INTRODUCTION

      Aging is manifested as a multisystemic deterioration with waning regenerative potential that leads to declining tissue and organ function (Lo´pez-Otı´n et al., 2013). Such aging phenotypes have been reversed in animalmodels, reported as rejuvenation interventions by targeting global mechanisms, most notably through the use of heterochronic parabiosis (HP) (Zhang et al., 2020). In the HP mouse model, the circulatory systems of aged and young mice are surgically connected, creating a shared circulatory system in which blood-borne factors generated by one individual are shared with the other. In the pioneering work, the young blood was reported to be capable of reversing age-related structural erosion and molecular changes in the muscle, bone, liver, and nervous systems (Castellano et al., 2015). Thus, HP offers a unique experimental paradigm with which to study how an intact aged organism is rejuvenated by the infusion of ‘‘pro-youth factors,’’ and vice versa, how the young organism is affected by circulatory ‘‘pro-aging factors’’ (Brack et al., 2007; Castellano et al., 2015; Conboy et al., 2005; Davies et al., 2015; Eggel and Wyss-Coray, 2014; Kang and Yang, 2020; Villeda et al., 2011). Although several pro-youth factors have been identified, our knowledge about the cellular targets of blood-borne factors and how they mediate rejuvenation at the systemic level remains limited.

      Many hallmarks of systemic aging are associated with and have been attributed to the impairment and exhaustion of organ-specific adult stem cells (Lo´pez-Otı´n et al., 2013). For example, in organs such as bone marrow, skin, brain, and skeletal muscle, a rare population of adult stem cells replenish the tissues throughout life to repair age-related damage and maintain tissue homeostasis. In the high-turnover hematopoietic and immune system, which comprises multiple organs and tissues, including bone marrow, spleen, and peripheral blood, the hematopoietic stem cells (HSCs) and their immediate progeny generate the spectrum of immune cell types in this system (Haas et al., 2018; Wang et al., 2012). However, hematopoietic stem and progenitor cells (HSPCs) gradually lose their capacity for blood reconstitution upon transplantation (TX) during aging, which is reflected as a skewed differentiation potential and contributes to age-dependent hematopoietic disorders and low immunity (Geiger and Rudolph, 2009; Grover et al., 2016; Nikolich-Zugich, 2018; Rossi et al., 2005, 2007; Sudo et al., 2000). Similarly, other organs, such as the skin, skeletal muscle, and brain, also harbor different types of adult stem cells, i.e., hair follicle stem cells (HFSCs) and basal cells responsible for the regeneration of hair follicles and the epidermis, respectively, muscle stem cells (MuSCs), and fibro/adipogenic progenitor cells (FAPs) for the skeletal muscle, as well as neural stem cells (NSCs) for the brain (Ge et al., 2020; Oh et al., 2014; Zou et al., 2021). These adult stem cells gradually lose their regenerative ability with age, leading to hair loss and skin aging, as well as skeletal muscle and neurological degeneration. Thus, resident adult stem cells both in the hematopoietic system and in the peripheral organs are compromised by aging, but if and to what extent they can be rejuvenated require further investigation.

      Here, we systematically analyzed single-cell transcriptomes across the hematopoietic and immune system (the enriched HSPCs, bone marrow, peripheral blood, and spleen), as well as four solid tissues/organs (the skin, skeletal muscle, brain, and liver) affected by the hematopoietic and immune organs, sampled from heterochronic and isochronic parabionts. Through this approach, we were able to construct a multi-tissue single-cell transcriptomic atlas to study the effect of HP on aging and to explore the intricate changes and regulatory roles of the adult stem cells and their supportive niches.We dissected both how the young circulatory milieu rejuvenated the aged animal and how the aged circulatory milieu compromised the young animal to alter cell type populations, gene expression signatures, and cell-cell communications. Within our HP atlas, we found HSPCs to be one of the cell types most resistant to spatial re-distribution but most responsive to systemic regulation at the same time in the hematopoietic and immune systems. Moreover, we identified aging-rejuvenating factors and pathways with the potential for reversing age-related impairments. Our work constitutes a mineable resource for advancing our understanding of aging-related systemic factors and how they might be targeted to alleviate aging.

RESULTS

The rejuvenating effects of HP on the old parabiont

      To study the systemic regulation of aging and rejuvenation, we used the peritoneal method (Xiong et al., 2018) to generate heterochronic parabiotic pairs between 23-month-old (Het-O) and 2-month-old (Het-Y) male mice, and as controls, isochronic parabiotic pairs between young mice (2 months old) and old mice (23 months old), respectively (Iso-Y and Iso-O) (Figure 1A). Overall, the basic physiology, such as the weight and blood glucose levels, were not markedly changed in the heterochronic parabionts (Figure S1A). In Het-O parabionts, we found agingassociated phenotypes to be alleviated at the cellular level in multiple tissues/organs, reflected as reduced senescence-associated b-galactosidase (SA-b-gal)-positive cells in the spleen, skin, liver, and brain, decreased apoptotic cells in the spleen, skin, liver, and skeletal muscle (Figures S1B and S1C). HP also mitigated aging-related inflammation in the liver, alleviated fibrosis in the liver and spleen (Figures S1D and S1E), and restored the average diameter of myofibers in skeletal muscle and the number of hair follicles in the skin (which normally diminishes with age) to a ‘‘youthful’’ level (Figure S1D).

Construction of a multi-tissue single-cell transcriptomic atlas of HP

     To identify cellular mechanisms by which HP contributes to aging and its rejuvenation, we isolated and compared, at singlecell resolution, seven tissues/organs within and distal to the hematopoietic and immune system, i.e., the bone marrow, peripheral blood, spleen, skin, liver, skeletal muscle, and brain from young and old animals in HP, with corresponding tissue samples from the age-matched animals from isochronic pairs. To obtain sufficient numbers of HSPCs from the bone marrow, we sorted cells with broad gates of Lin- Sca-1+ c-Kit+ (LSK) to further enrich HSCs and immature progenitors, such that we would gain a comprehensive view on cell population alterations in hematopoiesis during aging and HP (Figure S2A). After stringent quality control (including doublet removal, batch effect correction, and normalization), we acquired a total of ~ 164,000 high-quality single-cell transcriptomes and annotated 108 major cell types across the seven tissues/organs for subsequent analyses (see STAR Methods; Figures 1B and S2B–S2F; Table S1).

      Focusing first on the hematopoietic and immune system, we annotated 33 clusters encompassing the hematopoietic and immune cell types present in Iso-Y, Iso-O, Het-Y, and Het-O groups (Figures 1B, S2D, and S2F). These included cell types belonging to HSPCs, myeloid and lymphoid lineages (Figure 1B). Specifically, multipotent HSPCs, at the top of the hematopoiesis differentiation hierarchy, were subdivided into populations corresponding to the long-term self-renewing HSCs (LT-HSCs), short-term self-renewing HSCs (ST-HSCs), and multi-potential progenitors (MPPs) (Figures 1B and S2F). In comparative analysis across different heterochronic or isochronic parabiotic mice groups, all designated cell types were detected (Figures S2G and S2H). To enable the visualization of the putative relationship and differentiation trajectory between cell types, we performed partition-based graph abstraction (PAGA) analysis (Figure S2I) to generate a force-directed graph of different cell types with HSPCs at the center. In this analysis, the generation of myeloid immune cells proceeded along a path from pro-granulocytes and granulocytes to neutrophils or from promonocytes and monocytes to macrophages (Figures S2I– S2M). On the other hand, HSPCs differentiate into lymphoid immune cells through intermediate cell states, whose gene expression gradients were also captured, such as from pro-B, pre-B, and naive B to B cells (Figures S2I–S2M).

      In the analysis of solid tissues, we identified tissue-specific stem/progenitor cells, such as epidermal stem cells (basal cell-1 and basal cell-2), proliferating basal cells (basal.mito), and HFSCs in the skin, FAPs, and MuSCs in the skeletal muscle and NSCs in the brain (Figure 1B). Notably, surrounding niche cells, such as endothelial cells (ECs) and fibroblasts/ mesenchymal cells (i.e., pericytes in skeletal muscle, papillary fibroblasts [PapiFib], and reticular fibroblasts [RetiFib] in the skin), were also identified (Figure 1B). As expected, immune cells, known to make major contributions to the stem cell niche microenvironment, were also found in all solid tissues analyzed, such as macrophages (Figure 1B). Other than these cell types, 48 tissue-specific cell types were classified, such as hepatocytes in the liver, hair follicle cells in the skin, fast-twitch muscle fibers IIA and IIX (FAST IIA and FAST IIX) in the skeletal muscle, and excitatory and inhibitory neurons in the brain (Figure 1B). Altogether, we established a multi-tissue atlas of cellular diversity, generating a comparative framework for further studies of aging and HP.

HP elicits a transcriptomic reset in multiple cell types

      To further discriminate cell-type-specific aging-rejuvenating (aging-R) or aging-promoting (aging-P) effects of HP, we initially characterized aging-related differentially expressed genes (DEGs) between isochronic pairs (aging DEGs, Iso-O versus Iso-Y) (Figure 1C). Then, using an integrative comparative analysis between heterochronic and isochronic pairs, we classified aging DEGs that were mimicked by Het-Y as aging-P DEGs and aging DEGs that were rescued in Het-O as aging-R DEGs (Figures 1C–1E; see STAR Methods).

