Targeting aberrant splicing in myelodysplastic syndromes: biologic rationale and clinical opportunity

Introduction

Myelodysplastic syndromes (MDS) are cancers of the bone marrow with diverse clinical presentations, collectively characterized by clonally restricted hematopoiesis, cytopenias, dysplastic cell morphology, functional blood cell defects, and a risk of clonal evolution including progression to acute leukemia.^1-3 Patients with MDS face complications related to or exacerbated by bone marrow failure and have a high symptom burden,^4,5 and higher-risk forms of MDS have a poor prognosis with a median survival measured ranging from months to a few years.^6,7 Treatment of lower-risk MDS is often focused on alleviating symptoms related to ineffective clonal hematopoiesis – most commonly centering on amelioration of anemia – while in higher-risk MDS, therapy is usually intended to alter the course of disease and prolong survival.⁸ Currently, therapeutic options for MDS are limited, and no new therapies have been approved by regulatory authorities for MDS-related indications for nearly 15 years,⁹ although that may change soon.

While almost all patients with MDS have somatic mutations in hematopoietic cells (mostly in genes encoding epigenetic modifiers, spliceosome components, and transcription factors) to date, no available therapies are particularly specific for a given molecularly defined subset of MDS. For instance, while the majority of patients with MDS with del(5q) are likely to have improvements to their transfusion burden when treated with lenalidomide¹⁰, some do not, and conversely a smaller proportion of patients without del(5q) may also see improvement in transfusion needs and transfusion independence during lenalidomide therapy.^11,12 Similarly, the hypomethylating agents (HMAs) azacitidine or decitabine are routinely employed as disease-modifying chemotherapeutics for patients with higher risk disease, but mutation patterns have only modest effect on clinical response and such data are not used in practice for treatment decision-making.^13,14,15 New therapies that selectively target malignant clones while sparing healthy bone marrow (and ideally thus providing favorable toxicity profile) are desperately needed.

A challenge in developing new clinical therapeutics has been the limitations of preclinical in vivo models of MDS¹⁶ – as a result, many therapeutics in MDS to date have been “borrowed” from acute myeloid leukemia or other hematological malignancies in which such models are better developed. This barrier to drug development has been changing thanks to a greater understanding of the genetic composition of MDS and novel mouse models that provide insight into MDS genetics.^17-19 Sequencing efforts have identified frequent recurrent mutations in MDS that allow for the identification of clonal origin and how disease evolves over time.^3,20-22 These efforts identified frequent, recurrent mutations in pre-mRNA splicing that appear to be early events in the development of MDS.^23-27 Splicing factor mutations are present in more than half of MDS and chronic myelomonocytic leukemia (CMML) cases, and are associated with specific phenotypic findings, such as SF3B1 and ring sideroblasts,^28,29 or SRSF2 enrichment in CMML.³⁰

In this review, we discuss the rationale for targeting pre-mRNA splicing in MDS, including pre-clinical insights into the effects of splicing factor mutations and how they may present therapeutic vulnerabilities, and we summarize early clinical data from studies in progress. The spliceosome is an appealing target due to the prevalence of splicing factor mutations in MDS, presence of mutations in early malignant progenitors, and potential for exploiting synthetic lethality strategies. Nonetheless, splicing is such a fundamental biological process that the therapeutic window of splicing modulation may be narrow, and much work is needed to understand how to target splicing and test novel therapeutic combinations in order to change the MDS treatment paradigm.

Preclinical Rationale

Somatic mutations in pre-mRNA splicing factors are seen in a number of cancers, including solid tumors (notably uveal melanoma) and lymphoid malignancies such as chronic lymphocytic leukemia (CLL).^{31,27,32,33,34} These mutations are enriched in myeloid neoplasms, specifically in MDS, CMML, and AML (especially secondary AML arising from MDS). In MDS, mutations in splicing factors occur most often in SF3B1, U2AF1, SRSF2, and less frequently ZRSR2.²⁵ Combined, splicing factor mutations represent the most frequently acquired recurrently mutated gene in MDS,³⁵ and thus are of interest as possible therapeutic targets.³⁶

