Report

Hepatic thrombopoietin is required for bone marrow hematopoietic stem cell maintenance

See allHide authors and affiliations

Science  06 Apr 2018:
Vol. 360, Issue 6384, pp. 106-110
DOI: 10.1126/science.aap8861

Signaling hematopoietic stem cells from afar

Throughout our entire life span, hematopoietic stem cells (HSCs) generate all of our blood cells. The bone marrow microenvironment, or niche, is key to activating stem cell activity. Decker et al. now show that thrombopoietin generated in the liver, but not from the local bone marrow niche, maintains HSCs in vivo in mice. Thus, systemic endocrine factors are needed to maintain somatic stem cells from a distance. These findings may be important when considering how to stimulate HSCs for therapeutic use.

Science, this issue p. 106

Abstract

Hematopoietic stem cell (HSC) maintenance depends on extrinsic cues. Currently, only local signals arising from the bone marrow niche have been shown to maintain HSCs. However, it is not known whether systemic factors also sustain HSCs. We assessed the physiological source of thrombopoietin (TPO), a key cytokine required for maintaining HSCs. Using TpoDsRed-CreER knock-in mice, we showed that TPO is expressed by hepatocytes but not by bone marrow cells. Deletion of Tpo from hematopoietic cells, osteoblasts, or bone marrow mesenchymal stromal cells does not affect HSC number or function. However, when Tpo is deleted from hepatocytes, bone marrow HSCs are depleted. Thus, a cross-organ factor, circulating TPO made in the liver by hepatocytes, is required for bone marrow HSC maintenance. Our results demonstrate that systemic factors, in addition to the local niche, are a critical extrinsic component for HSC maintenance.

Hematopoietic stem cells (HSCs) reside primarily in the bone marrow and are maintained by extrinsic cues that arise from supporting niche cells (1). Endothelial cells (2, 3) and perivascular mesenchymal stromal cells (26) are critical components of the bone marrow niche. Growing functional genetic evidence suggests that HSCs are maintained largely through signals arising directly from, or mediated through, these local niche cells (7). However, olfaction maintains hematopoietic progenitors through systemic γ-aminobutyric acid (GABA) levels in Drosophila (8), suggesting that long-range signals may be able to directly maintain mammalian HSCs.

No such distal maintenance factors have yet been identified in the mammalian hematopoietic system, although long-range cues, such as estrogen from the ovaries and erythropoietin from the kidneys, can acutely stimulate HSC proliferation and dictate HSC and progenitor differentiation (9, 10). Neurotransmitters from the nervous system can mobilize HSCs, but this effect is mediated through mesenchymal stromal cells in the niche (11). Therefore, evidence indicates roles for long-range cues that modify HSC behavior, but direct evidence for constant maintenance of HSCs by a cross-organ long-range systemic factor is lacking.

Signaling of the hematopoietic cytokine thrombopoietin (TPO) through its receptor c-MPL is essential for thrombopoiesis (1214) and HSC maintenance (1517). Patients with loss-of-function mutations in c-MPL or TPO develop congenital amegakaryocytic thrombocytopenia and subsequent bone marrow failure (1820). Tpo mRNA is expressed by multiple cell types, including hepatocytes (14, 21), osteoblasts (17), megakaryocytes (22, 23), and stromal cells (21, 24). However, Tpo is under stringent translational control by inhibitory elements in the 5′ untranslated region (25), so it is not clear whether any of the above-mentioned cell types actually synthesize TPO protein. Tpo has not been conditionally deleted from any cell types to assess its source for HSC maintenance. Thus, it is not clear how TPO maintains bone marrow HSCs in vivo. Loss of hepatic TPO leads to low platelet counts (26), showing that TPO from the liver regulates thrombopoiesis.

Using quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis, we found that Tpo transcripts were enriched in osteoblasts, mesenchymal stromal cells, and the liver (Fig. 1A and fig. S1, A and B), consistent with previous reports (14, 17, 21, 27). To systemically assess the expression of TPO protein, we generated TpoDsRed-CreER knock-in mice by replacing the stop codon of Tpo with a P2A-DsRed-P2A-CreER cassette (fig. S1, C to F). The P2A elements allow the translation of TPO, DsRed, and CreER recombinase under the control of Tpo endogenous regulatory elements. This arrangement enabled us to monitor the translational expression of TPO in vivo. We then generated TpoDsRed-CreER;loxpZsGreen mice (Fig. 1B). Consistent with the low expression level of Tpo in vivo (25), no DsRed fluorescence was detected (Fig. 1, C to F). However, upon tamoxifen (TMX) administration to 8-week-old mice, we detected broad and specific expression of ZsGreen in hepatocytes (Fig. 1, G to J, and fig. S1, G to O). We also observed rare ZsGreen+ cells in the kidney (fig. S1P). However, no ZsGreen+ bone marrow cells could be detected (Fig. 1, K to N, and fig. S1Q). Thus, TPO is generated by hepatocytes but not by cells in the bone marrow.