      Upon characterizing aging, aging-P and aging-R DEGs in each cell type, we observed marked transcriptomic changes across all tissues and cell types, suggesting a widespread impact of HP-induced changes, independently of their relationship to the shared circulation (Figures 1C–1E). As expected, the liver, the most perfused organ, harbors a large number of aging-R and aging-P DEGs (Figures 1C–1E). Strikingly, resident stem cells/ progenitors displayed a variable extent of transcriptional plasticity upon HP, especially for HSPCs (LT-HSCs, ST-HSCs, and MPPs) in Het-O bone marrow, as well as FAPs in Het-O skeletal muscle and basal cells and HSFCs in Het-O skin to a lesser extent (Figures 1C and 1E). Vascular ECs, which form a continuous layer lining the intima of the blood vessel wall, are directly exposed to the blood flow. As expected, we found that ECs were generally responsive to HP, especially in the liver (Figures 1C and 1E). Following HP, tissue-resident specialized immune cells, including Kupffer cells in the liver and microglia in the brain, also underwent marked transcriptomic reprogramming, which may strongly impact the microenvironment harboring resident stem cells (Figures 1C and 1E). In addition, other terminally differentiated cell types, such as FAST IIX myo-fibers in the skeletal muscle and oligodendrocytes in the brain, were also susceptible to HP (Figures 1C and 1E). Collectively, long-term and continuous exposure to the young milieu rejuvenates adult stem cells to a variable extent, concomitant with a reset of the global transcriptomic network across tissues.

HP mitigates aging pathways in HSPCs and rewires the continuum of HSPC differentiation

      Among the resident stem cells in various tissues tested, HSPCs in the bone marrow harbor the most aging-R DEGs (171 aging-R DEGs [149 upregulated and 22 downregulated], along with 540 aging DEGs [131 upregulated and 409 downregulated] and 119 aging-P DEGs [96 downregulated and 23 upregulated]) (Figures 1D, 2A–2C, and S3A; Table S2). To provide further insight into the underlying regulatory pathways modulated by HP in HSPCs, we analyzed functionally annotated aging, aging-P, and aging-R DEGs in HSPCs. We found that upregulated aging DEGs were commonly associated with neutrophil activation and antimicrobial response (Figure S3B). By contrast, downregulated aging DEGs were commonly associated with cytokine production, hemopoiesis, chromatin organization, circadian rhythm, and cell cycle (Figure 2D). These aging-related transcriptional changes were largely mimicked by parabiotic aging induced by exposure to old blood, as aging-P DEGs were preferentially enriched in similar gene ontology (GO) terms as aging DEGs (Figure 2D). Indeed, about 80% of aging-P or aging DEGs in HSPCs upon HP are downregulated genes, indicating that exposure to blood-borne factors from the old animal primarily functions to mimic aging-related gene repression (Figure S3A). By contrast, after exposure to young blood, genes silenced during aging become strongly reactivated in HSPCs (Figures 2A and 2C). Interestingly, global transcriptomic profiling of LT-HSCs, ST-HSCs, and MPPs in Het-O was closer to Iso-Y than Iso-O, further indicating that HP reshapes the transcriptomic landscape of aged HSPCs to a younger state (Figure 2B). GO analysis of the restored gene expression profiles, especially of upregulated aging-R DEGs, were enriched in pathways for regulation of hematopoiesis, cytokine production, and stem cell population maintenance in Het-O HSPCs (Figure 2D).

      When we overlapped the aging-R DEGs with aging-P DEGs, we identified 36 genes whose expression was altered by young or old blood in opposite directions (Figure 2A). We referred to these DEGs as heterochronic parabiosis-core DEGs (HP-core DEGs), as this gene set was most sensitive to the changing conditions in HP. In the HP-core DEGs, we identified young bloodaugmented chemokine or cytokines such as Ccl3 (MPP, common lymphoid progenitor cell [CLP], and granulocyte-monocyte progenitor [GMP]) and Cxcl2 (MPP, CLP, and common myeloid progenitors [CMP]), which function as major regulators of HSPC homeostasis (Figure S3C; Sinclair et al., 2016). In addition, we found that anti-inflammatory targets of the TNF-a pathway such as Tnfaip3 (LT-HSC and ST-HSC), a key regulator of HSCs and hematopoiesis (Nakagawa et al., 2018), as well as Nfkbia (LT-HSC and ST-HSC) and Nfkbiz (ST-HSC, MPP, and GMP), were decreased in Iso-O and Het-Y but recovered upon exposure to young blood in Het-O (Figure S3C). Notably, another subset of HP-core DEGs was also implicated in the maintenance of the HSC pool, such as Fosb (LT-HSC and MPP), Junb (LTHSC, ST-HSC, and MPP), Ier3 (LT-HSC, ST-HSC, and MPP), and Dusp2 (LT-HSC, ST-HSC, and MPP) (Figure S3C; Passegue´et al., 2004a, 2004b; Santaguida et al., 2009). Therefore, our results identified aging regulatory pathways and core genes whose expression perturbation was mitigated by HP to reinforce the maintenance of a healthy HSPC pool.

      Our inquiry into changes in the transcription profiles and related pathways suggested HSPCs as the key cellular mediator for reshaping the hematopoietic and immune systems in HP. To further reveal how hematopoiesis is affected by HP, we performed a pseudotime analysis of the single-cell transcriptomic atlas, which presents a continuum of cell state changes originating from HSPCs and branching out into the lymphoid and myeloid lineages (Figures 2E and S3D–S3F). We then classified clusters of genes that are differentially expressed along pseudotime and found that genes involved in HSPC hemopoiesis (cluster 1) were highly enriched in Iso-Y HSPCs, reduced in Iso-O, and reactivated in Het-O (Figures 2F and 2G; Table S3). Genes that are progressively upregulated along the molecular trajectory of B lymphopoiesis (cluster 5) and T lymphopoiesis (cluster 6) were expressed at lower levels in Iso-O and Het-Y HSPCs compared with that of Iso-Y, while their expression was restored to a younger state in Het-O HSPCs (Figures 2F and 2G), suggesting a stronger tendency of lymphopoiesis. Consistently, as revealed by calculation based on scRNA-seq and verified by fluorescence-activated cell sorting (FACS), in the bone marrow of Het-O parabiont, pro-B cells were replenished in the bone marrow and may promote adaptive immunity (Figures 2H, 2I, and S3G–S3I).

Transcriptional and chromatin modulators of HSPC rejuvenation

      To untangle how transcriptional regulatory networks were altered in aging and HP, we used single-cell regulatory network inference and clustering (SCENIC) to predict core transcription factors (TFs) that regulate aging, aging-P, and aging-R DEGs (Figures 3A, 3B, and S4A; Table S4). Of all the transcription factors found to be dysregulated during aging, more than 20% of the expression level changes were mimicked by exposure to old blood, and about half were restored by exposure to young blood (Figure 3A). Besides TFs identified as HP-core DEGs (i.e., Fos and Jun), we identified additional upregulated transcription factors for aging-R DEG expression in Het-O, including Atf3 that prevents stress-induced depletion of HSCs and Atf4 that plays a crucial role in HSC repopulation and antagonizing HSC aging (Sun et al., 2021; Zhao et al., 2015; Figures 3B–3D). Accordingly, downstream genes of Atf3 and Atf4 are involved in hemopoiesis, and leukocyte differentiation, namely Klf2, Klf4, Klf6, and Klf13, was upregulated in LT-HSC, ST-HSC, and MPP of Het-O (Figures 3E–3G). Taken together, our data suggest that HP, through these transcription factors, resets the regulatory programs that govern the stemness and differentiation potential of HSPCs.

      We also found that chromatin organization was among the enriched pathways in downregulated aging and aging-P DEGs (Figure 2D). Although a fraction of chromatin organization and remodeling factors can be corrected by exposure to young blood, such asKdm6b (frequency = 7) and Jmjd1c (frequency = 3) in Het-O (Figure S4A), aging DEGs that were not rescued by young blood were still primarily concentrated in these chromatin organization and remodeling factors (Figures S4B and S4C). Among them, Yy1 was also a core transcriptional regulator of aging DEGs (Figure 3B). When overexpressed in aged mouse HSPCs, YY1 promoted the long-term engraftment ability of aged HSPCs, highlighting the importance of epigenetic reprogramming in HSPC rejuvenation (Figures 3H, 3I, and S4D).

