Fig 2: PI3Kd-S but not PI3Kd-L is resistant to small-molecule inhibition of PI3K/AKT/mTOR signalling and proliferation.(a) Assessment of PI3K/AKT/mTOR signalling following treatment with vehicle (saline) or CAL-101 (100 nM, 24 h). PI3K/AKT/mTOR signalling was assessed by western blot analysis with phospho-antibodies to AKT (pAKT) and S6 ribosomal protein (pS6). ß-Actin served as a loading control. His-tag antibody was used to demonstrate equal expression of His-tagged variant PI3Kd protein in stably transfected cell lines. Representative images from at least three independent western blot experiments. Unprocessed western images shown in Supplementary Fig. 12. (b) Proliferation in VCaP and PC-3 cells stably overexpressing the PIK3CD-S variant or PIK3CD-L variant following treatment with vehicle (saline) or selective PI3Kd small molecule inhibitor CAL-101 (24 h). Data presented as mean±s.e.m. from at least four independent experiments for each treatment group. *Significantly different from -S variant, P<0.05 by ANOVA with Dunnett’s post hoc test. (c) Treatment of PIK3CD variant-overexpressing cells with vehicle (saline) or selective AKT small molecule inhibitor MK-2206 (24 h). Proliferation was assessed using a 5-bromodeoxyuridine (BrdU) labelling assay. Data presented as mean±s.e.m. from at least four independent experiments for each treatment group. * Or #significantly different from corresponding vehicle control, P<0.05 by ANOVA with Dunnett’s post hoc test. Variance was similar among groups being compared.

Fig 3: Selective knockdown of the AA-enriched PIK3CD-S variant or race-independent PIK3CD-L variant has opposing effects on tumorigenesis and AKT/mTOR signalling in AA PCa cells.(a) RT-PCR of PIK3CD-L and PIK3CD-S variants following knockdown in the EA PCa cell line VCaP and AA PCa cell line MDA PCa 2b. Specific knockdown of the PIK3CD-L variant (left panel) was accomplished with an exon 20-targeting siRNA (siP20), while knockdown of the PIK3CD-S variant (right panel) was achieved with an siRNA targeting the region spanning exons 19 and 21 (siPj). Closed arrowheads below exons represent primer location for qRT-PCR validation of alternatively spliced transcript knockdowns. Knockdown efficiency of siP20 and siPj siRNAs was determined by the S/L ratio derived from RT-PCR reactions; ns, nonsense siRNA treatment. Representative images of n=4 independent knockdown experiments. (b) Proliferation and invasion of VCaP and MDA PCa 2b following knockdown of the PIK3CD-S (left panel) or PIK3CD-L variant (right panel). Data presented as mean±s.e.m. from at least three independent experiments for each treatment group. *P<0.05 by ANOVA with post hoc Tukey. Variance was similar among groups being compared. (c) Western blot analysis of AKT/mTOR signalling following knockdown of the PIK3CD-L (top panel) or PIK3CD-S variant (bottom panel) in VCaP and MDA PCa 2b. Level of AKT, mTOR and ribosomal S6 kinase activities is reflected by the amount of phospho-AKT (pAKT), phospho-mTOR (pmTOR) and phospho-ribosomal protein S6 (pS6) immunoblotting, respectively. ß-Actin served as loading control. Representative images from at least three independent western blot experiments.

Fig 4: PI3Kd-S but not PI3Kd-L is resistant to small-molecule inhibition of xenograft growth and metastasis.(a) Growth of PC-3 cells stably overexpressing the PIK3CD-S variant in NOD-SCID mice is resistant to CAL-101 treatment (50 mg kg-1 i.p. 3 times a week). In contrast, growth of PC-3 cells stably overexpressing the PIK3CD-L variant is sensitive to CAL-101 treatment. Data represent the mean tumour size±s.e.m. of n=10 independent mice for each treatment group at each time point. *Significantly different from saline-treated group, P<0.05 by ANOVA with post hoc Tukey. Variance was similar among groups being compared. (b) Tumour weights and gross morphology of tumour xenografts from a. Box-and-whisker plot represents mean xenograft weight in mice after 30-day vehicle or CAL-101 treatment. *P<0.05 by ANOVA with Dunnett’s post hoc test; n=10 independent mice for each treatment group. Variance was similar among groups being compared. (c) Quantification of lung metastases in NOD-SCID mice. PC-3 cells stably overexpressing PIK3CD-L or PIK3CD-S were injected into the tail vein of NOD-SCID mice treated with vehicle or CAL-101 (50 mg kg-1 i.p. 3 times a week). After 8 weeks, lungs were collected and stained with India ink and bleached with Fekete’s solution for visualization of metastatic nodules (white-coloured areas). Data presented as mean±s.e.m. of n=10 for each treatment group. *P<0.05 by ANOVA with Dunnett’s post hoc test. Variance was similar among groups being compared. (d) Representative India ink-stained lungs from treatment groups analysed in c.

Fig 5: Representation of differential splicing events in a composite oncogenic signalling pathway.DS events in AA PCa versus EA PCa were frequently detected in multiple oncogenic signalling pathways. Depicted pathway is a composite of PI3K/AKT/mTOR, RAS/RAF/MAPK, CDK/RB1, MDM2/TP53 and WNT/GSK3/CTNNB1/APC signalling. Open figures indicate oncogenes and closed figures indicate tumour suppressor genes. ‘DS’ indicates that a differential alternative splicing event was detected for a particular oncogene or tumour suppressor gene.