Cancer synthetic vulnerabilities to PRMT inhibitors.

Abstract

Protein arginine methylation is an abundant post-translational modification involved in the modulation of essential cellular processes ranging from transcription, post-transcriptional RNA metabolism, and propagation of signaling cascades to the regulation of the DNA damage response.

Excitingly for the field, in the past few years there have been remarkable advances in the development of molecular tools and clinical compounds able to selectively and potently inhibit protein arginine methyltransferase (PRMT) functions.

In this review, we first discuss how the somatic mutations that confer advantages to cancer cells are often associated with vulnerabilities that can be exploited by PRMTs’ inhibition. In a second part, we discuss strategies to uncover synthetic lethal combinations between existing therapies and PRMT inhibitors.

1. Overview of protein arginine methylation

Positively charged amino acids (i.e. arginines, lysines and histidines) can be methylated (i.e. addition of -CH₃ group(s)) by members of the protein methyltransferase family. Specifically, the guanidinium moiety of arginines, can be post-translationally modified to monomethylarginine (Rme1/MMA), asymmetrical dimethylarginine (Rme2a/ADMA) and/or symmetrical dimethylarginine (Rme2s/SDMA) by Protein Arginine MethylTransferases (PRMTs), using S-adenosyl methionine (SAM) as a cofactor ¹ (Fig.1A). All PRMTs are able to catalyze the addition of a first -CH₃ moiety (from Rme0 to Rme1/MMA). PRMT7, the only type III PRMT, is not able to add additional CH₃ groups, while PRMT1/PRMT2/PRMT3/PRMT4/PRMT6 and PRMT8 (Type I PRMTs) are able to catalyze Rme2a/ADMA and PRMT5 and PRMT9 (Type II PRMTs) are able to catalyze Rme2s/SDMA (Fig.1B). The addition of one or two methyl moieties maintains the positive charge of the guanidine group, while adding hydrophobicity to the side chain and diminishing its ability to form hydrogen bonds.

Fig.1.

A. The SAM cycle and MET salvage pathway. All PRMTs utilize SAM to catalyze their reaction, leading to the synthesis of monomethylarginine (MMA), asymmetrical dimethylarginine (ADMA) and symmetrical dimethylarginine (SDMA). MTA accumulates in MTAP null tumors and inhibits PRMT5. Abbreviations: MET=Methionine, SAM=S-adenosylmethionine, SAH=S-adenosylhomocysteine, HcY=Homocysteine, MTR1P=Methythioribose-1-phosphate, MTA=Methylthioadenosine, MAT=Methionine adenosyltransferases, MTAP=Methylthioadenosine phosphorylase B. The PRMT family. All PRMTs share a MT domain. Type I PRMTs (PRMT1, PRMT2, PRMT3, PRMT4, PRMT6 and 8 are Type I enzymes catalyzing MMA and ADMA. PRMT5 and PRMT9 are type II enzymes generating MMA and SDMA. PRMT7 is a type III enzyme generating only MMA. Abbreviations: MT=MethylTransferase domain; SH3=Src homology 3; ZnF=zinc finger; TPR=tetratricopeptide. Numbers at the C-terminus correspond to the aminoacid length of human PRMTs.

PRMTs are involved in the modulation of essential cellular processes such as transcription, post-transcriptional RNA metabolism (i.e. splicing), intracellular signaling, and DNA damage response (DDR).

Given such fundamental functions, it is no surprise that their homozygous deletion in mice results in embryonic or perinatal lethality and that human germline or somatic inactivating mutations in PRMT genes are rare. On the contrary, upregulation of PRMTs by a variety of mechanisms is a common feature of cancer (see ¹ for review).

In this commentary, we discuss how cancer vulnerabilities, whether associated with somatic mutations or with existing therapies, can be exploited by targeting PRMT function. The plethora of available PRMT inhibitors, with different modes of action, used for research or in early clinical trials makes such a strategy appealing ^1,2 (Table.1).

Table. 1.