Fig. 1 TPO is expressed by hepatocytes but not by bone marrow cells.

(A) qRT-PCR analysis of Tpo transcript levels (n = 3 mice; error bars indicate SD). (B) Schema of TPO expression analysis in TpoDsRed-CreER;loxpZsGreen mice. LSL, loxp-Stop-loxp. (C to J) Confocal images of liver sections from sham-treated (+Sham) or TMX-treated TpoDsRed-CreER;loxpZsGreen mice. DAPI, 4′,6-diamidino-2-phenylindole. (K to N) Confocal images of femur sections from sham- or TMX-treated TpoDsRed-CreER;loxpZsGreen mice.

We generated a loss-of-function allele of Tpo (Tpogfp) by recombining a gene encoding enhanced green fluorescent protein (gfp) into the start codon of Tpo (fig. S2, A to C). As expected, Tpo transcripts were depleted from Tpogfp/gfp mouse livers (fig. S2D). Consistent with earlier reports (28, 29), whole-body loss of TPO led to reduced platelet counts (fig. S3, A to C) and reduced numbers of megakaryocytes (fig. S3, D to J). Bone marrow from Tpogfp/gfp mice had normal cellularity, but CD150+CD48LinSca1+cKit+ HSC frequency (the percentage of live whole bone marrow cells) decreased about 70-fold compared with the frequency in Tpo+/+ controls (fig. S3, K to M). CD150CD48LinSca1+cKit+ multipotent progenitor (MPP) (30) and LinSca1+cKit+ (LSK) hematopoietic progenitor frequencies declined by 10- and 3-fold, respectively (fig. S3, N and O). Lineage-restricted hematopoietic progenitors appeared normal, except that CD34+FcγRLinSca1cKit+ common myeloid progenitors (CMPs) were reduced (fig. S3P). Bone marrow and spleen cells from Tpogfp/gfp mice formed fewer colonies in methylcellulose (fig. S3Q). Spleen cellularity did not change, but spleen HSC frequency was reduced (fig. S3, R and S). Tpo+/gfp heterozygous mice displayed intermediate phenotypes (fig. S3). Thus, TPO is a major factor required for HSC maintenance.

We generated a floxed allele of Tpo (Tpofl) by inserting loxp sequences flanking exons 2 to 4 of Tpo (fig. S4, A to C). Recombination of the loxp sites will lead to the deletion of the start codon and the generation of a frameshift. We recombined the Tpofl allele in the germ line to generate Tpo through mating with EIIa-cre mice (fig. S4D). As expected, Tpo transcripts were absent from homozygous Tpo−/− mice (fig. S4E). Tpo−/− mice had significant reduction of HSCs, MPPs, and megakaryocytes (fig. S4, F to H) to levels nearly identical to those in the Tpogfp/gfp mice (fig. S3). Recombination of the Tpofl allele therefore gave a strong loss of Tpo function.

Megakaryocytes have been proposed to be a major source of TPO for HSCs (22, 23). However, we did not detect any Tpo expression in megakaryocytes by qRT-PCR (Fig. 1A) or by using reporter mice (Fig. 1, K to N). Nonetheless, we directly tested whether megakaryocytes (or any hematopoietic cells) are sources of TPO for HSC maintenance in vivo by generating Vav1-cre;Tpofl/gfp mice. Vav1-cre efficiently deleted Tpo from the hematopoietic system (fig. S5A). Eight-week-old Vav1-cre;Tpofl/gfp mice had normal blood cell counts (fig. S5B), cellularity, and HSC frequency (fig. S5, C and D) and normal restricted hematopoietic progenitors in the bone marrow (fig. S5, E to H). Spleen cellularity, HSC frequency, and megakaryocytic cells were also unaffected (fig. S5, I to K). Bone marrow cells from Vav1-cre;Tpofl/gfp mice formed normal numbers of colonies in methylcellulose and reconstituted irradiated recipients normally (fig. S5, L and M). Thus, hematopoietic cells, including megakaryocytes, are not a critical source of TPO for HSC maintenance.