Splicing factor mutations result in characteristics splicing alterations which can then be measured as a biomarker.³⁷ Normal splicing proceeds through the identification of a 5’ splice site and 3’ splice site at either end of each intron to be spliced together, as well as a branch point sequence upstream of the 3’ splice site and polypyrimidine tract that are critical for the identification of exons and splicing accuracy.³⁸ In the initial steps of mRNA splicing, the 5’ site is identified and bound to the branch point sequence by the U1 and U2 snRNPs, the latter of which contains SF3B1 which binds to the branch point sequence. U2AF1 is involved in recognizing the 3’ splice site, while SRSF2, a member of the serine/arginine (SR) protein family, functions as a splicing enhancer, with a role in exon recognition. ZRSR2 appears to have a role analogous to U2AF1 but as part of the minor splicing complex. Mutations in these splicing factors result in characteristic alterations in pre-mRNA splicing based on these functions:³⁹ mutations in SF3B1 result in alternative branchpoint usage,^40,41 mutations in U2AF1 are associated with altered splice site recognition,⁴² and mutations in SRSF2 result in altered recognition of exon splicing enhancers, inducing altered splicing of a number of proteins including critical regulators of hematopoiesis.⁴³

At the same time, splicing does not happen in a vacuum; rather, splicing is coupled with RNA transcription in a dynamic fashion.^44-46 This has the effect that any perturbations in splicing can cascade into the normal processing of mRNA, potentially slowing elongation. Alterations in RNA transcription may lead to increased RNA:DNA hybrid structures termed R-loops,⁴⁷ which can effect DNA integrity and leading to single strand breaks.^48,49 In response, ataxia telangiectasia and rad3 related (ATR) signaling is activated, with eventual R-loop resolution.⁵⁰ The downstream effects of alternative splicing on such integral cellular functions as pre-mRNA transcription may provide further targets for therapeutic intervention in splicing factor mutated cells.

Mutual Exclusivity of Splicing Factor Mutations

Nearly all MDS cases can be characterized by at least one recurrent somatic mutation using a relatively small panel of ~100 myeloid genes of interest,^35,51-53 the most frequently mutated family of genes being those in the splicing apparatus.⁵¹ Interestingly, these mutations are almost always mutually exclusive of one another;^26,31,54 meaning that (with rare exception) a patient sample will harbor only one splicing mutation. Instances with more than one splicing factor mutation identified may often represent separate subclonal events. In addition, splicing factor mutations are heterozygous events, and the presence of the wild-type allele appears necessary for cell survival.⁵⁵ These factors suggest that mutations in the splicing apparatus, although frequent in MDS, nonetheless result in a state of dependence upon some wildtype splicing function, with strong negative selection for homozygous mutations or mutations in multiple splicing factor genes in the same cell.

Splicing Factor Mutations as Early Clonal Events

Serial sequencing of MDS patient samples over time shows that dynamic clonal changes may occur and can be associated with changes in clinical and pathologic behavior.^56,57 Some mutations, including those commonly seen in secondary AML such as FLT3, PTPN11, WT1, IDH1, NPM1, IDH2 and NRAS, are more commonly late events, occurring around the time of disease transformation.⁵⁷ This contrasts with splicing mutations which are usually present early in the course of disease.^23,58,59 Mutations in SF3B1, SRSF2, or U2AF1 tend to persist after treatment even when clinical response occurs,^58,60 further supporting the hypothesis that these mutations are present in an early progenitor and critical to the initial development of MDS. This also suggests that therapies targeting splicing factor mutations may be more likely to impact the malignant progenitor clone and perhaps thus more likely to elicit durable responses.

Preclinical Models of Splicing Factor Mutant MDS

Efforts to recapitulate genomically accurate murine models of myelodysplasia have led to new models for testing therapeutic combinations for treating MDS and greater insights into its pathogenesis (Figure 1).^61-63 One such murine model utilized inducible SRSF2 P95H, a mutation hotspot in human MDS. In this murine model, the investigators showed that homozygous loss of SRSF2 resulted in failed hematopoiesis, but conditional expression of P95H mutations in SRSF2 yielded a myeloid dysplasia phenotype.⁴³ Splicing factor mutations do not appear to result in loss of function (which may be lethal), but rather alter splicing preferences; in the SRSF2 murine model, disruption of SRSF2 function resulted in alternative splicing of several key hematopoietic regulators like EZH2. In a different inducible murine model of U2AF1 S34F, heterozygosity for this mutation recapitulated an MDS phenotype and resulted in aberrant splicing involving genes recurrently mutated in MDS and AML, among others.⁶⁴ Curiously, these inducible models of splicing factor mutations can recapitulate some aspects of the MDS phenotype, but splicing mutant cells have a competitive disadvantage with wildtype cells in co-transplantation experiments, in contrast to what is seen in human MDS (Figure 1).