PRMT, protein arginine methyltransferase; SAM, S-adenosyl methionine; MDS, myelodysplastic syndromes; MTAP, methylthioadenosine phosphorylase

MANUFACTURER	COMPOUND	TARGET	MECHANISM OF ACTION	REFERENCE	CLINICAL INDICATION
	AMI-1	type I PRMT	Binding to substrates, prevents substrate recognition	3 4
JIN LAB-SGC	MS023	type I PRMT	substrate-binding pocket	5
GSK	GSK3368715	type I PRMT	substrate-binding pocket	6	PhI: NCT03666988: Relapsed or refractory DLBCL and MTAP null solid tumors
JIN LAB-SGC	SGC707	PRMT3	allosteric	7
TAKEDA	TP-064	PRMT4	substrate-binding pocket	8
GSK/EPIZYME	EZM2302 or GSK3359088	PRMT4	substrate-binding pocket and making additional contact with the SAM-binding pocket	9
JIN LAB-SGC	MS049	PRMT4/6	substrate-binding pocket	10
EPIZYME	EPZ020411	PRMT6	substrate-binding pocket	11
GSK/EPIZYME	EPZ015666	PRMT5	substrate-binding pocket and making additional contact with the SAM-binding pocket	12
GSK/EPIZYME	GSK3326595		substrate-binding pocket and making additional contact with the SAM-binding pocket	13	PhI NCT02783300: Solid tumours and NHL; PHI/II NCT03614728: MDS and AML PHII: NCT04676516 Early stage breast cancer
Eli LILLY	LLY-283	PRMT5	SAM-binding pocket	14
JANSSEN	JNJ- 64619178	PRMT5	SAM-binding pocket and reaches the substrate- binding pocket	15	PhI NCT03573310: Relapsed or refractory B cell non-Hodgkin lymphoma (NHL) and advanced solid tumors
PFIZER	PF-06939999	PRMT5	SAM competitive		PhI NCT03854227: Advanced solid and metastatic tumors
PFIZER	PF-06829927	PRMT5	SAM competitive	16
PRELUDE	PRT543	PRMT5	SAM competitive	17;18	PhI NCT03886831
PRELUDE	PRT811	PRMT5	SAM competitive		PhI NCT04089449
PRELUDE	C449	PRMT5	covalent inhibitor	19
ARGONAUT	T1–44	PRMT5	SAM cooperative & peptide substrate competitive	20
JIN LAB	MS4322	PRMT5	PROTAC degrader	21
MERCK	Comp 1A	PRMT5	Allosteric inhibitor	22
AURIGENE	AU-14755	PRMT5	SAM cooperative & peptide substrate competitive	WO2019180628 WO2019180631
CTXT		PRMT5		WO2016034673
ANGEX		PRMT5	SAM competitive	WO2019112719 WO2020243178
JUBILANT	JPRMT5i	PRMT5	SAM cooperative & peptide substrate competitive	WO2019102494
SELLERS/IANARI LABs	BRD0639	PRMT5	PBM-competitive Disrupts PRMT5-RIOK1 complex	23
SGC	SGC3027	PRMT7	cell permeable pro-drug converted to SGC8158 binding to the SAM-binding pocket	24

2. Synthetic lethality with PRMT inhibitors

The classical concept of synthetic lethality states that simultaneous, but not individual, genetic loss of two genes impacts cell viability. This has been extended in cancer biology to encompass pharmacological inhibition or degradation of a particular target, leading to the selective killing of cancer cells bearing mutations or loss of the other gene, but not of normal cells retaining a wildtype copy ²⁵(Fig.2).

Fig.2.

Synthetic vulnerabilities to PRMT inhibitors in cancer. A. In normal cells, the inhibition of gene B (in this example PRMT5), is compatible with cellular viability, granted that gene A is active. B. Synthetic dose lethality. In cells with oncogenic upregulation of a driver (in this example MYC), there is cellular dependence on gene B (i.e. PRMT5). C. Synthetic or Collateral vulnerability. In cells with mutation in a driver (e.g. SRSF2^P95H; U2AF1^S34F; SF3B1^K700E) there is cellular dependence on gene B (i.e. PRMT5). A specific sub-example of synthetic lethality, is when the deleted gene A is a passenger deletion (i.e. MTAP, co-deleted with the CDKN2A/B locus).