Osteoblasts have also previously been proposed to be the main source of TPO in the bone marrow (17), and osteoblasts express Tpo transcripts (Fig. 1A); however, we could not detect any Tpo translational activity in osteoblasts from our reporter mice (Fig. 1, K to N). We directly tested whether deletion of Tpo from osteoblasts affects HSCs. Consistent with previous reports (2, 3), Col2.3-cre recombined efficiently in bone-lining osteoblasts (fig. S6, A to G). Eight-week-old Col2.3-cre;Tpofl/gfp mice had normal blood cell counts (fig. S6H), cellularity, and HSC frequency in the bone marrow and spleens (Fig. 2, A and B). Restricted hematopoietic progenitors were also normal (fig. S6, I to L). Spleen megakaryocytic cells were unaffected (fig. S6M), as were the capacities of bone marrow and spleen cells to form colonies in methylcellulose (Fig. 2C and fig. S6N). Col2.3-cre;Tpofl/gfp bone marrow cells had normal capacities to reconstitute irradiated recipients (Fig. 2D and fig. S6, O and P). Thus, osteoblasts are not a critical source of TPO for HSC maintenance.

Fig. 2 Osteoblasts and mesenchymal stromal cells are not critical sources of TPO for HSC maintenance.

(A to C) Cellularity (A), HSC frequency (B), and bone marrow colony-forming cell frequency (C) were normal in Col2.3-cre;Tpofl/gfp mice (n = 4 to 5 mice). (D) Bone marrow cells (5 × 105) from Col2.3-cre;Tpofl/gfp mice gave normal donor cell reconstitution compared to controls (in two experiments with a total of eight to nine recipient mice per genotype). (E to G) Cellularity (E), HSC frequency (F), and bone marrow colony-forming cell frequency (G) were normal in Lepr-cre;Tpofl/gfp mice (n = 5 to 6 mice). (H) Bone marrow cells (5 × 105) from Lepr-cre;Tpofl/gfp mice gave normal levels of donor cell reconstitution compared to controls (in two experiments with a total of eight recipient mice per genotype). Controls (Con), Col2.3-cre;Tpo+/gfp, Lepr-cre;Tpo+/gfp, or Tpofl/gfp mice. Δ/gfp, Col2.3-cre;Tpofl/gfp or Lepr-cre;Tpofl/gfp mice. Data are means ± SD. ns, not significant (P > 0.10).

Bone marrow Lepr+ mesenchymal stromal cells are a critical source of HSC niche factors, including stem cell factor (SCF) and CXCL12 (2, 3). These cells express Tpo transcripts (Fig. 1A), although we could not detect Tpo translational activity (Fig. 1, K to N). We conditionally deleted Tpo from mesenchymal stromal cells by generating Lepr-cre;Tpofl/gfp mice. Tpo was efficiently deleted from these cells (fig. S7, A to G). Eight-week-old Lepr-cre;Tpofl/gfp mice had normal blood cell counts (fig. S7H), cellularity, and HSC frequency in the bone marrow and spleens (Fig. 2, E and F). Restricted hematopoietic progenitors were also normal (fig. S7, I to L). Spleen megakaryocytic cells were unaffected (fig. S7M). Bone marrow and spleen cells from Lepr-cre;Tpofl/gfp mice formed normal numbers of hematopoietic colonies in methylcellulose (Fig. 2G and fig. S7N). Bone marrow cells from Lepr-cre;Tpofl/gfp mice reconstituted irradiated recipients normally (Fig. 2H and fig. S7, O and P). Thus, bone marrow mesenchymal stromal cells are not a critical source of TPO for HSC maintenance.

