![[Düzenle]](?edit=/home/u514654129/domains/awesamohealth.com/public_html/wp-content/themes/pixwell/screenshot.png){kind=link}
![[Yeniden Adlandır]](?rename=/home/u514654129/domains/awesamohealth.com/public_html/wp-content/themes/pixwell/screenshot.png){kind=link}
Hematopoietic cells from differently oxygenated niches exhibit distinct properties
Each hematopoietic niche exists at different O2 tensions (Table 1) [6,7,8,9,10,11]. The BM ranges from <1–6% O2 while peripheral blood (PB) total oxygenation lies between 4 and 14% (with location dependent levels of hemoglobin-bound or dissolved O2 [31]) and is different in arterial or venous blood. CB exhibits ranges from 0.5–6% O2 [7, 8], though possibly up to 10% (this study, Fig. 1D). Other tissues where HSCs/HPCs reside like the lung are highly oxygenated, up to 17% [11, 32]. We infer that HSC/HPC responses to O2 could be a consequence of specialization of the tissue they reside in. To test differences in HSCs/HPCs from differently oxygenated niches, we performed ex vivo characterization of HSCs/HPCs from human CB, mPB, or BM CD34+ cells using immunophenotypic analysis by cell surface staining with flow cytometry and colony forming unit (CFU) assays (Fig. 1A–C) [33, 34]. This showed that the HSC/primitive HPC enriched CD34 + CD38- fraction is higher in mPB and CB compared to BM (Fig. 1A). There was also a significantly higher frequency of multipotent progenitors (MPPs) and multilymphoid progenitors (MLPs) in CB compared to BM (Fig. 1A, B). Conversely, BM contains a higher proportion of more lineage determined CD34 + CD38+ HPCs, with a significantly higher granulocyte-macrophage progenitors (GMPs) frequency compared to mPB (Fig. 1A, B). CB and mPB show increased trends of functional HPCs, demonstrated by numbers of CFU-granulocyte, macrophage (CFU-GM) and CFU-granulocyte, erythroid, macrophage, megakaryocyte (CFU-GEMM) compared to BM (Fig. 1B, C). These data suggest that CD34+ cells exhibit different proportions of HSCs/HPC subpopulations when isolated from different niches.
A Immunophenotypically defined HSCs/HPCs or B CFU numbers as a fraction of CD34+ cells from BM (n = 3), mPB (n = 3), and CB (n = 2). C Significance analysis of A using 1-way ANOVA with post hoc comparisons. D Dissolved O2 concentration in fresh whole CBUs compared to total volume of the CBU. Dissolved O2 levels associated with physiologically relevant local and atmospheric O2 tensions are indicated by dashed lines. E–G CFU from CB (n = 8), BM (n = 3), or mPB (n = 3) CD34+ cells performed in the indicated O2 tensions. 12 days after plating CFUs of different potentials were enumerated manually by counting. H CB (n = 5), BM (n = 3), or mPB (n = 3) CD34+ cells expanded in serum free media with growth factors for 7 days in the indicated O2 tensions. Total CD34+ cell numbers compared to the number of cells seeded by flow cytometry analysis. Stats: one-way ANOVA controlling for donor unit (indicated by p-value) with post hoc Tukey or Sidak comparisons. *p < 0.05. HSCs hematopoietic stem cells; HPCs hematopoietic progenitor cells; CFU colony forming unit; BM bone marrow; mPB mobilized peripheral blood; CB umbilical cord blood; O2 oxygen.
We have previously suggested that donor hematopoietic cells used in therapy and research are necessarily exposed to ambient O2 levels upon harvest and thus lose their relative hypoxic states [15,16,17, 35]. To test this directly, we measured the dissolved O2 concentration in > 70 fresh CBUs that were never frozen and were measured within 1 week of isolation. To our surprise, the dissolved O2 level of CBUs collected in preservation bags remains hypoxic relative to fully oxygenated liquids (Fig. 1D). The level of dissolved O2 directly correlates with the volume of the unit. This is likely the result of the physiologic oxygenation state of the blood being slightly diluted by the addition of fully oxygenated anti-coagulant preservatives in each CBU, which impacts low collection volumes the most. Thus, CB remains lowly oxygenated after collection. We thus focus on CB as a primary model for studying O2 dependency. We also examine BM and mPB to explore differential O2 sensitivity of HSCs/HPCs from different donor sources, inferring that their exposure to extra physiologic O2 is likely similar to that of CB.