Synthetic lethality was first described in Drosophila and yeast, where large scale genetic manipulations have historically been easier to achieve. With the advent of siRNA, shRNA, and, most recently, CRISPR/Cas9 technologies, multiple studies have extended the concept to mammalian cells ²⁶.

Here we summarize three examples of this concept that are relevant to PRMT inhibition.

2a. Tumors overexpressing MYC

MYC (or c-MYC) is part of the proto-oncogenic family of basic-helix–loop–helix leucine-zipper (bHLH-LZ) transcription factors together with NMYC and LMYC. They all recognize an E-box DNA element (i.e. CACGTG or non-canonical variants differing by one base pair) to promote or, less commonly, prevent transcription.

MYC levels are highly regulated at multiple stages (i.e. transcriptional, post-transcriptional, translational and post-translational), to allow rapid induction and degradation in response to signaling cascades influencing fundamental cellular processes (i.e. growth, proliferation, death/apoptosis, anabolism, nucleotide and protein biosynthetic pathways)²⁷.

MYC upregulation contributes to tumorigenesis and is a hallmark of multiple human malignancies ²⁷. Mechanistically, MYC regulates the differential expression of a discrete set of genes, mainly involved in metabolic processes such as RNA processing, ribosome biogenesis, and de novo purine synthesis ^28,29. In turn, this MYC-induced transcriptional program results in cellular rewiring, which ultimately affects global transcription levels as a secondary effect ³⁰.

An important set of genes directly activated by MYC is components of the splicing machinery, among which PRMT5 is a critical enzyme sustaining the kinetics of spliceosomal assembly ³¹. PRMT5 depletion results in cell cycle arrest followed by apoptosis, due to substantial deregulation of cellular splicing. Specifically, in the absence of PRMT5, Sm proteins are not symmetrically dimethylated and fail to be correctly assembled into snRNPs by the SMN complex. As a consequence, cells with reduced U1 levels fail to recognize weak 5′ splice sites, resulting in aberrant accumulation of intron retention and selective skipping of weak alternative exons³¹.

Importantly, by utilizing both genetic (heterozygous or homozygous PRMT5 deletion) and pharmacological (PRMT5 inhibitors) approaches, the community has reached a consensus on a “threshold model” to explain the role of PRMT5 in cancer (Fig.3A). In short, cancer cells typically require 2–5 fold more PRMT5 activity than normal cells. Critically, normal cells can still function with up to 15–20% residual PRMT5 activity, while cancer cells require >50%. Consistent with this model, heterozygous deletion of Prmt5 is well tolerated in mice while extending the latency of B cell lymphoma significantly. In the MYC-driven Eμ‐myc lymphoma model, the increased dependency on highly active core splicing machinery represents a specific vulnerability associated with MYC overexpression. A possible explanation for this increased dependency on PRMT5 and the splicing machinery in general ^31,32 is that in order to sustain the high metabolic rates, MYC-overexpressing cells have higher levels of transcribed mRNAs, which need to be correctly spliced to ensure cell survival. The vulnerability of MYC overexpressing cells to subtle perturbations of the splicing machinery (i.e. PRMT5 inhibition, or depletion of essential components such as SmB or PHF5A) has also been observed in patient-derived glioblastoma multiforme stem cells (GSCs)³³. Interestingly, this vulnerability can be recapitulated in Neural Stem Cells (NSC) ectopically overexpressing MYC, but not other oncogenic drivers (i.e. hTERT and dominant-negative p53, among others)³³.

Fig.3.

A. Threshold model of PRMT5 activity. Normal post-mitotic cells (black bar) require a certain PRMT5 activity, cancer cells typically require 2–5 fold more (blue bar). In addition, normal cells can tolerate low PRMT5 activity (indicated by the bottom black line), while cancer cells require more. The therapeutic window in red is indicated. B. Threshold model of PRMT5 activity in MTAP null cells. As in A. with the addition of the basal lower activity of PRMT5 in MTAP null cells. C. Modalities of PRMT5 inhibition.