Figure 1.

General Schema of Splicing Factor Mutated MDS Murine Models.

A prototype murine model of MDS. MDS progenitor cells may be introduced into mice through a number of methods, including using patient samples or engineering mice with inducible splicing factor mutations. After infusion/induction, these cells can be tested either comparing wildtype (WT) or mutant (SF^MUT) populations alone, or in combination with a drug of interest (e.g. spliceosome modulator, ATR inhibitor). Competitive transplantation studies show exhaustion of the clones harboring splicing factor mutations, that is hastened in homozygous mutations. Spliceosome modulation appears to improve survival when administered to mice with splicing factor mutated leukemic xenografts.

Synthetic Lethality in Splicing Factor Mutant Cells

A key question raised by the frequency of splicing factor mutations is whether and how these mutations could be targeted for therapeutic effect. Using murine models with mutated SRSF2 or U2AF1, the effects of specific chemotherapies have been testing for signals of selective activity against mutant cells (Figure 1).^62,63 In one study, mice engineered to express the SRSF2 P95H mutation received the spliceosome modulator E7107, an SF3B1 inhibitor. In this model, there was no significant difference in survival between SRSF2^P95H/+ mice and SRSF2^WT mice, but those mice with heterozygous SRSF2 P95H mutations who were exposed to E7107 had prolonged survival.⁶² Similarly, in a study evaluating hematopoietic cells expressing U2AF1^S34F/+, exposure to sudemycin compounds, which bind to the SF3B1 protein in the spliceosome, attenuated expansion of U2AF1 mutated progenitor cells in a mouse transplant model.⁶³ These and other studies have led to a hypothesized mechanism of action for spliceosome modulators: there is a degree of “tolerable” splicing dysfunction within a cell, but further alteration of splicing results in a state where the cell can no longer compensate and leads to selective cell death (Figure 2).

Figure 2.

Putative Therapeutic Mechanism of Spliceosome Modulator Therapies in Splicing Factor Mutated Cells.

At diagnosis, patients with MDS have largely malignant hematopoiesis arising from MDS progenitors; in splicing factor mutated MDS, normal splicing function will be diminished and maintained by the wildtype allele. Modulators of the spliceosome are thought to further alter splicing; healthy progenitor cells are able to “tolerate” this modulation, while cells with a splicing factor mutation have excessive spliceosome dysfunction leading to cell death. This hypothetically results in the expansion of the healthy cell compartment and restoration of normal hematopoiesis.

Interaction with Transcription and DNA Damage

Because splicing occurs in conjunction with mRNA transcription, alterations in splicing can affect the efficiency and integrity of transcription. As noted above, alternative splicing slows the replication fork and results in DNA:RNA hybrid structures termed “R-loops.” R-loops are unstable structures which displace the non-hybridized DNA strand, resulting in single-stranded DNA which activates a DNA damage response mediated through ataxia telangiectasia and Rad3 related (ATR) signaling.^48,65,66 Interrupting this DNA damage response may hamper genomic integrity and lead to selective apoptosis of splicing factor mutant cells.^67-69 Alternatively, cumulative DNA damage as well as the buildup of alternatively spliced isoforms may yield tumor enriched antigens that enhance immunogenicity on their own or in combination with immuno-oncologic therapies.^70,71

Clinical Studies

Several therapies have recently entered clinical trials to test the preclinical rationale of targeting the spliceosome itself or associated pathways (Table 1). Although clinical results are preliminary, early signals from these studies are likely to greatly influence our understanding of spliceosome function in both wildtype and mutated cells, including the role of splicing factor mutations in MDS pathogenesis and treatment.

Table 1.

