Humanizedin vivobone marrow models orchestrate multi-lineage human hematopoietic cell development

(161/160) 38 Hematopoiesis develops in the bone marrow (BM) where multiple interactions regulate 39 differentiation and preservation of hematopoietic stem/progenitor cells (HSPCs). Although 40 murine BM has been extensively analyzed, the human BM microenvironment remains less 41 understood. Immune-deficient murine models have enabled the analysis of molecular and 42 cellular regulation of human HSPCs, which remains limited as human hematopoietic cells 43 develop in xenogenic microenvironments. In this study, we thoroughly characterized a 44 humanized (h) in vivo BM model, based on mesenchymal stromal cell (MSC) differentiation 45 (called hOssicles (hOss)), and hematopoietic cell compartments generated 3 months post-46 transplant of CD34 + cells using single-cell RNA sequenci ng and cellular barcoding. Serial 47 isolation of MSCs and HSPCs from hOss and transplant experiments revealed the dynamic 48 nature of these hBM niches. hOss altered hum an hematopoietic development by modulating 49 myeloid/lymphoid cell production and HSPC levels. Clonal tracking highlighted hematopoietic 50 cell cross-talks between the murine BM and hOss, indicating the multipotent or more 51 restricted lineage origin of human hematopoiesis shared in the BM sites. 52 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 3

(5224 words) 53 Post-natal hematopoiesis takes place in the bone marrow (BM), where numerous cell 54 interactions promote the production of mature blood cells. Analysis of the BM 55 microenvironment in murine models has enabled t he detailed characterization of spatially 56 distinct niches of the various types of hematopoietic stem and progenitor cells (HSPCs), 57 including dormant/quiescent and activated hematopoietic stem cells (HSCs) (Ding and 58 Morrison, 2013; Cordeiro Gomes et al., 2016; Pi nho et al., 2018). Furthermore, single-cell 59 RNA sequence analysis of BM niches has re cently revealed the heterogeneous nature of 60 non-hematopoietic cells that support hematopoietic cell production (Tikhonova et al., 2019; 61 Baryawno et al., 2019; Baccin et al., 2020). 62 Studying the human BM microenvironment is more complicated than murine BM as it 63 involves analysis of BM biopsy sections, which enables only general characterization but not 64 genetic modification of the different cell components, thus greatly limiting the investigation of 65 cell function. Humanized (h)BM models, which are generated ex vivo and in vivo by 66 transplanting human BM cell components into immune-deficient mice, serve as valuable 67 tools in biomedical research (Abarrategi et al., 2018; Dupard et al., 2020). These models are 68 designed to replicate the complex cellular interactions that occur in the human BM 69 microenvironment. By recapitulating the physiological conditions, these models provide 70 researchers with a powerful platform to st udy hematopoiesis, immune cell development, and 71 disease progression. hBM models incorporate co mponents critical for hematopoiesis, such 72 as stromal cells, and extracellular matrix components (Chen et al., 2012; Antonelli et al., 73 2016; Groen et al., 2012; Abarrategi et al., 2017; Fritsch et al., 2018; Reinisch et al., 2016). 74 Inclusion of human endothelial cells that contribute to blood vessel development and are key 75 supportive cells of hematopoiesis in the BM (Ramasamy et al., 2016; Itkin et al., 2016) has 76 been successfully described in vivo (Chen et al., 2012; Passaro et al., 2017; Abarrategi et al., 77 2017). Although hBM models are chimeric with murine and human cell components, these 78 systems enable the study of normal and pathological conditions, including hematological 79 disorders, leukemia and immunodeficiency. Furthermore, these models which are based on 80 mesenchymal stromal/stem cells (MSCs) and hematopoietic cell differentiation, offer the 81 opportunity to study the effects of various factors, such as infection, therapeutic interventions 82 and response, and genetic mutations that can be incorporated into HSCs or progenitors 83 using CRISPR-Cas9 editing. By providing a controlled and reproducible system, these 84 models enable researchers to gain extensive insight into the mechanisms underlying 85 diseases and test potential therapies with greater precision. Moreover, the application of 86 hBM models has proven valuable in cancer research, enabling the analysis of tumor 87 progression, including infiltration and metastasis , in the context of a physiologically relevant 88 microenvironment (Martine et al., 2017; Thibaudeau et al., 2014; Grigoryan et al., 2022). By 89 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 4 providing a platform to evaluate the efficacy and toxicity of anti-cancer drugs, hBM models 90 also facilitate the development of personalized treatment strategies. 91 It has been hypothesized that species-specific HSC microenvironment interactions are likely 92 vital for human hematopoietic development (Holzapfel et al., 2015). However, the impact of 93 humanized niches on human hematopoietic cell recovery generated following CD34 + HSPC 94 transplant has not been assessed (Abarrategi et al., 2018). In this study, we aimed to 95 thoroughly characterize a robust and reproducib le human ossicles (hOss) hBM model 96 adapted from previous studies (Reinisch et al., 2017a). We generated hOss by using human 97 fetal (F) and post-natal (P-N) BM-derived MS Cs transplanted subcutaneously into immune-98 deficient mice. We found that F/hOss and P-N/hOss reproducibly supported human 99 hematopoietic cell development following peripheral injection of human CD34 + umbilical cord 1 00 blood cells. Using flow cytometry and single-cell RNA sequencing, characterization of mature 1 01 and immature human hematopoietic cells developing in hOss revealed a myeloid cell bias at 1 02 the expense of B lymphoid cells, contrary to that observed in the BM of gold standard NOD 1 03 scid gamma (NSG) mouse models (Pflumio et al., 1996; Henry et al., 2020), thus more 1 04 accurately recapitulating human BM cells. Importantly, recovered hOss contained hMSCs 1 05 capable of secondary hOss formation and human hematopoietic support, indicating that 1 06 hOss are dynamic BM structures that maintain functional MSCs. Furthermore, hOss better 1 07 supported immature human cells capable of secondary hematopoietic reconstitution after 1 08 transplant. Finally, using a cellular barcoding tracking strategy, we found that many human 1 09 hematopoietic cell clones generated in mice harboring hOss were myeloid biased. Most cell 1 10 clones were multipotent and produced human hematopoietic cells in both hOss and murine 1 11 BM (mBM), thus indicating cross-talk between BM sites. The dynamic hOss structure created 1 12 a favorable environment for human myelopoiesis and multipotent HSPCs. 1 13 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 5

1 14 Development of hOss originating from human bone marrow-derived fetal and post-1 15 natal hMSCs. hOss formation was carried out according to a previously developed protocol 1 16 (Reinisch et al., 2017a). Briefly, we first isolated primary hMSCs from F/BM and P-N/BM after 1 17 cell adhesion on plastic plates. hMSCs were amplified until confluent and split; a portion of 1 18 the cells were stored frozen at passage 0, while the remaining cells were further amplified for 1 19 <4 passages for downstream analyses in accordance with the protocol by Passaro et al. 1 20 (Passaro et al., 2017). Kinetic analysis of cell growth indicated enhanced expansion of the 1 21 majority F/hMSCs (4 of 5) compared with P-N/hMSCs ( Figure 1A). Flow cytometry analysis 1 22 of conventional hMSC markers, such as CD90, CD73, CD105 and CD44, and pan-1 23 hematopoietic CD45 and myeloid CD14/CD15 markers and endothelial CD31 confirmed the 1 24 non-hematopoietic mesenchymal origin of the isolated adherent cells of F/BM and P-N/BM 1 25 origin (Figure 1B and SupFig1A). No major phenotypic differences were observed between 1 26 F/hMSCs and P-N/hMSCs. These hMSCs were capable of ex-vivo multilineage 1 27 differentiation into adipogenic, osteogenic and chondrogenic cells upon exposure to the 1 28 appropriate cell culture media ( SupFig1B). To test the ability to generate hOss, a mixture of 1 29 hMSCs, cold-matrigel, human platelet lysate and hBMP-7 was subcutaneously injected into 1 30 immune-deficient NSG mice that were treated 5 days per week for 4 weeks with parathyroid 1 31 hormone (to enhance hOss formation and growth) ( Figure 1C ) (Reinisch et al., 2017b; 1 32 Martine et al., 2017). OsteoSense labeling and tomography analysis, which typically 1 33 indicates regions with high bone turnover, showed bone remodeling in hOss derived from 1 34 F/hMSCs and P-N/hMSCs ( SupFig1C) (Lambers et al., 2013; Martine et al., 2017). At 8 1 35 weeks post-hMSC implantation, hOss were reco vered from grafted mice and analyzed. The 1 36 F/hMSC-derived hOss were larger and 5 times heavier (median: 579 mg) than P-N/hMSC-1 37 derived hOss (median: 94 mg) ( Figure 1D-E ). Murine hematopoietic Ter119 + and CD45 + 1 38 cells were detected in F/hOss and P-N/hOss, albeit at slightly lower levels compared with 1 39 mBM, thus showing that murine blood cells perfused the implanted hBM (Figure 1F). 1 40 Histological analysis of hOss sections indica ted the presence of active bone structures, 1 41 including adipocytes and hematopoietic lodges, resembling BM ( Figure 1G ). Immuno-1 42 labeling with anti-mouse endomucin or anti-mouse Meca32 antibodies also revealed vascular 1 43 networks ( SupFig1D-E), thereby confirming that murine blood vessels vascularized hOss. 1 44 Overall, these results show efficient in vivo development of vascularized functional BM 1 45 structures derived from human F/hMSCs and P-N/hMSCs. 1 46 1 47 Human hematopoiesis in hOss after transplantation of human HSPCs . We next aimed 1 48 to analyze the supportive ability of F/hOss and P-N/hOss for human hematopoiesis following 1 49 intravenous transplantation of human CD34 + cells (105 cells/mouse) in sub-lethally irradiated 1 50 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 6 mice harboring hOss ( Figure 2A ). Human hematopoietic development was analyzed 12 1 51 weeks later, which is representative of long-t erm hematopoietic reconstitution in transplant 1 52 models (Dykstra et al., 2007). We found that the size and weight differences persisted 1 53 between F/hOss (median: 390 mg, at 12 weeks) and P-N/hOss (median: 78 mg, at 12 1 54 weeks) (SupFig2A) as observed in non-transplanted hOss ( Figure 1E ). Immuno-1 55 histochemistry labeling indicated high levels of hCD45 + cells in hOss, comprising 1 56 MPO+/CD14+ granulo-monocytic cells, CD61+ megakaryocytes, glycophorin C+ erythroid cells 1 57 and immature CD34 + HSPCs in F/hOss ( Figure 2B ). Qualitative and quantitative flow 1 58 cytometry analysis further confirmed human multi-lineage hematopoietic formation in both 1 59 F/hOss and P-N/hOss after 12 weeks HSPC post-transplant ( Figure 2C-I ). Erythroid cells 1 60 were enhanced in F/hOss (10%) and to a lesser extent (albeit not significant) in P-N/hOss 1 61 (6%) compared with mBM (2%) ( Figure 2D), whereas hCD45+ cell percentages were similar 1 62 in mBM from mice without and with hOss, regardless of the hMSCs origin (F/ or P-N/) 1 63 (Figure 2E) . In terms of absolute numbers, a median of 2.8x10 6 and 3x10 6 hCD45 + 1 64 cells/hOss were recovered from F/hOss and P-N/hOss at 12 weeks, respectively 1 65 (SupFig2B). The most striking difference was the decreased percentage in CD19 + B-cells in 1 66 F/hOss and P-N/hOss (46% and 53% of hCD45 + cells, respectively) compared with mBM 1 67 from mice without hOss, in which B-cells predominated (80%) ( Figure 2F ) as typically 1 68 observed in these humanized mouse models (Henry et al., 2020). Inversely, myeloid cell 1 69 levels (CD14 +/CD15+) were higher in hOss (34% and 27%) compared with mBM (10%) 1 70 (Figure 2G ). Only a minor difference was noted in B/myeloid/erythroid cell levels between 1 71 F/hOss or P-N/hOss. Comparing B/myeloid/er ythroid cell levels between mBM from mice 1 72 with and without hOss showed a similar, albeit less pronounced, decrease in B-cell levels 1 73 and an increase in myeloid cell levels in mBM from mice with hOss ( SupFig2E-H), thus 1 74 highlighting possible cross-talk between hOss and mBM sites. Analysis of immature human 1 75 cell compartments indicated similar levels of CD34 + HSPCs in F/hOss, P-N/hOss and mBM 1 76 without hOss at 12 weeks post-injection of CD34 + cells ( Figure 2H ). Assessing more 1 77 immature hCD45+Lin-CD34+CD90+ HSCs indicated slightly enhanced levels of F/hOss (2.7%, 1 78 p<0.05) compared with P-N/hOss (1.8%) and mBM (1.4%) ( Figure 2I). Compared with mBM 1 79 without hOss, CD34+ cells were significantly increased in mBM with F/hOss and mBM with P-1 80 N/hOss, while no differences were observed in hCD45 +Lin-CD34+CD90+ HSCs (SupFig2I-J). 1 81 These results show that hOss enable human hematopoiesis development, and differences 1 82 were detected primarily in phenotypically mature hematopoietic cells compared with mBM. 1 83 Interestingly, levels of B-cells, myeloid cells, and immature cells in hOss fell between the 1 84 levels observed in mBM and hBM from young healthy donors (median age: 11 years) 1 85 (Figures 2F-I), thereby suggesting that hOss better mimics normal human hematopoiesis. 