An alternative explanation of the synthetic lethality between MYC overexpression and PRMT5 depletion is that MYC leads to transcriptional upregulation of several components of the translation machinery, including ribosomal RNAs/proteins and elongation factors (e.g. eIF4E). Consistent with this dependency, MYC oncogenic activity is suppressed by haploinsufficiency of ribosomal protein rpl24 ³⁴, and PRMT5 activity is required for cap‐dependent translation by eIF4E¹.

Last, cells with high MYC levels are subject to replicative stress ²⁷. Reduction in PRMT5 activity could exacerbate accumulation of R-loops and DNA damage, ultimately leading to cell death (see section 3 for more detail)^35,36

2b. Tumors with point mutations in splicing factors

More than 90% of transcriptional units express alternatively spliced isoforms and deregulated alternatively spliced (AS) oncogenes and tumor suppressors are linked to cancer initiation or maintenance³⁷. Myelodysplastic syndromes (MDS) are particularly affected by this phenomenon. Specifically, SF3B1, U2AF1, SRSF2, and ZRSR2 constitute the four most commonly mutated spliceosomal components in MDS (i.e. SRSF2^P95H; U2AF1^S34F; SF3B1^K700E and ZRSR2 loss of function mutations), as well as other liquid and solid malignancies ³⁷. Splicing factor mutations occur as heterozygous point mutations at distinct residues and are generally mutually exclusive. These mutations alter the normal RNA binding and splicing preferences of splicing factor proteins in a sequence-specific manner, that is distinct from loss-of-function ³⁷. Importantly, cells become dependent on expression of the wildtype (WT) allele, which explains the heterozygous nature of these mutations³⁸. From a therapeutic standpoint, splicing factor mutant cells are preferentially sensitive to further perturbations of the splicing machinery, such as those obtained using SF3B1 inhibitors ³⁹, which led to early phase clinical trials in patients with CMML and MDS (NCT02841540). The initial report of this study showed no major adverse effects, but disappointingly no objective complete or partial remission, highlighting the need to inhibit spliceosome function in alternative and more efficient ways ⁴⁰.

Perturbation of asymmetric (ADMA) and symmetric arginine dimethylation (SDMA) of multiple RNA binding proteins and splicing factors, through inhibition of PRMT1 and PRMT5, respectively, provides such alternate means of therapeutic splicing inhibition ^31,35,41. Indeed, reducing splicing fidelity by inhibiting ADMA/SDMA results in strong preferential killing of splicing factor mutant AMLs over their wildtype counterparts ⁴², consistent with what was previously observed with SF3B1 inhibitors. Mechanistically, both type I (PRMT1, PRMT4 and PRMT6) as well as type II PRMT enzymes (i.e. PRMT5), methylate distinct arginines on an overlapping set of core and alternative splicing factors. These RBPs converge to regulate the alternative splicing of key downstream targets (e.g. a poison exon on EZH2), leading to the observed synthetic lethality in SRSF2 or SF3B1 mutant cells, and the synergistic effects between PRMT1 and PRMT5 inhibitors⁴².

Besides the convergence on alternative splicing events, there are at least two other possible explanations for the synthetic lethality. First, both SRSF2 and SF3B1 perturbation lead to the mis-splicing of genes important in regulating the NF-KB pathway ³⁸. Given that PRMT5 activity is necessary for proper NF-KB function, PRMT5 inhibition could lead to downregulation of an essential pro-survival pathway in AML and MDS (Fig.2).

Second, splicing factor mutant tumors are known to upregulate the formation of R-loops, leading to DNA damage ⁴³. Both type I and II PRMTs have been linked to R-loop biology, and their inhibition leads to R-loop accumulation ^36,44, potentially explaining the synthetic lethality observed in SRSF2 or SF3B1 mutant AMLs⁴². Specifically, PRMT5 mediated methylation of PolII is important to prevent R-loop formation, by mediating the recruitment of Senataxin to chromatin via methylation of RNAPolII R1810me2s and the SMN reader protein ³⁶. Type I PRMTs (i.e. PRMT1 and PRMT4) have instead been implicated in TOP3b recruitment, via methylation of H4R3me2a and RNAPolII R1810me2a and the subsequent binding of the TDRD3 reader protein ⁴⁵. As a consequence, PRMT inhibition would lead to excessive R-loop formation, tilting the balance towards apoptotic cell death (Fig.2).