The above data suggest that TPO produced locally by osteoblasts or mesenchymal stromal cells is not required for HSC maintenance. To test whether systemic TPO generated by the liver is important for HSC maintenance, we generated Alb-cre;Tpofl/fl mice. As expected (31), Alb-cre recombined specifically and efficiently in hepatocytes but not in the bone marrow (Fig. 3, A to D, and fig. S8, A to D). Eight-week-old Alb-cre;Tpofl/fl mice had a fivefold reduction in platelet count (Fig. 3E and fig. S8E) and a fivefold reduction in megakaryocytic cells in the bone marrow (Fig. 3F). Alb-cre;Tpofl/fl mice exhibited normal bone marrow and spleen cellularity (Fig. 3G). The frequency of bone marrow HSCs was reduced by a factor of 24 compared with that in controls (Fig. 3H). LSK progenitors and MPPs were similarly reduced (fig. S8, F and G). CMPs were reduced in the bone marrow, whereas other restricted hematopoietic progenitors appeared normal (fig. S8H). Bone marrow and spleen cells from Alb-cre;Tpofl/fl mice formed fewer colonies in methylcellulose (Fig. 3, I and J). Spleen HSC frequency was normal in Alb-cre;Tpofl/fl mice (Fig. 3H), suggesting that there was no compensatory extramedullary hematopoiesis. Spleen CD41+ cell frequency decreased (fig. S8I). Bone marrow cells from Alb-cre;Tpofl/fl mice had severe defects in their ability to reconstitute irradiated recipients (Fig. 3, K and L). Overall, these data show that hepatic TPO is critical for the maintenance of bone marrow HSCs.

Fig. 3 Hepatocyte-derived TPO is required for HSC maintenance.

(A to D) Confocal images of liver sections from Alb-cre;loxpTdTomato (Alb-cre;loxpTdT) mice. HNF4α, hepatocyte nuclear factor 4–α. (E) Platelet counts were decreased in Alb-cre;Tpofl/fl mice (n = 4). (F) Bone marrow CD41+ cell frequency was decreased in Alb-cre;Tpofl/fl mice (n = 5 to 7). (G) Cellularity was normal in the bone marrow and spleens from Alb-cre;Tpofl/fl mice (n = 5 to 6). (H) HSCs were depleted in the bone marrow but not in spleens from Alb-cre;Tpofl/fl mice (n = 6 to 8). (I and J) Colony-forming cell numbers were decreased in the bone marrow and trended lower in spleens (SPL) from Alb-cre;Tpofl/fl mice (n = 5 to 6). E, erythroid burst–forming unit; Mk, megakaryocyte; GM, granulocyte-macrophage; GEMM, granulocyte-erythrocyte-monocyte-megakaryocyte. (K) Bone marrow cells (5 × 105) from Alb-cre;Tpofl/fl mice gave lower levels of donor cell reconstitution (in two experiments with a total of eight to nine recipient mice per genotype). (L) Recipient mice from (K) showed significant decrease in donor bone marrow HSC chimerism 16 weeks posttransplant (n = 6 mice per genotype). Con, Alb-cre;Tpo+/+, Tpofl/fl, or Tpofl/+ mice; Δ/Δ, Alb-cre;Tpofl/fl mice. Data are means ± SD. ns, not significant (P > 0.10). *P < 0.05, **P < 0.01, ***P < 0.001.

During development, HSCs transiently reside in the fetal liver (32), where Alb-cre recombines (33). Thus, it is possible that bone marrow HSCs in Alb-cre;Tpofl/fl mice acquire a persistent defect during development, although global deletion of Tpo has no impact on fetal liver HSCs (16). Nonetheless, we conditionally deleted Tpo from adult hepatocytes by administering the hepatotropic Cre-bearing virus AAV8-TBG-cre (34). Consistent with a prior report (34), we observed specific and efficient recombination in hepatocytes after a single intravenous administration of AAV8-TBG-cre to adult mice (Fig. 4, A to D, and fig. S9, A to D).

Fig. 4 TPO from adult hepatocytes regulates bone marrow HSC maintenance and quiescence.

(A to D) Confocal images of liver sections from AAV8-TBG-cre–treated loxpTdTomato mice. (E) Bone marrow and spleen cellularity was normal in Alb-cre;Tpofl/fl mice (n = 7 to 8). (F) Bone marrow HSCs were depleted in AAV8-TBG-cre–treated Tpofl/fl mice (n = 4 to 5). (G) Spleen HSC frequency tended to decrease in AAV8-TBG-cre–treated Tpofl/fl mice. (H and I) Bone marrow and spleen colony-forming cell frequencies decreased in AAV8-cre;Tpofl/fl mice (n = 6 to 7). (J) Bone marrow cells (5 × 105) from AAV8-cre;Tpofl/fl mice gave lower levels of donor cell reconstitution than controls (in two to three experiments with a total of n = 16 for controls, n = 9 for 1 month after AAV8 treatment, and n = 14 for 2 months after AAV8 treatment). (K) Limit dilution assays showed a 6-fold reduction in functional HSCs in the bone marrow from AAV8-cre;Tpofl/fl mice compared with those in controls (in two independent experiments). (L) Annexin V+ DAPI bone marrow HSC frequency was not affected in AAV8-cre;Tpofl/fl mice (n = 5). (M) Bone marrow HSCs incorporated more bromodeoxyuridine (BrdU) in AAV8-cre;Tpofl/fl mice (n = 4). Con, Tpo+/+ mice treated with AAV8-TBG-cre or Tpofl/fl mice treated with phosphate-buffered saline; Δ/Δ, Tpofl/fl mice treated with AAV8-TBG-cre. Data are means ± SD. ns, not significant (P > 0.10). *P < 0.05, **P < 0.01, ***P < 0.001.