1 86 1 87 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 7 Fetal and post-natal hOss contain functional human mesenchymal stem cells . To 1 88 characterize the possible replenishment of hOss by engrafted resident hMSCs, we tested 1 89 whether isolation of hMSCs from primary hOss was possible and whether these recovered 1 90 hMSCs could generate secondary hOss. After crushing primary F/hOss or P-N/hOss, 1 91 adherent cells were recovered and amplified in plastic plates ( Figure 3A). Flow cytometry 1 92 analysis of adherent cells (using the same markers as displayed in Figure 1B ) confirmed 1 93 that the cells originated from human mesenchymal cells ( Figure 3B). The secondary isolated 1 94 hMSCs were injected into mouse flanks to form hOss, and human CD34 + cells (10 5 1 95 cells/mouse) were transplanted 8 weeks later ( Figure 3A). Upon sacrifice of the mice at 12 1 96 weeks post-transplant, secondary hOss had formed; interestingly, secondary F/hOss were 1 97 twice as large (mg) as secondary P-N/hOss ( Figures 3C-D). This difference was slightly less 1 98 pronounced than that observed in primary hOss ( Figure 1E and SupFig2B ). Murine 1 99 erythroid and hematopoietic cells were present in secondary hOss that did not receive CD34+ 2 00 umbilical cord blood cells, although lower levels were observed compared with mBM 2 01 especially in secondary P-N/hOss, within the limits of testing only a few hOss ( Figure 3E). 2 02 Human erythroid cells were enhanced in secondary hOss, as observed in primary hOss. 2 03 However, P-N/hOss were less permissive to this human lineage development than F/hOss 2 04 (Figure 3F ). The percentage of hematopoietic cells did not vary between mBM and 2 05 secondary F/hOss, but were significantly higher in secondary P-N/hOss compared with mBM 2 06 (Figure 3G). In terms of absolute number of hCD45 + cells A significant decrease was 2 07 observed in secondary P-N/hOss compared with F/hOss ( SupFig3A). Compared with mBM 2 08 from mice without hOss, secondary F/hOss displayed a significant decrease in CD19 + B-cells 2 09 and a significant increase in CD14 +/CD15+ myeloid/GPA+ erythroid cells, while a difference 2 10 was not observed in secondary P-N/hOss ( Figure 3H-I). Analysis of immature cells revealed 2 11 similar CD34+ HSPC levels, and increased CD34 +CD90+ HSC levels were observed only in 2 12 secondary P-N/hOss compared with mBM ( Figure 3J-K). As analyzed in mice with primary 2 13 hOss, mBM from mice with secondary hOss displayed similar human hematopoietic 2 14 reconstitution to that observed in secondary hOss, again indicating cross-talk between the 2 15 mBM and hOss sites ( Figure 3F-K ). Analysis of hematopoietic cells from primary and 2 16 secondary hOss further confirmed subtle but significant differences between F/hOss and P-2 17 N/hOss. Indeed, F/hOss appeared to effectively support balanced production of human 2 18 lymphoid and myelo-erythroid cells, whereas secondary P-N/hOss displayed reduced 2 19 development of human myelo-erythroid cells ( SupFig3B-E). Furthermore, no major 2 20 difference in the immature cell compartment s was observed between primary and secondary 2 21 hOss, although subtle but significan t differences were observed in CD34 +CD90+ cells 2 22 (SupFig3F-G). Overall, these results show that hOss are dynamic human BM-like structures 2 23 that contain hMSCs capable of serial production of hOss that continue to support human 2 24 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 8 hematopoietic development. Interestingly, the lympho-myeloid bias observed in primary P-2 25 N/hOss was no longer observed in secondary P- N/hOss, thus suggesting that exhaustion 2 26 may occur during the transplant process. 2 27 2 28 Immature human hematopoietic cells from hOss display enhanced hematopoietic 2 29 stem cell functional potential . We next characterized the immature cell function of hOss-2 30 engrafted CD34 + HSPCs using standard assays, such as myelo/erythroid colony forming 2 31 units (CFUs) to measure committed progenitor levels and secondary reconstitution of 2 32 immune-deficient mice to detect HSCs (Robin et al., 1999). Cells were obtained from primary 2 33 hOss and the BM of mice with and without hOss as controls. Human CD34 + HSPCs were 2 34 quantified and either plated in semi-solid methylcellulose medium or transplanted into NSG 2 35 mice ( Figure 4A ). According to the CFU results, we observed similar total numbers of 2 36 colonies generated by cells isolated from hOss and mBM ( Figure 4B ). Assessing colony 2 37 types (i.e., progenitors) between the three condi tions, we did not observe major differences 2 38 in progenitor composition ( Figure 4C and SupFig4 ). Analysis of the human hematopoietic 2 39 cell levels in secondary NSG mice injected with cells from mBM of mice with and without 2 40 hOss as well as primary and secondary hOss s howed increased reconstitution potential of 2 41 human HSPCs recovered from mice with hOss compared with mice without hOss, regardless 2 42 of whether the hOss were primary or secondary ( Figure 4D-I). Mice injected with cells from 2 43 mBM with hOss also displayed hCD45+ cells (Figure 4E and H), of which levels were greater 2 44 than in mice with cells from mBM without hOss and similar or reduced compared with that 2 45 observed in mice receiving cells from hOss ( Figure 4E and H) . The human hematopoietic 2 46 cells recovered from secondary NSG mice were primarily composed of B lymphoid cells, 2 47 regardless of origin (hOss and mBM) ( Figure 4F and I ). Overall, these results show that 2 48 hOss maintain stronger human HSC potential than mBM of mice without hOss. Interestingly, 2 49 the enhanced myeloid cell differentiation observed in hOss ( Figures 2 and 3) was lost in the 2 50 mBM of secondary NSG mice in the absence of the hOss environment. 2 51 2 52 Gene expression profiles reveal enhanced myeloid precursors/progenitors and 2 53 decreased lymphoid precursors/progenitors in immature human hematopoietic cells 2 54 from fetal and post-natal hOss . To further characterize the immature cell compartment 2 55 engrafting hOss compared with mBM from mice without hOss, we performed single-cell RNA 2 56 sequencing of hCD34+ cells sorted at 12 weeks post-transplant ( Figure 5A). A total of 9,894 2 57 cells, containing 3,188, 4,005 and 2,701 cells from mBM without hOss, F/hOss and P-2 58 N/hOss, respectively, were successfully integrated in UMAP ( SupFig5A-B). Seurat 2 59 unsupervised clustering (resolution levels = 0.2) was applied to separate cells into 13 2 60 clusters according to the gene expression profile (Hao et al., 2021). Comparing these results 2 61 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 9 with the expression profiles described in (Hay et al., 2018), the 13 clusters were classified 2 62 according to cell differentiation pathway, which highlighted lymphoid (CD34 + Multilin/CLP, 2 63 CD34+ pre-B cycling, CD34 + pro-B cycling, CD34 + pre-plasma cell (Pre-PC)), myeloid 2 64 (immature neutrophils/monocytes, neutrophils), erythroid and dendritic cell populations 2 65 (CD34+ early erythroblasts, erythroblasts, CD34 + MDP/pre-dendritic and dendritic) ( Figures 2 66 5B-C). Cell cluster annotation revealed that 70.5% of cells (6,975 cells) were either B-cell 2 67 lineage oriented (CD34 + pre-PC and pre-B cell cycling) or multilineage HSCs/progenitors 2 68 (CD34+ Multilin/CLP and CD34 + HSC/MPP/LMPP) ( Figure 5D ). The other immature cells 2 69 (immature neutrophils/monocytes, CD34+ MDP/pre-dendritic cells, CD34 + Eo-B-Mast, CD34+ 2 70 early erythroblasts, erythroblasts ) were detected at lower levels. The repartition of these cell 2 71 subpopulations/clusters indicated variations among cells isolated from hOss and mBM; for 2 72 example, early erythroblasts and neutrophils (c lusters 7 and 9) were frequent in P-N/hOss 2 73 cells, while CD34 + eosinophils-basophiles-mastocytes (clu ster 8) were highly represented in 2 74 F/hOss (SupFig5C). When progenitors/precursors were grouped into B lymphoid (including 2 75 CD34+ pre-PC, CD34 + pre-B cycling, CD34 + Multilin/CLP, CD34 + pro-B, CD34 + pre-B, 2 76 Follicular B cells), myeloid/erythroid (imm ature neutrophils/monocytes, neutrophils, CD34 + 2 77 MDP/pre-dendritic cells, CD34 + Eo-B-Mast, CD34 + early erythroblasts, erythroblasts) and 2 78 early HSPC (CD34 +HSC/MPP/LMPP) lineages, significant differences were observed 2 79 between mBM and hOss ( Figure 5E) that were similar to variations observed in the mature 2 80 cell compartments ( Figure 2E-I ). Indeed, this analysis showed that immature cells from 2 81 F/hOss and P-N/hOss comprise significantly reduced B lymphoid precursors/progenitors 2 82 (p<0.0001, χ 2 test) and increased myeloid/erythroi d precursors/progenitors (F/hOss vs mBM, 2 83 p<0.0001; P-N/hOss vs mBM, p<0.0001, χ 2 test) compared with mBM from mice without 2 84 hOss. We also observed a slightly enhanced proportion of HSC/MPP/LMPP cells in hOss 2 85 (Figure 5E, F/hOss vs mBM, p<0.0001, P-N/hOss vs mBM, p=0.0015, χ 2 test), in 2 86 accordance with the results of the functional assays displayed in Figure 4C-F. 2 87 We also assessed whether hOss could impact gene expression in the most immature CD34 + 2 88 HSC/MPP/LMPP cells of our data set. We analyzed the most differentially expressed genes 2 89 in the HSC/MPP/LMPP cell cluster isolated from hOss and mBM. Comparing P-N/hOss and 2 90 mBM, we identified 318 differentially expressed genes (p<0.05), of which 298 genes were 2 91 upregulated and 20 genes were downregulated in P-N/hOss-derived cells ( Figure 5F ). 2 92 Comparing F/hOss and mBM, we found 4,443 differential genes (1,130 upregulated genes 2 93 and 3,313 downregulated genes in F/hOss) (Figure 5G). The 50 most upregulated genes are 2 94 presented in SupTables 1 and 2. Examination of the two lists of upregulated genes revealed 2 95 47 common genes, of which 9 genes were included among the 50 most upregulated genes in 2 96 both F/hOss and P-N/hOss ( Figure 5H-I, SupTable 3, SupTable 1-2 ). Among these 9 2 97 genes, LRRC75A (Leucin-rich Repeat Containing 75A pr otein) has recently been implicated 2 98 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 0 in the regulation of VEGF in ischemic BM-MSC, and VEGF is a regulator of HSC function 2 99 (Miura et al., 2023). However, VEGF was not enhanced in our HSC/MPP/LMPP data set, 3 00 thus suggesting that LRRC75A enhancement does not impact VEGF transcriptional levels in 3 01 immature cells, in accordance with the finding that VEGF is typically produced by endothelial 3 02 cells. To further examine the modified processes in hOss-derived cells, we performed KEGG 3 03 and GO enrichment analyses using the 100 most deregulated genes in P-N/hOss and 3 04 F/hOss, compared with mBM. We found that metabolic pathways/OXPHOS-cell respiration 3 05 was enriched in P-N/hOss, and several pathways unrelated to HSC function/potential were 3 06 slightly enriched in F/hOss ( SupFig5B-C). We also assessed several other gene sets more 3 07 specific to hematopoietic cells (list in SupTable4) but did not observe enrichment in any 3 08 specific pathway in HSC/MPP/LMPP cells from hOss compared with mBM. Overall, these 3 09

