Hematopoietic stem cell aging promotes TET2 clonal hematopoiesis
Marco De Dominici, Lamis Naddaf, Angelica Varesi, Andrew Goodspeed, Bridget Hoag, Vadym Zaberezhnyy, Johannes Menzel, Yang Liu, Stephanie Xie, Sheng Li, James DeGregori
TL;DR
Aging hematopoietic stem cells become less fit, allowing mutated cells to expand more easily, leading to clonal hematopoiesis.
Contribution
The study reveals that aging reduces the fitness of healthy hematopoietic stem cells, promoting clonal expansion of Tet2-mutant cells.
Findings
Old donor Tet2 KO hematopoietic stem cells expand faster than young ones in a mouse model of clonal hematopoiesis.
Aged hematopoietic stem cells show enhanced RUNX1 activation and ribosomal protein gene expression, inducing a p53-mediated stress response.
Tet2 inactivation abrogates aging-related changes in hematopoietic stem cells, linking these changes to clonal hematopoiesis.
Abstract
Aging is strongly associated with the incidence of clonal hematopoiesis (CH) and myeloid malignancies. However, the role of aging in the clonal selection for CH mutations is not well understood. In a mouse model of CH, we observe that transplanted Tet2 KO hematopoietic stem cells (HSC) from old donor mice expand at a faster rate than young irrespective of the age of the recipient mice; that this acceleration is observed by middle age; and that it is primarily due to the aging-associated reduction in fitness of aged competitor non-mutant HSC. Mechanistically, in both mice and humans, we found that aged HSC exhibit enhanced activation of a RUNX1 transcriptional program and increased expression of ribosomal protein genes inducing a p53-mediated stress response, and that these changes are abrogated by Tet2/TET2 inactivation. Thus, aging creates the conditions that foster clonal expansion of…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAcute Myeloid Leukemia Research · Zebrafish Biomedical Research Applications · Hematopoietic Stem Cell Transplantation
Introduction
Aging of the hematopoietic system is thought to be caused by both cell intrinsic^1^ and cell extrinsic alterations^2^ and is associated with well-characterized phenotypes including increased hematopoietic stem cell (HSC) pool size, myeloid biased hematopoiesis and reduced stem cell clonality. Clonal hematopoiesis of indeterminate potential (CHIP), which is typically defined with a cutoff of 0.02 VAF (~ 4% nucleated cells) increases exponentially with age, and is observed in 10–20% of individuals > 70 years old^3–5^.
CHIP is associated with reduced overall survival^3^, a several-fold increase in the risk of leukemia^6^, and is a significant risk factor for multiple aging-associated diseases^5,7^. Notably, the rate of expansion of CHIP clones and their size correlates with increased risk of leukemia^8^ and higher VAF confers increased risk of cardiovascular disease^9^. Thus, therapies aimed at preventing or reducing the selective advantage of mutant clones hold the promise to significantly reduce morbidity and mortality in the elderly.
TET2 is the second most frequently mutated gene in CHIP after DNMT3A^10^. TET2 modulates chromatin accessibility through 5-methylcytosine oxidization^11^, but can also modulate RNA demethylation, which plays important roles in hematopoiesis^12,13^ and leukemia^14^. Furthermore, TET2 can regulate gene expression by direct interaction with other transcription factors and chromatin modifiers such as OGT^15^, RUNX1^16^ and HDAC2^17^.
There is abundant evidence that inflammatory cues can promote TET2 CH. Experiments in mouse models have shown that IL-6^18^ and IL-1^19^ signaling promote the expansion and malignant progression of Tet2 KO hematopoietic cells. In particular, gut microbial signals have been shown to play a role in the expansion of Tet2 KO cells by mediating IL-6 production^20^, and TNF signaling^21^. Aging is associated with low grade chronic inflammation, often termed inflammaging^22^, and TET2 loss of function has been shown to protect HSC from epigenetic reprogramming due to inflammation^23,24^ and inflammaging^25^. In mouse models, aging-associated gut dysbiosis increases IL-1 signaling in the bone marrow^26^ and the aging-related increase in IL-1 signaling have been shown to promote expansion of Tet2 mutant cells^27^. ADP-heptose has been identified as a specific gut microbial product that can promote the expansion of Dnmt3a and Tet2 mutant hematopoietic stem and progenitor cells (HSPC) as a consequence of loss of intestinal barrier function^28^ Other data point to a cell intrinsic mechanism for Tet2 mutant cell expansion in aged mice, depending on aging-related dysfunction of HSC due to activation of endogenous retroviral elements and induction of an aberrant interferon response, phenotypes that are abrogated by Tet2 inactivation^29^. Additional proposed mechanisms driving selection for Tet2 mutant cells include the role of obesity in creating a pro-inflammatory environment^30^; aberrant expression of the thrombopoietin receptor Mpl in Tet2 mutant cells^31^; upregulation of the transcription factor Id1^32^ and increased cell turnover associated with atherosclerosis^33^, although the latter was not corroborated in an epidemiological study^34^.
In summary, there are presently multiple molecular mechanisms that can explain the selection for TET2 mutant clones in CH and there is a rationale for increased selection for mutant clones with aging. However, it is not yet clear if these represent pleiotropic effects of TET2 or if they are connected by regulatory hubs that can be targeted in order to prevent or delay the development of CHIP. Here we utilized a CRISPR-based approach that allowed us to precisely determine the competitive advantage of Tet2 KO cells and dissect the role of the age of the donor cells or the recipient microenvironment. Our data show that the reduction in fitness of aged HSC creates the conditions fostering the expansion of Tet2 mutant clones. We further describe an aging-dependent mechanism associated with reduced HSC fitness involving aberrant expression of ribosomal protein genes (RPG) and p53 activation through increased RUNX1 activity which are reversed by Tet2/TET2 mutation.
Results
Tet2 KO but not Dnmt3a KO hematopoietic cells display an increased positive clonal selection with aging.
We established a mouse model to dissect the dynamics of Tet2 mutant clonal hematopoiesis. HSC from young (2 months old) or old (20 months old) donor C57BL/6j mice were cultured in a polyvinyl alcohol-based medium that has been shown to preserve the stem cell phenotype and transplantation efficiency of in vitro expanded HSC^35^. Optimized lentiviral vectors^36^ expressing sgRNAs targeting the mouse Tet2 or Dnmt3a gene or a non-coding region in the Rosa26 locus (as control), each expressing different fluorescent proteins, were introduced in separate batches of cells and indels were induced by Cas9 electroporation with high efficiencies (Extended Data Fig. 1a-1b). Subsequently, cells were pooled and injected in the tail vein of age-matched recipient mice conditioned with busulfan which is associated with only a transitory increase in bone marrow inflammation and no significant reduction in bone marrow cellularity^37^ (Fig. 1a).
