Splicing Factor Mutations in Clonal Hematopoiesis and Progression to Myeloid Neoplasms

ABSTRACT

Splicing factor (SF) mutations are recurrent driver alterations in myeloid neoplasms and also represent a distinctive class of mutations in clonal hematopoiesis (CH), an age‐associated expansion of hematopoietic clones carrying somatic driver mutations. SF‐mutant CH is characterized by relatively late emergence and accelerated clonal expansion in older individuals. This review summarizes current knowledge of the genetic features and clonal dynamics of SF‐mutant CH. We discuss disease‐associated molecular consequences of SF mutations, including dysregulation of innate immune and inflammatory signaling, R‐loop accumulation associated with replication stress, altered stress granule dynamics, impaired minor intron splicing, and cooperative genetic interactions. We further consider candidate mechanisms that may contribute to age‐dependent selection of SF‐mutant clones during CH, including attenuated interferon responsiveness and telomere‐dependent changes in replicative fitness. Together, current evidence highlights the unique biology of SF‐mutant CH and the unresolved mechanisms underlying its preferential expansion in older individuals. Clarifying these mechanisms may improve risk stratification, inform early‐intervention strategies, and advance our understanding of age‐associated clonal evolution and leukemogenesis.

Splicing factor (SF)‐mutant CH is characterized by relatively late emergence and accelerated clonal expansion in older individuals. This review summarizes its clonal dynamics, disease‐associated consequences of SF mutations in myeloid neoplasms, and candidate mechanisms that may contribute to age‐dependent selection.

Abbreviations

CH

Clonal hematopoiesis

CHIP

Clonal hematopoiesis of indeterminate potential

HSCs

Hematopoietic stem cells

MDS

Myelodysplastic neoplasms

NMD

Nonsense‐mediated mRNA decay

SF

Splicing factor

SGs

Stress granules

VAF

Variant allele frequency

1. Splicing Factor Mutations in Clonal Hematopoiesis and Myeloid Neoplasms

Somatic mutations accumulate in normal human cells throughout life and increase with age, constituting a fundamental biological process underlying both malignant and nonmalignant diseases [1]. Clonal hematopoiesis (CH) refers to the clonal expansion of hematopoietic stem cells (HSCs) that have acquired somatic mutations conferring a selective growth advantage [2]. When leukemia‐associated driver mutations are detected at a variant allele frequency (VAF) of ≥ 2% in individuals without overt hematologic malignancy or cytopenia, the condition is termed clonal hematopoiesis of indeterminate potential (CHIP), whose prevalence increases substantially with age [3, 4, 5, 6, 7]. Clinically, CH is associated with an increased risk of hematologic malignancies, elevated all‐cause mortality [4], and various nonmalignant diseases [8].

Over the past decade, studies have clarified the genetic landscape and molecular basis of CH emergence and clonal evolution. Among individuals with CHIP, mutations in epigenetic regulators such as DNMT3A, TET2, and ASXL1 account for more than two‐thirds of cases [7]. Splicing factor (SF) mutations represent the next most frequent class of driver alterations and are notable for their characteristic clonal behavior. Additional recurrent mutations affect genes involved in DNA damage response (e.g., PPM1D, TP53) and signaling pathways (e.g., JAK2). These findings highlight the biological heterogeneity of CH and the importance of understanding driver‐specific effects on clonal dynamics.

CH evolves through competition among mutant HSCs within the bone marrow microenvironment, with clonal fitness shaped by intrinsic and extrinsic factors. Although many clones remain stable, cooperative genetic alterations and environmental influences can promote progression to overt myeloid neoplasms [9] (Figure 1). Among the diverse driver lesions implicated in CH, SF mutations are of particular interest because their clonal behavior differs from that of more common CHIP drivers and becomes more prominent in older individuals.

FIGURE 1.

Gene‐dependent clonal expansion in CH. Schematic illustration of the development of CH from normal HSCs progression to hematologic malignancies. DNMT3A clones typically arise early and expand slowly, with later TET2 clones showing sustained growth across adulthood, whereas SF‐mutant clones generally emerge later and expand more rapidly in older individuals. Clonal expansion is influenced by both intrinsic and extrinsic factors, with cooperative mutations further accelerating progression to hematologic malignancy. (Created with BioRender.com).

