Mutations in the RNA Splicing Factor SF3B1 Promote Tumorigenesis Through MYC Stabilization

Published OnlineFirst March 18, 2020; DOI: 10.1158/2159-8290.CD-19-1330

Research Article

Mutations in the RNA Splicing Factor SF3B1 Promote Tumorigenesis through MYC Stabilization

Zhaoqi Liu1,2, Akihide Yoshimi3, Jiguang Wang4, Hana Cho3, Stanley Chun-Wei Lee3, Michelle Ki3, Lillian Bitner3, Timothy Chu1,2, Harshal Shah3, Bo Liu3, Anthony R. Mato5, Peter Ruvolo6, Giulia Fabbri7, Laura Pasqualucci7,8, Omar Abdel-Wahab3,5, and Raul Rabadan1,2

abstract Although mutations in the gene encoding the RNA splicing factor SF3B1 are frequent in multiple cancers, their functional effects and therapeutic dependencies are poorly understood. Here, we characterize 98 tumors and 12 isogenic cell lines harboring SF3B1 hotspot mutations, identifying hundreds of cryptic 3′ splice sites common and specific to different cancer types. Regulatory network analysis revealed that the most common SF3B1 mutation activates MYC via effects conserved across human and mouse cells. SF3B1 mutations promote decay of transcripts encoding the protein phosphatase 2A (PP2A) subunit PPP2R5A, increasing MYC S62 and BCL2 S70 phosphorylation which, in turn, promotes MYC protein stability and impair apoptosis, respectively. Genetic PPP2R5A restoration or pharmacologic PP2A activation impaired SF3B1-mutant tumorigenesis, elucidating a therapeutic approach to aberrant splicing by mutant SF3B1.

Significance: Here, we identify that mutations in SF3B1, the most commonly mutated splicing factor gene across cancers, alter splicing of a specific subunit of the PP2A serine/threonine phosphatase complex to confer post-translational MYC and BCL2 activation, which is therapeutically intervenable using an FDA-approved drug.

See related commentary by O’Connor and Narla, p. 765.

Introduction ref. 1). Analysis of the effects of SF3B1 mutations on RNA splicing has consistently identified that these mutations promote Mutations in genes encoding RNA splicing factors occur usage of aberrant branchpoint residues, which most commonly across a variety of hematologic malignancies and solid tumors, manifests in transcripts bearing aberrant intron-proximal 3′ and have provided evidence for the importance of aberrant splice site (3′ss; refs. 9, 10). To date, however, very few individual splicing to tumorigenesis (1). Mutations in the gene encoding misspliced transcripts that functionally link mutant SF3B1 to the core RNA splicing factor SF3B1 are the most common malignant transformation have been identified. In addition, across cancer types and occur as “hotspot” heterozygous point multiple distinct residues of SF3B1 are affected by hotspot mutations. SF3B1 hotspot mutations are among the most com- mutations, and many of these mutations exhibit cancer line- mon genetic alterations in patients with myelodysplastic syn- age specificity. For example, mutations at the R625 residue of dromes (MDS; refs. 2, 3), chronic lymphocytic leukemia (CLL; SF3B1 are strongly associated with uveal, acral, and mucosal refs. 4, 5), and uveal melanoma (UVM; refs. 6–8), and are also melanomas (6, 8), whereas mutations at position K700 are seen in 1% to 3% of many solid tumors, including breast-inva- present across myeloid leukemias, CLL, and epithelial cancers sive carcinoma (BRCA) and skin cutaneous melanoma (SKCM; (11). However, mechanistic differences between distinct hotspot mutations in SF3B1 are not well defined, nor is the basis for the 1Program for Mathematical Genomics, Columbia University, New York, cancer specificity ofSF3B1 mutant alleles understood. New York. 2Departments of Systems Biology and Biomedical Informatics, It is expected that functional changes associated with SF3B1 Columbia University, New York, New York. 3Human Oncology and Patho- genesis Program, Memorial Sloan Kettering Cancer Center, New York, New mutations are the result of aberrant splicing events in key genes York. 4Division of Life Science, Department of Chemical and Biological leading to deregulation in activity of regulatory networks. For Engineering, Center for Systems Biology and Human Health and State Key example, recent work identified robust aberrant splicing of Laboratory of Molecular Neuroscience, The Hong Kong University of Sci- the mRNA encoding the bromodomain protein BRD9 across 5 ence and Technology, Hong Kong SAR, China. Leukemia Service, Depart- cancer types and across individual hotspot mutations in SF3B1 ment of Medicine, Memorial Sloan Kettering Cancer Center, New York, New York. 6Division of Cancer Medicine, Department of Leukemia, The (12). However, the consequences of SF3B1 mutations on dysreg- University of Texas MD Anderson Cancer Center, Houston, Texas. 7Institute ulated gene expression, which poorly overlap with the mRNAs for Cancer Genetics, Columbia University, New York, New York. 8Depart- misspliced by mutant SF3B1 (9, 10), are not well defined. To ment of Pathology and Cell Biology, and the Herbert Irving Comprehensive this end, here we performed an analysis of the impact of mutant Cancer Center, Columbia University, New York, New York. SF3B1 on the activity of gene-regulatory networks, which may Note: Supplementary data for this article are available at Cancer Discovery be distinct from aberrant splicing changes associated with Online (http://cancerdiscovery.aacrjournals.org/). mutant SF3B1. In so doing, we elucidate the effects of SF3B1 Z. Liu and A. Yoshimi share first authorship of this article. hotspot mutations across cancer lineages at the level of both Corresponding Authors: Raul Rabadan, Columbia University Irving Medical mRNA splicing and expression. This effort identified a striking Center, 622 West 168th Street, PH18-200, New York, NY 10032. Phone: 212-305-3896; Fax: 212-851-5149; E-mail: [email protected]; effect of SF3B1 mutations on post-translational regulation of and Akihide Yoshimi, Memorial Sloan Kettering Cancer Center, Zuckerman multiple proteins with very well-established roles in tumori- Research Building, 417 E. 68th Street, New York, NY 10065. Phone: 646- genesis via inactivation of a specific regulatory subunit of the 888-3242; Fax: 646-422-0890; E-mail: [email protected] protein phosphatase 2A (PP2A) complex. The partial inacti- Cancer Discov 2020;10:1–16 vation of PP2A by aberrant splicing of a regulatory subunit doi: 10.1158/2159-8290.CD-19-1330 illuminated a potential therapeutic strategy for SF3B1-mutant ©2020 American Association for Cancer Research. malignancies via reactivation of residual PP2A activity.

JUNE 2020 CANCER DISCOVERY | OF2

RESEARCH ARTICLE Zhang et al.

AB112 6 8 6 Percent-spliced-in 2 6 E902K SF3B1 mutation Sample 3 61 6 K700E R625H 11 3 K700E 8 19 3 1 E902G H662R/D 4 −5 0 5 1 1 D894H K666N/E/T/R 2 Z-score R625C 8 2 G742D R625H/C 5 D781 E622D 1 E622D G742D D894H 1 D781E E902K/G H662D 1 Wild-type H662Q 3 H662R 1 K666E 2 1 1 6 3 K666N 1 K666R 1 K666T 4 Cohort CUMC CLL Leucegene AML TCGA UVM TCGA THYM dbGaP CLL TCGA AML TCGA SKCM TCGA LIHC ICGC CLL MSK K562 TCGA KIRC TCGA BRCA GEO Nalm-6 GEO MDS TCGA PRAD TCGA BLCA Oxford MDS TCGA SARC

C Splicing event 0.06 K666N K700E R625 0.04 00.6 ∆PSI Density 0.02 D

8 0.00 ave.exp(K700E) - ave.exp(R625) > 0 ave.exp(K700E) - ave.exp(R625) < 0 0255075 100 UBL7 EIF3B Distance (bp) from associated canonical 3′ss TPT1 E p53-dependent DNA damage response MHC class II antigen presentation 6 POLR1E Activation of NF-κB in B cells mTORC1-mediated signaling COQ4 Regulation of apoptosis Regulation of IFNα signaling ANAPC5 Heme metabolism NOTCH3/4 signaling HSPA8 WNT signaling TIE2 signaling AARS RPL12 NDRG3 PFDN5 4 SLC3A2 HIF1A Common MRPL3 CTCF DDIT3 UBA1 TTC3 UQCC1 Nonsense-mediated decay MXI1 PHF20 MRPS9 mRNA capping ARAF

( Q value) R625 vs. K700E PCM1 Polymerase II initiation ERGIC3 CIZ1 EDEM2 PPIH Nucleotide excision repair 10 2 NOP58 BET1L M/G1 transition ERK inactivation NDUFB1 RPS15 EFTUD2 RALBP1 DNA repair STARD3 DNA replication G2/M checkpoints Log FBXW5 IFI35 Mismatch repair Translation ERK/MAPK targets ERGIC3 Q = 0.05 mRNA processing Basal transcription factors Apoptotic cleavage DNA strand elongation of cellular proteins 0 SKCM −0.50 −0.25 0.00 0.25 0.50 mTOR signaling ERBB4 signaling R625 (UVM, SKCM) ∆PSI K700E (CLL, BRCA, MDS) Steroid hormones ERK/MAPK targets Prolactin receptor signaling

Figure 1. Differential aberrant splicing events across SF3B1-mutant cancers. A, Location and frequency of SF3B1 hotspot mutations from the pan- cancer dataset on the HEAT repeat domains (green hexagons) of SF3B1. The numbers in the center/on the edge of each circle denote the sample size of each mutation across all/specific to each cohort (see Supplementary Tables for abbreviations). B, Hierarchical clustering and heat-map analysis of differential 3′ss between SF3B1-mutant and WT samples. Rows and columns represent cryptic 3′ss events and samples, respectively. 1,401 cryptic 3′ss identified from the pan-cancer dataset are shown. Z-score in the matrix represents normalized PSI. C, Density plot of distance in base pairs from associated canonical 3′ss to cryptic 3′ss. D, Volcano plot representation of differentially spliced changes between SF3B1K700E mutations (CLL, BRCA, and MDS) and SF3B1R625 mutations (UVM and SKCM) showing the magnitude (difference of PSI; x-axis) and significance [-log10 (q value); y axis]. E, Representative gene sets significantly enriched in Gene Set Enrichment Analysis comparing all SF3B1-mutant and WT samples, as well as the results from comparisons done within each cancer type.