We determined the rate of clonal expansion by assessing the ratio of Tet2 KO or Dnmt3a KO cells to control cells in the peripheral blood and bone marrow normalized by the ratio in the HSC compartment at the time of transplantation. With this approach, we observed that Dnmt3a KO showed a clonal expansion in young and old mice in peripheral blood and bone marrow populations in both primary and secondary transplants, but the rate of expansion was not significantly affected by aging (Extended Data Fig. 1c-1e). By contrast, Tet2 KO cells expanded significantly faster in old mice injected with old donor cells than in young mice injected with young donor cells in all blood populations (Fig. 1b, 1c). We assessed the ratio of myeloid-biased differentiation of transplanted cells, and we observed that while aging promotes an increased myeloid differentiation in both the blood and bone marrow, Tet2 KO cells were not significantly different than control (WT) cells (Fig. 1d). In the bone marrow, Tet2 KO cells showed increased expansion with aging in all HSPC populations, observable in HSC and further increased in more differentiated populations. Notably, in young mice there was no net advantage of Tet2 KO in HSC since the ratio was close to the expected neutrality; however, old Tet2 KO HSC exhibited an approximately 4-fold relative expansion (Fig. 1e, rightmost panel). This implies that a Tet2 mutation acquired in HSC in youth is not expected to significantly expand over time (and a mutation in a committed clone would be extinguished due to differentiation), while in an aging context Tet2 mutations confers a clonal advantage to HSC with a consequent higher chance of establishing CHIP.
The clonal advantage of Tet2 mutant cells is observable by middle age and is dependent on the age of the donor cells.
To further characterize the age and sex-specific dynamics of Tet2 mutant clonal advantage, we performed additional experiment by transplanting Tet2/Rosa26 CRISPR cells of donor mice of different ages (2, 7, 13 and 20 months) and of both sexes in matched recipient mice. We observed that Tet2 KO cells show similar aging-dependent clonal advantage in both male and female mice. Furthermore, the acceleration of Tet2 KO expansion is observable by middle age (7–13 months) in the blood and bone marrow HSC and myeloid progenitor cells (MyP) of female (Fig. 2a, 2b) and male mice (Fig. 2c, 2d). To understand the contribution from the donor or the recipient age in the expansion of Tet2 KO cells we performed transplantations of Tet2/Rosa26 CRISPR cells in mice with all possible age combinations (Fig. 2e). These experiments revealed that the rate of expansion of Tet2 KO cells is entirely determined by the age of the donor mice and not the age of the recipient mice (Fig. 2f), and is thus HSC intrinsic. We also observed that both donor and recipient age contribute to the increased myeloid differentiation of transplanted cells, with a greater contribution of donor than recipient age (Fig. 2g).
The aging-dependent expansion of Tet2 KO hematopoietic cells is due to reduced fitness of old WT cells
The aging-associated expansion of Tet2 KO hematopoietic cells could be due to an increase in fitness conferred by Tet2 KO to old HSC, a reduced fitness of old WT HSC clones which are ineffective competitors, or a combination of these two factors. To disentangle these possibilities, we performed competitive transplants of donor HSC populations of different ages or Tet2 genotypes in young recipient mice (Fig. 3a). In agreement with our previous results, we observed a clear advantage of old Tet2 KO cells vs old WT cells in total peripheral blood cells and in neutrophils, which, given their shorter half-life should better reflect the dynamics in bone marrow HSPC pools^38^. When old Tet2 KO HSC were transplanted in competition with young WT HSC the rate of expansion was significantly diminished, suggesting that young WT HSC are better competitors than old WT HSC. Interestingly, old Tet2 KO HSC were at a disadvantage when compared to young Tet2 KO HSC, suggesting that Tet2 inactivation does not provide a greater fitness benefit when acquired in old HSC. Finally, old WT HSC were at a strong disadvantage when transplanted against young WT HSC (Fig 3b), confirming previous observation of reduced engraftment potential of HSC with aging^39,40^. We further analyzed the relative fitness of the transplanted population in the bone marrow, and the data largely overlap with the fitness differential in the peripheral blood. Notably, we observed that old Tet2 KO HSC in competition with either young or old WT HSC show a trend toward higher relative rates of expansion in more committed progenitors, particularly myeloid committed progenitors (MyP and MPP3), than in HSC (Fig. 3c). These results suggest that the competitive advantage of Tet2 KO cells is in part due to increased fitness in HSC and in part is due to further clonal expansion in more committed progenitors, in agreement with our previous observations (Fig. 1e).
The aging-dependent myeloid bias of HSC does not promote selection for Tet2 KO cells.
HSC aging is associated with an increase in the proportion of myeloid biased HSC (MyHSC) in mice and humans^41,42^. MyHSC can be defined by high expression of CD150 and CD41^43^ or IL27Ra^44^, and are characterized by reduced engraftment potential as opposed to the CD150^low^ lymphoid biased HSC (LyHSC)^44,45^. MyHSC frequency was shown to increase by middle age^43^ in mice which we also observed (Extended Data Fig. 2). Given that we observed increased selection for Tet2 KO HSPC in middle age, we sought to determine if the cell of origin in which the Tet2 mutation is acquired determines the rate of Tet2 selection. To test this hypothesis, we sorted MyHSC, LyHSC along with ST-HSC or MPP from 20 months old mice and performed Tet2/Rosa26 CRISPR in each population separately followed by transplantation in young mice (Fig. 4a). In the peripheral blood (6 weeks post transplantation), we did not observe a different rate of expansion of Tet2 KO cells from LyHSC or MyHSC. We did however observe that when the mutation is introduced in ST-HSC or MPP, there is no net selection for Tet2 KO cells (Fig. 4b). We confirmed that MyHSC (both Tet2 KO and WT) gave rise to higher proportion of myeloid to lymphoid differentiated cells as compared to transplanted LyHSC, confirming the conserved myeloid bias after transplant (Fig. 4c). In the bone marrow, transplanted cells from sorted donor LyHSC or MyHSC did not show significant differences in the rate of Tet2 KO expansion in any recipient population including in recipient LyHSC and MyHSC, which were both effectively reconstituted from either HSC donor population suggesting plasticity between the two compartments (Extended Data Fig. 3). Overall, these data indicate that the acquisition of the Tet2 mutation in MyHSC or in LyHSC results in similarly increased selection for Tet2 KO HSC with aging.
Tet2 inactivation increases the self-renewal of multipotent progenitor cells.
