Identification of highly penetrant Rb-related synthetic lethal interactions in triple negative breast cancer

Abstract

Although defects in the RB1 tumour suppressor are one of the more common driver alterations found in triple-negative breast cancer (TNBC), therapeutic approaches that exploit this have not been identified. By integrating molecular profiling data with data from multiple genetic perturbation screens, we identified candidate synthetic lethal (SL) interactions associated with RB1 defects in TNBC. We refined this analysis by identifying the highly penetrant effects, reasoning that these would be more robust in the face of molecular heterogeneity and would represent more promising therapeutic targets. A significant proportion of the highly penetrant RB1 SL effects involved proteins closely associated with RB1 function, suggesting that this might be a defining characteristic. These included nuclear pore complex components associated with the MAD2 spindle checkpoint protein, the kinase and bromodomain containing transcription factor TAF1, and multiple components of the SCF^SKP Cullin F box containing complex. Small-molecule inhibition of SCF^SKP elicited an increase in p27^Kip levels, providing a mechanistic rationale for RB1 SL. Transcript expression of SKP2, a SCF^SKP component, was elevated in RB1-defective TNBCs, suggesting that in these tumours, SKP2 activity might buffer the effects of RB1 dysfunction.

Fig. 1

Rb status in TNBC tumour cell lines. a Western blot illustrating Rb and p16 protein expression in 30 breast tumour cell lines (TCLs). BT549, MDAMB436, MDAMB468 and DU4475 each possess loss-of-function mutations in the RB1 gene. SUM149 cells express reduced RB1 mRNA. b Scatter plot illustrating quantification of Rb protein levels delineated from a. Protein expression in Rb-defective vs. not altered, p = 0.0484, Student’s t test. c Scatter plot illustrating Rb protein expression in defective and proficient TNBC TCLs estimated by mass spectrometry data from ref. [24]. TNBC TCLs were classified into “Rb-defective” and “not altered” groups according to western blot data from a; using this classification, normalised iBAQ Rb peptide scores were compared and are shown. p = 0.0002, Fishers exact test. d Scatter plot illustrating the correlation between Rb protein and mRNA transcript levels in TNBC TCLs. Rb protein levels from a and b are compared to Rb mRNA transcript levels obtained from the CCLE database [25]. Correlation r = 0.7, p = 0.0075, Pearson’s correlation. e Oncoprint illustrating Rb and breast cancer subtype status amongst 42 TNBC TCLs. f Volcano plot illustrating mRNAs that are differentially expressed (limma analysis p < 0.05) in Rb-defective (vs. Rb not altered) TNBC TCLs. Genes functionally related to Rb are highlighted, as is Rb itself. g Volcano plot of mRNAs that are differentially expressed (limma analysis p < 0.05) in 48 Rb-defective (vs. Rb not altered) triple-negative breast tumours from the TCGA study [27]. Genes functionally related to Rb are highlighted, as is Rb itself. h.Volcano plot of mRNAs that are differentially expressed (limma analysis p < 0.05) in 55 Rb-defective (vs. Rb not altered) triple-negative breast tumours from the Metabric study [28]. Genes functionally related to Rb are highlighted, as is Rb itself

Fig. 2

Identifying highly penetrant Rb synthetic lethal effects that operate in TNBC. a Schematic illustrating the data analysis workflow used. In the first instance, 16 056 gene zGARP scores from shRNA screens in 42 TNBC cell lines described in the Colt2 data set were analysed; parallel analyses were carried out using data from the DRIVE and Achilles data sets (see main text). b Scatter plot illustrating ERBB2 zGARP scores in 77 breast tumour cell lines partitioned according to effects in ERBB2-amplified and -non-amplified TCLs. ERBB2 shRNA elicits a p < 0.0001 oncogene addiction effect (siMEM) with 80% penetrance in ERBB2-amplified tumour cell lines (red). Coverage is also shown. c Scatter plot illustrating 1065 p < 0.05 significant siMEM Rb synthetic lethal effects identified from the siMEM analysis of TNBC TCLs in the Colt2 study (step one in a). p < 0.05 effects are ranked ordered by siMEM p value. E2F3 (synthetic lethal in Rb-defective), CDK6, CDK4 and the CDK4,6 cyclin partner, Cyclin D1 (CCND1) (dependencies in Rb not altered) are highlighted. d, e. Scatter plots illustrating Z scores in 42 TNBC TCLs for two siMEM p < 0.05 candidate Rb synthetic lethal effects, PSMD1 (D) and KLF16 (E), removed from further analysis by the use of Z score filters (steps two and three in a). zGARP scores for PSMD1 (preferentially target Rb-defective, siMEM p value 3 × 10⁻⁵) indicate all but two Rb not altered tumour cell lines exhibit Z score of <−2 (median Z in not altered group < −4). zGARP scores for KLF16 (preferentially target Rb-defective, siMEM p value 0.0002) indicate that the majority of Rb-defective TCLs exhibit Z score of >−1 (median Z in defective group = −0.8), despite median Z scores being significantly different in Rb not altered vs. deficient TCLs. f–h Scatter plots illustrating Z scores in 42 TNBC TCLs for three siMEM p < 0.05 candidate Rb synthetic lethal effects, GPS1, SNRPF and SNW1, where median Z in defective group < −1 and median Z in proficient group > −2 (steps two and three in a)

Fig. 3

Highly penetrant Rb synthetic lethal effects. Scatter plots illustrating Z scores in 42 TNBC TCLs for 30 candidate Rb synthetic lethal effects, which pass Z score threshold assessment and demonstrate a penetrance score of >90%, as summarised in steps 1–5 of Fig. 2a