Results that enable the identification and quantification of alternative 3′ss with enhanced sensitivity. To this end, we collected Pan-Cancer Analysis Identified Differential RNA sequencing (RNA-seq) data from 98 tumors and 12 Splicing Events Based on Lineage and isogenic cell lines harboring SF3B1 hotspot mutations in SF3B1 Mutant Allele the region encoding the C-terminal HEAT repeat domains To provide a more complete understanding of how SF3B1 together with 102 randomly chosen SF3B1 wild-type (WT) mutations alter splicing globally, we performed a compre- controls (Fig. 1A; Supplementary Table S1; Methods). Hot- hensive pan-cancer analysis using computational approaches spot mutations mostly clustered into two specific residues

OF3 | CANCER DISCOVERY JUNE 2020 AACRJournals.org

The Role of APOC2--CD36 Interaction in AML RESEARCH ARTICLE within SF3B1’s HEAT repeat domains (Fig. 1A). The most SF3B1R625 (UVM and SKCM; Fig. 1D; Supplementary Meth- frequent hotspot mutation, K700E, was the most common ods). This analysis identified a list of cryptic ′3 ss events that mutation in CLL, MDS, acute myeloid leukemia (AML), and are probably specific to differentSF3B1 mutational hotspots, BRCA, whereas mutations at position R625 were mostly suggesting that K700E and R625 positions may have a dif- restricted to UVM and SKCM. All the E902K/G variants were ferential preference of 3′ss usage. exclusively observed in bladder urothelial carcinoma (BLCA), We extended the above pan-cancer analyses of the effects whereas K666 mutations were present across multiple tumor of mutant splicing factors to established hotspot mutations types. The association of specificSF3B1 mutational hotspots in SRSF2 and U2AF1. We identified 52SRSF2 -mutant samples with distinct tumor types suggests the potential for allele- (including 29 SRSF2P95-mutant samples), as well as 59 U2AF1- specific splicing abnormalities and unique pathogenic roles mutant samples (5 U2AF1Q157-mutant and 32 U2AF1S34F- of each SF3B1 mutational hotspot within each lineage of mutant samples; Supplementary Fig. S2A; Supplementary cancer. To examine this possibility, unsupervised hierarchi- Table S3). Analysis of the effects of these mutations on splic- cal clustering was applied to the identified aberrant ′3 ss ing identified that most of the alternative splicing events in (Fig. 1B; Supplementary Table S2). Unsupervised clustering SF3B1-mutant samples were classified into alternative ′3 ss, clearly separated SF3B1 mutant from SF3B1 WT samples, whereas the predominant events in SRSF2- or U2AF1-mutant with the exception of all The Cancer Genome Atlas (TCGA) samples were cassette exons (Supplementary Fig. S2B), which BLCA (E902K/G) as well as one sarcoma case (TCGA SARC, suggested the uniqueness of mutant SF3B1’s role in pro- D894H), suggesting that these specific mutations do not lead moting cryptic 3′ss usage. Next, unsupervised hierarchical to aberrant 3′ss usage. In addition, the clustering dendro- clustering was performed on all SF3B1-, SRSF2-, and U2AF1- gram substructure revealed a difference in aberrant 3′ss usage mutant samples by combining the previously identified sig- between hematologic malignancies and solid tumors, as well nificantly differential splicing events (Supplementary Fig. as obvious tumor lineage specificity (Fig. 1B; Supplementary S2C). Consistent with our previous observation, SF3B1 hot- Fig. S1A; χ2 test; P value = 4.33e-51). For instance, patients spot mutations were separated from SRSF2 and U2AF1 with with SF3B1-mutated CLL clustered together despite coming the exception of very few R625 mutants (UVM). In contrast, from three independent cohorts and further grouped with SRSF2- and U2AF1-mutant samples had less distinguishable another cluster consisting of AML, MDS, K562 cells, and splicing patterns from one another. Overall, these pan-cancer Nalm-6 cells, indicating greater similarity in aberrant 3′ss analyses suggested the singularity of the aberrant splicing usage among hematologic malignancies. Meanwhile, UVM pattern in SF3B1 mutants. and SKCM formed a stable subgroup, exhibiting a different Given the splicing specificity ofSF3B1 hotspot mutations, splicing pattern from TCGA BRCA patients. At the same we performed gene set enrichment analysis (GSEA) against time, unsupervised clustering also demonstrated a distinct curated MSigDB gene sets to identify tumor- and SF3B1 preference for cryptic 3′ss associated with each mutational mutation–specific cellular processes. When we compared all hotspot (Supplementary Fig. S1A; χ2 test; P value = 1.40e- the SF3B1-mutant and WT samples across all 12 tumor types, 16). Consistent with our previous observation, clustering by significantly enriched gene sets included those involved in rows demonstrated that splicing events split into two groups, nonsense-mediated decay (NMD; core enriched genes: a set mainly dominated by events associated with K700E (CLL and of ribosomal genes), DNA repair (POLL, ERCC8, POLB, etc.), others) and R625 (UVM and SKCM) mutations. In addition, and RNA polymerase II initiation (CDK7, TAF9, TAF12, etc.; we applied a principal component analysis to the same data Fig. 1E; Supplementary Table S4). We also identified differ- matrix of aberrant 3′ss usage, which indicated a more distinct ential enrichment of signaling pathways depending on SF3B1 separation of samples harboring mutations affecting the mutation and tumor types, including activation of NOTCH K700 and R625 residues of SF3B1 (Supplementary Fig. S1B). signaling in SF3B1-mutant CLL (13) and heme metabo- Most cryptic 3′ss were associated with a modest change lism and NF-κB signaling (core enriched genes: proteasome in Percent-Spliced-In (PSI) between SF3B1-mutant and WT genes) in patients with SF3B1-mutant MDS (Supplementary tumors, usually less than 10% (10). As previously reported, Table S4; refs. 14, 15). In addition, activation of multiple the distance from cryptic 3′ss to canonical site showed a oncogenic pathways such as the ERK/MAPK, mTOR, and striking peak around 10 to 30 bp upstream exon (ref. 9; Sup- apoptosis-related pathways was suggested in SF3B1-mutant plementary Fig. S1C). Further evaluation of cryptic 3′ss usage tumorigenesis (Supplementary Table S4). In contrast, we specific to each mutational hotspot suggested that the dis- identified Gene Ontology terms (enriched with misspliced tribution of distances from the canonical to the cryptic 3′ss genes) associated with RNA processing, cellular metabolism, varies among the different hotspot mutations with altered and organophosphate biosynthesis in SRSF2-mutant sam- sequence motif associated with the aberrant 3′ss (Fig. 1C; ples, and macromolecule catabolism, cellular metabolism, Supplementary Fig. S1D and S1E). For example, SF3B1R625 establishment of protein localization, and metabolism in mutations seem to be associated with a shorter (<25 bp) and U2AF1-mutant samples (Supplementary Fig. S3A–S3C). more specific insertion, whereas K666N induces usage of longer inclusions. Accordingly, the sequence contexts associ- Activation of MYC Signaling in SF3B1-Mutant CLL ated with the cryptic 3′ss exclusive to SF3B1R625 are charac- To understand the functional role of mutant SF3B1 in terized by an upstream adenine enrichment, which was not promoting tumorigenesis from the view of transcriptional observed in SF3B1K666N-specific cryptic ′3 ss (Supplementary and post-translational interaction networks, we adopted a Fig. S1D). We next compared the cryptic 3′ss with differen- well-assembled transcriptional molecular interaction network tial usage between SF3B1K700E (MDS, BRCA, and CLL) and developed in human B cells (16) and performed a regulatory

JUNE 2020 CANCER DISCOVERY | OF4

RESEARCH ARTICLE Zhang et al.

A C Genes upregulated in SF3B1 mutant Genes upregulated in SF3B1 WT Genes upregulated 0.4 Transcription factor in SF3B1-mutant CLL Positive target 0.2 NES = 2.030 Negative target CEBPBB ***P < 0.001 ESR1 0.0 FDR < 0.001 Enrichment score

NR1H3N 3 1.0 Positively correlated MYCMYMYCC 0.5

AHRAHAHR 0.0 Negatively correlated ENO1ENENOEN −0.5 0 2,000 4,000 6,000 8,000 10,000 12,000 Ranked list metric Ranked

E2F5E2E2F5 0.0 MAXM X Genes downregulated in SF3B1-mutant CLL −0.2 NFATC3NFAA 3 E2F4 NES = 2.693 ***P < 0.001 −0.4 FDR < 0.001 Enrichment score

B Downregulated genes D Differentially expressed gene Gene expression level in SF3B1-mutant CLL

MYC positively −5 0 5 correlated targets 178 Z-score MYC-positive target MYC-negative target ***P = 9.5e-29 ICGC_CLL_WT_4 3,050 166 ICGC_CLL_WT_5 ICGC_CLL_WT_9 ICGC_CLL_WT_7 ICGC_CLL_WT_2 126 2,226 ICGC_CLL_WT_1 ICGC_CLL_WT_3 ***P = 8.1e-8 ICGC_CLL_WT_6 181 MYC negatively SF3B1 WT ICGC_CLL_WT_8 ICGC_CLL_K666N_3 correlated targets ICGC_CLL_K700E_6 ICGC_CLL_K700E_5 Upregulated genes ICGC_CLL_K700E_2

mutant ICGC_CLL_K700E_4 in SF3B1-mutant CLL ICGC_CLL_K700E_1 ICGC_CLL_H662D_2 ICGC_CLL_G742D_1 ICGC_CLL_K700E_3 SF3B1

Figure 2. Activation of MYC signaling in SF3B1-mutant CLL. A, Transcriptional regulation network of significantly enriched master regulators. Nodes indicate TFs or target genes differentially expressed between SF3B1-mutant and WT CLL. B, Venn diagram showing the overlap of MYC targets and differentially expressed genes in SF3B1-mutant samples (hypergeometric test; P = 8.1e-08 for positive regulated genes; P = 9.5e-29 for negative regulated genes). C, Results of preranked GSEA on MYC target genes. Genes up/downregulated in SF3B1-mutant CLL were tested against a preranked gene list which was generated based on the correlation with MYC. D, Heat map of unsupervised clustering of differentially expressed genes based on SF3B1 genotype. Analyses in B, C, and D were performed using an expression matrix of 294 patients with CLL from the International Cancer Genome Consortium Data Portal (Z-score > 0 represents upregulation in SF3B1-mutant samples). ***, P < 0.001. network analysis on patients with SF3B1-mutated CLL. In samples with SF3B1K700E mutations, whereas it was absent in total, 721 differentially expressed genes between SF3B1-mutant UVM and SKCM samples with SF3B1R625 mutations, or AML, and WT samples from the International Cancer Genome Con- lung adenocarcinoma, pancreatic adenocarcinoma, and uter- sortium (ICGC) cohort (Supplementary Table S5) were used ine corpus endometrial carcinoma samples with SRSF2/U2AF1 as input to test the enrichment of each transcription factor hotspot mutations (Supplementary Table S6). Taken together, (TF) network target. When taking into account the direction these observations suggested that tumors harboring SF3B1K700E of transcriptional changes, the top master regulator and the mutations (and potentially mutations affecting H662, K666, only statistically significant TF wasMYC (hypergeometric test, and G742) activate the MYC transcriptional program. q value = 1.76 × 10−7 for positive regulation, q value = 4.12 × 10−3 for negative regulation; Fig. 2A; Supplementary Meth- SF3B1 Mutations Promote MYC-Driven ods). This direction-specific association was validated from an Lymphomagenesis independent dataset of 294 patients with CLL from the ICGC We next sought to verify the effects of MYC activation CLLE-ES project (Fig. 2B). We further confirmed this using a by mutant SF3B1 in the B-cell lineage in vivo. To achieve preranked GSEA as well as unsupervised clustering (Fig. 2C this, we crossed Cd19-cre+/− mice with conditional Sf3b1K700E/+ and D). When repeating the analysis on additional cancer types knock-in mice (17) to generate Cd19-cre+/− Sf3b1K700E/+ mice. including UVM, SKCM, and BRCA, similar enrichment of the Detailed analysis revealed that there are no robust changes MYC transcriptional program was detected in TCGA BRCA in spleen weight, blood cell counts, lineage commitment, and