Hematopoietic cells exhibit O2 dependency ex vivo
Differently oxygenated hematopoietic niches have non-O2 related environmental factors that could account for changes to cellular properties. We thus sought to test the effects of O2 on hematopoietic cell growth and function in a single variable system using oxygen-controlled incubators. We maintained CD34+ cells in 1%, 3%, 5%, 14%, and 21% O2, which spans the range of O2 levels in tissues in which hematopoietic cells reside, migrate through, or are processed (Table 1). First, we measured the CFU potential of the CD34+ cell fraction and found that total CFU, CFU-GM, and CFU-GEMM numbers from CB, BM, and mPB are significantly altered by local O2 tension. Total CB CFU numbers are significantly higher in 5% O2 compared to 14% and 21% and CB CFU-GM numbers are higher in 5% compared to 1% (Fig. 1E), while BM and mPB CFU numbers show direct correlation to O2 tension, with increased CFU-GM, CFU-GEMM, and total CFU at 5%, 14%, and 21% O2 compared to lower tensions (Fig. 1F, G). Burst forming unit-erythroid (BFU-E) numbers were unaffected by variable O2. Next, we cultured CD34+ cells in serum free media with growth factors for 7 days to measure their proliferative capacity. CB and mPB CD34+ cells exhibit increased proliferation with increased O2 tension (Fig. 1H). Interestingly, at all tensions except 5% O2, CD34+ cells from mPB and/or CB exhibited significantly higher proliferation compared to BM.
Multipotent hematopoietic cells exhibit preferential expansion in different O2 ranges compared to lineage determined HPCs
CD34+ cells are heterogenous, so we next used immunophenotypic analysis following culture to determine how different HSC/HPC subpopulations expand in variable O2. For cell fractions derived from CB and mPB but not BM, total nucleated cells, total CD34+ cells and the multipotent cell enriched CD34 + CD38- fraction, and the lineage defined progenitor enriched CD34 + CD38+ fraction exhibited significantly increased expansion in increased O2 with higher total cell numbers at 5%, 14%, and 21% O2 compared to 1%, 3% and input (Fig. 2A–C; Fig. S1A–I). HSC growth was significantly affected by O2 tension with peak expanded cell numbers in 5% O2 in cells from mPB and BM, while HSCs from CB exhibited a similar trend (Fig. 2D–F). MPPs from all sources are also sensitive to changing O2, but the ideal growth oxygenation varies between CB, BM, and mPB (Fig. 2G–I). Interestingly, when combining all donor sources HSCs exhibit the highest frequency of HSCs in 5% O2, while MPP frequencies are highest in 1–3%, suggesting these cell types preferentially proliferate in different local O2 levels (Fig. 2J, K). An important caveat to immunophenotyping expanded cell populations is that cell surface markers may change in culture or as a result of changing local oxygenation. We thus sorted immunophenotypically defined HSCs/HPCs from CB prior to any ex vivo cell culture and seeded these pure populations in serum free media with growth factors at 1%, 5%, or 21% O2. Consistent with the previous results, only 5% O2 supported significant expansion compared to unmanipulated control input cells (Fig. 2L), and sorted MPPs showed trends toward higher expansion with higher oxygenation (Fig. S2A).
A–I CB (n = 5), BM (n = 3), or mPB (n = 3) CD34+ cells were expanded in serum free media with growth factors for 7 days in the indicated O2 tensions. Cells were then analyzed for enumeration of total numbers of the indicated immunophenotypically defined population by flow cytometry. J, K Percent of indicated immunophenotypic HSC/HPC population as a fraction of total cells in the well after 7 days of expansion. L CB HSCs were sorted prospectively from CD34+ cells prior to any culture. Due to cell number limitations, HSC sorted populations were pooled so that each expansion was seeded as a pool of two distinct CB units. Sorted pools were then expanded in serum free media with growth factors for 7 days in the indicated O2 tensions. Cells were analyzed for total viable cell expansion by counting using a hemacytometer (n = 4 CB pools). Two independent replicates from the HSC analysis were excluded due to detection of a contaminating cell population. This contamination was identified by outlier levels of total expansion (over 200-fold expansion specifically in 5%) combined with flow cytometry confirming 0% CD34+ cells in the expanded cell population. Stats: one-way ANOVA (indicated by p-value) with post hoc comparisons indicated by compact letter display (in compact letter display, groups that are not statistically different are indicated by matching letters). Each point/color indicates distinct biological donor sources as indicated by the included legends. HSCs hematopoietic stem cells; HPCs hematopoietic progenitor cells; CFU colony forming unit; BM bone marrow; mPB mobilized peripheral blood; CB umbilical cord blood; O2 oxygen; UnCon Unmanipulated control (input).
