Young donor hematopoietic stem cells revitalize aged or damaged bone marrow niche by transdifferentiating into functional niche cells

Abstract The bone marrow niche maintains hematopoietic stem cell (HSC) homeostasis and declines in function in the physiologically aging population and in patients with hematological malignancies. A fundamental question is now whether and how HSCs are able to renew or repair their niche. Here, we show that disabling HSCs based on disrupting autophagy accelerated niche aging in mice, whereas transplantation of young, but not aged or impaired, donor HSCs normalized niche cell populations and restored niche factors in host mice carrying an artificially harassed niche and in physiologically aged host mice, as well as in leukemia patients. Mechanistically, HSCs, identified using a donor lineage fluorescence‐tracing system, transdifferentiate in an autophagy‐dependent manner into functional niche cells in the host that include mesenchymal stromal cells and endothelial cells, previously regarded as “nonhematopoietic” sources. Our findings thus identify young donor HSCs as a primary parental source of the niche, thereby suggesting a clinical solution to revitalizing aged or damaged bone marrow hematopoietic niche.

Early studies of the bone marrow niche have been primarily focused on the identification and function of its cellular and molecular components in governing HSCs. Mesenchymal stromal/stem cells (MSCs) and endothelial cells (ECs) are the two major niche members, which have been regarded as "nonhematopoietic" sources Ding et al., 2012;Hooper et al., 2009;Kfoury & Scadden, 2015;Kiel et al., 2005;Kobayashi et al., 2010;Méndez-Ferrer et al., 2010;Roberts et al., 2013;Winkler et al., 2012).
Schwann cells and sympathetic neurons have also been suggested to be components of the bone marrow niche, supporting the function of HSCs Yamazaki et al., 2011). Additionally, osteoblastic cells and adipocytes are important for hematopoiesis despite their presence outside of the niche location (Calvi et al., 2003;Naveiras et al., 2009;Zhang et al., 2003;Zhou et al., 2017).
Although macrophages, megakaryocytes, and T cells are involved in mediating HSCs (Bruns et al., 2014;Crane et al., 2017;Winkler et al., 2010;Zhao et al., 2014), studies have reported that niche cells originate from nonhematopoietic sources. For example, the neuroepithelium and neural crest were suggested to supply MSCs and neural cells in the fetal bone marrow niche Nagoshi et al., 2008;Takashima et al., 2007).
Hematopoiesis by HSCs involves their integration of intrinsic programs with extrinsic signals from the niche. Increased metabolism, decreased autophagy capacity, and altered epigenetic regulation are major intrinsic drivers of HSC aging (Chandel et al., 2016;Fang et al., 2020;Ho et al., 2017;López-Otín et al., 2016;Verovskaya et al., 2019). Recent studies have overwhelmingly investigated the impact of the niche on HSCs and hematopoiesis. Bone marrow hematopoietic niche aging is characterized by degeneration of adrenergic nerves in bone marrow (Maryanovich et al., 2018), increased inflammatory signals (Verovskaya et al., 2019), and deregulated proliferation of stromal cells with reduced niche factors (Pinho & Frenette, 2019;Verovskaya et al., 2019). Niche aging promotes the transformation of HSCs and the development of malignant hematological disorders (Curto-Garcia et al., 2020;Duarte et al., 2018;Gnani et al., 2019;Hurwitz et al., 2020;Shlush, 2018;Zhan & Kaushansky, 2020). Niche function can be improved by engineered MSCs with the transcription factors Klf7, Ostf1, Xbp1, Irf3, and Irf7 (Nakahara et al., 2019), and long-term engraftment of primary bone marrow stromal cells repairs niche damage and improves HSC transplantation (Abbuehl et al., 2017). The young niche in the recipient mice is able to largely restore the transcriptional profile of aged donor HSCs but not their DNA methylation profiles. Therefore, restoration of the young niche is insufficient for rejuvenating HSC function (Kuribayashi et al., 2021).
However, studies on the active role of HSCs in reciprocally shaping the stem cell niche are still lacking. In particular, the parental source and its underlying mechanisms by which bone marrow niche cells are regenerated or repaired in mammalian adults remain fundamentally unexplored. This study attempted to address these issues.