OF5 | CANCER DISCOVERY JUNE 2020 AACRJournals.org

The Role of APOC2--CD36 Interaction in AML RESEARCH ARTICLE

A BC*** D *** K700E/+ *** K700E/+ Control (n = 6) Sf3b1 (n = 9) 15 *** 1,500 *** Control (n = 9) Sf3b1 (n = 9) 100 100

*** *** l l E -Myc m *** 75 (n = 6) 75 E -Myc (n = 10) 10 / µ L) 1,000 m 3 50 50 vs. K700E/+ K700E/+ **P = 0.0070 Sf3b1 +E -Myc Sf3b1 m Hb (g/dL) 5 500 vs. +E -Myc 25 (n = 8) ( × 10 PLT 25 m Percent survi va Percent Percent survi va Percent P = 0.052 (n = 10) 0 0 0 0 050 100 150 200 250 300 02040 -Myc -Myc Days after transplant Days Control K700E/+ m K700E/+ Control K700E/+ m K700E/+ (n = 9) E (n = 9) E = 5) (n = 9) (n = 5) (n = 7) (n = 9) (n (n = 7) Sf3b1 Sf3b1 Sf3b1 Sf3b1 -Myc -Myc m m E E Spleen Liver FG+ +E *** Cd19-Cre Em-Myc Cd19-Cre Sf3b1K700E/++Em-Myc 400 *** Control ** *** LPF HPF LPF HPF ** 300

K700E/+ Sf3b1 200

Spleen (mg) 100 Em-Myc 0 Sf3b1K700E/+ -Myc +Em-Myc Control = 3)K700E/+ m K700E/+ (n = 3)E (n (n = 6) Sf3b1 Sf3b1 m-Myc (n = 8) H +E Control Sf3b1K700E/+ Em-Myc Sf3b1K700E/++Em-Myc Lung Liver Spleen Bone marrow

Figure 3. SF3B1 mutations promote MYC-driven lymphomagenesis. A and D, Kaplan–Meier curves of survival of primary (A) and recipient (D) mice transplanted with bone marrow (BM) cells from mice with indicated genotype [Log-rank (Mantel–Cox) test]. B and C, Hemoglobin (Hb; B) and platelet (PLT; C) counts of recipient mice at day 28 after transplant (the mean value ± SD is shown). E and F, Representative pictures of spleen and liver of recipient mice (E; an inch ruler is shown together) and spleen weights of recipient mice (F) at day 28 (Cd19-Cre+/− control and Cd19-Cre+/− Sf3b1K700E/+ mice were sacrificed for comparison; the mean value± SD is shown). G, Hematoxylin and eosin staining of tissues from representative moribund primary mice [scale bar, 200 μm: Low power field (LPF); 50μ m: High power field (HPF)]. H, Representative cytomorphology of BM mononuclear cells of recipient mice (scale bar, 20 μm). **, P < 0.01; ***, P < 0.001.

B-cell maturation in Sf3b1K700E/+ mice compared with WT Nalm-6 cells, and sorted splenic B cells from recipient mice counterparts (Supplementary Fig. S4A–S4M). Next, Cd19- (Supplementary Fig. S5C–S5E). This revealed a significant cre+/− Sf3b1K700E/+ mice were crossed to mice transgenic for overlap in aberrant 3′ss events in SF3B1-mutant CLL sam- c-Myc driven by the IgH enhancer (Eμ-Myc mice; ref. 18) to ples, Nalm-6 cells, and murine B cells (Fig. 4A–D; Supple- generate control (Cd19-cre+/−), Sf3b1K700E-mutant (Cd19-cre+/− mentary Fig. S5F and S5G; Supplementary Tables S7–S9). Sf3b1K700E/+), Eμ-Myc (Cd19-cre+/− Eμ-MycTg/+), and Sf3b1K700E Most of these aberrant 3′ss were validated by RT-PCR in both Eμ-Myc double-mutant (Cd19-cre+/− Sf3b1K700E/+ Eμ-MycTg/+) samples from patients with CLL and isogenic Nalm-6 cells mice (Supplementary Fig. S5A). Although Cd19-cre+/− and resulted in decreased protein levels (Fig. 4E and F; Sup- Sf3b1K700E/+, Cd19-cre+/− Eμ-MycTg/+, and Cd19-cre+/− control plementary Fig. S6A–S6F). Given that MYC activation was mice did not develop B-cell malignancies within 300 days, 4 of observed specifically inSF3B1 -mutant CLL and BRCA where 8 Sf3b1K700E/+ Eμ-MycTg/+ double-mutant mice developed B-cell mutations predominantly affect the K700 residue, we hypoth- leukemia/lymphoma (Fig. 3A; Supplementary Fig. S5B). In esized that aberrant 3′ss usage specific toSF3B1 K700E-mutant serial transplantation into sublethally irradiated recipients, cancers might be responsible for MYC activation. Interest- mice transplanted with Cd19-cre+/− Sf3b1K700E/+ Eμ-MycTg/+ ingly, mRNA expression levels of MYC were not elevated in double-mutant cells developed severe anemia, thrombocy- SF3B1-mutant CLL, BRCA, or Nalm-6 cells compared with topenia, and splenomegaly, resulting in significantly shorter SF3B1WT counterparts, and SF3B1-mutant samples did not survival compared with single-mutant or control mice (Fig. harbor MYC amplification (Supplementary Fig. S7A–S7E). 3B–H). These data provide the first evidence thatSF3B1 muta- These observations led us to hypothesize that MYC activa- tions contribute to tumorigenesis in vivo. tion in SF3B1K700E-mutant cancers might occur via post-tran- To understand the molecular mechanisms for MYC acti- scriptional and/or post-translational mechanisms. Among vation across SF3B1-mutant cells and our mouse model, we genes whose product has previously been shown to result next analyzed RNA-seq data from patients with CLL, isogenic in increased stability of MYC protein (19–21), PPP2R5A was

JUNE 2020 CANCER DISCOVERY | OF6

RESEARCH ARTICLE Zhang et al.

ABC 6 25 CLL patient samples MAP3K7 Isogenic Nalm-6 cells MICAL1 SF3B1 WT vs. mutant SF3B1 WT vs. mutant PPP2R5A Common (n = 315) (n = 411) 20 Others Birc6 Fbxw7 ELP2 Map3k7 PPM1M 4 Tor1aip2 BIRC6, DLG1, EML3, EPG5, MAP3K7, SMURF2 146 15 TTI1 Adamts6 149 242 PPM1M, PPP2R5A, TOR1AIP2, value) value) CobII1 P FBXW7 KANSL3 P

( ( TRMT13, TTLL3, VARS2, etc. Dmxl1 15 10 CEP135 10 9 12 10 PFDN5 Ppp1m Log Log CHTF18 2 Ppp2r5a Overlapping − − DMXL1 Ttll3 220 CLL vs. Nalm-6: ***P = 2.20e-16 5 COBLL1 CLL vs. mouse: ***P = 1.72e-11 Nalm-6 vs. mouse: ***P = 2.02e-16 Mouse splenic B cells TRRAP CDK8 0 0 Sf3b1 WT vs. K700E (n = 256) 0.00.2 0.40.6 0.81.0 0.00.2 0.40.6 0.8 ∆PSI ∆PSI

D EFSF3B1 WT SF3B1 K700E PPP2R5A 3′ alternative splice site SF3B1 WT vs. mutant SF3B1 WT vs. mutant CLL patient 2 2% RT-PCR Aberrantly spliced genes Aberrantly spliced genes WT 102 CLL- 1 CLL- 2 CLL- 3 CLL- 4 CLL- 5 CLL- 6 CLL- 7 CLL- 8 CLL- 9 CLL-10 CLL-11 CLL-12 CLL-13 in SF3B1 mutant CLL in SF3B1 mutant Nalm-6 2 2% (n = 315) (n = 411) 20 118 29% MAP3K7 CDK8 CEP135 H662Q 13 49 19% CHTF18 SETD4 240 DLG1 165 5 57 5% SYVN1 141 LIG3 TTI1 MAP3K7 K666N 8 88 7% 17 3 9 103 CDK8

... 20% 2 6

4 sogenic Nalm-6 cell s K700E 35 I 7 18% PPP2R5A 00 31 1 1% CEP135 WT 90 421 11 409 DPH5 13 7% Validated ELP2 176 NDST2 1 1% PPP2R5A KANSL3 61 180 UBA7 48% MICAL Mut 36 65 Downregulated genes Downregulated genes 38% RNF2 atients with CLL 84 58 in SF3B1 mutant CLL in SF3B1 mutant Nalm-6 P 52% (n = 441) (n = 447) 78 SMURF2 PPP2R5A b 0204060 FBXW7

Exon 4 Exon 5 Not ~ ~ b × 100 (%) a a+b validated GAPDH

GH IJ

1 shControl (T1/2 = 4.78 ± 0.32) 3 **P = 0.0029 shUPF1-1 (T1/2 = 7.24 ± 0.65*) 0 ** 2.0 mRNA * shUPF1-2 (T = 9.21 ± 0.32**) 2 *P = 0.013 UPF1 -2 UPF1 -1 *** 1/2 ** ICGC 1.5 shControl sh sh −1 *** * 1 CUMC UPF1 1.0 −2 dbGap WB β-Actin 0 PPP2R5A 0.5

PPP2R5A 2 −3 −1

-score normalized 0.0 Log relative expression −4 Z −2 04812 WT (n = 7) (n = 6)

PPP2R5A relative expressio n K700E Hrs WT Mut (n = 21) (n = 22) K SF3B1 WT SF3B1 K700E L N CLL patient WB K700K K662Q K666N K700E 700K 666N 700E −+−+−+−+ K700K (T1/2 = 15.0 min) K H662Q K K PPP2R5A

CLL-14 CLL-15 CLL-16 CLL-17 CLL-18 CLL-19 CLL-20 PPP2R5A K700E (T1/2 = 39.5 min) pMYC (S62) pMYC (S62) c-MYC 103 pMYC (T58) MYC β-Actin MYC 102 pBCL2 (S70) pBCL2 (S70) BCL2 M 101 BCL2 SF3B1 K700K SF3B1 K700E PPP2R5A PPP2R5B PPP2R5A hrs 0 0.25 0.5 1.0 2.0 0 0.25 0.5 1.0 2.0 100 PPP2R5C β-Actin c-MYC 1 PPP2R5D MYC remaining (% ) 10− PPP2R5A 0.00.5 1.01.5 2.0 PPP2R5E -Actin -Actin β Hrs β