Lineage defined progenitors including MLPs, GMPs, and common myeloid progenitors (CMPs) expanded significantly more in high O2 tensions with CB showing the significantly higher expansion in 5-21% compared to lower tensions with similar trends in BM and mPB (Fig. 3A–F, Fig. S1J–L). Megakaryocyte-erythroid progenitor (MEP) expansion was not significantly affected by O2 tension (Fig. S1M–O). In combining all donor sources, frequencies of MLPs were lowest in 5% O2 but similar in all other tensions, while GMP proportions were highest in 14-21% O2, showing still different O2 dependencies from HSCs/MPPs (Fig. 3G, H). Sorted lineage defined MEPs showed significantly higher expansion in 21% O2 compared to lower tensions, and other HPCs including GMPs, MLPs, and CMPs exhibit a similar trend consistent with total CD34+ expansions (Fig. 3I, S2B–D). Interestingly, 1% O2 almost universally allowed seeded cells to retain the highest percent of initial immunophenotype (Fig. S2E–J). For example, sorted MPPs yielded the highest frequency of MPPs after 7 days of expansion in 1%. HSCs were the sole exception, which were best maintained even proportionally in 5% O2. To test the function of expanded HPCs, we performed CFU assays on CD34+ cells expanded from CB. CFU-GM and CFU-GEMM both expanded more in liquid culture at 5, 14, and 21% compared to lower O2 tensions (Fig. 3J, K). Together, these data show that immunophenotypic mature progenitors and functional CFUs exhibit increased expansion in higher O2 levels, though responses vary slightly with donor source. Importantly, HSCs and more primitive HPCs maintain higher proportional representation in the expanded pool of cells in low physiologic O2, though this varies by cell type.
A–F CB (n = 5), BM (n = 3), or mPB (n = 3) CD34+ cells were expanded in serum free media with growth factors for 7 days in the indicated O2 tensions. Cells were then analyzed for enumeration of total numbers of the indicated immunophenotypically defined population by flow cytometry. G,H Percent of indicated immunophenotypic HPC population as a fraction of total cells in the well after 7 days of expansion. I Cord blood (CB) GMPs were sorted prospectively from CD34+ cells prior to any culture. Due to cell number limitations, GMP sorted populations were pooled so that each expansion was seeded as a pool of two distinct CB units. Sorted pools were then expanded in serum free media with growth factors for 7 days in the indicated O2 tensions. Cells were analyzed for total viable cell expansion by counting using a hemacytometer (n = 6 CB pools). J, K CB CD34+ cells (n = 7) were expanded in serum free media with growth factors for 7 days in the indicated O2 tensions. Cells were then plated in CFU assays at the same tension (5%). 12 days after plating CFUs of different potentials were enumerated manually by counting. Stats: one-way ANOVA (indicated by p-value) with post hoc comparisons indicated by compact letter display (in compact letter display, groups that are not statistically different are indicated by matching letters). Each point/color indicates distinct biological donor sources as indicated by the included legends. HSCs hematopoietic stem cells; HPCs hematopoietic progenitor cells; CFU colony forming unit; BM bone marrow; mPB mobilized peripheral blood; CB umbilical cord blood; O2 oxygen; UnCon Unmanipulated control (input).