2c. MTAP null tumors

Despite the high frequency of loss-of-function genetic mutations in cancer, very few therapeutic approaches target tumor suppressor genes (TSGs) directly; there is instead a major focus on gain-of–function oncogenic events. This is often due to practical reasons, such as the absence of a protein to target upon TSG deletion, or the challenge of identifying small molecules able to restore TSG function in the case of inactivating mutations. These difficulties have led scientists to look at the problem from alternative angles: in particular, deletions or mutations of either TSG or neighboring genes might rewire cellular metabolism or the cancer epigenomic landscape, making tumor cells selectively vulnerable to specific drug treatments, concepts that have been referred to as synthetic and collateral lethality, respectively.

Chromosome 9p21 (chr9p21) is known to contain important TSGs such as p16/INK4a and p19/ARF and is homozygously deleted in a broad variety of solid tumors. The deletion varies widely both in frequency (e.g. reaching over 50% in mesotheliomas, gliomas and melanomas) and in amplitude, often including neighboring genes such as the methylthio-adenosine phosphorylase (MTAP) gene ⁴⁶.

MTAP is an enzyme that is part of the methionine salvage pathway and, as the name suggests, it generates methionine and adenine by phosphorylating 5′-methylthioadenosine (MTA)(Fig.1A). In an attempt to identify collateral vulnerabilities in MTAP null cells, three independent groups ^47–49 have identified PRMT5 and its interacting partners (i.e. WDR77/MEP50, RIOK1, COPR5, pICln), as key targets that, if downregulated, will selectively kill MTAP deficient tumors.

Mechanistically, all three groups were able to demonstrate that MTA (MTAP substrate), which accumulates in MTAP null cells at a concentration of 100 μM and additionally diffuses out in the tumor microenvironment, is a selective PRMT5 inhibitor, which occupies the SAM binding pocket and leads to downregulation of PRMT5-mediated symmetrically dimethylated proteins in the cell.

Importantly, from a therapeutic standpoint, the first PRMT5 inhibitor used in preclinical studies (EPZ015666) did not show any preferential killing of MTAP null cells. While at first these findings were surprising, it can be explained by the mechanistic action of this compound, which requires SAM in order to bind to the PRMT5 pocket ¹². Thus, EPZ015666 will not bind to a preformed PRMT5-MTA complex. For similar reasons, additional PRMT5 inhibitors that are currently in development and that target the SAM binding pockets are also predicted to show no or little selectivity toward MTAP null tumors.

In order to take advantage of the collateral vulnerability in MTAP null cells, we thus predict that several approaches could be successful to reduce PRMT5 activity in the context of MTAP null tumors (Fig.3B–C): 1] Develop allosteric PRMT5 inhibitors ²², not targeting the SAM pocket or not requiring a PRMT5/SAM complex to bind; 2] Develop PRMT5 inhibitors that selectively recognize and stabilize the PRMT5/MTA complex; 3] Develop therapeutics that will reduce PRMT5 levels in cells. These can be small molecule degraders ²¹, or alternative modalities such as antisense oligonucleotides (ASOs)⁵⁰. Both would mimic the effect reported in the case of PRMT5 shRNA-mediated downregulation ^47–49; 4] Develop inhibitors that interfere with PRMT5 activity by impacting the availability of SAM or other cofactors. One such example is the recently reported MAT2A inhibitor, able to impair growth of MTAP null tumor cells by perturbing PRMT5-mediated splicing and DNA damage control⁵¹; 5] Develop degraders or protein-protein interaction inhibitors directed against PRMT5 cofactors. These include: Methylosome Protein 50- MEP50/WDR77 (PRMT5 obligatory and stoichiometric partner, affecting PRMT5 enzymatic activity), Methylosome subunit pICln (directing PRMT5 methylation activity towards Sm proteins), RIO Kinase 1-RIOK1 (directing PRMT5 methylation activity towards Nucleolin), cooperator of PRMT5- COPR5 (directing PRMT5 methylation activity towards histone substrates), oxidation resistance gene 1 (OXR1) (directing PRMT5 methylation activity towards H3R2 methylation)⁵², Family With Sequence Similarity 47 Member E-FAM47E (enhancing PRMT5 chromatin association and histone methylation activity) ⁵³, and possibly others yet to be identified¹. The first in class of such class of inhibitors has recently been reported ²³.