Previous experiments have shown that multipotent progenitors are incapable of long-term engraftment after transplantation into irradiated mice^46,47^. However, Tet2 inactivation and especially Dnmt3a inactivation can promote increased reconstitution from transplanted ST-HSC (also known as MPP1)^47^. In our transplant protocol, we noticed efficient engraftment and hematopoietic reconstitution in the blood of WT and Tet2 KO stem and progenitor populations up to 6 weeks post-injection. While as expected LyHSC and MyHSC led to long term engraftment of both Tet2 KO and WT populations (albeit the WT population was gradually outcompeted by Tet2 KO cells), by 9 weeks, there was a significant decline of hematopoietic contributions from WT ST-HSC and complete loss of contributions from WT MPP. In contrast, Tet2 KO ST-HSC were unaffected, and Tet2 KO MPP only showed a moderate decline (Fig. 4d). In the bone marrow at 12 weeks, recipient mice showed strikingly reduced fractions of donor WT ST-HSC and almost no contribution from WT MPP while the Tet2 KO fractions were much less affected (Fig. 4e). Given that at earlier time points in the blood we did not observe selection for Tet2 KO cells, this is unlikely to be due to WT cells being removed by the competitive advantage of Tet2 KO cells. Rather, these observations suggest that Tet2 mutant progenitor cells have increased self-renewal potential compared to WT counterparts.
Tet2 inactivation alters myeloid cell response to aging and inflammation
To dissect the molecular alterations caused by Tet2 inactivation in hematopoietic cells, we performed single cell RNA-sequencing (scRNAseq) on Tet2 KO or WT HSPC from bone marrow of young (2 months old), middle aged (12 months old), or old mice (20 months old) following competitive transplantations.
First, we looked at the effect of Tet2 inactivation in HSC independent of age of the donor cells. We selected differentially expressed genes (DEG) common to all three age groups and performed over-representation analysis. Age-independent Tet2 KO upregulated genes in HSC are highly enriched for pathways affecting G2-M progression and sister chromatid separation (Fig. 5a). Overlaying the relative expression of the cell replication marker Ki67 on the cell Uniform Manifold Approximation and Projection (UMAP) representation, revealed widespread gain of proliferation in multiple Tet2 KO populations, particularly notable in middle age (Fig. 5b).
We next focused on myeloid cells given that a well characterized feature of aging is myeloid biased hematopoiesis, and TET2 inactivation results in significant alterations in granulocyte^48^ and monocyte functions^17,49,50^. In granulocytes, top DEG in Tet2 KO cells from all ages largely overlap with previously observed changes in neutrophils^48^ (Fig. 5c). Middle aged and old WT granulocytes showed upregulation of inflammatory pathways and reduced signatures related to cell cycle progression while Tet2 KO middle-aged and old granulocytes showed reversal of these aging-associated changes (Fig. 5d).
Macrophages exhibited a similar pattern in which aging-induced alterations were often reversed in Tet2 KO cells. Specifically, Tet2 KO ablated the age-related increases in pathways related to chemokine production, cell adhesion, LPS responses, and apoptosis while reversing the downregulation of E2F target genes, mitochondrial ribosomal components and response to IL4 (Fig. 5e), the latter being shown to play important role in the function of tumor-associated macrophages^51,52^. These data show that hematopoietic aging can create conditions that substantially alter the phenotypes of Tet2 KO myeloid cells.
Tet2 inactivation reverses aging-related alterations in HSC
We next looked at the transcriptional differences within HSC since our data suggest their relative fitness is a critical determinant for Tet2 mutant cells expansion. First, we estimated cell cycle distributions from transcriptional states, and we observed a larger fraction of cycling Tet2 KO HSC at older ages (Extended Data Fig. 4) consistent with previous results^29^.
Next, we looked more broadly at aging and Tet2-modulated transcriptional changes in HSC. We observed that aging in WT cells (OW vs YW) was associated with a predominant representation of RPG among upregulated genes (Fig. 6a). Interestingly, almost all aging upregulated DEG were coordinately downregulated in aged Tet2 KO cells (Fig. 6b).
Conversely, aging-dependent downregulated genes did not show restoration to young-like levels in Tet2 KO cells but were also slightly downregulated (Fig. 6c), consistent with widespread transcriptional repression following TET2 inactivaton^11,53^. RPG were upregulated by middle age in most hematopoietic populations; however, the reversal of this upregulation by Tet2 KO was mostly observed in HSPC (Fig. 6d–6e). We also observed an aging-dependent increase in nuclear encoded genes related to the electron transport chain (ETC), predominantly belonging to complex I (Fig. 6f). Tet2 KO reversed this upregulation in most hematopoietic populations.
To validate if these transcriptional changes result in altered mitochondrial and ribosomal functions, we performed additional tests on transplanted Tet2 KO and WT HSPC in young and old mice. We detected no difference in protein synthesis rate (OPP), mitochondrial membrane potential (TMRE), reactive oxygen species (CellROX, which correlate with the activity of ETC Complex I), and for mitochondrial content (Mitotracker Green, MTG) based on age or Tet2 genotype (Extended Data Fig. 5a-5d). We then sought to determine if pharmacological manipulation of these pathways can affect the rate of Tet2 KO cell expansion. Rapamycin, through inhibition of mTORC1, can reduce mitochondrial and ribosomal biogenesis and has been shown to extend life-span and health-span in mice^54,55^ and functionally rejuvenate HSC^56^. We therefore tested whether rapamycin treatment can reduce the selection of Tet2 KO cells with aging (Extended Data Fig. 5e). We first verified that a single injection of rapamycin (4 mg/kg) can modulate mTORC1 signaling by reducing S6 kinase phosphorylation in HSPC in vivo (Extended Data Fig. 5f). Rapamycin treatment was then given to donor mice before Tet2 CRISPR, in recipient mice after the transplant, or both, and in all cases, the treatment did not affect the rate of expansion of Tet2 KO HSPC cells (Extended Data Fig. 5g).
These data argue against direct effects of aging and Tet2 KO on mitochondrial and ribosomal functions. We then examined more closely the aging and Tet2-dependent transcriptional changes in HSC. The top signatures by GSEA included, as expected, Structural Constituent of the Ribosome and Oxidative Phosphorylation. We also noticed an aging-associated increase in TP53 Regulated Metabolic Genes, in RUNX1-Regulated HSC Genes and in Hematopoietic Stem Cell Differentiation (likely reflecting the increased RUNX1 transcriptional activation), which were reversed in Tet2 KO (Fig. 6g–6h). Many ribosomal proteins are known to increase p53 functions in conditions of nucleolar stress, when not assembled into ribosomes^57^. Interestingly, it has previously been shown that RUNX1 transcriptionally activates genes associated with ribosomal biogenesis and modulates p53 activation in HSC^58^. Furthermore, TET2 has been shown to colocalize with RUNX1 at the chromatin of HSC^16^. Together these findings lend support to a model in which TET2 in cooperation with RUNX1 positively regulates the expression of RPG and increases the activation of p53 with aging while Tet2 KO cells are protected from this stress signaling (Fig. 7a), contributing to selection for Tet2 mutations in aged HSC pools.
