A PML–PPAR-δ pathway for fatty acid oxidation regulates hematopoietic stem cell maintenance

Abstract

Stem-cell function is an exquisitely regulated process. Thus far, the contribution of metabolic cues to stem-cell function has not been well understood. Here we identify a previously unknown promyelocytic leukemia (PML)–peroxisome proliferator-activated receptor δ (PPAR-δ)–fatty-acid oxidation (FAO) pathway for the maintenance of hematopoietic stem cells (HSCs). We have found that loss of PPAR-δ or inhibition of mitochondrial FAO induces loss of HSC maintenance, whereas treatment with PPAR-δ agonists improved HSC maintenance. PML exerts its essential role in HSC maintenance through regulation of PPAR signaling and FAO. Mechanistically, the PML–PPAR-δ–FAO pathway controls the asymmetric division of HSCs. Deletion of Ppard or Pml as well as inhibition of FAO results in the symmetric commitment of HSC daughter cells, whereas PPAR-δ activation increased asymmetric cell division. Thus, our findings identify a metabolic switch for the control of HSC cell fate with potential therapeutic implications.

Figure 1. PPARδ is essential for HSC maintenance

(a) Overview of the experimental design for the deletion of Ppard. KSL, c-Kit^posSca-1^posLin^neg, MOI, multiplicity of infection, CAFC, cobblestone area forming cells, LTC-IC, long-term culture initiating cells. (b) Homing capacity to the bone marrow (BM) of indicated HSCs. The mean percentages ± S.D. of donor-derived CD34^negKSL cells are shown (n = 3). (c,d) Haematopoietic reconstitution capacity of Ppard-deleted KSL cells in BMT. Recipient mice were transplanted with 1.5x10³ Ppard^Δ/Δ KSL cells plus 2.0x10⁵ competitor cells in competition assays. Results represent the mean percentages ± s.d. of donor-derived cells (n = 4) (c) and multi-lineage haematopoietic contribution (myeloid cells (left), B cells (middle) and T cells (right)) 5 months after transplantation (d). (e,f), Limiting-dilution competitive repopulation analysis in second (e) and third (f) BMT with Ppard-ablated HSCs. (g), Cell cycle status of Ppard^lox/lox CD150^posCD48^negKSL cells by Pyronin Y staining 6 weeks after BMT (n = 4). All error bars indicate s.d.

Figure 2. Pharmacological activation of PPARδ enhances HSC maintenance

(a) Overview of the experimental design for GW treatment (Tx) in serial BMT. (b) Mean values ± s.d. of donor-derived CD150^posCD48^negCD41^negCD34^negFlt3^negKSL cells in recipient mice (left panels) and representative flow cytometry data and are shown (right upper panels) (n = 4). Negative Ctrl as well as fluorescence minus one (FMO) for CD48 are also demonstrated as control (right lower panels). (c–e). Donor-derived haematopoietic contribution 16 weeks after second round BMT with donor-derived KSL cells (c, d) or bone marrow mononuclear cells (BMMNCs, e) (as described in (a)) (n = 4). Results represent mean percentages ± s.d. of donor-derived cells in myeloid cells (left), B cells (middle) and T cells (right) in recipient mice 16 weeks after secondary BMT (n = 4) (d). (f) Effect of PPARδ agonist in the repopulation capacity of WT or Ppard-depleted KSL cells in 16 weeks after BMT. All error bars indicate s.d.

Figure 3. Pharmacological inhibition of mitochondrial FAO with Etomoxir induces HSC exhaustion

(a) FAO in undifferentiated (KSL) and differentiated (Lineage^pos) cells (n = 3). (b) Effect of etomoxir treatment on transplanted HSC function. After transplantation, recipient mice were treated with vehicle or Etomoxir for 4 weeks and BM was harvested 6 weeks after transplantation. Representative flow cytometry data (right) and mean numbers of donor-derived stem cell compartment in recipient mice (left) 6 weeks after BMT are shown. (c,d) Multi-lineage haematopoietic contribution. Recipient mice were treated with Etomoxir for 2 weeks. Results represent mean percentages of donor-derived cells in mononuclear cells (c) and in the indicated lineages (d) in recipient mice 24 weeks after transplantation. (e,f) Effect of etomoxir on the reconstitutive capacity of HSCs in secondary round BMT, following the experimental design outlined in Fig. 2a. Donor-derived haematopoietic contribution (e) and frequency of HSCs (f, n = 10) were determined 16 weeks after second round BMT. (g,h) Experimental design (g, left) and repopulation capacity of HSCs treated with Etomoxir in second round BMT. Donor-derived haematopoietic contribution (g, right) and functional HSC activity (h, n = 10). All error bars indicate s.d.