CD34+ expansion in variable O2 affects in vivo hematopoietic engraftment
Ex vivo analyses do not always recapitulate the in vivo function of human HSCs/HPCs, thus in vivo mouse models are required for full evaluation of potency. We examined the ability of cells expanded at variable O2 levels to repopulate a hematopoietic system in vivo (Fig. 4A). CD34+ cells were expanded in five O2 tensions for 7 days in serum free media with growth factors. Unmanipulated “input” cells (2.5×104 cells) or input cells plus their expanded progeny (variable numbers based on total cell expansion) were transplanted to NSG mice. While total human engraftment was not significantly different at 2 weeks post transplantation (Fig. 4B), early human neutrophil recovery measured in mouse PB, an analogous measure of successful patient transplantation, was increased when cells were expanded in low physiologic O2 tensions (1-5%) compared to input cells (Fig. 4C). When correcting for differential cell numbers transplanted due to differential expansion, early neutrophil recovery as a fraction of transplanted CD34+ cells was significantly higher in 1% O2 than in all higher tensions (Fig. 4D). 6 weeks post transplantation, mice transplanted with cells expanded at 1% O2 showed lower total engraftment in the PB compared to higher tensions (Fig. 4E), a transient effect no longer observed by 10 weeks post-transplantation (Fig. 4F), but which remained a trend in the BM at 16 weeks post transplantation (Fig. 4G). At 10 weeks, mice transplanted with expanded cells at all O2 tensions had significantly higher human myeloid/lymphoid ratio in their PB compared to those transplanted with unmanipulated input cells, with mice transplanted with cells expanded at 1% trending toward the most “balanced” ratio (Fig. 4H). No differences in myeloid/lymphoid balance were observed in the BM 16 weeks post transplantation, indicating either a transient effect or one that can only be observed in the PB (Fig. 4I). In secondary engraftment assays, CD34+ cells initially expanded at 14% O2 yielded the highest engraftment and was the only condition where all mice had observable secondary engraftment (Fig. 4J, data not shown). Together, these data show that different cell populations are responsible for different engraftment kinetics, such as early myeloid recovery and long-term reconstitution, and that these subpopulations are variably affected by physiologic levels of O2.
A Experimental scheme for in vivo studies. B–J CB CD34+ (n = 3) cells expanded in serum free media with growth factors for 7 days in the indicated O2 tensions. Input cells (25,000/ mouse) or input cells plus expanded progeny were transplanted to sublethally irradiated NSG mice (n = 9 each group). B,C Mice were monitored at 2 weeks for total human CD45+ chimerism and neutrophil recovery in PB by flow cytometry. D Neutrophil recovery percentage as a fraction of total cells transplanted (25,000 input cells + expanded progeny). E,F Mice were monitored at 6 and 10 weeks for total human CD45+ chimerism in PB by flow cytometry. G Mice were monitored at 16 weeks for total human CD45+ chimerism in BM by flow cytometry. H,I Mice were monitored at 10 weeks in the PB and 16 weeks in the BM for myeloid/lymphoid recovery by flow cytometry. Myeloid cells were defined by CD33+ cells while lymphoid cells were positive for CD3 and/or CD19. J Pooled BM cells from the primary transplant recipients were injected into sublethally irradiated secondary recipients (n = 5 each group). Engraftment was monitored at 16 weeks post transplantation in the BM of secondary recipients. Stats: one-way ANOVA (indicated by p-value) with post hoc comparisons indicated by compact letter display (in compact letter display, groups that are not statistically different are indicated by matching letters). Each color indicates distinct biological donor sources as indicated by the included legends. Each point indicates a recipient mouse. Notes on mouse numbers: the following groups lost mice due to experimental conditions (likely bone marrow failure or infection) during the course of the transplantation experiment: 1%- 1 mouse; 3%- 3 mice; 5%- 0 mice; 14%- 0 mice; 21%- 2 mice. 5 mice were lost from the 1% group due to husbandry complications unrelated to the experimental conditions. HSCs hematopoietic stem cells; HPCs hematopoietic progenitor cells; CFU colony forming unit; BM bone marrow; PB peripheral blood; CB umbilical cord blood; O2 oxygen; UnCon Unmanipulated control (input).
Expansion in variable O2 alters the hematopoietic transcriptome
Our in vivo data lead us to infer that local O2 affects HSC/HPC potency, particularly in cells responsible for early immune reconstitution and those that retain a more balanced myeloid/lymphoid ratio. We sought to identify subclusters of cells associated with these alterations in functional potency and determine the molecular effects of local O2 tensions on various subpopulations of HSCs/HPCs. We performed single cell RNA sequencing (scRNA-seq) under oxygen-controlled conditions using PIP-seq [30]. A total of four CBUs were expanded for 2 or 7 days at 5 O2 levels or were sequenced unmanipulated as input cells. The majority (60%) of unmanipulated CD34+ cells were transcriptomically distinct from cultured cells (Fig. 5A, Fig. S3A, Table S1) and were separated and sub-clustered for higher resolution (Fig. 5A, Fig. S3B).