HSC aging is associated with increased RUNX1 activity, RPG expression and p53 transcriptional programs in mice and humans
In order to more closely characterize the overlap between TET2 and RUNX1 transcriptional programs, we looked at the expression in our scRNAseq data of previously described RUNX1 target genes involved in ribosomal biogenesis^58^. Among this set of genes, 70% were increased in middle-aged (MW vs YW) and 55% were increased in old (OW vs YW) HSC. Of these upregulated genes, 92% were reduced in middle aged Tet2 KO HSC and 85% were reduced in old Tet2 KO HSC (Fig. 7b), again indicating that Tet2 inactivation reverses gene expression changes induced by aging. We then focused on a subset of RPG with known roles in regulating p53 (Supplementary Table 1) and we found that the genes fall into two clusters. Cluster 1 genes are upregulated by aging in both middle age and old WT HSC, and cluster 2 consist of genes that show no net change in middle age and a reduction in old age. Strikingly, almost all of the genes were downregulated by Tet2 KO in middle-aged and old mice, including cluster 2 genes in young mice (Fig. 7c), suggesting that Tet2 KO suppress the expression of a larger subset of p53-modulating RPG in addition to reversing the increased expression observed in aging.
To assess whether the aging-related alterations that we discovered are conserved in humans, we performed additional analyses from previously published sequencing datasets re-analyzed to include RPG^25,59^. We focused on key ribosomal signatures upregulated by aging and reduced by Tet2 KO, and observed clear upregulation in bone marrow HSC from middle aged and old individuals compared to young individuals (Fig. 7d). These previous studies^25,59^ have identified “inflammatory memory HSC” (HSC-iM), a subset of HSC enriched in aged individuals and in CHIP and characterized by increased inflammatory transcriptional responses. Ribosomal gene signatures appear to be upregulated specifically in HSC-iM and show lower enrichment in the non-inflammatory HSC-I subset. We also analyzed mutant TET2 cells compared to the WT counterpart within the same CHIP donor subjects obtained by TARGET-seq. Critically, we found a clear downregulation of these signatures in TET2-mutant cells (Fig. 7f).
It was previously observed that ribosomal RNA expression is reduced in old HSC^60^, which could exacerbate nucleolar stress by further unbalancing the stoichiometric ratio of ribosomal proteins to rRNA. To corroborate these findings, we performed rRNA FISH-Flow experiments with previously established protocols^61,62^, and we confirmed reduced expression of the 47S pre-rRNA as well as 18S and 28S rRNA in HSC and MPP but not in more mature MyP (Fig. 7g). Additionally, by GSEA we found consistent positive enrichment of the Hallmark p53 signature in aged HSC from multiple human scRNAseq studies^25,41,63–65^ (Fig. 7h) and downregulated p53 signaling in TET2 mutant HSC2 from Jakobsen et al.^25^ but not in HSC1 (corresponding to HSC-iM and HSC-I respectively in Zeng et al.^59^). These analyses support the presence in HSC of an aging-dependent TET2/RUNX1/RPG program, which together with reduced rRNA abundance, leads to p53 activation, particularly in the most inflammation-responsive HSC-iM/HSC-II subset.
RUNX1 or p53 inactivation neutralizes the competitive advantage of Tet2 KO cells in aging.
Our hypothesis suggests that the clonal advantage of Tet2 inactivation in aged HSC depends in part on the negative regulation of RUNX1 and p53 functions. Consequently, we expect that old Tet2 KO cells should display reduced fitness advantage when in competition with Runx1 KO or old Trp53 KO cells. In addition, given that RUNX1 and p53 pathway are activated in HSC with aging, we might expect increase selection for Runx1 and Trp53 KO HSC in an aged context. To test these hypotheses, we performed competitive transplants to assess the relative fitness advantages of these populations (Fig. 8a and 8b). In the peripheral blood, we observed accelerated expansion of Runx1 KO vs WT cells in old compared to young mice; old Tet2 KO cells showed the fastest rate of expansion when competing with old WT cells, but a significantly lower advantage when competing with old Runx1 KO cells (Fig. 8c). In the case of p53, Trp53 KO cells showed a clonal advantage in peripheral blood in both young and old mice, which was more pronounced in young mice. As before, the largest clonal advantage was observed in old Tet2 KO cells competing with old WT cells, but this advantage was essentially ablated when old Tet2 KO cells were competing with old Trp53 KO cells (Fig. 8d). In the bone marrow, Runx1 KO cells showed increased clonal selection with aging within HSC but not MyP, suggesting that Runx1 inactivation is more positively selected in aged HSC. Notably when old Tet2 KO cells were transplanted with Runx1 KO cells the competitive advantage of Tet2 KO HSC was completely ablated in stem and progenitor cells (Fig. 8e). A similar pattern was observed with regard to p53. Trp53 was more positively selected in old than in young HSC (but not MyP), and competitor Trp53 KO cells significantly reduced the clonal advantage of old Tet2 KO MyP and HSC (Fig. 8f). These data show that inactivation of p53 and RUNX1 in old HSC equalizes the competition with old Tet2 KO HSC supporting that these genes are part of a shared pathway that reduces the fitness of aged HSC.
Discussion
Rare clones of HSPC harboring CH mutation are found ubiquitously in healthy adults^66,67^, and yet CHIP (VAF ≥ 2%) is most prevalent in the elderly and its incidence increases exponentially with age. Previous studies based on direct observations and phylogenetic analyses concluded that CH mutations are commonly acquired early in life^68–70^, possibly even in utero, and expand on average at a slow pace over many decades. Whether this expansion is constant throughout life or if it can be modulated by cell-intrinsic or cell-extrinsic alterations in the bone marrow is still a matter of debate. It was proposed that CHIP clones show deceleration with aging^70^; however, this deceleration was less evident with TET2 CHIP and could be partially due to increased competition with highly fit clones lacking detectable CH mutations that become more frequent in old age^69,70^. Drawing firm conclusions is complicated by the fact that we currently lack a comprehensive characterization of the clonal dynamics of mutant clones across VAF spectrums and human lifespans.
In the present study, we observe accelerated expansion of Tet2 mutant HSC with aging noticeable in mice of both sexes by 7–13 months of age, corresponding to approximately 30–50 years old in humans^71^. WT HSC show remarkable reduction in fitness with aging which is tempered by Tet2 KO. The larger fitness differential resulting from Tet2 KO in old HSC pools is critical for promoting the clonal advantage of Tet2 KO cells. Notably, these fitness differentials are determined by the age of the host cells and are not affected by the recipient microenvironment, suggesting that HSC-intrinsic alterations determine the age-dependent clonal advantage of Tet2 KO cells. We previously showed that aging promotes selection of activated oncogenes in HSPC with a similar mechanism whereby reduced competitive fitness of the aged non-mutant cells leads to the expansion of Bcr-Abl expressing B-cell progenitors^72^. In contrast, in a model of Nras^G12V^ mutation mediated leukemogenesis, the fitness differential was modulated by aging-associated inflammation in the recipient mice^37^, suggesting that cell-extrinsic and cell-intrinsic mechanism play a context-dependent role in the aging-dependent selection for mutant hematopoietic clones.
