¡Descarga índice de contenidos, materia, año, curso, autor, profesor. Los documentos con una y más Esquemas y mapas conceptuales en PDF de Etología solo en Docsity!
Bone marrow niches orchestrate stem cell hierarchy and
immune tolerance
Kazuhiro Furuhashi *,2,3,10,11, Miwako Kakiuchi *,2,3, Ryosuke Ueda *,1,2,3, Hiroko Oda^1 , Simone Ummarino 4,5,6,7, Alexander K. Ebralidze 4,5,6,7, Mahmoud A. Bassal 5,7,8, Chen Meng 1,9, Tatsuyuki Sato^1 , Jing Lyu 1,9, Min-guk Han^1 , Shoichi Maruyama^10 , Yu Watanabe^10 , Yuriko Sawa^10 , Daisuke Kato^12 , Hiroaki Wake^12 , Boris Reizis^13 , John A. Frangos^14 , David M. Owens 15,16, Daniel G. Tenen 4,5,6,7,8, Ionita C. Ghiran^1 , Simon C. Robson^1 , Joji Fujisaki 1,2,3,17, (^1) Center for Inflammation Research, Department of Anesthesia, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 02215, USA; (^2) Columbia Center for Translational Immunology, Department of Medicine, Columbia University, Vagelos College of Physicians and Surgeons, New York, NY 10032, USA (^3) Columbia Stem Cell Initiatives, Columbia University, Vagelos College of Physicians and Surgeons, New York, NY 10032, USA (^4) Harvard Medical School Initiative for RNA Medicine, Harvard Medical School, Boston, MA, 02115, USA (^5) Harvard Stem Cell Institute, Harvard Medical School, Boston, USA (^6) Cancer Research Institute, Beth Israel Deaconess Medical Center, Boston, 330 Brookline Avenue Boston, MA 02215, USA (^7) Division of Hematology and Oncology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, USA (^8) Cancer Science Institute of Singapore, National University of Singapore, Singapore
(^18) Correspondence: Joji Fujisaki, Center for Inflammation Research, Department of Anesthesia, Beth Israel Deaconess Medical Center, Harvard Medical School, 3 Blackfan Cir, Boston MA, 02115, jfujisak@bidmc.harvard.edu. Author Contributions: KF, MYK, and RU performed flow cytometric analysis to characterize NOHi^ HSCs. KF, MYK, and HO performed competitive transplantation assays. KF and MYK and RU performed histological analyses. MYK and RU performed allogeneic HSC transplantation. SU and AKE performed CD200R knockdown. RU performed HSC transplantation following CD200R knockdown. KF, YW and YS performed tissue-scanning electron microscope studies. DK and HW assisted confocal microscope studies. CM, TS, JL, and MH assisted bone marrow isolation for transplantation and flow cytometric analysis. SM assisted in the interpretation of ciliated structures in electron microscope images. KF, MYK, and RU performed the data analysis of the transplantation assays, flow cytometric analyses, and histological analyses. MAB performed RNA sequencing analysis. BR provided the protocol and expertise for using Pdzk1ip1creER^ mice. RU and ICG performed immunostaining studies. ICG and JF interpreted the results of immunostaining. DMW provided the protocol and expertise for using CD200-flox mice. JAF provided the protocol and expertise for using eNOS-flox mice. DGT provided intellectual input into the design of knockdown studies. KF, MYK, RU, SCR, and JF interpreted the data. SCR provided intellectual input into the writing. JF wrote the paper. *KF, MK, and RU contributed equally Competing Interest Declaration : S.C.R. is a scientific founder of Purinomia Biotech Inc. and consults for eGenesis.; his interests are reviewed and managed by HMFP at Beth Israel Deaconess Medical Center following their conflict-of-interest policies. All other authors declare no competing interest. Additional information: Source data are provided with this manuscript. Supplementary Information : Supplementary methods, figures, videos, and notes concerning statistics and reproducibility are available.
HHS Public Access
Author manuscript
Nature. Author manuscript; available in PMC 2025 March 31.
Published in final edited form as:
Nature. 2025 February ; 638(8049): 206–215. doi:10.1038/s41586-024-08352-6.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
(^9) Department of Anesthesiology, Taihe Hospital, Hubei University of Medicine, Shiyan, Hubei, China (^10) Department of Nephrology, Nagoya University Graduate School of Medicine, Nagoya, Aichi 466-8550, JAPAN (^11) Nagoya University Institute for Advanced Research, Nagoya, Aichi 464-8601, JAPAN (^12) Department of Anatomy and Molecular Cell Biology, Nagoya University Graduate School of Medicine, Nagoya, Aichi 466-8550, JAPAN (^13) Translational Immunology Center, Department of Pathology, New York University, New York 10016, NY, USA (^14) La Jolla Bioengineering Institute, La Jolla, CA 92037, USA (^15) Department of Dermatology, Columbia University Irving Medical Center, Vagelos College of Physicians & Surgeons, New York, NY 10032, USA (^16) Department of Pathology & Cell Biology, Columbia University Irving Medical Center, Vagelos College of Physicians & Surgeons, New York, NY 10032, USA (^17) Division of Clinical Immunology, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 02215, USA
Abstract
Stem cells reside in specialized microenvironments, termed niches, at several different locations within tissues^1 –^3. The differential functions of heterogeneous stem cells and niches are important given the increasing clinical applications of stem cell transplantation and immunotherapy. It remains unknown whether hierarchical structures amongst stem cells at distinct niches exist and further control aspects of immune tolerance. Here, we propose novel hierarchical arrangements within hematopoietic stem cells (HSCs) and bone marrow (BM) niches that dictate both regenerative potential and immune privilege. High-level nitric oxide-generating (NOHi) HSCs are refractory to immune attack and exhibit “delayed” albeit robust long-term reconstitution. Such highly immune-privileged, primitive NOHi^ HSCs colocalize with distinctive capillaries, characterized by primary ciliated endothelium and high immune checkpoint molecule CD levels. These capillaries regulate the regenerative functions of NOHi^ HSCs through ciliary protein IFT20/CD200/eNOS/autophagy signals, further mediating immune protection. Notably, previously described niche constituents, sinusoidal cells, and Type H vessels^2 –^10 co-localize with less immune-privileged and less potent NOLow^ HSCs. Therefore, we have identified highly immune-privileged, “late-rising” primitive HSCs and characterized their immunoprotective niches constituted of specialized vascular domains. Our studies suggest the niche may orchestrate hierarchy in stem cells and immune tolerance, delineating future immunotherapeutic targets.
