The myelodysplastic syndromes (MDS) are common myeloid 64 malignancies characterized by
 ineffective hematopoiesis and blood cytopenias, with patients showing increasing bone
 marrow blasts with disease progression [1]. Mutations in genes involved in pre-mRNA
splicing (SF3B1, SRSF2, U2AF1 and ZRSR2) are the most common mutations found in MDS,
 occurring in over 50% of all cases [2-4]. There is evidence that some spliceosome
 components play a role in the maintenance of genomic stability [5]. Splicing is a transcription
 coupled process; splicing factor mutations affect transcription and may lead to the
 accumulation of R-loops (RNA-DNA hybrids with a displaced single stranded DNA) [6].
 Mutations in the splicing factors SRSF2 and U2AF1 have been recently shown to increase R73
loop formation in leukemia cell lines, resulting in increased DNA damage, replication stress
 and activation of the ATR-Chk1 pathway [7,8]. SF3B1 is the most frequently mutated
 splicing factor gene in MDS, with mutations occurring in 25-30% of MDS patients [9-11].
 SF3B1 mutations are also found at lower frequency in other hematological malignancies
 [4,11] and in some individuals with clonal hematopoiesis of indeterminate potential [12]. A
 role of SF3B1 mutations in R-loop accumulation and DNA damage has not yet been reported
 in hematopoietic cells. Here, we investigated the effects of SF3B1 mutations on R-loop
 formation and associated DNA damage response in MDS and leukemia cells, and we also
 explored potential therapeutic implications.
 Firstly, we investigated the effects of SF3B1 mutations on the formation of R-loops, as
 measured by immunofluorescence staining using the S9.6 antibody (Supplementary Materials
 and Methods) [7]. K562 cells (a myeloid leukemia cell line) with the SF3B1K700E mutation
 showed a significant increase in the number of S9.6 foci, indicating increased R-loops,
 compared to isogenic SF3B1K700K K562 cells (Fig. 1A, S1A). We then analyzed induced
 pluripotent stem cells (iPSCs) that we generated (and characterized, Fig. S2) from bone
 marrow CD34+ cells of one SF3B1 mutant MDS patient and of one healthy control
(Supplementary Materials and Methods). A significant 89 increase in R-loops was observed in
 an iPSC clone harboring the SF3B1 mutation compared to another iPSC clone without the
 SF3B1 mutation obtained from same MDS patient, and to iPSCs from the healthy control
 (Fig. 1B). We have also analyzed the levels of R-loops in bone marrow CD34+ cells from
 three SF3B1 mutant MDS patients, three splicing factor wildtype MDS patients and three
 healthy controls. Importantly, CD34+ cells from SF3B1 mutant MDS patients (Table S1)
showed a significant and marked increase in R-loops compared to CD34+ cells from MDS
 patients without splicing factor mutations (2.4-fold) and from healthy controls (2.6-fold) (Fig.
1C, S1B). Our results demonstrate that an accumulation of R-loops occurs in association with
 the presence of SF3B1 mutations in MDS and leukemia cells.
 We then investigated the effects of SF3B1 mutations on the DNA damage response, as
 measured by immunofluorescence staining using anti-γ-H2AX antibody (Supplementary
 Materials and Methods) [7]. K562 cells with the SF3B1K700E mutation showed a significant
 increase in the number of γ-H2AX foci, indicating increased DNA damage, compared to
 isogenic control SF3B1K700K K562 cells (Fig. 1D, S3A). Similarly, an iPSC clone harboring
 the SF3B1 mutation showed increased DNA damage as compared to another iPSC clone
 without the SF3B1 mutation obtained from same MDS patient and to iPSCs from a healthy
 control (Fig. 1E). Bone marrow CD34+ cells from SF3B1 mutant MDS patients showed
 significantly increased DNA damage compared to CD34+ cells from MDS patients without
 splicing factor mutations and from healthy controls (Fig. 1F, S3B).
 To investigate whether the observed DNA damage in SF3B1 mutant K562 cells results
 from induced R-loops, we overexpressed RNaseH1 (encoding an enzyme that degrades the
 RNA in RNA:DNA hybrids) to resolve R-loops in these cells. RNaseH1 overexpression
 significantly reduced the number of S9.6 foci in SF3B1K700E K562 cells compared to
SF3B1K700E K562 cells expressing an empty vector (Fig. 1G, 1H). Furthermore, SF3B1K700E
K562 cells expressing WKKD m 114 utant RNaseH1 (lacking hybrid binding and RNaseH1
 activity) did not show a decrease in the number of S9.6 foci (Fig. 1G, 1H). RNaseH1
 overexpression also significantly reduced the number of γ-H2AX foci (Fig. 1I, 1J) in
 SF3B1K700E K562 cells compared to SF3B1K700E K562 cells expressing an empty vector.
 Further, western blot analysis of γ-H2AX levels in SF3B1K700E K562 cells expressing
 RNaseH1 also showed decreased levels of γ-H2AX (Fig. S4). These data demonstrate that
 increased levels of R-loops result in increased DNA damage in SF3B1 mutant leukemia cells.
Next, to investigate the signaling events related to DNA damage in SF3B1 mutant cells,
 we have studied ATR and ATM signaling, two pathways that are frequently activated
 following DNA damage [13]. The ATR signaling pathway was analyzed by measuring the
 levels of Chk1 phosphorylation at Ser345, a hallmark of activation of the ATR pathway, in
 K562 cells (Supplementary Materials and Methods). We observed increased phosphorylation
 of Chk1 in K562 cells with the SF3B1K700E mutation compared to isogenic SF3B1K700K K562
 cells (Fig. S5A). Suppression of R-loops by RNaseH1 overexpression resulted in decreased
 Chk1 phosphorylation, indicating suppression of ATR pathway activation in SF3B1K700E
 K562 cells (Fig. S5A). In contrast, we did not observe activation of the ATM signaling
 pathway in the SF3B1K700E K562 cells as analyzed by measuring the levels of ATM
 phosphorylation at Ser1981, Chk2 phosphorylation at Thr68 and RPA32 phosphorylation at
 Ser4/8 (Fig. S5B). These data demonstrate that SF3B1 mutations are associated with
 downstream activation of ATR but not ATM signaling.
