Vitamin B5 and succinyl-CoA improve ineffective erythropoiesis in SF3B1-mutated myelodysplasia

Abstract

Patients with myelodysplastic syndrome and ring sideroblasts (MDS-RS) present with symptomatic anemia due to ineffective erythropoiesis that impedes their quality of life and increases morbidity. More than 80% of patients with MDS-RS harbor splicing factor 3B subunit 1 (SF3B1) mutations, the founder aberration driving MDS-RS disease. Here, we report how mis-splicing of coenzyme A synthase (COASY), induced by mutations in SF3B1, affects heme biosynthesis and erythropoiesis. Our data revealed that COASY was up-regulated during normal erythroid differentiation, and its silencing prevented the formation of erythroid colonies, impeded erythroid differentiation, and precluded heme accumulation. In patients with MDS-RS, loss of protein due to COASY mis-splicing led to depletion of both CoA and succinyl-CoA. Supplementation with COASY substrate (vitamin B5) rescued CoA and succinyl-CoA concentrations in SF3B1^mut cells and mended erythropoiesis differentiation defects in MDS-RS primary patient cells. Our findings reveal a key role of the COASY pathway in erythroid maturation and identify upstream and downstream metabolites of COASY as a potential treatment for anemia in patients with MDS-RS.

Figure 1. RNA splicing analysis is performed on SF3B1^mut haematopoietic progenitors derived from colony forming assay.

(A) Presence of SF3B1 mutations across cancers in TCGA and MSKCC (MDS) 2020 cohorts. (B) Schematic of the experimental strategy used to identify mis-splicing events associated with SF3B1 mutations. BM CD34⁺ cells from healthy donors (HD1, HD2, HD3) and 3 patients with MDS SF3B1^mut were seeded in methylcellulose to generate colonies under hypoxic or normoxic conditions. Colonies from each condition were harvested and used for RNA-sequencing. Splicing analysis with rMATS identified splicing events with a FDR<0.05. Splicing events were classified into 5 subcategories: skipped exon (SE), alternative 5’ splice site (A5SS), alternative 3’ splice site (A3SS), mutually exclusive exon (MXE), and retained intron (RI). (C) Clinical and SF3B1 mutations details of the 3 patients used for RNA sequencing analysis. (D) Total number of colonies derived from the BM CD34⁺ cells from healthy donors (n=3) and patients with MDS-RS (n=3) per 1000 CD34⁺ cells seeded. Colony forming cell (CFC) assays were performed under hypoxic (3% O₂) or normoxic conditions (20% O₂). Data are the mean± S.E.M. ** p<0.01, *** p<0.001. (E) Bar chart showing the number of BFU-E/G/M/GM colonies obtained from 3 healthy donors or 3 MDS-RS SF3B1^mut patients’ samples per 1000 CD34⁺ cells plated and cultured upon normoxia or hypoxia (n=3).

Figure 2. SF3B1 mutations induce global mis-splicing of downstream genes in bone marrow haematopoietic stem and progenitor cells of patients with MDS-RS.

(A) Pie chart representing the proportion of the mis-splicing events identified in the samples. (B) Heatmap of the differential splicing events captured between healthy donor colonies cultured under normoxic and hypoxic conditions. (C) Heatmap of the differential splicing events captured between healthy donor colonies and MDS SF3B1^mut colonies grown in normoxia (left panel) or hypoxia (right panel). (D) Volcano plot showing the genes that were differentially mis-spliced between MDS SF3B1^mut and HD samples in hypoxic and normoxic conditions with a p< 0.05 and Inclusion Level ≥ |0.2|. (E) Dotplot showing the top most significant Gene Ontology Biological Processes for each comparison (hypoxia MDS SF3B1^mut versus healthy donors, and normoxia MDS SF3B1^mut versus healthy donors) with adj. p< 0.05. (F) Venn diagram showing overlap between genes with significant splicing events identified in healthy donors and SF3B1^mut patients with MDS-RS. A total of 200 gene mis-splicing events that were present in patients with MDS SF3B1^mut but not in HDs were analyzed further. (G) GSEA analysis of the differentially expressed genes demonstrated defects in heme metabolism (top panel) and the citric acid cycle (TCA)/respiratory electron transport chain (bottom panel) in patients with MDS SF3B1^mut.