In addition to testing PRMT5 inhibitors, several groups have now highlighted the synergistic effect of combining PRMT5 selective inhibitors with type I inhibitors ^42,54. Indeed, type I PRMT inhibitors preferentially kill MTAP null tumors, which have an impaired PRMT5 activity, highlighting an additional vulnerability of such tumors^6,54. It will be interesting to test in the future whether MTAP null tumors will be preferentially sensitive to other inhibitors of the spliceosome or to inhibitors converging on additional PRMT5-regulated pathways (i.e. DDR).

3. Synergistic drug combinations with PRMT inhibitors

In addition to identifying synthetic lethality between PRMT inhibition and genetic alternations in cancer (as described above), several groups have uncovered potential drug combinations that would synergize with PRMT inhibitors.

3a. PRMTi and the DNA double-strand break response (DDR) pathway

Different PRMT family members are able to modulate the DDR pathway, and as such, multiple groups have first characterized the mechanism of action and then tried to combine PRMT inhibition with DNA damage-inducing chemotherapy agents, ionizing radiation (IR), or PARP inhibitors, to mention a few.

It is well established that PRMT5 inhibition leads to DNA damage accumulation and induction of a p53 response ³⁵. This can be due to several proposed mechanisms, which are likely happening simultaneously to coordinate a robust DDR. First, PRMT5 can directly methylate RUVBL1 at R205; this SDMA event is essential to modulate 53BP1 release at sites of homologous recombination (HR)-mediated double-strand break (DSB) repair ⁵⁵. Additionally, PRMT5 inhibition leads to aberrant splicing of multiple genes involved in HR, including Tip60 ⁵⁶. An important corollary of this latter study is that, similarly to other documented HR deficiencies (i.e. BRCA1 loss of function), PARP inhibition can be used in combination with PRMT5 inhibitors to synergistically kill tumor cells. Third, PRMT5 directly methylates 53BP1, affecting its stability, and coordinating non-homologous end joining (NHEJ) repair following etoposide treatment ⁵⁷. Last, as pointed out earlier, both type I PRMTs and PRMT5 are important to prevent R-loop accumulation and subsequent DNA damage^36,45.

In breast cancer, PRMT5 regulates cancer cell stemness ⁵⁸, a function that has recently been linked to doxorubicin resistance⁵⁹, while in pancreatic cancer, PRMT5 was identified by a CRISPR screen as a druggable target whose inhibition synergizes with gemcitabine (Gem) to induce DNA damage. PRMT5 can also directly methylate p53, Rad9 (cell cycle checkpoint control protein), Fen1 (Flap endonuclease-1), and TDP (tyrosyl-DNA phosphodiesterase)(summarized in¹), further corroborating the relevance of PRMT5 in DDR.

It is noteworthy that PRMT5-mediated suppression of p53 activity is also relevant to the PRMT5/DDR axis. This was first documented in the developing brain, where PRMT5 depletion led to mis-splicing of a key exon in Mdm4, leading to depletion of MDM4 and activation of a P53 transcriptional program ³⁵. This was later confirmed by multiple studies in the context of solid and hematopoietic tumors, highlighting the importance of an intact MDM4/p53 axis for the activity of PRMT5 inhibitors ^13,31,60,61.

PRMT1 function has also been clearly linked to the DDR. Specifically, the MRE11 and 53BP1 GAR motif, methylated by PRMT1, is important for proper DNA repair, making cells and organisms deficient in such ADMAs more sensitive to IR ^62–66.