A–E CB CD34+ cells expanded in serum free media with growth factors for 2 days (n = 2) or 7 days (n = 2) in the indicated O2 tensions, or unmanipulated input cells (n = 2) were subjected to scRNA-seq using particle-templated instant partition sequencing (PIP-seq). A UMAP showing annotated clusters from all cells and conditions. B UMAP plots with cells from the indicated expansion condition highlighted. C Dot Plot showing top three gene markers used to annotate HSC/HPC clusters. D,E Top differentially expressed genes from the comparison of 1 vs 21% for the indicated cell populations after 2 days (D) or 7 days (E) in culture. Annotated are the genes with known proximal HIF binding sites in hematopoietic cell lines. Genes are only shown if they were significantly different (padj<0.05) in pseudobulked differential expression analysis. The full list of genes can be found in Table S2. HSCs hematopoietic stem cells; HPCs hematopoietic progenitor cells; CB umbilical cord blood; O2 oxygen; UMAP uniform manifold approximation and projection; HIF hypoxia inducible factor.
Our final single cell object consisted of 26 annotated clusters of cells (Fig. 5A, Fig. S4, Table S2). Clusters comprised primarily of unmanipulated CBUs were termed “unexpanded” HSCs, MPPs, myeloid MPPs, CMPs, GMPs, and CMP-GMP transitioning. Other than unmanipulated input, only cells expanded at 1% O2 for 2 days maintained significant proportions of “unexpanded” transcriptomic characteristics (Fig. 5B). By day 7 of expansion, few cells in any expanded group were found in these clusters, though low O2 tensions still maintained higher a higher proportion of cells with these molecular characteristics (Table S2). This confirms that ex vivo expansion of CD34+ cells dramatically alters their transcriptomic profile, but this can be mitigated for short time periods by maintaining the cells at very low physiologic (1%) O2. We annotated cells enriched for primitive HSC/MPP markers in expanded conditions on day 2 or day 7 and termed them “expanded early HSC/MPP” (Fig. 5C). Notably, CD34+ cells expanded at lower O2 tensions (1–5%) exhibit a higher proportion of these cells after the full seven days of expansion compared to input and higher O2 levels (Table S2).
We examined genes that are altered in response to differences in local O2 tension, using 1% O2 and 21% O2 as differential conditions. We focused on two cell populations of interest comprised of primitive HSCs/HPCs that contain the functionally potent cells we seek to assess and that include sufficient cell numbers from each condition for differential analysis. The first was a pooled population defined by characteristics of unmanipulated CD34+ cells containing “Unexpanded” HSCs, MPPs, myeloid MPPs, CMPs, GMPs, and CMP-GMP along with MEPs, MKPs, MLPs, and ERPs (which do not clustered separately based on cell culture conditions). We also analyzed the individual cluster “Expanded Early HSCs/MPPs”. In the “Expanded Early HSCs/MPPs” cluster and the “Unexpanded” pooled population, 5 and 51 genes, respectively, were significantly altered by oxygen tension after 2 days in culture, while 16 and 79 genes were altered after 7 days of culture (Fig. 5D-E representative genes, Table S2 full list). Given the importance of Hypoxia inducible factor (HIF) proteins in regulating hypoxia responses, we examined differentially expressed genes for predicted regulation by HIF1a and HIF2a by mining previously published ChIP-seq data [36]. Across the two analyzed cell populations, 29.7% of differentially expressed genes are bound by HIF1a, with nearly half of those being co-bound by HIF2a (Fig. 5D-E, Table S2). Most (>90%) of the HIF peaks localized to genes that were only altered after 7 days of culture (Fig. 5E). These findings suggest that HIF contributes to the regulation of a subset of O2-dependent genes in these HSCs/HPCs subpopulations and that its influence becomes more apparent during longer-term culture. Interestingly, we observed that “Early Expanded HSCs/MPPs” exhibit O2-dose-dependent gene expression responses up to 5%, but at 5% O2 and above, they show similar expression patterns (Fig. 6A). We infer that in this cell population specifically there are molecular switches in the low physiologic O2 space somewhere between 3 and 5% O2 responsible for oxygen-dependent gene expression regulation that may or may not be HIF-dependent. By western blot we confirmed changes in expression in bulk CD34+ cells at the protein level of select O2 dependent molecules including enolase 1 (ENO1), BCL2 interacting protein 3 (BNIP3), and lactate dehydrogenase A (LDHA) (Fig. 6B, S5A–C). All together, these analyses reveal O2 dependent genes associated with potent subpopulations of cells that may be important for HSC/HPC functions.