Figure 3. SF3B1 mutations result in mis-splicing of COASY gene in patients with MDS-RS.

(A) Exon structure of the different refseq annotated COASY isoforms. Area highlighted in blue depicts the mis-splicing target 5′ UTR region. The red arrows highlights the junction which is alternatively used in MDS SF3B1^mut cells. Sashimi plots underneath represent the read coverage in RPKM at exon-exon junction in COASY 5’UTR transcript for 2 patients with MDS SF3B1^mut (in red) and 1 healthy donor (in orange). Inclusion level (IncLevel) and numbers of reads covering the junction are displayed on the plot. (B) Representative gel image of COASY 5’UTR RT-PCR in MDS SF3B1^WT (n=3) and MDS SF3B1^mut (n=3) patients. (C) Quantitative PCR analysis of NM_001042532.4 isoform encoding COASY β in healthy donors (HD) (n=3), patients with MDS SF3B1^WT (n=5), and patients with MDS SF3B1^mut (n=17). Data presented here are the mean± S.E.M. ** p<0.01.

Figure 4. COASY transcript mis-splicing induces protein loss and reduction of CoA synthesis in SF3B1^mut cells.

(A) CRISPR-cas9 editing strategy used to insert SF3B1 heterozygous K700E mutation in K562 cell line. (B) Quantitative PCR analysis of NM_001042532.4 isoforms, encoding COASY β protein, in K562 SF3B1^wt and K562 SF3B1^mut clones (Clone1 and Clone2) (n=3). (C) Representative Western blot showing COASY protein abundance in K562 SF3B1^wt and K562 SF3B1^mut clones (C1 and C2). Actin B protein was used for all Western blots as a loading control. (D) Quantification of COASY protein expression in K562 SF3B1^wt and K562 SF3B1^mut clones (Clone1 and Clone2). Data here are normalized to Actin B (n=3). (E) Representative Western blot showing COASY protein expression in SF3B1^wt AML cell lines (OCI AML3 and THP-1) and SF3B1^mut AML cell lines (HNT-34). (F) Schematic representing the dual luciferase vector pDualLuc (top panel). Firefly and Renilla luciferases (FLuc and RLuc) are under the control of two independent promoters, hPGK and EF1A, respectively. The 5’UTR from the 4 different COASY isoforms (NM_001042529.3, NM_001042532.4, NM_025233.7, XM_011525300.2) were cloned upstream of FLuC. Rluc expression was used for normalisation. We quantified the impact on translation of the 5’UTR of each isoform by dual luciferase assay (n=3) (bottom panel). (G) LC-MS/MS analysis of phospho-pantothenate, CoA, succinyl-CoA, and glycine concentrations in K562 SF3B1^wt and K562 SF3B1^mut (N=5) (H) LC-MS/MS analysis of phosphopantothenate, CoA, succinyl-CoA, and glycine abundance in K562 SF3B1^wt and K562 SF3B1^mut. Metabolites were labelled by incubating cells with 0.25mg/L vitamin B5 stable isotope (13C6, 15N2) for 24h (N=5). (I) Schematic representation of the serine-glycine synthesis pathway. 3-phosphoglycerate from glycolysis is converted into 3-phospho-pyruvate by the phosphoglycerate dehydrogenase enzyme (PHGDH). Successively 3-phospho-pyruvate is converted into 3-phosphoserine, serine, and glycine. (J) LC-MS/MS analysis of serine abundance in K562 SF3B1^wt and K562 SF3B1^mut (N=5). (K) Representative Western blot showing PHGDH protein abundance in K562 SF3B1^wt and K562 SF3B1^mut (Clone1 and Clone2) (n=3). (L) Representative Western blot showing PHGDH protein abundance in SF3B1^wt AML cell lines (OCI AML3 and THP-1) and SF3B1^mut AML cell lines (HNT-34). Data presented are the mean± S.D. **** p<0.001, * p<0.05.

