triplicate). Data are expressed as mean ± s.d. *P < 0.05 (Student's t-test). Expression of miR-548c-3p CB cells 3 days after transfection with miR-548c-3p. (n = 2 BPH and 3 PCa, each sample in triplicate). Data are expressed as mean ± s.d. ***P < 0.001 (Student's t-test). Supplementary Fig. 3

7. Supplementary Fig. 4 – The potential miR-548c-3p targets were predicted using miRWalk algorithm and gene ontology analysis was performed on them. Note: Such in silico analysis has known shortcomings and further experimental proof is necessary to extend these conclusions.

8. References: 1.Taylor B, Schultz N, Hieronymus H, et al. Integrative genomic profiling of human prostate cancer. Cancer Cell. 2010;18(1):11-33. 2.Rane JK, Droop AP, Pellacani D, et al. Conserved two-step regulatory mechanism of human epithelial differentiation. Stem cell reports. 2014;2(2):180-8.