Heterogeneity within somatic stem cells may be linked to their respective niches. Different sites within the brain, gastrointestinal tract, liver, lung, and BM have been reported, albeit controversially, to serve as niches^1 –^3. Earlier reports suggested HSC colocalization with endosteum osteoblasts^11 ,^12 , while more recent studies have demonstrated HSCs at central marrow sinusoids^2 –^8. Still, others have suggested that HSCs co-localize with BM
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
NOHi, NOLowCD200R+, or NOLowCD200R−^ HSCs (10 cells/recipient; CD45.1) were
transplanted into lethally-irradiated CD45.2congenic mice, with competitor BM cells
(200,000 cells/recipient; CD45.2). Serial transplantations were performed twice at five- month intervals (Extended Data Fig. 2a).
For the first two months, NOHi^ HSCs and NOLowCD200R+^ HSCs exhibited comparable levels of multi-lineage hematopoietic reconstitution, whereas NOLowCD200R−^ HSCs exhibited limited reconstitution (Fig. 1g; Extended Data Fig. 2b). However, NOHi^ HSCs increased reconstitution over time. More than three months post-transplantation, NOHi^ HSCs
exhibitedrobust reconstitution. Secondary transplantationfurther increased NOHi^ HSC
reconstitution, while NOLow^ HSC reconstitution was largely abolished (Fig. 1h–i; Extended Data Fig. 2b–c).
Analysis of reconstitution trends within individual recipients confirmed “increasing
reconstitution” throughout the first and secondary transplantation exclusively in NOHi^ HSC
recipients (4/17) (Fig. 1h–i; Extended Data Fig. 2c–d).
Notably, certain NOHi^ HSC recipients (3/17) displayed non-significant levels of chimerism at two to five months after the initial transplantation, while serial transplantation markedly induced high levels of chimerism (10–70%) (Fig. 1i; Extended Data Fig. 2c–d). This
pattern of “sleeping beauty-like reconstitution” suggested NOHi^ HSCs are enriched with
“late-rising” HSCs, which remain initially dormant while exhibiting robust reconstitution
over time, particularly after the serial transplantation.
In contrast, NOLow^ HSCs frequently exhibited “exhaustion” and/or minimal engraftment (Fig. 1i; Extended Data Fig. 2c–d). At four months after the secondary transplantation, engraftment was sustained in only a few NOLow^ HSC recipients (1/37) but frequently found in NOHi^ HSC recipients (10/17). Such differential trends were sustained following the third transplantation (Extended Data Fig. 2e).
These data indicate NO expression distinguishesrobust, late-rising HSCs fromless potent,
exhausted HSCs.
These observations were reproduced by additional transplantation studies using a higher dose of competitor BM cells (300,000 cells/recipient). Of note, the use of these competitor
dosesmore frequently induced“sleeping beauty-like reconstitution” from NOHi^ HSCs
(6/20). These results were found inall four independent studies conducted, supporting
reproducibility (Extended Data Fig. 3a–e).
Moreover, this high competitor dose inducedseveral different “late-rising” patterns
(Extended Data Fig. 3d–e). Certain recipients (2/20) initiated NOHi^ HSC reconstitution three to four months after the first transplantation. Other NOHi^ HSC recipients (2/20)
demonstrated “reconstitution re-emergence” four to five months after the perceived
“engraftment loss.” In total, 10 out of 20 NOHi^ HSC recipients showed some “late-rising” pattern.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
These observations suggest the distinct properties of NOHi^ HSCs: to stay dormant; even
revert to a dormant status after being activated; and then (re-)initiate reconstitution in a later
phase. The abundance of other cycling HSCs augments such “late-rising” features.
NOHi^ HSCs generated both NOHi^ and NOLow^ HSCs (Fig. 1j–k); whereas NOLow^ HSCs only generated multipotent progenitor cells (MPPs) (Fig. 1j). Alterations in lineage skewing were not found (Extended Data Fig. 2f). These results indicate high NO levels identify highly primitive, “late-rising” potent HSCs.
NOHi^ fractions could not be further enriched by the use of additional HSC markers: Flt2−; CD34−; ESAM+; and EPCR+^ (not shown). Enrichment was only feasible using surface staining with anti-C1q antibodies (Extended Data Fig. 4a–b). NOHi cells comprised 80–100% of C1q+CD200R+SSCHiCD150+CD48−Lin−Sca1HicKit+^ HSCs.
C1q+CD200R+SSCHiCD150+CD48−Lin−Sca1HicKit+^ HSCs, sorted without NO dyes,
exhibited robust, time-dependent increasing reconstitution (Extended Data Fig. 4c–d), excluding possible confounding effects of the dye.
Intravenous injection of APC-conjugated CD45 antibodies labeled virtually all peripheral blood cells, while over 99% of BM NOHi^ and NOLow^ HSCs were not stained (Extended Data Fig. 4e). These results indicate NOHi^ HSCs isolated from BMs reside predominantly in BM. Additionally, NOHi^ and NOLow^ HSCs expressed equivalent levels of G-CSF receptors and CXCR4 receptors, implying comparable mobilization capacity (Extended Data Fig. 4f).