HP re-establishes cell-cell communications essential for hematopoiesis

      Immune cells communicate with each other through cytokines such as chemokines, interleukins, and interferons, which are all required for constructing the hematopoietic hierarchy and mounting an effective immune response (Florian et al., 2013). We found that cell-cell communication pathways were generally uncoupled in aging but became re-established upon exposure to young blood (Figures 4A and 4B; Table S5). In particular, agingrelated loss of communication among HSCs (LT-HSCs, STHSCs, and MPPs) and dendritic cells, immature NK, and pro-B cells were markedly restored by exposure to young blood in Het-O, which may be involved in rebalancing lymphoid differentiation potential (Figures 4A and 4B). GO and pathway enrichment analyses highlighted cytokine and cytokine receptor interactions, chemotaxis, and leukocyte proliferation in those restored cell-cell communications (Figure 4C). Especially, CD86 expressed on LT-HSCs, ST-HSCs, and MPPs could interact with either CD28 or CTLA4 from various lymphoid cells and dendritic cells, endowing HSCs with lymphopoietic potential (Figure S4E; Shimazu et al., 2012).

      We also noticed that genes enriched in the KEGG pathway‘‘cytokine-cytokine receptor interaction’’ including Ccl3 (frequency = 3), Ccl4 (frequency = 5), and Vegfa (frequency = 5), all of which have been reported to contribute to HSC maintenance, were restored upon exposure to young blood in Het-Omice (Figures 4D and 4E; Lee et al., 2018; Rehn et al., 2011; Staversky et al., 2018; Yamashita and Passegue´, 2019). Among them, Ccl3 is also among HP-core DEGs, which are the most sensitive to the changing conditions in HP. To study the effect of CCL3 on the biological function of aged HSCs, we sought to reconstitute CCL3 through lentiviral vector-based overexpression of CCL3 in aged mouse HSCs. The long-term competitive TX assay showed higher T lymphopoietic potential of the engineered HSCs (Figures 4F, 4G, and S4F). Collectively, these data suggest that reestablishing cell-cell interactions in aged HSPCs by supplementing key cytokines may be an effective way to facilitate hematopoiesis.

      Lastly, to explore the biological implications of our findings in human aging, we compared aging-R DEGs to available differential gene expression datasets of aged human HSCs (Figure 4H) (Adelman et al., 2019). Interestingly, we found that a fraction of aging-R DEGs overlapped with those in aged human HSCs (Adelman et al., 2019), whose expression was altered in opposite directions, including Atf3/4 target genes Klf2, Klf6, and Klf13 and circadian factors Per1 (Figure 4H). In addition, several genes (such as Vegfa, Klf4, and Klf6) that have been implicated in aging-related hematopoietic disorders were also included in the aging-R gene set (Figures 4H and S4G). Our observations suggest that the young blood-born factors identified here may help alleviate aspects of the transcriptional disturbance underlying human hematopoietic aging or aging-related diseases.

HP revitalizes resident HSPCs in the aged Het-O parabionts

      Since parabionts share each other’s blood circulation, we wondered if hematopoietic cells from one parabiont can relocate and reside in the more isolated bone marrow compartment of the other. To answer this question, we used the CD45.1/CD45.2 congenic system to analyze lymphocyte exchange of hematopoietic cells originating from the aged (CD45.1 alloantigen, Het-O) and young (CD45.2 alloantigen, Het-Y) partners with a shared circulatory system. It is worth noting that only a small fraction of CD45.2-positive cells from Het-Y bone marrow was detected in Het-O bone marrow, compared with peripheral blood and spleen (Figures 5A and 5B). This observation emphasizes that, even with a connected circulatory system, the bone marrow is still a more isolated compartment, probably due to the mechanical bone-marrow-blood barrier formed by bone marrow ECs that regulate cellular trafficking (Tavassoli, 1979; Zhao and Li, 2016). However, the restrictive niche environment might not necessarily limit the parabiosis-mediated exchange of cytokines and other blood-borne factors that may act on the resident autologous cells to reshape their transcription regulatory network and contribute to the observed phenotypes.

      Nevertheless, to eliminate any confounding effects resulting from immune cell crossover between Het-Y and Het-O parabionts, we leveraged the CD45.1/CD45.2 congenic system to trace the origin of bone marrow cells or enriched HSPCs (Figures 5A–5C and S5A–S5E). Consistent with the FACS data, CD45.1/CD45.2-based scRNA-seq transcriptomes further demonstrate that HSPCs are seldomly exchanged between parabionts (Figures 5A and 5D). Although CD45.1/CD45.2-based scRNA-seq revealed that a considerable fraction of Het-Y B and T cells translocated into the Het-O bone marrow compartment, most cell types in the bone marrow originated from the host organism itself (Figure 5E), suggesting that autologous HSPCs make critical contributions to the changes in bone marrow cellular composition we observed during HP.

      As expected, the transcriptomic changes were basically the same with or without expelling the rarely detected crossover cells (CD45.1 cells in Het-Y and CD45.2 cells in Het-O) in the HSPCs (Figure S5F). Again, this analysis demonstrated that aging-related regulatory programs of Het-O-derived CD45.1 HSPCs in the Het-O were effectively rejuvenated to a younger state with a strong potential for lymphopoiesis by exposure to young blood (Figures 5F–5H). Overall, HP mitigates aging pathways enriched in the regulation of hematopoiesis, cytokine production, and circadian rhythmic processes in Het-O HSPCs (Figure 5G). More importantly, key transcription factors such as Atf3, Atf4, and the cytokine factor Ccl3 were also restored in CD45.1 Het-O HSPCs (Figure 5I). Altogether, with the power of an independent dataset based on the CD45.1/CD45.2 congenic system, our data support the conclusion that at least at the transcriptional level, HP enables functional rejuvenation of old HSPCs to a younger state.

Dissecting the re-distribution of immune cells due to the conjoined circulation

      In contrast to the low infiltration rate of nonautologous cells in the bone marrow, we detected a high infiltration rate in both peripheral blood and spleen using FACS (Figure 5B). To trace the origin of these circulatory immune cells, we again resorted to CD45.1/ CD45.2-based scRNA-seq. Intriguingly, we found that the majority of naive CD4+ T and NK cells in Het-O blood were CD45.2 positive and therefore originated from Het-Y (Figures 5J and S5G). Accordingly, lymphocytes that decreased in aging, such as NK cells, were replenished in Het-O peripheral blood (Figures 5J and S5G). In addition, we found that more than half of naive CD4+ T and B cells were derived from Het-Y and recolonized in the spleen of Het-O animals (Figures 5J and S5G). Such replenishment in blood and spleen of Het-O animals was to a substantial extent attributed to translocated CD45.2 cells from the Het-Y parabionts.

      By comparison, the majority of CD45.1-positive memory CD8+ T cells were traced back to Het-O peripheral blood and spleen (Figure 5J). As confirmed by FACS, most memory CD8+ T cells in both Het-Y and Het-O parabionts originated from Het-O (Figures 5K and 5L). Recently, the accumulation of granzyme K (GZMK)-expressing T cells was reported to be a hallmark of the aged immune system, as well as causing the consequent low-grade chronic inflammation grade (‘‘in-flammaging’’) in both mice and humans (Mogilenko et al., 2021). Here, we found that most GZMK+ cells were memory T cells residing in the spleen, the proportion of which increased in Het-Y but not in Het-O parabionts were derived from the Het-O parabiont (Figures S5H–S5L). Altogether, through tracing the changes in the cellular composition, we showed that HP leads to the re-distribution of circulatory immune cells, which in turn facilitates correction of the agingassociated skewed ratio of naive to memory T cells, thus restoring the aged immune system to a younger state (Figures S5M and S5N).

      We then asked whether aged blood and spleen cells, at the transcriptional level, could be rejuvenated by exposure to young blood. Overall, the number of aging-R genes was negligible in most circulatory blood and spleen cell types in Het-O, even after eliminating the confounding effect of cell crossover (Figure 5F), indicating that blood and spleen cells are resistant to rejuvenation by young blood exposure. Confirming these data, the PCA dissecting the transcriptomic relationship between CD45.1- and CD45.2-positive cells in Het-O and Het-Y parabionts shows the limited reprogramming effects of HP on these circulatory immune cells (Figure S5O). Therefore, unlike HSPCs in the more isolated bone marrow compartment, the transcriptional status of the more differentiated immune cells in peripheral blood and spleen is more recalcitrant; instead, HP achieves rejuvenating-like phenotypes in Het-O peripheral blood and spleen through cellular trafficking between the young and aged compartments.

Young blood environments promote solid tissue rejuvenation in aged mice

      Since the revitalized immune system is an integral part of the systemic environment, we proceeded to investigate its contribution to the systemic reversal of aging. As a proof of concept, we chose to study the systemic environment of solid tissues/organs that are distal to but are influenced by the shared circulation, such as the skin, liver, skeletal muscle, and brain. Broadly, scRNA-seq analysis shows that an aging-R effect reshaped the transcriptomic landscape in Het-O compared with Iso-O, and vice versa, that an aging-P effect was observed in Het-Y compared with Iso-Y (Figure 6A). Upon exposure to the young systemic milieu, genes involved in apoptosis were downregulated across multiple tissues, including the skin, liver, and skeletal muscle (Figures 6B and S6A–S6C). We also observed suppression of SASP-associated genes, such as Il1b, Mif, and Serpinb2, which were upregulated in Iso-O skin but reversed in Het-O skin (Figure S6C). Meanwhile, genes associated with blood vessel development and tissue morphogenesis were among upregulated aging-R genes in the Het-O parabiont (Figures S6A and S6B).