SF mutations are among the most frequent genetic alterations in myelodysplastic neoplasms (MDS), most commonly affecting four genes—SF3B1 , U2AF1, SRSF2, and ZRSR2—which encode proteins involved in 3′ splice‐site recognition during pre‐mRNA splicing. Mutations in three of these factors (SF3B1, SRSF2, and U2AF1) are typically heterozygous missense variants clustered at hotspot residues and act as neomorphic alleles that alter splice‐site selection specificity. In contrast, ZRSR2 mutations are distributed throughout the coding region and frequently result in truncation, consistent with loss of function. These alterations are largely mutually exclusive, suggesting convergence on shared downstream consequences. Functionally, SF mutations induce widespread splicing abnormalities—including alternative 3′ or 5′ splice‐site usage, exon inclusion or exclusion, and intron retention—affecting genes that perturb hematopoiesis and contribute to the MDS phenotype, although their full pathogenic spectrum remains incompletely defined [10, 11]. Clinically, SF mutations display marked genotype–phenotype correlations within myeloid neoplasms. For example, SF3B1 mutations are detected in the majority of MDS cases with ring sideroblasts, whereas SRSF2 mutations are frequently observed in chronic myelomonocytic leukemia [10].

This review focuses on SF‐mutant CH and its progression to myeloid neoplasms, with particular attention to clonal dynamics, disease‐associated molecular consequences, and candidate mechanisms that may contribute to age‐dependent clonal expansion.

2. Age‐Associated Clonal Dynamics of SF‐Mutant CHIP

CHIP clones harboring SF mutations exhibit distinctive patterns of age at onset and clonal expansion compared with other major driver mutations. A targeted sequencing study by McKerrell and colleagues first highlighted age‐related differences among CHIP drivers, showing that DNMT3A and JAK2 mutations could be detected in young adults (as early as 25 years), whereas SF mutations were predominantly observed in individuals older than 70 years [6].

Fabre et al. longitudinally followed 385 adults aged 55–93 years (median follow‐up, 13 years) without prior hematologic malignancy and analyzed clonal dynamics using targeted sequencing. DNMT3A‐ and TP53‐mutant clones expanded at approximately 5% per year, TET2‐mutant clones at roughly 10% per year, whereas SRSF2‐ or U2AF1‐mutant clones grew more rapidly at 15%–20% per year, in some cases exceeding 50% [12].

Whole‐genome sequencing of single‐cell–derived colonies enabled estimation of mutation timing. DNMT3A‐mutant clones were inferred to arise early in life (around the third decade) and expand gradually, with growth decelerating later in life. TET2 mutations appeared throughout adulthood and maintained relatively stable growth in older individuals. In contrast, SRSF2‐ and U2AF1‐mutant clones were estimated to arise after age 30, with a median age of clonal expansion onset in the late 50s. However, as these estimates were based on mathematical modeling, it remains unclear whether SF mutations are acquired later in life or instead arise earlier but gain a selective advantage in the aging hematopoietic environment [12].

Similar trends were observed in an exome sequencing study of 4187 participants (aged 44–84 years; mean age 55 years) from the Atherosclerosis Risk in Communities (ARIC) cohort, followed for a median of 21 years. CHIP clones harboring SF or TET2 mutations exhibited significantly faster growth compared with clones carrying non‐R882 DNMT3A mutations. Sex‐specific differences were also observed, with SF and ASXL1 mutations more common in men and DNMT3A mutations more frequent in women [13]. The mechanisms underlying these sex differences remain poorly understood, but potential contributors include differential carcinogen exposure, gene‐dosage effects involving X‐linked genes such as ZRSR2, and hormonal influences on HSPC biology and leukemic potential [14, 15].

Consistent with these findings, a prospective analysis of the population‐based Lifelines cohort evaluated 7045 serial samples from 3359 individuals aged ≥ 60 years (mean, 68; range, 60–91) with peripheral blood count abnormalities. Over a median follow‐up of 3.6 years, SF‐ and JAK2‐mutant clones exhibited the highest annual growth rates (approximately 10%–18% and 9.8%, respectively), whereas DNMT3A‐ and TP53‐mutant clones expanded only marginally (1.9% and 3.6%) [16].

Taken together, SF‐mutant CH demonstrates a characteristic pattern of relatively late emergence and accelerated expansion in older individuals compared with other major CHIP drivers (Figure 1).