Active Clinical Trials for Splicing-Factor Mutated MDS

Drug	Target	NCT Number	Notes
H3B-8800	Small molecule modulator of SF3B1		Dose escalation on two schedules: 5 days on, 9 days off, or 21 of 28 days.
GSK3326595	Reversible inhibitor of PRMT5		Dose escalation followed by three parallel study expansion arms in monotherapy or in combination with azacitidine
JNJ-64619178	Inhibitor of PRMT5		Dose escalation and tolerability, evaluating in lower-risk MDS
AZD6738	Small molecular ATR inhibitor		Single agent AZD6738 in higher-risk MDS, CMML, and lower-risk MDS

Modulating Spliceosome Function

Some of the first human studies to evaluate a spliceosome modulator investigated the compound E7107. Two phase 1 studies of E7107 enrolled patients with advanced solid tumors (, ).^72,73 Although no patients on these studies had a diagnosis of MDS, the toxicities encountered may be informative to spliceosome modulators in general. A total of 56 patients were enrolled between the two studies; the maximum tolerated dose of E7107 as an intravenous infusion was 4.3mg/m² when given on days 1 and 8 of a 21-day cycle, or 4mg/m2 when dosed on days 1, 8, and 15 of a 28-day cycle. Dose-limiting toxicities (DLTs) across the two studies included gastrointestinal toxicity (diarrhea, vomiting, nausea, abdominal cramping), anorexia, dehydration, and myocardial infarction. Also noted were three patients who experienced irreversible visual toxicity; one with blurred vision and central scotomas, and two others with vision loss; these events resulted in cessation of further study of this agent. Of note, investigators did observe a dose dependent change in mRNA splicing in cells from E7107 treated patients.⁷³

H3B-8800 is an orally bioavailable small molecule modulator of SF3B1 which shares a pladienolide chemical backbone but is molecularly distinct from E7107. Preclinical data confirmed selective cytotoxic activity against splicing factor mutated leukemia in xenograft models.⁷⁴ H3B-8800 has been studied in a phase I dose escalation trial enrolling patients with AML, MDS, and CMML, with preliminary data reported to date ().^75,76 A total of 81 patients were enrolled in dose escalation cohorts evaluating doses of 1-40mg administered either 5 days on/9 days off or on days 1-21 of a 28 day cycle. DLTs included bone marrow failure in a patient with lower-risk MDS treated at 7mg on the 5 days on/9 days off schedule, and nausea and QTcF interval prolongation at 20mg on the schedule of days 1-21 of 28. There was a dose-dependent increase in plasma levels of H3B-8800, as well as evidence of on-target alternative splicing in hematopoietic cells proportional to dose increases.⁷⁶ The drug was generally tolerable with several patients receiving treatment for over two years duration.

Other spliceosome modulators are under development including sudemycin D6,⁷⁷ Pladienolide-B, and FD-895, although none have yet been tested in human subjects;⁷⁸ there is also significant interest in novel combination therapies incorporating spliceosome modulation with other therapies in MDS.

Inhibition of PRMT5

Protein arginine methyltransferase 5 (PRMT5) is essential to the formation of small nuclear ribonucleoproteins (snRNPs) and the spliceosome machinery. Inhibition of PRMT5 results in increased alternative splicing patterns,⁷⁹ and also may impair homologous recombination DNA repair.⁸⁰ As such, PRMT5 is being explored as a target among tumors with spliceosome mutations.

GSK3326595 is a specific and reversible inhibitor of PRMT5 and has preclinical activity in a number of tumors; cell lines treated with GSK3326595 showed increased alternative splicing, and p53 activation due to alternative splicing of MDM4 was observed.⁸¹ This compound has been studied in a phase I dose escalation study in solid tumors and non-Hodgkin’s lymphoma ()⁸² and more recently is under investigation in myeloid neoplasms including AML, MDS, and CMML ().

Another PRMT5 inhibitor in early human study is JNJ-64619178. This agent is orally bioavailable and shows preclinical efficacy in a number of tumors including AML.⁸³ A study in patients with non-Hodgkin lymphoma and advanced solid tumors, as well as lower-risk MDS, is currently enrolling ().