      As the global transcriptomic profiles revealed, both resident adult stem cells (i.e., basal cells and HFSCs in the skin) and their niches were reprogrammed to a state closer to Iso-Y at the transcriptomic level (Figure 6C). Immunohistochemistry staining of the skin tissues confirmed a reduced EC population in Iso-O relative to Iso-Y and its restoration in Het-O (Figure 6D). Concordantly, we observed that the global cell-cell interactions were reset to a youthful state as well, with enhanced interaction between the basal cells or HFSCs with cell types residing nearby them, namely ECs as well as fibroblasts (Figures 6E and S6D). For example, R-Spondin-2/4 (RSPO2/4) interactions with leucine-rich repeats containing G protein-coupled receptors (LGRs), known to activate Wnt/b-catenin signaling pathway critical for epithelial tissue regeneration, were re-established between basal cells and fibroblasts (Figure S6D). Thus, in parabiosis models, a young systemic milieu may also re-establishe youthful cell-cell interaction in tissues distal to the hematopoietic and immune organs, ultimately providing a more favorable microenvironment for stem cell maintenance and reducing tissue aging.

      Finally, we investigated whether manipulating specific gene expression can exert rejuvenation effects on solid tissues. We analyzed aging-R genes that were detected in at least two tissue types (Figure S6E). Notably, Tsc22d3 (also known as Gilz), an anti-inflammatory transcription factor that inactivates the NF-kB pathway, was downregulated in aged basal cells, RetiFib, and hepatocytes but reactivated by exposure to the young blood environment (Figure 6F). The siRNA-mediated knockdown of Gilz increased apoptosis and suppressed the proliferative ability of fibroblasts, in the presence of the pro-inflammatory factor IL-6 (mimicking the inflammatory microenvironment associated with aging) (Figures 6G–6I and S6F–S6H). These results support a beneficial role for Gilz suppression in maintaining a youthful state in solid tissues. Altogether, our study describes an extensive single-cell transcriptomic atlas of aging and responses to HP, through which we identified previously unappreciated factors that mediate the rejuvenation effects.

DISCUSSION

       An increasing number of studies, and in particular, in the aged skeletal muscle, brain, liver, lung, and skin, have reported on the rejuvenation effects of HP (Eggel and Wyss-Coray, 2014; Kang and Yang, 2020), and some also point to stem cells and the niche environment as critical mediators. However, a systematic survey across multiple organs that probes molecular changes in all cell types is a much-needed resource for the field to fully understand the systemic effects of HP-induced aging and rejuvenation. By constructing an HP single-cell atlas, our work reveals how blood exchange exert profound effects on systemic and global regulation of hematopoiesis and tissue homeostasis. Thus, our work opens new vistas to further explore mechanisms underlying the mysterious pro-youth effects of young blood and identifies factors as potential candidates for systemic rejuvenation.

      Whether or not aged HSPCs are responsive to HP or the young bone marrow niche has long been debated (Kuribayashi et al., 2021; Pa´lovics et al., 2022). In a very recent study, aged HSCs, after exposure to young blood, still displayed reduced regenerative capability with a myeloid-biased output in TX assays, thus claiming that HSCs are refractory to interventions mediated by blood-borne factors (Ho et al., 2021). An earlier study reported that the aged immune system appeared not to be reactivated by HP, reaching a similar conclusion (Pishel et al., 2012). However, by leveraging the robustness of single-cell technology, it is possible to detect molecular changes per cell and help identify the fraction of HSPCs that are most responsive to the changing milieus, as confirmed by a very recent HP study also using single-cell technologies (Pa´ lovics et al., 2022). Moreover, we used the CD45 congenic system to trace cell origins in parabionts and verified the rejuvenation effects of autologous HSPCs upon HP in Het-O bone marrow. In addition, we identified the underlying mechanisms and key mediators of rejuvenation, followed by validation in adult stem cells (e.g., HSCs).

      To promote rejuvenation and delay aging, cellular targets have been considered, but an alternative and attractive intervention strategy could instead be the direct delivery of pro-youth factors. Identifying such candidate molecules relies on an in-depth analysis of extrinsic blood-borne factors and molecular changes intrinsic to target cells. Our data suggest that proinflammatory chemokines CCL3 and CCL4 are associated with HP-mediated rejuvenation of HSPCs. Although not studied in HSPC aging, prior studies suggest that CCL3 may prevent HSPCs from over-proliferation and contribute toward maintaining a steady-state healthy HSPC pool (Staversky et al., 2018). Our study found that the overexpression of CCL3 in aged HSPCs indeed enhanced their differentiation ability toward T cells after long-term engraftment, which to some extent reversed the age-related myeloid bias of hematopoietic differentiation. Therefore, our study confirms and extends the list of factors reported to delay aging (such as VEGF, TIMP2, THBS4, and SPARCL1, etc.) or promote aging (such as b2-microglobulin and IL-6) based on the HP model (Baht et al., 2015; Castellano et al., 2017; Gan and S € udhof, 2019; Valletta et al., 2020; Grunewald et al., 2021). Interestingly, we also analyzed genes that cannot be rescued by HP, reflecting aging memories refractory to the changing environment, and found that these genes were mainly involved in epigenetic regulation, such as Yy1. More importantly, we overexpressed YY1 in aged mouse HSPCs, which enhanced their engraftment ability in a competitive TX assay, suggesting that YY1 is a bottleneck mechanism for rewriting aging memories of HSPCs. Thus, our comprehensive HP atlas, with which we identified key rejuvenating factors, holds the promise to provide new therapeutic interventions for systemic aging.

Limitations of the study

       Although we and Pa´lovics et al. (2022) have demonstrated the effect of young blood exposure via HP to alleviate aging in multiple tissues across the whole organism at single-cell resolution, more detailed experimental verifications with a larger cohort at both tissue and cellular levels await future investigation. Furthermore, since HP-related surgical procedures limit free movements between parabionts, which inevitably lead to certain stress conditions, HP-based rejuvenation is not exactly equivalent to that achieved by supplementing young blood in aging individuals. With these caveats in mind, our study provides critical clues for identifying biomarkers, potential intervention targets, and strategies against systemic aging.

STAR+METHODS

Detailed methods are provided in the online version of this paper and include the following:

● KEY RESOURCES TABLE

● RESOURCE AVAILABILITY

 Lead contact

 Materials availability

 Data and code availability

 EXPERIMENTAL MODEL AND SUBJECT DETAILS

 Ethical statement

 Experimental animals

 Cell culture

 METHOD DETAILS

 Parabiosis surgery

 Tissue dissociation and cell isolation

 Droplet-based scRNA-seq using the 10x Genomics Chromium Platform

 Nuclei isolation and snRNA-seq on the 10x Genomics Chromium Platform

 Flow cytometry analysis

 SA-b-gal staining

 Hematoxylin and eosin (H&E) staining

 Masson’s trichrome staining

 TUNEL staining

 Immunohistochemistry staining

 Plasmids

 Lentivirus packaging

 LSK cell isolation and lentiviral transduction

 Bone marrow cell counts

 Transplantation

 Western blot analysis

 Knockdown of Tsc22d3 / Gilz in fibroblasts

 qRT-PCR

 Immunofluorescence staining

● QUANTIFICATION AND STATISTICAL ANALYSIS

 Data processing

 Clustering and identification of cell types

 Cell composition variation analysis

 Differential expression and cell type-specific DEG network analyses

 Single-cell trajectory analysis

 Transcription factor (TF) regulatory network analysis

 Gene set score analysis

 GO analysis

 Cell-cell communication analysis

 Statistical analysis

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j. stem.2022.04.017.

ACKNOWLEDGMENTS

      We thank J.J. for assisting with the animal experiments; Y. Zheng for his help with the bioinformatic analysis; and L. Bai, R. Bai, Q. Chu, J. Lu, J. Chen, Y. Yang, L. Tian, and X. Li for their administrative assistance. This work was supported by the National Key Research and Development Program of China (2020YFA0804000), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA16000000), the National Natural Science Foundation of China (81921006, 82125011, 92149301, 92168201, 92049116, 32121001, 82192863, 91949209, 92049304, 82122024, 82071588, 82001477, 31900523, 81861168034, 81922027, and 81601233), the National Key Research and Development Program of China (2018YFC2000100, 2018YFA0107203, 2021ZD0202401, 2021YFF1201005, 2019YFA0802202, 2020YFA0803401, and 2020YFA0113400), the Program of the Beijing Natural Science Foundation (Z190019 and JQ20031), the K. C. Wong Education Foundation (GJTD-2019-06 and GJTD-2019-08), the Young Elite Scientists Sponsorship Program by CAST (YESS20210002 and YESS20200012), the CAS Project for Young Scientists in Basic Research (YSBR-012), Youth Innovation Promotion Association of CAS (E1CAZW0401), the Informatization Plan of Chinese Academy of Sciences (CAS-WX2021SF-0301), the Tencent Foundation (2021-1045), the Milky Way Research Foundation (MWRF), the Beijing Municipal Science & Technology Commission (Z200022), and the Project for Extramural Scientists of State Key Laboratory of Agrobiotechnology from China Agricultural University (2021SKLAB6-3 and 2021SKLAB6-4).

