A branched chain amino acid rheostat controls hematopoietic stem cell replicative lifespan Free

Hematopoietic stem cells (HSCs) self-renew and give rise to all mature blood cells upon activation. However, studies using the histone H2B-GFP label-retention model have shown that HSCs progressively lose self-renewal ability with each division, demonstrating that their self-renewal capacity is limited. Additionally, we previously showed that after replicative stress in transplantation studies, HSCs retain abnormal mitochondria that have low mitochondrial membrane potential, and this leads to ineffective hematopoiesis. Our goal is to understand how replicative stress-induced metabolic remodeling affects HSC function with the ultimate goal of targeting these pathways to improve HSC function.

To better understand the effects of cell division on HSC function, we used the doxycycline-inducible H2B-GFP label retaining model and integrated this with mitochondrial activity using tetratmethylrodamine ethyl ester (TMRE) dye to measure mitochondrial membrane potential (MMP). Five months following doxycycline removal, the H2B-GFP labeled population could be subdivided into four functionally distinct subpopulations, differing based on GFP levels and mitochondrial activities generating GFP(G)^Hi;TMRE(T)^lo, G^Hi;T^Hi, G^Lo;T^Lo, G^Lo;T^Hi, which are unexpectedly very heterogeneous in functional behavior. G^Hi;T^Hi and G^Lo;T^Hi engrafted poorly. G^Hi;T^Lo HSCs possessed the highest level of repopulation potential with balanced lineage production in serial competitive repopulation assay. G^Lo;T^Lo HSCs had significantly decreased and myeloid-biased repopulating potential. We focused on the comparison between G^Hi;T^Lo and G^Lo;T^Lo HSCs, as they differ in division history yet maintain similar mitochondrial states. Single cell RNA-sequencing analyses of the bone marrow generated by each group reveal that G^Lo;T^Lo HSCs produce a distinct differentiation landscape with imbalanced mature lineage frequency. Remarkably, each lineage cell derived from G^Lo;T^Lo donor cells had differential gene expression signatures compared to G^Hi;T^Lo donor cells, suggesting committed hematopoietic lineages retain memory of HSC division history.

To understand mechanism, we performed bulk RNA-sequencing. This revealed a significant upregulation of genes involved in branched-chain amino acid (BCAA) catabolism as well as mTOR suppression in G^Hi;T^Lo HSCs compared to G^Lo;T^Lo HSCs. Protein expression of BCAT2, branched chain amino acid transaminase 2 that converts BCAAs into branched chain keto-acids, was higher in G^Hi;T^Lo than in G^Lo;T^Lo HSCs. Consistent with this, global metabolomics analysis showed lower levels of leucine and valine but higher levels of BCAA-derived acylcarnitines in G^Hi;T^Lo HSCs compared to G^Lo;T^Lo HSCs, suggesting enhanced BCAA catabolism. Interestingly, G^Lo;T^Lo HSCs also contained significantly higher levels of ATP, as well as mTOR target phospho-S6 expression, suggesting a higher energy and anabolic state.

Suppressing BCAT2 activity using pharmacological and genetic approaches was sufficient to decrease HSC functions. To determine if G^Lo;T^Lo HSC functions can be rescued, we treated G^Lo;T^Lo HSCs with the leucine catabolic product alpha ketoisocaproate (KIC) in vitro. Upon transplantation, KIC-treated G^Lo;T^Lo HSCs restored their regenerative potential and lymphoid cell production. Importantly, metabolomics analysis showed that G^Lo;T^Lo HSCs treated with KIC had restored low levels of BCAAs and low ATP, but high levels of acylcarnitines, similar to those of G^Hi;T^Lo HSCs, thus showing that KIC treatment boosts BCAA catabolism to maintain lower metabolic activity in HSCs. Mechanistically, the time-to-division in G^Lo;T^Lo HSCs was faster than G^Hi;T^Lo. G^Lo;T^Lo HSCs had increased proportion of EdU positive cells and higher CDK6 expression upon activation. KIC treatment slowed the time-to-division of G^Lo;T^Lo HSCs and decreased the proportion of EdU positive cells and expression of CDK6, suggesting that BCAA usage couples HSC stemness to cell cycle speed.

Taken together, these data suggest that mitochondrial catabolic activity is coupled to stemness and a lower energy state through a branched chain amino acid rheostat, and this is defective as HSCs undergo more cell divisions. Importantly, boosting mitochondrial catabolism with KIC maintains stemness. This provides important insights into the use of metabolite replacement to target division-dependent loss of HSC function, which has therapeutic implications for HSC regenerative medicine.