A Representative dot plot showing genes that are up and downregulated in low and high O2 tensions after 7 days of culture in the Early Expanded HSCs/MPPs subcluster of cells. B Representative western blot of whole cell lysates using antibodies targeted against indicated proteins after 7 day expansion in the shown O2 conditions. Densitometry analysis of band intensity was performed and is calculated relative to the 21% bands from the same antibody blot, relative to ACTB as a normalizing housekeeping protein. The numbers from this analysis are under the images of the western blot. C,D Top ranked gene sets from GSEA analysis from the indicated cell populations cultured for 7 days in 1% vs 21% O2. Gene sets were first filtered by significance (padj<0.05), were sorted by normalized enrichment score, and were then reduced for redundancy by removing gene sets with >50% intersection. Top five upregulated gene sets are shown; only one downregulated gene set passed the statistic filters. The full list of gene sets can be found in Table S2. E–G CB CD34+ cells expanded in serum free media with growth factors for 2 days (n = 2) or 7 days (n = 2) in the indicated O2 tensions, or unmanipulated input cells (n = 2) were subjected to scRNA-seq using particle-templated instant partition sequencing (PIP-seq). E,F Representative GSEA plots showing the enrichment for the indicated gene programs in 1% versus 21% O2 in the subcluster Early Expanded HSCs/MPPs. G Transcriptomic prediction of cell cycle status for all cells in the 2-day expanded groups for the indicated O2 tensions. H–M CB CD34+ cells expanded in serum free media with growth factors for 2 or 7 days in the indicated O2 tensions. H Cell cycle analysis using Ki67/DAPI flow cytometry analysis after 2 days expansion (n = 3). I Intracellular reactive O2 species (ROS) measurement after 2 days expansion using CM-H2DCFDA ROS indicator and flow cytometry analysis (n = 5). J–L Seahorse metabolic flux analysis after 7 days expansion using J) ATP rate assay to show ATP production (n = 4) or K,L mitochondrial stress test (n = 4) to show calculated basal and maximum respiration rates. M CD34+ cells were treated with 0.02 µM cytarabine (n = 3) for 2 days in variable O2 tensions and counted by Trypan Blue viability staining. Stats: one-way or two-way ANOVA (indicated by p-value) with post hoc comparisons indicated by compact letter display (in compact letter display, groups that are not statistically different are indicated by matching letters). Colors used are described in figure legends. HSCs hematopoietic stem cells; HPCs hematopoietic progenitor cells; CB umbilical cord blood; O2 oxygen; IB immunoblot; GSEA gene set enrichment analysis; ROS reactive oxygen species.
O2 affects hematopoietic cell stress
We next performed unbiased gene set enrichment analyses (GSEA). To reduce redundancy inherent in these results, we reported the top gene sets with positive and negative enrichment scores that had less than 50% overlap in their leading edge genes with other top gene sets (Fig. 6C, D). As expected, the most enriched gene programs in cells grown under low O2 were associated with hypoxia, and this gene program was redundant with specific pathways including HIF [37] and mechanistic target of rapamycin complex (MTORc) [38] signaling (Fig. 6C, D, S5D, E, Table S2). Interestingly, gene programs associated with the MDM4 regulator of p53 (MDM4) pathway (which has recently been implicated in hematopoietic cell potency [39]), as well as specific types of immune function and/or differentiation states (e.g., thymic macrophage vs epithelial macrophage; dendritic Cell stimulation), were also enriched in cells grown at 1% O2. These data suggest that hematopoietic cell function and differentiation are influenced by local O2 tension. In contrast, cells expanded in 21% exhibited transcriptomes enriched for genes associated with oxidative phosphorylation, which overlapped with gene programs associated with oxidative stress and mitochondrial functions, such as ATP synthesis. This pattern suggests an underlying accumulation of acute metabolic stress in cells grown at higher O2 tensions (Fig. 6C–F, S5D, E). Finally, cells grown at lower O2 levels are predicted by their transcriptomes to have a higher proportion of cells in G1 (Fig. 6G), which may also indicate quiescence.