AUTHOR CONTRIBUTIONS

      G.-H.L., W.Z., and J.Q. conceptualized this project and supervised the overall experiments. S.M. performed the bioinformatics analysis of the sc/snRNAseq. S.W. and J.R. performed the data analysis of experiments. S.W., W.L., L.Z., and J. Lei. performed the animal experiments, immunostaining, and tissue section analyses. Y. Ye, R.C., M.X., and Y. Yang performed the flow cytometry analyses, LSK cell isolation, lentiviral transduction, and TX. J. Li. and Y. Zuo. performed the quality control of the sc/snRNA-seq raw data. Q. Zhao. and G.S. performed the cell culture and knockdown experiments of the fibroblasts. Y.J. performed the single nucleus isolation. S.S. performed the single-cell isolation of mouse tissues. Q.W. performed the illustrations of mouse and tissue in figures. X.L. provided the human fibroblast, and G.-H.L., W.Z., J.R., S.W., J.Q., M.S., S.Y., P.C., J.W., Q. Zhou., and J.C.I.B. wrote, reviewed, and edited the manuscript.

DECLARATION OF INTERESTS

The authors declare no competing interests.

Received: August 8, 2021

Revised: March 18, 2022

Accepted: April 25, 2022

Published: May 24, 2022

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to Dr. Guang-Hui Liu (该Email地址已收到反垃圾邮件插件保护。要显示它您需要在浏览器中启用JavaScript。).

Materials availability

This study did not generate new unique reagents.

Data and code availability

      The accession numbers for the raw sc/snRNA-seq data reported in this paper are in GSA (Genome Sequence Archive). Accession numbers are listed in the key resources table.

      No specific custom code was used in this manuscript. All the codes are publicly available and the source is annotated in the text and/or in the STAR Methods.

      Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Ethical statement

      All experimental procedures were approved by the Animal Care and Use Institutional Committee of the Chinese Academy of Sciences and the Peking Union Medical College Hospital Institutional Review Board.

Experimental animals

      The information on age and sex of three groups of C57BL6/J mice analyzed (SPF Biotechnology Co., Ltd., Beijing) are as follows: Iso-Y pairs (2-month-old, n = 7, male mice), Het-Y & Het-O pairs (2-month-old & 23-month-old, respectively, n = 5, male mice), and Iso-O pairs (23-month-old, n = 9, male mice) groups. In addition, we acquired heterochronic pairs of Het-Y (2-month-old, n = 4, CD45.2 male mice) and Het-O (23-month-old, n = 4, CD45.1 male mice) using CD45.1 / CD45.2 congenic system to trace the origin and distinguish the donor/host cells. The mice were raised in a certificated SPF grade facility with individually ventilated cages. The room was maintained at a controlled temperature of 20-25℃, a humidity of 30%-70% and light exposure cycles of 12 h light-dark cycle. Each pair of parabionts were housed in the same cage for at least two weeks prior to the parabiosis surgery. Body condition and overall appearance of the mice were checked daily.

Cell culture

      Mouse fibroblasts were isolated from the skin of 7-week-old male mouse as described previously (Cai et al., 2020). Briefly, mouse skin was minced with scissors and incubated in 8 mg/mL collagenase IV (Gibco) for 1.5 h at 37  C. After enzyme treatment, cells were collected by centrifugation and resuspended in fibroblast culture medium containing high-glucose DMEM (Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco), 2 mM GlutaMAX (Gibco), 0.1 mM NEAA (Gibco) and 1% penicillin/streptomycin (Gibco). Human skin fibroblasts were derived from the skin tissue of a 27-year-old healthy donor as previously described (Zou et al., 2021). In brief, the cells were cultured in fibroblast culture medium under permissive conditions to about 80% confluence, and then the culture medium was changed every other day. For cell passaging, 0.25% Trypsin (Gibco) was used. No mycoplasma contamination was observed during cell culture.

METHOD DETAILS

Parabiosis surgery

      Administration of analgesia (buprenorphine, 0.1 mg/kg subcutaneously) was performed before and after the surgical manipulation. The parabiosis surgery was performed as previously described in a clean animal surgery room (Xiong et al., 2018). In brief, two mice for parabiosis were anesthetized by isoflurane (RWD, R510-22), and monitored to ensure adequate depth of anesthesia. Each mouse was shaved along a contiguous line from the elbow, the flank, to the knee on the side to be joined, and the skin was scrubbed three times with iodine and 75% alcohol. The skin incisions were performed along the flank, ranging from skin proximal to the knee to skin proximal to the elbow without damaging the muscle tissue underneath using 4-0 polydioxanone suture material (Jinhuan medical, F406). The triceps and quadriceps of the parabionts were joined with two interrupted sutures. Finally, the skin of the parabionts was closed with interrupted sutures. The mice were then placed on a heated pad in a supine position, and allowed to wake up in ambient air. After the parabiosis surgery, each pair was housed individually, with 0.25% bupivacaine in 1 mL saline administered subcutaneously daily for 3 days. In the heterochronic parabiosis group, the young mice were joined on the left and the old mice on the right.

Tissue dissociation and cell isolation

      For single-cell isolation, around five to six weeks after the parabiosis surgery, tissues were isolated from randomly selected mice in Iso-Y pairs (n = 4, male mice) and Iso-O pairs (n = 4, male mice), as well as mice in Het-Y & Het-O pairs (n = 4, male mice). Briefly, the mice were anesthetized and then systemically perfused with saline through the heart. Then the tissues were collected by Consortium including Qianzhao Ji, Huifang Hu, Lanzhu Li, Ruochen Wu, Min Wang, Shijia Bi, Shikun Ma, Ping Yang, linging Geng, Hongkai Zhao, Feifei Liu, Zhiran Zou, Cui Wang, Lixiao Liu, Yusheng Cai, Fangshuo Zheng, Yaobin Jing, Yanling Fan, Shanshan Yang, Yiyuan Zhang, Sheng Zhang, Hezhen Shan, Jianli Hu, Hongyu Li, Fang Cheng, Daoyuan Huang, Xiaojuan He, Yifang He, Zehua Wang, Hui Zhang, Qun Chu, Jing Lu, Miyang Ma, Tianling Cao, Chengyu Liu, Jian Yin, Hao Li, Kaowen Yan, Xuebao Wang, Xiaoyu Jiang, Xiaoyan Sun, Yingjie Ding, Xiuping Li, Yixin Zhang, Baohu Zhang, Jinghao Hu, and Xiangmei Jin. Afterwards, selected tissues were separated and processed for single-cell isolation as follows.

      Peripheral blood was collected and EDTA-2Na was added to prevent coagulation. Erythrocytes were removed by incubating the cells with red blood cell lysis buffer (BD Biosciences) for 15–20 min at room temperature. The resultant cells were washed twice with cold phosphate buffer saline (PBS, Sigma-Aldrich) and resuspended in cold PBS with 2% FBS at a density of 1 3 106 cells/mL. To trace the origin of young (CD45.2) / aged (CD45.1) cells, freshly isolated cells were stained with CD45.1 (A20; BioLegend) and CD45.2 (104; Invitrogen) antibodies. CD45.1+ cells or CD45.2+ cells were sorted by FACS (BD Influx), and resuspended in cold PBS with 2% FBS.

      Bone marrow (BM) was isolated by crushing limb bones, including femur and tibia bones. The marrow was carefully flushed out from the bone cavity using PBS containing 2% FBS and 2 mM EDTA with a 5-ml syringe on ice. The cells were collected and filtered through a 40-mm strainer (BD Falcon) and centrifuged at 1, 200 rpm at 4 C for 5 min. The resuspended cells were incubated with red blood lysis buffer (BD Biosciences) for 5 min, washed twice with cold PBS, and finally resuspended in cold PBS with 2% FBS at a density of 1 3 106 cells/mL. To trace the origin of young (CD45.2) / aged (CD45.1) cells, freshly isolated BM cells were stained with CD45.1 (A20; BioLegend) and CD45.2 (104; Invitrogen) antibodies. CD45.1+ or CD45.2+ cells were sorted by FACS (BD Influx), and resuspended in cold PBS with 2% FBS.

      To isolate LSK cells, bone marrow cells were stained with antibodies against Lineage cocktail (BioLegend), c-Kit (2B8; BioLegend), and Sca-1 (D7; BioLegend) in PBS containing 2% FBS. The cells were then sorted through a flexible Influx cell sorter (BD Biosciences) according to manufacturer’s instructions (Figure S2A). Lineage markers included: anti-mouse CD3 (145-2C11), B220 (RA3-6B2), CD11b (M1/70), Ly6G/Ly-6C (RB6-8C5), Ter-119 (Ter-119). To trace the origin of young (CD45.2) / aged (CD45.1) cells, freshly isolated LSK cells were stained with CD45.1 (A20; BioLegend) and CD45.2 (104; Invitrogen) antibodies. Then, the CD45.1+ or CD45.2+-LSK cells were sorted by FACS (BD Influx) and resuspended in cold PBS with 2% FBS. Due to the limited cell numbers of CD45.1- positive LSK cells from the Het-Y mice and CD45.2-positive LSK cells from the Het-O mice, we supplemented HEK293T cells to each sample. 