NOHi^ HSCs reside at CD200Hi^ capillaries The BM vasculatures were characterized by developing 3D two-photon/confocal microscopy methods using whole-mount long BMs. Such high-resolution imaging identified fine capillary subsets (< 3–4 μm in diameter), which expressed distinctly high levels of CD (Fig. 2a; Extended Data Fig. 5a–d; Supplementary Figure 2a–h; Supplementary Videos 1– 3). CD200Hi^ capillaries are preferentially localized within the metaphyseal endosteum, less frequently at the diaphyseal endosteum. These unique capillaries created dense vascular beds surrounding transitional regions, from arterioles (> 7 μm) into metaphyseal EMCN+^ Type H vessels or diaphyseal sinusoids.
Flow cytometric analysis confirmed CD200Hi^ capillaries, comprising 0.03–0.1% of BM endothelial cells and enriched in the metaphyseal areas (Fig. 2b–d; Extended Data Fig. 6a; Supplementary Figure 3). CD200Hi^ capillaries exhibited distinct cytometric profiles (CD200HiEMCN−Sca1HiCD140bInt–HiCD31HiCD45−) when compared to other BM vessels: Type H vessels; other arterial fractions (CD200Neg^ capillaries, arterioles, and arteries); and sinusoids (Fig. 2b; Extended Data Fig. 6a–b). CD200Hi^ capillaries expressed far higher levels of CD200 than various other hematopoietic and mesenchymal populations, including the following putative niche constituents: Type H vessels^9 ,^10 ; sinusoids^2 –^8 ; arterioles^13 – (^15) ; and CD140b+CD31−CD45− (^) perisinusoidal mesenchymal cells, which exclusively overlapped with leptin receptor- (Lepr-) expressing cells^5 (Fig. 2a, 2c).
The observation of interwoven capillary networks prompted us to investigate whether
CD200Hi^ capillaries express putative shear-stress sensorsviz. primary cilia, which are
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
CD200Hi^ capillaries, while less potent NOLow^ HSCs are noted at Type H vessels or sinusoids.
IFT20/CD200/eNOS/autophagy axis
The roles of CD200 and endothelial cell cilia in HSC regulation are unknown. Conditional
deletions of CD200 and the cilia-related protein IFT20 were achieved using VECadcreERT2^ × CD200flox/flox^ mice and VECadcreERT2^ × IFT20flox/flox^ mice, respectively. Both CD200 and IFT20 targeted deletions in these vessels decreased BM cellularity; the abundance of HSCs and NOHi^ HSCs together with levels of eNOS; but not nNOS (Fig. 3a–b; Extended Data Fig. 9a–c). NOLow^ HSC frequencies were not altered (Extended Data Fig. 9d).
These deletions limited the long-term reconstituting potential of HSCs and BM cells (Fig. 3c) without inducing lineage skewing (Extended Data Fig. 9e). These data indicate critical roles of vascular CD200 and IFT20 in the maintenance of NOHi^ HSC pools.
The interdependency between IFT20 and CD200 was suggested by the double deletions that did not show additional effects over the single deletion (Fig. 3a, 3c). Moreover, IFT deletion in vessels decreased expression levels of CD200 in Sca1Hi^ capillaries (which express CD200 at the steady state) without altering frequencies of the capillaries and other vessels (Fig. 3d–e; Extended Data Fig. 9f).
Direct NOHi^ HSC regulation via CD200R was also indicated by antisense-mediated CD200R knockdown, which decreased eNOS levels in HSCs and abolished NOHi^ HSC reconstitutions without altering homing frequencies (Fig. 3f; Extended Data Fig. 9g–j).
The contributions of other mesenchymal cells were assessed using conditional deletions of CD200 in NG2creERTM-expressing cells^13 ,^14 ; Leprcre-expressing cells^5 and BmxcreERT2- expressing arteries and arterioles^36. None of these deletions altered BM cellularity, abundance of HSCs or NOHi^ HSCs (Extended Data Fig. 9k). These results are consistent with our observation that CD200Hi^ cells are confined to BM capillaries. Such observations support the possibility that CD200-mediated NOHi^ HSC regulation is largely attributed to capillary CD200.
These results suggest IFT20 maintains endothelium expression of CD200, which upregulates eNOS levels in HSCs, thereby modulating NOHi^ HSC abundance.
Abundant expression of angiocrine molecules in CD200Hi^ capillaries (Extended Data Fig. 6h–i) suggested the co-existence of CD200-independent regulation. Of note, IFT20 deletion
in vesselsdecreased vascular integrity and the levels of red marrow at metaphysis without
grossly altering adipocyte numbers there (Fig. 3e; Extended Data Fig. 9l; Supplementary Figure 5a–b). Such observations imply that IFT20 governs the development and organization of metaphyseal vascular niches^38 and may also control metabolic zonation.
However, IFT20 deletion in vessels did not alter the capillary expression of angiocrine molecules or staining levels of the hypoxia probe pimonidazole in capillaries and HSCs (not shown). These results suggest IFT20-mediated NOHi^ HSC regulation may be independent of these angiocrine molecules or hypoxia-signaling. Nonetheless, angiocrine and metabolic
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
factors could be integrated with IFT20-CD200-CD200R signaling to regulate NOHi^ HSC functions finely. The IFT20-mediated maintenance of the vascular integrity might also help to stabilize CD200/CD200R interactions and maintain the homeostasis of NOHi^ HSCs.