Fig. 4

TAF1 and SKP2 as central nodes in highly penetrant Rb synthetic lethal networks. a Scatter plot illustrating Z scores in 42 TNBC TCLs for TAF1 from the data analysis illustrated in Fig. 2a. b Cytoscape network plot illustrating 33 highly penetrant (>80% penetrance) Rb synthetic lethal effects identified as TAF1 transcription factor target genes, as annotated by ENCODE data [26, 44]. c–h Scatter plots illustrating Z scores in 42 TNBC TCLs for SKP1, SKP2, COPS1,3,4 and CKS1B from the data analysis in Fig. 2a. i Pathway diagram highlighting the role of multiple highly penetrant Rb synthetic lethal effects in the control of p27 activity. j Volcano plot illustrating mRNAs from highly penetrant Rb SL genes that are differentially expressed (limma analysis p < 0.05) in 48 Rb-defective vs. 92 Rb not altered triple-negative breast tumours from the TCGA study [27]. Four highly SCF^SKP/COP9 complex genes, highlighted in red, demonstrate significantly higher mRNA expression levels in the Rb-defective cell lines. k As per j, using data from the Metabric study [28]. l Box plots illustrating elevated SKP2 or GPS1 (COPS1) mRNA expression in Rb-defective TNBC from both the TCGA [27] and Metabric studies [28]. p-values shown calculated with Wilcox rank sum test

Fig. 5

Small-molecule inhibition of SKP2 in Rb-defective breast cell lines is synthetic lethal. a Western blot illustrating loss of Rb expression in isogenic MCF10A non-tumour breast epithelial cells with constitutive expression of either a control non-targeting shRNA (shCONTROL) or an Rb-targeting shRNA (shRB1). b Western blot illustrating loss of SKP2 protein expression in MCF10A cells 48 h after transfection with SKP2 or control, non-targeting, siRNAs (siCON1 or siCON2). c Bar chart illustrating cell inhibitory effects in isogenic MCF10A cells with/without stable expression of Rb shRNA transfected with SKP2 siRNA. Cells were reverse transfected with siRNAs as shown and cultured for 5 continuous days, at which point cell viability was assessed by use of CellTitre-Glo reagent. SKP2 siRNA caused significant cell inhibition (p < 0.001, Student’s t test) in cells with stable Rb silencing, but not in cells with wild-type Rb expression. d Dose response survival curves illustrating the cell inhibitory effects of the SKP2 small-molecule inhibitor, SKPinC1, in isogenic MCF10A cells with/without stable expression of Rb shRNA. Cells were exposed to SKPinC1 for 5 continuous days at which point cell viability was assessed by use of CellTitre-Glo reagent. Rb-defective cells demonstrated a profound sensitivity, compared to Rb wild-type cells (p < 0.0001, two-way ANOVA, SF₅₀ = 1 and >10 μM in Rb-defective and wild-type cells, respectively). e Tumour cell inhibitory effect of SKPinC1 in 13 TNBC TCLs classified according to Rb status. Cells were exposed to 1 μM SKPinC1 as in h. Surviving fractions are shown (p = 0.0022, Student’s t test). f Western blot illustrating p27 protein levels in Rb-defective cells exposed to 0.5 and 1 μM SKPinC1 for 24 h

Fig. 6

SKP2 identified as a highly penetrant Rb synthetic lethal effect in multiple, independently derived, data sets. a–c Scatter plots comparing p-values (−log₁₀) for Rb synthetic lethal effects identified in Colt2 [22], Achilles [23] and DRIVE [29] data sets. p < 0.05 effects in any single screen are shown. p < 0.05 effects in two screens are shown in the top right hand quadrant of each plot. SKP2 and SART3, which were identified in all three screens as synthetically lethal with Rb defects, are highlighted in red. p < 0.05 effects in Colt2 data were identified by siMEM, followed by Z score and penetrance filtering (Fig. 2a); p < 0.05 effects in DRIVE and Achilles data were identified by median permutation t test, followed by Z score and penetrance filtering (Fig. 2a). d Scatter plots illustrating Z scores in 42 TNBC TCLs for SKP2 from the Colt2 data analysis. e Scatter plots illustrating Z scores in 16 TNBC TCLs for SKP2 from the Achilles data analysis. f Scatter plots illustrating Z scores in 12 TNBC TCLs for SKP2 from the Drive data analysis

Fig. 7

SKP2 identified as a highly penetrant Rb synthetic lethal effect in other histologies in two independently derived data sets. a Scatter plot of 775 p < 0.05 significant Rb synthetic lethal effects identified from the MP test analysis of 373 non-breast cancer TCLs in the Drive study (step one in Fig. 1a). All 775 p < 0.05 effects are ranked ordered by MP test p value. SKP2 and E2F3 are highlighted. b Scatter plot illustrating RSA scores in 373 non-breast TCLs with Rb annotation for SKP2 sensitivity from the Drive data analysis. c Scatter plot of 1467 p < 0.05 significant Rb synthetic lethal effects identified from the MP test analysis of 467 non-breast TCLs in the Achilles study (step one in Fig. 1a). All 1467 p < 0.05 effects are ranked ordered by MP test p value. SKP2 and E2F3 are highlighted. d Scatter plots illustrating Demeter scores in 1467 non-breast TCLs with Rb annotation for SKP2 sensitivity from the Achilles data analysis. e, f Scatter plots illustrating RSA and Demeter scores in 63 and 115 lung TCLs with Rb annotation for SKP2 sensitivity from the Drive and Achilles studies, respectively. g Scatter plot of intersect of cell line between the two data sets showing SKP2 sensitivity in Drive RSA scores (x axis) and Achilles Demeter scores (y axis) for selected histologies with only a single Rb-defective line. This graph illustrates a trend between Rb defects and sensitivity to SKP2 shRNA across seven different histotypes