Reduced fitness of HSC with aging is a well characterized phenotype in mice and humans^39,40,45^. Previous data showed that the reduction of fitness in old HSC is primarily observed in myeloid-biased HSC^45^. Our data do not support increased selection for Tet2 mutant clones originating from MyHSC as opposed to LyHSC, although we cannot exclude that other markers or transcriptional states can separate HSC subpopulations with different fitness levels and different propensities in fostering mutant Tet2 clonal expansion, as shown by the fact that HSC become increasingly heterogenous with age^25,59^. In that regard, it is notable that in human samples we detected the aging-dependent alteration in ribosomal/p53 pathways in the HSC-iM/HSC2 population but not in the HSC-I/HSC1 population, consistent with the notion that aging does not uniformly affect the entirety of the HSC population and a subset of HSC retains young like features^25,73,74^.
Our scRNAseq analysis showed that aging and Tet2 inactivation result in opposite transcriptional effects in several hematopoietic populations, suggesting that Tet2 KO can promote increased fitness by restoring a young-like transcriptional profile. A recent in vivo CRISPR-Cas9 screen has suggested that the clonal advantage of Tet2 KO HSC is dependent on iron availability to promote mitochondrial metabolism^75^. Interestingly, labile iron concentration is increased in old HSC and compromises their fitness, which might create the conditions for increased selection for Tet2 mutant clones^76^. Although we did not observe significant differences in markers of translation or mitochondrial metabolism in ex vivo WT and Tet2 KO from young or old mice, we cannot exclude that more sophisticated analyses (which are technically challenging from a low number of engrafted HSC) might show phenotypic differences contributing to the differential fitness of Tet2 KO HSC in aging.
We observed instead clear enrichment of RUNX1 and p53 signatures in old HSC which are repressed by Tet2/TET2 inactivation in mice and humans. It was previously showed that TET2 and RUNX1 interact to coordinately regulate gene transcription^16^, and activation of TET2 in HSC with retinoic acid and ascorbate was shown to induce RUNX1 transcriptional activity^77^. Furthermore, RUNX1 positively regulates the expression of ribosomal biogenesis genes in HSC including a number of RPG^58^. Increased RPG expression in old HSC has been previously described in mice^78^ and humans^65^. However, opposite effects on RUNX1 and RPG regulation in aged HSC have also been observed^79^. We hypothesized that these transcriptional changes suggest increased nucleolar stress in aged HSC, which is a stress response characterized by the increased expression of unassembled ribosomal proteins activating p53 activity by direct association with MDM2^57^ or with MDM2-independent mechanisms^80^.
In support of this hypothesis, we validated increased p53 transcriptional signatures in multiple datasets of aged human HSC (Fig. 6g and ref^81^) and reduced p53 activation in TET2 mutant inflammatory HSC. It should be noted that p53 activation in aged HSC can be induced by multiple insults, including increased DNA damage which is regarded as hallmark of aging in multiple tissues. Interestingly, aged HSC have been shown to accumulate γH2AX selectively in the nucleolus which has been proposed to cause reduced rRNA synthesis^60^, and Tet2 KO has recently been reported to reduce γH2AX foci in HSC^82^. Coupled with our observed increased expression of RPG with aging, these observations point to a cooperative mechanism leading to increased p53 activation in aged HSC reversed by TET2 inactivation.
Previous studies have largely focused on the role for inflammation in driving TET2 CH^18–20,23,25,30^, and in an aging context IL-1 was proposed as an important factor in driving selection of Tet2 KO HSC^27^. A different study described a cell-intrinsic mechanisms based on TET2-dependent dysregulation of the heterochromatin and aberrant expression of retroviral elements with aging as a driver for Tet2 KO clonal expansion^29^. Interestingly, it has recently been shown how inflammatory cues can induce epigenetic modifications in HSC driving LINE-1 expression and increasing selection for Tet2 mutations^82^, thus providing a connection between cell-extrinsic and cell-intrinsic alterations. In a similar vein, we speculate that transcriptional alterations that we observed in old HSC are the product of an altered microenvironment in aged mice, possibly connected to lifelong inflammatory cues. Cell extrinsic cues likely become “hardwired” in HSC through epigenetic changes leading to aging phenotypes. Several aging-dependent cell-intrinsic alterations in HSC have been described, some of which can be modulated by pharmacological agents^1^. Treatments or interventions aimed at preserving HSC fitness by targeting age-associated changes hold the promise of reducing or delaying the onset of CHIP, myeloid malignancies and other CHIP-associated aging-related diseases.
Online Methods
Mouse models
Young C57BL/6j mice (2 months old) were obtained from Jackson laboratories or from the National institute on Aging (NIA) repository. Old C57BL/6J mice (20–22 months old) were obtained from the NIA. Mice were acclimatized for 2–4 weeks at the University of Colorado Anschutz Medical Campus Laboratory Animal Shared Resources before being used for experiments. Mice were housed in specific pathogen-free animal facilities, maintained at 21 °C (±1 °C) and 35% humidity with a 14 h:10 h light:dark cycle (light 06:00–20:00). Animal protocols were approved by the CU Anschutz Institutional Animal Care and Use Committee in accordance with the NIH Guidelines for the Care and Use of Laboratory Animals.
Rapamycin intervention
Rapamycin was used as previously described^83^. Rapamycin was dissolved in DMSO at 20 mg/mL and stored at −80 °C. At the moment of use, Rapamycin solution was diluted with 10% PEG-300, 5%Tween 80, 85% PBS to a final concentration of 0.4 mg/mL. Mice were treated injected intraperitoneally (4 mg/kg) 3 days per week for 8 consecutive weeks.
Bone marrow isolation
Mice were sacrificed by CO_2_ asphyxiation and immediately placed on ice. For transplantation experiments mouse carcasses were disinfected with ethanol and the tissues processed in a sterile environment in a biosafety cabinet. Bones (femurs, tibia, iliac crests and spine) were removed with a scalpel, placed in a sterile mortar containing 15 mL of FACS buffer (PBS, 2% FBS, 2 mM EDTA) ad crushed. The suspension of bone marrow cells was passed through a 70 μm cell strainer, centrifuged at 500 g for 5 minutes at 4 °C. Red blood cells were hemolyzed with ammonium-chloride-potassium (ACK) lysis buffer (8 g/L ammonium chloride, 1g/L potassium hydrogen carbonate, 0.2 mM EDTA).