IFT20 deletion in vessels decreased the number of ciliated structures in metaphyseal capillaries (Extended Data Fig. 9m; Supplementary Figure 6a–b). The global deletion of another ciliary protein IFT88^39 using tamoxifen treatment in UBCcreERT2^ × IFT88flox/flox mice, decreased BM cells, HSCs, NOHi^ HSCs, and CD200Hi^ vessels (Extended Data Fig. 9n). The results are consistent with the findings in the targeted IFT20 deletion. These observations support the central role of cilia in maintaining homeostasis in these models, but cilia-independent regulations involving IFT20 and IFT88 cannot be ruled out.
Next, the intrinsic HSC regulation was investigated. CD200 and IFT20 deletions in vessels lowered eNOS and did not impact nNOS levels in HSCs (Fig. 3b; Extended Data Fig. 9c). This prompted us to study eNOS. NOHi^ HSCs were sorted from Pdzk1ip1creER^ × eNOSflox/flox^ mice (CD45.2) and transplanted into lethally-irradiated CD45.1 mice with CD45.1 competitor BM cells. Targeted eNOS deletion in transplanted HSCs decreased the reconstitution of NOHi^ HSCs but not of NOLow^ HSCs, while homing frequencies were not altered (Fig. 3g; Extended Data Fig. 9o–q). The eNOS deletion also converted NOHi^ HSC reconstitution trends towards the “exhaustive” phenotype (Extended Data Fig. 9o).
Furthermore, tamoxifen treatment in Pdzk1ip1creER^ × eNOSflox/flox^ × ROSA26TdTomato^ mice
decreased NOHi^ HSC abundance and the long-termendogenous hematopoiesis, as measured
by TdTomato+^ hematopoietic cell frequencies (Fig. 3h–i; Extended Data Fig. 9r). As before, these eNOS deletions did not induce lineage skewing (Extended Data Fig. 9p).
These results indicate that eNOS in HSCs maintains abundance, reconstitution, and late- rising features of NOHi^ HSCs.
Metabolic features of NOHi^ HSCs concerning autophagy were investigated. NOHi^ HSCs exhibited high staining levels of an autophagic dye CYTO-ID and high levels of LC3 and several autophagy regulators, including CTSA (Extended Data Fig. 10a). These results suggest high levels of basal autophagic activities within NOHi^ HSCs. Immunostaining
assays followingex vivo incubation further suggest high levels ofinduced autophagic
activities within NOHi^ HSCs (Extended Data Fig. 10b). In NOHi^ HSCs, NO signals
frequently localized within LC3+ autophagosomes, whereCD200R signals aggregated
following autophagy induction. These results indicate the crucial roles of the CD200R/
eNOS/autophagy axis in HSCs.
The eNOS deletion in HSCs decreased levels of CYTO-ID, LC3, and CTSA in HSCs (Extended Data Fig. 10c). Consistently, CD200 deletion in vessels lowered CYTO-ID levels in HSCs (Extended Data Fig. 10d). These results suggested eNOS signaling maintains autophagic activities in NOHi^ HSCs. Furthermore, conditional deletion of the autophagy gene Atg7 in NOHi^ HSCs abolished NOHi^ HSCs’ reconstitution (Extended Data Fig. 10e) but NOHi^ HSC abundance was not altered (Extended Data Fig. 10f). These results suggest eNOS may control the regenerative function of HSCs, at least in part, via modulation of autophagic activities.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
These results suggested capillary CD200 maintains activated niche Treg pools, boosting the immunoprotection of NOHi^ HSCs. Consistent with the described Treg regulation by CD200^44 , activated niche Tregs expressed high levels of CD200R (Extended Data Fig. 11m), implying direct Treg regulation by capillaries.
Intravenous injection of CD200Hi^ capillary cells (B6) improved engraftment of allo-HSCs (BALB/c) but not of syn-HSCs (Fig. 4f), while lineage skewing was not noted (Extended Data Fig. 11n). Such observations support immunoprotective properties of the capillaries (Extended Data Fig. 12), further highlighting CD200Hi^ capillary cell transfer as a promising strategy to improve engraftment.
Donor CD200Hi^ capillary cells homed at sites of recipient CD200Hi^ endothelial cells (Extended Data Fig. 11o); but these donor cells were not detected at day 30 (not shown). It is feasible that donor capillary cells prevented allo-HSC rejection in the early phases, boosting long-term engraftment.
Finally, we investigated whether decreases in niche Tregs in tamoxifen-treated VECadcreERT2^ × CD200flox/flox^ and VECadcreERT2^ × IFT20flox/flox^ mice impacted NOHi^ HSCs (Fig. 3a). Niche Tregs were depleted using previously-established FoxP3creCXCR4flox/flox^ mice^18 to delete CXCR4 in Tregs, a chemokine receptor required for BM homing of Tregs. This deletion resulted in ~90% decrease in niche Treg pools^18.
Importantly, CXCR4 deletion in Tregs did not alter the abundance of NOHi^ HSC (Extended Data Fig. 11p). However, CXCR4 deletion in Tregs increased BM cellularity^18 ; reconstitution potentials of BM cells^18 together with pool sizes of MPPs^18 and NOLow^ HSCs (Extended Data Fig. 11p). In contrast, CD200 and IFT20 deletions in vessels decreased: BM cellularity; reconstitution potentials of HSCs and BM cells and the numbers of NOHi
HSC (Fig. 3a, 3c). These observations indicateniche Tregs do not directly control the
NOHi^ HSC pools. Hence, decreases in NOHi^ HSCs following the CD200 and IFT20 genetic
deletions are unlikely to be attributed to Treg deficiency. It is feasible that high levels of immunomodulatory molecules within the niche may compensate, at least in part, for Treg functions under basal conditions.