In summary, clinically promising strategies include combination of PRMT inhibitors with DNA damage-inducing agents or drugs blocking an effective DDR response.

3b. PRMTi and combination with other precision therapies

Both PRMT1 and PRMT5 activity is important for pro-survival and proliferative signaling cascades, such as BMP, TGFbeta, EGFR, and PDGFR (recently reviewed in ¹), making the case for combining PRMT inhibitors with available kinase inhibitors or biologics (i.e. cetuximab) targeting these signaling cascades. As an example, PRMT5 methylation of proteins in the PI3K/AKT pathways led to testing a combination of PRMT5 and mTOR inhibitors as a potential therapeutic approach in glioblastoma (GBM) ⁶⁷.

PRMT1-mediated methylation of EGFR promotes EGFR signaling. To exploit this in a therapeutic setting, several groups have combined PRMT1 knockdown (KD) or inhibitors with EGFR inhibitors (i.e. erlotinib or cetuximab). In one study, the authors show that PRMT1 inhibition increased non-small cell lung cancer (NSCLC) sensitivity to erlotinib and reversed EMT to overcome resistance to the kinase inhibitor⁶⁸. In other studies, PRMT1 depletion sensitized cells to cetuximab treatment ^69,70. While the authors used PRMT1 KD, it will be important in the future to test if similar effects are observed with PRMT1 inhibitors (i.e. GSK3368715 or MS023). Overall, these studies have demonstrated encouraging efficacy in vitro and in mouse models. However, future investigations are necessary to assess the degree of cumulative toxicity, or potential mechanisms of tumor resistance.

PRMT1-mediated FLT3 methylation sustains AML maintenance. As such, combination of a potent FLT3 inhibitor, Quizartinib (AC220) with MS023 (a pan-type I PRMT inhibitor, primarily inhibiting PRMT1) efficiently eliminated FLT3-ITD(+) AML cells and would be an interesting approach to test in the clinic ⁷¹.

Additionally, a recent publication assessed the efficacy of a novel PRMT5 inhibitor (C220) in JAKV617F-mutant myeloproliferative neoplasm (MPN). C220 was very effective in inducing DNA damage and apoptosis, both as a monotherapy, and in combination with JAK inhibition (ruxolitinib) ¹⁸.

4. Future outlook

Overall, it is still early days for PRMT inhibitors in the clinic and these coming years should be focused on identifying additional synthetic lethality strategies to widen the therapeutic windows for these compounds. An important aspect to take into consideration for the effectiveness of synthetic lethality strategies will be tumor heterogeneity. Ideally, the majority of tumor cells should carry the given somatic alteration, such that the vulnerability being exploited is clonal, rather than subclonal. In the examples discussed above, splicing factor mutations tend to occur early, typically in the founding clone of MDS and other myeloid malignancies ⁷²; similarly, MTAP deletion occurs early on in tumor progression, concurrently with that of the CDKN2A/B locus ⁷³. Last, MYC amplification and/or its increased expression is a hallmark of cancer, and is also present clonally in multiple cancer types ²⁷, making all of these alterations viable options to target with PRMT inhibitors to exploit their synthetic lethality.

In the near future, in addition to cell intrinsic effects of PRMT inhibitors, it will be paramount to test PRMT inhibitors in combination with immunotherapy. The rationale for such combination stems from the known link between splicing deregulation and creation of neo-antigens ^74–76 and the fact that PRMT1 and PRMT5 inhibition leads to widespread inclusion of novel exons as well as intron retention. A recent study has indeed tested the combination of PRMT5 inhibition with anti-PD-1 therapies, demonstrating high efficacy in a murine melanoma model (B16cells) due to the negative correlation between PRMT5 activity and both MHCI antigen presentation and PD-L1 expression ⁷⁷.

At the same time, the potential effect of PRMT inhibitors on T-cell fitness and activation of an interferon response will have to be carefully evaluated ^16,17,78,79 to further pave the way for more extensive testing in the clinic.