Figure 4. Aberrant splicing of PPP2R5A leads to MYC activation via post-translational regulation. A and B, (Half) volcano plot showing alternative 3′ss events in SF3B1-mutant CLL (A) and mouse Sf3b1K700E splenic B cells (B). Genes highlighted in green represent genes that are misspliced in both human SF3B1-mutant CLL and mouse Sf3b1K700E splenic B cells. C, Venn diagram of numbers of differentially spliced genes in indicated datasets (Fisher exact test). D, Venn diagram of numbers of differentially spliced and downregulated genes in SF3B1-mutant CLL and Nalm-6 samples. E, Sashimi plots of PPP2R5A 3′ss in isogenic Nalm-6 cells (top) and representative CLL samples (bottom) with or without SF3B1 mutations. F, Representative RT-PCR results of aberrantly spliced transcripts in BM samples from patients with CLL with or without SF3B1 mutations. G, Western blot (WB) analysis of Nalm-6 SF3B1K700E knock-in cells transduced with shRNAs against UPF1 (representative results from three biologically independent experiments with similar results). H, Half-lives of PPP2R5A transcripts with alternative 3′ss were measured by qPCR (n = 3; the mean value ± SD is shown; P values* compared with shControl). I, Box plot of PPP2R5A expression from patients with CLL with or without SF3B1 mutations based on RNA-seq data (sources indicated with different colors; the mean value ± SD is shown). J and K, Quantitative real-time PCR (J) and WB (K) analysis of primary CLL patient samples with or without SF3B1 mutations (the mean value ± SD is shown). L and N, WB analysis of protein lysates from isogenic Nalm-6 cells (representative results from three independent experiments are shown). M, Isogenic Nalm-6 cells were treated with 100 μg/mL cycloheximide, and cell lysates were prepared at the indicated time points after treatment. The amount of MYC protein at each time point (left) was quantified using ImageJ, and protein half-life was calculated using Prism 7 (right). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

OF7 | CANCER DISCOVERY JUNE 2020 AACRJournals.org

The Role of APOC2--CD36 Interaction in AML RESEARCH ARTICLE the only gene whose aberrant splicing events were specifically totic function (24), and (iii) BCL2 has been shown to play a observed in SF3B1K700E-, SF3B1H662D-, and SF3B1K666N-mutant role in blocking MYC-induced apoptosis (25, 26). We first veri- cells and were validated in human and mouse SF3B1-mutant fied that the S70-phosphorylated BCL2 is increased inSF3B1 - cells (Fig. 4F; Supplementary Fig. S6D–S6F). In addition, mis- mutant cells compared with WT cells, which was canceled by splicing in PPP2R5A was more robust in SF3B1K700E-mutant PPP2R5A restoration (Fig. 4L and N). Intriguingly, mRNA CLL compared with that in SF3B1R625-mutant UVM (Supple- expression of proapoptotic BH3-only BCL2 family genes mentary Fig. S7F). These results led us to focus on the roles of such as BAD, BIK, BMF, and BBC3 (PUMA) was decreased in PPP2R5A loss in MYC activation in more detail. SF3B1-mutant Nalm-6 cells (Fig. 5A–E; Supplementary Table S10), suggesting that S70-phosphorylated BCL2–mediated Aberrant Splicing of PPP2R5A Leads to MYC antiapoptotic function coupled with transcriptional down- and BCL2 Activation via Post-Translational regulation of proapoptotic BCL2 family might contribute to Modification the antiapoptotic effect of MYC. Consistent with this, SF3B1- PPP2R5A is a regulatory B subunit of the major serine/ mutant cells were more resistant to apoptotic stimulus, such threonine PP2A protein complex (which consists of a core as that induced by a topoisomerase inhibitor, camptothecin trimer between the scaffolding A subunit, catalytic C sub (CPT; Fig. 5F–H). To evaluate the role of BCL2 phospho- unit, and a regulatory B subunit; reviewed in ref. 22). PP2A rylation, we utilized S70-mutant forms of BCL2 that either complex containing PPP2R5A dephosphorylates serine 62 mimic (S70E) or prevent (S70A) BCL2 S70 phosphorylation (S62) of MYC, resulting in an unstable, singly threonine 58 (23). Expression of BCL2 S70E cDNA significantly protected (T58)–phosphorylated form of MYC that is a substrate for SF3B1WT cells from CPT-induced apoptosis and enhanced cell ubiquitination by FBXW7 and degradation by the 26S pro- growth, whereas BCL2 S70A expression had minimal effects teasome (19). As expected, aberrant PPP2R5A transcript was compared with BCL2 WT. On the other hand, expression of confirmed to result in NMD (Fig. 4G and H), and a decrease BCL2 S70A induced more apoptosis upon treatment with in PPP2R5A mRNA and protein expression was observed, CPT in SF3B1K700E cells, resulting in inhibition of cell growth which was associated with MYC upregulation in SF3B1K700E- (Fig. 5I; Supplementary Fig. S9A–S9C). Next, we investigated mutant CLL, Nalm-6 cells, and mouse splenic B cells (Fig. how BCL2 phosphorylation enhances antiapoptotic func- 4I–K; Supplementary Fig. S7G–S7J; Supplementary Tables tions. Protein half-lives of BCL2 WT, S70A, and S70E were S10 and S11). Among genes encoding regulatory B subu- almost identical, suggesting that BCL2 phosphorylation does nit family members, PPP2R5A was specifically misspliced not affect protein stability of BCL2 (Supplementary Fig. and downregulated in SF3B1-mutant tumors compared S9D). As BCL2 is known to bind BAK and antagonize the with WT counterparts across cancers (Fig. 4L; Supplemen- effects of BAK (which is required for mitochondrial per- tary Fig. S8A–S8G). The half-life of PPP2R5A protein was meabilization during apoptosis), we compared the binding also diminished in SF3B1-mutant cells (Supplementary affinity of either WT, S70A-mutant, or S70E-mutant BCL2 Fig. S8H), although the mechanism for this is unknown. protein with BAK by immunoprecipitation assays following Consistent with the dramatic increase of the S62-phospho- treatment with CPT for 5 hours. This revealed that BCL2 rylated form of MYC in SF3B1-mutant cells compared with S70E had higher binding affinity to BAK, whereas BCL2 that in WT cells (Fig. 4L; Supplementary Fig. S8I and S8J), S70A had reduced affinity compared with BCL2 WT protein. SF3B1K700E expression significantly decreased the rate of These data suggest that BCL2 phosphorylation status at S70 MYC protein degradation following inhibition of protein regulates binding with BAK (Fig. 5J). Levels of cytochrome C synthesis with cycloheximide (Fig. 4M). In contrast, there release in the isogenic cell lines above were analyzed by intra- was no difference in MYC protein synthesis rate between cellular flow cytometry following treatment with CPT for 5 SF3B1 WT and mutant cells (Supplementary Fig. S8K). hours. Most likely as a result of BCL2 activation, cells with Importantly, restoration of PPP2R5A expression in SF3B1- BCL2 S70A expression demonstrated a significant increase in mutant cells almost completely reduced phosphorylation cytochrome C release compared with cells with BCL2 WT or levels of MYC at S62, leading to protein degradation (Fig. S70E expression, which is consistent with our previous data 4N). To further evaluate the role of MYC phosphoryla- that expression of BCL2 S70E cDNA significantly protected tion, we utilized S62-mutant forms of MYC that either cells from CPT-induced apoptosis and enhanced cell growth, mimic (S762D) or prevent (S62A) MYC S62 phosphoryla- whereas BCL2 S70A expression had minimal effects com- tion. When we overexpressed HA-tagged MYC WT, S62A, pared with BCL2 WT (Fig. 5K). Overall, these findings suggest or S62D in SF3B1WT and SF3B1K700E cells, S62A muta- that the mutant SF3B1–PPP2R5A axis regulates post-transla- tions significantly shortened MYC half-life, whereas S62D tional modification of BCL2 at S70 and plays an important mutations prolonged MYC half-life in both SF3B1WT and role in the antiapoptotic phenotype of SF3B1-mutant cells. SF3B1K700E cells. These data are consistent with our obser- Given that numerous missplicing events occur in the setting vations that the differential status of S62 phosphorylation of mutant SF3B1, we next sought to evaluate the link among of MYC between SF3B1WT and SF3B1K700E cells altered MYC mutations in SF3B1, PPPR5A missplicing and expression, and protein stability (Supplementary Fig. S8L). MYC activity in more detail. We first evaluated the effects of We were further interested in potential BCL2 involvement restoring PPP2R5A levels in Sf3b1K700E Eμ-Myc double-mutant in the context of MYC activation in mutant SF3B1–mediated tumors. We expressed an empty vector or Ppp2r5a cDNA in tumorigenesis program because (i) the serine 70 (S70)–phos- Cd19-cre+/− Sf3b1K700E/+ Eμ-MycTg/+ tumors and transplanted phorylated form of BCL2 is another target of PPP2R5A (23), these cells into recipients (Fig. 6A). Recipients transplanted (ii) phosphorylation at S70 is required for BCL2’s antiapop- with Cd19-cre+/− Sf3b1K700E/+ Eμ-MycTg/+ cells with restored

JUNE 2020 CANCER DISCOVERY | OF8

RESEARCH ARTICLE Zhang et al.

A BC BAD BIK −Log10 (P value) 5 5 30 4 4 3 3 *** 2 2 *** ****** *** *** *** *** *** *** *** *** *** *** *** 1 *** 1 * 20 0 0 Relative expression Relative expression PPP2R5A 0 µmol/L5 µmol/L10 µmol/L 0 µmol/L5 µmol/L 10 µmol/L Camptothecin ( mol/L) Camptothecin ( mol/L) K700K CDKN1A µ µ H662Q CDK8 K666N BIK DE K700E BMF 10 5 BMF 5 BBC3 (PUMA) BBC3 (PUMA) 4 4 BAD 3 3 2 ** *** 2 *** *** *** *** 1 ** *** *** 1 *** *** *** *** *** ** *** *** −5 −4 −3 −2 −1012345 0 0 Relative expression 0 µmol/L5 µmol/L 10 µmol/L Relative expression 0 µmol/L5 µmol/L 10 µmol/L Log2 (fold change) Camptothecin (µmol/L) Camptothecin (µmol/L)

F GH PI K700K H662Q K666N K700E 0.032 0.096 0.050 0.13 0.046 0.046 0.075 0.075 30 Early apoptosis 50 Late apoptosis Necrotic Late apoptotic 40 K700K mol/L H662Q µ Alive 20 *** 0 Early 30 *** K666N % apoptotic *** % *** 20 K700E 99.1 0.74 99.1 0.73 99.3 0.60 99.2 0.67 10 *** *** 0.35 29.4 0.48 8.2 0.2 7.8 0.682 7.7 *** 10 *** *** 0 0 *** mol/L

µ 0510 0510 5 Camptothecin (µmol/L) Camptothecin (µmol/L) Camptothecin 47.9 22.3 85.3 6.0 86.2 5.8 85.2 6.4 1.6 36.9 1.8 20.2 0.66 18.0 2.12 18.4 I SF3B1K700K SF3B1K700E BCL2 WT S70A S70EWT S70A S70E mol/L Relative cell number