| HSC dysfunction by autophagy disruption led to an aging-like bone marrow niche lacking niche factors
To explore the role of HSCs in the bone marrow niche, we employed a conditional mouse model (Atg7 f/f ;Vav-iCre) in which HSC function is impaired and features a speedy aging-like phenotype F I G U R E 1 Deletion of the autophagy gene Atg7 led to an abnormal bone marrow microenvironment and niche cell populations lacking niche factors, comparable to those in old mice. (a) Representative micro-CT reconstructed three-dimensional pictures of femur trabecular (top) and cortical bone (bottom). Femora were collected from 10-week-old mice, 72-week-old mice and 10-week-old Atg7 −/− mice. n = 3, from three independent experiments. (b) H&E staining of femur paraffin sections. Femurs were collected from young mice (10-week-old), Atg7 −/− mice (10-week-old), and old mice (72-week-old) for paraffin sectioning and immunohistochemistry. Bar = 50 μm. n = 3, from three independent experiments. (c) Representative flow cytometry plots (left) with quantification of CD45 − Ter119 − cells, CD45 − Ter119 − CD31 + cells, and CD45 − Ter119 − CD31 − PDGFRα + cells in the bone marrow of young, Atg7 −/− , and old mice (right). (d) The transcriptional levels of niche factors in hematopoietic cells (CD45 + Ter119 − ) and the bone marrow stroma cells (CD45 − Ter119 − ) were detected by quantitative real-time PCR. (e) Protein levels of niche factors in bone marrow were quantified by ELISA. (f) PCA summarizing the top 2000 most variable genes among the gene expression profiles of young, old and Atg7 −/− mouse MSCs. (g) Left panel, intersection of differentially expressed genes in two comparison schemes (young vs. old and young vs. Atg7 −/− ). Right panel, Venn diagram between the intersecting gene sets from the left panel and differentially expressed genes from old versus Atg7 −/− mice. (h) Heatmap depicting the expression of niche factors in MSC transcriptomic profiling. Expression levels are row-normalized. ns p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; unpaired twotailed t test. All error bars indicate SD.
in hematopoiesis due to the selective deletion of the autophagyessential gene Atg7 (Cao et al., 2015;Fang et al., 2020;Mortensen et al., 2011). Dysfunction of HSCs in 10-week-old mice led to an abnormal bone marrow architecture, manifested as an obvious loss of trabecular structures and apparent cortical thinning in the Atg7 −/− mice observed in microcomputed tomography scanning images

| Transplantation of young but not aged HSCs restored the genetically or physically damaged bone marrow niche and rejuvenated the aged bone marrow niche in mice
To examine whether young donor HSCs are able to regenerate a functional hematopoietic niche in the host bone marrow whose niche is genetically predisposed, we transplanted 2000 HSCs from young To support the observation of the HSC capacity to restore the "nonhematopoietic" niche that was genetically disrupted by deletion of Atg7 in the hematopoietic system, we also examined the capacity of young donor HSCs to restore the niche destroyed by irradiation ( Figure S1a). Irradiation destroyed the whole hemato- To examine whether young donor HSCs are able to regenerate functional niche in physiologically aged niche in host mice, HSCs from 8-week-old mice were transplanted into 72-week-old mice. As expected, three major niche factors (SCF, CXCL12, and E-selectin) and three telomerases (Tert, Terf1, and Terf2), which are antiaging markers in bone marrow stromal cells that were previously reduced in old mice, were rescued in the 72-week-old host at the protein level  (Figure 3e,f). Therefore, young but not aged or dysfunctional donor HSCs are able to restore the "nonhematopoietic" niche, and restoration of the niche depends on the integrity of the autophagy machinery.

