Thrombopoietin from hepatocytes promotes hematopoietic stem cell regeneration after myeloablation

  1. Longfei Gao
  2. Matthew Decker
  3. Haidee Chen
  4. Lei Ding  Is a corresponding author
  1. Columbia Stem Cell Initiative, Columbia University Medical Center, United States
  2. Department of Rehabilitation and Regenerative Medicine, Columbia University Medical Center, United States
  3. Department of Microbiology and Immunology, Columbia University Medical Center, New York, United States, United States

Abstract

The bone marrow niche plays critical roles in hematopoietic recovery and hematopoietic stem cell (HSC) regeneration after myeloablative stress. However, it is not clear whether systemic factors beyond the local niche are required for these essential processes in vivo. Thrombopoietin (THPO) is a key cytokine promoting hematopoietic rebound after myeloablation and its transcripts are expressed by multiple cellular sources. The upregulation of bone marrow-derived THPO has been proposed to be crucial for hematopoietic recovery and HSC regeneration after stress. Nonetheless, the cellular source of THPO in myeloablative stress has never been investigated genetically. We assessed the functional sources of THPO following two common myeloablative perturbations: 5-fluorouracil (5-FU) administration and irradiation. Using a Thpo translational reporter, we found that the liver but not the bone marrow is the major source of THPO protein after myeloablation. Mice with conditional Thpo deletion from osteoblasts and/or bone marrow stromal cells showed normal recovery of HSCs and hematopoiesis after myeloablation. In contrast, mice with conditional Thpo deletion from hepatocytes showed significant defects in HSC regeneration and hematopoietic rebound after myeloablation. Thus, systemic THPO from the liver is necessary for HSC regeneration and hematopoietic recovery in myeloablative stress conditions.

Introduction

Hematopoietic stem cells (HSCs) generate all blood and immune cells, and play critical roles in the regeneration of the blood system after stress. They reside in the bone marrow niche where Leptin Receptor+ (LepR+) perivascular stromal cells and endothelial cells are key components. These stromal cells are the major sources of factors that promote HSC maintenance (Ding and Morrison, 2013; Ding et al., 2012; Lee et al., 2017). Following myeloablative stress, the bone marrow niche supports HSC regeneration, which is essential for the recovery of hematopoiesis (Zhou et al., 2015; Hooper et al., 2009; Kopp et al., 2005; Li et al., 2008). Several signaling pathways, including Notch, EGF/EGFR, pleiotrophin, Dkk1, angiogenin, and SCF/c-KIT, are essential for the regeneration of HSCs and hematopoiesis (Butler et al., 2010; Doan et al., 2013; Goncalves et al., 2016; Himburg et al., 2017; Himburg et al., 2010; Kimura et al., 2011). Many of the secreted cytokines activating these pathways arise from the local bone marrow niche and regulate HSC regeneration and hematopoietic recovery in a paracrine fashion (Guo et al., 2017; Himburg et al., 2017; Poulos et al., 2013; Zhou et al., 2017). However, it is not clear whether systemic signals that originate outside the bone marrow also promote HSC regeneration and hematopoietic recovery after stress.

Thrombopoietin (THPO) is critical for bone marrow HSC maintenance and hematopoietic recovery (Decker et al., 2018; Mouthon et al., 1999; Qian et al., 2007; Yoshihara et al., 2007). It was first identified as the ligand to the MPL receptor and a driver of megakaryopoiesis and platelet production (de Sauvage et al., 1994; Kaushansky et al., 1994; Lok et al., 1994). Later results show that the THPO/MPL signaling is also required for the maintenance of HSCs (Kimura et al., 1998). Circulating levels of THPO are inversely related to platelet counts (de Graaf et al., 2010; Kaushansky, 2005), consistent with a ‘receptor sink’ model where HSCs are stimulated to proliferate and self-renew by free systemic THPO ligand, which is gradually soaked up by the expanding population of MPL-expressing progenies until equilibrium is reached. In line with this model, deletion of the MPL receptor from megakaryocyte-lineage cells does not lead to a loss of megakaryocytes or platelets, instead, leads to an expansion of HSCs with accompanying megakaryocytosis and thrombocytosis (Ng et al., 2014). By conditionally deleting Thpo from the bone marrow or liver, we have recently showed that steady-state HSC maintenance depends on hepatocyte-derived THPO (Decker et al., 2018), further highlighting the importance of systemic THPO on HSCs.

The THPO/MPL signaling also plays a crucial role in hematopoietic stress response, particularly after myeloablation. The characteristic hematopoietic progenitor rebound at around 10 days following administration of the antimetabolite drug 5-fluorouracil (5-FU) is dependent on MPL (Li and Slayton, 2013). After irradiation, the THPO/MPL signaling is similarly essential for hematopoietic recovery and survival (de Laval et al., 2014; de Laval et al., 2013; Mouthon et al., 1999; Wang et al., 2015). Indeed, THPO mimetic drugs, such as romiplostim and eltrombopag, have been shown to improve recovery after ablative challenge, and have also been used clinically to support hematopoiesis in diseases such as immune thrombocytopenic purpura and aplastic anemia (Desmond et al., 2014; Gill et al., 2017; Rodeghiero and Ruggeri, 2015; Yamaguchi et al., 2018).

The regulation of THPO production has been extensively investigated, but the in vivo source of THPO for HSC regeneration and hematopoietic recovery after myeloablation is not clear. Previous studies have found that bone marrow cell populations such as stromal cells and osteoblasts may upregulate THPO in hematopoietic stress conditions, whereas the liver produces Thpo transcripts at a constant level (Sungaran et al., 1997; Yoshihara et al., 2007). However, other investigators have found no significant changes in bone marrow Thpo transcript levels after 5-FU-mediated myeloablative treatment (Li and Slayton, 2013). Because Thpo expression is under heavy translational control (Ghilardi et al., 1998), it is not clear what cells produce THPO protein for HSC regeneration and hematopoietic recovery after myeloablation. Furthermore, although upregulation of THPO may be a key mechanism of the bone marrow response to hematopoietic stress, the role of local THPO from bone marrow niche or systemic THPO from the liver for HSC and hematopoietic recovery has not been functionally investigated in vivo. Nonetheless, most studies proposed that local THPO derived from the bone marrow niche is critical for HSC and hematopoietic recovery after myeloablation (Kaushansky, 2005; Yoshihara et al., 2007). Here, we genetically dissected the in vivo source of THPO for HSC regeneration and hematopoietic recovery following myeloablative stress.

Results

Myeloablation induced by 5-FU drives THPO-dependent hematopoietic recovery and HSC expansion

5-FU is a commonly used chemotherapy agent that leads to myeloablation. To test whether THPO is required for hematopoietic recovery after 5-FU treatment, we administrated 5-FU to Thpo knockout (Thpogfp/gfp) (Decker et al., 2018) and wild-type control mice at 150 mg/kg via a single intraperitoneal injection. Because Thpogfp/gfp mice have hematopoietic phenotypes compared with wild-type controls without any treatment (Decker et al., 2018), we also normalized hematopoietic parameters with baseline mice of the same genotype without any treatment. Ten days after the 5-FU administration, treated wild-type mice showed normal leukocyte and neutrophil counts, normal reticulocyte frequency, but significantly increased platelet counts, while treated Thpogfp/gfp mice had no platelet expansion (Figure 1—figure supplement 1A-H). When normalized to baseline levels, Thpogfp/gfp mice had a significant reduction of leukocyte, neutrophil, and platelet counts as well as reticulocyte frequency compared with wild-type controls (Figure 1—figure supplement 1A-H), suggesting that Thpo is required for hematopoietic recovery. Normalized post-treatment bone marrow and spleen cellularity did not differ significantly in Thpogfp/gfp mice compared with wild-type controls (Figure 1—figure supplement 1I-L). The frequencies of bone marrow and spleen HSCs (Lineage- Sca-1+ c-kit+ CD150+CD48-, LSKCD150+CD48-) were increased by more than 12 fold in 5-FU-treated wild-type mice compared with baseline controls, while 5-FU-treated Thpogfp/gfp mice showed no change from the baseline (Figure 1—figure supplement 1M-P). Overall, 5-FU stimulated a 3.4-fold increase of total HSC number in wild-type mice, but no significant effects were observed in Thpogfp/gfp mice (Figure 1—figure supplement 1Q). Compared with wild-type controls, the HSC increase in response to 5-FU in Thpogfp/gfp mice was reduced by 6 fold (Figure 1—figure supplement 1R). The effects of 5-FU on bone marrow and spleen hematopoietic progenitor cells (LSK) were also blunted in Thpogfp/gfp mice (Figure 1—figure supplement 1S-V). These data confirm that hematopoietic recovery, expansion of HSCs, and hematopoietic progenitors following chemoablation mediated by 5-FU are THPO-dependent.

Hepatic but not bone marrow THPO is required for HSC and hematopoietic recovery after 5-FU-mediated myeloablation

To identify the source of THPO, we investigated Thpo expression following 5-FU treatment. It has been reported that the bone marrow upregulates Thpo mRNA in stress conditions (Sungaran et al., 1997). Although bone marrow stromal cells (CD45- Ter119-) express Thpo transcripts, we did not observe a significant upregulation of Thpo transcripts 14 days after 5-FU injection compared with baseline levels (Figure 1—figure supplement 2A). THPO protein production is under strong translational controls (Ghilardi et al., 1998). To assess the translation of THPO protein following 5-FU treatment, ThpocreER; Rosa26LSL-ZsGreen translational reporter mice (Decker et al., 2018) were injected with 5-FU and immediately started on a 10-day course of tamoxifen treatment. Although robust expression of ZsGreen was observed in the liver (Figure 1A), no ZsGreen fluorescence was appreciated in the bone marrow (Figure 1B). Within the liver, ZsGreen were from HNF4a+ hepatocytes (Figure 1C and D). These data suggest that hepatocytes but not the bone marrow are the major source of THPO after 5-FU treatment.

Figure 1 with 2 supplements see all
Bone marrow thrombopoietin (THPO) is not required for hematopoietic recovery and hematopoietic stem cell (HSC) expansion after 5-fluorouracil (5-FU) chemoablation.

(A) Confocal images of liver sections from tamoxifen-treated ThpocreER; Rosa26LSL-ZsGreen mice 10 days after 5-FU injection. (B) Confocal images of femur sections from tamoxifen-treated ThpocreER; Rosa26LSL-ZsGreen mice 10 days after 5-FU injection. (C and D) Confocal images showing immunostaining of HNF4α on liver sections from tamoxifen-treated ThpocreER; Rosa26LSL-ZsGreen mice 10 days after 5-FU injection. (D) Enlarged image of the region denoted in C. Arrows point to individual HNF4α+ hepatocytes. (E–H) Blood counts of mice with Thpo conditionally deleted from osteoblasts (Col1a1-cre), bone marrow stromal cells (Lepr-cre), or both (Prrx1-cre), and controls after 5-FU challenge. n = 3–10 mice. (I–L) Cellularity and frequencies of HSCs in the bone marrow and spleens from mice with Thpo conditionally deleted from osteoblasts, bone marrow stromal cells, or both, and controls after 5-FU challenge. n = 3–11 mice. (M) Total numbers of HSCs in the bone marrow and spleens from mice with Thpo conditionally deleted from osteoblasts, bone marrow stromal cells, or both, and controls after 5-FU challenge. n = 3–10 mice. Con: Thpogfp/+ control mice. Col1a1-cre: Col1a1-cre; Thpofl/gfp mice. Lepr-cre: Leprcre; Thpofl/gfp mice. Prrx1-cre: Prrx1-cre; Thpofl/gfp mice. Data represent mean ± SEM. Statistical significance was calculated with one-way ANOVA with Dunnett’s test. Each dot represents one independent mouse in E–M.

Figure 1—source data 1

Numerical values of the data plotted in panels E-M.

https://cdn.elifesciences.org/articles/69894/elife-69894-fig1-data1-v2.xlsx

Although we did not detect appreciable Thpo translation in the bone marrow after 5-FU treatment (Figure 1B), osteoblasts and LepR+ mesenchymal stromal cells are the major bone marrow cells expressing Thpo transcripts (Decker et al., 2018; Guerriero et al., 1997; Sungaran et al., 1997), and based on antibody staining, it has been reported that osteoblasts express THPO protein after 5-FU treatment (Yoshihara et al., 2007). To formally test the function of THPO from these cells, we conditionally deleted Thpo from osteoblasts (Col1a1-cre; Thpofl/gfp) or bone marrow mesenchymal stromal cells (Leprcre; Thpofl/gfp). Deletion of Thpo from osteoblasts or LepR+ stromal cells had no significant effects on leukocytes, neutrophils, platelets, reticulocytes, bone marrow and spleen cellularity, HSCs, or LSKs after 5-FU treatment (Figure 1E–M, Figure 1—figure supplement 2B and C). These results show that hematopoietic recovery and HSC regeneration from 5-FU treatment do not require THPO production by bone marrow osteoblasts or mesenchymal stromal cells.

It is possible that osteoblasts and mesenchymal stromal cells serve as redundant sources of THPO. We thus also generated Prrx1-cre; Thpofl/gfp mice. Prrx1-cre drives recombination in mesenchymal lineage cells in murine long bones, including both osteoblasts and mesenchymal stromal cells (Ding and Morrison, 2013; Greenbaum et al., 2013; Logan et al., 2002), allowing conditional deletion of Thpo from both osteoblasts and mesenchymal stromal cells. Compared with littermate controls, adult baseline Prrx1-cre; Thpofl/gfp mice had normal blood cell counts, normal bone marrow cellularity, normal HSC and LSK frequencies, and showed no signs of extramedullary hematopoiesis in the spleen (Figure 1—figure supplement 2D-N), consistent with the notion that THPO from the bone marrow is not required for HSC maintenance or hematopoiesis (Decker et al., 2018). We then administrated 5-FU to these mice. Deletion of Thpo from both osteoblasts and mesenchymal cells in Prrx1-cre; Thpofl/gfp mice had no effects on leukocytes, neutrophils, platelets, reticulocytes, bone marrow and spleen cellularity, HSCs, or LSKs after 5-FU chemoablation (Figure 1E–M, Figure 1—figure supplement 2B and C). These results strengthen our finding that the bone marrow is not a source of THPO for hematopoietic recovery and HSC regeneration following 5-FU treatment.