µ 2.0 2 10

38.5 23.1 61.7 16.2 69.4 12.0 69.7 9.8 4 1.5

Annexin V - APC Da y 6 1.0 8 0.5 JK ** 25,000 5% input HA-IP *** Isotype 20,000 BCL2 WT WT S70A S70E WT S70A S70E BCL2 S70A 15,000 BAK

BCL2 S70E MF I HA-BCL2 Count 10,000

5,000

0 Alexa647-Cytochrome C WT S70A S70E

Figure 5. Impaired apoptosis in SF3B1-mutant cells linked to increased BCL2 S70 phosphorylation. A, Volcano plot showing differentially expressed genes in isogenic SF3B1-mutant Nalm-6 cells |Log2 (Fold change) | = 0.5 and –log10(P value) = 2 are used as thresholds; genes with |x| > 5 and/or y > 30 are shown on |x| = 5 or y = 30]. B–E, qPCR of BAD (B), BIK (C), BMF (D), and BBC3 (PUMA; E) in isogenic Nalm-6 cells upon treatment with CPT at indicated concentrations for 5 hours (n = 3; the mean value ± SD is shown; P values are shown for comparisons of K700K vs. each mutant). F, Representative flow cytometry analysis of isogenic Nalm-6 cells upon treatment with CPT for 5 hours. G and H, Percentages of cells in early (Annexin V+ PI–; top, G) and late (Annexin V+ PI–; bottom, H) apoptotic phases with different concentrations of CPT (n = 3; the mean value ± SD is shown; P values are shown for comparisons of K700K vs. each mutant). I, Heat map was made using Prism 7 based on the cell growth upon CPT treatment at 1 μmol/L (n = 3; scale represents the relative cell number at each day relative to BCL2 WT–expressing cells). J, Parental Nalm-6 cells were transduced with either WT, S70A-mutant, or S70E-mutant BCL2 (all HA-tagged), treated with 5 μmol/L CPT for 5 hours, and protein lysates were immunoprecipitated with an HA antibody. Eluted protein (95% was used for BAK blotting and 5% was used for HA blotting) as well as 5% input protein were analyzed by WB. K, Isogenic Nalm-6 cells from J were treated with 5 μmol/L CPT for 5 hours, and cytochrome C release was evaluated by intracellular flow cytometry staining [left; representative flow cytometry analysis from four independent experiments; mean fluorescence intensity (MFI) is provided on the right;n = 4; the mean ± SD]. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

OF9 | CANCER DISCOVERY JUNE 2020 AACRJournals.org

The Role of APOC2--CD36 Interaction in AML RESEARCH ARTICLE

A Infection Transplantation Serial transplantation Cd19-cre+/− MSCV-IRES-GFP Sf3b1 K700E 200 cGy 200 cGy 10,000 cells 50,000 cells Em-Myc tumor GFP ×9 GFP ×5 sorting sorting Splenic cells 200 cGy 200 cGy 10,000 cells 50,000 cells GFP ×10 GFP ×5 Cd45.2 sorting sorting Survival analysis MSCV-PPP2R5A Cd45.1 mice Analysis on day 60 Cd45.1 mice -IRES-GFP Day −2, −1 Day 0

B CD E 15 * 2,000 300 *** *** EV PPP2R5A L) L)

µ 1,500

EV µ

PPP2R5A / / 3

10 3 200 PPP2R5A

× 10 1,000 × 10 (

β-Actin 5 T( 100 500 PL WBC pleen weight (mg) 0 0 S 0

EV EV EV

PPP2R5A PPP2R5A PPP2R5A F G DAPI− -gated, % DAPI− cells 100 BM Spleen Liver PB *** *** EV PPP2R5A 58.2 63.4 32.2 39.5 *** 80 *** 100 60

28.9 16.1 21.2 60.2 (% ) EV

40 50 GFP + vs.

20 ercent surviva l **P = 0.0026

36.0 24.7 12.3 7.79 P 0 0 47.6 42.1 19.6 91.6

GFP (EV/PPP2R5A) 0510 15 20 25 EV EV EV EV

PPP2R5A Days after transplant PPP2R5A PPP2R5A PPP2R5A PPP2R5A PE-CD45.1 BM Spleen Liver PB

Figure 6. Restoration of PPP2R5A expression in Sf3b1-mutant Eμ-Myc tumors reduces tumorigenicity. A, Schematic representation of the study design. B, Confirmation of PPP2R5A expression inCd19-cre +/− Sf3b1K700E/+ Eμ-MycTg/+ cells with expression of an empty vector (EV) or PPP2R5A cDNA. C and D, white blood cell (WBC; C) and PLT (D) counts of recipients (the mean value ± SD is shown). E, Representative photo of spleen (left) and spleen weight of recipients (right; the mean value ± SD is shown; an inch ruler is shown together). F, Percentages of GFP+ DAPI− cells in BM, spleen, liver, and peripheral blood (PB; right) with representative flow cytometry analysis (left; the mean value± SD is shown). G, Kaplan–Meier curves of survival of recipients transplanted with Sf3b1K700E/+ Eμ-Myc tumor ± PPP2R5A expression [Log-rank (Mantel–Cox) test]. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

PPP2R5A expression had lower white blood cell (WBC) counts mor therapy. For this purpose, we assessed the therapeutic and less severe thrombocytopenia, splenomegaly, and tumor potential of the FDA-approved oral PP2A activator FTY- cell infiltration in multiple organs compared with control mice 720 (also known as fingolimod; refs. 27–29).SF3B1 -mutant (Fig. 6B–F). In addition, serial transplantation of Sf3b1K700E/+ Nalm-6 cells were more sensitive to FTY-720 treatment than Eμ-MycTg/+ tumor cells revealed that PPP2R5A overexpression their SF3B1WT counterparts, experiencing growth inhibition significantly prolonged survival (Fig. 6G). Conversely, deple- at lower concentration compared with SF3B1WT cells in vitro tion of PPP2R5A in SF3B1WT Nalm-6 cells enhanced phos- (Fig. 7A and B; Supplementary Fig. S11A–S11D). Similar phorylation of both MYC and BCL2, and conferred growth results were obtained with other PP2A activators including advantage in vitro (Supplementary Fig. S10A–S10C). Similarly, the recently reported DT-061 (30, 31) as well as perphena- genetic depletion of Ppp2r5a in Sf3b1WT Eμ-MycTg/+ tumors zine (32) in vitro (Supplementary Fig. S11E–S11H). Moreo- enhanced disease progression in vivo (Supplementary Fig. ver, both S62-phosphorylated MYC and S70-phosphorylated S10D–S10K). Overall, these data strongly suggest that altera- BCL2 decreased in a dose-dependent manner upon treat- tions in PPP2R5A levels due to aberrant splicing by mutant ment with FTY-720 (Fig. 7C). MYC protein half-life was SF3B1 promote MYC-driven tumorigenesis. shortened with FTY-720 treatment, which was accompanied by downregulation of gene expression of MYC target genes SF3B1-Mutant Cells Are Preferentially Sensitive (Supplementary Fig. S11I and S11J). In contrast, sensitiv- to Pharmacologic PP2A Activation ity to FTY-720 was diminished in SF3B1-mutant cells with The molecular and functional consequences of loss of PPP2R5A overexpression (Supplementary Fig. S11K–S11M). PPP2R5A-related MYC and BCL2 dephosphorylation sug- Finally, to assess the efficacy of FTY-720in vivo, we serially gested that activation of PP2A complex containing the transplanted Cd19-cre+/− Sf3b1+/+ or Sf3b1K700E/+-mutant Eμ- PPP2R5A regulatory B subunit may be exploited for antitu- MycTg/+ tumors into sublethally irradiated CD45.1+ recipient

JUNE 2020 CANCER DISCOVERY | OF10

RESEARCH ARTICLE Zhang et al.

A B C FTY-720 SF3B1 K700K SF3B1 K700E K700K (IC50 = 14.9 µmol/L) (µmol/L) 012481632 100 Relative cell number H662Q (IC50 = 3.8 µmol/L) Day 2 1.0 FTY-720 µmol/L40112 20 4 K666N (IC = 2.6 µmol/L) K700K 50 Day 4 0.8 pMYC (S62) K700E (IC50 = 3.0 µmol/L) Day 2 50 H662Q 0.6 1.00 0.64 0.48 0.41 1.43 0.68 0.55 0.50 Day 4 MYC K666N Day 2 0.4 1.00 0.69 0.87 0.70 2.17 1.99 1.83 1.67 Day 4 pBCL2 (S70) Day 2 0.2 1.00 0.55 0.30 0.32 2.59 1.93 0.54 0.10 0 K700E Cell viability at 72 hrs (% ) −20 24 Day 4 0.0 BCL2 Log(FTY-720), µmol/L 1.00 0.98 0.95 0.96 1.05 1.06 1.07 1.04 PPP2R5A 1.00 1.05 0.93 0.92 0.64 0.79 0.30 0.65 β-Actin 1.00 0.87 0.93 0.92 0.94 0.94 1.00 1.05 D 200 cGy Tail-vein injection Vehicle 20,000 cells +/ − ×6 E FG FTY-720 (3 mg/kg/day) 30 ** *** 16 1,500 ** * 20,000 cells *** ** L)

Sf3b1 WT ×6 L) Cd19 - cre µ / E m Myc tumor µ 3 /

20 14 3 1,000

CD45.1 mice Day 8–12, 15–19 10 10 or × ( Tail-vein injection × Vehicle 10 12 ( 500 T 20,000 cells Hb (g/dL) +/ − WBC Analysis on day 19 6 PL

× Follow-up for surviva l FTY-720 (3 mg/kg/day) 0 10 0 20,000 cells Veh FTY Veh FTY Veh FTY Veh FTY Veh FTY Veh FTY ×6 Cd19 - cre Sf3b1 K700E

E m Myc tumor WT K700E WT K700E WT K700E 10 doses/2 weeks CD45.1 mice

K DAPI− -gated BM Spleen Liver PB H WT K700E Veh FTY Veh FTY 69.1 57.4 33.8 62.2 28.2 21.0 42.9 37.4 Vehicle Sf3b1 WT

68.3 53.9 52.1 63.1 28.3 26.0 32.3 36.1 36.1

FTY-720 26.1 I J Sf3b1 WT 400 *** *** 1,600 * **

300 1,400 PCCy7-CD45.2 95.6 78.5 89.0 99.5 A 1.0 0.1 0.0 0.0 200 1,200

Vehicle 0.0753 0

100 1,000 Sf3b1 K700E Spleen weight (mg ) Liver weight (mg) 0 800 20.6 39.9 23.4 42.7 Veh FTY Veh FTY Veh FTY Veh FTY 76.3 27.4 56.0 56.5 WT K700E WT K700E 56.5 FTY-720 76.3 Sf3b1 K700E