This work has discovered novel regenerative and tolerance mechanisms, which indicate hierarchical structures within HSCs and distinct niches. In essence, we have identified highly immune-privileged, late-rising primitive NOHi^ HSCs, which are further shielded from immune attack by distinctive immune-protected niches containing ciliated CD200Hi capillaries (Extended Data Fig. 12). These innovative mechanisms appear to underpin elements of stem cell homeostasis as well as boosting regenerative capacity and immune tolerance.
NO expression distinguishesdistinctly potent, “late-rising” HSCs from less potent,
exhaustion-prone HSCs. Transplanted NOHi^ and NOLow^ HSCs exhibited equivalent levels of reconstitution initially. However, NOHi^ HSCs exhibited boosts in reconstitution over time, whereas NOLow^ HSCs demonstrated trends to exhaustion. More than three months
post-transplantation, NOHi^ HSCs demonstrated robust reconstitution, which wasfurther
increased by serial transplantation.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Notably, in certain recipients, NOHi^ HSCs demonstrated unique“sleeping beauty-like
late-rising reconstitution.” The higher doses of competitor cells more frequently induced
“sleeping beauty-like reconstitution’ and even elicited “reconstitution re-emergence” long
after the perceived“engraftment loss.”
These observations suggest NOHi^ HSCs possess the distinct inclination to remain dormant, revert to the dormant status after being activated, and then (re-)initiate later robust reconstitution. Such “late-rising” patterns appear augmented by other cycling HSCs. NOHi HSCs generate both NOHi^ HSCs and NOLow^ HSCs, whereas NOLow^ HSCs do not generate NOHi^ HSCs. These results therefore identify novel NOHi^ BM HSCs as highly primitive late-rising HSCs, amongst others.
“Sleeping beauty-like late-rising” multipotent reconstitution has not been previously demonstrated. Prior studies have included single-cell HSC transplantation studies^45 using “young” mice, where “myeloid-restricted repopulating progenitors” from aged mice may exhibit similar late-rising trends. The discordant observations may arise because conventional BM isolation by “flushing” bones may limit NOHi^ HSC isolation from near epiphyseal lines.
Our whole-mount imaging has provided unprecedented BM mapping that links stem cell hierarchy to differential niches. NOHi^ HSCs localize at endosteal, ciliated CD200Hi capillaries, particularly within the metaphysis. CD200Hi^ capillaries exhibit sprouting vessel- like molecular profiles and robust expression of immunoregulatory molecules and HSC
regulators. Such capillaries aredistinct from reported niche constituents: sinusoids; Type H
vessels and arterioles^2 –^10 ,^13 –^15.
Indeed, less potent NOLow^ HSCs preferentially localize at sinusoids or Type H vessels. Such results differ from other descriptions of HSC localization at Type H vessels or sinusoids^2 –^10 , including prior detailed studies using the whole-mount BM imaging of an HSC-reporter mouse^7. The discrepancies may arise from low NOHi^ HSC frequencies and the difficulty in visualizing the fine CD200Hi^ capillaries. We have validated our observations by use of whole-mount imaging using CD150 staining and the HSC-reporter^37. While lymphoid-biased HSCs have been previously demonstrated near arterioles^13 ,^15 , our imaging has distinguished CD200Hi^ capillaries from arterioles based on their differential sizes and CD200 expression. The high-resolution imaging identified these specialized BM capillary niches for the highly primitive, multipotent HSCs.
Our study has characterized intrinsic cytometric/metabolic profiles to define “late-rising” HSCs and has further deciphered underlying regulatory mechanisms. Ciliary protein IFT maintains capillary expression of CD200, which upregulates eNOS levels in HSCs via CD200R. These pathways control abundance as well as reconstitution and further modulate “late-rising” features of NOHi^ HSCs. The roles of NOHi^ HSCs in endogenous hematopoiesis are further indicated by endogenous hematopoiesis tracking following the targeted eNOS deletion in HSCs. eNOS regulates regenerative functions of NOHi^ HSCs, at least in part, via modulation of autophagic activities.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Medical Science). Cdh5(PAC)-CreERT2 (VECadcreERT2) mice^9 and C57BL/6J-Tg(Bmx- CreERT2)1Rha (BmxcreERT2) mice were kindly provided by Dr. Ralf Adams (^49). UBCcreERT2 (^) × IFT88flox/flox^ mice were kindly provided by Dr. Bradley Yoder^39. Mice were studied at 6– weeks of age. Tamoxifen treatment in VECadcreERT2^ × CD200flox/flox^ mice, VECadcreERT × IFT20flox/flox^ mice, VECadcreERT2^ × CD200flox/flox^ × IFT20flox/flox^ mice, and UBCcreERT × IFT88flox/flox^ mice was initiated when mice were at 5–6 weeks old and studied at 11– 12 weeks old. Pdzk1ip1creER^ × eNOSflox/flox^ mice and Pdzk1ip1creER^ × Atg7flox/flox^ mice received tamoxifen treatment at 7–10 weeks old of age. For the analyses of NOHi^ HSC
distribution (imaging, flow cytometric analysis/transplantation of metaphysisvs. diaphysis
HSC), 8–10 weeks-old mice were studied. Serial transplantation studies were conducted when mice were 8–10 weeks old of age.
All animal studies were conducted with approval from the Institutional Review Boards and Animal Care and Use Committees at Columbia University and Beth Israel Deaconess Medical Center.