| Host bone marrow nonhematopoietic niche cells carried markers from young donor HSCs
To study how young donor HSCs contribute to regeneration of the bone marrow niche, the "nonhematopoietic" cells, we used young donor HSCs of Rosa mT/mG mice to serve as a cell tracing system in a host since this animal has universal expression of tdTomato fluorescence in all tissues; hence, the daughter cells from young donor HSCs of Rosa mT/mG mice can be tracked via tdTomato fluorescence.
We transplanted 2000 Rosa mT/mG HSCs with 0.2 million CD45.1 bone marrow cells into Atg7 −/− mice or lethally irradiated mice via tail vein injection to examine the reconstitution capacity of HSCs (Figures 4a and S1a). Compared with the nontransplant control, we found that more than 40% of ECs and approximately 20% of MSCs had tdTomato fluorescence in host mice ( Figure 4b). ImageStream measurement quantified over 90% colocalization between tdTomato fluorescence and CD31 (from CD45 − Ter119 − ) and approximately 25% colocalization between tdTomato and PDGFRα (from in bone marrow were quantified by ELISA (right). ns p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; unpaired two-tailed t test. All error bars indicate SD. and MSC-derived adipocytes (Perilipin) and osteoblasts (OPN) in the host bone marrow also revealed colocalization with tdTomato fluorescence in these cells, again supporting that HSCs are the source of these niche cells ( Figure 4d).
Next, we used quantitative PCR to trace the male-specific gene Ddx3y on the Y chromosome of the young donor HSCs of male mice in the "nonhematopoietic" niche cells of the recipient female mice.
The CD45 − Ter119 − stromal cells of the Atg7 −/− female host acquired the Ddx3y gene, similar to the CD45 + Ter119 − hematopoietic cells carrying Ddx3y from the donor HSCs (Figure 4e). Bone marrow CD45 − Ter119 − stromal cells have long been regarded as "nonhematopoietic." However, it appears here that HSCs are transdifferentiated to these niche cells. This notion is further supported by the fact that the Rosa mT/mG genotype in CD45 − Ter119 − stromal cells was changed from homozygous (in the donor) to heterozygous (in the host) (Figure 4e). Apart from analysis of the generation of the niche predisrupted by Atg7 deletion, we also observed that the host bone marrow niche cells, predamaged by irradiation, carried markers of young donor HSCs ( Figure S2). Thus, young donor HSCs are transdifferentiated to the "nonhematopoietic" niche populations in the host whose bone marrow niche was previously damaged by autophagy disruption or irradiation.
MSCs express cell markers, such as CD105, CD73, CD90, and CD29 and are always positive for CD44 and negative for CD45 (Kfoury & Scadden, 2015;Su et al., 2014). We examined these MSCspecific markers in the HSCs of nontransplanted wild-type mice and found that HSC-enriched hematopoietic cells (HSPCs) also express these markers (Figure 4f), and a previous study found that HSCs express Ang1, an EC marker, to regulate their niche (Zhou et al., 2015).
These data together suggest that under physiological conditions, bone marrow MSCs and ECs may also originate from bone marrow HSCs.

| Host bone marrow nonhematopoietic niche cells carried transcriptomic imprinting from young donor HSCs
To extend the identification of the niche cells from tracing single markers to a wider expression landscape, we performed transcriptomic profiling, which clearly indicates that while substantial contrast exists between donor HSCs and tdTomato − bone marrow stromal cells that are nondonor source, the transcriptomic pattern of High autophagy activity is associated with juvenescence (Ho et al., 2017), and autophagy is significantly downregulated in aged HSCs  and aged bone marrow .
Consistent with this finding, aged tdTomato − stromal cells displayed low expression of autophagy-associated genes, but stromal cells, excluding tdTomato − stromal cells, expressed high levels of autophagy-related genes as donor HSCs (Figure 4k), suggesting that host niche cells were rejuvenated after HSC transplantation.
Together, based on the identification of donor single markers and transcriptomic patterns in the host niches, the host nonhematopoietic niche cells MSCs and ECs carry the HSC donor markers, and the regenerated niche displays a distinct expression pattern overlapping with donor HSCs but different from the host own aged niche cells.
Therefore, the above results support the notion that "nonhematopoietic" niche cells are derived from donor HSCs via a transdifferentiating trajectory and become rejuvenated.

| HSC-derived "nonhematopoietic" niche cells in the host can be excluded from macrophagemediated phagocytosis or cell fusion between donor HSCs and niche cells
To examine whether the specific markers and transcriptomic pat- To test whether the regenerated "nonhematopoietic" niche cells carrying tdTomato fluorescence were derived from cell-cell fusion between the donor HSCs and nonhematopoietic niche cells of the host source, we transplanted HSCs from Rosa mT/mG mice to irradiated wildtype mice. Bone marrow stromal cells (tdTomato + CD45 − Ter119 − ) from the host mice showed a homozygous Rosa mT/mG genotype by PCR amplification, consistent with that of the donor Rosa mT/mG mice, and no hybrid genotype between donor HSCs and host nonhematopoietic cells was found (Figure 5a). tdTomato + CD45 − Ter119 − cells were transdifferentiated from donor mice showing Rosa mT/mG homozygous genotyping.
Furthermore, tdTomato + CD45 − Ter119 − cells from host mice were stained with Hoechst and analyzed for chromosome ploidy by FACS, and no polyploidy was found in these HSC-derived cells (Figure 5b