      To isolate spleen cells, spleen tissues were dissected and placed in 10 mL PBS containing 2% FBS, and gently disaggregated through a 40-mm cell strainer (BD Falcon) using a 5 mL syringe plunger. All spleen cells were collected and centrifugated at 1,200 rpm 4℃ for 5 min. The red blood cells were removed by lysing with 5 mL lysing buffer (BD Biosciences) for 3-5 min at room temperature. After centrifugation at 1,200 rpm 4℃ for 5 min, cells were washed twice with cold PBS and resuspended in cold PBS containing 2% FBS at a density of 1 3 106 cells/mL (Xie et al., 2020). To trace the origin of young (CD45.2) / aged (CD45.1) cells, freshly isolated spleen cells were stained with CD45.1 (A20; BioLegend) and CD45.2 (104; Invitrogen) antibodies. CD45.1+ or CD45.2+ cells were sorted by FACS (BD Influx), and resuspended in cold PBS with 2% FBS.

      To isolate skin cells, skin tissues were collected and stored in ice-cold PBS. After a wash with cold PBS, tissues were minced with scissors into small pieces in PBS on ice, transferred into 15-mL centrifuge tubes, and incubated at 37℃ for 1 h in digestion solution (PBS supplemented with 1 mg/mL collagenase I, 1 mg/mL collagenase IV, 1 mg/mL dispase and 0.125% trypsin-EDTA). The digestion was stopped by adding DMEM (Gibco) containing 10% FBS. Dissociated cells were collected by centrifugation at 1,000 rpm for 5 min at 4℃, resuspended and incubated in red blood cell lysis buffer (BD Biosciences) for 5 min, centrifuged at 1,000 rpm at 4 C for 5 min and resuspended in cold PBS containing 10% FBS. The cell suspension was passed through a 40-mm strainer (BD Falcon), washed twice with PBS, centrifuged at 1,000 rpm at 4 C for 5 min, and resuspended in cold PBS containing 10% FBS.

      Dissociated cells were FACS sorted (BD Influx) to exclude cell debris and Propidium iodide (PI)-positive dead cells, and then counted with a dual fluorescence cell counter (LUNA-FLTM, Logos Biosystems), after which the resultant single-cell suspension was subjected to 10x Genomics-based single-cell RNA-sequencing. All the antibodies used in this study are listed in key resources table.

Droplet-based scRNA-seq using the 10x Genomics Chromium Platform

      Single cells were encapsulated into droplet emulsions using a Chromium Single-Cell instrument (10x Genomics), and scRNA-seq libraries were constructed following the 10x Genomics protocol using a Chromium Single-Cell 3’ Gel Bead and Library V2 Kit. The isolated single cells were then loaded in each channel, with a target output of 5,000 cells per sample. Reverse transcription and library preparation were performed in a Bio-Rad C1000 Touch thermal cycler with a 96-deep well reaction module. A total of 12 cycles were used for cDNA amplification and the sample index PCR step. Amplified cDNA and final libraries were assessed on a fragment analyzer using a High Sensitivity NGS Analysis Kit (Advanced Analytical). The average fragment length in the 10x cDNA libraries was quantified in a fragment analyzer (AATI) and by qPCR with a Kapa Library Quantification Kit for Illumina. Each individual library was diluted to 2 nM, and pools from equal volumes of 16 libraries were sequenced for each run in the NovaSeq 6000 Sequencing System (Illumina 20012866).

Nuclei isolation and snRNA-seq on the 10x Genomics Chromium Platform

      Fresh frozen brain, liver and skeletal muscle tissues were homogenized in 1 mL homogenization buffer using a freezing multi-sample tissue grinding system (60 Hz, 60 s, 2 times for brain; 60 Hz, 60 s, 3 times for liver and skeletal muscle) (TissueLyser-24, Jingxin Industrial Development, China) according to a published protocol with modifications (Krishnaswami et al., 2016). Homogenization buffer contains 250 mM sucrose, 25 mM KCl, 25 mM MgCl2, 10 mM Tris buffer, 1 mM DTT, 1 X protease inhibitor, 0.4 U/mL RNaseIn (Thermo Fisher Scientific), 0.4 U/mL Superasin (Thermo Fisher Scientific), 0.1% Triton X-100 (v/v), 1 mM PI (Thermo Fisher Scientific), and 10 ng/mL Hoechst 33342 (Thermo Fisher Scientific). After being passed through a 40-mm filter (BD Falcon), nuclei were centrifuged at 500 g for 5 min, and resuspended in PBS supplemented with 0.3% BSA (Gemini), 0.4 U/mL RNaseIn (Promega N2615) and 0.4 U/mL Superasin (Invitrogen, AM2694). Nuclei positive for Hoechst 33342 and PI were sorted by FACS (BD Influx) and counted using a dual fluorescence cell counter (Luna-FLTM, Logos Biosystems). The nuclei from the same tissues (n = 4) were pooled before single-nucleus capture using the 10x Genomics Single-Cell 3’ system. Approximately 7,000 nuclei per sample were captured following the standardized 10x capture and library preparation protocol (10x Genomics) and sequenced in a NovaSeq 6000 Sequencing System (Illumina, 20012866).

Flow cytometry analysis

       Cells were isolated freshly or recovered from frozen-stock samples in 10% DMSO plus 90% serum. The cells were incubated with the following fluorescent chromogen-conjugated monoclonal antibodies: CD45.1 (A20; BioLegend), CD45.2 (104; eBioscience), B220 (RA3-6B2; BioLegend), IgM (RMM-1; BioLegend), CD43 (S11; BioLegend), CD3 (145-2C11; BioLegend), CD8 (53-6.7; BioLegend), CD62L (MEL-14; BioLegend), and CD44 (IM7; BioLegend) for 30 min at 4 C, and then washed twice with PBS. 7-AAD was used to exclude dead cells. Flow cytometry data were acquired on Fortessa (BD Biosciences) and analyzed using FlowJo software (TreeStar). The antibodies used were listed in key resources table.

      The freshly collected fibroblasts were stained with Annexin V-EGFP Apoptosis Detection Kit (Vazyme Biotechnology, A211-02) following the manufacturer’s instructions and then subjected for flow cytometry by BD LSRFortesa flow cytometer, and data were analyzed by FlowJo software (TreeStar).

SA-b-gal staining

       SA-b-gal staining was performed as previously described (Zhang et al., 2019). In brief, the frozen sections of the mouse tissue were recovered to room temperature, and allowed to air dry, washed with PBS, and fixed with 2% formaldehyde and 0.2% glutaraldehyde at room temperature for 15 min. Then, the samples were incubated with 1 mg/mL X-gal staining solution at 37 ℃ for 48-96 h. The slides were then mounted with 80% glycerol. Images were collected under a microscope (Nikon), and the percentage or intensity of SA-b-gal-positive area was quantified by ImageJ.

Hematoxylin and eosin (H&E) staining

      Hematoxylin and eosin (H&E) staining was performed as previously reported (Li et al., 2020; Wang et al., 2020; Zhang et al., 2020). The paraffin-embedded tissues were sectioned at a thickness of 5 mm with a rotary microtome. Sections were deparaffinized in xylene and rehydrated in gradient alcohols (100%, 100%, 95%, 80%, 70%), incubated in hematoxylin solution, rinsed in running water to remove the excess hematoxylin, differentiated with 1% acid alcohol for 5-10s and then rinsed in running water for 10 min. Lastly, sections were stained with eosin, dehydrated in gradient ethanol and xylene, and mounted with cytoseal-60 (Stephens Scientific).

Masson’s trichrome staining

      Masson’s trichrome staining was performed according to the manufacturer’s instructions (Solarbio, G1346). After embedding in paraffin, tissues were sectioned at a thickness of 5 mm, deparaffinized with xylene, and hydrated with 100% alcohol, 95% alcohol, 70% alcohol and running tap water. Then the sections were stained with potassium bichromate solution overnight. After rinsing with running tap water for 10 min, the sections were stained in iron hematoxylin working solution for 5 min and Ponceau-acid fuchsin solution for 10 min. Then, the slides were differentiated in phosphomolybdic-phosphotungstic acid solution for 10 min and stained with aniline blue solution for 3-5 min. This step was followed by a brief rinse in distilled water and differentiation in 1% acetic acid solution for 2-5 min. Then the sections were dehydrated with 70% alcohol, 95% alcohol and 100% alcohol, then removed with xylene and covered with a cover glass, and mounted with cytoseal-60. The image was captured by a slice scanner (Leica, CS2).

TUNEL staining

      TUNEL staining was performed using a TUNEL cell apoptosis detection Kit (Beyotime, C1088) according to the manufacturer’s instructions (Wang et al., 2020). Images were acquired using a Leica SP5 laser scanning confocal microscope, and the percentages of positive cells were quantified using ImageJ.