Flow cytometry of NO levels in HSCs.— BM cells were isolated by crushing long bones (tibia and femur). Following treatment with RBC lysis buffer (BioLegend), the cell suspension (2 × 10^6 cells) was plated onto 96-well plates and incubated with PBS containing three μM DAF-FM (Invitrogen) for 20 min at 37°C. Subsequently, BM cells were washed by PBS, resuspended into the staining buffer (PBS-2% FBS, 2 mM EDTA), and incubated with the following surface antibodies: CD200R-Alexa Flour 647 (OX110; BD Biosciences), CD48-Alexa Flour 700 (BioLegend), anti-mouse Lineage Cocktail-Pacific Blue (BioLegend), CD117 (c-Kit)-APC-eFluor 780 (Invitrogen), Ly-6A/E (Sca1)-Brilliant Violet 605 (BioLegend), Zombie Aqua (BioLegend), and CD150-PE/Cy7 (BioLegend). For the C1q staining, cells were incubated with C1qA-biotin (Novus Biologicals; 1x dilution) or biotin mouse IgG2b isotype for 30min, followed by the staining with Brilliant Violet 711- streptavidin (Biolegend). Flow cytometry was performed using an LSRII (BD Biosciences), LSRFortessa (BD Biosciences), FACSymphony A1 (BD Biosciences), or CytoFLEX LX (Beckman Coulter), followed by analysis using FlowJo software (Tree Star Inc.). To stain HSCs with MitoSOX (Invitrogen; 5 μM) or CYTO-ID (Enzo Life Sciences; × 500 dilution), each dye was added to the DAF-FM-containing culture media of freshly harvested BM cells and incubated for 20 min before staining with surface antibodies.
For the intracellular staining, after cell surface staining and DAF-FM staining, BM cells were incubated overnight in fixation buffer FM Fix (Fujifilm) at 4°C. After washing, cells were incubated with permeabilization buffer (eBioscience; 2x) for 20 min at room temperature; washed once; and incubated again with permeabilization buffer (eBioscience; 2x) for 20 min. After washing, samples were stained for 60 min with the following antibodies: eNOS-Alexa Fluor 647 (Abcam) or rabbit IgG-Alexa Fluor 647 isotype (Abcam); nNOS-Alexa Fluor 647 (Abcam) or rabbit IgG-Alexa Fluor 647 isotype (Abcam); G-CSFR-PE (Novus Biologicals) or Rat IgG2a-PE isotype (Biolegend); CXCR4-PE (Biolegend) or rat IgG2b κ-PE isotype (Biolegend); Protective protein/Cathepsin A (CTSA)- Alexa Fluor 647 (Abcam) or rabbit IgG-Alexa Fluor 647 isotype (Abcam); LAMTOR1- Alexa Fluor 647 (Novus Biologicals) or rabbit IgG-Alexa Fluor 647 isotype (Novus Biologicals); or ROBLD3 (LAMTOR2)-Alexa Fluor 647 (Novus Biologicals) or rabbit IgG-
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Alexa Fluor 647 isotype (Novus Biologicals). For LC3B staining, after the above fixation step, BM cells were stained with rabbit anti-LC3B antibody (Invitrogen) or rabbit polyclonal IgG isotype antibody (Invitrogen) for 60 min. After washing, cells were incubated with goat anti-rabbit IgG-Alexa Fluor 647 (Invitrogen) for 30 min.
Flow cytometry of BM mesenchymal cells.— BM cells were isolated by crushing the tibia and femurs, followed by filtration through 100 μm cell strainers. The bone fragments remaining on the strainers were isolated, incubated with HBSS containing Collagenase IV (Worthington; 2mg/ml) and DNase I (Worthington; 2500 Kunitz/ml) for 30 min at 37°C, and resuspended into the staining buffer. Supernatants were filtered through 100 μm cell strainers and combined with BM cells. After washing, cells were treated with RBC lysis buffer and washed again with PBS. Subsequently, cells were resuspended into the staining buffer, and incubated with the following surface antibodies: CD200-PE (Invitrogen), EMCN-Alexa Fluor 488 (Santa Cruz Biotechnology), Sca1-Brilliant Violet 605 (BioLegend), CD140b- biotin (BioLegend), DAPI (BioLegend), CD45-Pacific Blue (BioLegend), TER119-Pacific Blue (BioLegend), and CD31-PE/Cy7 (BioLegend). Samples were washed and stained with APC/Cy7-streptavidin (BioLegend).
For the cell surface staining with anti-Tie1 antibodies, non-fixed BM cells were stained with rabbit anti-Tie1 antibody (Proteintech) or rabbit polyclonal IgG isotype antibody (Invitrogen) for 30 min. Samples were washed and stained with goat anti-rabbit IgG- Alexa Fluor 647 (Invitrogen) for 30 min. For the cell surface staining with anti-VEGFR antibodies, non-fixed BM cells were stained with VEGFR2-APC (BioLegend) or rat IgG2a κ-APC isotype (BioLegend) for 30 min.
For the intracellular staining of BM mesenchymal cells, after the cell surface staining, BM cells were incubated overnight in fixation buffer FM Fix (Fujifilm) at 4°C. After washing, cells were incubated with permeabilization buffer (eBioscience; 8x) for 20 min at room temperature; washed once; and incubated again with permeabilization buffer (eBioscience; 8x) for 20 min. After washing, samples were stained with the following antibodies: CD106- APC (BioLegend) or rat IgG2a κ-APC isotype (BioLegend); CD54-APC (BioLegend) or rat IgG2b κ-APC isotype (BioLegend); LYVE1-eFluor 660 (Invitrogen) or rat IgG1 κ-eFluor 660 isotype (Invitrogen); SCF-Alexa Fluor 647 (Bioss) or rabbi IgG-Alexa Fluor 647 isotype (Bioss); Tie2-APC (BioLegend) or rat IgG1 κ-APC isotype (BioLegend); VEGFR1- APC (R&D Systems) or rat IgG2b κ-APC isotype (BioLegend); and VEGFR3-APC (R&D Systems) or goat IgG-APC isotype (R&D Systems). Cells were stained with all these antibodies for 60 min, except for Tie2-APC, VEGFR1-APC, and VEGFR3-APC; for which 90 min staining was needed.
