Integrative RNA-omics Discovers GNAS Alternative Splicing as a Phenotypic Driver of Splicing Factor-Mutant Neoplasms - PubMed (original) (raw)

. 2022 Mar 1;12(3):836-855.

doi: 10.1158/2159-8290.CD-21-0508.

Emily C Wheeler # 1 2 3, Shailee Vora # 4 5 6 7, Andriana G Kotini 4 5 6 7, Malgorzata Olszewska 4 5 6 7, Samuel S Park 1, Ernesto Guccione 4 5 6, Julie Teruya-Feldstein 9, Lewis Silverman 5 7, Roger K Sunahara 8, Gene W Yeo 1 2 3, Eirini P Papapetrou 4 5 6 7

Affiliations

Integrative RNA-omics Discovers GNAS Alternative Splicing as a Phenotypic Driver of Splicing Factor-Mutant Neoplasms

Emily C Wheeler et al. Cancer Discov. 2022.

Abstract

Mutations in splicing factors (SF) are the predominant class of mutations in myelodysplastic syndrome (MDS), but convergent downstream disease drivers remain elusive. To identify common direct targets of missplicing by mutant U2AF1 and SRSF2, we performed RNA sequencing and enhanced version of the cross-linking and immunoprecipitation assay in human hematopoietic stem/progenitor cells derived from isogenic induced pluripotent stem cell (iPSC) models. Integrative analyses of alternative splicing and differential binding converged on a long isoform of GNAS (GNAS-L), promoted by both mutant factors. MDS population genetics, functional and biochemical analyses support that GNAS-L is a driver of MDS and encodes a hyperactive long form of the stimulatory G protein alpha subunit, Gαs-L, that activates ERK/MAPK signaling. SF-mutant MDS cells have activated ERK signaling and consequently are sensitive to MEK inhibitors. Our findings highlight an unexpected and unifying mechanism by which SRSF2 and U2AF1 mutations drive oncogenesis with potential therapeutic implications for MDS and other SF-mutant neoplasms.

Significance: SF mutations are disease-defining in MDS, but their critical effectors remain unknown. We discover the first direct target of convergent missplicing by mutant U2AF1 and SRSF2, a long GNAS isoform, which activates G protein and ERK/MAPK signaling, thereby driving MDS and rendering mutant cells sensitive to MEK inhibition. This article is highlighted in the In This Issue feature, p. 587.

©2021 American Association for Cancer Research.

PubMed Disclaimer

Conflict of interest statement

Conflict of interest statement

GWY is co-founder, member of the Board of Directors, on the SAB, equity holder, and paid consultant for Locanabio and Eclipse BioInnovations. GWY is a visiting professor at the National University of Singapore. GWY’s interest(s) have been reviewed and approved by the University of California, San Diego in accordance with its conflict of interest policies. EPP has received honoraria from Celgene and Merck and research support from Incyte. JTF is on the advisory board of Astellas and Blueprint Medicines and a consultant for Curio Science, Histowitz and Scopio Labs. The authors declare no other competing financial interests.

Figures

Figure 1:

Figure 1:. SF-mutant iPSC-HSPC models recapitulate cellular phenotypes and motif preferences of SF-mutant MDS.

A. Schematic overview of the generation of isogenic clonal iPSC lines with the canonical U2AF1 S34F and SRSF2 P95L mutations and allele-specific epitope tags through CRISPR/Cas9-mediated gene editing. B. Percentage of mutant allele of total U2AF1 or SRSF2 transcripts (from RNA-Seq data) confirming heterozygous state and showing equal expression of the mutant and WT alleles. Mean and SEM of 3-4 replicates is shown (Supplementary Table S1). C. Number of colonies from 5,000 cells seeded in methylcellulose assays on day 14 of hematopoietic differentiation of the indicated iPSC lines. Mean of 1-6 independent differentiation experiments for each line is shown. D. Fraction of CD15+, CD14+, and CD16+ myeloid cells on days 12 and 14 of hematopoietic differentiation of WT and SF-mutant iPSC lines (WT-1, S34F-6, P95L-2). Mean and SEM of 2-4 independent differentiation experiments are shown. E. Competitive growth assay. The cells were mixed 1:1 at the onset of hematopoietic differentiation with an isogenic WT iPSC line stably expressing GFP (derived from the parental line WT-1). The relative population size was estimated as the percentage of GFP- cells (measured by flow cytometry) at each time point (days 4-12 of differentiation), relative to the population size on day 2. Results from 1 or 2 independent experiments per line are shown. F. iPSC lines were differentiated along the hematopoietic lineage and CD34+/CD45+ HSPCs were sorted for RNA-Seq and allele-specific eCLIP analyses. G. Number of alternative splicing (AS) events detected for each genotype in comparison to WT cells (FDR < 0.05, delta PSI > 5%). SE: skipped exon; A3SS: alternative 3’ splice site; A5SS: alternative 5’ splice site; RI: retained intron. H. Sequence logos of 3’ splice sites flanking skipped exon events and downstream exons. “WT exons” and “S34F exons” denote exons skipped or included, respectively, in U2AF1-mutant cells. I. Count of skipped exon events containing the indicated 3’ splice site sequence. “WT exons” and “S34F exons” denote exons skipped or included, respectively, in U2AF1-mutant cells. “Unchanged” denotes a randomly selected control set of exons that are not differentially spliced in U2AF1-mutant cells. J. HOMER motif enrichment of sequence preferences within skipped exons and downstream control exons. “WT exons” and “P95L exons” denote exons skipped or included, respectively, in SRSF2-mutant cells. P-values were calculated with a cumulative binomial distribution. K. Individual 5-mer sequences with most significant differential enrichment in “WT exons” and “P95L exons”. p < 0.005 by chi-squared test with Bonferroni-correction for all 5-mers plotted.