| Transplantation of HSCs reversed the decline in the bone marrow niche in leukemia patients
Receiving HSC transplantation from an old donor is associated with poorer outcome in patients diagnosed with hematological malignancies (Bastida et al., 2015;Murthy et al., 2022). This is in accordance with our result that HSCs from aged mice failed to restore niche function and hematopoiesis in the host (Figure 2). However, direct evidence of bone marrow niche aging in humans is inadequate.
To address this, we examined healthy young (average age 32) and old (average age 67) humans with their bone marrow (Table S1) and found that the function of the bone marrow niche declined in the old group, with significantly reduced niche factors, including SCF, CXCL12, E-selectin, OPN, Ebf3, and Angpt1, and significantly increased inflammatory factors, including IL-1, IL-6, TNFα, and TGFβ, compared with the young group (Figure 6a (f) HSPCs sorted against Lin − Sca-1 + c-Kit + from the bone marrow of WT mice were prepared for subsequent RNA extraction and reverse transcription-PCR assay. n = 3, from three independent experiments. (g-k) RNA sequencing with donor HSCs and host tdTomato − stromal cells, host tdTomato + stromal cells and total stromal cells of the donor bone marrow. All expression data in the heatmap are row normalized. Heatmap illustrating clusters of genes with high abundance in transcripts. Columns/samples were clustered using complete-linkage clustering, and rows/genes were clustered using kmeans (See method for details) (g); heatmap illustrating transcription profile of genes related to cell cycle (h); heatmap illustrating the transcription profile of genes related to HSC signatures (i); heatmap illustrating the transcription profile of proinflammatory genes (j); heatmap illustrating the expression profile of genes related to autophagy (k). tdT: tdTomato; ns p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; unpaired two-tailed t test. All error bars indicate SD.

F I G U R E 5
Aged or damaged bone marrow niche can be restored by donor HSCs, not via phagocytosis or cell fusion. (a) Genotyping of Rosa mT/mG and host mice. PCR analysis of Rosa mT/mG genotype from tdTomato + CD45 − Ter119 − cells in host mice. Rosa mT/mG mice served as a positive control. Wild-type mice served as a negative control. n = 3, from three independent experiments. (b) Detection of DNA ploidy. tdTomato + CD45 − Ter119 − cells in the bone marrow from Rosa mT/mG and host mice were stained with Hoechst and subjected to FACS analysis. Left, representative flow cytometry plots; right, DNA ploidy statistics. (c) Confocal detection of coenocytes. Representative confocal images of tdTomato + CD45 − Ter119 − cells from host mice stained for 6-diamidino-2-phenylindole (DAPI). (d) Karyotype FISH of Rosa mT/mG and host mice. Chromosome count from donor Rosa mT/mG and host mice. The above image shows the number and morphology of chromosomes at metaphase, and the following table shows the statistics of chromosome number and the percentage of cell fusion in Rosa mT/mG and host mice. (e) Representative confocal images of donor and host cells stained with anti-sodium potassium ATPase antibody (membrane marker), MitoTracker (mitochondrial marker), and 4,6-diamidino-2-phenylindole (DAPI). ns p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; unpaired two-tailed t test. All error bars indicate SD.