3. Molecular Consequences of SF Mutations in Myeloid Neoplasms

SF mutations primarily alter RNA processing and lead to diverse downstream consequences in myeloid neoplasms. This section summarizes major disease‐associated effects of SF mutations identified in patient samples and experimental models, as well as cooperative events implicated in the pathogenesis of myeloid neoplasms.

3.1. Aberrant Splicing and Activation of Innate Immune Signaling

In innate immune signaling, Toll‐like receptors recognize pathogen‐ or damage‐associated molecular patterns (PAMPs or DAMPs), leading to recruitment of MyD88 and assembly of the myddosome complex, which includes IRAK4 and IRAK1. This cascade activates TRAF6 and MAP3K7 (TAK1), ultimately triggering NF‐κB signaling.

IRAK4 has a long isoform, IRAK4‐L, which retains the N‐terminal death domain, hinge domain, and kinase domain, and shorter isoforms generated by alternative splicing. U2AF1 mutations alter 3′ splice‐site recognition and promote inclusion of IRAK4 exon 4, thereby preserving the death domain, whereas SF3B1 mutations induce inclusion of exon 6, which encodes part of the kinase domain. Both events increase expression of the active IRAK4‐L isoform. IRAK4‐L enhances myddosome assembly and activates downstream NF‐κB signaling, thereby promoting the survival and proliferation of SF‐mutant cells [17, 18] (Figure 2A).

FIGURE 2.

Distinct inflammatory related mechanisms associated with SF mutations. (A) SF mutations promote aberrant splicing of IRAK4, MAP3K7, and CASP8, collectively enhancing NF‐κB–mediated inflammatory signaling. (B) SRSF2 mutations promote accumulation of R‐loops and cytoplasmic micronuclei, thereby activating the cGAS–STING–IRF3–inflammasome axis. (C) SRSF2 mutations attenuate responsiveness to IFN–JAK–STAT1 signaling via STAT1 isoform imbalance. (Created with BioRender.com).

Since SF mutations are largely mutually exclusive in myeloid neoplasms, their functional convergence and mutual exclusivity have been investigated using mouse models harboring both Sf3b1 p.K700E and Srsf2 p.P95H mutations [19]. Individual SF mutations converge on activation of innate immune signaling through distinct aberrant splicing events. In SF3B1‐mutant cells, aberrant 3′ splice‐site selection in MAP3K7 exon 5 generates a frameshifted transcript subject to nonsense‐mediated mRNA decay (NMD), reducing MAP3K7 expression and enhancing NF‐κB activity. In contrast, SRSF2 mutations induce exon skipping in CASP8, producing a truncated isoform that lacks the catalytic domain but retains the N‐terminal death‐effector domain (DED), thereby augmenting NF‐κB signaling upon TRAIL stimulation (Figure 2A).

Thus, coexistence of Sf3b1 and Srsf2 mutations results in excessive NF‐κB signaling that, together with additional splicing defects, compromises hematopoietic cell survival. In contrast, NF‐κB activation driven by a single mutation permits cell survival while inducing MDS‐like phenotypes. These findings provide a mechanistic explanation for the mutual exclusivity of major SF mutations and identify innate immune signaling as a convergent pathway in SF‐mutant myeloid neoplasms [19].

3.2. R‐Loop–Associated Replication Stress and Innate Immune Activation

R‐loops are three‐stranded nucleic acid structures formed by RNA:DNA hybrids during transcription, leaving the non‐template DNA strand exposed as single‐stranded DNA. They function as essential regulators of transcription and chromatin organization [20].

In cellular and disease models, SF mutations, including those in SRSF2 and U2AF1, disrupt transcriptional elongation by impairing RNA polymerase II (Pol II) pause release, resulting in persistent promoter‐proximal pausing and consequent R‐loop accumulation. Such transcriptional stress compromises replication fork progression, activates the ATR–CHK1 DNA damage response, and induces G2/M cell cycle arrest, thereby impairing HSC proliferation and contributing to MDS pathogenesis [21].

In addition, accumulated R‐loops and cytoplasmic micronuclei activate DNA‐sensing pathways including cGAS–STING, leading to type I interferon and NF‐κB signaling, inflammasome activation, and pyroptosis—an inflammatory form of programmed cell death [22] (Figure 2B). These findings link SF mutation‐associated transcriptional stress to DNA damage responses and innate immune activation in SF‐mutant myeloid neoplasms.