L PE-CD45.1 WT K700E Veh FTY Veh FTY MYC BCL2 M FTY-720 β-Actin 100 Sf3b1 WT Vehicle l FTY 3 *** *** 1.5 P = n.s. Sf3b1K700E 50 2 1.0 vs. Vehicle *P = 0.022 FTY

1 0.5 Percent surviva vs. **P = 0.0010 0 0 0.0 010203040 Relative MYC expression Veh FTY Veh FTY Relative BCL2 expressio n Veh FTY Veh FTY Days after transplant WT K700E WT K700E

OF11 | CANCER DISCOVERY JUNE 2020 AACRJournals.org

The Role of APOC2--CD36 Interaction in AML RESEARCH ARTICLE mice and treated these mice with either vehicle or FTY-720 expression of specific regulatory B subunits. Consistent with (Fig. 7D). Similar levels of PP2A activation from in vivo FTY- this, loss of specific regulatory B subunits has been repeat- 720 treatment were seen in recipients of both Sf3b1WT and edly shown to result in cellular transformation. Interestingly, Sf3b1-mutant tumors (Supplementary Fig. S12A and S12B). the specific subunit of PP2A targeted for decay in SF3B1- Sf3b1K700E/+ Eμ-Myc tumors were more aggressive compared mutant cells, PPP2R5A (also known as B56α), is well known with Sf3b1+/+ tumors as evidenced by elevated CD45.2+ B cells to regulate human cell transformation from systematic RNAi in peripheral blood (PB), increased tumor burden in spleen screens for oncogenic PP2A regulators (34), studying PP2A and liver, increased CD45.2 chimerism across organs, and targets of polyoma T antigens (which block interaction of worsened anemia and thrombocytopenia, consistent with B56 subunits with the PP2A holoenzyme; ref. 35), and genetic the results above. However, FTY-720 treatment in vivo selec- depletion studies in mice (36). In fact, mice hypomorphic for tively improved these phenotypes in recipients of Sf3b1K700E/+ Ppp2r5a exhibit spontaneous skin tumorigenesis, lymphoid Eμ-Myc tumors, diminished MYC expression and induced cell expansion, and increased clonogenic potential of hemat- apoptosis to a greater extent in Sf3b1K700E/+ Eμ-Myc tumors opoietic precursors (36). Despite these abundant data high- than in Sf3b1WT Eμ-Myc tumors, and prolonged their survival, lighting PPP2R5A as a critical tumor-suppressive subunit whereas the effects on Sf3b1+/+ Eμ-Myc tumors were minimal of PP2A, however, a genetic basis for the loss of PPP2R5A in (Fig. 7E–M; Supplementary Fig. S12C–S12G). human cancers was absent until this current study. One additional observation from this study was that distinct hotspot mutations in SF3B1 have unique effects on Discussion RNA splicing and gene expression. For example, samples Despite abundant genetic data that suggest that mutations with mutations affecting the H4–H8 HEAT repeat domains in SF3B1 represent oncogenic drivers in malignancy, clear bio- of SF3B1 had much clearer effects on aberrant intron proxi- logical contribution of mutations in SF3B1 to tumorigenesis mal 3′ss usage than samples with mutations at the D894 has not been demonstrated. In fact, evaluation of the role of and E902 residues. Although the precise mechanistic basis mutant versus WT alleles of SF3B1 has failed to identify a role for this is not yet clear, amino acid residues within HEAT of these mutations in cancer maintenance and instead has repeats H4–H8 of SF3B1 are clustered in a small area and highlighted the requirement of the WT allele in cells with het- are exposed to solvent, suggesting that these mutations may erozygous expression of SF3B1 mutants (33). Here, through disrupt SF3B1 interactions with other spliceosomal pro- combined evaluation of the effects of the SF3B1 mutation teins or SF3B subunits (37) to induce aberrant branchpoint on splicing, gene expression, and TF regulatory networks usage. In contrast, D894 and E902 residues are located in across cancer types, we identify a novel mechanism by which the H11 domain, which may not directly affect RNA interac- the mutant SF3B1–mediated cotranscriptional alterations in tion. These computational predictions will be important to RNA splicing contribute to stabilization and activation of evaluate in future functional studies which will likely require oncogenic MYC. Activation of the MYC pathway was molecu- structural and biochemical evaluation of each mutant form larly conserved across human and mouse B-cell malignancy of SF3B1 independently. models and identifies a clear contribution of aberrant splic- Although prior work has highlighted MYC and BCL2 as ing driven by mutant SF3B1 to a biological mechanism of critical substrates of PPP2R5A-containing PP2A, it is important tumorigenesis. to acknowledge that there are additional known substrates The high frequency of change-of-function mutations in for this specific PP2A complex. Future work comprehensively SF3B1 in cancer and the reliance of SF3B1-mutant cells on defining the full spectrum of substrates of PPP2R5A-contain- otherwise WT splicing have resulted in efforts to target RNA ing PP2A and how these are altered in SF3B1-mutant can- splicing for therapeutic means in cells bearing mutations cers may therefore be very important. In addition, given the in SF3B1 and other splicing factors. However, the clinical striking redundancy in B subunits of PP2A, where 15 genes viability of targeting RNA splicing in patients is unclear, encode 26 B subunits potentially assembling >100 distinct and further efforts to dissect mechanistic dependencies in PP2A complexes, it will be important to determine how loss SF3B1-mutant cells are needed. To this end, we identify that of PPP2R5A affects other PP2A complexes in SF3B1-mutant a specific enzymatic activity, PP2A phosphatase activity, is cells. Nonetheless, the striking redundancy in PP2A com- altered in SF3B1-mutant cancers. PP2A’s substrate specificity, plexes allows for pharmacologic interventions to reactivate function, and enzymatic activities are regulated in part by residual PP2A activity in SF3B1-mutant tumors. FTY-720, the

Figure 7. Preferential sensitivity of SF3B1-mutant cells to PP2A reactivation. A, Dose–response curves of isogenic Nalm-6 cells to FTY-720 (n = 3; the mean value ± SD is shown). IC50 value for each genotype is indicated. B, Heat map was made using Prism 7 based on the cell growth upon FTY-720 treatment with increasing doses (n = 3; scale represents the relative cell number at each dose of FTY-720 relative to vehicle-treated cells with 2 or 4 days of FTY-720 treatment). C, WB analysis of protein lysates from isogenic Nalm-6 cells treated with FTY-720 at various concentrations for 24 hours (representative results from three independent experiments are shown). D, Schematic representation of the study design. E–G, WBC (E), Hb (F), and PLT (G) counts of mice (n = 6 per genotype; the mean value ± SD is shown). H–J, Representative photo of spleens (H) and weight of spleen (I) and liver (J) in recipients (n = 6 per genotype; the mean value ± SD is shown; an inch ruler is shown together). K, Representative flow cytometry analysis of CD45.2+ DAPI− cells in BM, spleen, liver, and PB (quantification of %CD45.2+ DAPI− cells is shown in Supplementary Fig. S12C–S12F). L, CD45.2+B220+ splenic cells from D at day 19 were FACS-sorted and analyzed (top; 3 mice per group; n = 1 experiment). The amount of MYC and BCL2 protein was quantified using ImageJ (bottom; the mean value ± SD). M, Kaplan–Meier curves of survival of recipient mice treated with either vehicle or FTY-720 [n = 5 per genotype; treatment period is marked in light blue; Log-rank (Mantel-Cox) test]. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

JUNE 2020 CANCER DISCOVERY | OF12

RESEARCH ARTICLE Zhang et al. therapeutic modality used here, activates PP2A by binding SET, files were downloaded from the National Center for Biotechnology a negative regulator of PP2A, and directly causing its displace- Information (NCBI) Sequence Read Archive (SRA; accession number ment from and assembly of active PP2A holoenzymes (38). GSE67039). Nine samples of SF3B1-mutant Nalm-6 cell lines were Our data highlight an important novel therapeutic approach also employed from GEO accession GSE72790, together with three K700E targeting the impact of mutant SF3B1 on post-translational WT samples. Three cases of SF3B1 K562 cells and three WT cell lines were downloaded from GEO accession GSE95011 (39). Data modification and stabilization of oncogenic MYC. The specific from six SF3B1-mutant samples (four K700E, one K666T, and one compound used here is far from the only means to reactivate K666R) and six patients with WT MDS were obtained from GEO PPP2R5A-containing PP2A, however, and future efforts to tar- accession GSE85712 (17). Another cohort of 19 SF3B1K700E MDS get SF3B1-mutant tumors by complimentary means of reacti- samples with the same number of randomly chosen WT samples were vating via allosteric activation of its phosphatase activity may added from GSE114922 (40). Six SF3B1K700E and one SF3B1K666E case also be fruitful. Finally, besides PPP2R5A, recent studies identi- of patients with CLL were accessed from dbGaP under the project of fied robust alternative splicing changes inBRD9 (12) and DVL2 The UC San Diego CLL Study (phs000767.v1.p1). (13) that also play important roles in tumorigenesis of SF3B1- Collectively, we have established transcriptomic characterization of mutant cancers. Aberrant 3′ss usage in these two genes was also pan-cancer SF3B1 mutations on 110 samples covered by 12 tumor types. observed in the current study (Supplementary Fig. S13A and S13B). The genetic effects of restoring PPP2R5A expression in Identification of Novel Cryptic ′3 ss Usage SF3B1-mutant tumors here, together with recent studies above, In order to annotate novel 3′ss that are not present in the current continue to highlight the need to study the impact of individual genome, we adopted the splice junction read output by STAR align- missplicing events generated by mutant SF3B1 to promote ment (41). This preprocess is motivated by DeBoever and colleagues disease development. (10) to enable a more sensitive detection of novel sites, specifically designed for 3/5′ss identification. To this end, all the bam files were transformed back to .fastq format (samtools) and realigned using a Methods splice junction database (10). Counts of junction reads from SJ.out.tab file (STAR output) were merged into a unique matrix, with each row For complete experimental details and computational analyses, see indicating one splice junction and column for each sample. We further also Supplementary Methods. filtered out junctions with a read coverage of< 50 summed across all samples. Then, for each junction, we defined it as a novel splice site if Pan-Cancer SF3B1-Mutant Sample Collection it only has one end (either 3′ or 5′) annotated in the provided known and Sequencing genome, whereas the other end is not. Corresponding canonical junc- RNA-seq data from patient samples or isogeneic cell lines with tion is found if it shares the same annotated end of the cryptic junction or without SF3B1 hotspot mutations were collected from Columbia and the other end also annotated. If this kind of canonical junction University Medical Center (CUMC), Memorial Sloan Kettering Can- did exist, we calculated the distance between the novel end and the cer Center (MSKCC), and public portal resources including TCGA, corresponding canonical end. If there was more than one distance, we ICGC database of Genotypes and Phenotypes (dbGaP), and Gene utilized the minimum. After adding the information of transcript’s Expression Omnibus (GEO). Detailed sample information is shown strand and relative position to the canonical site, we determined in Supplementary Table S1. whether a cryptic site is on 3′ss or 5′ss, upstream or downstream. For Six patients with CLL with SF3B1K700E mutation and six WT cases mouse RNA-seq, we reproduced this essential preprocess using a stand- were identified from CUMC. This study is approved by the Insti- ard mm10 genome (Ensembl: Mus_musculus. GRCm38.75). Junctions tutional Review Boards of Columbia University and performed in with both ends unannotated were discarded from the analysis. accordance with the Declaration of Helsinki. All patients gave their Next, with a clear relationship of each cryptic junction to its informed written consent. Poly-A pulldown was performed to enrich canonical one, we calculated the PSI using raw read counts for each mRNA from total RNA samples, and libraries were prepared using the sample, transforming the reads matrix into a PSI matrix with values Illumina RNA TruSeqv2 PE100 60M (2 × 100 bp) and sequenced using between 0 and 1. Then, for two groups of interested comparison, the Illumina HiSeq2000 instrument at the Columbia Genome Center. t test was applied for all novel cryptic splice site using PSI, rather Real-time analysis (Illumina) was used for base calling and bcl2fastq than read counts, followed by BH multiple test correction. The dif- (version 1.8.4) for converting BCL to fastq format, coupled with adap- ference of averaged PSI between two groups was also adopted as an tor trimming. important metric for potential differential splicing. A bar plot of log2 For TCGA RNA-seq data, we downloaded aligned bam files from distance in base pairs was used to demonstrate cryptic 3′ splicing pat- the cancer genomics cloud in November 2017 (cgc.sbgenomics.com). tern on significantly differentially used novel sites. This cohort includes 16 UVM, 10 BRCA, seven BLCA, five SKCM, To compare aberrant 3′ss usage between human CLL and mouse three AML, two kidney renal clear cell carcinoma, two liver hepato- B cells, mouse genes were mapped into human symbols. A hypergeo- cellular carcinoma, two prostate adenocarcinoma, two sarcoma, and metric test was used to evaluate the significant level of overlapped two thymoma samples. In the meantime, an equal number of WT common genes. samples was randomly chosen with the closest name ID of TCGA barcode for each mutated sample. Patient Samples The ICGC CLLE-ES cohort consists of six cases of SF3B1K700E, Studies were approved by the Institutional Review Boards of one SF3B1G742D, one SF3B1H662D, one SF3B1K666N, and nine randomly CUMC and MSKCC and conducted in accordance with the Declara- chosen WT samples. Raw fastq files were downloaded from the ICGC tion of Helsinki protocol. Written informed consent was obtained Data Portal (https://dcc.icgc.org) with the project name “CHRONIC from all participants. LYMPHOCYTIC LEUKEMIA - ES.” Controlled access to these data is approved with application number DACO-1023313. The sequencing Animals data are paired-end 2 × 76 bp libraries. All animals were housed at MSKCC. All animal procedures were In addition, we collected 2 patients with AML (K666N, R625C) completed in accordance with the Guidelines for the Care and Use as well as two WT samples from Leucegene datasets. Raw fastq of Laboratory Animals and were approved by the Institutional