To remove dead cells and debris cells were centrifuged on top of a layer of Histopaque-1119 and the upper layer was collected. For HSC sorting, bone marrow cells were incubated with CD117 magnetic microbeads and CD117^+^ cells were enriched by magnetic separation on LS columns (Miltenyi). HSC were defined ad lineage^NEG^ (Ter119, CD3e, CD8, B220, CD11b, GR1), Sca^+^, Kit^+^, CD48^NEG^, CD150^+^. HSC were sorted with the MoFlo XDP100 or the Sony MA900 cell sorter.
Lentiviral production
HEK293T cells were cultured in DMEM high glucose, pyruvate, supplemented with 10% FBS, 1% Pen/Strep in 10 cm dish at 37 °C, 5 % CO_2_ to 50–70% confluence. The medium was changed 2 hours prior to the transfection and 15 μM chloroquine was added. A transfection mix was prepared by combining the lentiviral plasmids: pLCV2-gRNA (10 μg/plate), pSPAX2 (10 μg/plate), pMDG2.G (6 μg/plate) with PEI Max (at a 1:3 mass ratio) in OptiMEM medium (1 mL per 10 cm plate). The mix was shaken by vortex for 30 seconds, incubated for 15 minutes at room temperature and then added drop by drop to the HEK293T cells and mixed by swirling. The medium was changed at 24 hours after the transfection and collected at 48 hours. Fresh medium was added at 48 hours and collected at 72 hours. The collected medium was filtered (0.45 μm) aliquoted in high-speed PPCO centrifuge tubes. Following a previous protocol^84^ a layer of 5 mL sterile sucrose-TNE (20% (w/v) sucrose, 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 0.5 mM EDTA) was gently deposited at the bottom of the tube, then tubes were centrifuged at 40,000 g for 2 hours at 4 °C. The supernatant was carefully aspirated, the pellet was resuspended in a Polyvinyl alcohol based medium (HAM-F12 medium supplemented with 100 U/mL Penicillin/Streptomycin, 2mM L-Glutamine, 10 mM HEPES, 1X Insulin, Transferrin, Selenium, Ethanolamine, 0.1% PVA) and stored at −80 °C.
CRISPR protocol
Sorted HSC cells were plated in fibronectin coated plates in HAM-F12 medium supplemented with 100 U/mL Penicillin/Streptomycin, 2 mM L-Glutamine, 10 mM HEPES, 1X Insulin, Transferrin, Selenium, Ethanolamine, 0.1% Polyvinyl alcohol, 10 ng/mL recombinant mouse SCF and 100 ng/mL recombinant mouse TPO. The medium was changed every 2–3 days by gently aspirating the supernatant, spinning down the detached cells and adding them back to the well containing fresh medium. 3–5 days after sorting cells were transduced with lentiviruses on fresh fibronectin plates by adding the lentiviruses and spinning the plates at 1000 g for one hour at 32 °C. The amount of lentivirus was established by titering on mouse HSPC cells to obtain > 90% transduction efficiency. The cells were incubated with the lentivirus for 48 hours, then the medium was replaced and 48 hours later cells were electroporated with Cas9 mRNA. The Cas9 mRNA was produced in house by in vitro transcription with the T7-FlashScribe^™^ Transcription Kit (Cellscript) using the pCas9-PolyA plasmid (digested with SpeI and NheI) as template, the RNA was purified with the Monarch RNA clean-up kit (NEB) and 5’ capped with the ScriptCap^™^ Cap 1 Capping System (Cellscript), then purified again, resuspended in RNase-free water at 1.8 μg/μL and stored at −80 °C. The electroporation was carried out with the Neon^™^ Transfection System 10 μL Kit (ThermoFisher). Batches of up to 500,000 cells were resuspended in 12 μL of buffer T containing 1.8 μg of Cas9 mRNA, of which 10 μL were electroporated at 1600 V, 10 ms, 3 pulses and plated in PVA medium without antibiotics. Cells were expanded for 7 days and then transplanted into recipient mice. An aliquot of the cells used for transplant was analyzed by flow cytometry to assess the EGFP/dTomato ratio and used as the time zero normalizer to assess the ratio of expansion. A portion of the cells was also used for genomic DNA purification with the DNeasy blood and tissue kit (Qiagen) followed by PCR with primer spanning the CRISPR target locus. PCR products were Sanger sequenced and analyzed by TIDE for assessing CRISPR efficiency.
Bone marrow transplantation
Mice were conditioned by intraperitoneal injection of busulfan 4 days prior to the cell injection and anti-CD4, anti-CD8 2 days prior to the cell injection based on our previous optimizations^37,85^. Busulfan was dissolved in DMSO at 20 mg/mL. At the moment of administration, the busulfan solution was diluted to 2 mg/mL with pre-warmed sterile saline solution (0.9 g/L NaCl in water) and kept warm (~60 °C). Mice were injected intraperitoneally with 10 μL/g of body weight with the diluted busulfan. Anti-CD4 and anti-CD8 antibodies were diluted in sterile PBS and mice were administered 30 μg of each antibody in 200 μL by intraperitoneal injection. Cells were detached from fibronectin plates by pipetting and collecting the medium, then saline, 2 mM EDTA was used to rinse the wells and combined with the cell suspension. Cells were centrifuged at 500 g for 5 minutes at room temperature, resuspended in sterile PBS and 200 μL of cell suspension was injected into the tail vein of conditioned recipient mice. Given that the engraftment potential of HSC on a per cell basis varies with age rather than injecting a specific cell number, a constant fraction of donor to recipient bone marrow was utilized within experiments (typically one donor for 5 to 10 recipients).
Peripheral blood analysis
Peripheral blood was obtained by facial vein puncture and collection of about two drops of blood in a 15 mL conical tube containing 4 mL of FACS buffer. Tubes were centrifuged at 500 g, 5 minutes at 4 °C. Red blood cells were removed by hemolysis with 2 mL ACK buffer for 5 minutes at room temperature, then 8 mL of FACS buffer was added and cells were centrifuged at 200 g for 5 minutes at 4 °C. pellets were washed one more time, strained through a 50 μm membrane, stained with antibodies cocktail (see Supplementary Methods for details) in 50 μL of FACS buffer for 30 minutes, at 4 °C, in the dark. Cells were washed with 1 mL of FACS buffer, resuspended in 80 μL of FACS buffer + DAPI 100 ng/mL and analyzed.