Figure 2:

Figure 2:. Allele-specific eCLIP identifies direct targets of mis-splicing in human HSPCs from isogenic iPSC models of SRSF2- and U2AF1- mutant MDS.

A. Count of significantly enriched binding sites (eCLIP peaks) identified in each genic region indicated. Significantly enriched peaks are peaks with fold change > 4 relative to input and p-value < 0.001 (chi-squared test) in at least one of two replicate experiments. B. Top enriched motifs from all peaks identified per genotype. P-values are reported to the right of each motif (cumulative binomial distribution). C,D. Relative abundances of 6-mer sequences counted in all eCLIP peak regions. Differentially used 6-mers were identified with a chi-squared test and FDR < 0.05 and are colored according to the 6-mer category defined in the legend. 6-mers that are not differentially used are plotted in light grey. E,F. Splicing maps of eCLIP binding density in regions flanking skipped exons that are differentially used between WT and U2AF1-mutant (E) or SRSF2 mutant (F) cells. A random background set of exons that are unchanged between WT and mutant cells was used as control. Binding density is averaged across regions (50 bp on each end of exon sequences and 300 bp into surrounding introns) and normalized to the input sample. Data from one of two replicate samples are shown. (The other replicate is shown in Supplementary Fig. S8A and S8B.) AU: arbitrary units

Figure 3:

Figure 3:. Intersectional analysis of mis-splicing and mRNA binding reveals a long isoform of GNAS (GNAS-L) as a direct convergent effect of both mutant SFs.

A. Upper panel: schematic of strategy for identification of skipped exon events with differential peaks in regulatory regions (upstream intron for U2AF1 and cassette exon for SRSF2). Skipped exons are categorized in 3 groups: “WT peak” include peaks bound by the WT SF only; “Mutant peak” include peaks bound by the mutant SF only; and “Both peak” include sites bound by both WT and mutant SF. Lower panels: Delta PSI (WT–mutant) of skipped exon events divided among the 3 binding categories. A chi-square test was used to determine dependence of “WT peak” or “Mutant peak” status on the direction of exon inclusion (positive Delta PSI in “WT peak” and negative Delta PSI in “Mutant peak”). B. Workflow showing the integration of splicing analyses in both iPSC-HSPC genotypes with datasets from _U2AF1_- and _SRSF2_- mutant MDS patient cells and with eCLIP analyses of differentially bound exons. 41 splicing events, affecting 40 genes, were common to both _U2AF1_- and _SRSF2_- mutant iPSC-HSPCs. Of those, 20 had sufficient read coverage to allow evaluation of splicing in publicly available data from MDS patients (Pellagatti et al. 2018(12)). 15 of the 20 genes were differentially spliced in the same direction in both _U2AF1_- and _SRSF2_- mutant cells, compared to cells from MDS patients without SF mutations (SF-WT). Of those 15, the 3 genes indicated (GNAS, ITGB3BP and PSMA4) also contained a differential eCLIP peak between mutant and WT factor. C. Delta PSI of the 20 AS events common to both iPSC-HSPC genotypes that could be evaluated in MDS patient cells (sufficient coverage). Black dots represent AS events trending in the same direction in both genotypes vs SF-WT in the MDS patient cells, as well as in our iPSC-HSPC models of the two mutations. Blue circles show events that also contain a differential eCLIP peak in the regulatory region (as determined in a). D. Visualization of G/A and G/C-rich sequence motifs in exons 2–4 of the GNAS transcript. Only exon 3, which is more included in SRSF2 P95L cells, contains a G/C-rich sequence motif. E. eCLIP binding density at exons 2 - 5 of the GNAS transcript showing normalized read density for one replicate of WT U2AF1 (green) and U2AF1-S34F (blue) and sashimi plots showing inclusion levels of exon 3 in WT (green), S34F (blue) and P95L (orange) iPSC-HSPCs. Read counts are from one representative sample of each genotype and Ψ (Percent spliced in) is the average of all biological replicates. An eCLIP peak at exon 3 can be seen in the U2AF1-S34F but not the WT U2AF1 track. (There were no peaks for either WT or mutant SRSF2 called in this exon due to insufficient coverage. The SRSF2 eCLIP read density from this region is shown in Supplementary Fig. S9E.) Ψ: Percent spliced in. F, G. PSI (percent spliced in) of GNAS exon 3 in iPSC-HSPCs (F) or MDS patient cells (from Pellagatti et al. 2018(12)) (G). Data points represent independent lines (F) or patients (G). Boxes represent the IQR (25th, 50th and 75th percentiles) and whiskers represent 1.5 times the IQR from the 25th and 75th percentiles. ***p<0.005, *p<0.05, n.s.: not significant (Wilcoxon rank-sum test). H. Quantification of GNAS isoforms by RT-PCR in WT and SF-mutant iPSC-HSPCs, as indicated. Upper: schematic of the two isoforms. Middle: Ethidium bromide staining showing the two isoforms. Lower: Quantitation by image quantitation software of the two isoforms from the gel shown in the middle panel. I. Immunoblot showing the long and short forms of Gαs in WT and SF-mutant iPSC-HSPCs.