OF13 | CANCER DISCOVERY JUNE 2020 AACRJournals.org

The Role of APOC2--CD36 Interaction in AML RESEARCH ARTICLE

Animal Care and Use Committees at MSKCC. The number of mice Statistical Analyses in each experiment was chosen to provide 90% statistical power with Statistical significance was determined by (i) unpaired two-sided K700E/+ a 5% error level. Generation and genotyping of the Sf3b1 con- Student t test after testing for normal distribution, (ii) one-way or ditional knock-in mice on C57BL/6 background were as described two-way ANOVA followed by Dunnett, Tukey, Sidak, or Dunnett (17). Eμ-Myc mice have been described previously (18). Complete multiple comparison test, or (iii) Kruskal–Wallis tests with Uncor- blood count analysis was performed on PB collected from subman- rected Dunn test where multiple comparisons should be adjusted dibular, using a Procyte Dx Hematology Analyzer (IDEXX Veterinary (unless otherwise indicated). Data were plotted using GraphPad Diagnostics). Prism 7 software as mean values, with error bars representing SD. Bone Marrow Transplantation Assays Data Availability Freshly dissected femurs and tibias were isolated from Cd19- The data that support the findings of this study are available from cre+/−, Cd19-cre+/− Sf3b1K700E/+, Cd19-cre+/−Eμ-MycTg/+, and Cd19-cre+/− the corresponding author upon reasonable request. The RNA-seq Sf3b1K700E/+ Eμ-MycTg/+ CD45.2+ mice. Bone marrow (BM) was flushed data have been deposited in NCBI SRA under accession number with a 3-cc insulin syringe into cold PBS supplemented with 2% BSA GSE116391. to generate single-cell suspensions. BM cells were pelleted by centrifugation at 1,500 rpm for 4 minutes, and red blood cells were lysed in Disclosure of Potential Conflicts of Interest ammonium chloride–potassium bicarbonate lysis buffer for 3 minutes on ice. After centrifugation, cells were resuspended in PBS/2% A.R. Mato is an advisory board member of AbbVie, Genentech, J BSA, passed through a 40 μm cell strainer, and counted. For trans- and J, Pfizer, AstraZeneca, Loxo, Sunesis, and Pharmacyclics; is an plantation experiments, 1.0 × 106 BM cells from Cd19-cre+/−, Cd19- advisory board/DSMB of TG Therapeutics; is DSMB of Celgene; and cre+/− Sf3b1K700E/+, Cd19-cre+/−Eμ-MycTg/+, and Cd19-cre+/− Sf3b1K700E/+ reports receiving other commercial research support from AbbVie Eμ-MycTg/+ Cd45.2+ mice were transplanted via tail-vein injection into and Celgene. G. Fabbri is an employee of AstraZeneca and has stock 8-week-old sublethally irradiated (450 cGy) CD45.1+ recipient mice. ownership and/or stock options or interests in AstraZeneca. O. Abdel- For each bleeding, whole blood cell counts were measured on an Wahab is a consultant at H3 Biomedicine, Foundation Medicine, automated blood analyzer. Merck, and Janssen, and has a consultant/advisory board relationship with Envisagenics Inc. No potential conflicts of interest were disclosed FTY-720 Treatment In Vivo and Measurement by the other authors. of PP2A Activity FTY-720 (Sigma Aldrich) was dissolved in water and administered Authors’ Contributions intraperitoneally (3 mg/kg/day). The dose was optimized and deter- Conception and design: Z. Liu, A. Yoshimi, O. Abdel-Wahab, mined by multiple pilot tests to minimize weight loss (such that all R. Rabadan mice maintained ≥85% of their weight after 2 weeks of treatment). Development of methodology: Z. Liu, A. Yoshimi, R. Rabadan Protein lysates extracted from the spleen of treated mice were assayed Acquisition of data (provided animals, acquired and managed to evaluate the efficacy of FTY-720 to activate PP2A complexin vivo, patients, provided facilities, etc.): Z. Liu, A. Yoshimi, H. Cho, using the PP2A Immunoprecipitation Phosphatase Assay Kit as S.C.-W. Lee, M. Ki, L. Bitner, H. Shah, B. Liu, O. Abdel-Wahab directed by the manufacturer (Millipore Sigma). Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Z. Liu, A. Yoshimi, J. Wang, mRNA Stability Assays H. Cho, S.C.-W. Lee, T. Chu, A.R. Mato, P. Ruvolo, O. Abdel-Wahab, mRNA stability assays were performed as previously described R. Rabadan (42). Briefly, anti–UPF1 shRNA- or control shRNA lentivirus–infected Writing, review, and/or revision of the manuscript: Z. Liu, Nalm-6 SF3B1K700E/+ knock-in cells were generated using puromy- A. Yoshimi, H. Shah, A.R. Mato, P. Ruvolo, O. Abdel-Wahab, cin selection (1 μg/mL) for 7 days, and shRNAs against UPF1 were R. Rabadan expressed using doxycycline (2 μg/mL) for 2 days. GFP (shRNA)- Administrative, technical, or material support (i.e., reporting or positive cells were FACS sorted, treated with 2.5 μg/mL actinomycin organizing data, constructing databases): H. Cho, M. Ki, G. Fabbri, D (Life Technologies), and harvested at 0, 2, 4, 8, and 12 hours. L. Pasqualucci, O. Abdel-Wahab Study supervision: R. Rabadan Metabolic Labeling and Capture of Newly Synthesized Protein Acknowledgments This was done as previously described (43). Briefly, newly synthe- We thank Esther A. Obeng and Benjamin L. Ebert for sharing sized proteins were labeled using the Click-iT Protein Labeling Kit Sf3b1K700E conditional knock-in mice, and Goutham Narla, Michael (Invitrogen). Isogenic Nalm-6 cells were cultured in fresh medium for G. Kharas, and Piro Lito for technical advice. Z. Liu is supported by 24 hours, washed twice with PBS, and resuspended in methionine- NIH grant P01CA087497. A. Yoshimi is supported by grants from the free RPMI 1640 medium (Gibco) supplemented with 10% FCS for 30 Aplastic Anemia and MDS International Foundation (AA&MDSIF), minutes, at which point the methionine analog l-azidohomoalanine the Lauri Strauss Leukemia Foundation, and the Japanese Society (AHA; Invitrogen) was added (50 μmol/L) to allow incorporation of for the Promotion of Science (JSPS) Overseas Research Fellowships. AHA into nascent proteins. MG-132 (Sigma, M7449; 5 μmol/L) was A. Yoshimi and S.C.-W. Lee are supported by the Leukemia and Lym- added at 4 hours before the methionine replacement. Protein was phoma Society Special Fellow Award. S.C.-W. Lee is supported by the extracted from the cells, and 150 μg of total protein was used in the NIH/NCI (K99 CA218896) and the American Society of Hematology cross-linking of AHA-labeled nascent proteins to alkyne-derivatized (ASH) Scholar Award. O. Abdel-Wahab is supported by grants from biotin in Click-iT Protein Reaction Buffer (Invitrogen) according to NIH/NHLBI (R01 HL128239), the Department of Defense Bone the manufacturer’s instructions. Biotin–cross-linked nascent pro- Marrow Failure Research Program (W81XWH-16-1-0059), the Starr tein was captured overnight with streptavidin-coated Dynabeads Foundation (I8-A8-075), the Henry & Marilyn Taub Foundation, the (M-280, Invitrogen) and eluted. The whole volume of AHA-labeled, Edward P. Evans Foundation, the Leukemia and Lymphoma Society, biotin–cross-linked, streptavidin-precipitated protein was separated and the Pershing Square Sohn Cancer Research Alliance. R. Rabadan by SDS–PAGE. is supported by NIH grants (R01 CA185486, R01 CA179044, U54

JUNE 2020 CANCER DISCOVERY | OF14

RESEARCH ARTICLE Zhang et al.