 We sought to explore the functional importance of ATR activation associated with SF3B1
 mutation and determine whether this could represent a therapeutic vulnerability. We
 evaluated the sensitivity of SF3B1 mutant cells to VE-821 (Supplementary Materials and
 Methods). SF3B1K700E K562 cells showed preferential sensitivity to the ATR inhibitor VE-
 821 compared to isogenic SF3B1K700K K562 cells (Fig. S6A). Chk1 is a critical substrate of
ATR, and we next chose to investigate the effects of Ch 139 k1 inhibition in SF3B1K700E K562
 cells. Interestingly, SF3B1K700E K562 cells demonstrated preferential sensitivity to the Chk1
 inhibitor UCN-01 compared to SF3B1K700K K562 cells, suggesting that ATR-Chk1 activation
 is important for the survival of SF3B1 mutant cells (Fig. 2A). The effect of RNaseH1
 overexpression on the sensitivity of SF3B1K700E K562 cells towards UCN-01 was also
 examined. We found that the sensitivity of SF3B1K700E K562 towards UCN-01 decreased
 after overexpressing RNaseH1 (Fig. S6B). Treatment with an ATM inhibitor (KU-55933) did
 not show a significant difference in the sensitivity of SF3B1K700E K562 cells compared to
 isogenic SF3B1K700K K562 cells (Fig. S6C). Notably, bone marrow CD34+ cells from SF3B1
 mutant MDS patients showed preferential sensitivity towards UCN-01 (Fig. 2B) and VE-821
 (Fig. 2C) compared to CD34+ cells from MDS patients without splicing factor mutations and
 from healthy controls. These results show that activation of ATR, but not ATM, plays an
 important role for the survival of SF3B1 mutant cells, and that these cells are vulnerable to
 Chk1 inhibition.
 Preferential sensitivity of splicing factor mutant cells towards splicing modulators has
 been reported previously [4,14,15]. Thus we investigated whether a splicing modulator could
 increase the sensitivity of SF3B1 mutant cells to ATR or Chk1 inhibition. The splicing
 modulator Sudemycin D6 has been shown to preferentially kill U2AF1 mutant cells [14], but
 its effects on myeloid leukemia cells with the SF3B1 mutation have not been evaluated. Here
 we showed that SF3B1K700E K562 cells were preferentially sensitive to Sudemycin D6
 compared to isogenic SF3B1K700K K562 cells (Fig. 2D). Bone marrow CD34+ cells from
 SF3B1 mutant MDS patients showed preferential sensitivity to Sudemycin D6 compared to
CD34+ cells from MDS patients without splicing factor mutations and from healthy controls
 (Fig. 2E). We then tested the effects of Sudemycin D6 in combination with an ATR inhibitor
 or a Chk1 inhibitor. The effects of VE-821 and UCN-01 on SF3B1 mutant K562 cells were
enhanced by Sudemycin D6 (Fig. 164 2F, 2G). We have also determined the synergy scores of
 Sudemycin D6 and UCN-01 (Fig. S7A-B), and Sudemycin D6 and VE-821 (Fig. S7C-D) on
 SF3B1K700K and SF3B1K700E K562 cells. Various dose combinations showed a positive
 synergy score (δ-score), indicating synergy of Sudemycin D6 with VE-821 and UCN-01,
 with higher scores for SF3B1K700E K562 cells. Importantly, bone marrow CD34+ cells from
 SF3B1 mutant MDS patients also showed preferential sensitivity towards the combination of
 Sudemycin D6 with UCN-01 (Fig. 2H) or with VE-821 (Fig. 2I).
 In summary, we show for the first time that mutations of SF3B1, the most commonly
 mutated splicing factor gene in MDS [2,3,9,11], lead to the accumulation of R-loops and
 associated DNA damage, resulting in activation of the ATR pathway in MDS and leukemia
 cells. The suppression of R-loops rescued cellular defects including DNA damage and ATR175
Chk1 activation. Our current study on mutant SF3B1, and previous studies by others on
 mutant U2AF1 and SRSF2 [7,8], together demonstrate that different mutated splicing factors
 in hematopoietic cells all have convergent effects on R-loop elevation leading to DNA
 damage. It is possible that this R-loop induced DNA damage may give rise to deleterious
 mutations in MDS hematopoietic stem and progenitor cells, contributing to the clonal
 advantage of splicing factor mutant cells in human bone marrow. Future studies seeking to
 compare R-loop levels in the CD34+ cells of SF3B1 mutant MDS cases with those in CD34+
 cells of MDS patients with mutations of other splicing factors (SRSF2, U2AF1, ZRSR2) are
 warranted.
 This is the first study showing that splicing factor mutant MDS and leukemia cells are
 preferentially sensitive to the Chk1 inhibitor UCN-01, suggesting that Chk1 inhibition, alone
 or in combination with splicing modulators, may represent a novel therapeutic strategy to
 target splicing factor mutant cells. This strategy could also be potentially extended to
 therapeutically target other types of cancers known to harbour SF3B1 mutations. This study
provides preclinical evidence that MDS 189 patients with spliceosome mutations may benefit
from Chk1 inhibition to exploit the R-loop-associated vulnerability induced by these
 mutations.