Based on these findings, we tested whether hematopoietic cells exhibit changes in stress associated properties in different local O2 tensions. By flow cytometry, we analyzed the cell cycle status of CD34+ cells grown in variable O2 tensions. Cells grown in 1% and 3% O2 for 2 days had increased proportions in G0 compared to higher O2 tensions (Fig. 6H), confirming the transcriptomic prediction. CD34+ cells grown in 5%, 14%, and 21% O2 also exhibited increased accumulation of intracellular reactive O2 species, a marker for oxidative stress (Fig. 6I). Given the oxygen-dependent effects on reactive O2 species and mitochondrial gene expression (Fig. 6A, E, F, I), we next sought to examine the O2 associated metabolic properties of HSCs/HPCs using the Seahorse metabolic flux analyzer. ATP rate assays showed that CD34+ cells generate significantly more ATP from glycolysis in lower O2 tensions (1%, 3%, and 5%) compared to 14% and 21%, at the expense of ATP generated from oxidative phosphorylation (Fig. 6J). Cells grown in 1% had the lowest ratio of oxidative phosphorylation to glycolysis compared to all other expansion conditions (Fig. 6J). To further examine the oxidative phosphorylation pathway, we examined the effects of inducing acute stress using the mitochondrial stress test (Fig. S6A). CD34+ cells exhibited oxygen-dose dependent effects on mitochondrial respiration, with cells grown at 1% O2 exhibiting significantly lower basal and maximal respiration rates as well as lower spare respiratory capacity compared to 14% and 21% (Fig. 6K-L, Figure S6A).
These data show that CD34+ cells grown at higher O2 tensions are more metabolically active, more proliferative, and accumulate more stress related molecular properties compared to those maintained in lower O2. Thus, we hypothesized that CD34+ cells in low physiologic O2 are more resistant to acute toxicities. To test this, we treated CD34+ cells with the broadly cytotoxic chemotherapy cytarabine in variable O2 levels. CD34+ cells grown in low O2 tensions, particularly 1% oxygen, were highly resistant to cytarabine induced cell death compared to higher O2 levels, as indicated by a significantly higher percentage of viable cell number after 2 days of treatment compared to vehicle (Fig. 6M) and higher expansion over time in the presence of cytarabine (Fig. S6B). These data provide insights into hematopoietic toxicities following chemotherapy and resistance of malignant hematopoietic cells to treatment.
O2 dependencies are partially regulated through MTORc
Several genes and gene programs that exhibit stabilized expression in low O2 tensions in HPC populations are linked to MTORc signaling, (Figs. 5D, E, 6A–D), which regulates processes critical to HSC/HPC fate including mitochondrial function, biogenesis, and degradation [40]. Paradoxically, genes altered by low O2 include both repressors of MTORc signaling (BNIP3 and DNA damage inducible transcript 4 (DDIT4)) [41,42,43] and genes activated by MTORc (LDHA and ENO1) [44, 45]. We sought to determine the degree to which this pathway may contribute to the observed O2-dependent phenotypes. We found by western blot that canonical targets of MTOR, S6K [46] and 4E-BP1 [40] exhibited inconsistent trends toward increased phosphorylation with higher oxygenation (Fig. 7A, S7A). Despite this, CD34+ cells grown in high O2 in the presence of MTORc inhibitor Rapamycin consistently showed significantly decreased expansion of total nucleated cells (Fig. 7B) and modest, though significant, increases in proportions of the potent HSC/HPC enriched CD38- population (Fig. 7C). Consistent with an MTOR role in hematopoietic potency, published work showed that Rapamycin treatment increases in vivo engraftment of human CD34+ cells expanded in ambient air [47]. However, Rapamycin treatment did not affect O2-dependent mitochondrial respiration measured by Seahorse and did not phenocopy low O2 stabilization of BNIP3, LDHA, or ENO1 at the transcription level (Fig. S7B–I). Taken together, our data suggest that MTOR plays a partial role in the observed O2-dependent phenotypes, and that its effects on hematopoietic potency may be mediated through non-canonical targets. The modest effects observed with MTOR inhibition further suggest that a dynamic interplay among multiple regulatory- including but not limited to pathways regulated by HIF, MDM4, and MTORc revealed by our single cell transcriptomic data- contributes to molecular O2 sensing and the regulation of hematopoiesis (Fig. 6C, D). Full mechanistic investigation of each O2 associated pathway is warranted in future studies.