Immunohistochemistry staining

       After paraffin embedding, the tissues were sectioned at a thickness of 10 mm, deparaffinized with xylene, and hydrated with 100% alcohol, 95% alcohol, 85% alcohol, 75% alcohol, 50% alcohol, and running tap water. Antigen retrieval was performed by microwaving in 10 mM sodium citrate buffer (pH 6.0) three times for 5 min each. After cooling down to room temperature, the sections were rinsed with PBS three times. After permeabilization in 0.4% Triton X-100 for 1.5 h, endogenous peroxidase blockade was performed with 3% H2O2 for 20 min. Then the sections were blocked in 5% donkey serum in PBS for 1 h at room temperature, and then incubated with the primary antibodies overnight at 4 C. Sections were then incubated with secondary antibodies at room temperature for 1 h, followed by colorimetric detection with DAB Kit (ZSJQ-BIO) and counterstaining with hematoxylin, differentiation with 1% hydrochloric acid and alcohol. Finally, sections were dehydrated with 70% alcohol, 95% alcohol and 100% alcohol, and xylene and mounted in the neutral resinous mounting medium. The image was captured by a slide scanner (Leica, CS2). The antibodies used in this study are listed in key resources table.

Plasmids

      The cDNAs of mouse Yy1 and Ccl3 were amplified by PCR and cloned into SF-LV-EGFP vector. Primers were synthesized by Tsingke Biotech (Beijing, China). The primer pairs are listed in Table S6.

Lentivirus packaging

      Lentivirus was packaged by the plasmid psPAX2 (#12260, Addgene) and pMD2G (#12259, Addgene). For lentivirus production, HEK293T cells were plated in a 10 cm tissue culture plate and transfected with 10 mg lentiviral plasmid, 10 mg psPAX2 plasmid and 3 mg pMD2G plasmid using Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific). After 8 h, the medium was changed, and viral particles were collected at 24 h, 48 h and 72 h by a 2.5-hr ultracentrifugation at 19,400 g at 4℃. Pellets were re suspended in SFEM medium (Stem Cell Technology, catalog No.09650) with 1% penicillin/streptomycin, 20 ng/ml mouse stem cell factor (SCF), and 20 ng/ml mouse thrombopoietin (TPO) and aliquots were stored at -80℃.

LSK cell isolation and lentiviral transduction

      Bone marrow cells were isolated from 24-month-old mice by crushing the bones, including femur, tibia, and pelvis, with pestle and mortar in Hanks balanced salt solution (HBSS) containing with 2% FBS and 1% N-2-hydroxyethylpiperazine-N0 -2-ethanesulfonic acid (HEPES) buffer (HBSS+ ) on ice. Cells were acquired after incubation with c-Kit (2B8; Invitrogen) antibody and enrichment by magnetic cell sorting (MACS, Miltenyi Biotec, Germany) from whole bone marrow cells, and c-Kit+ cells were labelled by Sca-1 and lineage antibodies. Lineage markers included: anti-mouse CD3 (145-2C11; BioLegend), CD4 (RM4-5; BioLegend), CD8 (53-6.7; BioLegend), B220 (RA3-6B2; BioLegend), CD11b (M1/70; BioLegend), Ly6G/Ly-6C (RB6-8C5; BioLegend) and Ter-119 (Ter-119; BioLegend). All lineage Sca-1+ c-Kit + (LSK) cells were FACS sorted (BD Fusion), and 2 3 105 LSK cells were cultured in SFEM medium with 1% penicillin/streptomycin (Gibco), 20 ng/ml mSCF (Peprotech), and 20 ng/ml mTPO (Peprotech), in a 96-well plate (Nest Biotech, catalogue No. 701111). Lentivirus suspensions were added into LSK cells at a calculated multiplicity of infection (MOI). After 72 h culture, cells were collected and stained with Sca-1 (D7; Thermo) and CD48 (HM48-1; BioLegend) antibodies. 40 ,6-diamidino-2-phenylindole (DAPI, Invitrogen) staining was used to exclude dead cells. EGFP+ Sca-1+ CD48- cells were sorted by FACS (BD Fusion) and injected into lethally irradiated recipients. The antibodies used for LSK isolation were listed in key resources table.

Bone marrow cell counts

      Bone marrow cells were harvested from mice in HBSS+ buffer on ice and cell counts were analyzed using Vi-Cell XR cell viability analyzer (Beckman Coulter).

Transplantation

       For the competitive bone marrow transplantation experiments, the recipient mice (CD45.2) were lethally irradiated (10 Gy gamma irradiation) using X-ray irradiator (RS-2000, Rad Source Technologies) before transplantation. 2,000 EGFP+ Sca-1+ CD48- cells (CD45.2) together with 2 3 105 whole bone marrow cells from 2 to 3-month-old mice (CD45.1) as a competitor were transplanted intravenously into 8-week recipient mice. Donor-derived chimerism (including myeloid cells, B cells, and T cells) was monitored every 4 weeks by FACS analysis of the peripheral blood from the recipient using a cocktail of anti-CD45.1 (A20; BioLegend), anti-CD11b (M1/70; BioLegend), anti-B220 (RA3-6B2; BioLegend) and anti-CD3 (145-2C11; BioLegend) antibodies. Propidium iodide, or 40 ,6-diamidino-2-phenylindole (DAPI) was used to exclude dead cells. The antibodies used to evaluate peripheral blood chimerism were listed in key resources table.

Western blot analysis

       Infected LSK cells after 72 h culture were lysed with 1 3 SDS lysis buffer (62.5 mM Tris-HCl, pH 6.8, 2% [wt/vol] SDS) and heated at 105℃ for 10 min. Western blot was performed as previously described. Briefly, total protein concentration was determined via BCA following the manufacturer’s instructions, and the protein lysates were subjected to SDS-PAGE. The PVDF (polyvinylidene fluoride) membranes (Millipore) were blocked with 5 % skimmed milk powder (BBI Life Sciences) in 1 3 TBST, and incubated with antibodies against YY1 (Santa Cruz), or CCL3 (Santa Cruz) overnight at 4  C. The PVDF membranes were incubated with the HRP-conjugated secondary antibodies for 1 h at room temperature, and followed by visualization using the ChemiDoc XRS system (Bio-Rad). The antibodies used were listed in key resources table.

Knockdown of Tsc22d3 / Gilz in fibroblasts

      siRNA molecules specifically targeting the mRNA of mouse Gilz and human GILZ and the negative control (NC) duplex were purchased from RIBOBIO (China). The sequences were listed in key resources table. Fibroblasts were transfected with a negative control duplex or siRNAs against target gene using Lipofectamine RNAiMAX transfection reagent (Thermo Fisher Scientific) following the manufacturer’s instructions for 48 h and then treated with 50 ng/mL IL-6 for 24 h, which mimics the inflammatory environment in the aged tissues. The cells were collected for qRT-PCR, apoptosis analysis and Ki67 immunostaining.

qRT-PCR

      TRIzol Reagent (Thermo Fisher Scientific) was used to extract total RNA and 2 mg total RNA was used for cDNA synthesis with reverse transcription master mix (Promega). qRT-PCR was conducted with the iTaq Universal SYBR Green Super Mix (Bio-Rad) by a Real-Time qPCR (quantitative PCR) system (Thermo Fisher Scientific). All data were normalized to the internal control transcript and calculated by 2-DDCq method. The primer pairs used in this study are listed in Table S6.

Immunofluorescence staining

      The fibroblasts were fixed with 4%PFA for 25 min, permeabilized with Triton X-100 (0.4% in PBS) for 25 min, incubated with blocking buffer (10% donkey serum in PBS) for 1 h at RT, and stained with Ki67 antibody overnight at 4 C. Then, the cells were incubated with secondary antibodies for 1 h and Hoechst 33342 (Thermo Fisher Scientific) for 5 min at RT. The image was captured by a confocal laser-scanning microscope (Zeiss 900 confocal system). The antibodies used in this study were listed in key resources table.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data processing

      The sequencing data generated by the NovaSeq platform were first processed using bcl2fastq (version 2.20.0.422) to convert BCL files into FASTQ format. Next, the reads were aligned to the mm10 reference genome, and the filtered count matrix was obtained for further analysis by calculating gene counts using the default parameters of Cell Ranger (version 3.1.0) (https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/3.1/what-is-cell-ranger). In addition, because of supplementation of human HEK293T cells in the CD45.1-positive cells of LSK from Het-Y sample and CD45.2-positive cells of LSK from Het-O sample, the raw data of these two samples were aligned with the merged reference genome of both hg19 and mm10 by using Cell Ranger.

Clustering and identification of cell types

      The single-cell expression matrix calculated by Cell Ranger of LSK, bone marrow, peripheral blood, spleen, and immune cells of skin was performed with the Scanpy (version 1.4.4) for filtering low quality data, integration, normalization, dimensionality reduction, clustering, and gene differential expression analysis (Wolf et al., 2018). We first processed the data using the tutorial of scrublet (version 0.2.1) package to detect and filter the doublets (Wolock et al., 2019). The filtered scRNA-seq matrixes were analyzed in an order of the following steps:

Step 1: We discarded cells with gene number below 500 and above 6,000, or UMI number below 500 and above 40,000, or a mitochondrial genes ratio of greater than 10%, or a ribosomal genes ratio of greater than 40%.