Splicing Induced Immune Responses

Splicing factor mutations may result in generation of clonal cell-restricted neoantigens that can be recognized by the immune system.⁷¹ A number of studies are actively investigating the role of immunotherapy in MDS as monotherapy or in combination with hypomethylating agents, including the anti-CTLA-4 therapy ipilimumab, and the anti-PD-1 or PD-L1 therapies nivolumab, pembrolizumab, and atezolizumab.^84,85 While not specific to MDS harboring splicing factor mutations per se, analysis of mutation patterns and clinical responses in patients enrolled on these studies will be informative. Preliminary data suggests relatively low single agent response rates to immune checkpoint inhibitors, restricted largely to arms containing CTLA-4 inhibition.⁸⁴ Moreover, a number of patients have experienced immune-related toxicities, including pulmonary complications, that will need to be mitigated if these agents have a future in MDS.

DNA Damage Response

As noted, splicing factor mutations have effects on RNA transcription resulting in the activation of a DNA damage response. Early data suggests that this occurs by recruitment of RPA to the area of R-loop formation, activating ATR signaling and the CHK1-Wee1 axis.⁶⁸ Inhibiting this DNA damage response may provide another target in splicing factor mutated as well as splicing wildtype MDS.

AZD6738 is an orally bioavailable selective inhibitor of ATR that has been studied in solid tumors as monotherapy and, since it may potentiate chemotherapy-induced DNA damage, paired with a number of other agents,^86,87 including other DNA repair agents like poly-(ADP ribose) polymerase (PARP).⁸⁸ Mutations in the spliceosome result in constitutively increased R-loop formation and resultant ATR activation compared to wildtype cells.⁶⁸ Blunting the DNA damage response by inhibiting ATR results in synthetical lethal death preferentially in splicing factor mutated cells.⁶⁷ A clinical trial of AZD6738 in MDS and CMML is currently underway to test this hypothesis (). Other targets in this pathway, such as CHK1 or Wee1, may provide avenues for future study as well.

Summary/Discussion

The discovery of recurrent genetic mutations in MDS patient samples promises to bring an immense arsenal of targeted therapies to this extremely heterogeneous and treatment-refractory group of hematologic malignancies. A curious finding is the central, prominent role that mutations in the spliceosome machinery seem to play in this disease. These mutations occur early in the evolution of MDS and persist during treatment and over time. Splicing factor mutations are mutually exclusive of one another, and while each mutated splicing factor results in distinct and different downstream effects on splice isoforms, there are a number of common mis-spliced regulatory proteins. For all that has been discovered regarding splicing factor mutations in MDS, numerous questions remain, including how spliceosome mutations confer a clonal advantage and how best to exploit these mutations in the treatment of patients with this cancer. Nonetheless, the field is dynamic and evolving, with a number of currently enrolling trials targeting the pathways outlined in this review. Moreover, the identification of multiple potential targets against splicing factor mutations suggests the promise of novel:novel targeted combinations for patients with MDS, which may spare traditional chemotherapies altogether.

Synopsis.

Myelodysplastic syndromes are enriched for somatic mutations in the pre-mRNA splicing apparatus, with recurrent acquired mutations most commonly occurring in SF3B1, SRSF2, U2AF1, or ZRSR2. These mutations appear to be early events in the pathogenesis of disease, and given their frequency and central role in leukemogenesis, are of interest as potential therapeutic targets. Currently, clinical trials are exploring targets that either directly affect the spliceosome (splicing modulators e.g. H3B8800 or PRMT5 inhibitors e.g. GSK3326595) or that exploit vulnerabilities created by alternative splicing (ATR inhibitors e.g. AZD6738). Future research is needed to explore novel approaches and combinations of therapies and understand how these mutations lead to clonal dominance.

Key Points.

1. Myelodysplastic syndromes (MDS) are enriched for somatic mutations in proteins involved in pre-mRNA splicing, specifically SF3B1, SRSF2, U2AF1, or ZRSR2

2. Alterations in pre-mRNA splicing in MDS may expose therapeutic vulnerabilities related to splicing itself or to effects caused by altered splicing in the cells

3. Identification of multiple targets in splicing factor mutated cells may lead to novel drugs or novel drug combinations for the treatment of MDS and other myeloid neoplasms in the future