3.3. Impaired Minor Intron Splicing and HSC Self‐Renewal

Most introns are removed by the major spliceosome, whereas rare but evolutionarily conserved U12‐type introns are excised by the minor spliceosome. ZRSR2 is a key minor spliceosome component and is therefore mechanistically distinct from SF3B1, SRSF2, and U2AF1, which mainly alter major spliceosome‐dependent splice‐site recognition. In ZRSR2‐mutant myeloid neoplasms, impaired minor intron excision leads to widespread U12‐type intron retention [23]. In a mouse model of Zrsr2 loss, Inoue et al. showed that impaired minor intron excision enhances HSC self‐renewal. CRISPR screens designed to mimic nonsense‐mediated decay of minor intron‐containing transcripts further identified LZTR1, a regulator of RAS‐related GTPases, as a downstream mediator [24]. These findings suggest that minor intron recognition regulates hematopoietic stem cell function and that ZRSR2 loss may confer a self‐renewal advantage through dysregulation of selected minor intron‐containing genes.

3.4. Stress Granules Formation and Clonal Fitness

Stress granules (SGs) are membraneless cytoplasmic organelles composed of RNAs and RNA‐binding proteins that assemble in response to cellular stress and enhance stress adaptation [25].

In U2AF1‐mutant leukemia cell lines, integrative analyses of RNA binding, splicing, and transcript turnover revealed that approximately 30% of genes affected by the S34F and Q157R mutants were shared and enriched for SG‐related functions, indicating that aberrant RNA processing by mutant U2AF1 converges on SG biology. U2AF1 mutations increase the stability or synthesis of SG‐enriched transcripts, thereby promoting SG assembly. Enhanced SG formation may improve cellular stress adaptation in U2AF1‐mutant cells and contribute to their fitness advantage in myeloid malignancies [26].

3.5. Cooperative Genetic Interactions in Disease Progression

Large‐scale genomic analyses have revealed that myeloid neoplasms frequently harbor multiple coexisting driver mutations. Importantly, these mutations do not accumulate randomly, but exhibit gene‐specific patterns of co‐mutation and mutual exclusivity (Table 1) [27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37]. Recurrent co‐mutation combinations likely reflect positive selection, whereas mutual exclusivity suggests synthetic lethality or deleterious interactions.

TABLE 1.

Co‐mutation and mutual exclusivity patterns of SF mutations in myeloid neoplasms.

Splicing factor mutation	Frequently co‐occurring mutations	Mutually exclusive mutations	References
SF3B1	DNMT3A, TET2, JAK2	IDH1/2, STAG2, TP53, U2AF1, SRSF2, ZRSR2	Ogawa [27], Bersanelli et al. [28], Bernard et al. [29], Malcovati et al. [30]
U2AF1	ASXL1, BCOR	SF3B1, SRSF2, ZRSR2	Ogawa [27], Bernard et al. [29], Badar et al. [31]
SRSF2	TET2, IDH1/2, ASXL1, STAG2, RUNX1	EZH2, SF3B1, U2AF1, ZRSR2	Ogawa [27], Bersanelli et al. [28], Kim et al. [32], Yoshimi et al. [33], Ochi et al. [34]
ZRSR2	TET2, EZH2	DNMT3A, IDH2, SF3B1, U2AF1, SRSF2	Ogawa [27], Bersanelli et al. [28], Garcia‐Ruiz et al. [35]

In CH, additional driver lesions can also accumulate over time, increasing clonal complexity and contributing to cooperative genetic interactions. Population‐based studies indicate that approximately 10%–18% of individuals with CH carry two or more driver mutations—occasionally up to five to seven—and that the prevalence of such multi‐mutant clones rises with age [12, 13, 16, 38]. Although greater clonal complexity is associated with an elevated risk of progression to myeloid neoplasms, many individuals remain clinically stable for years [39], indicating that genetic complexity alone does not determine disease trajectory.

Consistent with this concept, our conditional knock‐in mouse model carrying the Srsf2 P95H mutation did not develop MDS under steady‐state conditions but developed overt MDS under replicative stress following bone marrow transplantation [40], indicating that SF mutations alone do not drive overt disease without cooperating genetic alterations or environmental pressures.