Step 2: Integration and normalization across all samples were conducted using the standard Scanpy procedure and the batch effect was corrected by using the ‘‘sc.external.pp.bbknn’’ function.

Step 3: The first 75 principal components (PCs) were used to execute the dimensionality reduction and clustering by using ‘‘sc.tl.umap,’’ ‘‘sc.pp.neighbors,’’ and ‘‘sc.tl.louvain’’ functions of Scanpy.

Step 4: Differential expression analysis for each cluster was performed using the Wilcoxon rank-sum test as statistical method in the ‘‘FindAllMarkers’’ function with the cutoff of adjusted P-values < 0.05 and |logFC| > 0.5 by using Seurat (version 3.2.3) (Satija et al., 2015).

      We exclude the low-quality clusters with high expression of Gm42418 or AY036118 from the atlas and repeat the above steps (2-4). The details of the markers for each cell type are listed in Table S1. Cell types were assigned to each cluster using canonical marker genes.

      Through the above pipeline, we processed the scRNA-seq data of 74,323 high-quality cells to create a single-cell atlas of mouse hematopoietic and immune system.

      For the samples from the CD45.1 and CD45.2 system, firstly, mouse-derived LSK cells with other single-cell expression matrix calculated for bone marrow, peripheral blood and spleen and were also subjected to filtering for low quality data, integration, normalization, dimensionality reduction, clustering, gene differential expression analysis and cell type annotation as mentioned above. Through the pipeline, we processed the scRNA-seq data of 54,049 high-quality cells to create a single-cell atlas of mouse hematopoietic and immune system from CD45.1 and CD45.2 system.

      The analysis of skin, liver, skeletal muscle, and brain samples were performed with Seurat (version 3.2.3). We detected and filtered doublets using the DoubletFinder (version 2.0.2) package (McGinnis et al., 2019), and excluded the cells with gene number below 200 or a mitochondrial genes ratio of greater than 5% (the samples from skin were excluded the cells with gene number below 500 or a mitochondrial genes ratio of greater than 15%). Each sample dataset was normalized using the Seurat’s ‘‘SCTransform’’ function. Integration and normalization across all samples for each tissue were performed using the ‘‘PrepSCTIntegration’’ and ‘‘FindIntegrationAnchors’’ function of Seurat. The first 22-31 PCs (30 for skin, 22 for liver, 31 for skeletal muscle and 30 for brain) were used to execute the dimensionality reduction and clustering by using ‘‘RunUMAP’’, ‘‘FindNeighbors’’ and ‘‘FindClusters’’ functions of Seurat. Differential expression analysis for each cluster was performed as same as the above step 4 and exclude the low-quality clusters that were highly expression Gm42418 or AY036118 from the atlas and repeat the above steps (2-4). The details of the markers for each cell type are listed in Table S1. Cell types were assigned to each cluster using canonical marker genes.

Cell composition variation analysis

      Prior to analysis of cell ratio changes, we excluded cell types with fewer than 2.5% of the total number of cells in one tissue. The number of cells for each type in different groups (Iso-Y, Het-Y, Iso-O, and Het-O) was counted and divided by the total number of cells in the same group to calculate the cell type ratio. Based on these ratios, the percentage of specific cell types in each group was calculated. The Log2FC of cell type ratios between the Iso-O and Iso-Y groups was calculated to identify the cell types altered during aging (|Log2FC| > 0.5), the Log2FC of cell type ratios between the Het-Y and Iso-Y or Het-O and Iso-O groups was calculated to identify the cell types altered by parabiosis (|Log2FC| > 0.5). In addition, for the samples from the CD45.1 and CD45.2 systems, the cell proportion in single-cell data for each sample was firstly multiplied by the mean percentage of FACS for each sample.

Differential expression and cell type-specific DEG network analyses

      Differential expression analysis for each cell type between different groups (Iso-O/Iso-Y, Het-Y/Iso-Y and Het-O/Iso-O) was performed using the Wilcoxon rank-sum test as statistical method in the ‘‘FindMarkers’’ function of the Seurat package (version 3.2.3). Before performing the differential expression analysis, we filtered out cell types with three or fewer cells or missing from the comparison groups (Iso-O/Iso-Y, Het-Y/Iso-Y and Het-O/Iso-O). First, DEGs between the Iso-O and Iso-Y groups were calculated to generate an aging-related DEG dataset (aging DEGs) (|LogFC| > 0.25, adjusted p value < 0.05). Next, DEGs between the Het-Y and Iso-Y groups were calculated to generate a HY DEGs (|LogFC| > 0.25, adjusted p value < 0.05), and DEGs between the Het-O and Iso-O groups were calculated to generate a HO DEGs (|LogFC| > 0.25, adjusted p value < 0.05) (see the detailed list of DEGs in Table S2). Two genes (Gm42418 and AY036118) were removed from the DEGs list because of that may not the real biological changes (Kimmel et al., 2020). Based on these results, ‘‘aging-P DEGs’’ were defined as a subset of HY DEGs that were changed in the same direction as aging DEGs; ‘‘aging-R DEGs’’ were defined as a subset of HO DEGs that were changed in the opposite direction as aging DEGs.

Single-cell trajectory analysis

       To build the lineage tree in the mouse hematopoietic and immune system single-cell atlas, we used the PAGA module in Scanpy (Wolf et al., 2019); with iter = 1,000, and constructed the trajectory using the ‘fa’ layout. Pseudotime analysis was implemented on cells of randomly sampled (downsample = 3,000 from each group) atlas data using the Monocle2 R package (Qiu et al., 2017). Gene ordering was calculated using expression in at least 10 cells as a cutoff and a combination of inter-cluster differential expression and dispersion with q-value less than 0.01. The structure of the trajectory was drawn in a 2-dimensional space using the DDRTree dimensionality reduction algorithm and cells described above were ordered in pseudotime. We identified the top 1000 differentially expressed genes (DEGs) (p value < 0.001) by ‘‘BEAM_rest’’ function of Monocle2 and fell into 6 distinct gene clusters and depicted successive waves of gene expression by ‘‘plot_genes_branched_heatmap’’ function of Monocle2 (see the detailed list of genes in Table S3).

Transcription factor (TF) regulatory network analysis

      Transcription factor regulatory network analysis was implemented via SCENIC workflow (version 1.1.2.2) with default parameters using mm10 database from RcisTarget (version 1.6.0) as a reference (Aibar et al., 2017; Herrmann et al., 2012). GENIE3 (version 1.6.0) was invoked to build gene regulatory networks based on the DEGs across cell types of bone marrow, peripheral blood, and spleen. Enriched TF-binding motifs, predicted candidate target genes (regulons) and regulon activity were acquired from RcisTarget. Cytoscape (version 3.7.2) was used to visualize the transcription regulatory network (Shannon et al., 2003; Table S4).

Gene set score analysis

      Public gene sets were acquired from GSEA (http://www.gsea-msigdb.org/gsea/msigdb/search.jsp) database. Gene sets were used to score each input cell with the Seurat function ‘‘AddModuleScore’’. Changes in the scores across Iso-O, Iso-Y, Het-Y and Het-O group were calculated by ggpubr package using Wilcoxon test (https://github.com/kassambara/ggpubr) (version 0.2.4).

GO analysis

       Metascape (version 3.5) (Zhou et al., 2019) (http://metascape.org) was used to perform GO biological process and pathway enrichment analyses. The results were further visualized with the ggplot2 R package (p value < 0.01) (Wickham, 2016). GO enrichment analysis of multiple tissues/cell types was performed by selecting the top 10 of the top 30 most significantly enriched GO terms or pathways (p value < 0.01). Heatmaps and gene expression profile cluster plots were acquired using the pheatmap R package. Graphics were generated with ggplot2, and figures were prepared with Inkscape.

Cell-cell communication analysis

       Cell-cell communication analysis based on the scRNA-seq data was performed using CellPhoneDB software (version 1.1.0) (www. cellphonedb.org) (Vento-Tormo et al., 2018). Only receptors and ligands expressed in at least 10% cells of a given cell type were further evaluated, while the interaction was considered nonexistent if either the ligand or the receptor was not qualified. The average expression of each ligand-receptor pair was compared between different cell types, and only those with p value < 0.05 were used for subsequent prediction of cell-cell communication in the three groups (Iso-Y, Iso-O, and Het-O) across tissues (Table S5).

Statistical analysis

       All data were statistically analyzed using one-tailed t test, two-tailed t test, Wilcoxon test to compare differences between groups, with the assumption that PRISM software (GraphPad 6 Software) or R packages have equal variances. A p value < 0.05 was considered statistically significant. p values are presented in indicated figures as appropriate. ***p < 0.001, **p < 0.01; *p < 0.05; ns, not siginficant.

This article is excerpted from the Cell Stem Cell 29, 990–1005.e1–e10, June 2, 2022 by Wound World.