| DISCUSS ION
The bone marrow niche regulates HSC homeostasis through cellcell interactions and, in particular, the secretion of niche factors.
A single transplanted donor HSC is capable of reconstituting the entire hematopoietic system in lethally irradiated mice (Osawa et al., 1996;Uchida et al., 1994), and in zebrafish, HSCs were found to trigger physical remodeling of the perivascular niche to form a surrounding pocket (Tamplin et al., 2015). An important but unanswered question pertinent to clinical significance thus far is what type of cells is responsible for the renewal of the damaged niche in the irradiated host bone marrow and how the host niche is regenerated in the course of HSC engraftment for recovery of hematopoiesis in irradiated mammals. An extended question to the above is whether physiologically or pathologically aged niche can clinically be revitalized in adult bone marrow by donor HSC transplantation.
Our recent study showed that HSCs are indispensable in maintaining the macroenvironment in the bone marrow since impairment in HSCs by genetic intervention damages H-vessels in the bone and promotes osteoporosis, a typical bone aging-associated syndrome .
In this study, we found that young donor HSCs are capable of reversing a damaged or an aged niche, the microenvironment for HSC activity in the host bone marrow. Using a donor lineage fluorescence-tracing mouse model in which the progeny cells of fluorescent HSCs can be tracked along with their differentiation trajectory in the host and a niche-defective model in which bone marrow HSCs and the niche are physiologically aged or functionally impaired by autophagy disruption or by irradiation, we provide evidence that MSCs and ECs, the two major niche cell members in the host, previously regarded as "nonhematopoietic," originate from donor HSCs (outlined in Figure 7).
Several studies suggest niche lesions as initiating cascades in hematological malignancies (Duhrsen & Hossfeld, 1996;Méndez-Ferrer et al., 2020). An inflammatory microenvironment with reduced niche factors is an important predisposition factor for many hematological disorders as we age. For example, abnormal alterations in MSCs were observed in patients with myelodysplastic syndrome and acute myeloid leukemia (Blau et al., 2007;Kim, Jekarl, et al., 2015;Kim, Shim, et al., 2015;von der Heide et al., 2017). Dysfunctional niches facilitate mutant hematopoietic cell survival and expansion, leading to malignancy development and progression, and possibly protect malignant cells from chemotherapy, ultimately leading to relapse (Duhrsen & Hossfeld, 1996;Méndez-Ferrer et al., 2020). Overexpression of transcription factors in MSCs was attempted to improve niche function (Nakahara et al., 2019). Although the young niche in the recipient mice partially restored the transcriptional profile of aged donor HSCs, it was unable to normalize the DNA methylation profiles and function of HSCs (Kuribayashi et al., 2021). In our present study, transplantation of young donor HSCs was shown to improve aged or damaged niche in both animal models and leukemia patients. This is particularly important because patients with hematological malignancies often receive radiotherapy and long-term chemotherapy, which inevitably harm and even severely impair the bone marrow hematopoietic niche. Therefore, transplantation of younger HSCs F I G U R E 7 Schematic cartoon illustrating that aged or damaged bone marrow can be rejuvenated or repaired by young donor HSCs. Bone marrow niche aging is characterized by the accumulation of silenced and dysfunctional niche cells. Young donor HSCs are capable of regenerating aged or damaged "nonhematopoietic" niche cells. Transdifferentiation of young donor HSCs to their niche cells in the host depends on autophagy machinery integrity. may play a dual role in the reconstitution of both hematopoiesis and the hematopoietic niche.
The molecular mechanism by which HSCs are transdifferentiated into their niche cells was not elucidated in this study. However, the possibility of phagocytosis can be precluded because the HSCderived niche cells MSCs and ECs in the host did not harbor markers (F4/80 + CD11b + ) specific to macrophages, thus excluding false signals for colocalization of the markers from macrophages and HSCs.
HSC-derived cells maintained a homozygous genome and diploidy; confocal microscopy did not find coenocytes; karyotype assay confirmed the normal morphology of the host niche cells and number of chromosomes, thereby ruling out the possibility of cell-cell fusion to give false signals (Figure 5a,b,d).
Therefore, we argue that transdifferentiation may be responsible for the remodeling of donor HSCs to their niche cells. In addition to donor lineage fluorescence-tracing analysis, bioinformatics analysis revealed transcriptomic imprinting from donor HSCs in host niche cells. The transdifferentiation trajectory may be triggered by aging or damaging stresses, particularly by the accumulation of inflammation in the niche, which is in agreement with the pioneering report on bone marrow-derived cell transdifferentiation to nonbone marrow cells by Krause and colleagues (Harris et al., 2004;Krause et al., 2001). Bone marrow consists of a long array of cell types with and without blood lineage. It is not surprising that certain types of bone marrow cells can fuse with cells from nonhematopoietic organs (Alvarez-Dolado et al., 2003;Terada et al., 2002). Our results suggest that donor HSCs are responsible for the transdifferentiation potential of bone marrow cells into niche cells. Recently, donor HSCs were found to differentiate into 28 cell types in myeloablated recipients at an early stage one week after transplantation (Dong et al., 2020), a number much more than that currently known as "hematopoietic cells," suggesting that Autophagy has been documented to decelerate hematopoietic aging Ho et al., 2017). The present study shows that regeneration of the niche by HSC transdifferentiation depends on the integrity of the autophagy machinery since deletion of Atg7, an essential autophagy gene, disabled the capacity of HSCs to generate niche cells in the host; in particular, autophagy defects in the hematopoietic system apparently cause bone marrow niche aging ( Figure 1). Therefore, the maintenance of the niche by HSCs depends on autophagy in both transplant and nontransplant settings.
In summary, the present study establishes that bone marrow aging is characterized by extensively silenced and dysfunctional niche cells and that young donor HSCs can serve as the parental source for regenerating pathologically damaged or physiologically aged niches, supporting an increased multipotency of HSCs for transdifferentiation to the "nonhematopoietic" lineage in the case of niche damage or decline. Young donor HSCs can flexibly orchestrate the balanced production of their progeny cells between the blood lineage and bone marrow niche lineage, thereby securing nichesupported hematopoiesis. Therefore, young HSC transplantation may be used to improve both HSC hematopoiesis and its supporting niche in leukemia patients. Nevertheless, future study is warranted to determine the molecular mechanism driving such a transdifferentiation trajectory.