Bone marrow analysis
Bone marrow was isolated by crushing bones (femurs, tibias, iliac crests) in 15 mL FACS buffer. After red blood cell lysis and debris removal with Histopaque 1119 as described in the section “Bone marrow isolation” a portion of bone marrow cells was stained with “mature” and “progenitors” antibody panels. The remaining cells were CD117 enriched and stained for “stem” markers (see Supplementary Methods for details).
rRNA FISH-Flow.
FISH-Flow for ribosomal RNA was performed as described previously^62^ with minor modifications. Bone marrow cells were isolated and CD117 enriched as described above from young and old mice. Cells were stained with cell surface antibody cocktail (see Supplementary Methods for details) for 30 minutes at 4 °C. Cells were fixed in 5 mL of 4% formaldehyde in PBS for 15 minutes at RT with gentle rocking. Samples were centrifuged at 500 g for 5 minutes at RT, washed with 1 mL FACS buffer and then with 1 mL FISH-Flow wash buffer (2X SSC buffer with 10% formamide). Cells were resuspended in 100 μL of hybridization buffer (2X SSC buffer with 10% formamide and 10% dextran sulfate) with a subset of the 18S, 28S and 47S rRNA probes described in Anthony et al.^61^ and incubated for 3 hours at 30 °C. Specifically, probes were obtained from Biosearch Technologies and resuspended in TE buffer at 25 μM. Fluorescein conjugated rRNA probes for 18S (#1 and #8) were used at 1.25 μM final concentration each. Quasar 570 conjugated 28S probes (#2 and #8) were used at 1.25 μM final concentration each and Quasar 670 conjugated 47S probes (#4 and #7) were used at 2.5 μM final concentration each. All probes were used in the same mix and distinguished by the different colors. Following the incubation, 1 mL of FISH-Flow wash buffer was added, cells were centrifuged at 1000 g for 5 minutes, resuspended in 1 mL FISH-Flow wash buffer and incubated with mild agitation for 30 minutes at 30 °C to remove the unbound probes. Samples were centrifuged and resuspended in 500 μL of FISH-Flow wash buffer containing 1 μg/mL DAPI and incubated for 30 minutes at 30 °C with mild agitation. Cells were washed with 1 mL of 2X SSC buffer, then resuspended in FACS buffer with 0.1 μg/mL DAPI and analyzed by flow cytometry.
Ribosomal and mitochondrial functional assays
To assess ribosomal translational efficiency Tet2 KO GFP+ cells and Rosa26 dTomato+ cells were competitively transplanted into young and old mice from age-matched donors. Staining was performed as described previously^86^. At the end point mice were injected intraperitoneally with O-Propargyl Puromycin 50 mg/kg and exactly one hour later were sacrificed by CO_2_ asphyxiation, immediately placed on ice and bone marrow isolated as described above. CD117-enriched bone marrow cells were stained with antibody cocktail (See Supplementary Methods for details) in 100 μL of FACS buffer for 30 minutes on ice. Cells were washed once with 1 mL of FACS buffer, fixed (15 minutes at room temperature, in the dark) and permeabilized (5 minutes at room temperature, in the dark) with the BD Cytofix/Cytoperm Fixation/Permeabilization Kit. Then to visualize intracellularly bound OPP, cells were stained with the Click-&-Go Plus 647 OPP kit. The Alexa Fluor 647 mean fluorescent intensity was tested in GFP+ or in dTomato+ HSC (lineage^NEG^, c-Kit^+^, Sca^+^, CD48^NEG^, CD150^+^).
For mitochondrial membrane potential assay, non-competitive transplants were set up with Tet2 KO GFP+ or Rosa26 GFP+ cells from young or old mice into age-matched recipients. At the end point mice were sacrificed, bone marrow cells isolated, CD117 enriched and stained with 100 nM Tetramethylrhodamine Ethyl Ester (TMRE) with 50 μM Verapamil (to avoid efflux of the dye^87^) in 500 μL of HAM-F12 medium at 37 °C for 30 minutes in the dark. Next cells were stained with cell surface antibody cocktail (see Supplementary Methods for details) in 100 μL of FACS buffer with 50 μM verapamil for 15 minutes at room temperature in the dark. Cells were washed once with 1 mL room temperature FACS buffer with 50 μM verapamil, then resuspended in 200 μL of FACS buffer with 50 μM verapamil and analyzed by flow cytometry. The TMRE mean fluorescent intensity (PE channel) was assessed in dTomato+ HSC cells (Near-IR^NEG^, lineage^NEG^, c-Kit^+^, Sca^+^, CD135^NEG^, CD48^NEG^, CD150^+^) from Tet2 KO or control transplanted mice.
For mitochondrial content and reactive oxygen species quantification, non-competitive transplants were set up with Tet2 KO dTomato+ or Rosa26 CRISPR dTomato+ cells from young or old mice into age-matched recipients. At the end point mice were sacrificed, bone marrow cells isolated, CD117 enriched and stained with Mitotracker Green (MTG) 30 nM in the presence of Verapamil 50 μM and CellRox 5 μM in 500 μL of HAM-F12 medium for 30 minutes at 37 °C in the dark. Then cells were stained with cell surface antibody cocktail (see Supplementary Methods for details), in 100 μL of FACS buffer with 50 μM Verapamil for 15 minutes at room temperature in the dark. Cells were washed once with 1 mL FACS buffer with 50 μM Verapamil, then resuspended in 200 μL of FACS buffer with 50 μM Verapamil and analyzed by flow cytometry. The mean fluorescent intensity of MTG (FITC channel) and of CellRox (APC channel) was assessed in dTomato+ HSC cells (lineage^NEG^, c-Kit^+^, Sca^+^, CD135^NEG^, CD48^NEG^, CD150^+^) from Tet2 KO or control transplanted mice.
Immunoblot
Freshly isolated bone marrow cells were lysed in Laemmli buffer (63 mM Tris, 2% sodium dodecyl sulphate, 0.01 % bromophenol blue, 10% glycerol, pH = 6.8) at 10,000 cell/μL. 20 μL of lysate were loaded on a 4–20% acrylamide gel, then protein were transferred to nitrocellulose membranes, incubated with antibodies diluted in 5% BSA in TBS-T (150 mM NaCl, 20 mM Tris, 0.1% Tween-20, pH = 7.6) overnight at 4 °C. Membranes were washed three times with TBS-T, incubated with horseradish peroxidase conjugated secondary antibodies, washed three times with TBS-T, then chemiluminescent substrate was added and the membrane were imaged with a Bio-Rad ChemiDoc MP imager.
Single cell RNA sequencing
Raw sequencing data for mouse scRNAseq were processed using Cell Ranger v6.0.0 (10x Genomics) to generate gene-barcode count matrices. Downstream analysis was conducted in R using the Seurat package. Quality control steps included filtering out cells with fewer than 500 detected genes, as well as genes expressed in fewer than three cells. Additionally, cells with over 20% mitochondrial or more than 50% ribosomal gene content were excluded to remove potentially low-quality or stressed cells. Doublets were detected and removed using the DoubletFinder package. Individual sample objects were merged using Seurat’s merge() function.