The functional consequences of cooperative mutations have been investigated in compound mutant mouse models. Combined loss of Zrsr2 and Tet2 synergistically exacerbated splicing abnormalities—particularly in MAPK pathway genes—thereby enhancing bone marrow repopulation and clonal dominance, ultimately leading to MDS in mice [35]. Similarly, mice harboring both Srsf2 and Idh2 mutations develop lethal MDS with greater clonal dominance and leukemogenic potential than single‐mutant mice [33]. Mechanistically, mutant SRSF2 exhibits altered RNA‐binding specificity, with increased preference for CCNG motifs over GGNG motifs [32]. This altered sequence preference promotes mis‐splicing of INTS3, while mutant IDH2 produces 2‐hydroxyglutarate (2‐HG), inducing DNA hypermethylation and RNA Pol II pausing that further amplify the mis‐splicing of INTS3. Aberrant INTS3 transcripts are degraded by NMD, compromising Integrator complex‐dependent snRNA processing and Pol II regulation, thereby disrupting myeloid differentiation and oncogenic signaling programs (Figure 3).

FIGURE 3.

Cooperative SRSF2 and IDH2 mutations drive INTS3 mis‐splicing and leukemogenesis. Schematic model showing that mutant SRSF2 induces INTS3 intron 4/5 retention and exon 5 skipping, while mutant IDH2–mediated DNA hypermethylation enhances RNA pol II pausing, collectively amplifying INTS3 mis‐splicing. Aberrant INTS3 transcripts undergo NMD, resulting in INTS3 depletion and Integrator dysfunction, impairing snRNA processing and transcriptional programs, thereby promoting clonal dominance and leukemogenesis. (Created with BioRender.com).

Together, these findings suggest that cooperative mutations may modify the fitness consequences of SF mutation‐associated RNA processing defects and thereby promote progression toward myeloid neoplasms.

4. Candidate Mechanisms for Age‐Dependent Selection of SF‐Mutant CH

HSCs harboring SF mutations exhibit impaired self‐renewal and proliferative capacity under steady‐state conditions, as shown in patient‐derived cells and young SF‐mutant mouse models [32, 40, 41, 42, 43, 44, 45]. These findings indicate that SF mutations do not intrinsically confer a proliferative advantage to young HSCs. The preferential expansion of SF‐mutant clones in older individuals therefore suggests that their fitness advantage is context‐dependent and may emerge only when age‐associated stresses constrain normal HSC function. In this section, we discuss candidate mechanisms that may contribute to this age‐dependent selection.

4.1. Context‐Dependent Inflammatory Pressure and Attenuated Interferon Responsiveness

Transcriptomic analyses of human induced pluripotent stem cell (iPSC) models and patient‐derived cells harboring SRSF2 mutations revealed altered alternative polyadenylation of STAT1 mRNA, resulting in reduced expression of the shorter STAT1β isoform. Because STAT1β stabilizes STAT1α, its reduction results in decreased total STAT1 protein levels. As a key mediator of type I and II interferon signaling, diminished STAT1 attenuates interferon‐induced phosphorylation and overall pathway responsiveness. Although interferon signaling normally suppresses HSC proliferation, this inhibitory effect is blunted in SRSF2‐mutant cells, potentially conferring a competitive advantage under interferon stimulation [46] (Figure 2C). Thus, attenuated interferon responsiveness may represent a candidate mechanism for age‐dependent selection of SRSF2‐mutant CH.

4.2. Telomere Shortening and SF‐Mutant CH

HSCs undergo progressive telomere shortening with each cell division, and critical telomere attrition activates DNA damage responses and impairs replicative capacity [47]. Telomere length in CH can be assessed at the individual level using genetically predicted or directly measured leukocyte telomere length, and at the clonal level by measuring telomere length in single HSPC‐derived colonies from the same person.

At the individual level, a large‐scale genetic epidemiologic analysis of UK Biobank data showed that genetically predicted leukocyte telomere length differed by CH driver: individuals with DNMT3A‐, TET2‐, and JAK2‐mutant CH tended to have longer telomeres, whereas those with SF‐, PPM1D‐, and TERT promoter‐mutant CH were enriched among individuals with shorter telomeres. Measured leukocyte telomere length further distinguished SF‐mutant CH from other common CH subtypes. High‐VAF DNMT3A‐, TET2‐, and JAK2‐mutant clones were associated with shorter telomeres, consistent with telomere attrition during clonal expansion. In contrast, SF‐mutant CH did not show comparable VAF‐associated telomere shortening; in SRSF2‐mutant CH, higher VAF was instead associated with longer measured leukocyte telomere length [48].

