Sphingolipid Modulation Activates Proteostasis Programs to Govern Human Hematopoietic Stem Cell Self-Renewal

Abstract

Cellular stress responses serve as crucial decision points balancing persistence or culling of hematopoietic stem cells (HSCs) for lifelong blood production. Although strong stressors cull HSCs, the linkage between stress programs and self-renewal properties that underlie human HSC maintenance remains unknown, particularly at quiescence exit when HSCs must also dynamically shift metabolic state. Here, we demonstrate distinct wiring of the sphingolipidome across the human hematopoietic hierarchy and find that genetic or pharmacologic modulation of the sphingolipid enzyme DEGS1 regulates lineage differentiation. Inhibition of DEGS1 in hematopoietic stem and progenitor cells during the transition from quiescence to cellular activation with N-(4-hydroxyphenyl) retinamide activates coordinated stress pathways that coalesce on endoplasmic reticulum stress and autophagy programs to maintain immunophenotypic and functional HSCs. Thus, our work identifies a linkage between sphingolipid metabolism, proteostatic quality control systems, and HSC self-renewal and provides therapeutic targets for improving HSC-based cellular therapeutics.

Keywords: DEGS1; StemRegenin-1; UM171; autophagy; fenretinide; hematopoietic stem cell; lipidomics; sphingolipid metabolism; umbilical cord blood; unfolded protein response.

Graphical abstract

Figure 1

DEGS1 Contributes to the Distinct Wiring of Sphingolipid Synthesis in the Human Hematopoietic Hierarchy and Is Functionally Required In Vivo (A) Heatmap of mRNA expression for 36 lipid genes that are significantly differentially expressed (FDR < 0.05 and fold change > 1.5) between LT-HSCs and megakaryocyte erythroid progenitor (MEP)/GMP/common myeloid progenitor (CMP)/multi-lymphoid progenitor (MLP) in the dataset from Laurenti et al. (2013). SpL genes are in bold. (B) SpL distribution in the indicated CB populations (n = 3–5). See Figure S1 for direct comparisons of each SpL. (C) Normalized dhCer profiles for stem and progenitor cells with the indicated fatty-acyl chain (n = 3). (D) Log2 ratio of Cer to dhCer %. Significance to stem cells is in red and to progenitors is in blue. (E) qRT-PCR of DEGS1 at 0 h and 6 h in culture from CB subpopulations. (F) Human engraftment (hCD45⁺BFP⁺) at 4 weeks xenotransplantation for shCtrl or shDEGS1 marked by BFP (5 biological replicates; n = 5 mice per replicate; see Figures S1J–S1M for transduction input, hCD45 chimerism, and BFP%). (G) Fold change of BFP marked transduced cells relative to input in human CD45⁺ cells. Unpaired t test; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.

Figure 2

Sphingolipid Modulation of DEGS1 Alters HSC Function and Lineage Balance In Vitro (A) Total Cer and dhCer levels in lin⁻ CB progeny cultured for 8 days with control or 4HPR (n = 2). (B and C) LT-HSC, ST-HSC, or GMP CFC assays in the presence of Ctrl or 2 μM 4HPR showing (B) colonies/100 cells and (C) colony distribution (n = 12; CB). (D and E) Flow cytometry for (D) monocytic (CD14⁺) and (E) erythroid (GlyA⁺) markers in live cells from pooled CFC colonies (n = 6; CB). (F) Colony distribution in CFC assays with LT-HSC, ST-HSC, or GMP cells transduced with shCtrl or shDEGS1 (n = 4; CB). (G) Flow cytometry for erythroid cells from pooled CFC colonies in (F). (H and I) hCD45⁺BFP⁺ lacking the CD34⁺CD19⁻ population were isolated from shCtrl or shDEGS1 mice at 4 weeks post-transplant (n = 2 CB) and profiled by LC/MS for (H) total Cer and dhCer levels and (I) normalized Cer and dhCer profiles with the indicated fatty-acyl chain. BQL, below quantitation level. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.

Figure 3

Ex Vivo Treatment with 4HPR Maintains HSC Function following Xenotransplantation (A) Experimental scheme for ex vivo culture of lin⁻ CB followed by xenotransplantation with vehicle control or 4HPR. (B) Total cell counts at 8 days culture for the 16-LTRC dose of a representative experiment (n = 3; CB cultured in triplicate). (C–F) The number of (C) viable cells, (D) CD15⁺ myeloid cells, (E) GlyA⁺ erythroid cells, and (F) CD34⁺ cells injected/mouse were calculated for the 16-LTRC dose by flow cytometry analysis for control or 4HPR treatment prior to xenotransplantation (n = 5 CB pools, 4 in technical triplicate, marked with different symbols). (G) Human CD45⁺ engraftment at 16 weeks post-transplant in injected femurs and non-injected bones (n = 4 CB pools, 5 mice/drug treatment for each CB pool). (H–K) Lineage analysis of mice engrafted with control or 4HPR-treated cells from Figure 2G for (H) B lymphoid, (I) myeloid, (J) erythroid, and (K) CD34⁺ lacking CD33 or CD19 markers at the 16-LTRC dose. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.