revealed only a discreet qualitative distinction between the HSC compartments 3 10 recovered from hOss and mBM. However, we observed a slightly increased proportion of 3 11 HSCs in hOss and a bias toward B lymphoid/myelo-erythroid differentiation at the 3 12 progenitor/precursor levels, which may contribute to the observed mature cell variations and 3 13 the enhanced functional HSC. 3 14 3 15 Clonal tracking shows that hOss promote the human myelopoietic cell lineage in both 3 16 hOss and the mBM of mice with hOss. We next used lentiviral barcoding to label and track 3 17 CD34+ HSPC-driven hematopoietic reconstitution at the clonal level in hOss and mBM 3 18 (Figure 6A ). CD34 + cells from umbilical cord blood were transduced with 21-bp lentiviral 3 19 barcode libraries containing 40×10 4 different barcodes expressing GFP (Eisele et al., 2022). 3 20 Low transduction efficiency was achieved (7.8-16.8% GFP + cells, SupFig6A and 3 21 SupTable5) ensuring that most cells were labeled with one barcode. Transduced cells were 3 22 injected into mice with or without hOss and hematopoietic GFP + CD19+ B-cells, CD14+CD15+ 3 23 myeloid cells and CD34 + immature cells were sorted from hOss and mBM at 12 weeks after 3 24 injection (Figure 6A, SupFig6B ). At that time point, GFP + and GFP - HSPCs had generated 3 25 the same proportion of each cell lineage fraction ( SupFig6C), thus showing similar 3 26 differentiation from the transduced and non-transduced cells. After filtering and quality control 3 27 (SupFig6D), we obtained variable barcode numbers between mice, which was neither 3 28 correlated with the presence of hOss nor the hOss origin (SupTable5). Based on the number 3 29 of injected CD34+ umbilical cord blood cells and the transduction efficiency, we estimated the 3 30 frequency of HSCs at 1/1000 to 1/100 ( SupTable5), which is in accordance with a previous 3 31 study using barcoded human HSCs (Cheung et al., 2013). We then assessed hematopoietic 3 32 variations between mice with and without hOss. In the mBM without hOss, the progeny of 3 33 barcoded HSPCs was primarily biased toward B- cells, whereas in mice with hOss, the 3 34 barcoded HSPCs more uniformly yielded B-cells and myeloid cells ( Figure 6C), which is in 3 35 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 1 agreement with the flow cytometry results ( Figure 2). In mice bearing hOss, we found an 3 36 increased proportion of clones containing myeloid cells, which was only statistically 3 37 significant for mice with P-N/hOss (SupFig6E). No difference was observed in total barcodes 3 38 (Figure 6B) or the number of cells produced per barcode (clone size) when hOss and mBM 3 39 of mice with hOss were compared with mBM without hOss, with the exception of myeloid 3 40 clones in the P-N/hOss condition, which were significantly smaller ( SupFig6F). Thus, the 3 41 increased myeloid to B lymphoid ratio observed at the population level in P-N/hOss was 3 42 associated with an increase in small myeloid-biased clones and not to an increased 3 43 production of total myeloid cells per clone. Nex t, we investigated the potential relationship 3 44 between human hematopoiesis in hOss and mBM in the same mice. More than 40% of 3 45 clones/barcodes were common between the hOss and mBM ( Figures 6D-E , blue boxes, 3 46 SupFig6G), thus indicating that many hematopoietic cells originating from the same HSPC 3 47 were distributed in both sites. These observations provide insight into the mechanisms of 3 48 phenotypic similarities between mBM from mice with hOss and hOss-containing cells at the 3 49 population level ( SupFigures 2C-F). The common cells constituted mainly large multipotent 3 50 clones in which CD34 + immature cells were also detected, which is indicative of self-renewal 3 51 properties ( Figures 6D-E ). Clones uniquely present in one BM site were also detected 3 52 (Figures 6D-E, red boxes, SupFig6G ) and found at similar levels in mBM and F/hOss, 3 53 whereas more unique clones were detected in P-N/hOss ( SupFig6G). Interestingly, while 3 54 mice with F/hOss were primarily reconstituted with mBM/hOss common large multipotent 3 55 clones (Figure 6D, SupFig6H), mice with P-N/hOss displayed clones that varied in size, with 3 56 a trend of enhanced myeloid-restricted clones that were either common or localized in P-3 57 N/hOss (Figure 6E, SupFig6I). Overall, our results indicate that a diversity of multipotent or 3 58 more lineage-restricted HSPC clones reconstitute human hematopoiesis in both the mBM 3 59 and hOss in these models, many of which were common between BM sites. The presence of 3 60 hOss creates a more favorable environment for human myelopoiesis in large clones in 3 61 F/hOss and small clones in P-N/hOss. 3 62 3 63

3 64 In this study, we show that hBM hOss models support the capacity of human hematopoietic 3 65 cell differentiation and production in immune-deficient mice harboring hOss. These hOss also 3 66 exhibit intrinsic self-replenishing potential by maintaining implanted hMSCs capable of serial 3 67 functional hBM formation and hematopoietic s upport. These results are concordant with 3 68 previously described hBM models that display efficient human hematopoietic support 3 69 following HSPC transplant, including balanced production of mature myeloid cells compared 3 70 with mature lymphoid cells (Chen et al., 2012; Abarrategi et al., 2017; Fritsch et al., 2018; 3 71 Grigoryan et al., 2022), thus demonstrating that the results between different hOss models 3 72 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 2 are reproducible. Such a myeloid bias was detec ted, regardless of whether the MSCs used 3 73 to form hOss were of fetal or P-N origin. However, P-N/MSC-derived hOss and mBM 3 74 displayed smaller myeloid-biased HSPC clones compared with mice with F/hOss, in which 3 75 myeloid cells were produced from more multi potent clones. These results suggest that even 3 76 if the developmental origin of MSCs did not significantly impact hematopoietic cell type 3 77 recovery at the population level, there are subtle differences between F/hOss and P-N/hOss 3 78 niches in terms of the type of clones that reconstitute human hematopoietic cells. 3 79 It was previously unknown whether this bias in lineage cell production originated from biased 3 80 differentiation of HSCs or progenitors towards the myeloid lineage at the expense of the 3 81 lymphoid lineage. In our study, we assessed gene expression profiles in CD34 + cells isolated 3 82 from hOss and mBM at the single-cell level. Our results indicate that HSPCs are not lineage 3 83 bias at the transcriptional level, as specific lineage commitment programs were not observed 3 84 in HSPCs isolated from hOss, compared with mBM. The single-cell RNA sequencing results 3 85 rather highlighted the presence of lar ger myeloid and erythroid progenitor/precursor 3 86 compartments, which is probably supported by human factors produced by the humanized 3 87 microenvironment of hOss (Chen et al., 2012; Reinisch et al., 2016; Grigoryan et al., 2022). 3 88 These results are in accordance with previ ous findings from a BM model generated by 3 89 implanting collagen scaffolds carrying human MSCs, which displayed enhanced myeloid 3 90 progenitors, as assessed by cell surface mark er and CFU-C analysis (Fritsch et al., 2018). 3 91 However, when secondary transplants were perfo rmed in regular NSG mice (without hOss), 3 92 even though HSC levels were enhanced in both models (our results and Fritsch et al., 2018), 3 93 we did not observe an increased myeloid bias in human CD45 + cells derived from hOss, thus 3 94 indicating that without the hOss influence, the myeloid potential of human HSPCs is no 3 95 longer detected. The weak enhancement in my eloid production observed in previous studies 3 96 was probably due to the recipient mouse models used (STRG, MSTRG), which produce 3 97 human myeloid cytokines (Fritsch et al., 2018). Our results indicate that mice with hOss can 3 98 substitute cytokine-humanized mice in terms of myeloid cell lineage production from 3 99 engrafted HSCs. 4 00 Indeed, using NSG recipient mice that do not produce human factors, we found that primary 4 01 hOss had a strong impact on human hematopoiesis, which was detected from mature cells to 4 02 immature cells, and was observed from hOss to the mBM of mice with hOss. Importantly, this 4 03 impact on lineage properties persisted in MSCs isolated from primary hOss, as observed in 4 04 human hematopoietic cells recovered from secondary hOss, especially when hOss formation 4 05 was of F/MSC origin, thus suggesting that F/MSCs are more potent than P-N/hOss in the 4 06 long term. These results indicate that hOss are dynamic multicellular structures that maintain 4 07 MSCs with the ability to differentiate in hOss capable of perpetuating cell lineage in a 4 08 balanced manner, which is an established feature of BM. Interestingly, in one study in which 4 09 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 3 hOss were implanted after HSPCs had engrafted mBM, the balanced myeloid-lymphoid 4 10 progeny was observed in hOss (Reinisch et al., 2015). This finding is in line with our results 4 11 of mBM from hOss-bearing mice, thus displaying the impact of hOss both in proximity and at 4 12 a distance. Using clonal tracking of hematopoietic cell production, we show that in hOss-4 13 grafted mice, many human hematopoietic cells recovered from mBM and hOss originated 4 14 from multipotent HSPCs common between mBM and hOss, especially in F/hOss. Such 4 15 clones were also detected independently in hOss or mBM, although they were rarer, thus 4 16 highlighting that at 12 weeks post-transplant, human hematopoiesis in such models occurs 4 17 through differentiation of multipotential HSCs. Furthermore, smaller unipotent myeloid cell 4 18 clones were present in hOss, especially in P-N/hOss, potentially associated with different 4 19 waves of progenitor differentiation or concomitant production of hematopoietic cells by a 4 20 diversity of HSPCs, as previously described (Cheung et al., 2013). The extent to which hOss 4 21 and mBM compete to attract HSCs versus committed progenitors remains to be assessed 4 22 using hematopoiesis kinetic analyses. Whether the common clonally-related hematopoietic 4 23 cells are due to cross-talk of differentia ted cells between mBM and hOss or are rather 4 24 generated from single barcoded HSCs that divided, trafficked through the blood stream and 4 25 settled in the mBM and hOss to differentiate remains to be examined. 4 26 In this study, we also investigated whether the developmental age of BM from which MSCs 4 27 were recovered impacted hOss formation and hematopoietic support. Previous studies on 4 28 human F/MSCs, pediatric MSCs and adult MSCs have reported variations in cell surface 4 29 marker expression and fibroblast CFUs, outlining specificities in F/MSCs (Maijenburg et al., 4 30 2012). In mice, several age-related BM micr oenvironment factors modulate HSC potency, 4 31 including osteopontin, which enhances levels in young adults compared with older adults -4 32 attenuating aging of old HSCs, and the CCL5/RANTES, which is a secreted matrix protein 4 33 that contributes to the bias in old HSCs toward myeloid differentiation (Guidi et al., 2017; 4 34 Ergen et al., 2012). In our experiments, both F/MSC and P-N/MSC reproducibly formed 4 35 hOss, with increased size and weight in F/MSC-derived hOss, which was likely associated 4 36 with enhanced cell proliferation and thicker bone structure. Such differences did not 4 37 dramatically impact hematopoietic cell production at the population level, at least in primary 4 38 hOss, with a trend towards myeloid progenitors only detected by single-cell RNAseq in P-4 39 N/MSC-derived hOss compared with F/MSC-der ived hOss. Interestingly, secondary hOss 4 40 perpetuated the variations in size albeit to a lesser extent, and P-N/hOss lost the myeloid-4 41 lymphoid lineage balance, indicative of exhausti on most likely mediated by stress-induced 4 42 serial transplantation. Further research is necessary to understand the drift mechanisms in 4 43 primary MSCs undergoing hOss formation. The possible aging caveat of serial tissue culture-4 44 isolated MSCs and the difficulty to reproduce hOss formation from one lab to the next has 4 45 recently been challenged (Côme et al., 2020). A human telomerase-immortalized 4 46 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 4 mesenchymal cell line, Mesenchymal Sword of Damocles, has been shown to recapitulate 4 47 the various multipotent and hematopoietic s upportive properties of primary hMSCs by 4 48 generating hOss in vivo after ex-vivo priming to chondrocyte or osteocyte differentiation 4 49 (Grigoryan et al., 2022). Such a cell line is a valuable tool as it produces >1000 hOss 4 50 originating from a clonal cell population (Grigoryan et al., 2022). hOss obtained from 4 51 Mesenchymal Sword of Damocles cells and primary MSC samples are also useful recipients 4 52 of pathologic cells, such as acute myeloblastic leukemia and BM-infiltrating solid tumor cells 4 53 (breast cancer and neuroblastoma). This appr oach enables analysis of the relationship 4 54 between the BM microenvironment and pathologic cells in a humanized context, including 4 55 the soluble factors involved (Antonelli et al., 2016; Abarrategi et al., 2017; Reinisch et al., 4 56 2016; Vaiselbuh et al., 2010; Martine et al., 2017; Grigoryan et al., 2022). 4 57 These results contribute novel findings regarding the distinct impact of hBM models 4 58 generated from hMSCs isolated from normal F and P-N donors on normal hematopoietic 4 59 development after transplantation of hum an HSPCs. Continued advancements in these 4 60 models, such as the incorporation of human endot helial cell-derived vasculature (Passaro et 4 61 al., 2017), will undoubtedly further impact normal and abnormal hematopoietic cell 4 62 development. However, detailed characterization at the clonal levels, as described here, is 4 63 essential to validate the physiological relevance of these models to study normal and 4 64 abnormal human cells. 4 65

(3554 words) 4 66 Human samples. Umbilical cord blood (CB) samples were collected from healthy infants 4 67 with the informed written consent of the mothers based on the declaration of Helsinki. 4 68 Samples were obtained in collaboration with Clinique des Noriets, Vitry-sur-Seine and the 4 69 Cell Therapy Department in Hôpital Saint Louis, Paris, France. Human BM used in Figure 4 70 2F-J were collected from allografts (mostl y children sibling donors) at the Cell Therapy 4 71 Department in Hôpital Saint Louis, Paris, France. Samplings and experiments were 4 72 acknowledged by the Institutional Review Board of INSERM (Opinion number 13-105-1, 4 73 IRB00003888). CB and BM cells were subjected to Ficoll gradient. CB mononucleated cells 4 74 were enriched for CD34+ cells using the human CD34 MicroBeads Kit following manufacturer 4 75 instructions (130-046-703, Miltenyi Biotec), assessed for purity ( ≥ 70% CD34 + cells), and 4 76 used directly in experiments or froze in fetal bovine serum (FBS, F9665, Sigma-Aldrich) 4 77 supplemented with 10% DMSO (D5879, Sigma-Aldrich) for later usage. hBM cells were 4 78 processed for Flow cytometry directly after Ficoll. 4 79 Fetal bone marrow (F/BM) samples to 12 weeks post-conception were obtained after 4 80 abortion processes with the informed written consent of the parents in accordance to 4 81 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 5 guidelines approved by the French Agence de Biomédecine (PFS18-009). Samples were 4 82 obtained in collaboration with the department of Foeto-anatomo-pathology of the Hospital 4 83 Antoine Beclère, Clamart, France. 4 84 Post-natal BM (P-N/BM) MSC samples were collected from healthy children and adult 4 85 allografts at Necker Enfants Malades hospital, France. 4 86 4 87 Human MSC (hMSC) isolation and culture. Human F/BM samples were crushed in DPBS 4 88 with 1% of penicillin-streptomycin (PS). Following centrifugation, the BM fragments were 4 89 resuspended in α -MEM medium (M4526, Sigma-Aldrich) containing 0.22 μm-filtered 10% 4 90 human platelet lysate provided by the Department for Transfusion Medicine, Paracelsus 4 91 Medical University, Salzburg, Austria or purchased from StemCell TM Technologies (Stem 4 92 Cells Lysate, 06962), 1% PS and glutamine, 2 units/mL heparin (hereafter called MSC 4 93 medium) and incubated in plastic dishes. P-N/BM cells were extracted from filters, 4 94 centrifuged and directly plated in MSC medium without Ficoll process. After amplification, 4 95 adherent cells were split for further expansion (pas sage 0) or directly plated for experiments. 4 96 At every passage, the culture medium was gently removed, plastic dishes were carefully 4 97 washed with DPBS followed by addition of trypsin/EDTA solution (0.05%, 25300-054, Gibco) 4 98 and incubated at 37°C for 5-10 minutes. Once hMSC were detached, α MEM medium 4 99 containing 10% of FBS was added. Recovered adherent cells were resuspended in MSC 5 00 medium after centrifugation and counted usin g the GuavaEasycyte 8HT apparatus in Muse 5 01 Count & viability kit (MCH600103, Cytek Biosciences) or with trypan blue (T8154, Sigma-5 02 Aldrich). Part of cells were frozen at 1x10 6 cells/vial in FBS containing 10% DMSO. Adherent 5 03 cells were characterized by flow cytometry to measure their MSC content. To expand hMSC 5 04 in cultures, a minimal number of 500 cells per cm2 were set at start. Adherent cells were 5 05 amplified during maximal 4 passages in MSC medium. Secondary hMSC, isolated from 5 06 primary hOss, were obtained using the same me thod as for primary hMSC isolation, e.g. 5 07 after crushing of primary hOss, adhesion of cells on plastic dishes, expansion in hMSC 5 08 medium and immunophenotyping to ensure their hMSC content. 5 09 To measure cell proliferation, hMSC were seeded at 500 cells/cm2 in 24-well plates in 5 10 quadruplicates, three-days prior the counting wa s started. Thereafter adherent cells were 5 11 detached every 24 hours and counted using the G uavaEasycyte 8HT with Muse Count & 5 12 viability kit. 5 13 To test for adipocyte, osteocyte and chondrocyte differentiation, hMSC were cultured in 5 14 conditions provided by Miltenyi Biotec using StemMACSTM Adipodiff media Human (130-5 15 091-677), StemMACS TM OsteoDiff media Human (130-091-678) and StemMACS TM 5 16 ChondroDiff media Human (130-091-679). Oil Red staining kit (0843-SC, Cliniscience) was 5 17 used to reveal adipocyte differentiation, and Alizarin Red S Staining kit (0223-SC, 5 18 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 6 Cliniscience) helped showing osteocyte different iation. All imaging were performed with the 5 19 EVOS FL Auto Imaging System (Thermo Fisher Scientific). 5 20 hMSC flow cytometry antibody panel. All antibodies were used at the 1/100 dilution except 5 21 for CD44 labelling (1/500). Detached adherent cells were labelled with antibodies directed 5 22 against human CD14 (clone 61D3), CD44 (IM7, both from Invitrogen) CD15 (W6D3), CD31 5 23 (WMD59), CD105 (SN6H), CD73 (AD2), CD45 (HI30, all from BioLegend®), CD146 (541-5 24 0B2, Miltenyi Biotech), and CD90 (5E10, BD Biosciences) markers to determine the 5 25 hematopoietic- Lin- MSC fraction. Zombie AquaTM Dye (423101, BioLegend® ) was used as a 5 26 viability marker (1/500). 5 27 5 28 Mice. All experimental procedures were done in accordance with the recommendations of 5 29 the European Community (2010/63/UE) and French Ministry of Agriculture regulations 5 30 (animal facility registration number: A9203202) for the care and use of laboratory animals. 5 31 Experimental procedures on animal were approv ed by the French National Animal Care and 5 32 Use Committee (Project A21_021, extension of Project A17_009, APAFIS#9458-5 33 2017033110277117). NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG), originally purchased from 5 34 the Jackson Laboratory (Bar Harbor, Maine, USA), were housed and bred in specific 5 35 pathogen-free animal facilities (Commissariat à l’Energie Atomique et aux Energies 5 36 Alternatives (CEA), Fontenay-aux-roses, France). Experiments were performed in 8 to 12-5 37 week-old NSG females. At dedicated time points, NSG mice were sub-lethally irradiated at 5 38 2Gy using a GSRD1-irradiator (137 CS source , GSM, dose rate 0.97gy/min, irradiation 5 39 plateform of Institut de Radiobiology Cellulaire et Moléculaire, Fontenay-aux-Roses) 4 hours 5 40 before intra-venous retro-orbita l injection of human CB CD34 + cells under isoflurane 5 41 anesthesia. 5 42 5 43 Human Ossicles. The human ossicle (hOss) formation pr otocol was adapted from previous 5 44 works (Reinisch et al., 2017b). Briefly, hOss were generated after subcutaneous injection of 5 45 2x106 MSC. MSC were centrifuged 5 min at 300g and resuspended in 50µL human platelet 5 46 lysate and the cell suspension was kept on ice. Osteogenic factor, hBMP-7 (5 μg, 130-108-5 47 988, Miltenyi Biotec), was added in the mix of cells and lysate. The cell suspension was then 5 48 mixed to 250µL cold matrigel (ECM625, Mercks Millipore) and the final 300µL cell mixture 5 49 was injected subcutaneously in one flank of isoflurane-anesthetized mice. Every mouse 5 50 received 2 injections to form 2 hOss (allowed by ethical committee rules). Bone formation 5 51 was tested after injection of 4 nmol of OsteoSense 750 ex (NEV10053EX, Perkin Elmer) per 5 52 mouse 24 hours before tomography analysis. 5 53 5 54 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 7 Human hematopoietic development in NSG mice. Two to 8 weeks after hOss 5 55 implantation, 1x105-1x106 CB CD34+ cells (100 µL DPBS) were injected intravenously (IV) in 5 56 the sublethally irradiated mice to constitute human hematopoiesis in vivo. Three-months 5 57 later, mice were sacrificed, 2 tibiae and 2 femurs (4 long bones, called mBM) and hOss were 5 58 harvested from euthanized animals and hematopoi etic cells were isolated by flushing the 5 59 mBM with DPBS and by gently crushing hOss. The isolated cells were maintained on ice 5 60 during the analysis. Immunophenotyping of human cells was performed using flow cytometry 5 61 (see below). Counting was realized with a Guava Easycyte 8HT with Muse count & Viability 5 62 kit. Serial transplantation was carried out after flow cytometry measurement of the percent of 5 63 hCD34+ cells and the counting of total BM/hOss ce lls, by intra-venous injection of 3-5x10 5 5 64 hCD45+CD34+ cells, recovered from primary reci pient mice, in secondary sublethally 5 65 irradiated animals. 5 66 5 67 Immunostaining and coloration. Tissues were dehydrated in graded ethanol dilution baths, 5 68 cleared with xylene and embedded in paraffin wax. Microtome sections (5 µm, Leica 5 69 RM2245) were mounted on adhesive slides (Klinipath- KP-PRINTER ADHESIVES), 5 70 deparaffinized, and stained with hematoxylin, eosin and saffran (HES) for morphology and 5 71 Fast Green-Safranin O to detect proteoglycan. Slides were scanned with the Panoramic 5 72 Scan 150 (3D Histech) and analyzed with the CaseCenter 2.9 viewer (3D Histech). 5 73 To visualize the vascularization of hOss, we froze directly a piece of ossicles in cryo-gel. We 5 74 used CryoJane Tape-Transfert Systems of Leica Biosystems on cryotome sections of 16µm. 5 75 Samples were fixed with acetone during 8 min at -20°C, dried during 3 min at RT and 5 76 washed twice during 2 min and 10 min. Then a saturation was set during 30 min with normal 5 77 goat serum (NGS, 15%, 005-000-121, Jackson ImmunoResearch) in PBS and blocked twice 5 78 15 min with Avidin/biotin blocking system (927301, Biolegend ® ). Following was a one-night 5 79 incubation at 4°C with Meca32-biotin (1/250, 558773, BD Pharmingen) in PBS with 5% NGS 5 80 and 0.05% Tween20. Next, primary antibodies were washed out twice during 5 min with PBS 5 81 0.1% Tween20 and 5 min with PBS. An additional incubation occurred with a Streptavidin 5 82 AlexaFluor-594 (1/500, S32356, Invitrogen) diluted in PBS with 5% NGS, 0.05%Tween20 5 83 during 1h at RT following by three 5 min washes with PBS 0.1% Tween20 at RT. DAPI 5 84 (1/2000) was next deposited for 5 min at RT and the samples were mounted on glass slides 5 85 using Prolong Gold antifade reagent and a coverslip (P36934, Invitrogen). Samples were 5 86 imaged using Spining Disk confocal microscope and analyzed by ImageJ. 5 87 Then hOSS were paraffin embedded, sectioned (4 μ m) and stained with MPO (Clone Poly, 5 88 DAKO), human anti-CD45 antibody (clone 2B11+PD7/26, DAKO), human anti-CD34 5 89 antibody (clone QBend-10, DAKO), human anti-CD14 antibody (clone 7, BIOSB-Diagomics), 5 90 human anti-Glycophorin C (clone Ret40f, DAKO), human anti-CD61 (clone 2F2, LEICA), and 5 91 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 8 mouse anti-endomucin (clone V.1C7.1, Abcam). Images were acquired using a Lamina slide 5 92 scanner (Perkin Elmer) and analyzed with CaseViewer software (Version 2.4.0.119028). 5 93 5 94 Flow cytometry phenotyping of human/murine hematopoietic cells. 1x10 6 cells from 5 95 mBM and from hOss were centrifuged, re suspended in 100µL D PBS and incubated with 5 96 antibodies (1/100, all from BioLegend®, otherwise indicated) directed against human CD19 5 97 (HIB19), CD14 (61D3), CD15 (W6D3), CD34 (581), CD90 (5E10, BD Biosciences), CD3 5 98 (SK7), CD38 (HB-7), CD45 (HI30) and mouse TER119 (TER-119, 1/500, eBioscience), 5 99 mouse CD45 (30-F11, 1/500, BD Pharmingen). Zombie Aqua™ Dye was used as a live/dead 6 00 cell marker (dilution 1/500). All data were acquired on BD FACSCanto™ II and BD LSR II 6 01 SORP machines with the DIVA software. Compensation controls were performed with single-6 02 stained compensation beads. After acquisition, live cells were gated and antibody-labelling 6 03 analysed using FlowLogic 7 software. 6 04 6 05 Colony Forming Units (CFU) assays. As for secondary transplantation, tested cell 6 06 suspensions were characterized by flow cytometry and the percent of CD34 + cells carefully 6 07 monitored. Based on these results, an equivalent of 1,000 human CD34 + cells from mBM 6 08 and hOss were plated in 1 mL of methylcellulose-based medium (130-125-042, 6 09 StemMACS™ HSC-CFU Assay Kit; Miltenyi Biotec) that contained human SCF, IL-3, IL-6, 6 10 Erythropoietin (EPO), G-CSF, GM-CSF and 1% Penicillin/Streptomycin. Each experiment 6 11 was performed in technical triplicates. The cells were incubated for 14 days at 37°C. After 14 6 12 days, human hematopoietic colonies, also called CFU, were counted and qualified by their 6 13 shape and the morphology of the containing cells using an inverted microscope. 6 14 6 15 Single cell RNA-seq. 10 5 human CD34 + HSPCs from a single CB source were injected 6 16 intravenously into NSG mice, previously irradiated with sub-lethal doses (2Gy). Mice were 6 17 either hOss-bearing or hOss-free. hOss were obtained from F/hOss (n=1) or P-N/hOss (n=2). 6 18 As control condition, we used NSG mice without hOSS (mBM W/o hOss, n = 2). Three 6 19 months after HSPC injection, mice were sacrificed and mononuclear cells (MNC) that had 6 20 engrafted into hOSS and murine BM were pooled for each condition. In total, we had 3 6 21 samples of murine BM (mBM W/o hOss, mBM P-N/hOss and mBM F/hOss) and 2 samples 6 22 of pooled ossicles (F/hOss and P-N/hOss). These 5 samples were then characterized by flow 6 23 cytometry to assess i) the human engraftment (%hCD45 +), ii) the repartition of the 6 24 populations and iii) the percentage of hCD34 expressing cells (%CD45 +/CD34+). To enrich 6 25 samples in CD34 + fraction, immunomagnetic sorting was performed (130-046-703, CD34 6 26 MicroBead Kit, human, Miltenyi Biotec). After checking the purity and viability of the sorted 6 27 cells, the CD34+ cell fraction was used to prepare scRNA-seq libraries. 6 28 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 1 9 Libraries preparation. Human CD34 + cells were washed with PBS containing 0.04% BSA. 6 29 Cell concentration and viability were determined microscopically with a Malassez counting 6 30 chamber cell after staining with trypan blue. Libraries were prepared with the Chromium 6 31 system of 10× Genomics, with the Chromium Single Cell 3’ v3.1 kit according to the 6 32 manufacturer’s protocol (www.10xgenomics.co m). The samples were processed on the 6 33 same chromium chip. The number of cells targeted was 8000 per sample. Sequencing of the 6 34 libraries was performed on a Miseq sequencer (Illumina) in pair-end, dual-index mode, with a 6 35 target of 400 million reads per sample. 6 36 Bioinformatic analysis. All bioinformatics analyses are based on the recommendations 6 37 issued by (Luecken and Theis, 2019). 6 38 Pre-processing of scRNA-seq data. The Galaxy interface was used to construct the gene-6 39 barcode (= Gene-cell) expression matrix. Briefly, FastQ files were aligned to the reference 6 40 genome (GRCh38) via STAR (Spliced Transcripts Alignment to a Reference) (Dobin et al., 6 41 2013), with gene annotation from ENSEMBL. After quantifying the UMIs (Unique molecular 6 42 identifiers) for each barcode, the unfiltered gene-barcode matrix was generated. From this 6 43 unfiltered matrix, we created a filtered matrix (without empty cells and cells with a low 6 44 number of UMIs (UMI < 50)). The rest of the analysis was carried out using Seurat (Version 6 45 4.1.0)316. Quality controls are based on the co mbined use of 3 variables: the number of 6 46 UMIs per barcode, the number of genes per barcode and the fraction of reads from 6 47 mitochondrial genes per barcode. We have chosen to retain for further analysis only 6 48 barcodes expressing at least 200 UMI, 200 genes , with a percentage of mitochondrial genes 6 49 below 20%. UMI counts were normalized with the “NormalizeData” Seurat function. A linear 6 50 transformation of gene expression was then app lied to prevent certain highly expressed 6 51 genes from masking the expression of others ("Scaling"). 6 52 Dimensionality reduction and Integration. For each matrix, we used the "vst" method to 6 53 determine the Highly Variable Features. The selection of these genes then enabled the 6 54 application of algorithms specially dedicated to dimensional reduction like PCA and UMAP. 6 55 To enable comparative analysis between different experimental conditions, we then 6 56 integrated data sets. Integration involves a pr eliminary step of identifying "anchors", which 6 57 correspond to "pairs" of biologically similar cells between the different datasets. These 6 58 integrated data were then used for the graphical representations. 6 59 Clustering and Cell annotation. Based on the UMAP generated on the integrated data, 6 60 cells were grouped into clusters using the k Nearest Neighbors (KNN) algorithm. The genes 6 61 most differentially expressed by each cluster were determined and used to define a list of 6 62 biomarker genes. Using these lists, which we co mpared with publicly available data (Hay et 6 63 al., 2018a), we assigned each cluster to a particular cell type. We chose not to perform cell 6 64 cycle regression, so that we could study whether hOss could have an effect on the cell cycle. 6 65 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 0 Differential gene expression. After specifying the subpopulation of interest (subset.ident) 6 66 and the samples to be compared (ident.1 and identi.2), we performed a Wilcoxon test. 6 67 Seurat's "FindMarkers" function was used for these differential expression analyses. EnrichR 6 68 was used for enrichment analysis. 6 69 6 70 Barcode virus production, infection and injection. 106 fresh CB CD34+ cells/ml were pre-6 71 stimulated in BIT 9500 Serum Substitute (09500, StemCell TM Technologies) supplemented 6 72 with cytokines (hSCF 100ng/mL, 130-046-703; hFlt3-Ligand 100ng/mL, 130-096-479; hIL-3 6 73 60ng/mL, 130-095-069; TPO 10nM, 130-094-013, al l from Miltenyi Biotec) and 1% PS during 6 74 24 hours. Sulfate protamine (5µg/mL) and 6µ L barcode-containing lentivirus vector 6 75 suspension were added per well on the next day. The lentiviral vector mix was produced as 6 76 before (Eisele et al., 2022) by infecting HEK293T cells with the barcode plasmid, p8.9-QV 6 77 and pVSVG in DMEM-GlutaMAX (Gibco) supplemented with 10% FCS (Eurobio), 1% MEM 6 78 NEAA (Sigma), and 1% sodium pyruvate (Gibco) using polyethyleneimine (Polysciences). 6 79 Supernatant was 0.45 μ m filtered, concentrated 35 times by 1.5 hr ultracentrifugation at 6 80 31,000 × g, and frozen at –80°C. The lentiviral vector mix included a DNA stretch of 180 bp 6 81 with a 20 bp ‘N’-stretch, called the barcode and a GFP fluorescent reporter. It contained 6 82 around 40,000 different barcodes. CB CD34 + cells were spinoculated with the lentiviral 6 83 particles during 1h30 at 2000rpm at room temperature and thereafter incubated during 4h30 6 84 at 37°C. The cells were then washed in α MEM media supplemented by 10% FBS, 6 85 resuspended at 10 6 cells/mL of DPBS and 10 5 cells were injected per NSG mouse carrying 6 86 or not hOss. This protocol was designed to reach a single genomic integration of the 6 87 lentivector per cell, and thus the transduction was meant to be <30% efficiency that was 6 88 verified by flow cytometry 72h after the end of transduction. 6 89 Cell progeny isolation for barcoding . When clonal tracking was to be done, engrafted 6 90 human hematopoietic progenitor cells (GFP +/hCD45+/CD34+), B cells (GFP +/hCD45+/CD19+) 6 91 and myeloid cells (GFP +/CD45+/CD14+/CD15+) were sorted using a BD InfluxTM Cell Sorter 6 92 (BD Biosciences) using antibodies described before. Sorted cells were mixed with PCR lysis 6 93 buffer (301-C, Euromedex) supplemented with proteinase K solution (20mg/ml, 25530-049, 6 94 Invitrogen), incubated at 55°C for 2 hours, 85°C for 30 min, 95°C for 5 min, and finally stored 6 95 at 4°C. 6 96 Lysis, barcode amplification, and sequencing. Lysed samples were then split into two 6 97 replicates, and a three-step nested PCR was performed to amplify barcodes and prepare for 6 98 sequencing as in (Eisele et al., 2022). In summary: the Taq DNA Polymerase Recombinant 6 99 500U (Invitrogen™) was used to perform all the PCRs. The first step amplifies barcodes (top-7 00 -LIB [5 ′ TGCTGCCGTCAACTAGA ACA-3 ′ ] and bot--LIB [5 ′ GATCTCGAATCAGGCGCTTA-7 01 3′ ]). The second step adds unique 8 bp plate indices as well as Read 1 and 2 Illumina 7 02 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 1 sequences (PCR2--Read1-plate-index-partial-Top--forward 5 ′ ACAC TCTT TCCC TACA 7 03 CGAC GCTC TTCC GATC TNNN NNN NNCTA GAAC ACTC GAGATCAG 3 ′ and PCR2-7 04 -Read2-partial-Bot--reverse 5′ GTGA CTGG AGTT CAGA CGTG TGCT CTTCCGAT CGAT 7 05 CTCG AATCAGGC GCTTA3 ′ ). In the third step, P5 and P7 flow cell attachment sequences 7 06 and one of 96 sample indices of 7 bp are added (PCR3--P5-forward 5 ′ AATGATA CGGC 7 07 GACC ACCG AGAT CTAC ACTC TTTC CCTA CACG ACGC TCTT CCGATCT3′ and PCR3-7 08 -P7-sample--index--reverse 5 ′ CAAG CAGA AGAC GGCA TACG AGAN NNNN NNGT 7 09 GACT GGAG TTCAGA CGTGCTCTTCCGATC3′ ). (PCR program: hot start 5 min 95°C, 15 s 7 10 at 95°C; 30 s at 57.2°C; 30 s at 72°C, 5 min 72°C, 4°C forever. 30 cycles [PCR1--2] or 15 7 11 cycles [PCR 3]). 7 12 Both index sequences (sample and plate) were designed based on Faircloth and Glenn, 7 13 2012 such that sequences differed by at least 2 bp. To avoid lack of diversity at the 7 14 beginning of the reads, we performed the PCR2 with a mix of 4 different plate indexes for the 7 15 PCR2-Read1-plate-index-partial-Top--forward primer (ACGGAATG, CTAACTCC, 7 16 GATGGTCA, TGGCAGAA). Primers were or dered desalted as high-performance liquid 7 17 chromatography purified. During lysis and each PCR, a mock control was added. The DNA 7 18 amplification by the three PCRs was monitored by the run on a 2% agarose gel. PCR3 7 19 products for each sample and replicate were pooled, purified with the Agencourt AMPure XP 7 20 system (ratio 1,2x) (Beckman Coulter) and diluted to <10nM. A High Sensitivity D1000 7 21 ScreenTape® (Agilent) was performed on the purified PCR3 pool, and sequenced on a 7 22 NovaSeq system (Illumina) (SR--65bp) at the Institut Curie facility with 10% of PhiX spike-in. 7 23 Cellular barcode sequence preprocessing. The barcodes in the NGS reads were 7 24 identified by locating the variable sequence be tween the constant sequences to all barcodes 7 25 5' CTAGAACACTCGAGATCAG and 3' TGTGGTATGATGTAT using CellBarcode package 7 26 (Sun et al., 2024). We only kept the barcodes that appeared in the reference list generated 7 27 from sequencing the lentiviral library prov ided in (Eisele et al., 2022), and removed any 7 28 barcodes not detected in both technical replic ates. For each technical replication, we 7 29 normalize the total sequencing reads to 100k reads. Then we used the mean of the 7 30 normalized barcode reads from two technical rep licates for further analysis. Further filtering 7 31 included the removal of barcodes with less than 100 total reads across all samples. 7 32 Additionally, if a barcode represented less than 1% of the reads from a specific sample 7 33 compared to the total reads across all samples of that barcode, it was set to zero in the 7 34 corresponding sample. To validate our filtering process, we assessed the heatmap of reads 7 35 counts across samples within each batch that was generated using the heatmap package 7 36 (Kolde, 2012). Finally, the read numbers were renormalized to the number of GFP positive 7 37 sorted cells and any barcode with less than 2 cells were set to zero. The remaining barcodes 7 38 were utilized for barcode analysis in Figure 6B-E and Figure S6E-I for which the data is 7 39 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 2 renormalized to 100k reads. For the clone size analysis in Figure 6C and S6F, the data was 7 40 renormalized to the total number of barcoded cells extracted during cell isolation and 7 41 analyzed by cytometry to take into account the total size of the different compartments. 7 42 Barcode analysis, plotting and statistics. We calculated the engrafted cells by dividing 7 43 barcode number to estimated injected GFP positive cells (injected cell number multiplied by 7 44 the fraction of GFP positive cells measured after transduction). For lineage bias analysis in 7 45 Figure S6E, barcode clones bias towards “M” were defined as containing less than 10% of B 7 46 and 10% of CD34 cells out of their total output, similarly to previous classification used in 7 47 (Eisele et al., 2022). The same applies for t he “B” biased clones defined as containing less 7 48 than 10% of M and 10% of CD34 cells. In the anal ysis of compartment-specific barcodes in 7 49 Figure S6H-I , first the count for each barcode occurring in multiple cell types within one 7 50 organ were summed. Then if a barcode had zero reads in either the bone marrow or the 7 51 ossicles then it was classified as under the “hOss” group or the “mBM” group respectively. 7 52 The other barcodes that were present in bot h organs are categorized as the "both" groups. 7 53 To produce the figures after filtering, R4.3.1 was used. The ternary plot was done by ggtern 7 54 package (Hamilton and Ferry, 2018) ; boxplot and other general visualization was done with 7 55 ggplot2 package (Wickham, 2011). When appropriate and as indicated in the legend, t-test 7 56 was used to compare between conditions across mice, and paired t-test was used to 7 57 compare conditions within mice. 7 58 7 59 Statistical analysis. Statistical analysis was performed using GraphPad Prism version 9 7 60 software. Values are presented as the median with 95% of Cl or mean ± SEM. Statistical 7 61 comparisons between conditions were determined using Kruskal-Wallis without correction or 7 62 Mann-Whitney test. Differences with p<0.05 (*), p<0.01 (**), p<0.001 (***) and p<0.0001 7 63 (****) were considered statistically significant. 7 64 7 65

7 66 We acknowledge the help of B Burroni, from the Hôpital Cochin, Paris, France (immuno-7 67 histo-chemestry labelling of hOss sections), the people from the technical platforms of Institut 7 68 de Radiobiology Cellulaire et Moléculaire, IBFJ/CEA, Fontenay-aux-Roses, France, in 7 69 particular V Ménard from the Irradiation Platfo rm, D Busso and G Piton from the Molecular 7 70 Bioengineering platform, J Baijer and A Schmitz from the flow cytometry platform, and J 7 71 Rivière and M Vilotte from INRAE for hOss sections and histochemestry. We are grateful to F 7 72 Duconge who helped with tomography and to C Antoniewski and L Bellenger from ARTbio 7 73 Bioinformatics Analysis Facility from Sorbonne University. We acknowledge the 7 74 fetopathology department of Antoine Béclère Hospital, Clamart, France, The clinique des 7 75 Noriets, Vitry-sur-Seine, France and the Cell Therapy department from Hôpital Saint Louis, 7 76 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 3 Paris, France, for providing human cord blood and BM samples. We thank the patients for 7 77 agreeing to help scientific research. We thank S Teinrera Bento for help with barcode PCR in 7 78 the troubleshooting phase of the project and F Cayrac from the BMBC facility for the barcode 7 79 lentivirus production. We thank E Solary, N Drouin and C Jego (Gustave Roussy, Villejuif, 7 80 France) and C Dussiau (Institut Cochin, Paris) who participated to the early days of the 7 81 project. C Moore edited the manuscript for english language. 7 82 This work was supported by institutional grants from INSERM, CEA, Université Paris Saclay 7 83 and Université Paris Cité. The study was also supported by ITMO-Cancer (3R Program, 7 84 édition 2020 and MIC Program, édition 2022), CONECT-AML Pair-Pédiatrie program, Ligue 7 85 Nationale Contre le Cancer (Labellisation Program) and the Fondation ARC pour la 7 86 recherche contre le Cancer (ARC). 7 87 The authors declare no competing financial interests 7 88 7 89 Authors participation 7 90 LR, CF, KG, CC, EP, VB, SD and ND performed experiments. LR, WS, CF and JC 7 91 performed scRNAseq and barcoding bioinformatic analyses; KS, LK, AR, JM, AM and LF 7 92 provided extremely important materials; LP, OK and FP supervised the work; LR, CF and 7 93 WS made the figures; LR and FP wrote the manuscript; all authors read and approved the 7 94 manuscript content. 7 95 7 96 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 4

7 97 Abarrategi, A., K. Foster, A. Hamilton, S.A. Mian, D. Passaro, J. Gribben, G. Mufti, and D. 7 98 Bonnet. 2017. Versatile humanized niche model enables study of normal and malignant 7 99 human hematopoiesis. J Clin Invest. 127:543–548. doi:10.1172/JCI89364. 8 00 Abarrategi, A., S.A. Mian, D. Passaro, K. Rouault-Pierre, W. Grey, and D. Bonnet. 2018. 8 01 Modeling the human bone marrow niche in mice: From host bone marrow engraftment to 8 02 bioengineering approaches. J Exp Med. 215:729–743. doi:10.1084/jem.20172139. 8 03 Antonelli, A., W.A. Noort, J. Jaques, B. de Boer, R. de Jong-Korlaar, A.Z. Brouwers-Vos, L. 8 04 Lubbers-Aalders, J.F. van Velzen, A.C. Bloem, H. Yuan, J.D. de Bruijn, G.J. Ossenkoppele, 8 05 A.C.M. Martens, E. Vellenga, R.W.J. Groen, and J.J. Schuringa. 2016. Establishing human 8 06 leukemia xenograft mouse models by implanting human bone marrow-like scaffold-based 8 07 niches. Blood. 128:2949–2959. doi:10.1182/blood-2016-05-719021. 8 08 Baccin, C., J. Al-Sabah, L. Velten, P.M. Helbling, F. Grünschläger, P. Hernández-Malmierca, 8 09 C. Nombela-Arrieta, L.M. Steinmetz, A. Trumpp, and S. Haas. 2020. Combined single-cell 8 10 and spatial transcriptomics reveal the molecular, cellular and spatial bone marrow niche 8 11 organization. Nat Cell Biol. 22:38–48. doi:10.1038/s41556-019-0439-6. 8 12 Baryawno, N., D. Przybylski, M.S. Kowalczyk, Y. Kfoury, N. Severe, K. Gustafsson, K.D. 8 13 Kokkaliaris, F. Mercier, M. Tabaka, M. Hofree, D. Dionne, A. Papazian, D. Lee, O. 8 14 Ashenberg, A. Subramanian, E.D. Vaishnav, O. Rozenblatt-Rosen, A. Regev, and D.T. 8 15 Scadden. 2019. A Cellular Taxonomy of the Bone Marrow Stroma in Homeostasis and 8 16 Leukemia. Cell. 177:1915-1932.e16. doi:10.1016/j.cell.2019.04.040. 8 17 Chen, Y., R. Jacamo, Y. Shi, R. Wang, V.L. Battula, S. Konoplev, D. Strunk, N.A. Hofmann, 8 18 A. Reinisch, M. Konopleva, and M. Andreeff. 2012. Human extramedullary bone marrow in 8 19 mice: a novel in vivo model of genetically controlled hematopoietic microenvironment. Blood. 8 20 119:4971–4980. doi:10.1182/blood-2011-11-389957. 8 21 Cheung, A.M.S., L.V. Nguyen, A. Carles, P. Beer, P.H. Miller, D.J.H.F. Knapp, K. Dhillon, M. 8 22 Hirst, and C.J. Eaves. 2013. Analysis of the clonal growth and differentiation dynamics of 8 23 primitive barcoded human cord blood cells in NSG mice. Blood. 122:3129. 8 24 doi:10.1182/blood-2013-06-508432. 8 25 Côme, C., A. Balhuizen, D. Bonnet, and B.T. Porse. 2020. Myelodysplastic syndrome 8 26 patient-derived xenografts: from no options to many. Haematologica . 105:864–869. 8 27 doi:10.3324/haematol.2019.233320. 8 28 Cordeiro Gomes, A., T. Hara, V.Y. Lim, D. Herndler-Brandstetter, E. Nevius, T. Sugiyama, S. 8 29 Tani-Ichi, S. Schlenner, E. Richie, H.-R. Rodewald, R.A. Flavell, T. Nagasawa, K. Ikuta, and 8 30 J.P. Pereira. 2016. Hematopoietic Stem Cell Niches Produce Lineage-Instructive Signals to 8 31 Control Multipotent Progenitor Differentiation. Immunity. 45:1219–1231. 8 32 doi:10.1016/j.immuni.2016.11.004. 8 33 Ding, L., and S.J. Morrison. 2013. Haematopoietic stem cells and early lymphoid progenitors 8 34 occupy distinct bone marrow niches. Nature. 495:231–235. doi:10.1038/nature11885. 8 35 Dobin, A., C.A. Davis, F. Schlesinger, J. Drenkow, C. Zaleski, S. Jha, P. Batut, M. Chaisson, 8 36 and T.R. Gingeras. 2013. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 29:15–8 37 21. doi:10.1093/bioinformatics/bts635. 8 38 Dupard, S.J., A. Grigoryan, S. Farhat, D.L. Coutu, and P.E. Bourgine. 2020. Development of 8 39 Humanized Ossicles: Bridging the Hematopoietic Gap. Trends Mol Med . 26:552–569. 8 40 doi:10.1016/j.molmed.2020.01.016. 8 41 Dykstra, B., D. Kent, M. Bowie, L. McCaffrey, M. Hamilton, K. Lyons, S.-J. Lee, R. Brinkman, 8 42 and C. Eaves. 2007. Long-term propagation of di stinct hematopoietic differentiation 8 43 programs in vivo. Cell Stem Cell. 1:218–229. doi:10.1016/j.stem.2007.05.015. 8 44 Eisele, A.S., J. Cosgrove, A. Magniez, E. Tubeuf, S. Tenreira Bento, C. Conrad, F. Cayrac, 8 45 T. Tak, A.-M. Lyne, J. Urbanus, and L. Perié. 2022. Erythropoietin directly remodels the 8 46 clonal composition of murine hematopoietic multipotent progenitor cells. Elife. 11:e66922. 8 47 doi:10.7554/eLife.66922. 8 48 Ergen, A.V., N.C. Boles, and M.A. Goodell. 2012. Rantes/Ccl5 influences hematopoietic 8 49 stem cell subtypes and causes myeloid skewing. Blood. 119:2500–2509. doi:10.1182/blood-8 50 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 5 2011-11-391730. 8 51 Fritsch, K., S. Pigeot, X. Feng, P.E. Bourgine, T. Schroeder, I. Martin, M.G. Manz, and H. 8 52 Takizawa. 2018. Engineered humanized bone organs maintain human hematopoiesis in vivo. 8 53 Exp Hematol. 61:45-51.e5. doi:10.1016/j.exphem.2018.01.004. 8 54 Grigoryan, A., D. Zacharaki, A. Balhuizen, C.R. Côme, A.G. Garcia, D. Hidalgo Gil, A.-K. 8 55 Frank, K. Aaltonen, A. Mañas, J. Esfandyari, P. Kjellman, E. Englund, C. Rodriguez, W. 8 56 Sime, R. Massoumi, N. Kalantari, S. Prithiviraj, Y. Li, S.J. Dupard, H. Isaksson, C.D. Madsen, 8 57 B.T. Porse, D. Bexell, and P.E. Bourgine. 2022. Engineering human mini-bones for the 8 58 standardized modeling of healthy hematopoiesis, leukemia, and solid tumor metastasis. Sci 8 59 Transl Med. 14:eabm6391. doi:10.1126/scitranslmed.abm6391. 8 60 Groen, R.W.J., W.A. Noort, R.A. Raymakers, H.-J. Prins, L. Aalders, F.M. Hofhuis, P. 8 61 Moerer, J.F. van Velzen, A.C. Bloem, B. van Kessel, H. Rozemuller, E. van Binsbergen, A. 8 62 Buijs, H. Yuan, J.D. de Bruijn, M. de Weers, P.W.H.I. Parren, J.J. Schuringa, H.M. Lokhorst, 8 63 T. Mutis, and A.C.M. Martens. 2012. Recons tructing the human hematopoietic niche in 8 64 immunodeficient mice: opportunities for studying primary multiple myeloma. Blood. 120:e9–8 65 e16. doi:10.1182/blood-2012-03-414920. 8 66 Guidi, N., M. Sacma, L. Ständker, K. Soller, G. Marka, K. Eiwen, J.M. Weiss, F. Kirchhoff, T. 8 67 Weil, J.A. Cancelas, M.C. Florian, and H. Geiger. 2017. Osteopontin attenuates aging-8 68 associated phenotypes of hematopoietic stem cells. EMBO J . 36:1463. 8 69 doi:10.15252/embj.201796968. 8 70 Hamilton, N.E., and M. Ferry. 2018. ggtern/i6 : Ternary Diagrams Using ggplot2. J. Stat. Soft. 8 71 87. doi:10.18637/jss.v087.c03. 8 72 Hao, Y., S. Hao, E. Andersen-Nissen, W.M. Mauck, S. Zheng, A. Butler, M.J. Lee, A.J. Wilk, 8 73 C. Darby, M. Zager, P. Hoffman, M. Stoeckius, E. Papalexi, E.P. Mimitou, J. Jain, A. 8 74 Srivastava, T. Stuart, L.M. Fleming, B. Yeung, A.J. Rogers, J.M. McElrath, C.A. Blish, R. 8 75 Gottardo, P. Smibert, and R. Satija. 2021. Integr ated analysis of multimodal single-cell data. 8 76 Cell. 184:3573-3587.e29. doi:10.1016/j.cell.2021.04.048. 8 77 Hay, S.B., K. Ferchen, K. Chetal, H.L. Grimes, and N. Salomonis. 2018a. The Human Cell 8 78 Atlas bone marrow single-cell interactive web portal. Exp. Hematol. 68:51–61. 8 79 doi:10.1016/j.exphem.2018.09.004. 8 80 Hay, S.B., K. Ferchen, K. Chetal, H.L. Grimes, and N. Salomonis. 2018b. The Human Cell 8 81 Atlas bone marrow single-cell interactive web portal. Exp Hematol . 68:51–61. 8 82 doi:10.1016/j.exphem.2018.09.004. 8 83 Henry, E., I. Souissi-Sahraoui, M. Deynoux, A. Lefèvre, V. Barroca, A. Campalans, V. 8 84 Ménard, J. Calvo, F. Pflumio, and M.-L. Arcangeli. 2020. Human hematopoietic 8 85 stem/progenitor cells display reactive oxygen species-dependent long-term hematopoietic 8 86 defects after exposure to low doses of ionizing radiations. Haematologica. 105:2044–2055. 8 87 doi:10.3324/haematol.2019.226936. 8 88 Holzapfel, B.M., D.W. Hutmacher, B. Nowlan, V. Barbier, L. Thibaudeau, C. Theodoropoulos, 8 89 J.D. Hooper, D. Loessner, J.A. Clements, P.J. Russell, A.R. Pettit, I.G. Winkler, and J.-P. 8 90 Levesque. 2015. Tissue engineered humanized bone supports human hematopoiesis in vivo. 8 91 Biomaterials. 61:103–114. doi:10.1016/j.biomaterials.2015.04.057. 8 92 Itkin, T., S. Gur-Cohen, J.A. Spencer, A. Schajnovitz, S.K. Ramasamy, A.P. Kusumbe, G. 8 93 Ledergor, Y. Jung, I. Milo, M.G. Poulos, A. Kalinkovich, A. Ludin, O. Kollet, G. Shakhar, J.M. 8 94 Butler, S. Rafii, R.H. Adams, D.T. Scadden, C.P. Lin, and T. Lapidot. 2016. Distinct bone 8 95 marrow blood vessels differentially regulate haematopoiesis. Nature. 532:323–328. 8 96 doi:10.1038/nature17624. 8 97 Kolde, R. 2012. Pheatmap: pretty heatmaps. R package version. 1:726. 8 98 Lambers, F.M., F. Stuker, C. Weigt, G. Kuhn, K. Koch, F.A. Schulte, J. Ripoll, M. Rudin, and 8 99 R. Müller. 2013. Longitudinal in vivo imaging of bone formation and resorption using 9 00 fluorescence molecular tomography. Bone. 52:587–595. doi:10.1016/j.bone.2012.11.001. 9 01 Luecken, M.D., and F.J. Theis. 2019. Current bes t practices in single-cell RNA-seq analysis: 9 02 a tutorial. Mol Syst Biol. 15:e8746. doi:10.15252/msb.20188746. 9 03 Maijenburg, M.W., M. Kleijer, K. Vermeul, E.P.J. Mul, F.P.J. van Alphen, C.E. van der 9 04 Schoot, and C. Voermans. 2012. The composition of the mesenchymal stromal cell 9 05 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 6 compartment in human bone marrow changes during development and aging. 9 06 Haematologica. 97:179–183. doi:10.3324/haematol.2011.047753. 9 07 Martine, L.C., B.M. Holzapfel, J.A. McGovern, F. Wagner, V.M. Quent, P. Hesami, F.M. 9 08 Wunner, C. Vaquette, E.M. De-Juan-Pardo, T.D. Brown, B. Nowlan, D.J. Wu, C.O. 9 09 Hutmacher, D. Moi, T. Oussenko, E. Piccinini, P.W. Zandstra, R. Mazzieri, J.-P. Lévesque, 9 10 P.D. Dalton, A.V. Taubenberger, and D.W. Hutmacher. 2017. Engineering a humanized bone 9 11 organ model in mice to study bone metastases. Nat Protoc . 12:639–663. 9 12 doi:10.1038/nprot.2017.002. 9 13 Miura, T., T. Kouno, M. Takano, T. Kuroda, Y. Yamamoto, S. Kusakawa, M.S. Morioka, T. 9 14 Sugawara, T. Hirai, S. Yasuda, R. Sawada, S. Matsuyama, H. Kawaji, T. Kasukawa, M. Itoh, 9 15 A. Matsuyama, J.W. Shin, A. Umezawa, J. Kawai, and Y. Sato. 2023. Single-Cell RNA-Seq 9 16 Reveals LRRC75A-Expressing Cell Population Involved in VEGF Secretion of Multipotent 9 17 Mesenchymal Stromal/Stem Cells Under Ischemia. Stem Cells Transl Med . 12:379–390. 9 18 doi:10.1093/stcltm/szad029. 9 19 Passaro, D., A. Abarrategi, K. Foster, L. Ariza-McNaughton, and D. Bonnet. 2017. 9 20 Bioengineering of Humanized Bone Marrow Microenvironments in Mouse and Their 9 21 Visualization by Live Imaging. J Vis Exp. 55914. doi:10.3791/55914. 9 22 Pflumio, F., B. Izac, A. Katz, L.D. Shultz, W. Vainchenker, and L. Coulombel. 1996. 9 23 Phenotype and function of human hematopoietic cells engrafting immune-deficient CB17-9 24 severe combined immunodeficiency mice and nonobese diabetic-severe combined 9 25 immunodeficiency mice after transplantation of human cord blood mononuclear cells. Blood. 9 26 88:3731–3740. 9 27 Pinho, S., T. Marchand, E. Yang, Q. Wei, C. Nerlov, and P.S. Frenette. 2018. Lineage-9 28 Biased Hematopoietic Stem Cells Are Regulated by Distinct Niches. Dev Cell . 44:634-9 29 641.e4. doi:10.1016/j.devcel.2018.01.016. 9 30 Ramasamy, S.K., A.P. Kusumbe, T. Itkin, S. Gur-Cohen, T. Lapidot, and R.H. Adams. 2016. 9 31 Regulation of Hematopoiesis and Osteogenesis by Blood Vessel-Derived Signals. Annu Rev 9 32 Cell Dev Biol. 32:649–675. doi:10.1146/annurev-cellbio-111315-124936. 9 33 Reinisch, A., N. Etchart, D. Thomas, N.A. Hofmann, M. Fruehwirth, S. Sinha, C.K. Chan, K. 9 34 Senarath-Yapa, E.-Y. Seo, T. Wearda, U.F. Hartwig, C. Beham-Schmid, S. Trajanoski, Q. 9 35 Lin, W. Wagner, C. Dullin, F. Alves, M. Andreeff, I.L. Weissman, M.T. Longaker, K. 9 36 Schallmoser, R. Majeti, and D. Strunk. 2015. Ep igenetic and in vivo comparison of diverse 9 37 MSC sources reveals an endochondral signature for human hematopoietic niche formation. 9 38 Blood. 125:249–260. doi:10.1182/blood-2014-04-572255. 9 39 Reinisch, A., D.C. Hernandez, K. Schallmoser, and R. Majeti. 2017a. Generation and use of 9 40 a humanized bone-marrow-ossicle niche for hematopoietic xenotransplantation into mice. 9 41 Nat Protoc. 12:2169–2188. doi:10.1038/nprot.2017.088. 9 42 Reinisch, A., D.C. Hernandez, K. Schallmoser, and R. Majeti. 2017b. Generation and use of 9 43 a humanized bone-marrow-ossicle niche for hematopoietic xenotransplantation into mice. 9 44 Nat Protoc. 12:2169–2188. doi:10.1038/nprot.2017.088. 9 45 Reinisch, A., D. Thomas, M.R. Corces, X. Zhang, D. Gratzinger, W.-J. Hong, K. Schallmoser, 9 46 D. Strunk, and R. Majeti. 2016. A humanized bone marrow ossicle xenotransplantation 9 47 model enables improved engraftment of healthy and leukemic human hematopoietic cells. 9 48 Nat Med. 22:812–821. doi:10.1038/nm.4103. 9 49 Sun, W., M. Perkins, M. Huyghe, M.M. Faraldo, S. Fre, L. Perié, and A.-M. Lyne. 2024. 9 50 Extracting, filtering and simulating cellular barcodes using CellBarcode tools. Nat Comput 9 51 Sci. 1–16. doi:10.1038/s43588-024-00595-7. 9 52 Thibaudeau, L., V.M. Quent, B.M. Holzapfel, A.V. Taubenberger, M. Straub, and D.W. 9 53 Hutmacher. 2014. Mimicking breast cancer-induced bone metastasis in vivo: current 9 54 transplantation models and advanced humanized strategies. Cancer Metastasis Rev . 9 55 33:721–735. doi:10.1007/s10555-014-9499-z. 9 56 Tikhonova, A.N., I. Dolgalev, H. Hu, K.K. Sivaraj, E. Hoxha, Á. Cuesta-Domínguez, S. Pinho, 9 57 I. Akhmetzyanova, J. Gao, M. Witkowski, M. Guillamot, M.C. Gutkin, Y. Zhang, C. Marier, C. 9 58 Diefenbach, S. Kousteni, A. Heguy, H. Zhong, D.R. Fooksman, J.M. Butler, A. Economides, 9 59 P.S. Frenette, R.H. Adams, R. Satija, A. Tsirigos, and I. Aifantis. 2019. The bone marrow 9 60 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 7 microenvironment at single-cell resolution. Nature. 569:222–228. doi:10.1038/s41586-019-9 61 1104-8. 9 62 Vaiselbuh, S.R., M. Edelman, J.M. Lipton, and J.M. Liu. 2010. Ectopic human mesenchymal 9 63 stem cell-coated scaffolds in NOD/SCID mice: an in vivo model of the leukemia niche. Tissue 9 64 Eng Part C Methods. 16:1523–1531. doi:10.1089/ten.tec.2010.0179. 9 65 Wickham, H. 2011. ggplot2. WIREs Computational Statistics . 3:180–185. 9 66 doi:10.1002/wics.147. 9 67 9 68 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 8 Figure legends 9 69 9 70 F igur e 1. H u man M SCs fro m fetal an d p os t -natal bo ne marr ow gen erat es hOs s . 9 71 A. Growth curve of hMSCs isolated fr om fetal b one ma rro w (BM) (5 hMSC s, age <12 po s t-9 72 c on c ept ion week s ( PC Ws ) , pa s s age 2) and pos t -natal BM (4 hMSCs, age s : 5 , 9, 24 and 5 1 9 73 y ear s, pas sa g e 2). Ea ch culture w as c arr ied out in quadruplicate. Th e mean numb er of c ells 9 74 for ea ch time point for every t es ted MSC is indicated; the med ian growth of fet al (in blue) 9 75 and post-na t al (in orange) MSCs are indic at ed. ns, not signific ant ; mult iple Ma nn-W hit ney 9 76 test. 9 77 B. Phenot ype analys i s of adher ent f e tal (F28 sample, 11.6 PC Ws, pas s a ge 2) and post-n a t a l 9 78 (ALL O3 sample, 5 year s old, p as sage 2) human BM (hB M) cells isolated from fetal and post-9 79 natal BM. Cells were gated on hCD4 5 - CD14 - CD 15 - CD31 - c ell s (>90%). Rep r es ent a t iv e of >5 9 80 tested hMSC sa mples . The results were obt a ined fr om t he gating strat egy s hown in 9 81 S u pFig1A . 9 82 C. Experimental pr otocol of hOs s gen er ation using immune-defic ient N S G mic e. h MSC s wer e 9 83 isolated fro m human BM s amp les ob t ained be for e (fe tal, F) a n d afte r (post-nat al, P-N ) bir t h 9 84 and a mplified in plas tic wells (pass a ges ≤2), befor e mix ing 2.10 6 cells wit h 300µl c old-9 85 matrigel c on taining hu man plat ele t lys ate ( 50µl) and hB MP7 (5µ g/ 5 µl) and sub cutaneous 9 86 transplantat ion in mic e. Eight week s lat er, hOss were harvested and analyzed. 9 87 D. Repr es ent ative pic t ures of hOss ge nerated from f eta l (F/ hO s s, F28, 1 1.6 PCWs , passa g e 1) 9 88 and post- natal ( P-N/h Oss , AL LO2, 9 years old, pa ssa ge 1) hMS Cs . h Os s were recovered 8 9 89 w eek s after impla ntation in mice. 9 90 E. W ei ght compar is on of F/hOss a nd P-N/ h Oss. The re sults are s h own f rom F/hOs s (9 9 91 in dependent ex per iments, 11 hMSC samples; 50 h Oss) a n d P-N/ h Os s ( 4 independent 9 92 ex per imen ts, 5 hMSC samples ; 20 h Os s ) . Each do t r epresents a s ingle h Os s, and the line 9 93 indic at es the media n value. * ***, p< 0.0001 , Mann-Whitney test. 9 94 F . Flow c y t ometr y analy si s of mur ine c ells re covered fro m hO s s. P er c e n tage of c ells in 9 95 S SC/ FS C ga t ed viable cells ar e sho wn. Mu rine red blood cells ar e Ter 119 + , and mur ine 9 96 hematopoie t ic c ells are Ter119-/ mCD45 + . Results were obtained from F/ hOs s (4 9 97 in dependent experimen t s , 8 hM SC samples; 25 h Os s ) and P - N /h Oss (3 independent 9 98 ex per imen ts, 4 hM SC s amples; 16 h Os s ) . L ines indi cate t he median value of each condit io n. 9 99 *, p<0.05; ****, p<0.0001 ; Krus kal-Wallis t es t w it hout corr ection. 1 000 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 2 9 G . Representat iv e s ect ions of F/ hOss (left) and P-N/ hOss (right) with h istological staining 1 001 us ing HE S (upper panels) and Safr a nin O / Fast Green (lower panels ) . HL , hemato poietic 1 002 lodge; V, vessel; B, bone; Ad, adipoc y tes. Amplif ication x10. 1 003 1 004 F igur e 2. H u man fet a l h Os s and p os t -natal hOs s su ppor t nor mal human hematop oies is. 1 005 A. Exper imental des ign of xenogr afts of umbilical cord blood CD34 + cells in mice c ar rying 1 006 hOs s. Umbilic al c or d bloo d CD34 + c e l l s ( 1 0 5 cells /mou se) wer e injected int o mice c ar rying 1 007 hOs s and non- hOs s contro ls , les s tha n 24 hours aft er sublethal ir rad iatio n, 2-8 week s after 1 008 hOs s implantation. Human hema t op oiesis wa s analy zed 12 week s af ter umbilical cord bloo d 1 009 HSPC injection into t he bone marr ow of mice (mBM) and the hOss. 1 010 B . Um b i l i c a l c o r d b l o o d C D 3 4 + in ject ed in F/ h Oss at 12 week s wit h his t olog y immunolabeling 1 011 of s ections: human hemato poietic cells (h CD45), myeloid cells (hMPO, CD1 4), 1 012 megak ar yocyt es (CD6 1) , r ed cells (Glycophor in C ) , imma t ure HSPC s (CD34) . Amplific at io n, 1 013 x 10. 1 014 C. A r epresentative flow c ytometry a naly s is of the human hematopoietic cells develo ping in 1 015 mBM (upper panel) and F/hOs s (lower panel). Analys i s at 12 week s post-injection o f 1 016 umbilic al cord blood CD34 + cells. 1 017 D. Percentage of human G PA + cells in mice with and without hOss. Ce lls were ga t ed o n 1 018 F SC/ SS C following t he gating strat egy in C. E ach dot repr es ent s ind ividual mice ( n=21 mice, 1 019 w it hout hOs s in bla c k a n d g r a y from 2 indep endent experim ent s ; n =25 mic e, with F/h Oss, 4 7 1 020 hOs s obtained from 5 diff erent MSCs; n=28 mi ce with P- N/hOss , 51 h Os s obtained f rom 4 1 021 differ ent MSCs). Gra y and black dots: mBM fro m mice without hOs s ; blue d ots: F/hOs s; 1 022 orange dots: P- N/ h Os s . 1 023 E. Per c entage of cells expres s ing hCD45 + us ing the gating s t rategy s ho wn in C for mBM 1 024 (pooled 4 long bones) and h Oss (2/mous e, treat ed separately) 1 2 week s post-t ransplant o f 1 025 umbilic al cor d blood CD34 + cells. The results repr es ent the same mice as those displayed in 1 026 D. 1 027 F - G . Percenta ge of hu man (F) l ymp hoid CD19 + B-cells and ( G ) myeloid CD14 + /CD 15 + c e l l s . 1 028 Cells were gated on hCD45 + as de s cribed in C. hBM c ells from healt hy donors (age <18 year s, 1 029 n=9) were included as cont rols. The r esults represent t he s ame mice as th ose dis pla yed in D-1 030 E. The g at in g str ateg y is shown in S up F ig 2D and Figur e 2 C. 1 031 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 3 0 H-I. Per c entage of CD34 + ( H ) and CD34 + CD90 + HSCs (I) gated on hCD45 + Li n - (CD19 - CD14 - CD 15 -1 032 ) cells , ba sed on t he gating strat e g y des cribed in C . hBM cells from healthy donor s (age <18 1 033 y ear s, n = 9 ) were included a s c on t rols. The results repr es en t the sa me mice a s those 1 034 dis played in D- E. 1 035 ns , no statistical dif ference; *, p<0. 05; **, p<0. 001; ****, p<0. 0001; K ruskal-Wallis tes t 1 036 w it hout corr ection. 1 037 F igur e 3. H u man prim ary hOss co mp rise hM SC s c a pable of s eco n dary h O s s de velopm ent. 1 038 A. Experiment a l design of sec on dar y h Os s f ormat io n a n d grafts of CD 34 + c e l l s i n m i c e 1 039 c ar ryin g s e condar y h Os s . Se condar y hMSC s wer e isolated after crushing pr imary hOss, 1 040 ex pand ed a s adher ent cells and tr an splant ed as indicated in Figure 1C. Ei g ht w eeks later, 10 5 1 041 umbilic al cor d blood CD34 + c ells were in ject ed into mic e car rying s e cond ar y hOs s and non-1 042 hOs s contr ols les s than 24 hours af ter sublethal (2 Gy/ mous e) ir radiation. Human 1 043 hematopoie t ic dev elopment was ana lyzed 12 week s la t er in t he bone ma r row ( BM) o f mice 1 044 and hOs s. 1 045 B. Flow c ytometry analy s i s of se c o nd ary hMS C s i s o lat ed from primar y F/ a nd P- N/ h Os s . Cells 1 046 w er e gated on hCD45 - CD 1 4 - CD 15 - CD 31 - c ells (>90%). Rep resentative of > 4 tes t ed sec ond a r y 1 047 hMSC samples . The results were obt a ined f ro m t he ga t ing s t rategy s ho w n in SupFig1A. 1 048 C. Representat iv e examples of se con dary h Oss obtained 10- 12 wee k s po st-injection o f 1 049 s e condary hMS C s. The image s di s p lay the results for t he samples F28 ( f etal hMSC or igin , 1 050 11.6 P CWs) and ALLO3 ( pos t -natal hMSC o rigin, 5 year s old). 1 051 D. Weight comp a r is on of secondary F/hOss and P- N /h Os s. The r es ult s wer e obta ined from 1 052 18 hOs s (individual blue dots, media n value: b lack lin e) generated f r om secondary F/hMSC s, 1 053 derived f ro m 2 MSC s sa mples , in 3 independent experimen ts , and obtained from 8 hOss 1 054 (individ ua l red d ots, median value : blac k line) gener ated fr om secondary P- N /hMSC s , 1 055 derived from 1 hMSC sa mple in 2 independent exper imen ts . 1 056 E. Per centage of mur ine cells analy z ed by f low cyt ometr y , recovered from s e condar y h Oss 1 057 w it hout injection of umbilical c o rd blood CD34 + c e l l s . G a t i n g w a s c a r r i e d o u t o n v i a b l e 1 058 muri ne red bl ood cel ls, Ter119 + , and via b le mur ine he mat opoietic cells (in Ter119 - ) , mCD 45 + . 1 059 The results ar e shown f or 1 2 h Oss gener a t ed with 1 secondar y F/hMS C in 2 independent 1 060 ex per imen ts and 2 hOss from 1 se con dary P- N/ hMS C in 1 ex per iment. 1 061 F . Percenta ge of human G PA + e r y t h r o i d c e l l s i n m B M f r om m i c e w i t h a n d w i t h o u t s e co n da r y 1 062 (II) hOs s . The gating str ategy i s the s ame as in F igur e 2C. The result s shown r epresent t he 1 063 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 3 1 in div idual va lues of mBM from 21 mice without hOs s ( gra y dots: the same mice as di splayed 1 064 in Fig u re 2D), 8 h Oss gener ated by 2 se condary F /hMSC s in 4 mice (d ark blue dot s ), 6 hOss 1 065 ge n erated by 1 se condary P- N /hMSC in 3 mic e (dar k r ed do ts ) , 6 mBM f rom mice wit h 1 066 s e condary F/h Os s (light blue dot s ) an d 3 mBM fr om th e mic e wit h s e condary P- N /h Os s (pink 1 067 dots). Black lines indi c at e the median values . 1 068 G . Percentage of hCD 45 + cells as se ss ed via flow c yt ometry in se c on dary hOss at 12 week s 1 069 pos t -tr ans plant of umbil ical cord blo od CD34 + cells . Th e results repr es ent the same mice as 1 070 those dis played in F. The ga t ing s tr ategy i s the same a s t hat shown in F igur e 2C. 1 071 H-I. Percentage of human (H) lymph oid B -c ells (C D19 + ) and ( I) myeloid c ells (CD14 + /C D15 + ) 1 072 ga t ed on hCD45+ c ells ac c o rding t o the s t rategy shown in Figur e 2C. Th e results repr es ent 1 073 the sa me mic e as those di s played in F -G . 1 074 J-K. Percentage of (J) CD3 4 + and (K) CD34 + C D 9 0 + HSC s gated on hC D4 5 + Lin - (CD19 - CD 1 4 -1 075 CD15 - ) cells , ba s ed on the same ga ting strat egy a s th a t shown in F igu r e 2C. The results 1 076 represent t he s ame mice as t hose dis played in F - G . 1 077 *, p<0.05; **, p <0. 01; ***, p<0.0 01; **** , p<0.0 001; F-K. Kru skal- W allis te s t without 1 078 c or rection. D: Mann-Whitney t es t. 1 079 1 080 F igur e 4. Human pr imary hOss disp la y enhan ced lev els of f unctional immat u re hum an 1 081 cells. 1 082 A. Experimental design of the func tion a l analysis of the imma t ure cell comp artment in 1 083 primary hOs s compared wit h mu rine bone ma r r ow (mBM) fr o m m ice with and without hOs s. 1 084 CD34 + cells recovered from crushed hOs s and mBM 12 week s aft er C D34 + cell tr ans plant 1 085 w er e analyzed by f low c ytometr y and eithe r plated in met hylc ellulose semi-solid medium 1 086 (1x10 3 hCD34 + cells /plate) or r e -injected into s econdar y NSG mi c e (3- 5.10 5 hCD34 + 1 087 c ells/ mo u s e). Two wee k s lat er, colonies wer e scored and char act eriz ed as granuloc ytic (CF U -1 088 G ), monoc ytic (CFU - M), gr a n ulo-monocyti c (CFU- G M), er yt hroid ( BFU -E) or mult ipo ten t 1 089 (CFU - GEMM). Human hemato poies is was anal y z ed in the BM of sec onda r y mouse recipients 1 090 12 w eek s pos t -injection. 1 091 B. Tot al number of colonies generated by 3, 000 plated CD34 + cells at 2 weeks of cultur e, 1 092 isol at ed f r om dif f erent B M si t es ( mBM wi t h out hOss , gr a y dot s; F/h O s s, bl ue dot s; P - N/ h O ss, 1 093 orange dots; mBM wit h F/h Oss, gr e en dot s ; mBM with P-N/ h Os s , pink do ts ) . The result s 1 094 w er e obt ained from 3 independen t experimen t s and each sample was tested in tr iplic at e. 1 095 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 3 2 The individual mean o f tr ip lic ates f or each t es t ed s ample is s hown; black lines indicate t he 1 096 median v alues. ns, not signif icant ; Krus ka l- W allis t es t wit hout corr ec t ion. 1 097 C. Tot al number of c olon ies per colo ny type generat ed by 3, 000 plated CD34 + c e l l s a t 2 1 098 w eek s of s emi- s olid cult ures (mBM with hO s s, gray dot s; F /hOs s , blue d ots; and P- N /hOss, 1 099 orange dots ) . Bla ck lines indi c at e the median values . ns , not significant; Krus k al-Wallis te s t 1 100 w it hout corr ection. 1 101 D. Repr es ent ative ex amples of FACS plots of hC D45 + cells detected in t he mBM of NSG mice 1 102 in j ect ed wit h cell s f r o m mBM of mi ce wi t h hOs s ( pr i mar y) an d hOs s ( pr i mar y) . Th e cel l s wer e 1 103 ga t ed on FSC/SS C paramet ers. 1 104 E. U pper panel: rat io of s econd a r y mic e en g r afted with >0.1% of hCD 45+ cells on all in jec t ed 1 105 mic e. The a s soc iat e d pr oport io n wa s calc ulat ed . Lower panel: per centage of hCD45 + c e l l s 1 106 analyzed in the BM o f NS G mic e inj e ct ed with c ells fr om mB M mic e w it hout (7 mice) and 1 107 w it h prima ry hOss (15 mice) and f r om pr i mary hOs s (18 mice). Black lines in d ic at e t he 1 108 median v alues. Krus kal-Wa llis test without cor rection. 1 109 F . Relative percentage of B ly mphoid (CD19 + ) and myelo id ( C D14 + /CD15 + ) cells from th e 1 110 pooled mice s h own in E. Only t he mice with reliable engr a f tmen t (≥0.1% hCD45 + cells ) wer e 1 111 analyzed. Bla ck lines indicat e the me dian values . 1 112 G . Repr es en tative ex amp les of F ACS plots of hCD45 + cells detected in the mBM o f NS G mice 1 113 in jec t ed wit h cells fr om m BM of mice with hOs s ( secondar y) and h Os s ( s e c ondary). The c ells 1 114 w er e gated on FSC/ S SC par a meters. 1 115 H. Upper panel: r atio of secondary mice engraft ed with >0.1% of h CD45 + c ells on all inject ed 1 116 mic e. The a s soc iat e d pr oport io n wa s calc ulat ed . Lower panel: per centage of hCD45 + c e l l s 1 117 a n a l y z e d i n t h e B M of N S G m i ce i n j e c t e d w i t h ce l l s f r o m se c o n da r y h Os s ( 9 m i c e , ) a n d m B M 1 118 of t he same mice (with hOss , 9 mic e). Blac k lines indicate t he me d ian values . *, p<0.0 5; 1 119 Mann-Whitney test. 1 120 F . Relative percent a ge of B-lymp hoid ( CD19 + ) and my eloid (CD14 + /CD1 5 + ) c ells from t he 1 121 pooled m ice s ho w n in H. Only the mice with r eliable engr aftmen t (≥0.1% hCD45 + cells ) wer e 1 122 analyzed. Bla ck lines indicat e the me dian values . 1 123 1 124 Fi g u r e 5 . Single cell a n alysis show s th at im matur e cells recov er ed f ro m hO s s d i splay an 1 125 enhanced bias t o ward m yelo/ er ythr oid progen itors and inc reased H SC l e vels. 1 126 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 3 3 A. Exper i ment al desi gn. Thr ee m ont hs af t er in j ect io n of 10 5 um bi li cal cord bl oo d CD34 + cel l s , 1 127 mononuclear cells wer e isolated f rom mB M and hOSS and pooled ac c o rding to or igin. CD34 + 1 128 c ells were then s or ted with a pu r ity > 70% and viability >90%. Sorted cells were then u s ed t o 1 129 prepare single-cell RN A- s equenc ing librar ies . 1 130 B. U MAP r epresentat io n of t he 13 clus t ers identified v ia Seur at. Each clus t er is repr es ent ed 1 131 by a diff e rent color . Clus t ers w er e an notat ed by compar ing th e gene exp re s s ion pr ofiles wit h 1 132 those of the hematopo iet ic populations de scribed b y Hay et al. ( Hay et al., 2018a) (HSC = 1 133 hematopoie t ic stem c ell, MPP = multipotent pr ogenito r, L MPP = lymphoid- primed mu lti-1 134 potent ia l progenit or, MDP = monocy t e/ dend ritic cell pr ogenitor, CLP = c ommon lymph oid 1 135 progenito r, Mult i- L in = multi- lin ea ge pr ogenitor, p r e-PC = pre - plas ma cells , Eo-B- Mast = 1 136 eos ino- bas o- mas t cell s). 1 137 C. Anno tation o f th e 13 clus t ers a cc ordin g to gene markers as previously d escribed (Hay et 1 138 al., 2018 b; Ba ccin et al., 2020). The color code is shown in the f igur e leg end. 1 139 D. C omparison of populatio n distrib ut ion (absolute nu mber s ) between the mB M and hOss 1 140 c ompartme n ts. Each bar corresponds to the tot al number o f cells in a given sub-populat io n. 1 141 The sum o f each bar corr es po nds to the t ota l n umber of c ells in our d ataset. Ea ch b a r i s 1 142 div ided in t o 3 s egment s corr es ponding to the or iginal samples. 1 143 E. C omparison of the pr op or tion o f 3 major progenitor groups (HSC/ MPP / LMP P, B cell 1 144 precursor s/progenitor s and myeloid pr ec u rsor/progenitor cells ) in the mBM and hOss 1 145 c on ditions. Propor tions were calculated wit hin eac h s ample using the data fr om E. "B 1 146 Precur sor/Progenit or" inclu de CD34 + p r e - P C , C D 3 4 + p r e - B cy c l i n g , C D 3 4 + M ultilin/CLP, CD3 4 + 1 147 pro-B , CD34 + pre -B, follicular B cells. " M y eloid pr ec ursor/progenitor" cont ain immatu r e 1 148 neutro phils /monoc ytes , neut rophils, CD34 + MDP/pre-dendrit ic cells, CD34 + Eo- B-Mas t , 1 149 CD34 + early eryt hroblast s, erythr oblas t s . Ear ly HSPC ar e c o ntained in C D3 4 + HSC/ MPP/L M PP 1 150 c ells. The prop ortions were comp a r ed u s ing a χ2 t es t . For ea ch t es t , * *p < 0.01, ***p < 1 151 0.001, ****p < 0 .001. 1 152 F - G . D i f f e r e n t i a l l y e x p r e s s e d g e n e s b e t w e e n p o s t - n a t a l h O s s ( F ) o r f e t a l h O s s ( G ) a n d m B M 1 153 c ells within the CD34 + HS C / MP P/ L MPP compart ment. Ea c h volcanop lo t indicates t he 1 154 ex pr es sion of ge n es that were s i gnificantly downregulated (p < 0 .05, r ed dot s ) or 1 155 upregulated (blue dot s) in hOss comp a r ed w it h mBM. Genes that were not significantly 1 156 underexpressed or overexpressed are shown in gr een. 1 157 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint 3 4 H. Ven n Diagram s howing t he 47 common upregulated genes in the CD3 4 + HSC / MP P/L MPP 1 158 c ompartme nt f rom P-N/ hOss (blue) and F/hOss (red), both compared with mBM. 1 159 I. Venn diagram showin g 9 out o f the 47 gene li st ( green) from H iden tified among the 50 1 160 mos t upr eg u lat ed genes, P-N/ h Os s c ompar ed with mBM ( blue), and F - hO s s compared wit h 1 161 mBM (r ed). 1 162 F igur e 6 Cellular ba rcoding shows th at hOss s u p port mor e balanced human h ematop oietic 1 163 dev elopment . 1 164 A. Ex per imental des ign. Inje ction of 10 5 umbilic al cord blood CD34 + c e l l s w a s do n e a f t e r 1 165 barcode-tr ans d uc t ion. Th ree months later , mononucleated c ells were iso lat ed f rom mBM 1 166 and hOss and pooled ac cording t o or igin. Tran sdu ced (GFP+) B, M y eloid ( M) and CD34 + c e l l s 1 167 w er e then sort ed, lyzed and barcode s we r e amplif ied by PCR for identification and analys i s. 1 168 B. N umber of u nique ba r c od e s ident ified pe r mous e in eac h condit ion. ns, not significant; t-1 169 test. 1 170 C. Ternar y plot of th e fraction of c ells produced per barcode clone in each of the 3 cell type s 1 171 (CD34 + , B-cells, and my eloid (M) cell s). Ea c h dot represents a dis tinct ba rcode. The size of 1 172 the d ot in dic ates the n umber of c ells or iginating f rom eac h barcoded p rogenit or in ject ed 1 173 (clone s ize). 1 174 D-E. Number of reads per barcode clones det ec t ed in the hOss and b one marr ow for F/hOss 1 175 (D) and P- N /h O s s (E) in the va r ious cell types. Ea c h dot repr es ent s a dist inc t barcode. The 1 176 red lin e rep resents the point w h e re barcodes wou ld have the same c lon e siz e in bot h 1 177 organs. The blue b ox es c ont ain t he commo n bar c od es / c lones and t he r ed bo x es cont ain 1 178 barcod es/clone s spec if ic to mBM and hOs s . The ax i s i s lo g10 t ransformed of t he 1 179 renor malized r ead count+1. Th e thr e shold f or lin eage conten t is 1%, and the min imal clo ne 1 180 s ize is 10 cells. 1 181 1 182 1 183 1 184 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint Ter119+ mCD45+ in Ter119- 0 20 40 60 80 100Percentage of cells Fig 1f v3 sans OA v13 mBM F/hOss P-N/hOss ✱ ✱ ✱✱✱✱E F Figure 1 A B C x10 G HES F/hOss P-N/hOss Safranine O/ Fast green V V V V B B BAd Ad Ad HL HL F/hOss P-N/hOss Cold matrigel BM MSC isolation and amplification 2.106 hMSC Adherence step Subcutaneous injection 8 weeks Analysis of hOss composition 4 weeks PTH F/hOss P-N/hOss 0 100 200 300 500 1000 1500 Weight (mg) Fig1f_v7 without ✱✱✱✱ Fetal Post-Natal Event count related to % CD90 CD73 CD105 CD44 Primary hMSC 5 mm 0 1 2 3 4 5 6 7 0 1×105 2×105 3×105 4×105 5×105 Days Number of viable cells Week 1 (P2) : proliferation of MSC Fetal hBM MSC Post-Natal hBM MSC D ns was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint mBM W/o hOss F/hOssP-N/hOss hBM 0 2 4 6 8 10 12 14% hCD34+/hCD90+ cells Lin-/CD34+/CD90+ ✱✱✱ ✱✱✱ ✱ mBM W/o hOss F/hOssP-N/hOss hBM 0 20 40 60 80% hCD34+ cells Lin-/CD34+ ns mBM W/o hOss F/hOssP-N/hOss 0 20 40 60 80% hGPA+ cells Erythroid cells ✱ mBM W/o hOss F/hOssP-N/hOss hBM 0 20 40 60 80 100% hCD19+ cells B cells ✱✱ ✱✱✱✱✱ ✱✱✱✱ ✱✱ ✱✱✱✱ mBM W/o hOss F/hOssP-N/hOss 0 20 40 60 80 100% hCD45+ cells hCD45+ ns Figure 2 A B Human Hematopoiesis F/hOss or P-N/hOss UCB CD34+ cells mouse Bone Marrow (mBM W/o hOss) mBM F/hOss or P-N/hOss 12 weeks UCB CD34+ cells D E F G H I hMPO hCD14 hCD61 Glycophorin C hCD34 hCD45 x10 C SSC FSC mBM F/hOss hOss SSC hCD45 CD14+15 CD19 GPA hCD45 mBM CD34 CD90 in hCD45+ cells F/hOss in hCD45+Lin- cells mBM W/o hOss F/hOssP-N/hOss hBM 0 20 40 60 80 100 % hCD14 +15 + cells Myeloid cells ✱✱✱✱ ✱✱✱✱ ✱✱✱✱ ✱ ✱✱ was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint mBM W/o hOss F/hOss IIP-N/hOss II mBM F/hOss IImBM P-N/hOss II 0 2 4 6 8 Prog Lin-/CD34+/CD90+ cells % hCD34+/hCD90+ cells ✱✱✱ ✱✱ ✱ ✱✱ mBM W/o hOss F/hOss IIP-N/hOss II mBM F/hOss IImBM P-N/hOss II 0 20 40 60 80 100 Myeloid Cells % hCD14+15+ cells ✱✱✱✱✱ ✱✱✱ Ter119+ mCD45+ in Ter119- 0 20 40 60 80 100Percentage of cells Fig 3d Murin IIaire v13 mBM F/hOss II P-N/hOss II ✱✱ ✱✱ mBM W/o hOss F/hOss IIP-N/hOss II mBM F/hOss IImBM P-N/hOss II 0 10 20 30 40 Erythroid cells % hGPA+ cells ✱✱✱ ✱✱✱✱✱ mBM W/o hOss F/hOss IIP-N/hOss II mBM F/hOss IImBM P-N/hOss II 0 20 40 60 80 Prog CD34+ cells % hCD34+ cells ✱✱ ✱ ✱ ✱✱ mBM W/o hOss F/hOss IIP-N/hOss II mBM F/hOss IImBM P-N/hOss II 0 20 40 60 80 100 B cells % hCD19+ cells ✱✱ ✱✱✱✱✱ mBM W/o hOss F/hOss IIP-N/hOss II mBM F/hOss IImBM P-N/hOss II 0 20 40 60 80 100 hCD45+ % hCD45+ cells ✱ ✱ ✱ Cold matrigel B Figure 3 A Primary hOss engrafted mice hMSC isolation and amplification 2.106 secondary hMSC NSG mouse hOss crushed Subcutaneous Injection hCD34+ cells 8 weeks 12 weeks Human Hematopoisesis Secondary F/hOss Secondary P-N/hOss D E F G H I J K Fetal Post-Natal Count CD90 CD73 CD105 CD44 C F/hOss IIP-N/hOss II 0 100 200 300 400 Fig3b Weight (mg) ✱✱✱✱ Secondary hMSC Secondary hOss 5 mm 4 weeks PTH was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted April 11, 2024. ; https://doi.org/10.1101/2024.04.08.588553doi: bioRxiv preprint mBM W/o hOssmBM W/ hOss hOss 0.01 0.1 1 10% hCD45+ cells 0.0506 0.0591 mBM w/hOss hOss 0.01 0.1 1 10% hCD45+ cells Hemto II IIaire✱ mBM W/o hOss F/hOssP-N/hOssmBM F/hOssmBM P-N/hOss 0 200 400 600 800 Number of colonies per 3000 CD34+ cells ns mBMhOssmBMhOss 0 20 40 60 80 100% of cells Pheno Hemato IIaire B cellsM cells Figure 4 B C D E A F PrimaryNSG mice SecondaryNSG mice12 wksHuman Hematopoiesis Colonies CFU assay 14 days G F/hOssP-N/hOss mBMW/o hOss mBMF/hOssP-N/hOss 12 wks 12 wks mBM W/o hOss F/hOssP-N/hOss 0 50 100 150 200 250BFU-E Number of colonies per 3000 CD34+ cells nsBFU-E mBM W/o hOss F/hOssP-N/hOss 0 50 100 150Number of colonies per 3000 CD34+ cells CFU-Mns CFU-M mBM W/o hOss F/hOssP-N/hOss 0 100 200 300 400Number of colonies per 3000 CD34+ cells CFU-GnsCFU-G mBM W/o hOss F/hOssP-N/hOss 0 10 20 30 40 50 CFU-GM Number of colonies per 3000 CD34+ cells nsCFU-GM mBM W/o hOss F/hOssP-N/hOss 0 5 10 15Number of colonies per 3000 CD34+ cells CFU-GEMMnsCFU-GEMM mBMW/o hOssF/hOssP-N/hOssmBMF/hOssmBMP-N/hOss Injectedcellsfrom F/hOssIIP-N/hOssIImBMF/hOssIImBMP-N/hOssII mBMhOssmBMhOss 0 20 40 60 80 100% of cells Pheno Hemato II IIaire B cellsM cells H I +/total (>0.1%)2/79/159/18(%)29%60%50%Primary hOss mBMw/ primary hOss Donorcellsfrom SSChCD45 Secondary hOss SSChCD45 mBMw/ secondary hOss Donorcellsfrom+/total (>0.1%)7/99/9(%)78%100% Injectedcellsfrom Figure 5B D E A F/hOss: 48% UCB CD34+ cells mBM: 60% 12 weeks W/o hOssN=2 W/ F/hOssN=1 W/ P-N/hOssN=2P-N/hOss: 60% CellrecoverySortingof hCD34+ mBM: 89% F/hOss: 91% P-N/hOss: 77% C F 298 genesUP 20 genesDOWN 3313 genesDOWN 1130 genesUP Upregulatedgenesin P-N/hOssvs mBM P-N/hOssvs mBMF/hOssvs mBM Common genes(fromH) P-N/hOssvs mBMF/hOssvs mBM G H I LRRC75A 3188 cells 4005 cells 2701 cells %CD34+%CD45+ HSC/MPP/LMPP50 mostupregulatedgenesUpregulatedgenesin F/hOssvs mBM 020406080100 HSC/MPP/LMPP Myeloid Precursor/progenitor B Precursor/Progenitor MO vs hOss color % of total number of cell mBM W/o hOssF/hOssP-N/hOss ✱✱✱✱✱✱✱✱ ✱✱✱✱✱✱✱✱ ✱✱✱✱✱ Figure 6A B C D PooledF/hOss UCB CD34+ cellstransducedwithbarcodes Pool mBMW/o hOss 12 weeks MicewithouthOss MicewithF/hOss MicewithP-N/hOss Pool mBMF/hOss PooledP-N/hOss Pool mBMP-N/hOss CellsrecoverySortingcells B cells(B) Myeloid(M) Progenitor(CD34) W/o hOssW/ F/hOssW/ P-N/hOss E Clone Size Output bias Output bias MOI = 1F/hOssP-N/hOss mBMP-N/hOss mBMF/hOss