Figure 4:

Figure 4:. GNAS-L is a phenotypic MDS driver and encodes a hyperactive Gαs protein.

A. SF-mutant iPSC-HSPCs were transduced with a lentiviral vector encoding either shRNA specifically targeting GNAS-L exon 3 or a scrambled shRNA. GNAS isoform expression was quantified by RT-PCR 48 hours later. One representative experiment out of 3 is shown. The upper and lower bands correspond to GNAS-L and GNAS-S, respectively. The bars show quantification of the respective bands. B. Number of methylcellulose colonies from 5,000 SF-mutant iPSC-HSPCs transduced with GNAS-L shRNA or scrambled shRNA. Mean from 6 (WT), 3 (GNAS-L shRNA with two different shRNAs) and 2 (Scrambled shRNA) experiments is shown. C. Methylcellulose colonies from WT iPSC-HSPCs transduced with GNAS-L or GNAS-S. Mean and SEM of 3 different WT lines (WT-1, WT-2, WT-3) is shown. D. Crystal structure of Gαs bound to GTPγS (PDB: 1AZT). Demarcated is the disordered region (dotted line) encoded by exon 3. E. Time course of [35S]GTPγS binding to Gαs-S and Gαs-L forms with the R201C mutation, the R201H mutation or no mutation. Mean and SEM of 3 experiments is shown. (See also Supplementary Table S4) F. Real-time measurement of BodipyGTPγS-FL binding to R201 mutant or WT, Gαs-S or Gαs-L forms, as indicated. Mean and SEM from 3 experiments is shown. AU: arbitrary units. (See also Supplementary Table S4) G. Dose-response of cAMP accumulation under isoproterenol stimulation in HEK293 cells lacking endogenous GNAS(29) and stably expressing a cAMP sensor transfected with the indicated constructs. FU: fluorescence units. Data are mean and SEM of 6-7 independent transfections. H. Basal and maximal adenylyl cyclase activity (cAMP accumulation) estimated by non-linear curve fit of the dose-response curves shown in G. Mean and SEM of 6-7 independent transfections is shown. *p<0.05, **p<0.005, ***p<0.0001 (See also Supplementary Table S5)

Figure 5:

Figure 5:. GNAS-L is non-redundant to the R201 mutant form.

A. Mutational co-occurrence in combined MDS and AML patient cohorts from Papaemmanuil et al. 2013(4); Papaemmanuil et al. 2016(30) and Haferlach et al. 2014(31). p value was calculated with a Fisher’s exact test. OR: odds ratio. B. PKA substrate phosphorylation in K562 cells transduced with GNAS-L or GNAS-S with or without the R201H mutation, as indicated (EV: empty vector; S-WT: Short WT; L-WT: Long WT; S-R201H: Short Mutant; L-R201H: Long Mutant). (Note that multiple bands are visible and expected in these blots because a phospho-PKA substrate antibody is used.) C. PKA substrate phosphorylation in CB CD34+ cells transduced with the same vectors as in B. D. PKA substrate phosphorylation in WT (WT-1) and SF-mutant (P95L-4 and S34F-5) sorted CD34+/CD45+ iPSC-HSPCs. E. PKA substrate phosphorylation in primary cells from AML and MDS patients with SRSF2 mutation with or without GNAS R201H mutation.