Figure 4

Sphingolipid Modulation Restricts Expansion of Committed Progenitors during Ex Vivo Culture to Enhance HSC Self-Renewal (A) Number of colonies for LT-HSC CFC assays with control (−) or the indicated combination of 4HPR, UM171, or SR1 at 10 days, unpaired t test relative to control (Ctrl, black) or to the presence or absence of 4HPR treatment. (B) Flow cytometry for the % and count (no.) of CD34⁺ cells subsequently transplanted per mouse following 8 days culture. (C) Representative flow cytometry plots for CD90 and CD45RA within the CD34⁺ fraction of lin⁻ CB progeny at day 8 with the indicated treatments. (D) Distribution of cultured CD34⁺CD90⁺CD45RA⁻ (cLT), CD34⁺CD90⁻CD45RA⁻ (cST), and CD34⁺CD90⁻CD45RA⁻ (cProg) for the treatments at day 8 following CD34 enrichment (n = 4); unpaired t test relative to control. (E and F) Graph of LTRC frequencies for control and 4HPR (E) and table summarizing results of secondary assays (F) with CD45⁺ cells isolated from 16-LTRC dose mice and transplanted at limiting doses for 16 weeks to calculate LTRC frequencies (5 biological experiments for control and 4HPR; 2 biological experiments for U+S and 3-Factor). Human CD45⁺ marking of >0.1% was considered positive for secondary engraftment. p value was by extreme limiting dilution analysis (ELDA). ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.

Figure 5

Sphingolipid Modulation with 4HPR Treatment Upregulates Cellular Stress Pathways and Remodels Cellular Metabolism in HSPCs during Ex Vivo Culture (A) NES scores for selected pathways that are significantly enriched following 4HPR treatment at day 2. (B) Pathway analysis (GSEA) of SpL/cer, cholesterol biosynthesis ER stress/UPR, protein folding, and ROS pathway modules for 4HPR compared to control. (C) GSEA analysis of autophagy, ER stress/UPR, protein folding, or ROS pathways in uncultured LT-HSCs versus ST-HSCs. (D and E) LT-HSCs from Velten et al. were clustered as cell cycle primed or non-primed as described in STAR Methods, (D) the gene expression of CDK6 (top) and DEGS1 (bottom), and (E) signature scores representing relative expression of pathways in (C) for single LT-HSCs; Wilcoxon rank sum test for (D) and (E); ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001. (F) Heatmap showing gene expression at day 2 of the top 4 genes in gene sets in (A) organized as pathways. We included ATF4-related genes gene sets from the ER stress/UPR pathway. (G and H) Median pEIF2S1 staining intensity quantified from microscopy images for BFP⁺ (G) LT-HSCs and (H) ST-HSCs isolated from 4-week xenografts engrafted with shCtrl or shDEGS1 stem cells for 3 CBs (60 cells/CB, except shDEGS1 LT-HSCs, 13–17 cells/CB). (I and J) Flow cytometry analysis at day 2 post-treatment with indicated concentrations of 4HPR in the progeny of CD34⁺CD38⁻ stem or CD34⁺CD38⁺ progenitor CB cells from 4 CBs for (I) ROS with CellROX and (J) mitochondrial membrane potential with TMRE. Paired t test; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001 in (G)–(J).

Figure 6

4HPR Activates Autophagy in HSPCs during Ex Vivo Culture (A and B) LC3II staining intensity of (A) LT-HSCs (174 cells) and GMPs (312 cells) isolated from CB (n = 3) or (B) mPB (165 cells; n = 3 mPB) LT-HSCs compared to CB LT-HSCs. (C) Representative microscopy images of DAPI (blue) and LC3II staining (green) for (C) CD34⁺CD38⁻ (stem) cells following 2 days of treatment with DMSO control or 2 μM 4HPR and ±BAF. Scale bar is 5 μm. (D and E) LC3II foci area for (D) stem and (E) prog cells from one of three CB. The mean in control stem cells without BAF is shown with a dotted line. (F) Relative LC3II foci in the presence of BAF for stem and Prog populations. (G) Relative Cyto-ID flux for 0 (DMSO), 0.2 μM, and 2 μM 4HPR in stem and prog populations at 2 days post-treatment (n = 4). (H) Relative Cyto-ID MFI measurements for stem-enriched samples to analyze autophagic flux with and without cytokine withdrawal with indicated drug treatments at day 2. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.

Figure 7

Sphingolipid Modulation by 4HPR Activates a Coordinated Proteostatic Pro-survival Response (A–D) Flow cytometry analysis at day 8 culture following autophagy inhibition with BAF and/or ISR inhibition with ISRIB in lin⁻ CB (n = 3, in duplicate) beginning at day 0 for control or 4HPR-treated cells for number of (A) live cells, (B) CD34⁺ cells, (C) CD14⁺ cells, and (D) GlyA⁺ cells; represented as relative % to control; mean ± SEM. (E) Total number of cells at day 8 after BAF was added either starting at day 0 or day 2. (F) Autophagic flux at 20 h with indicated drugs was assayed with Cyto-ID MFI relative to control (2 CB, in triplicate). (G) Model for how 4HPR maintains stemness during ex vivo culture. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.