For the intracellular staining with anti-BMP2, anti-BMP4, or anti-DLL4 antibodies, after the above fixation step, BM cells were stained for 90 min with the following antibodies: rabbit anti-BMP2 antibody (Abcam) or rabbit monoclonal IgG isotype (Abcam); rabbit anti-BMP antibody (Abcam) or rabbit polyclonal IgG isotype (Invitrogen); and rabbit anti-DLL antibody (Novus Biologicals) or rabbit polyclonal IgG isotype (Invitrogen). After washing, samples were stained with goat anti-rabbit IgG-Alexa Fluor 647 (Invitrogen) for 30 min; and washed twice with the permeabilization buffer.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Peripheral blood samples from recipients were periodically analyzed, as described in the competitive HSC transplantation assays.
Tamoxifen treatment for studies using Pdzk1ip1creER^ × eNOSflox/flox^ or Pdzk1ip1creER^ × Atg7flox/flox^ mice.— To achieve conditional deletions in donor HSCs isolated from Pdzk1ip1creER^ × eNOSflox/flox^ or Pdzk1ip1creER^ × Atg7flox/flox^ mice,
tamoxifen treatment was performedex vivo in donor HSCs before transplantation, and
subsequently in recipients following transplantation. HSCs were sorted from Pdzk1ip1creER × eNOSflox/flox^ or Pdzk1ip1creER^ × Atg7flox/flox^ mice and were incubated with 4- hydroxytamoxifen ((Sigma-Aldrich; 1 μM) + HSC media (10 ng/mL IL-6, 10 ng/mL IL-3, 25ng/mL IL-7, 50 ng/mL SCF, and 50 ng/mL Flt-3l in StemSpan SFEM [Stemcell Technologies]) for six hours. Subsequently, NOHi^ HSCs were intravenously injected ( cells/recipient) into lethally irradiated SJL recipients (950cGly in total) with competitor SJL BM cells (200,000 cells). The recipients received tamoxifen treatment (intraperitoneal; 1 mg; days 0–3). For the analysis of endogenous hematopoiesis, Pdzk1ip1creER^ × eNOSflox/ flox (^) mice received tamoxifen treatment (1 mg; oral gavage).
HSC transplantation into non-conditioned recipients.— BM cells were isolated from B6 mice by crushing the femur, tibia, and iliac bones. HSCs, NOHi^ HSCs or NOLow HSCs were sorted using Influx and FACSAria II (BD Biosciences), as in the competitive HSC transplantation assay. Sorted HSCs were stained with 10 μM DiD (Invitrogen) in PBS with 0.1% fetal bovine serum for 20 min at 37°C, and washed once in PBS with 0.1% fetal bovine serum. DiD-labeled B6 HSCs (3000 cells/recipient; Fig. 1a), NOHi^ HSCs or NOLow^ HSCs (400/recipient; Extended Data Fig. 11a) were injected into the tail vein of non-irradiated B6, BALB/c or B10.A mice. Confocal images of the skull and long bones were obtained. In Figure 4d and Extended Data Fig. 11a–b, the numbers of donor HSCs in the 3D images at the identical regions of the skull BMs (1500 μm [x] × 1800 μm [y] × 120 μm [z]) were counted as described^20 ,^21.
Allo-NOHi^ HSC transplantation following nonmyeloablative conditioning.— Age-matched B10.A (H2Dd+) and B6 (H2Kb+; CD45.2) recipient mice received 3-Gy irradiation, followed by intravenous injection of NOHi^ and NOLow^ B6 SJL HSCs (500/each; CD45.1; H2Kb+), together with B6 competitor BM cells (2 × 10^7 /each; CD45.2; H2Kb+). Anti-mouse CD8 mAbs (2.43; 1.44 mg/each) was injected intraperitoneally on Day −1^50. On Day 0, anti-mouse CD40L mAbs (MR1; 2mg/each) was injected intraperitoneally (i.p.) following TBI (total body irradiation). Anti-CD8 mAbs and anti-mouse CD40L mAbs were purchased from BioXcell (West Lebanon, NH, USA). The H2Kb+H2Dd−CD45.2−CD45.1+ donor blood frequencies were analyzed.
Nonmyeloablative allo-HSC transplantation with CD200Hi^ capillary cell transfer.— Age-matched B6 recipient mice (CD45.2; H2Kb+) were treated with a nonmyeloablative dose of TBI (3 Gy) followed by an intravenous injection of MHC- mismatched BALB/c (allo; H2Dd+) or B6 SJL (syn; H2Kb+; CD45.1+) BM cells (2 × 107 /each)^50. B6 CD200Hi^ capillary cells (15,000/each) were sorted from B6 BM cells and intravenously injected into irradiated recipients on Day −2. Anti-mouse CD8 mAbs (2.43;
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
1.44 mg/each) was injected intraperitoneally on Day −1. On Day 0, anti-mouse CD40L mAbs (MR1; 2mg/each) was injected i.p. following TBI. The frequencies of H2Kb−H2Dd+ or H2Kb+H2Dd−CD45.2−CD45.1+^ donor blood cells were analyzed following allo- or syn- BM transplantation, respectively.
Confocal microscope imaging of whole-mount BMs.— Samples for whole-mount BM imaging were prepared as described in the Supplementary Method. Imaging was performed using a Leica SP8 MP microscope (Leica Microsystems), as in our previous studies. The maximum imaging depth of the whole-mount BM reached ~80 μm from the surface. Montages of 3D image stacks (~2500 μm [x] × ~5000 μm [y] × 300 μm [z]; 1–5 μm of each z step) were obtained to cover all the long BMs. GFP/Alexa 488 signal was excited with a 488-nm laser and detected with 500–540 nm. The Alexa 568 signal was excited with a 561-nm laser and detected at 580–620 nm. Alexa 647 was excited with a 633-nm laser and detected at 680–730 nm. The TdTomato signal was excited with a 561-nm laser and detected at 580–640 nm. mCherry was excited with a 633-nm laser and detected at 645– nm. Q655 was excited with a 488 nm laser and detected at 625–680 nm. BV421 was excited with a two-photon excitation at 780–790 nm and detected at 410–430 nm.