Figure 6:

Figure 6:. GNAS-L drives MDS through ERK/MAPK pathway activation.

A, B. ERK, MEK and AKT phosphorylation in WT iPSC-HSPCs (WT-1) (A) and CB CD34+ cells (B) transduced with lentiviral vectors expressing GNAS-L or GNAS-S with or without the R201H mutation, as indicated (S-WT: Short WT; L-WT: Long WT; S-R201H: Short Mutant; L-R201H: Long Mutant). mCherry is co-expressed with GNAS in all lentiviral constructs through a P2A peptide. The long and short forms of Gαs (Gαs-L, Gαs-S) can also be seen. C. ERK, MEK, CRAF phosphorylation and total DUSP6 in WT and SF-Mutant iPSC-HSPCs (lines WT-1, S34F-6, and P95L-2). D. ERK and CRAF phosphorylation and total DUSP6 in WT iPSC-HSPCs transduced with an empty vector (EV) or a lentiviral vector expressing GNAS-L. E. ERK phosphorylation in SRSF2 P95L and U2AF1 S34F iPSC-HSPCs transduced with lentiviral vectors expressing a GNAS-L shRNA or scrambled shRNA. F. ERK phosphorylation and Gαs isoform expression at the protein level in mononuclear cells from three MDS patients with U2AF1 mutations (S34Y, Q157P and S34F in patients U2AF1-MUT 1, U2AF1-MUT 2 and U2AF1-MUT 3, respectively, see also Supplementary Table S6), two MDS patients with SRSF2 mutations (P95H and P95R in patients SRSF2-MUT 3 and SRSF2-MUT 6, respectively, see also Supplementary Table S6) and 3 MDS patients without any SF mutations (SF-WT 1-3, see also Supplementary Table S6), as indicated.

Figure 7:

Figure 7:. SRSF2- and _U2AF1_- mutant cells are sensitive to MEK inhibition.

A. Representative IC50 curves of treatment of SRSF2 P95L iPSC-HSPCs (P95L-2) with the MEK inhibitors trametinib, cobimetinib, selumetinib and CH5126766 (or the mutant-BRAF inhibitor vemurafenib as negative control) for 3 days. B. IC50 values of SRSF2 P95L (P95L-2) and U2AF1 S34F (S34F-6) iPSC-HSPCs treated with the MEK inhibitors trametinib, cobimetinib, selumetinib and CH5126766 (and the mutant-BRAF inhibitor vemurafenib as negative control) for 3 days. Mean and SEM of 3-6 replicates for each MEK inhibitor from independent differentiation experiments is shown. C. IC50 values of WT (WT-1), SRSF2 P95L (P95L-2) and U2AF1 S34F (S34F-6) iPSC-HSPCs treated with the indicated MEK inhibitors (and the mutant-BRAF inhibitor vemurafenib as negative control) for 3 days. Mean and SEM of 3-8 replicates for each MEK inhibitor from independent differentiation experiments is shown. D. IC50 curves from primary SF-WT or SRSF2-mutant mononuclear cells from MDS and sAML patients (see Supplementary Table S6 for details) treated with the indicated MEK inhibitors.

References

    1. Graubert TA, Shen D, Ding L, Okeyo-Owuor T, Lunn CL, Shao J, et al. Recurrent mutations in the U2AF1 splicing factor in myelodysplastic syndromes. Nat Genet. 2011. Dec 11;44(1):53–7. -PMC -PubMed
    1. Papaemmanuil E, Cazzola M, Boultwood J, Malcovati L, Vyas P, Bowen D, et al. Somatic SF3B1 mutation in myelodysplasia with ring sideroblasts. N Engl J Med. 2011. Oct 13;365(15):1384–95. -PMC -PubMed
    1. Yoshida K, Sanada M, Shiraishi Y, Nowak D, Nagata Y, Yamamoto R, et al. Frequent pathway mutations of splicing machinery in myelodysplasia. Nature. 2011. Oct 6;478(7367):64–9. -PubMed
    1. Papaemmanuil E, Gerstung M, Malcovati L, Tauro S, Gundem G, Van Loo P, et al. Clinical and biological implications of driver mutations in myelodysplastic syndromes. Blood. 2013. Nov 21;122(22):3616–27; quiz 99. -PMC -PubMed
    1. Kim E, Ilagan JO, Liang Y, Daubner GM, Lee SC, Ramakrishnan A, et al. SRSF2 Mutations Contribute to Myelodysplasia by Mutant-Specific Effects on Exon Recognition. Cancer Cell. 2015. May 11;27(5):617–30. -PMC -PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources