AA2P-mediated DNA demethylation synergizes with stem cell agonists to promote expansion of hematopoietic stem cells Authors: Javed K. Manesia, Harinad B. Maganti, Sakhar Almoflehi, Suria Jahan, Tanvir Hasan, Roya Pasha, Chelsea McGregor, Nellie Dumont, Josée Laganière, Julie Audet, Nicolas Pineault Published: 2023 ABSTRACT SummarySmall molecules have enabled expansion of hematopoietic stem and progenitor cells (HSPCs), but limited knowledge is available on whether these agonists can act synergistically. In this work, we identify a stem cell agonist in AA2P and optimize a series of stem cell agonist cocktails (SCACs) to help promote robust expansion of human HSPCs. We find that SCACs provide strong growth-promoting activities while promoting retention and function of immature HSPC. We show that AA2P-mediated HSPC expansion is driven through DNA demethylation leading to enhanced expression of AXL and GAS6. Further, we demonstrate that GAS6 enhances the serial engraftment activity of HSPCs and show that the GAS6/AXL pathway is critical for robust HSPC expansion. Introduction Hematopoietic stem cell (HSC) transplantation is a curative therapy for over 30 hematological diseases. Umbilical cord blood (CB) is an important source of allogeneic cells for patients with complex and rare human leukocyte antigen (HLA) haplotypes. However, CB grafts are associated with lower cell dose that leads to slower platelet and leukocyte engraftment. Ex vivo expansion has emerged as a promising strategy to facilitate timely and durable engraftment from a single CB unit with favorable outcomes.1,2,3 Moreover, natural and synthetic small molecules led to breakthroughs in cell culture engineering for the expansion of HSCs. Interestingly, some of these “stem cell agonists” promote HSC and progenitor expansion through different biochemical pathways, raising the possibility that their combination may elicit greater amplification. The original aim of this study was to develop stem cell agonist cocktails (SCACs) to maximize HSC and hematopoietic stem and progenitor cell (HSPC) expansion. Even though progress was made with HSC expansion protocols, many approaches failed to favor self-renewal over differentiation, resulting in the outgrowth of more committed progenitors.4,5 However, recent studies identified small molecules that synergize with early-acting cytokines to promote expansion of HSPCs that reflect both molecular and functional HSC properties. These include StemRegenin1 (SR1), an aryl hydrocarbon receptor (AHR) antagonist, and UM171, a pyrimidoindole derivative that acts by maintaining the balance between pro- and anti-inflammatory pathways and by suppressing LSD1/RCOR1 suppressor complex via proteosomal degradation leading to the preservation of important epigenetic marks within HSPCs.6,7,8,9,10 Chromatin-modifying agents (e.g., 5-aza-2-deoxycytidine) were among the first molecules reported to promote HSC expansion via the transcriptional regulation of core HSC genes.11,12 Valproic acid (VPA) is an epigenetic modifier that inhibits histone deacetylases (HDACs) with strong stem cell agonist properties,11,12 and a phase I clinical trial is currently underway with VPA-expanded CB units (NCT03885947). The lysine methyltransferase inhibitor UNC0638 (specific inhibitor to G9a13 and GLP) was shown to promote the maintenance of CD34+CD38−CD45RA− cells and increase 8-week SRCs by 2-fold, or 5-fold when combined with SR1.14 Nicotinamide is another small molecule that has been reported to increase expansion of HSCs.15 Nicotinamide and VPA also prime HSPCs for bone marrow (BM) homing via the upregulation of CXCR4.11,15 To date, UM171, SR1, and nicotinamide have proved beneficial in clinical trials aiming to accelerate engraftment through HSC expansion technologies.1,2,16,17 L-ascorbic acid 2-phosphate (AA2P) is a phosphorylated form of L-ascorbic acid. AA2P is more resistant to oxidation and is a stable alternative to ascorbic acid in culture. Ascorbic acid and N-acetyl cysteine have both been reported for their anti-oxidant properties.18 Interestingly, ascorbic acid was recently reported to regulate HSC function in vivo.19 Ascorbic acid was found to be a natural metabolite highly enriched in HSCs and multipotent progenitors (MPPs). Functionally, ascorbic acid was found to regulate HSC activity and suppress leukemogenesis by acting as a cofactor for the ten-eleven-translocation (TET) 2 enzyme. TETs are chromatin-modifying enzymes that catalyze the oxidation of cytosine residues, which precedes DNA demethylation.19,20 TET2 is frequently mutated within hematopoietic malignancies, and TET2 mutations are often associated with DNA hypermethylation.21 Treatment with ascorbic acid was shown to enhance TET2 function, resulting in formation of 5-hydroxymethylcytosine and suppression of leukemia progression in a mouse model.19,20 It is unclear if AA2P or ascorbic acid would have any regulatory roles on HSCs in culture. Herein, we investigated whether AA2P could act as a stem cell agonist for the expansion of umbilical CB HSCs. Another HSC expansion platform of interest is the fed-batch culture system developed to minimize the negative impacts of paracrine factors on HSPC expansion22 and also shown to complement UM171-expansion cultures.1,7 However, most studies to date have focused on the use of a single small molecule to promote HSC amplification. The few studies that combined small molecules did not characterize the impact of interaction effects between small molecules on HSPC expansions.7,23 In this study, we used advanced experimental design methods to screen for individual and interactive (synergistic or antagonist) activities of small molecules on CB HSCs to develop a series of SCACs to maximize HSC expansion. Serial and limiting dilution assays (LDAs) demonstrated superior engraftment activity for SCAC grafts produced with the lead SCAC X2A. We also highlight that the expansion of HSCs with X2A is driven in part by the growth arrest-specific factor 6 (GAS6) acting through its receptor AXL, previously implicated in the self-renewal of leukemic stem cells (LSCs).24,25 Results AA2P promotes the expansion of CB HSPCs We tested whether AA2P could act as a stem cell agonist given the role of ascorbic acid in HSC regulation in mice in vivo.19,20 CB CD34+ cells were cultured with increasing doses of AA2P in serum-free medium (SFM) supplemented with stem cell factor (SCF), FL and TPO (complete medium referred to as STFL). AA2P had no significant impact after 7 days of culture but raised expansions of total nucleated cells (TNCs) and CD34+CD38−CD45RA− HSPCs up to 3-fold by day14 (Figures 1A and 1B), a property shared by ascorbic acid (Figures 1C and S1A). Expansion of CD34+CD38−CD45RA− cells was maximal at 500 μM of AA2P, which was used in subsequent analyses (Figure 1B). AA2P also supported a significant increase in the total number of colony-forming unit (CFU) progenitors (Figure 1D). Phenotypic analyses revealed that AA2P promoted the development of most progenitor subsets (Figures 1E and 1F) with a 2.5-fold increase in the number of CD34+CD38−CD45RA+CD90− lymphoid-primed multipotent progenitor cells (LMPPs; Figure 1F).Figure 1AA2P supports the growth of CB HSCPs in vitro and improves their engraftment(A and B) CB CD34+ cells were cultured with vehicle or with AA2P at different concentrations for 14 days. Mean ± standard error of the mean (SEM) presented for total nucleated cells (TNCs) and CD34+CD45RA−CD38− cell expansions. Two-way ANOVA, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; n = 3.(C) Impact of vitamin C (Vit-C) on CD34+CD38−CD45RA− expansion after 14 days of culture (mean ± SEM, n = 3, two-way ANOVA, ∗∗p < 0.01, ∗∗∗∗p < 0.0001).(D) Net number of CFU produced per starting cells in indicated cultures (mean ± SEM, n = 4). Paired t test, ∗p < 0.05.(E) Net production of CD34+CD45RA−CD38−CD90+ and CD34+CD45RA−CD38−CD90+CD49F+ per starting cells treated with AA2P for 14 days (mean ± SEM, n = 4).(F) Net production per starting cells of hematopoietic progenitor subsets. MPP, CD34+CD38−CD45RA−CD90−; LMPP, CD34+CD38−CD45RA+CD90−; CMP, CD34+CD38+CD45RA−CD123+; MEP, CD34+CD38+CD45RA−CD123−; and GMP, CD34+CD38+CD45RA+CD123+. Paired t test, ∗p < 0.05, mean ± SEM, n = 4.(G) Mean fluorescence intensity (MFI) ± SEM of CXCR4 on cell subsets produced in presence of AA2P or not (day 14 presented, n=4).(H) Migration of AA2P expanded subsets toward recombinant SDF-1a after 4-h incubation (mean ± SEM, n = 4).(I) Levels of human platelet and leukocyte engraftment in peripheral blood at 3, 10, and 18 weeks after transplantation. NOD scid gamma (NSG) mice were transplanted with the progeny of 5,000 cells cultured in presence or absence of AA2P for 14 days. Each symbol corresponds to an individual mouse, and the horizontal line presents median. Five to seven mice per group per experiment, n=3. Mann-Whitney test; ∗p < 0.05, ∗∗p < 0.01.(J) Levels of human BM engraftment in humanized NSG mice at 22–24 weeks post transplantation. Two-way ANOVA, ∗∗p < 0.01. Each symbol corresponds to an individual mouse, and the horizontal line presents median. Five to seven mice per group per experiment, n = 3.(K) Long-term human engraftment in NSG mice transplanted with the progeny of 5,000 CD34+ cells expanded in the presence or absence of Vit-C or AA2P for 14 days. Percentage of CD45+ BM cells 22 weeks post transplant (bars represent median of four or five mice per group, one-way ANOVA, ∗p < 0.05). AA2P supports the growth of CB HSCPs in vitro and improves their engraftment (A and B) CB CD34+ cells were cultured with vehicle or with AA2P at different concentrations for 14 days. Mean ± standard error of the mean (SEM) presented for total nucleated cells (TNCs) and CD34+CD45RA−CD38− cell expansions. Two-way ANOVA, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; n = 3. (C) Impact of vitamin C (Vit-C) on CD34+CD38−CD45RA− expansion after 14 days of culture (mean ± SEM, n = 3, two-way ANOVA, ∗∗p < 0.01, ∗∗∗∗p < 0.0001). (D) Net number of CFU produced per starting cells in indicated cultures (mean ± SEM, n = 4). Paired t test, ∗p < 0.05. (E) Net production of CD34+CD45RA−CD38−CD90+ and CD34+CD45RA−CD38−CD90+CD49F+ per starting cells treated with AA2P for 14 days (mean ± SEM, n = 4). (F) Net production per starting cells of hematopoietic progenitor subsets. MPP, CD34+CD38−CD45RA−CD90−; LMPP, CD34+CD38−CD45RA+CD90−; CMP, CD34+CD38+CD45RA−CD123+; MEP, CD34+CD38+CD45RA−CD123−; and GMP, CD34+CD38+CD45RA+CD123+. Paired t test, ∗p < 0.05, mean ± SEM, n = 4. (G) Mean fluorescence intensity (MFI) ± SEM of CXCR4 on cell subsets produced in presence of AA2P or not (day 14 presented, n=4). (H) Migration of AA2P expanded subsets toward recombinant SDF-1a after 4-h incubation (mean ± SEM, n = 4). (I) Levels of human platelet and leukocyte engraftment in peripheral blood at 3, 10, and 18 weeks after transplantation. NOD scid gamma (NSG) mice were transplanted with the progeny of 5,000 cells cultured in presence or absence of AA2P for 14 days. Each symbol corresponds to an individual mouse, and the horizontal line presents median. Five to seven mice per group per experiment, n=3. Mann-Whitney test; ∗p < 0.05, ∗∗p < 0.01. (J) Levels of human BM engraftment in humanized NSG mice at 22–24 weeks post transplantation. Two-way ANOVA, ∗∗p < 0.01. Each symbol corresponds to an individual mouse, and the horizontal line presents median. Five to seven mice per group per experiment, n = 3. (K) Long-term human engraftment in NSG mice transplanted with the progeny of 5,000 CD34+ cells expanded in the presence or absence of Vit-C or AA2P for 14 days. Percentage of CD45+ BM cells 22 weeks post transplant (bars represent median of four or five mice per group, one-way ANOVA, ∗p < 0.05). Previous studies reported that cytokine-based culture can impair engraftment due to homing defects.26,27 To investigate whether AA2P could influence migration properties of expanded HSPCs, we measured the expression of CXCR4 and the chemotaxis activity of CB by flow cytometry and an SDF-1 transwell migration assay, respectively. AA2P treatment slightly increased CXCR4 expression on CD34+ cells (Figure 1G) but insufficiently to affect the migratory activity of HSPC subsets toward SDF-1 (Figure 1H). Next, we assessed the impact of AA2P on the engraftment activities of expanded grafts. AA2P-expanded HSPCs provided higher levels of human platelets at all time points with the greatest increase measured at week 18 (>4-fold, Figure 1I, platelet analysis Figure 4A). However, no differences were observed in leukocyte multilineage engraftment (Figure 1I, data not shown). Long-term BM engraftment investigated at week 22–24 revealed normal multilineage differentiation but superior human CD45+ chimerisms in AA2P recipients (Figure 1J). Ascorbic acid also provided a slight improvement in human engraftment (Figures 1K, S1B, and S1C). Taken together, these results suggest that AA2P may act as a stem cell agonist. Modeling of small molecules identifies key stem cell agonists and their optimal doses We set out to optimize a series of SCACs using design of experiments. We identified 11 small molecules that had been reported to possess stem cell agonist activity or properties that may promote stem cell expansion. SR1, UM171, VPA, AA2P, UNC0638, trichostatin A, nicotinamide, BIRB796, Ly2228820, GSH, NSC23005, and DEAB were initially screened individually in STFL medium to identify the most potent molecules for the expansion of CD34+ subsets. SR1, UM171, VPA, AA2P, and UNC0638 provided the best expansion of CD34+CD38−CD45RA− cells (data not shown). Nicotinamide was not retained as it only weakly promoted expansion of CD34+CD38−CD45RA− cells (+40% over control, data not shown).3,16 A two-level full factorial design experiment was then applied to characterize their individual and interactive activities (under-additive and supra-additive) of the five factors on the expansion of CB cells (i.e., TNCs), progenitor- (CD34+ cells) and HSC-enriched subsets identified as CD34+CD45RA− and CD34+CD45RA−EPCRHigh.28 This screen consisted of all combinations at two levels (present or not) and four replicated center points. The measured cell expansion data for each cell subset were used to fit a linear model that included block effects (i.e., independent experiment), main factor effects, and factor interactions (up to five-factor interactions). The results of these analyses are presented in volcano plots (Figure 2A) where dominant positive effects are on the top right quadrant (vertical axis; false discovery rate [FDR] log worth statistical significance) and the dominant negative effects on the top left.29Figure 2Individual and interaction effects of stem cell agonists determined by full 25 factorial design(A) Individual and interactive effects of SR1, UM171, VPA, UNC0638, and AA2P on the expansion of TNC, CD34+, CD34+CD45RA−, and EPCRHigh are presented as volcano plots. The effects are depicted on x axis, while statistical significance as FDR log worth is represented on y axis. Interactive effects are presented as factor∗factor in legend. Non-significant activity not presented to simplify result presentation (n = 3).(B) Correlation between CCD RS models prediction and experimental data. A series of 60 conditions were modeled and then tested experimentally in independent cultures (coefficient of correlation R2 provided, n = 3). Scatterplot shows experimental cell expansion values and the values predicated by RS model for TNC, CD34+, CD34+CD45RA−, and EPCRHigh subsets.(C) Composition of the SCAC and their growth-promoting attributes. SM, surface model. Individual and interaction effects of stem cell agonists determined by full 25 factorial design (A) Individual and interactive effects of SR1, UM171, VPA, UNC0638, and AA2P on the expansion of TNC, CD34+, CD34+CD45RA−, and EPCRHigh are presented as volcano plots. The effects are depicted on x axis, while statistical significance as FDR log worth is represented on y axis. Interactive effects are presented as factor∗factor in legend. Non-significant activity not presented to simplify result presentation (n = 3). (B) Correlation between CCD RS models prediction and experimental data. A series of 60 conditions were modeled and then tested experimentally in independent cultures (coefficient of correlation R2 provided, n = 3). Scatterplot shows experimental cell expansion values and the values predicated by RS model for TNC, CD34+, CD34+CD45RA−, and EPCRHigh subsets. (C) Composition of the SCAC and their growth-promoting attributes. SM, surface model. The aryl AHR antagonist SR1, the HDAC inhibitors VPA, and the suppressor of LSD1/RCOR1 complex UM171 were consistently the most potent factors for the expansion of all cell populations, except for TNC, for which UM171 had a strong negative effect (Figure 2A). Of note, UM171 and VPA had under-additive effects on all populations (except on TNC), as indicated by their significant negative interactions (UM171∗VPA). The lysine methyltransferase inhibitor UNC0638 had a weak, borderline significant effect (either a positive or negative effect) on all cell CD34+ populations but had a stronger, more significant positive effect on TNC expansion. Similarly, AA2P consistently had a weak, borderline significant positive effect on all cell populations except for EPCRHigh cells (Figure 2A). The negligible effects included most of the factor interactions (two-, three- and four-factor interactions). These results supported the selection of SR1, UM171, VPA, and AA2P for the next investigation aimed at identifying optimal concentrations. The SCAC X2A and SM6 promote robust ex vivo expansion of CD34+CD45RA− and EPCRHigh cells Diagnostic tests on the model derived from the two-level factorial designs indicated severe curvature in the dose-response curves of the factors. Therefore, the identification of optimal concentrations required another series of experiments with a larger number of doses. Specifically, factor combinations that corresponded to a central composite design (CCD) were tested. The CCD included five different doses and wider concentration range (Table S1A). The cell expansion data for each subset of interest were used to fit a quadratic model (Equation 1), which made it possible to explore in detail the dose-response relationship and interactions of SR1, UM171, VPA, and AA2P. Diagnostic tests on the model fit suggested that there was a high consistency between the experimental cell expansion values and those predicted by the CCD models (Figure 2B), although there was also significant lack of fit for combinations that had the highest UM171 doses (stars signs; Figure 2B). Nonetheless, the models were used in numerical simulations to predict the doses of the four factors that, in combination (i.e., SCACs; Figure 2C) would maximize the expansion of TNC and the three HSC and progenitor subsets (CD34+, CD34+CD45RA−, and EPCRHigh cells; Figure 3). The response surface (RS) model for CD34+CD45RA− cells predicted that a condition consisting of SR1, VPA, and AA2P at highest dose with lowest UM171 dose would maximize expansion (surface model A [SMA]; Figure 2C). Conversely, the EPCRHigh RS model predicted maximal expansion with SR1, UM171, and AA2P at highest concentration and VPA at lowest (SM2; Figure 2C). In agreement with the two RS model predictions, independent experiments revealed that SMA and SM2 showed remarkably high levels of CD34+CD45RA− and EPCRHigh expansion, respectively (Figures 3B, 3E, and 3F).Figure 3Impact of SCAC on the expansion of HSC- and progenitor-enriched subsets(A) Net fold expansion of TNCs in indicated cultures.(B) Heatmap representation of the expansion of the four cell subpopulations tracked in indicated cultures.(C) Representative flow cytometry analysis of ex vivo expanded CB HSPCs showing the three HSPC subsets tracked (CD34+, CD34+CD45RA−, and EPCRHigh) in indicated cultures.(D–F) Frequency of indicated HSPC subsets (left graphs) and expansion of HSPC subsets (right graphs) in cultures. Expansions of CB subsets were calculated based on respective population on day 0. Control STFL cultures consisted of the same complete media without small molecules. For (A) and (D–F), mean ± SEM presented, one-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; n = 10. Impact of SCAC on the expansion of HSC- and progenitor-enriched subsets (A) Net fold expansion of TNCs in indicated cultures. (B) Heatmap representation of the expansion of the four cell subpopulations tracked in indicated cultures. (C) Representative flow cytometry analysis of ex vivo expanded CB HSPCs showing the three HSPC subsets tracked (CD34+, CD34+CD45RA−, and EPCRHigh) in indicated cultures. (D–F) Frequency of indicated HSPC subsets (left graphs) and expansion of HSPC subsets (right graphs) in cultures. Expansions of CB subsets were calculated based on respective population on day 0. Control STFL cultures consisted of the same complete media without small molecules. For (A) and (D–F), mean ± SEM presented, one-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; n = 10. The cell expansion data in response to combinations that contained the highest UM171 doses seemed associated with the highest experimental cell expansion values. Consequently, to further explore this region of the parameter space, additional SCACs were created by modifying the top factor combinations identified using the model and these were tested in subsequent experiments. Moreover, even though SM2 improved EPCRHigh HSC-enriched cell expansion, the concentrations of SR1, UM171, and AA2P were very high, which may not be clinically relevant. Therefore, additional experiments also included combinations with different doses of SR1 (1,000–5,000 nM), UM171 (38–62 nM), and AA2P (250–2,000 μM). This made it possible to identify the SCACs X2A and SM6 (Figure 2C). X2A supported a similar level of expansion of EPCRHigh cells to SM2 but provided significantly superior expansion of TNC, CD34+, and CD34+CD45RA− cells when compared to SM2 (Figures 3A–3F). X2A also matched the maximal expansion of CD34+CD45RA− cells (Figure 3E). Conversely, SM6 was found to maximize expansion of CD34+CD45RA− cells but, like SMA, it poorly supported the expansion of EPCRHigh cells due to the lower dose of UM171 and higher dose of VPA (Figures 3E and 3F). In addition, both X2A and SM6 supported the highest level of cell growth (Figure 3A) and nearly doubled CD34+ cell expansion over other SCACs (Figure 3D). Finally, the functional contribution of AA2P to stem and progenitor expansion was further investigated in the context of SCACs. Lowering AA2P concentration to non-functional concentration (i.e., the lowest range) resulted in a 50% reduction in the expansion of TNC, CD34+, and EPCRHigh cells, confirming AA2P as a stem cell agonist (X2B vs. X2A; Figures 3A–3E). This activity was not the result of greater magnesium (Mg) concentration in X2A coming from AA2P, as similar Mg concentration supplemented in X2B had no significant impact on cell growth (data not shown). The growth-promoting attributes of X2A, X2B, and SM6 on CB CD34+ cells were confirmed by an external collaborator (J.L., Héma-Québec, Innovation department) with X2A and SM6 maximizing expansion of TNCs, CD34+ cells, and CD34+CD45RA− HSPCs (Figure S2A). Last, we confirmed that expansion of CB CD34+ cells with X2A provided significant improvement in the expansion of most subsets tracked over single-agonist expansion platforms (Figure S2B). Grafts of SCACs X2A and SM6 show superior short- and long-term engraftment Prompt and durable engraftment are critical for the well-being of patients. We used the serial transplantation assay to investigate the engraftment properties of grafts produced by the five lead SCACs (Figure 2C). Each SCAC induces a specific pattern of cell growth and expansion of CD34 subsets, but it is unclear which would provide superior engraftment. Engraftment was first tracked by measuring human platelets and leukocytes. X2A-expanded CB HSPC provided superior short-term platelet and leukocyte engraftment compared to other SCAC grafts (Figure 4A). The low level of human platelet chimerism was in line with the low cell doses transplanted herein and previous studies.30,31,32 Reducing AA2P to non-functional level in X2B resulted in a significant decline in engraftment activity (Figures 4A and 4B). Mouse cohorts transplanted with X2A and SM6 grafts exhibited the highest level of human CD45+ leukocyte chimerism at week 16 compared to other SCAC-expanded grafts (Figure 4B; middle panel). Conversely, SMA grafts showed low short-term engraftment that progressively increased over time (Figures 4A and 4B).Figure 4Engraftment properties of SCAC-expanded HSPC grafts(A and B) Representative human platelet and leukocyte analyses. Levels of human platelet (top) and proportion of human CD45+ leukocytes (bottom) in indicated humanized NSG mice groups. The progeny of 1,250 CD34+ cells cultured in presence of SCACs for 14 days were transplanted per mouse. Peripheral blood analyses were done at week 3 (A) and 16 (B) post transplantation.(C and D) Multilineage lymphomyeloid engraftment analysis of human BM cells in serial humanized transplants. Primary (C) and secondary (D) mice were analyzed 18 and 12 weeks after transplantation, respectively.For (A) to (D), each symbol corresponds to an individual mouse and the horizontal line presents median, 12–18 mice total per condition (n = 2–3). Ordinary ANOVA or Kruskal-Wallis one-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Engraftment properties of SCAC-expanded HSPC grafts (A and B) Representative human platelet and leukocyte analyses. Levels of human platelet (top) and proportion of human CD45+ leukocytes (bottom) in indicated humanized NSG mice groups. The progeny of 1,250 CD34+ cells cultured in presence of SCACs for 14 days were transplanted per mouse. Peripheral blood analyses were done at week 3 (A) and 16 (B) post transplantation. (C and D) Multilineage lymphomyeloid engraftment analysis of human BM cells in serial humanized transplants. Primary (C) and secondary (D) mice were analyzed 18 and 12 weeks after transplantation, respectively. For (A) to (D), each symbol corresponds to an individual mouse and the horizontal line presents median, 12–18 mice total per condition (n = 2–3). Ordinary ANOVA or Kruskal-Wallis one-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. BM analysis at 18 weeks revealed that X2A and SM6 grafts provided superior engraftment and all grafts retained multilineage lympho-myeloid engraftment activities (Figure 4C). Mice transplanted with grafts from X2B (X2A with minimal AA2P) showed significantly diminished human engraftment compared to mice infused with X2A grafts (Figure 4C). However, SM2 grafts provided the lowest levels of engraftment (Figures 4A–4C). The self-renewal potentials of the primary grafts were tested in secondary transplantation with BM engraftment analysis done at week 12 (Figure 4D; Table S4). Multilineage secondary engraftment was superior within cohorts transplanted with grafts from X2A, SM6, and SMA (Figure 4D). Taken together, these results demonstrate that, while all SCACs supported serial multilineage engraftment, primary grafts produced with the SCACs X2A and SM6 provided superior engraftment. Expansion with the SCAC X2A enhances the engraftment activity of CB grafts Culture of CB HSCs in presence of growth factors was reported to reduce their short-term engraftment activity.33 To determine whether expansion with X2A could have a similar effect, we compared the engraftment properties of fresh non-cultured (NC) CD34+ cells to that of their progeny expanded with X2A. X2A was selected as lead SCAC since it provided better short-term engraftment of platelets and leukocytes than the others. Platelet and leukocyte engraftments were significantly superior at week 3 in mice transplanted with X2A grafts (>2-fold, p < 0.05; Figure 5A) and a similar non-significant trend was observed at week 16 (Figure 5B). Moreover, long-term BM chimerism was increased in X2A-expanded recipients over the NC counterpart in primary and secondary recipients with normal multilineage engraftment (Figures 5C and 5D).Figure 5Expansion of CB CD34+ cells with X2A enhances the engraftment of CB grafts(A and B) Levels of human platelet (left) and proportion of human leukocytes (right) in NSG mice transplanted with 1,250 fresh/non-cultured CD34+ (NC) or their progeny produced in the presence of X2A for 14 days. Chimerism was measured at week 3 (A) and 16 (B) post transplantation. Each symbol corresponds to an individual mouse, and the horizontal line presents median; 11 mice total per cohort (n = 2). ∗p < 0.05, ∗∗∗ <0.001, ∗∗∗∗p < 0.0001; one-way and two-way ANOVA analysis for platelet levels and leucocyte frequencies, respectively.(C) Multilineage lymphomyeloid engraftment analysis of human BM cells in primary transplants. Mice were analyzed 18 weeks after transplantation (n = 2). ∗p < 0.05, ∗∗∗p < 0.001, two-way ANOVA.(D) Multilineage lymphomyeloid engraftment analysis of human BM cells in secondary transplants. Mice were analyzed 12 weeks after transplantation (n = 2).(E) LT-SRC frequency estimated by Poisson analysis. SRC frequencies and 95% confidence intervals estimated by LDA of fresh CD34+ or the progeny of cells cultured in presence of SCACs X2A and C6 for 14 days. Human BM chimerism was analyzed at 20 weeks after transplantation (n = 2 for SCACs, n = 1 NC [pooled from two donors], two to five mice per group per experiment).(F) Net number of SRCs in fresh CD34+ cells or in indicated cultures from 1,000 starting CD34+ cells. One-way ANOVA; ∗p < 0.05, ∗∗∗ <0.001, ∗∗∗∗p < 0.0001 (n = 2). Expansion of CB CD34+ cells with X2A enhances the engraftment of CB grafts (A and B) Levels of human platelet (left) and proportion of human leukocytes (right) in NSG mice transplanted with 1,250 fresh/non-cultured CD34+ (NC) or their progeny produced in the presence of X2A for 14 days. Chimerism was measured at week 3 (A) and 16 (B) post transplantation. Each symbol corresponds to an individual mouse, and the horizontal line presents median; 11 mice total per cohort (n = 2). ∗p < 0.05, ∗∗∗ <0.001, ∗∗∗∗p < 0.0001; one-way and two-way ANOVA analysis for platelet levels and leucocyte frequencies, respectively. (C) Multilineage lymphomyeloid engraftment analysis of human BM cells in primary transplants. Mice were analyzed 18 weeks after transplantation (n = 2). ∗p < 0.05, ∗∗∗p < 0.001, two-way ANOVA. (D) Multilineage lymphomyeloid engraftment analysis of human BM cells in secondary transplants. Mice were analyzed 12 weeks after transplantation (n = 2). (E) LT-SRC frequency estimated by Poisson analysis. SRC frequencies and 95% confidence intervals estimated by LDA of fresh CD34+ or the progeny of cells cultured in presence of SCACs X2A and C6 for 14 days. Human BM chimerism was analyzed at 20 weeks after transplantation (n = 2 for SCACs, n = 1 NC [pooled from two donors], two to five mice per group per experiment). (F) Net number of SRCs in fresh CD34+ cells or in indicated cultures from 1,000 starting CD34+ cells. One-way ANOVA; ∗p < 0.05, ∗∗∗ <0.001, ∗∗∗∗p < 0.0001 (n = 2). Next, LDA was used to measure the level of HSC expansion achieved with X2A. As a benchmark control, we used the previously reported combination of UM171 and SR1 shown to enhance expansion of LT-HSC,7 referred herein as condition 6 (C6; Figure 2C). The transplantation of increasing numbers of fresh CD34+ or the progeny of those cultured in presence of X2A and C6 resulted in varying levels of human chimerism. The LT-SRC frequency in X2A cultures and others is presented in Figures 5E and S2C. Notably, SRC frequencies of day-0 equivalent were 4-fold and 15-fold superior in X2A compared to C6 and fresh (NC) controls, respectively. When we calculated the net SRC number per 1,000 starting CD34+ cells, X2A showed significantly higher number of SRC unit compared to C6 and NC control (Figure 5F). The transcriptome profiles of fresh and SCACs expanded CD34+CD45RA− cells are distinct Next, we performed transcriptomic analyses of day 14 CD34+CD45RA− purified HSPCs from NC, STFL, C6, X2A, X2B, and SM6 cultures to explore the impact of SCACs on the transcriptional profiles of HSPCs and how they differ from the fresh HSPCs. Principal-component analysis (PCA) and hierarchical clustering revealed that the transcriptome of NC CD34+CD45RA− cells was very distinct from cultured cells (Figures 6A and 6B) with an average of ∼11,900 differentially expressed genes (DEGs) between NC replicates and the cultured conditions (Figure 6C). Within the cultured cells, the replicates from STFL condition clustered separately and there was an average of ∼2,800 DEGs between STFL replicates and the different SCACs (Figure 6D). Within the SCACs, replicates from SM6 and C6 clustered together, apart from X2A and X2B (Figure 6B). The majority of HSC cell-surface markers investigated were found expressed in all SCACs except for a few, such as KIT and PROM1/CD133, which were expressed at higher level in NC or STFL controls (Figure 6E).Figure 6Bulk RNA-seq reveals unique and overlapping gene expression programs in SCAC-expanded CD34+CD45RA− cells(A and B) PCA (A) and hierarchal clustering (B) identifies SCAC cultured CD34+CD4RA− cells to cluster close to each other compared to STFL and NC CD34+CD45RA− cells (n = 2).(C and D) Consistent within PCA and hierarchical clustering analysis, the differentially expressed gene (DEG) analysis comparing (C) SCACs vs. NC and (D) SCACs vs. STFL identifies a significant number of genes that are differentially upregulated and downregulated (mean, n = 2).(E–H) Heatmaps showing selected HSC cell-surface markers (E), erythroid markers (F), megakaryocyte markers (G), and genes implicated in AHR pathway (H) within the RNA-seq datasets (n = 2).(I–L) Enrichment map analysis of top GO terms identified within X2A vs. STFL (I), X2B vs. STFL (J), C6 vs. STFL (K), and SM6 vs. STFL (L) DEG comparisons from the CD34+CD45RA− RNA-seq datasets.(M) Pearson correlation analysis of known RTKs within the signal transduction GO term from (I) and (J) identified AXL and EGFR as some of the top outliers (n = 2).(N) RNA-seq traces display increased mRNA levels of AXL within X2A compared to X2B (showing data from one biological replicate).(O) Relative expression (RT-qPCR) of AXL, GAS6, and PROS1 normalized to STFL in sorted CD34+CD45RA− cells in culture supplemented with indicated SCACs. Two-way ANOVA, ∗p < 0.05, ∗∗ <0.01 (mean ± SD, n = 3).(P) GAS6 protein was detected in the supernatant of cultures and at much higher level in X2A cultures.(Q) CD34+CD45RA− cell subpopulations within SCAC X2A cell cultures at day 14 that were positive for AXL and pAXL were analyzed using intracellular image-flow cytometry. Representative images are shown.(R) Proportion of pAXL+ cells within the CD34+CD45RA− cell subpopulations in indicated cell cultures at day 14. One-way ANOVA; ∗∗∗∗p < 0.0001, mean ± SD, n = 3. Bulk RNA-seq reveals unique and overlapping gene expression programs in SCAC-expanded CD34+CD45RA− cells (A and B) PCA (A) and hierarchal clustering (B) identifies SCAC cultured CD34+CD4RA− cells to cluster close to each other compared to STFL and NC CD34+CD45RA− cells (n = 2). (C and D) Consistent within PCA and hierarchical clustering analysis, the differentially expressed gene (DEG) analysis comparing (C) SCACs vs. NC and (D) SCACs vs. STFL identifies a significant number of genes that are differentially upregulated and downregulated (mean, n = 2). (E–H) Heatmaps showing selected HSC cell-surface markers (E), erythroid markers (F), megakaryocyte markers (G), and genes implicated in AHR pathway (H) within the RNA-seq datasets (n = 2). (I–L) Enrichment map analysis of top GO terms identified within X2A vs. STFL (I), X2B vs. STFL (J), C6 vs. STFL (K), and SM6 vs. STFL (L) DEG comparisons from the CD34+CD45RA− RNA-seq datasets. (M) Pearson correlation analysis of known RTKs within the signal transduction GO term from (I) and (J) identified AXL and EGFR as some of the top outliers (n = 2). (N) RNA-seq traces display increased mRNA levels of AXL within X2A compared to X2B (showing data from one biological replicate). (O) Relative expression (RT-qPCR) of AXL, GAS6, and PROS1 normalized to STFL in sorted CD34+CD45RA− cells in culture supplemented with indicated SCACs. Two-way ANOVA, ∗p < 0.05, ∗∗ <0.01 (mean ± SD, n = 3). (P) GAS6 protein was detected in the supernatant of cultures and at much higher level in X2A cultures. (Q) CD34+CD45RA− cell subpopulations within SCAC X2A cell cultures at day 14 that were positive for AXL and pAXL were analyzed using intracellular image-flow cytometry. Representative images are shown. (R) Proportion of pAXL+ cells within the CD34+CD45RA− cell subpopulations in indicated cell cultures at day 14. One-way ANOVA; ∗∗∗∗p < 0.0001, mean ± SD, n = 3. The SCAC X2A and X2B repress differentiation programs Interestingly, a targeted lineage analysis revealed that many of the genes driving the clustering of replicates were involved in differentiation; genes associated with erythroid and megakaryocytic differentiation were downregulated in CD34+CD45RA− cells cultured with X2A and X2B but not with C6 and SM6 (Figures 6F and 6G). Conversely, the transcriptional profile of SCAC-expanded HSPC partially confirmed the mean of action of SR1 and UM171, such as downregulation of AHR signaling pathway genes (Figure 6H) in SR1 containing SCACs (C6, SM6, X2A, and X2B) and a dose-dependent increase in the expression of UM171-regulated genes such as PROCR/EPCR, CD86, and Ly96 (data not shown). Gene Ontology analysis identifies key pathways critical for CD34+CD45RA− cell expansion To elucidate the molecular programs driving the expansion of the CD34+CD45RA− HSPCs within the different SCACs, we performed Gene Ontology (GO) analysis using the DEGs. Enrichment map analysis revealed immune response, inflammatory response, and signal transduction as some of the top pathways that were affected in X2A and X2B HSPCs when compared to STFL (Figures 6I and 6J). In comparison, SRP-dependent co-translational, translation initiation, and inflammatory response were some of the top pathways in SM6 and C6 (Figures 6K and 6L). Inflammatory response has previously been reported to be critical for HSC expansion.8 Since, receptor tyrosine kinases (RTKs) are also potent regulators of stem cell fate decisions,34 we dissected DEGs listed under the GO term signal transduction with known RTKs (GSEA: Gene Set: Reactome signaling by RTK was used) between X2A and X2B, which identified three RTKs increased in X2A HSPCs, namely AXL, EGFR, and FLT3 (Figure 6M). Interestingly, AXL was identified to play a pivotal role in the self-renewal and expansion of LSCs.24,25 AXL was upregulated within all the SCACs but at the highest level within X2A HSPCs (Figures 6N and 6O). Moreover, the natural ligand of AXL (GAS6) was found co-expressed and at higher level within X2A (Figures 6O and 6P). In line with this, GAS6 protein was detected in the supernatant of SCAC cultures and more abundantly in X2A cultures (Figure 6P), and X2A CD34+CD45RA− HSPCs had an increased number of cells positive for phospho-AXL (pAXL) (Figures 6Q and 6R). Single-cell transcriptomics of SCAC-expanded HSPC and profiling of graft composition To characterize the CD34+ graft composition of the SCAC-expanded cells, we performed single-cell RNA sequencing (scRNA-seq) on day 14 fluorescence-activated cell sorting (FACS)-purified CD34+ cells from X2A (9,055 cells), X2B (8,357), C6 (11,682) cultures, and 7,569 NC CD34+ cells. Genes with high variance were used to compute principal components as inputs for projecting cells in two dimensions by using t-SNE and performed clustering using graph-based clustering35 with a range of resolution parameters. These analyses identified 22 clusters in X2A, 23 clusters in X2B, 22 clusters in C6, and 22 clusters in NC CD34+ cells (Figures S3A–S3D). To identify the cell clusters containing HSC-enriched cells (HSCe: HSC and MPP), LMPPs, common myeloid progenitors (CMPs), granulocyte monocyte progenitors (GMPs), megakaryocyte-erythroid progenitors (MEPs), and mature precursors (MPs) within the scRNA-seq data, we combined all cells and repeated the clustering analysis, which identified 21 clusters; 16 clusters were relatively homogeneous containing X2A, C6, and X2B cells, and five clusters (cluster 0, cluster 6, cluster 9, cluster 15, cluster 20) contained mostly cells belonging to C6 (Figures S3E and S3F). Interestingly, the numbers of genes detected within these clusters were found to be significantly lower compared to the median genes detected within the overall data (Figures S4A–S4D). The NC scRNA-seq data were excluded because NC cells are mostly non-cycling and express a unique transcriptional profile (data not shown). Independently validated gene markers (see STAR Methods and Figure S5 for details) were then used to define the clusters within the scRNA-seq data (Figures S4E–S4I). Graft composition analysis revealed that X2A-expanded CD34+ cells contained an increased percentage of LMPPs and GMPs compared to C6. X2B graft contained a higher percentage of LMPPs compared to X2A and C6. C6 graft contained a higher percentage MEPs (∼75%) and CMPs compared to X2A and X2B (Figures S3G and S5). The highest percentages of HSCe cells were found in the NC cells followed by X2A and C6 (Figures S3G and S5). HSCe clusters from NC, X2A, X2B, and C6 were extracted and reanalyzed (Figures S3A–S3D, S3H, and S4E). Trajectory analysis of this dataset predicted a linear, multi-state trajectory with one branchpoint (Figure S3I). Pseudotime analysis revealed that at one end were NC HSCe, while on the other end were the C6 HSCe (Figures S3I and S4J). Interestingly, X2A/X2B HSCe located in the middle extending toward both ends (Figure S3I). Considering that the frequency of long-term SRCs was greater within X2A compared to C6 (Figure 5E), we sought to answer whether these differences were only due to numbers or inherent alterations in the transcriptional programs within the HSPCs. We therefore extracted and reanalyzed cluster 7 from X2A and C6 (Figure S3J). Differential gene analysis revealed that X2A HSPCs and C6 HSPCs displayed 492 DEGs (fold change [FC] > 1, p < 0.01, data not shown). GO analysis of the upregulated DEGs within X2A revealed biological processes such as immune response and interferon-gamma-mediated signaling to be enriched within X2A compared to C6 (Table S2). Conversely, cell division, cell proliferation, and DNA replication were enriched within C6 compared to X2A (Table S3). Consistent with the bulk-RNA-seq data, AXL was expressed at higher levels within X2A HSCe than C6 (Figure S3K). GAS6 promotes expansion of CD34+CD45RA− HSPCs through AXL signaling AXL signaling plays a pivotal role in the self-renewal and expansion of LSC in both AML and CMLs.24,25,36 RNA-seq analyses, RT-qPCR, and flow cytometry data convincingly revealed the upregulation of AXL in X2A HSPCs (Figures 6M–6R), raising the possibility that AXL signaling drives the expansion of CD34+CD45RA− HSPC within X2A. This hypothesis was tested with an AXL-specific kinase inhibitor. Bemcentinib at 0.2 μM inhibited activation of AXL signaling and significantly reduced the expansion of CD34+CD45RA− HSPCs within the SCAC X2A only (Figures 7A and 7B), but higher bemcentinib concentration also inhibited C6. Similarly, addition of a GAS6 neutralizing antibody reduced AXL signaling and reduced the expansion of CD34+CD45RA− HSPCs within the X2A cultures only (Figures 7C–7E). In addition, adding GAS6 to X2A, C6, and X2B cultures enhanced the expansion of CD34+CD45RA− HSPCs (Figures 7C–7E). In contrast, lentiviral-mediated knockdown of AXL (>75%) dramatically blunted the expansion of CD34+CD45RA− cells induced by X2A and rendered HSPCs insensitive to GAS6 supplementation (Figure S6). Cell-cycle analysis revealed that both bemcentinib and the GAS6 neutralizing antibody reduced the proliferation capacity of CD34+CD45RA− HSPCs in X2A cultures (Figures S7A and S7B) without affecting their viability (Figures S7C and S7D). The growth-promoting activity of GAS6 was not limited to SCAC cultures as addition of GAS6 to all HSPC expansion platforms tested, including STFL control, stand-alone agonists, C6, and SCAC cultures, raised the expansion of CB TNC, CD34+, and CD34+CD45RA− HSPCs (Figure S8). These findings indicate that GAS6 can act as a growth factor for HSPC and that its levels are even limiting in X2A cultures.Figure 7AA2P-driven GAS6/AXL signaling axis is critical for HSC and progenitor expansion(A and B) Impact of the AXL kinase inhibitor bemcentinib on the expansion of CD34+CD45RA− cells in indicated cell cultures (mean ± SEM, n = 3). One-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.(C–E) Impact of GAS6 antibody-mediated neutralization or GAS6 supplementation on the expansion of CD34+CD45RA− cells in indicated cultures (mean ± SEM, n = 3). One-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.(F and G) Peripheral blood analysis of NSG mice transplanted with the progeny from 750 starting CD34+ cells cultured in indicated culture conditions. Human platelet levels (F) and CD45+ leukocytes (G) presented (mean ± SEM, n = 2, 9–10 mice/group). Two-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.(H–J) Frequency of human CD45+ BM cells (H) and the net CD34+ BM cell numbers (I) and human BM CFU (J) colonies in indicated mice 18 weeks post transplant (mean ± SEM, n = 2, 9–10 mice/group). One-way ANOVA, ∗∗∗∗p < 0.0001.(K–M) Frequency of human CD45+ cells (K) and the net CD34+ cell numbers (L) and human CFU (M) colonies from BM of mice 16 weeks post secondary transplant (mean ± SEM, n = 1, four mice/group). One-way ANOVA, ∗∗p < 0.01, ∗∗∗∗p < 0.0001.(N–P) Intra-nuclear flow cytometry analysis of CD34+CD45RA−cells revealed that cultures supplemented within AA2P show loss of DNA methylation. Addition of TET protein inhibitor BobCat (BOB 30) to cultures containing AA2P reversed the global loss of DNA methylation triggered by AA2P (mean ± SEM, n = 3). One-way ANOVA, ∗∗∗∗p < 0.0001.(Q–S) BobCat reverses AA2P-induced transcriptional de-repression of AXL and GAS6 within (Q) X2A but not C6 (R) and X2B cultures (S) (mean ± SEM, n = 3). Two-way ANOVA; ∗∗p < 0.01, ∗∗∗p < 0.001. AA2P-driven GAS6/AXL signaling axis is critical for HSC and progenitor expansion (A and B) Impact of the AXL kinase inhibitor bemcentinib on the expansion of CD34+CD45RA− cells in indicated cell cultures (mean ± SEM, n = 3). One-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. (C–E) Impact of GAS6 antibody-mediated neutralization or GAS6 supplementation on the expansion of CD34+CD45RA− cells in indicated cultures (mean ± SEM, n = 3). One-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. (F and G) Peripheral blood analysis of NSG mice transplanted with the progeny from 750 starting CD34+ cells cultured in indicated culture conditions. Human platelet levels (F) and CD45+ leukocytes (G) presented (mean ± SEM, n = 2, 9–10 mice/group). Two-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (H–J) Frequency of human CD45+ BM cells (H) and the net CD34+ BM cell numbers (I) and human BM CFU (J) colonies in indicated mice 18 weeks post transplant (mean ± SEM, n = 2, 9–10 mice/group). One-way ANOVA, ∗∗∗∗p < 0.0001. (K–M) Frequency of human CD45+ cells (K) and the net CD34+ cell numbers (L) and human CFU (M) colonies from BM of mice 16 weeks post secondary transplant (mean ± SEM, n = 1, four mice/group). One-way ANOVA, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. (N–P) Intra-nuclear flow cytometry analysis of CD34+CD45RA−cells revealed that cultures supplemented within AA2P show loss of DNA methylation. Addition of TET protein inhibitor BobCat (BOB 30) to cultures containing AA2P reversed the global loss of DNA methylation triggered by AA2P (mean ± SEM, n = 3). One-way ANOVA, ∗∗∗∗p < 0.0001. (Q–S) BobCat reverses AA2P-induced transcriptional de-repression of AXL and GAS6 within (Q) X2A but not C6 (R) and X2B cultures (S) (mean ± SEM, n = 3). Two-way ANOVA; ∗∗p < 0.01, ∗∗∗p < 0.001. Next, we investigated whether GAS6-mediated activation of AXL improves engraftment by comparing the engraftment of the progeny of 750 CD34+ cells expanded in X2A cultures with or without GAS6 supplementation or GAS6 neutralization from day 10 to day 14. GAS6 supplementation in cultures significantly enhanced short- and long-term platelet and leukocyte engraftment (Figures 7F and 7G) and BM engraftment (Figures 7H–7L). Recipients of X2A + GAS6 HSPCs had a pronounced increase in the proportion of CD34+ BM cells without preventing their differentiation (data not shown). Secondary engraftment was also increased by 11-fold by GAS6 (Figure 7K; Table S4). Impressively, the net number of human CD34+ BM cells and CFUs in primary and secondary recipients were increased by 7- and 15-fold (p < 0.0001) in primary, and by 20- and 26-fold (p < 0.0001) in secondary, transplants by GAS6 supplementation in culture (Figures 7I, 7J, 7L, and 7M). In contrast, GAS6 neutralization reduced human chimerism in all responses investigated (Figures 7F–7M). These results show that GAS6/AXL signaling is a significant driving force for the expansion of CD34+CD45RA− HSPCs within the SCAC X2A and that GAS6 is a potent growth factor for the expansion of HSPCs that retain serial engraftment activity. AA2P promotes increased expression of GAS6 and AXL via DNA hypomethylation Ascorbic acid is a cofactor for the TET2 enzyme known to induce DNA hypomethylation in HSPCs.19,20 Flow cytometry analysis revealed that AA2P promoted global DNA hypomethylation in CD34+CD45RA− HSPCs (Figures 7N–7P). Similar levels were measured in X2A, which were dependent on AA2P as it was lost in X2B. In contrast, HSPCs produced with UM171 and SR1 (modestly) had increased global DNA methylation levels (Figure S9). UM171 and AA2P as single molecules or in SCACs triggered increased acetylation of H3K27 (H3K27ac) and demethylation of H3K4 (H3K4me2) levels within the CD34+CD45RA− cell population (Figure S9).9 In contrast, SR1 did not appear to influence the global levels of H3K27ac and H3K4me2 (Figure S9). TET inhibition with the chemical inhibitor BobCat rescued the global levels of DNA methylation in AA2P and X2A cultures (Figures 7N and 7P). Next, we investigated whether AA2P triggered changes in the transcript levels of GAS6 and AXL was due to TET-mediated DNA hypomethylation. Methyl-qPCR results confirmed that AA2P promoted hypomethylation within the CpG islands located in the promoters of both genes and that TET protein inhibition with BobCat restored the methylation status of these CpG islands and repressed AXL and GAS6 transcripts (Figures 7Q–7S and S10). Altogether, these results demonstrate that, within CD34+CD45RA− HSPCs, AA2P promotes the expression of GAS6 and AXL via TET-mediated DNA hypo-methylation of the CpG islands located at the proximity of their promoters. Discussion Several chromatin-modifying drugs have been shown to promote ex vivo expansion of HSPCs, some of which led to improved engraftment outcomes in patients.1,2,17 Herein, we identified a stem cell agonist in AA2P and developed a series of SCACs that maximize the expansion of CB cell subpopulations deemed highly relevant for engraftment. Subsequent functional investigations revealed that AA2P-mediated enhanced expansion of HSPCs with X2A is in part driven through the activation of the AXL signaling pathway. Furthermore, epigenetic analysis revealed that AA2P promoted TET-mediated DNA hypomethylation of CpG islands proximal to the promoter region of AXL and GAS6. Finally, we showed that GAS6-mediated AXL activation promotes robust expansion of human HSPCs including serial engrafting HSCs resulting in unexpectedly high engraftment. Ascorbic acid is a metabolite highly enriched in HSCs and MPP when compared to downstream progenitors in mice and human. Mice can synthesize ascorbic acid, whereas human acquire it through diet. Its suppression in Gulo−/− mice was shown to lead to an increase of the HSC pool,19 but its role as a stem cell agonist was unknown. In culture, AA2P is frequently used as an alternative to ascorbic acid.37 AA2P increased the expansion of several CD34+ cell subsets in a dose-response manner during the second week, revealing that its impact on cell growth is not as acute as other agonists such as SR1. AA2P also provided a clear benefit to ensuing engraftment of platelets, which have a short half-life and provide a good correlation with engraftment and HSC activity,38 and for human BM chimerism. This rise in engraftment was likely due to the increased doses of HSPCs transplanted since AA2P had modest impact on the chemotaxis properties of HSPCs. Ascorbic acid treatment has previously been shown to promote widespread DNA CpG demethylation.39,40 Indeed, ascorbic acid is a cofactor for TET enzyme, which promotes demethylation,41 a finding corroborated in two recent studies linking ascorbic acid and TET2 enzyme to the regulation of HSC self-renewal and leukemogenesis.19,20 In line with this, AA2P induced DNA hypomethylation in HSPC strictly in a TET-dependent manner. DNA methylation of CpG islands is a common form of epigenetic modification essential for normal development.42,43 Chemical inhibition of TET enzymes also blunted AA2P growth-promoting activity, showing a direct correlation between the two in AA2P-containing cultures. Conversely, the levels of reactive oxygen species and the proportion of HSPCs positive for those in AA2P cultures were not significantly different (data not shown). These results confirm that the mean of action of ascorbic acid and AA2P on stem cells are principally dependent on their DNA-demethylation-promotion effects.19,39 UM171 and VPA promote HSPC expansion via epigenetic reprogramming.9,11 We present indications that UM171, SR1, and AA2P also diverge significantly in their ability to influence the epigenetic landscape of CD34+CD45RA− HSPCs. Consistent with previous studies, UM171 promoted increased levels of H3K27ac and H3K4me2 within CD34+CD45RA−HSPCs.9 AA2P had modest impact on the global levels of H3K27ac and H3K4me2 but triggered a global loss in DNA methylation levels in HSPCs. In contrast, HSPC produced with UM171 and SR1 showed elevated global DNA methylation especially with UM171, but SR1 had no impact on other epigenetic marks investigated herein.6 Moving forward, it would be interesting to investigate whether the increase in DNA methylation seen with UM171 limits its capacity for HSC expansion. Statistical design of experiments was used to develop a series of agonist cocktails.44 Factorial designs revealed that SR1, UM171, and VPA were the dominant factors for the expansion of HSPC subsets. Surprisingly, our results indicated that very few, if any, of the small molecules tested showed synergistic actions but, rather, some had clear antagonistic activities (e.g., UM171∗VPA). The RS models combined with further investigations revealed that X2A, which uses similar levels of VPA and AA2P to SM2, but with SR1 and UM171 at half of their concentrations, uniquely provided similar maximal expansion of CD34+CD45RA− and EPCRHigh subsets to SMA and SM2, respectively. The limitations of the prediction models are in part the result of the single test condition for the maximal and minimal concentration points. This resulted in a model overstating the importance of UM171 for the expansion of EPCRHigh cells. Serial transplant assays indicated that the SCACs X2A and SM6 provided the best overall engraftment properties to CB HSPCs produced in culture. SMA also provided excellent engraftment properties, where SM2 produced poor engraftment due to the toxic effects of elevated UM171 concentration.8 The functional contribution of AA2P to X2A serial engraftment properties was clearly established with X2B where AA2P is essentially absent. Importantly, short- and long-term engraftment was superior with X2A-expanded HSPCs compared to that achieved with NC fresh CD34+ cells. These findings were confirmed in LDA transplants, which indicated that long-term SRCs (i.e., HSCs) were increased 15-fold over input level by X2A and 4-fold over the combination of SR1 and UM171.7 We identified three relevant RTK genes highly expressed within X2A HSPCs when compared to X2B in FLT3, AXL, and EGFR. FLT3 expression enriches human long-term HSCs and support survival of human HSCs ex vivo.45 Conversely, AXL plays a significant role in the self-renewal of LSCs,24,25,36 and its natural ligand GAS6 was also upregulated by X2A. Interestingly, ectopic expression of GAS6 was shown to allow for murine progenitor maintenance in co-culture, but the role of GAS6/AXL in human HSCs was never investigated.46 We found that TET-mediated hypomethylation of AXL and GAS6 promoter by AA2P led to their increased expression in HSPCs. Both chemical inhibition and genetic knockdown of AXL significantly impaired the expansion of HSPCs and also blocked the capacity of exogenous GAS6 to raise HSPC expansion. Moreover, we confirmed the presence of GAS6 in SCAC cultures and the significant activation of AXL in X2A HSPC-enriched cells. These results show that AXL is activated by GAS6 in a para- and/or autocrine manner in X2A cultures. Interestingly, GAS6 levels even in X2A cultures were limiting since exogenous GAS6-supplementation enhanced HSPC expansion and ensuing engraftment 10-fold increase in serial engraftment. Remarkably, the net numbers of human CFU and CD34+ BM cells were increased by 7- and 15-fold in primary, and by 20- and 26-fold in secondary, transplants by GAS6, respectively. Our findings also suggest that the AXL/GAS6 axis is important for most serum-free HSPC expansion platforms since addition of GAS6 raised HSPC expansion in all condition tested, and that GAS6 neutralization, at least in X2A cultures, impaired HSPC expansion and downstream engraftment. Together, these findings suggest that GAS6 can act as an unrecognized potent growth factor for human HSPCs including HSCs. GAS6′ biological activity is highly dependent on N-terminal carboxylation, which is mediated by the vitamin K-dependent gamma-glutamyl carboxylase (GGCX) enzyme.47,48 We confirmed that GGCX is expressed in X2A and X2B cultured HSPCs and that GAS6 is carboxylated in X2A cultures. Moreover, GGCX knockdown reduced the carboxylation of GAS6, suggesting that the regulation of GAS6′ biological activity in HSPCs is similar to what was previously reported (unpublished results, Maganti et al., 2023).47,48 The impact of expansion on graft composition is unclear to date but likely to affect engraftment outcomes.49 For the first time to our knowledge, single-cell transcriptomics was used to characterize the composition of SCAC-expanded CD34+ cells. First, we had to identify gene markers and validated their robustness to detect functionally divergent HSPC subpopulations in both NC and SCAC-expanded CD34+ cells.50,51,52 These markers can now act as a resource for other groups to identify the various HSPC populations. While X2A provides significantly higher expansion of HSPCs and engraftment compared to X2B, this difference was mostly due to the greater proliferation with X2A as only minor differences between X2A and X2B grafts were seen. Pseudotime trajectory analysis also revealed that X2A HSPC grafts were more closely related to their NC HSPCs than C6 grafts, which is consistent with the higher frequencies of SRC measured in X2A. In conclusion, we developed a series of small-molecule cocktails supporting superior expansion of HSPC in culture. Included in those is AA2P, a derivative of vitamin C shown to improve HSPC expansion through DNA hypomethylation and elucidated a previously underappreciated role of AXL and its ligand GAS6 in promoting HSC self-renewal and expansion. Limitations of the study A current limitation is that only HSCs from CB were used to optimize and test the SCACs. It is possible that the growth responses of adult HSCs to stem cell agonists tested herein may differ. Hence, future work will need to determine which SCACs promote the best maintenance and expansion of HSCs derived from BM and/or peripheral blood stem cells. Finally, while our data demonstrate that GAS6 supplementation in culture enhances the serial engraftment of expanded HSPCs, LDAs will be required to truly quantitate the impact of GAS6 on human HSC expansion in culture. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesHuman anti-CD34, PE conjugated, clone 581BD BiosciencesCat# 555822; RRID: AB_396151Human anti-CD38, APC conjugated, clone HIT2BD BiosciencesCat# 555462; RRID: AB_398599Human anti-CD38, FITC conjugated, clone HIT2BD BiosciencesCat# 555459; RRID: AB_395852Human anti-CD45, APC conjugated, clone HI30BD BiosciencesCat# 555485; RRID: AB_398600Human anti-CD45, FITC conjugated, clone HI30BD BiosciencesCat# 555482; RRID: AB_395874Human anti-CD3, APC conjugated, clone UCHT1BD BiosciencesCat# 555335; RRID: AB_398591Human anti-CD19, PE conjugated, clone HIB19BD BiosciencesCat# 555413; RRID: AB_395813Human anti-CD19, FITC conjugated, clone HIB19BD BiosciencesCat# 555412; RRID: AB_395812Human anti-CD33, PE conjugated, clone WM53BD BiosciencesCat# 555450; RRID: AB_395843Human anti-CD90, PE-Cy7 conjugated, clone 5E10BD BiosciencesCat# 561558; RRID: AB_10714644Human anti-CD49f, PerCp-Cy5.5 conjugated, clone GoH3BD BiosciencesCat# 562495; RRID: AB_11151910Human anti-CD56, PE conjugated, clone NCAM16.2BD BiosciencesCat# 340363; RRID: AB_400017Human anti-CD41a, FITC conjugated, clone HIP8BD BiosciencesCat# 555466; RRID: AB_395858Human anti-CD41a, PE conjugated, clone HIP8BD BiosciencesCat#; 557297; RRID: AB_396624Human anti-CD123, PerCp-Cy5.5 conjugated, clone 7G3BD BiosciencesCat# 560904; RRID: AB_1645547Human anti-CD14, PE conjugated, clone M5E2BD BiosciencesCat# 555398; RRID: AB_395799Human anti-CD201(EPCR), APC conjugated, clone RCR-401BiolegendCat# 351906; RRID: AB_11126139Human anti-CD184(CXCR4), APC conjugated, clone 12G5BD BiosciencesCat# 555976; RRID: AB_398616Human anti-CD45RA, APC-H7 conjugated, clone HI100BD BiosciencesCat# 560674; RRID: AB_1727497Human anti-CD45RA, FITC conjugated, clone HI100BD BiosciencesCat# 555488; RRID: AB_395879Human anti-CD10, PerCP-Cy5.5 conjugated, clone HI10aBioLegendCat # 312216; RRID: AB_10642819Anti-Human GAS6 antibodyR&D SystemsCat# AB885; RRID: AB_354376Human anti-AXL, unconjugatedR&D SystemsCat# AF154; RRID: AB_354852Human anti-phospho-Axl (Y779)R&D SystemsCat# AF2228; RRID: AB_2062560Anti-5-methylcytosine antibody, clone 33D3Millipore SigmaCat# MABE146; RRID: AB_10863148anti-H3K27AcCell Signaling TechnologyCat# 15562; RRID: AB_2798746anti-H3K4me2AbcamCat# 7766; RRID:AB_2560996Bacterial and virus strainsBiological samplesUmbilical Cord Blood UnitsCanadian Blood ServicesREB 2018.015Chemicals, peptides, and recombinant proteinsStemReginin1STEMCELL TechnologiesCat# 72342UM171STEMCELL TechnologiesCat# 72912Valproic Acid (Sodium Salt)STEMCELL TechnologiesCat# 72292L-Ascorbic AcidSTEMCELL TechnologiesCat# 72132L-Ascorbic acid 2-phosphate sesquimagnesium salt hydrateSigmaCat# A8960StemSpan™ SFEMSTEMCELL TechnologiesCat# 09650Ficoll-Paque PLUS density gradient mediaCytivaCat# 17144003EasySep™ Human CD34 Positive Selection Kit IISTEMCELL TechnologiesCat# 17856MethoCult Classic H4434STEMCELL TechnologiesCat# 04434BobCat-339 free baseMedKooCat# 408006BemcentinibMedChemExpressCat# HY-15150PolyethyleniminePolysciencesCat# 23966Human SCFPeproTechCat# 300-07-250UGHuman FLT3LPeproTechCat#300-19-250UGHuman TPOPeproTechCat#300-18-250UGHuman LDLSTEMCELL TechnologiesCat# 02698Human SDF-1αPeproTechCat# 300-28AGibco™ Penicillin-StreptomycinThermoFisher ScientificCat# 15140122Human GAS6R&D SystemsCat# 885-GSB-050Anti-GAS6 Rabbit mAbCell SignalingCat# 67202; RRIB: AB_2799720Peroxidase IgG Fraction Monoclonal Mouse Anti-Rabbit IgG, light chain specificJackson Immuno ResearchCat# 211-032-1714x Laemmli Sample BufferBioRadCat# 161-07472-MercaptoethanolBioRadCat# 161-0710GelThermoFisher ScientificCat# NP0321K2SuperSignal™ West Pico PLUS Chemiluminescent SubstrateThermoFisher ScientificCat# 34580Halt™ Protease and Phosphatase Inhibitor CocktailThermoFisher ScientificCat# 78442BD Pharm Lyse™ solutionBD BiosciencesCat# 555899AccuCheck counting beadsThermoFisher ScientificCat# PCB100Hoechst 33342 Solution (20 mM)ThermoFisher ScientificCat# 62249SYTOX™ AADvanced™ Dead Cell Stain KitThermoFisher ScientificCat# S10274Annexin V, Alexa Fluor 488 conjugatedThermoFisher ScientificCat# A13201RosetteSep™ Human Cord Blood Progenitor Enrichment KitSTEMCELL TechnologiesCat# 152766.5 mm Transwell® with 5.0 μm Pore Polycarbonate Membrane InsertCorningCat# 3421Critical commercial assaysPierce™ Protein ConcentratorThermoFisher ScientificCat# 88527iScript™ Reverse Transcription SupermixBio-RadCat# 1708840SsoAdvanced universal SYBR green supermixBio-RadCat# 1725270miRNeasy Micro KitQiagenCat# 217084TruSeq Stranded mRNAIlluminaCat# 20020594PicoPure™ RNA isolation kitThermoFisher ScientificCat# KIT0204OneStep qMethyl™ KitZymo ResearchCat# D5310Deposited dataRNA-Seq Raw data, bulkThis paperGEO: GSE202577RNA-Seq Raw data, single-cellThis paperGEO: GSE212578Experimental models: Organisms/strainsNOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG)The Jackson LaboratoryStrain# 005557; RRID: IMSR_JAX:005557OligonucleotidesOligonucleotides used in this paperThis paperSee Table S1Recombinant DNApGIPZHorizon discoveryClone ID:V2LHS_238359V2LHS_202535pMD2GAddgeneCat# 12259pPAX2AddgeneCat# 12260Software and algorithmsJMP Pro 14.0JMP®, Version JMP Pro 14. SAS Institute Inc., Cary, NC, 1989–2021https://www.jmp.com/en_gb/software/new-release/preview-jmp14.htmlELDA: Extreme Limiting Dilution AnalysisHu, Y et al.35,53http://bioinf.wehi.edu.au/software/elda/index.htmlSalmon v1.0R Patro et al.54https://combine-lab.github.io/salmon/DESeq2 (v1.28.1)Love MI et al.55http://www.bioconductor.org/packages/release/bioc/html/DESeq2.htmlDAVID (version 6.8)https://david.ncifcrf.gov/home.jspGraphPad Prism v 8.0https://www.graphpad.com/scientific-software/prism/MORPHEUS softwareBroad institutehttps://software.broadinstitute.org/morpheus/Cell Ranger V3 software10X Genomicshttps://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/installationSeurat v2Laboratory of Rahul Satjahttp://satjalab.org/seurat/; RRID:SCR_016341Monocle v2Trapnell C et al.56http://cole-trapnell-lab.github.io/monocle-release/OtherI-Blot 2 dry blotting systemThermoFisher ScientificCat# IB21001BD FACS Aria IIIBD BiosciencesN/AAttune® CytometerThermo Fisher ScientificN/AAmnis ImageStreamX Mk II Imaging Flow CytometerLuminexRRID:SCR_020142CFX96 touch real-time PCR detection systemBio-RadRRID:SCR_018064 REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesHuman anti-CD34, PE conjugated, clone 581BD BiosciencesCat# 555822; RRID: AB_396151Human anti-CD38, APC conjugated, clone HIT2BD BiosciencesCat# 555462; RRID: AB_398599Human anti-CD38, FITC conjugated, clone HIT2BD BiosciencesCat# 555459; RRID: AB_395852Human anti-CD45, APC conjugated, clone HI30BD BiosciencesCat# 555485; RRID: AB_398600Human anti-CD45, FITC conjugated, clone HI30BD BiosciencesCat# 555482; RRID: AB_395874Human anti-CD3, APC conjugated, clone UCHT1BD BiosciencesCat# 555335; RRID: AB_398591Human anti-CD19, PE conjugated, clone HIB19BD BiosciencesCat# 555413; RRID: AB_395813Human anti-CD19, FITC conjugated, clone HIB19BD BiosciencesCat# 555412; RRID: AB_395812Human anti-CD33, PE conjugated, clone WM53BD BiosciencesCat# 555450; RRID: AB_395843Human anti-CD90, PE-Cy7 conjugated, clone 5E10BD BiosciencesCat# 561558; RRID: AB_10714644Human anti-CD49f, PerCp-Cy5.5 conjugated, clone GoH3BD BiosciencesCat# 562495; RRID: AB_11151910Human anti-CD56, PE conjugated, clone NCAM16.2BD BiosciencesCat# 340363; RRID: AB_400017Human anti-CD41a, FITC conjugated, clone HIP8BD BiosciencesCat# 555466; RRID: AB_395858Human anti-CD41a, PE conjugated, clone HIP8BD BiosciencesCat#; 557297; RRID: AB_396624Human anti-CD123, PerCp-Cy5.5 conjugated, clone 7G3BD BiosciencesCat# 560904; RRID: AB_1645547Human anti-CD14, PE conjugated, clone M5E2BD BiosciencesCat# 555398; RRID: AB_395799Human anti-CD201(EPCR), APC conjugated, clone RCR-401BiolegendCat# 351906; RRID: AB_11126139Human anti-CD184(CXCR4), APC conjugated, clone 12G5BD BiosciencesCat# 555976; RRID: AB_398616Human anti-CD45RA, APC-H7 conjugated, clone HI100BD BiosciencesCat# 560674; RRID: AB_1727497Human anti-CD45RA, FITC conjugated, clone HI100BD BiosciencesCat# 555488; RRID: AB_395879Human anti-CD10, PerCP-Cy5.5 conjugated, clone HI10aBioLegendCat # 312216; RRID: AB_10642819Anti-Human GAS6 antibodyR&D SystemsCat# AB885; RRID: AB_354376Human anti-AXL, unconjugatedR&D SystemsCat# AF154; RRID: AB_354852Human anti-phospho-Axl (Y779)R&D SystemsCat# AF2228; RRID: AB_2062560Anti-5-methylcytosine antibody, clone 33D3Millipore SigmaCat# MABE146; RRID: AB_10863148anti-H3K27AcCell Signaling TechnologyCat# 15562; RRID: AB_2798746anti-H3K4me2AbcamCat# 7766; RRID:AB_2560996Bacterial and virus strainsBiological samplesUmbilical Cord Blood UnitsCanadian Blood ServicesREB 2018.015Chemicals, peptides, and recombinant proteinsStemReginin1STEMCELL TechnologiesCat# 72342UM171STEMCELL TechnologiesCat# 72912Valproic Acid (Sodium Salt)STEMCELL TechnologiesCat# 72292L-Ascorbic AcidSTEMCELL TechnologiesCat# 72132L-Ascorbic acid 2-phosphate sesquimagnesium salt hydrateSigmaCat# A8960StemSpan™ SFEMSTEMCELL TechnologiesCat# 09650Ficoll-Paque PLUS density gradient mediaCytivaCat# 17144003EasySep™ Human CD34 Positive Selection Kit IISTEMCELL TechnologiesCat# 17856MethoCult Classic H4434STEMCELL TechnologiesCat# 04434BobCat-339 free baseMedKooCat# 408006BemcentinibMedChemExpressCat# HY-15150PolyethyleniminePolysciencesCat# 23966Human SCFPeproTechCat# 300-07-250UGHuman FLT3LPeproTechCat#300-19-250UGHuman TPOPeproTechCat#300-18-250UGHuman LDLSTEMCELL TechnologiesCat# 02698Human SDF-1αPeproTechCat# 300-28AGibco™ Penicillin-StreptomycinThermoFisher ScientificCat# 15140122Human GAS6R&D SystemsCat# 885-GSB-050Anti-GAS6 Rabbit mAbCell SignalingCat# 67202; RRIB: AB_2799720Peroxidase IgG Fraction Monoclonal Mouse Anti-Rabbit IgG, light chain specificJackson Immuno ResearchCat# 211-032-1714x Laemmli Sample BufferBioRadCat# 161-07472-MercaptoethanolBioRadCat# 161-0710GelThermoFisher ScientificCat# NP0321K2SuperSignal™ West Pico PLUS Chemiluminescent SubstrateThermoFisher ScientificCat# 34580Halt™ Protease and Phosphatase Inhibitor CocktailThermoFisher ScientificCat# 78442BD Pharm Lyse™ solutionBD BiosciencesCat# 555899AccuCheck counting beadsThermoFisher ScientificCat# PCB100Hoechst 33342 Solution (20 mM)ThermoFisher ScientificCat# 62249SYTOX™ AADvanced™ Dead Cell Stain KitThermoFisher ScientificCat# S10274Annexin V, Alexa Fluor 488 conjugatedThermoFisher ScientificCat# A13201RosetteSep™ Human Cord Blood Progenitor Enrichment KitSTEMCELL TechnologiesCat# 152766.5 mm Transwell® with 5.0 μm Pore Polycarbonate Membrane InsertCorningCat# 3421Critical commercial assaysPierce™ Protein ConcentratorThermoFisher ScientificCat# 88527iScript™ Reverse Transcription SupermixBio-RadCat# 1708840SsoAdvanced universal SYBR green supermixBio-RadCat# 1725270miRNeasy Micro KitQiagenCat# 217084TruSeq Stranded mRNAIlluminaCat# 20020594PicoPure™ RNA isolation kitThermoFisher ScientificCat# KIT0204OneStep qMethyl™ KitZymo ResearchCat# D5310Deposited dataRNA-Seq Raw data, bulkThis paperGEO: GSE202577RNA-Seq Raw data, single-cellThis paperGEO: GSE212578Experimental models: Organisms/strainsNOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG)The Jackson LaboratoryStrain# 005557; RRID: IMSR_JAX:005557OligonucleotidesOligonucleotides used in this paperThis paperSee Table S1Recombinant DNApGIPZHorizon discoveryClone ID:V2LHS_238359V2LHS_202535pMD2GAddgeneCat# 12259pPAX2AddgeneCat# 12260Software and algorithmsJMP Pro 14.0JMP®, Version JMP Pro 14. SAS Institute Inc., Cary, NC, 1989–2021https://www.jmp.com/en_gb/software/new-release/preview-jmp14.htmlELDA: Extreme Limiting Dilution AnalysisHu, Y et al.35,53http://bioinf.wehi.edu.au/software/elda/index.htmlSalmon v1.0R Patro et al.54https://combine-lab.github.io/salmon/DESeq2 (v1.28.1)Love MI et al.55http://www.bioconductor.org/packages/release/bioc/html/DESeq2.htmlDAVID (version 6.8)https://david.ncifcrf.gov/home.jspGraphPad Prism v 8.0https://www.graphpad.com/scientific-software/prism/MORPHEUS softwareBroad institutehttps://software.broadinstitute.org/morpheus/Cell Ranger V3 software10X Genomicshttps://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/installationSeurat v2Laboratory of Rahul Satjahttp://satjalab.org/seurat/; RRID:SCR_016341Monocle v2Trapnell C et al.56http://cole-trapnell-lab.github.io/monocle-release/OtherI-Blot 2 dry blotting systemThermoFisher ScientificCat# IB21001BD FACS Aria IIIBD BiosciencesN/AAttune® CytometerThermo Fisher ScientificN/AAmnis ImageStreamX Mk II Imaging Flow CytometerLuminexRRID:SCR_020142CFX96 touch real-time PCR detection systemBio-RadRRID:SCR_018064 Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Nicolas Pineault (Nicolas.pineault@blood.ca) Materials availability No new plasmids were generated in this study. All the lenti-plasmids were commercially purchases and catalog numbers are provided in the key resources table. Data and code availability (1)Single-cell RNA-seq and bulk-RNAseq data have been deposited at GEO (GSE212578) and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request.(2)This paper does not report original code.(3)Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. (1)Single-cell RNA-seq and bulk-RNAseq data have been deposited at GEO (GSE212578) and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request. (1)Single-cell RNA-seq and bulk-RNAseq data have been deposited at GEO (GSE212578) and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request. (2)This paper does not report original code. (2)This paper does not report original code. (3)Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. (3)Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Experimental model and study participant details All animal work was preapproved by the Animal Care Committee of University of Ottawa. All mice experiments were done using irradiated (300 cGy) 8 weeks old female NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG, Jackson Laboratory, Bar Harbor, USA) mice. While the impact of estrogen on hematopoietic stem cell engraftment is not observed in the clinic, estrogen promotes HSC engraftment in xenotransplant mice models and despite this artifact most preclinical studies use female NSG mice. Keeping this in mind and for the ease of comparing results, we used only female NSG mice in our studies. However, to account for the impact of sex, CD34+ cells from both male and female cord blood samples were used. Umbilical cord blood CD34+ cells were transplanted into irradiated mice via tail vein injections. A total of 2–3 cord blood samples were used for each experiment. Method details Collection and purification of CB CD34+ cells The project was preapproved by the Canadian Blood Services Ethic review board and cord blood (CB) units were obtained after written informed consent from the Canadian Blood Services CB for Research Program. An informed consent was obtained from all participants. CB units contained a minimum of 750 x106 nucleated cells and were obtained less than 24 h post-collection. Mononuclear cells were isolated by Ficoll-Paque Plus (GE, Pittsburgh, PA) and CD34+ CB cells were enriched using the EasySep Human CD34 Positive Selection Kit II (StemCell Technologies, Vancouver, Canada). All independent CD34+ cell preparations used in this study were prepared from 2 to 3 CB units. A total of 42 CB units were used during the course of this study. Within 42 CB units, the sex of 22 babies were male, and 20 were female. Gender could not be determined. The ancestry, race, socioeconomic status of the donors was not recorded at the time of collection. Small molecules StemReginin1 (SR1, Cat 72342), UM171 (Cat 72912), L-ascorbic acid (L-AA, Cat 72132) and VPA (Cat 72292) were purchased from StemCell Technologies (Vancouver, Canada) and AA2P from Sigma (Cat 8960) and reconstituted in dimethyl sulfoxide (DMSO). VPA and AA2P were solubilized in PBS. All molecules were aliquoted and stored at −80°C for single use. The final concentration of DMSO in culture never exceeded the cytotoxic level of 0.01% (v/v). Cell culture Pooled CD34+ cells (two donors) were cultured in 24-well plate (Corning, New York, United States) in serum-free medium (SFM) StemSpan SFEM (StemCell Technologies) supplemented with stem cell factor (SCF), thrombopoietin (TPO), FMS-like tyrosine kinase 3 ligand (FLT3L) each at 100 ng/ml, low-density lipoprotein (LDL, Stem cell Technologies) at 10 μg/ml and 1% Penicillin-Streptomycin (Gibco) at a concentration of 12000 cells/mL. All cytokines were purchased from Peprotech (Rocky Hill, NJ, USA). This medium is referred to as STFL or control. Cultures were maintained at 37°C, 5% CO2 with fresh medium added on day 4, 7 and 10. Cell expansion was monitored on day 7 and 14 by cytometric analysis. Absolute number of stem and progenitor cells were determined on day 7 and 14, and net fold expansions were derived by dividing net final cell numbers by the starting cell numbers. In indicated cultures, 30uM BobCat-339 (Cat# 408006, Medkoo Biosciences, NC, USA), 2uM Bemcentinib (MedChemExpress LLC, Cat# HY-15150 NJ, USA), 10 ng/mL neutralizing GAS6 antibody (Cat# AB885, R&D systems, MIN, USA), recombinant GAS6 (R&D Systems, cat# 885-GSB-050, optimal concentration determined at 10 ng/mL) or lentivirus (MOI 10) were added on day 10 and cultured for 4 days. On day 14, all the cells were harvested and analyzed for cell expansion, phenotype, viability and apoptosis and cell cycle analysis. Lenti-virus production 2.5 X 106 293T cells were seeded in 15-cm dishes (Corning Falcon). After 48 h, cells were co-transfected with second-generation packaging lentiviral plasmids pMD2G, pPAX2, and either pGIPZ encoding shRNAs (Thermo Scientific) in the presence of polyethylenimine (4.1 μmol/L) and NaCl (2.25 × 10−4 mol/L). Supernatant containing the virus was collected 48 and 72 h after transfection and concentrated using Lenti-X concentrator (Takara bio, Cat# 631232) and resuspended in a final volume of 300 μL of STFL medium. Cells were infected at a functional MOI of 10. ShRNA screen and clone selection and Western blot analysis CD34+ cells were cultured for 14 days in SCAC X2A. On day 10, cells were transduced independently with 6 different shRNA clones and GFP+ CD34+ cells were sorted 96 h post transduction. Transduced cells were lysed and subjected to RT-qPCR analysis (Tables S1B and S1C). Two of the selected shRNA clones (Table S1B) with highest knockdown efficiency based on RT-qPCR data were validated by Western blot analysis. Total protein was extracted from 1 million CD34+GFP+ expanded cells for Western blot using NP-40 lysis buffer (Cat: J60766.AP) supplemented with 100X Halt protease and phosphatase inhibitor cocktail (Cat: 78440). 45ul of the retentate was then combined with 15ul of 4X SDS Laemmli Sample Buffer (Cat#161–0747, BioRad) and 2-Mercaptoethanol (Cat# 161–0710, BioRad) and heated for 7 min at 90°C. 40uL of the sample was then loaded into a 10 well gel (Cat#NP0321K2, ThermoFisher Scientific) and Western blot analysis was performed. The gel was then transferred onto methanol activated polyvinylidene difluoride membrane (Millipore) using I-Blot 2 dry blotting system (ThermoFisher Scientific). After, the membrane was blocked for 1 h with 5% bovine serum albumin (BSA; Wisent Bio products) in Tris-Buffered saline (TBS) with 0.1% Tween 20 (TBS-T; BioBasic). The primary antibody cocktail was made in TBS-T with 5% BSA (1:1000 dilutions of AXL (clone: C89E7, Cell Signaling) and GAPDH (clone: 14C10, Cell Signaling)) and the membrane incubated overnight at 4°C. The membrane was then washed with TBS-T and incubated at room temperature in the secondary antibody cocktail (1:10,000 dilution with 5% BSA) of Peroxidase IgG Fraction Monoclonal Mouse Anti-Rabbit IgG, light chain specific (Cat#211-032-171, Jackson Immuno Research) for 1 h. Subsequently the blots were treated with 500ul of SuperSignal West Pico PLUS Chemiluminescent substrate and visualized using ChemiDoc imaging system. Concentration of supernatant from SCAC cultures and western blot analysis Collected SCAC X2A, C6 supernatants (10 mL) or SCAC X2B supernatants (20 mL) during the 14 days expansion period were pooled and then concentrated using Pierce Protein Concentrator (Cat#88527, ThermoFisher Scientific). In brief, supernatant was spun in the 10K column until 100 μl of retentate was left. 45ul of the retentate was then combined with 15ul of 4X SDS Laemmli Sample Buffer and 2-Mercaptoethanol and heated for 7 min at 90°C. 40 μL of the sample was then loaded into a western gel which was performed as mentioned above with the following change; the primary antibody cocktail was made using 1:1000 dilutions of GAS6 (clone: D3A3G. Cat#672025, Cell Signaling) in TBS-T with 5% BSA. Colony-forming unit assay Fresh CD34+ enriched cells or ex vivo expanded CB cells were plated in MethoCult Classic H4435 (StemCell Technologies). Colonies were scored according to their morphology as granulocyte (G), macrophage (M), granulocyte/macrophage (GM), and multilineage colonies containing erythroid and non-erythroid cells (GEMM) following a 14-day incubation period at 37°C. Chemotaxis assay CB CD34+ cells were cultured with indicated or absence of small molecules for 14 days. Cell count was performed using Accucheck beads (Thermofisher) pre- and post migration. 1 × 105 ex vivo expanded cells were in 100 μL SFEM and placed on the upper chamber of the transwell (5.0-μm pores, Corning), while pre-warmed SFEM (650 μL) containing SDF1-a (50 ng/ml) was added to the lower chamber. The 24-well transwell plates were incubated at 37°C for 4 h. After incubation, cells were harvested from lower chamber and stained with human anti-CD34-PE, CD45RA-APC-H7, CD38-FITC, CD90-PECY7 and CXCR4-APC.The percentage migration of indicated cell subpopulations was determined by flow cytometry. Statistical experimental designs First, a factorial design corresponding to a 25 screening design with 5 factors (SR1, UM171, VPA, AA2P and UNC628) was generated with the software JMP Pro 14.0 (SAS, NC, USA). This design was composed of 32 factor combinations to which 4 replicated center points were added. The individual doses (i.e., present) for the factors used were 700nM for SR1, 35nM for UM171, 0.25mM for VPA, 500nM for UNC0638, and 100μM for AA2P. Center points are combinations composed of all 5 factors at half their maximum dose that can be used to detect non-linearity in the small molecule dose-response curves (e.g., curvature) and to perform different diagnostic tests. This is necessary to evaluate the limitations of the initial statistical models to describe accurately the interplay of factor effects and, consequently, their ability to identify, by interpolation, improved factor-dose combinations. When a linear polynomial model is unable to fit the center points, this also provides a rationale for using more advanced techniques such a central composite design and quadratic models, in the subsequent steps of the study.57 The design was replicated 3 times, as 3 independent orthogonal blocks. This screen was done under fixed concentrations of SCF, FL and TPO (100 ng/mL). The number of cells of specific phenotypes were counted after culture in each of the 3 blocks composed of 36 conditions. This data was log transformed and the fitted to a linear polynomial model that represented the contributions of the block effects, main factor effects, factor interaction effects (up to 5-factor interactions). A multi-factor ANOVA was used to perform a null hypothesis test on each of the regression coefficients in the model to determine the level of statistical significance of the factor effects and interactions effects. Residual analysis and lack-of fit tests were performed to evaluate the goodness-of-fit of the model. Second, a four-factor central composite design (CCD) was generated in the software JMP Pro 14.0 composed of 16 conditions with 12 center points and 8 axial points (overall, factors are tested at 5 different doses in 36 cultures in parallel).58 The number of center points and the axial scaling (α, the ratio of the levels of the axial to factorial points) have been chosen in order for this to be orthogonal and rotatable (α = 2).59 The CCD was replicated 3 times, as 3 independent orthogonal blocks. As it was desirable to cover a large range of factor doses, the concentrations tested were established by log transformations of the concentrations in ng/mL and obtaining 5 coded level values equal to −2, −1, 0, 1, 2 (Table S11). The number of cells of specific phenotypes were counted after culture in each of the 3 blocks composed of 36 conditions. This data was log transformed and the fitted to a polynomial model that represented the contributions of the block effects, main factor effects, two-factor interaction effects and main factor quadratic effects. A multi-factor ANOVA was used to perform a null hypothesis test on each of the regression coefficient in the model (Equation 1) to determine the level of statistical significance of the factor effects and interactions effects. Residual analyses and lack-of fit tests were performed to evaluate the suitability of the model.(Equation 1)Y=K+Tp+∑j=1Dβjxj+∑i=1D∑j=1Dβijxixj+∑j=1Dβjjxj2+ϵ In Equation 1, Y corresponded to the log-transformed values of cell expansion, K corresponded to the intercept, Tp was the constant corresponding to the pth block, D corresponded to the number of factors (D = 4), βj corresponded to the main effect coefficient for factor j, βij corresponded to the interaction effect coefficient between factors i and j, xj corresponded to the coded dose [-2, −1, 0, 1, 2] for factor j, xi corresponded to the coded dose [-2, −1, 0, 1, 2] for factor i, and ε corresponded to the random error (residuals).29 Flow cytometry Flow cytometry analysis of all cultures was done using an Attune Cytometer (Thermo Fisher Scientific, Nepean, Canada)) while cell sorting was carried out using BD FACS Aria III (BD Biosciences). A portion of cultured cells on day-7, 14 and fresh day-0 were analyzed by flow cytometry. Unless stated otherwise, all antibodies were from Becton Dickinson Pharmingen (Mississauga, Ontario, Canada). The antibodies used included CD19−phycoerythrine (PE, clone HIb19), CD14-PE (clone M5E2), CD45-allophycocyanine (APC, clone HI30), CD33-PE (clone WM53), CD34-PE (clone 581), CD38-APC (clone HIT2), EPCR-APC (clone RCR-401, Biologend, MA, USA), CD90-PECy7 (clone 5E10) and CD123-PerCP-Cy5.5 (clone 7G3), CD3-APC (clone UCHT1), human CD41a (GPIIb)-FITC (clone HIP8), CD56-PE (clone NCAM16.2), human AXL (Cat# AF154, R&D Systems). Cells were stained in FACS buffer (PBS+2% FBS) and stained with antibodies for 30 min on ice, washed and resuspended in FACS buffer. Antibodies were titrated before use while, fluorescent minus one controls were used to set gates and/or quadrants. For indicated cultures, the following subsets were purified by FACS; HSC-enriched (CD34+CD38−CD45RA-CD38-+CD49f+), lymphoid-primed multipotent progenitor cells (LMPP, CD34+CD38+CD45RA + CD90−CD10+), multipotent progenitors (MPP, CD34+CD38−CD45RA-CD10−), common myeloid progenitors (CMP, CD34+CD38+CD45RA-CD90−CD123low), granulocyte monocyte progenitors (GMP, CD34+CD38+CD45RA + CD90−CD123+), megakaryocyte-erythroid progenitors (MEP, CD34+CD38+CD90−CD45RA-CD123-).50,51,52 Intracellular staining Day 14 cultured cells were harvested, stained with CD34 antibody, then fixed using 4% PFA and then stained for CD45RA. Cells were then permeabilized with 0.3% Triton containing Halt Protease and Phosphatase Single use inhibitor cocktail (Thermo Scientific, Cat# 78442) and incubated with indicated antibodies: anti-AXL (Cat# AF154), anti-p-AXL (R&D systems, Cat# AF2228), anti-5-methylcytosine (clone 33D3, Millipore Sigma, Cat# MABE146), anti-H3K27Ac (Cell Signaling Technology, Catalog #15562), and anti-H3K4me2 (Abcam, Cat# 7766) and appropriate secondary antibody. Protein expression was determined by flow cytometry compared with an isotype only control. Data analysis was performed using FlowJo (v10.7, BD) to compare mean fluorescent intensity (MFI) values. In vivo transplantation assay All animal work was preapproved by the Animal Care Committee of University of Ottawa. For SCAC serial transplantation, 1,250 CD34+ NCs or the total progeny of 1,250 CD34+ starting cells were transplanted i.v. into irradiated (300 cGy) 8 weeks old NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG, Jackson Laboratory, Bar Harbor, USA) mice. Secondary recipients were transplanted with 80% of the BM collected from individual primary recipients. For LDA transplantation, fresh CD34+ NCs (250, 500, 1000 cells) or the progeny of 25, 100, 250, 500 starting CD34+ cells (day 0 equivalent) cultured for 14 days were transplanted along with a fixed dose of inactivated helper cells (200,000CB mononuclear cells irradiated at 15 Gy).30,60 All primary transplants were performed using female NSG mice. For GAS6 supplementation transplants, GAS6 (10 ng/mL) or GAS6 neutralizing antibody (5 μg/mL) was added to X2A cultures on day 10. On day 14, the total progeny of 750 starting CD34+ cells per mouse from the indicated cultures were transplanted into irradiated NSG mice. The lower cell dose was used to prevent engraftment saturation and was based on LDA experiments. Net total nucleated cell (TNC) doses in serial transplant assays are summarized in Table S5. Human engraftment analyses Human platelet and leukocytes were tracked in the murine blood from saphenous vein 3 to 18 weeks after transplantation at indicated time points. Murine blood was collected in heparin coated capillary and then transferred into heparin containing tube. Human platelet levels were measured following a single platform protocol61 (representative plot shown in Figure 4A). In short, 5 μL of blood was diluted 10 times with PBS and stained for 20 min at 4°C with 2 μL of human CD41a PE-conjugated (BD). After incubation, red cells were lysed by an incubation of 15 min at RT with 250 μL of 1X BD Pharm Lyse solution (BD). Samples were then diluted with 250 μl of PBS followed by addition of 50 μl of Accucheck Beads (Thermofisher). A fix volume of 450 μL per samples were acquired by FACS. For leukocytes, mouse blood was stained with human anti-CD45-APC, CD33-PE and CD19-FITC, then red cells lysed with 1X BD Pharm Lyse solution (BD), after which samples were analyzed by FACS (representative plot shown in Figure 4A). LT BM engraftment analysis for serial transplants was performed past 18 weeks as indicated. BM cells were harvested from hind legs, red blood cells lysed with 1X BD Pharm Lyse solution (BD) and BM cells stained with indicated antibodies. The antibodies used included CD45-APC, CD19-FITC, CD33-PE, CD45-FITC, CD3-APC, CD14-PE, CD34-PE, CD38-FITC, and CD45-FITC (Becton Dickinson Pharmingen, Mississauga, Ontario, Canada). LDA mice were sacrificed at 20-week post-transplantation and mice were considered positive for human engraftment when human CD45+CD33+ cells were ≥0.10%. The limiting dilution analyses were performed using the ELDA software from the Walter and Eliza Hall Institute of Medical Research software http://bioinf.wehi.edu.au/software/elda/index.html.53 Differences in HSC frequencies were analyzed by chi-square test. ∗p values < 0.05 were considered significant. Imaging flow cytometry Day 14 FACS sorted CD34+CD45RA-cells were fixed and stained with Hoechst 33342 (Thermofisher). Cells were then stained with AXL and p-AXL antibodies as described above cells were analyzed using the ImageStream imaging flow cytometer (Amnis). The Amnis IDEAS analysis software was used to measure the percentage double-positive (AXL+p-AXL+) and single positive (AXL+) cells. Apoptosis assay Apoptosis was assessed by the dual staining of Annexin V and SYTOX-AADvanced (Thermofisher). The analysis of the treated samples was performed consistently, using the same gating strategy throughout. Viable cells were categorized as being AnnexinV and 7-AAD negative, late apoptotic/dead cells were categorized as being AnnexinV and 7-AAD positive. Cell cycle assay Cells are harvested and cell surface stained at appropriate time points. Cells were permeabilized using the aforementioned protocol. Post permeabilization, the cells were treated with RNAse (Thermofisher) for 10 min, washed and stained with Sytox (Thermofisher) to assess cell cycle state. Furthermore, cells were also stained with Ki67 to separate G0 and G1 cell cycle states. RNA sequencing and data analysis RNA-sequencing was performed on 1 or 1.5 x 105 sorted CD34+CD45RA-cells obtained from different culture conditions after 14-day expansion and non-cultured cells (NC). RNA was isolated using miRNeasy Micro Kit (Qiagen, Hilden, Germany) including DNAse treatment. cDNA libraries were constructed using TruSeq mRNA stranded protocol (Illumina) and sequencing was performed using Illumina HiSeq 2000 instrument at StemCore Laboratories (Ottawa Hospital Research Institute) following their standard operating procedures. Reads were assigned to transcripts from the GENCODE v32 assembly (GRCh38) using Salmon v1.0 (https://combine-lab.github.io/salmon/),54 transcript quantifications were loaded into R using tximport,62 and expression was modeled using DESeq263 (v1.28.1). Transcript abundance was compared between pairs of conditions (two replicate pools per condition) and fold changes were calculated using the DESeq2 lfcShrink function applying the apeglm method.55 Multiple testing correction was performed using the Benjamini Hochberg method and significantly up- or down-regulated genes were identified based on false discovery rate (FDR) < 0.05. The sequencing data can be accessed using Gene Expression Omnibus using the accession number GSE202577. Gene Ontology (GO) and heatmap analysis GO analysis was performed using significantly differentially expressed genes (unless mentioned) using DAVID (version 6.8). Top terms identified within biological processes were reported and plotted via bubble plot analysis using GraphPad Prism v 8.0. Heat maps were generated using Morpheus software (Broad Institute). Reverse transcription and RT-qPCR RNAseq results were validated via RT-qPCR. 1 microgram of RNA was isolated from corresponding SCACs and converted into cDNA using the iScript Reverse Transcription Supermix for RT-qPCR (Cat#1708840) as per manufacturers protocol. The cDNA was then used to perform RT-qPCR using the SsoAdvanced universal SYBR green supermix (Cat: 1725270). The assays were done using the CFX96 touch real-time PCR detection system (BioRad). Primers presented in Tables S1B–S1D. Single cell RNA-seq (ScRNA-Seq) analysis pipeline cDNA libraries were sequenced on an Illumina HiSeq2500 at the Ottawa Hospital Research Institute (Ottawa, ON) at an estimated read depth of 11,000–18,000 reads/cell. The sequencing data can be accessed using Gene Expression Omnibus using the accession number GSE212578 Sequenced FASTQ reads were processed using Cell Ranger V3 software (cell ranger count function) with settings as recommended by the manufacturer (10X Genomics). Read alignment was performed using a reference Ensembl human genome (GRCh38). Following Cell Ranger data processing, digital gene expression (DGE) matrices were analyzed to determine the average numbers of genes and transcripts (UMIs) detected per cell using genes detected in at least three cells. X2A, C6 and NC consisted of 2104, 992 and 1,640 genes/cell and 6826,2278 and 5426 UMIs/cell, respectively. DGE matrices were processed through a previously published and well cited computational pipeline.64,65,66 FindMarker function in Suerat was used to identify genes enriched in the various clusters. Data processing pipeline Briefly, cells with low UMI counts, doublets, and cells with relatively high mitochondrial DNA content were removed. Cell cycle phase of each cell was predicted using the Cyclone method.67 Genes detected in less than three cells were removed and cell transcriptomes were subsequently normalized as described68 according to their pool and deconvolution normalization method in the scran R package in order to correct for differences in read depth and library size. Normalized expression matrices were imported into Seurat (1.4.0.16) where PCA was performed using highly variable genes in the data. The Seurat function Run-tSNE was used to generate 2-dimensional t-distributed stochastic neighbor embedding (tSNE) projections using the top principal components detected in the dataset. The same top principal components were subsequently used to iteratively carry out SNN-Cliq inspired clustering (FindClusters function in Seurat) at increasing resolutions ensuring the number of differentially expressed genes (determined by the FindMarkers function in Seurat; p < 0.01, family wise error rate [FWER], Holm method) between most similar clusters was greater or equal to 30 genes. ScRNA-seq HSPC subset analysis For the X2A dataset, clusters were assigned at resolution 1.5 (22 clusters identified with at least 1900 differentially expressed genes between most similar clusters, FWER <0.01). For the X2B dataset, clusters were assigned at resolution 1.5 (23 clusters identified with at least 700 differentially expressed genes between most similar clusters, FWER <0.01). For the C6 dataset, clusters were assigned at resolution 1.0 (22 clusters identified with at least 400 differentially expressed genes between most similar clusters, FWER <0.01). For the NC dataset, clusters were assigned at resolution 1.5 (23 clusters identified with at least 500 differentially expressed genes between most similar clusters, FWER <0.01). To generate the combined dataset, all cells from the X2A, X2B, C6 datasets were extracted from their respective raw DGE matrices and merged into one dataset and subsequently run through the pipeline. Cell cycle regression was performed on the same dataset using Cyclone. Clusters were assigned at a resolution of 1.0 (21 clusters identified with at least 2300 differentially expressed genes between most similar clusters, FWER <0.01). To generate the X2A/C6 HSPC combined dataset, X2A, and C6 HSPCs were extracted from their respective raw DE matrices and merged into one dataset and subsequently run through the pipeline. Cell cycle regression was performed on the same dataset using Cyclone. Clusters were assigned at a resolution of 0.10 (2 clusters identified). 1400 differentially expressed genes between most similar clusters, FWER <0.01). Validated markers for graft composition analyses Seurat’ FindMarker function identified 14 marker genes enriched in the clusters identified in SCAC-HSPC (Figures S5A–S5E). These markers were validated in day 14 FACS-sorted HSPC subpopulations from independent cultures using previously characterized cell surface markers (Figure S5A). RT-qPCR analysis revealed that AVP and PRSS2 were expressed at high levels within HSCe, DNTT and MME were expressed at high levels within LMPPs (Figures S4E and S4F). AZU1, ELANE and PRTN3 were expressed at highest within CMPs, IRF8, MPO and TCF4 within GMPs and finally, HBG1, HBG2, HPGDS and TPSB2 were highest within MEPs (Figures S4G–S4I). Moreover, these gene marker patterns were found to hold true with fresh NC cells (Figures S5C and S5D; Table S1D). Trajectory inference and pseudotime ordering Single cell pseudotime trajectories were constructed using a modified version of the dpFeature method in Monocle v2.56 In short, cell barcodes from desired datasets were extracted from the raw digital gene expression matrices and merged prior to normalization using Monocle’s size factor normalization method. PCA was performed using the highly variable genes that were obtained from the pipeline (data processing pipeline described above) and the cells were projected into 2-dimensional space using the tSNE algorithm. Cells were subsequently assigned into distinct clusters using Monocle’s density peak clustering algorithm. A set of ordering genes was obtained by testing each gene for differential expression between the clusters in the dataset and selecting the top 1500 significantly differentially expressed genes. Expression profiles were reduced to 2 dimensions using the DDRTree algorithm included in Monocle 2 and cells were ordered using these genes to obtain a trajectory. Gene expression was plotted over pseudotime using Monocle’s plot genes in pseudotime function in which the sm.ns function in the VGAM package was used to model gene expression over pseudotime as a continuous non-linear function. Cluster assignment statistical analysis Single cells were assigned to distinct clusters using SNN-Cliq-inspired clustering (FindClusters function in Seurat) at increasing resolutions ensuring the number of differentially expressed genes (determined by the FindMarkers function in Seurat) between most similar clusters was greater or equal to 30 genes. In this analysis, an adjusted p value (FWER) smaller than 0.01 was considered statistically significant (Bonferroni-Holm method). Cluster Validation For SCACs, HSC-enriched [HSCe = HSC+MPP] (HSC: CD34+CD45RA-CD90+CD49f+; MPPs: CD34+CD45RA-CD90−), LMPPs (CD34+CD45RA + CD90−), CMPs (CD34+CD45RA-CD123+), GMPs (CD34+CD45RA + CD123+) and MEPs (CD34+CD45RA-CD123-), were FACS sorted. RNA was extracted from 20,000 FACS sorted cells using PicoPure RNA isolation kit (Cat# KIT0204) and RT-qPCR analysis was performed using total extracted RNA. For freshly isolated non cultured (NC) samples, lineage depleted HSPCs were isolated from umbilical cord blood using the manufacturers protocol (RosetteSep, Cat# 15276). Lineage depleted cells were stained and FACS sorted for the various populations: HSCe (HSC: CD34+CD38−CD45RA-CD90+CD49f+, MPP:CD34+CD38−CD45RA-CD90−), LMPPs (CD34+CD38−CD45RA + CD90−), CMPs (CD34+CD38+CD45RA-CD123+), GMPs (CD34+CD38+CD45RA + CD123+) and MEPs (CD34+CD38+CD45RA-CD123-). RNA was extracted from 20,000 FACS (∼200 for HSCs) sorted cells using PicoPure RNA isolation kit (Cat# KIT0204) and RT-qPCR analysis was performed using total extracted RNA. For HSCe subpopulations, equal number of HSCs and MPPs were combined prior to RNA extraction. Please refer Table S1D for information pertaining to genes tested and primers used. DNA methylation qPCR Methylation analyses were performed using the OneStep qMethyl Kit (Zymo Research, Cat# D5310) according to manufacturer’s instructions. Briefly, 250,000 CD34 + 45RA-cells cultured in the different SCACs were sorted on day 7 and day 14 and DNA was isolated. DNA was digested according to the manufacturer’s instructions and qPCR was performed using the BioRad CFX96 instrument. The primers used for qPCR targeted the CpG islands near the promoter of our target genes. The primers are listed in extended Data Table (Table S1D). Quantification and statistical analysis Data are shown as mean values ±standard error of the mean (SEM) unless stated otherwise. Statistical analyses (other than that of the factorial design experiments) were performed using GraphPad Prism v 8.0. Two-tailed Student’s t-tests were performed for statistical analysis between two groups. One-way or two-way ANOVA was used to compare the difference in means between more than two groups where indicated. A p value of less than 0.05 was considered as statistically significant. Only statistically significant differences are indicated in figures using indicated symbols. Statistics used to test differential gene expression in the ScRNA-Seq data was performed using the Seurat FindMarkers function. An adjusted p value (FWER) smaller than 0.01 was considered statistically significant (Bonferroni-Holm method).