As in our previous study^18 , images were processed and analyzed with Imaris (Bitplane, version 7.7.1) to identify NOHi^ HSCs, NOLow^ HSCs, CD200Hi^ capillaries, Type H vessels, and sinusoids, and to calculate the distance between cell centers in an automated fashion. NOHi^ HSCs were defined as HSCs that exhibited DAF-FM signals at levels in the top 15% of whole HSCs, whereas NOLow^ HSCs were in the bottom 85%. In assays using CD staining, HSCs were defined as HSCs that were stained with CD150-Alexa 568 but not with Lin/CD48-BV421. For HSC imaging using Pdzk1ip1creER^ × ROSA26TdTomato mice, long bones were harvested from Pdzk1ip1creER^ × ROS26TdTomato mice three days after tamoxifen treatment (oral gavage, 1 mg), when TdTomato expression was exclusively found within HSCs^42. HSCs were defined as HSCs that expressed TdTomato but were not stained with Lin/CD48-BV421. BM vasculatures were analyzed using the “Filament Tracer Module of Imaris.” CD200Hi^ capillaries were defined as fine vessels (CD200HiIB4+; < 3–4 μm in diameter) that expressed CD200 at levels in the top 0.01–0.05% of all BM cells. Type H vessels were defined as EMCN+IB4+^ vessels within 500 μm of the endosteal surface at the epiphyseal line. Sinusoids were defined as EMCN−IB4+^ vessels. “Metaphysis < 500 μm” was defined as the region within ~500 μm from the endosteal surface at the epiphyseal line. “Metaphysis 500–1500 μm” was defined as the region within 500–1500 μm from endosteal surface at the epiphyseal line. “Whole metaphysis” was defined as the region within ~ μm from the endosteal surface at the epiphyseal line. “Diaphysis” was defined as the region beyond 1500 μm of the endosteal surface at the epiphyseal line.
Statistics and Reproducibility.— Statistical parameters, including several biological replicates and repeat experiments, data dispersion, and precision measures (mean and SD), are reported in the figure legends. Statistical analysis was performed using GraphPad Prism
- Statistical significance was determined using two-tailed t-test for parametric two-group analysis, One-way ANOVA with post hoc test with Bonferroni correction for parametric multi-group comparison, Two-way ANOVA with post hoc test with Bonferroni correction
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Extended Data Figure 2: NO expression indicates robust but late-rising HSCs a. Serial transplantation assays. NOHi, NOLowCD200R+, or NOLowCD200R−^ HSCs (10 cells/recipient; B6 SJL; CD45.1) were injected into tail veins of lethally irradiated B6 congenic recipient mice (CD45.2), together with competitor B6 BM cells (200,000 cells/recipient; CD45.2). Serial transplantation was performed twice at five-month intervals by transplanting BM cells (5 × 106 /each) of the recipient mice into irradiated B6 recipient mice (950cGly). Refer to Fig. 1g–k. b. CD45.1+^ donor blood chimerism in B6 recipients transplanted with NOHi, NOLowCD200R+, or NOLowCD200R−^ HSCs (B6 SJL). Donor chimerism was analyzed monthly. The results of the first and secondary transplantation are pooled from three independent experiments (three donors in total; 17– recipients per group). The third transplantation experiments includes 6–8 recipients/group. Data were analyzed by 1-way ANOVA with post hoc test with Bonferroni correction. 1st: first transplantation. 2nd: secondary transplantation. 3rd: third transplantation. c. Differential patterns of blood reconstitution derived from NOHi, NOLowCD200R+, or NOLowCD200R−^ HSCs. Reconstitution patterns were categorized as in Extended Data Fig. 2d.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
d. Differential reconstitution patterns. e. Trend of CD45.1+^ donor chimerism over the 1st, 2nd, and 3rd^ transplantation. Experiments include 6–8 recipients/group. The colors of the trend lines correspond to those of the differential reconstitution patterns shown in Extended Data Fig. 2d. f. Lineage skewing analyses. These were done using CD45.1+^ donor blood cells in B6 recipient mice at 16 weeks after the transplantation of NOHi^ or NOLowCD200R+^ HSCs (B6 SJL). N = 5‒10. The results were pooled from three independent experiments. Frequencies of myeloid cells, T cells, and B cells in CD45.1+^ donor blood cells were analyzed by two-sided Student’s t-test. All data are presented as mean ± SD. Asterisks indicate significant differences (p-value <0.05, *; <0.01, **; <0.001, ***; <0.0001, ****; N.S., >0.05).
Extended Data Figure 3: Use of the higher dose of competitor cells frequently induced late-rising reconstitution from NOHi^ HSCs a-e. Competitive transplantation assays. NOHi^ or NOLow^ HSCs (10 cells/recipient; B6 SJL; CD45.1) were injected into tail veins of lethally irradiated B6 congenic recipient mice (CD45.2), together with competitor B BM cells (300,000 cells/recipient; CD45.2). The secondary (2nd) transplantation of the recipient BM cells (5,000,000 cells/recipient) to lethally irradiated B6 mice was performed four months after the first (1st) transplantation. Data are pooled from four independent experiments (four donors in total; 20 recipients per group). a. Blood reconstitution patterns derived from NOHi^ or NOLow^ HSCs.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