| Mouse models and xenografts
C57BL/6J, CD45.1, Rosa mT/mG and Atg7 f/f ;Vav-iCre mice were used in this study. Rosa mT/mG mice were generated by the laboratory of Dr.

| Human samples
Human samples were collected from the affiliated hospitals of Soochow University in accordance with the University's code on Medical Ethics. The sample information is detailed in Table S1.

| DNA isolation and genotyping
Genomic DNA of the indicated cells was isolated using a Genomic DNA Mini Preparation Kit with a Spin Column (Beyotime).
Genotyping via PCR was performed using the following primers:

| Bone immunostaining and image flow cytometry
Hind limbs were collected from the indicated mice, and the soft tis-

| Micro-CT analysis
The hind limb from the same side of each mouse was fixed in 2% polyvinylpyrrolidone (PVP) for 48 h, and the limbs were scanned on a SkyScan micro-CT system (SkySacn, Antwerp, Belgium).

| Flow cytometry analysis and cell sorting
BM cells from Atg7 f/f ;Vav-iCre mice, control littermates, and old mice were collected. The HSCs were analyzed after exclusion of lineage

| RNA sequencing analysis
MSCs (CD45 − Ter119 − CD31 − PDGFRα + ) were sorted from 10-week-old mice, 72-week-old mice and Atg7 −/− mice. HSCs were sorted from 8-week-old Rosa mT/mG mice, tdTomato + and tdTomato − stromal cells were sorted from host mice, and total stromal cells were sorted from WT mice. The experimental procedure was as follows: (1) mRNA enrichment and purification: Oligo dT selection to enrich the mRNA (for total RNA extracted from human whole blood, globin mRNA is depleted); (2) RNA fragmentation and cDNA synthesis (second-strand cDNA synthesis with dUTP instead of dTTP); (3) end repair, addition of A and adaptor ligation; (4) PCR; (5) circularization and DNB; and (6) sequencing on the DNBSEQ platform. Sequencing data filtering used the software SOAPnuke developed by BGI independently for filtering, and these data were subjected to quality control (QC) to guarantee suitability for analysis.
Unsupervised hierarchical clustering on expression profile data used in the clustering process was filtered with TPM > 20 (in at least one sample), and we selected the 2000 most variable genes to conduct cluster analysis. Complete-linkage hierarchical clustering was used to cluster the expression profiles, and the distance metric we used was Euclidean distance. The expression heatmap was plotted using filtered TPM expression and was normalized by gene. The clustering result was visualized as dendrograms in the expression heatmap.
Similar gene expression patterns were clustered into 5 clusters using k-Means. Enrichment analysis was conducted to investigate the function of each clustered module.

| Statistical analysis
Statistical analyses were performed using SPSS version 25.0. The clinical data of leukemia patients and healthy individuals were collected with blood cell count, classification, sex, and age. Experimental data are presented as the means ± standard deviations (SDs), which were evaluated using unpaired Student's t tests. p values are reported as nonsignificant when p ≥ 0.05. All statistical analyses were performed using GraphPad Prism v. 8 software. Graphs were plotted using Adobe Illustrator CS. (LT).

CO N FLI C T O F I NTER E S T S TATEM ENT
None declared.