m6a-Son-CCL5 Axis Controls HSC Fate and Inflammation

Blood stem cells intricately regulate their fate, balancing self-renewal and differentiation through symmetric and asymmetric divisions. When exposed to various stresses they can rapidly undergo symmetric commitment divisions to generate differentiated progenitors for immune regeneration or driving inflammation. In previous studies, we and others found that m⁶A RNA methylation controls symmetric commitment and inflammation in hematopoietic stem cells (HSCs) (Cheng and Luo et al., Cell Reports 2019). However, the precise molecular mechanisms underlying the role of m⁶A and its influence on the inflammatory program in stem cells remain elusive.

To uncover the dynamic RNA methylation during HSC commitment, we employed a recently developed tool called DART-seq, that fuses the YTH domain of the m⁶A reader protein YTHDF2 with the RNA editing protein APOBEC1. We identified 300 m⁶A sites and approximately 250-300 gene targets within HSCs and MPPs (padj beta binomial < 0.05). We found m⁶A sites in HSCs were enriched for: innate immune response pathways including NFκB and Toll-like receptor cascades, cellular signaling processes encompassing RhoGTPase and ERK/MAPK signaling targets, and metabolic pathways associated with amino acid transport and fatty acid oxidation. In contrast, the m⁶A targets enriched in MPPs were associated with lineage differentiation and myeloid programs such as MLL, GATA1, PU.1, and CEBPD.

By integrating our findings with two additional m⁶A mapping datasets, Son was a shared m⁶A target with increased m⁶A modification during the transition from HSCs to MPPs. SON, an RNA binding protein residing in nuclear speckles, has been implicated in a myriad of cellular processes including splicing control and transcriptional repression. Notably, de novo heterozygous loss-of-function variants in SON (ZTTK Syndrome) results in hematological symptoms, neurological impairments, and developmental delays.

The genetic deletion of METTL3 (Mettl3 conditional knockout mice; Mettl3 cKO) or METTL3 inhibitor STM2457 treatment, resulted in a twofold increase in Son transcripts in both HSCs and MPPs, but reduced SON protein abundance by 50%. Notably, SON exhibited asymmetric segregation during HSC division, with its abundance displaying a strong correlation with the commitment marker NUMB. Collectively, these findings suggest that m⁶A modification regulates the abundance of SON protein and implicates SON in HSC commitment fate.

Reintroduction of SON in m⁶A-deficient HSCs effectively rescued commitment defects (cKO EV 9.74% vs. cKO+SON 21.43%). Furthermore, SON overexpression partially restored in vivo engraftment defects in Mettl3 cKO LSKs (cKO EV 1.5% vs. cKO+SON 4.5%). Intriguingly, the functional rescue was accomplished using a fragment containing the RNA-binding domain of SON (cKO EV 1.2% vs. cKO+SON-RB 12.3%), highlighting the significance of SON's RNA-binding ability.

Additionally, deletion of SON in RosaCas9 WT LSKs using sgRNAs resulted in a substantial reduction in engraftment (WT EV 35.4% vs. WT+sgSON 11%), whereas SON overexpression in WT LSKs enhanced engraftment (WT EV 6% vs. WT+SON 39%). These results demonstrate that SON is a positive regulator of stem cell engraftment and function.

Moreover, SON overexpression effectively rescued specific clusters identified through scRNA-seq analysis in Mettl3 KO (KOsp1-1 and CCL5+) by 50%. In bulk RNA-seq analysis, we identified 257 upregulated genes and 227 downregulated genes between the Mettl3 cKO EV and Mettl3 cKO SON groups (padj<0.05). Remarkably, SON overexpression significantly downregulated inflammatory pathways that were aberrantly upregulated upon m⁶A loss. Furthermore, SON overexpression indirectly rescued dsRNA formation in Mettl3 cKO LSKs, aligning with the role of m⁶A in suppressing dsRNA formation in HSPCs.

SON overexpression reduced Ccl5 through binding to its transcript and decreasing nascent transcription. Strikingly, depletion of CCL5 significantly rescued Mettl3 cKO LSK engraftment (cKO 1.6% vs. cKO+shCCL5 5.3%), indicating that CCL5 is sufficient to mediate the effects downstream of SON. Moreover, co-culturing WT HSCs with CCL5 resulted in a symmetric commitment defect similar to that observed with m⁶A loss. In summary, our study delineates the critical m⁶A-SON-CCL5 axis governing HSC symmetric commitment fate and inflammation control.