Figure 4. Forced pharmacological activation of PPARδ rescues the maintenance defect of Pml-deficient HSCs

(a) Relative expression of Cpt1a (left) and Acox1 (right) in Pml^−/− CD34^negKSL cells compared with Pml^+/+ CD34^negKSL cells. β-Actin was used as an internal control. (b) FAO in Pml^+/+ or Pml^−/− in KSL cells. Mean fold change from Pml^+/+ KSL cells over Etomoxir values are shown. (c) Overview of the experimental design for GW treatment (Tx) in the serial BMT. (d) Representative flow cytometry (right) and mean numbers of donor-derived CD150^posCD48^negCD41^negCD34^negFlt3^negKSL cells in recipient mice 6 weeks after BMT. (e) Reconstitutive capacity of Pml^−/− HSCs treated with GW in secondary BMT. Donor-derived haematopoietic contribution 16 weeks after second round BMT. (f) CAFC frequencies (in LDA) and LTC-IC ability (relative to vehicle-treated Pml^+/+) in Pml^+/+ and Pml^−/− CD34^negKSL cells treated with vehicle (Ctrl), GW or L165 (n = 3–4). All error bars indicate s.d.

Figure 5. An immunophenotypical assay to characterize the asymmetric division in HSCs

(a) Tie2 and CD48 expression in 50 randomly selected CD150^posCD48^negCD41^negFlt3^negKSL cells (n = 3). (b) Division patterns of Tie2^posCD48^neg HSCs (100 randomly selected cells, Representative data from n = 3). (c) Functional difference of two daughter cells after an initial cell division of CD150^posCD48^negCD41^negFlt3^negCD34^negKSL cells was investigated in in vitro paired daughter cell assay by long-term culture initiating cell (LTC-IC) analysis (18 divisions -Total; 36 daughter cells-analysed per experiment, n = 3). (d) In vivo paired daughter cell assay experimental design. Right panel depicts a representative division pattern proportion from 36 divisions. All error bars indicate s.d.

Figure 6. Ppard and fatty acid oxidation regulate asymmetric division in the HSC compartment

(a–b) Expected division patterns (a, left) and division pattern of Ppard^+/+ and Ppard^F/F stem cells infected with empty or Cre vector (n = 3, 50 randomly selected divisions per experiment). Scale bar, 10μm. (c) Functional difference of two daughter cells after an initial cell division of Ppard-ablated cells in in vitro paired daughter cell assay (n = 3, 18 divisions from three mice in each experiment). (d) Division pattern in CD150^posCD48^neg CD41^negFlt3^negCD34^negKSL cells treated with Etomoxir. Scale bar, 10μm. (e) In vivo paired daughter cell assay upon Etomoxir treatment. Percentage for asymmetric division, in which one daughter cell holds long-term repopulation capacity but another does not, is shown. (f) Single cell deposition of CD150^posCD48^negCD41^negFlt3^negCD34^negKSL cells from mice treated with GW for one week was conducted followed by in vitro treatment with GW and/or Etomoxir (+; 10 μM, ++; 20 μM). Division pattern in each treatment was investigated (n = 9). (g) Division pattern of WT or Pml^−/− HSCs. Scale bar, 10μm. (h) CD48^neg cells after first division of Pml^−/− HSCs in vivo. Sorted CD150^posCD48^negCD41^neg Flt3^negCD34^negKSL were stained with CSFE followed by their transplantation. 3 days after BMT, CD48 positivity in cells which undergo one division was determined. (i) Division pattern in Pml^−/− HSCs treated with vehicle or GW. (j) A model for regulation of asymmetric division by PML-PPARδ-FAO. All error bars indicate s.d.