At the clonal level, whole‐genome sequencing of single HSPC‐derived colonies showed that SF‐mutant colonies had longer telomeres than wild‐type or non–SF‐mutant colonies from the same individuals, with an estimated increase of approximately 300 bp. These findings support a model in which age‐related telomere attrition creates a selective pressure, allowing SF‐mutant cells with relatively preserved telomeres to avoid critical telomere shortening and gain a relative advantage (Figure 4).

FIGURE 4.

Proposed model of age‐associated telomere attrition and selection of CH clones. Telomere length progressively shortens in HSCs with age. In individuals with shorter‐than‐average telomere length, age‐related telomere attrition may create a selective pressure under which SF‐mutant HSCs with relatively preserved telomeres gain an advantage over wild‐type HSCs by avoiding critical telomere shortening. In contrast, DNMT3A‐, TET2‐, and JAK2‐mutant clones are more often observed in individuals with longer‐than‐average telomere length; although these clones undergo telomere shortening during expansion, their longer initial telomere reserve may allow continued survival during driver‐mediated clonal expansion before telomeres become critically short. It should be noted that the mechanisms and causal relationships underlying these telomere‐associated selective patterns remain incompletely defined.

The mechanism underlying this relative telomere preservation remains unresolved. One possibility is altered telomerase regulation, as SF3B1 mutation has been reported to increase TERC and TERT expression [49]. Reduced proliferative activity of SF‐mutant HSCs may also limit cumulative cell divisions and thereby attenuate telomere attrition. Clarifying the causal relationship between telomere dynamics and SF‐mutant clonal evolution will require longitudinal and functional studies.

5. Conclusion and Perspectives

Accumulating evidence indicates that SF mutations define a biologically distinctive subset of CH and myeloid neoplasms, characterized by relatively late emergence, accelerated clonal expansion in older individuals, and an increased risk of progression to myeloid neoplasms.

In this review, we summarized molecular consequences of SF mutations, including altered RNA processing, dysregulated innate immune signaling, and R‐loop accumulation, as well as cooperative genetic interactions implicated in myeloid neoplasm pathogenesis. We also discussed CH‐focused and population‐based studies suggesting that age‐associated selective pressures may contribute to the preferential expansion of SF‐mutant clones.

Although not discussed in detail in this review, SF mutation‐derived neoantigens provide an additional hypothesis‐generating link between aberrant splicing and adaptive immunity. Studies in SF‐mutant myeloid neoplasms have shown that recurrent mis‐splicing events can generate HLA‐presented peptides recognized by CD8⁺ T cells, including neoantigens derived from SRSF2‐ or ZRSR2‐associated aberrant splicing events (Figure 5) [50]. Hotspot SF mutant proteins may also generate HLA‐presented peptides recognized by CD8⁺ T cells [51, 52]. Whether such immune recognition influences SF‐mutant clonal dynamics during CH remains unknown. Age‐associated T‐cell dysfunction could theoretically weaken immune control of neoantigen‐bearing clones, but current evidence does not establish neoantigen‐mediated immune surveillance as a mechanism of SF‐mutant CH selection.

FIGURE 5.

Mis‐splicing–derived neoantigens elicit CD8⁺ T‐cell recognition. In SRSF2‐mutant myeloid malignancies, recurrent mis‐splicing events generate shared neoantigen peptides presented on HLA class I molecules. Neoantigen‐reactive CD8⁺ T cells exhibit impaired cytotoxic function, accompanied by downregulation of NF‐κB target genes. (Created with BioRender.com).

Future studies should directly examine the causal roles of these mechanisms in clonal selection during CH. Clarifying the features that distinguish SF‐mutant CH from other driver‐defined subtypes may refine risk stratification and early intervention, while advancing our understanding of age‐associated clonal evolution and leukemogenesis.

Author Contributions

Yang Lyu: writing – original draft, visualization. Yutong Zhang: writing – original draft, visualization. Ayana Kon: conceptualization, supervision, visualization, writing – review and editing.

Funding

This work was supported by research grants awarded to A.K. from the Japanese Society of Hematology, the SGH Foundation, the Japan Philanthropic Foundation, the Cell Science Research Foundation, the Takeda Science Foundation, the Mitsubishi Foundation, and the Ono Pharmaceutical Foundation for Oncology, Immunology, and Neurology.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The authors have nothing to report.