Figure 5. COASY deficiency impairs erythroid differentiation.

(A) COASY knock-down strategy used in healthy donor CD34⁺ umbilical cord blood cells. Cells were transduced with shRNA-tagged eGFP reporter against COASY (shCOASY) or a scrambled sequence (shControl) and expanded for 4 days before FACS sorting of CD34⁺/GFP⁺ cells. shCOASY and shControl CD34⁺ cells were seeded in methylcellulose to assess colony-forming potential or in liquid culture for erythroid differentiation. (B) Pictures of colonies obtained with shControl cells (left) and shCOASY transduced CD34⁺ cells (right), at day 14. Progenitor cells from colony-forming assay were harvested and pellets are shown as representative picture. (C) Total colonies per 500 CD34⁺ cells seeded in methylcellulose for shControl and shCOASY KD CD34⁺ cells (n=3), at day 14. (D) BFU-E and G/M/GM colonies per 500 CD34⁺ cells seeded in methylcellulose for control and shCOASY KD HSPCs cells (n=3), at day 14. (E) Representation of the different erythroid differentiation stages and associated surface markers acquired during the differentiation process. (F) Flow cytometry analysis of CD71/CD235a acquisition in shControl and shCOASY KD cells at day 4, 7, 10, and 14 of erythroid differentiation. (G) Quantitative PCR analysis of total COASY expression in CD34⁺ HSPCs cells at day 4, 7, 10, and 14 of erythroid differentiation (n=3). (H) Quantitative PCR analysis of total COASY expression in CD34⁺ HSPCs cells transduced with shControl or shCOASY1 at day 4, 7, 10, and 14 of erythroid differentiation (n=3). (I) Percentage of cells in quadrants I, II, III and IV, as depicted in (F), were quantified for shControl and shCOASY KD HSPCs cells at day 14 of erythroid differentiation (n=3). (J) Heme quantification for shControl and shCOASY KD HSPCs cells at day 14 of erythroid differentiation (n=3). (K) Representative image for quantification of erythroid cell morphology using Giemsa staining of CD34⁺ HSPCs cells after erythroid differentiation, at day 10. Scale bar, 20μm. (L) Quantification of types of erythroid cells based on morphological features using Giemsa staining of shControl or shCOASY CD34⁺ HSPCs cells following erythroid differentiation, at day 10. Basophilic/polychromatic and orthochromatic erythroblasts were counted from 10 representative fields for each replicate (n=3). Data are the mean ±S.E.M. ****p<0.001, ***p<0.005, **p<0.01, *p<0.05.

Figure 6. Vitamin B5 and succinyl-CoA rescue the erythroid differentiation defect observed in MDS-RS SF3B1^mut cells.

(A) Schematics of the strategy used to rescue erythroid differentiation defect in MDS SF3B1^mut cells. CD34⁺ cells were isolated from MDS SF3B1^mut bone marrow MNCs and cultured in erythroid differentiation medium, supplemented with/without additional vitamin B5 or succinyl-CoA, for 14 days and then analyzed by flow cytometry. (B) Flow cytometry analysis of CD71/CD235a acquisition in BM CD34⁺ from adult healthy donor and MDS SF3B1^mut cells from patients, treated with/without additional vitamin B5 or succinyl-CoA, at day 14 of erythroid differentiation. (C) Percentage of cells during erythroid differentiation. Data are represented as the ratio [(quadrant III + quadrant IV) / (quadrants II + quadrants I)] of cells, as quadrants are depicted in (B), and quantified for MDS SF3B1^mut cells from patients, treated with/without vitamin B5 or succinyl-CoA, at day 14 of erythroid differentiation. 4 independent patients’ samples (MDS-RS #4, #5, #6, #7) were analyzed with a minimum of 3 replicates for each condition. Data presented here is the mean± S.E.M. *p<0.05. (D) Relative heme content, expressed as fold change, per 200,000 MDS SF3B1^mut cells treated with vitamin B5 or succinyl-CoA, compared to untreated at day 14 of erythroid differentiation (n=2).