We administered AAV8-TBG-cre virus once to 8-week-old Tpofl/fl mice and wild-type littermate control mice and analyzed the mice 4 to 8 weeks later. AAV8-TBG-cre–treated wild-type mice had phenotypes identical to those of buffer-treated Tpofl/fl mice. These mice were pooled together as controls. Deleting Tpo from adult hepatocytes (AAV8-cre;Tpofl/fl) significantly reduced platelet counts (fig. S9E). Bone marrow and spleen cellularity did not change (Fig. 4E). However, AAV8-cre;Tpofl/fl mice had significant reductions in bone marrow HSC, LSK, MPP, CMP, and CD41+ cell frequencies (Fig. 4F and fig. S9, G to I). The colony-forming capacities of bone marrow and spleen cells were also reduced (Fig. 4, H and I). Spleen HSC and CD41+ megakaryocytic cell frequencies trended lower (Fig. 4G and fig. S9J). Bone marrow cells from AAV8-cre;Tpofl/fl mice 2 months after the virus treatment had a significant decrease in their ability to reconstitute irradiated recipients, whereas bone marrow cells from AAV8-cre;Tpofl/fl mice 1 month after the virus treatment had an intermediate phenotype (Fig. 4J and fig. S9K). To quantify the frequency of functional HSCs, we performed limit dilution assays. In these assays, the frequencies of long-term multilineage reconstituting cells in bone marrow cells from control and AAV8-cre;Tpofl/fl mice were 1/87,660 and 1/498,020, respectively, corresponding to a 6-fold reduction (Fig. 4K). AAV8-cre;Tpofl/fl mice had no significant difference in Annexin V+ bone marrow HSCs compared with controls (Fig. 4L and fig. S9L), but HSCs from AAV8-cre;Tpofl/fl mice cycled more (Fig. 4M and fig. S9M). These data indicate that adult HSCs lacking hepatocyte-derived TPO signal are depleted through loss of quiescence.

Tpo transcription is up-regulated in the bone marrow during hematopoietic stress (21). We injected TMX-treated TpoDsRed-CreER;loxpZsGreen mice with 5-fluorouracil but failed to detect ZsGreen expression in the bone marrow (fig. S10), suggesting that bone marrow is not a major source of TPO in stress, at least under the conditions we tested. Our data show that hepatocytes are the major functional source of systemic TPO for bone marrow HSC maintenance under steady-state conditions. Further studies are required to determine the role of local and systemic TPO in nonhomeostatic conditions.

Supplementary Materials

www.sciencemag.org/content/360/6384/106/suppl/DC1

Materials and Methods

Figs. S1 to S10

References (3540)

References and Notes

Acknowledgments: We thank V. Lin at the Columbia University transgenic core for helping generate Tpogfp, TpoDsRed-CreER, and Tpofl mice. We thank S. Ho at the Columbia Center for Translational Immunology, A. Figueroa in the Columbia University Department of Microbiology and Immunology, and M. Kissner at the Columbia Stem Cell Initiative for flow cytometry. Funding: This work was supported by the Rita Allen Foundation and the National Heart, Lung, and Blood Institute (grant R01HL132074). M.D. was supported by the Columbia Medical Scientist Training Program and the NIH (grant 1F30HL137323). Author contributions: M.D., J.L., and L.D. performed experiments. Q.L. helped generate the targeting vectors. M.D. and L.D. designed the experiments, interpreted the results, and wrote the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the manuscript or the supplementary materials. The AAV8-TBG-cre virus is available from the Penn Vector Core under a material transfer agreement with the University of Pennsylvania.
View Abstract

Stay Connected to Science

Navigate This Article