CA193313, and U54 209997) and the NSF/SU2C/V-Foundation Ideas 18. Adams JM, Harris AW, Pinkert CA, Corcoran LM, Alexander WS, Cory Lab Multidisciplinary Team (PHY-1545805). S, et al. The c-myc oncogene driven by immunoglobulin enhancers induces lymphoid malignancy in transgenic mice. Nature 1985;318: The costs of publication of this article were defrayed in part by 533–8. the payment of page charges. This article must therefore be hereby 19. Yeh E, Cunningham M, Arnold H, Chasse D, Monteith T, Ivaldi G, marked advertisement in accordance with 18 U.S.C. Section 1734 et al. A signalling pathway controlling c-Myc degradation that impacts solely to indicate this fact. oncogenic transformation of human cells. Nat Cell Biol 2004;6: 308–18. Received November 14, 2019; revised February 15, 2020; accepted 20. Arnold HK, Sears RC. Protein phosphatase 2A regulatory subunit March 12, 2020; published first March 18, 2020. B56alpha associates with c-myc and negatively regulates c-myc accu- mulation. Mol Cell Biol 2006;26:2832–44. 21. Mori K, Maeda Y, Kitaura H, Taira T, Iguchi-Ariga SM, Ariga H. MM- 1, a novel c-Myc-associating protein that represses transcriptional References activity of c-Myc. J Biol Chem 1998;273:29794–800. . 1 Seiler M, Peng S, Agrawal AA, Palacino J, Teng T, Zhu P, et al. Somatic 22. Perrotti D, Neviani P. Protein phosphatase 2A: a target for anticancer mutational landscape of splicing factor genes and their functional therapy. Lancet Oncol 2013;14:e229–38. consequences across 33 cancer types. Cell Rep 2018;23:282–96. 23. Ruvolo PP, Clark W, Mumby M, Gao F, May WS. A functional role 2. Papaemmanuil E, Cazzola M, Boultwood J, Malcovati L, Vyas P, for the B56 alpha-subunit of protein phosphatase 2A in ceramide- Bowen D, et al. Somatic SF3B1 mutation in myelodysplasia with ring mediated regulation of Bcl2 phosphorylation status and function. J sideroblasts. N Engl J Med 2011;365:1384–95. Biol Chem 2002;277:22847–52. 3. Yoshida K, Sanada M, Shiraishi Y, Nowak D, Nagata Y, Yamamoto R, 24. Deng X, Gao F, Flagg T, May WS Jr. Mono- and multisite phospho- et al. Frequent pathway mutations of splicing machinery in myelod- rylation enhances Bcl2’s antiapoptotic function and inhibition of cell ysplasia. Nature 2011;478:64–9. cycle entry functions. Proc Natl Acad Sci U S A 2004;101:153–8. 4. Quesada V, Ramsay AJ, Lopez-Otin C. Chronic lymphocytic leukemia 25. Patel JH, McMahon SB. BCL2 is a downstream effector of MIZ-1 essen- with SF3B1 mutation. N Engl J Med 2012;366:2530. tial for blocking c-MYC-induced apoptosis. J Biol Chem 2007;282: 5. Wang L, Lawrence MS, Wan Y, Stojanov P, Sougnez C, Stevenson K, 5–13. et al. SF3B1 and other novel cancer genes in chronic lymphocytic 26. Wagner AJ, Small MB, Hay N. Myc-mediated apoptosis is blocked by leukemia. N Engl J Med 2011;365:2497–506. ectopic expression of Bcl-2. Mol Cell Biol 1993;13:2432–40. 6. Furney SJ, Pedersen M, Gentien D, Dumont AG, Rapinat A, Desjar- 27. Matsuoka Y, Nagahara Y, Ikekita M, Shinomiya T. A novel immuno- dins L, et al. SF3B1 mutations are associated with alternative splicing suppressive agent FTY720 induced Akt dephosphorylation in leuke- in uveal melanoma. Cancer Discov 2013;3:1122–9. mia cells. Br J Pharmacol 2003;138:1303–12. 7. Harbour JW, Roberson ED, Anbunathan H, Onken MD, Worley LA, 28. Brinkmann V, Davis MD, Heise CE, Albert R, Cottens S, Hof R, et al. Bowcock AM. Recurrent mutations at codon 625 of the splicing fac- The immune modulator FTY720 targets sphingosine 1-phosphate tor SF3B1 in uveal melanoma. Nat Genet 2013;45:133–5. receptors. J Biol Chem 2002;277:21453–7. 8. Martin M, Masshofer L, Temming P, Rahmann S, Metz C, Bornfeld 29. Chiba K. FTY720, a new class of immunomodulator, inhibits lym- N, et al. Exome sequencing identifies recurrent somatic mutations phocyte egress from secondary lymphoid tissues and thymus by in EIF1AX and SF3B1 in uveal melanoma with disomy 3. Nat Genet agonistic activity at sphingosine 1-phosphate receptors. Pharmacol 2013;45:933–6. Ther 2005;108:308–19. 9. Darman RB, Seiler M, Agrawal AA, Lim KH, Peng S, Aird D, et al. Can- 30. Kauko O, O’Connor CM, Kulesskiy E, Sangodkar J, Aakula A, Izad- cer-associated SF3B1 hotspot mutations induce cryptic 3′ splice site mehr S, et al. PP2A inhibition is a druggable MEK inhibitor resist- selection through use of a different branch point. Cell Rep 2015;13: ance mechanism in KRAS-mutant lung cancer cells. Sci Transl Med 1033–45. 2018;10:pii:eaaq1093. 10. DeBoever C, Ghia EM, Shepard PJ, Rassenti L, Barrett CL, Jepsen K, 31. Tohme R, Izadmehr S, Gandhe S, Tabaro G, Vallabhaneni S, Thomas et al. Transcriptome sequencing reveals potential mechanism of cryp- A, et al. Direct activation of PP2A for the treatment of tyrosine kinase tic 3′ splice site selection in SF3B1-mutated cancers. PLoS Comput inhibitor-resistant lung adenocarcinoma. JCI Insight 2019;4:pii: Biol 2015;11:e1004105. 125693. 11. Dvinge H, Kim E, Abdel-Wahab O, Bradley RK. RNA splicing factors 32. Gutierrez A, Pan L, Groen RW, Baleydier F, Kentsis A, Marineau J, as oncoproteins and tumour suppressors. Nat Rev Cancer 2016;16: et al. Phenothiazines induce PP2A-mediated apoptosis in T cell acute 413–30. lymphoblastic leukemia. J Clin Invest 2014;124:644–55. 12. Inoue D, Chew GL, Liu B, Michel BC, Pangallo J, D’Avino AR, et al. 33. Zhou Q, Derti A, Ruddy D, Rakiec D, Kao I, Lira M, et al. A chemi- Spliceosomal disruption of the non-canonical BAF complex in can- cal genetics approach for the functional assessment of novel cancer cer. Nature 2019;574:432–6. genes. Cancer Res 2015;75:1949–58. 13. Wang L, Brooks AN, Fan J, Wan Y, Gambe R, Li S, et al. Transcrip- 34. Sablina AA, Hector M, Colpaert N, Hahn WC. Identification of PP2A tomic characterization of SF3B1 mutation reveals its pleiotropic complexes and pathways involved in cell transformation. Cancer Res effects in chronic lymphocytic leukemia. Cancer Cell 2016;30:750–63. 2010;70:10474–84. 14. Lee SC, North K, Kim E, Jang E, Obeng E, Lu SX, et al. Synthetic lethal 35. Chen Y, Xu Y, Bao Q, Xing Y, Li Z, Lin Z, et al. Structural and bio- and convergent biological effects of cancer-associated spliceosomal chemical insights into the regulation of protein phosphatase 2A by gene mutations. Cancer Cell 2018;34:225–41. small t antigen of SV40. Nat Struct Mol Biol 2007;14:527–34. 15. Yoshimi A, Abdel-Wahab O. Splicing factor mutations in MDS RARS 36. Janghorban M, Langer EM, Wang X, Zachman D, Daniel CJ, Hooper and MDS/MPN-RS-T. Int J Hematol 2017;105:720–31. J, et al. The tumor suppressor phosphatase PP2A-B56alpha regulates 16. Lefebvre C, Rajbhandari P, Alvarez MJ, Bandaru P, Lim WK, Sato M, stemness and promotes the initiation of malignancies in a novel et al. A human B-cell interactome identifies MYB and FOXM1 as murine model. PLoS One 2017;12:e0188910. master regulators of proliferation in germinal centers. Mol Syst Biol 37. Cretu C, Schmitzova J, Ponce-Salvatierra A, Dybkov O, De Laurentiis 2010;6:377. EI, Sharma K, et al. Molecular architecture of SF3b and structural 17. Obeng EA, Chappell RJ, Seiler M, Chen MC, Campagna DR, Schmidt consequences of its cancer-related mutations. Mol Cell 2016;64: PJ, et al. Physiologic expression of Sf3b1(K700E) causes impaired 307–19. erythropoiesis, aberrant splicing, and sensitivity to therapeutic spli- 38. Saddoughi SA, Gencer S, Peterson YK, Ward KE, Mukhopadhyay A, ceosome modulation. Cancer Cell 2016;30:404–17. Oaks J, et al. Sphingosine analogue drug FTY720 targets I2PP2A/

OF15 | CANCER DISCOVERY JUNE 2020 AACRJournals.org

The Role of APOC2--CD36 Interaction in AML RESEARCH ARTICLE

SET and mediates lung tumour suppression via activation of PP2A- 41. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. RIPK1-dependent necroptosis. EMBO Mol Med 2013;5:105–21. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 2013;29: 39. Seiler M, Yoshimi A, Darman R, Chan B, Keaney G, Thomas M, et al. 15–21. H3B-8800, an orally available small-molecule splicing modulator, 42. Yoshimi A, Lin KT, Wiseman DH, Rahman MA, Pastore A, Wang B, induces lethality in spliceosome-mutant cancers. Nat Med 2018;24: et al. Coordinated alterations in RNA splicing and epigenetic regula- 497. tion drive leukaemogenesis. Nature 2019;574:273–7. 40. Pellagatti A, Armstrong RN, Steeples V, Sharma E, Repapi E, Singh S, 43. Vu LP, Prieto C, Amin EM, Chhangawala S, Krivtsov A, Calvo-Vidal et al. Impact of spliceosome mutations on RNA splicing in myelodys- MN, et al. Functional screen of MSI2 interactors identifies an essen- plasia: dysregulated genes/pathways and clinical associations. Blood tial role for SYNCRIP in myeloid leukemia stem cells. Nat Genet 2018;132:1225–40. 2017;49:866–75.

JUNE 2020 CANCER DISCOVERY | OF16

Mutations in the RNA Splicing Factor SF3B1 Promote Tumorigenesis through MYC Stabilization

Zhaoqi Liu, Akihide Yoshimi, Jiguang Wang, et al.

Cancer Discov Published OnlineFirst March 18, 2020.

Updated version Access the most recent version of this article at: doi:10.1158/2159-8290.CD-19-1330

Supplementary Access the most recent supplemental material at: Material http://cancerdiscovery.aacrjournals.org/content/suppl/2020/03/18/2159-8290.CD-19-1330.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerdiscovery.aacrjournals.org/content/early/2020/05/12/2159-8290.CD-19-1330. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.