Normalization was performed using the LogNormalize() method. To correct for batch effects across samples, integration was carried out using Harmony, with a theta value of 1.0. Dimensionality reduction was performed via principal component analysis (PCA) using the top 30 principal components. Clustering was carried out using the Louvain algorithm implemented in Seurat’s FindNeighbors() and FindClusters() functions, with a resolution parameter set to 0.5. For visualization, Uniform Manifold Approximation and Projection (UMAP) was applied using the Harmony-corrected PCA embeddings.
Clusters were annotated based on unsupervised clustering and the expression of well-characterized marker genes corresponding to distinct hematopoietic lineages.
Differential gene expression (DGE) analysis was conducted using Wilcoxon Rank Sum test embedded in Seurat’s FindMarkers() function. Comparisons were made between Old Wildtype (OW) vs. Young Wildtype (YW), Middle-aged Wildtype (MW) vs. Young Wildtype (YW), Old Tet2 KO (OT) vs. Old Wildtype (OW), and Middle-aged Tet2 KO (MT) vs. Middle-aged Wildtype (MW) across different cell types. For data in Figure 7d–f, Wilcoxon Rank Sum test in Seurat’s FindMarkers() function was used for DGE analysis by comparing HSC cells from Old vs Young donors. DEG for data in Figure 7d–f was obtained as reported in the respective original publications (Zeng et al 2023^59^ and Jackobsen et al 2024^25^), with the exception that all the genes were considered in the DGE analysis.
Gene set enrichment analysis (GSEA) was performed based on DEG lists for selected populations using fold-change and the fgsea R package^88^ with Hallmarks, GO-MF, GO-BP and Reactome gene sets from the Molecular Signatures Database, which were downloaded using the msigdbr R package (https://CRAN.R-project.org/package=msigdbr). Heatmaps were generated by using heatmapper.ca.
UMAP Grid-Based Mapping of Gene Set Expression Shifts
To visualize how the Tet2 mutation changes the expression of specific gene sets during aging on the UMAP, we analyzed expression patterns across the UMAP embedding. For each gene set (e.g., ribosomal protein genes), we first computed the average expression score per cell using normalized, log-transformed values.
The UMAP space was then divided into a grid of sliding square windows (0.5 × 0.5 units). Within each window, we calculated the mean expression score for each group: YW, MW, OW, YT, MT and OT.
To assess age- and Tet2 KO-related expression changes, we generated difference maps by subtracting the local mean expression of each group from that of the YW group, used as the reference. These differences were visualized as UMAP heatmaps with diverging color scales centered at zero, reflecting up- or down-regulation relative to the YW baseline.
Human sequencing data re-analysis
Gene Set Enrichment Analysis (GSEA) for Jakobsen et. al^25^, Ainciburu et al.^63^ and Aksoz et al.^41^ was performed using gene lists derived from human datasets reported in Supplementary Tables S6r–S6w of the Jakobsen et al study. For each dataset, genes were ranked either by the reported log__2__ fold change or using a significance-weighted approach that incorporated log-transformed p-values from the same tables, while preserving fold-change directionality. The ranked gene lists were then analyzed against curated gene sets from the Molecular Signatures Database (MSigDB v2025.1), including Hallmark pathways, Gene Ontology terms, Reactome pathways, and canonical pathway sets. We re-analyzed publicly available single-cell RNA sequencing (scRNA-seq) datasets of human hematopoietic stem and progenitor cells (HSPCs) from Zhang et al. 2022^64^ and Li et al. 2025^65^, corresponding to accession GSE189161 and the online resource http://scrna.sklehabc.com/HSPC/, respectively. Raw gene expression matrices, generated using Cell Ranger, were processed in R using the Seurat package. Standard preprocessing steps were applied as described earlier. Cell type annotations were assigned using metadata provided by the original studies. Developmental stage labels (Child, Adult, Aging) from the Zhang dataset and donor age information (ranging from 2 to 77 years) from the Li dataset were incorporated into the analysis. For the Li dataset, hematopoietic stem cells (HSCs) were stratified into “young” (2–25 years), and “old” (53–77 years). Differential gene expression (DGE) analysis was performed using Seurat’s FindMarkers() function, which applies the Wilcoxon Rank Sum test. For the Zhang dataset, comparisons were made between Aging vs. Child HSC populations. The resulting gene-level statistics, including log fold-change, adjusted p-values, and average expression values, were used to perform gene set enrichment analysis (GSEA) as described above.
Statistical analyses
For experiments with two groups, a two-tailed Student’s t-test was used; for experiments with more than one group, one-way ANOVA test with Sidak multiple comparison correction was used from the GraphPad prism software unless otherwise indicated. Histograms show biological replicates from distinct mice and represent the average ± standard error of the mean. Adjusted p-values are shown above charts.
Supplementary Material
Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Mejia-Ramirez E. & Florian M. C. Understanding intrinsic hematopoietic stem cell aging. Haematologica 105, 22–37 (2020). 10.3324/haematol.2018.21134231806687 PMC 6939535 · doi ↗ · pubmed ↗
- 2Ho Y. H. & Méndez-Ferrer S. Microenvironmental contributions to hematopoietic stem cell aging. Haematologica 105, 38–46 (2020). 10.3324/haematol.2018.21133431806690 PMC 6939521 · doi ↗ · pubmed ↗
- 3Genovese G. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med 371, 2477–2487 (2014). 10.1056/NEJ Moa 140940525426838 PMC 4290021 · doi ↗ · pubmed ↗
- 4Jaiswal S. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med 371, 2488–2498 (2014). 10.1056/NEJ Moa 140861725426837 PMC 4306669 · doi ↗ · pubmed ↗
- 5Jaiswal S. & Ebert B. L. Clonal hematopoiesis in human aging and disease. Science 366 (2019). 10.1126/science.aan 4673 · doi ↗
- 6Abelson S. Prediction of acute myeloid leukaemia risk in healthy individuals. Nature 559, 400–404 (2018). 10.1038/s 41586-018-0317-629988082 PMC 6485381 · doi ↗ · pubmed ↗
- 7Evans M. A. & Walsh K. Clonal hematopoiesis, somatic mosaicism, and age-associated disease. Physiol Rev 103, 649–716 (2023). 10.1152/physrev.00004.202236049115 PMC 9639777 · doi ↗ · pubmed ↗
- 8Watson C. J. The evolutionary dynamics and fitness landscape of clonal hematopoiesis. Science 367, 1449–1454 (2020). 10.1126/science.aay 933332217721 · doi ↗ · pubmed ↗