A Representative western blot of whole cell lysates using antibodies targeted against indicated proteins after expansion in the shown O2 conditions. Densitometry analysis of phosphor 4E-BP1 band intensity was performed and is calculated relative to total 4E-BP1. The numbers from this analysis are under the images of the representative western blot, and all replicates (n = 3 cord blood units) are shown in the graph underneath the western blot. B Total nucleated cells and C CD34 + CD38- cells were enumerated by flow cytometry after 7 days of treatment with 25 nM Rapamycin in the indicated O2 tensions. D Feature plot showing PRSS2 expression in primitive HSCs/HPCs clusters from scRNA-seq. E Expression of PRSS2 in different O2 tensions in HSCs/MPPs by scRNA-seq. F qPCR measuring PRSS2 expression relative to 18S housekeeping gene of 3 technical replicates of CD34+ cells from one CBU grown at the indicated oxygen tensions. G Expression of PRSS2 in immunophenotypically defined and sorted HSCs/HPCs by bulk RNA-seq. H Expression of PRSS2 in HSC/HPC populations from a publicly available database. I,J PRSS2 expression in HSCs correlated with long-term engraftment capacity in a mouse model of human CB transplantation. DESeq2 used for significance of differential expression analysis. Stats: (B,C) one-way or two-way ANOVA (indicated by p-value) with post hoc comparisons indicated by compact letter display (in compact letter display, groups that are not statistically different are indicated by matching letters). Colors used are described in figure legends. G/J DESeq2 statistical analysis. H one-way ANOVA (indicated by p-value) with post hoc comparisons indicated by compact letter display (in compact letter display, groups that are not statistically different are indicated by matching letters). HSCs hematopoietic stem cells; HPCs hematopoietic progenitor cells; CB umbilical cord blood; O2 oxygen; UnCon Unmanipulated control (input).
O2 dependent transcripts predict engraftment outcomes
We inferred based on the known phenomena of HSCs/HPCs losing overall engraftment potential during long-term ex vivo culture that “unexpanded” HSC/HPC clusters are enriched with functionally potent cells. Interestingly, in addition to known markers THY1 and AVP [48, 49], unexpanded HSCs also exhibited high expression of serine protease 2 (PRSS2), which encodes pancreatic Trypsin 2 protein and has well described digestive functions but does not have a well characterized hematopoietic function [50, 51]. The role of PRSS2 in HSC/HPC biology is unclear, though its hematopoietic expression has been reported. PRSS2 expression is primarily confined to unexpanded cell clusters as well as “Expanded Early HSCs/MPPs” cluster, which we predict contains cells responsible for engraftment from expansion conditions (Fig. 7D). Further analysis of scRNA-seq data as well as qPCR data shows that PRSS2 is stabilized in low O2, consistent with it being associated with increased HSC/HPC potency (Fig. 7E-F). To confirm expression of PRSS2 in HSCs/HPCs, we sequenced immunophenotypically defined HSC/HPC subpopulations from freshly isolated CBUs and found that PRSS2 is most highly expressed in MLPs followed by HSCs (Fig. 7G, Table S3). In independently published data, HSCs exhibit significantly higher expression of PRSS2 compared to mature HPC populations (Fig. 7H). We hypothesized that PRSS2 might be a transcriptional marker of HSC potency. To test this, we performed bulk RNA-seq on HSCs from biologically distinct CBUs with known engraftment outcomes in mouse models of transplantation (Fig. 7I; Table S4,S5), modeling the CD34+ and HSC transcriptomes as a function of SCID repopulating cell (SRC) frequencies. These data revealed that high PRSS2 expression in HSCs accurately predicts high SRC frequency (Fig. 7J). Notably, the MTORc regulator DDIT4 also trended toward higher expression in CBUs with higher SRC frequencies, but only in the broader CD34+ population (Figure S8). Whether these genes regulate hematopoiesis in an O2-dependent manner remains to be determined. Together, these data showed that specific O2 dependent genes mark potent HSCs/HPCs in context of transplantation.