Hepatic THPO is required for steady-state bone marrow HSC maintenance as conditional deletion of Thpo from hepatocytes leads to HSC depletion (Decker et al., 2018). We found that hepatocytes are also the major source of THPO after 5-FU treatment (Figure 1A–D). To address whether hepatic THPO is required for hematopoietic recovery and HSC regeneration after 5-FU-mediated stress, we treated adult Thpofl/fl mice with replication incompetent hepatotropic AAV8-TBG-cre viral vector (AAV) concurrently with 5-FU injection. A single dose of this cre-bearing viral vector drives efficient and specific recombination in hepatocytes (Figure 2—figure supplement 1A and Decker et al., 2018). Ten days after the treatment, AAV-treated mice showed a trend to reduction of platelet and reticulocyte expansion, with normal leukocyte and neutrophil responses to 5-FU relative to controls (Figure 2—figure supplement 1B-I). Although the response of spleen HSC frequency to 5-FU was not significantly impacted by AAV treatment (Figure 2G and H), the responses of bone marrow HSC frequency and the total number of HSCs were significantly decreased in AAV-treated mice compared to controls (Figure 2A–J), indicating a compromised HSC recovery after 5-FU treatment. Of note, we observed a significant increase of the responses of spleen cellularity to 5-FU by AAV treatment (Figure 2E and F), indicating possible enhanced extramedullary hematopoiesis. Consistent with reduced HSC regeneration, bone marrow cells from these AAV-treated mice after 5-FU administration showed a significant reduction in reconstituting irradiated recipient mice compared with controls (Figure 2K). The responses of bone marrow and spleen LSK frequencies to 5-FU were largely normal in AAV-treated mice compared with controls (Figure 2—figure supplement 1J-M), suggesting that the effects of acutely disrupted THPO signaling are most pronounced on HSCs, rather than restricted hematopoietic progenitors. Altogether, these data suggest that hepatic but not bone marrow-derived THPO is required for hematopoietic and HSC recovery after 5-FU treatment.

Figure 2 with 1 supplement see all
Hepatic thrombopoietin (THPO) is required for hematopoietic recovery and hematopoietic stem cell (HSC) expansion after 5-fluorouracil (5-FU) chemoablation.

(A–H) Cellularity and frequencies of HSCs (A, C, E, G) in the bone marrow and spleens from mice with Thpo conditionally deleted from hepatocytes (AAV) and controls, with and without 5-FU challenge. Normalized fold changes (relative to no 5-FU challenge baseline) (B, D, F, H) are also shown. n = 4–11 mice. (I and J) Total numbers of HSCs (I) in the bone marrow and spleens from mice with Thpo conditionally deleted from hepatocytes and controls, with and without 5-FU challenge. Normalized fold changes (relative to no 5-FU challenge baseline) (J) are also shown. n = 4–11 mice. (K) Numbers of total recipients and long-term multilineage reconstituted recipients after transplantation of 50,000 bone marrow cells from PBS- or AAV-treated Thpofl/fl and 5-FU challenged mice along with 500,000 unchallenged competitor bone marrow cells. Recipients were scored as reconstituted if they showed donor-derived myeloid, B and T cells in the peripheral blood 16 weeks after transplantation. Data were from recipients of three independent donor pairs from two independent experiments. PBS: PBS-treated Thpofl/fl or wild-type control mice. AAV: AAV8-TBG-cre-treated Thpofl/fl mice. Data represent mean ± SEM. *p < 0.05, **p < 0.01, ****p < 0.0001. Statistical significance was assessed with two-way ANOVA with Turkey’s test (A, C, E, G, I), Student’s t-test (B, D, F, H, J), or Fisher’s exact test (K). Each dot represents one independent mouse in A–J.

Figure 2—source data 1

Numerical values of the data plotted in panels A-J.

https://cdn.elifesciences.org/articles/69894/elife-69894-fig2-data1-v2.xlsx

Hematopoiesis from THPO-deficient mice are sensitive to ionizing radiation

Ionizing radiation is another common myeloablative treatment. To test whether THPO is required for hematopoietic recovery in this condition, we exposed mice to a single 5.5 Gy dose of radiation. Four weeks after the irradiation, Thpogfp/gfp mice had significant reduction of leukocyte and neutrophil recovery with normal platelet and reticulocyte responses compared with wild-type controls (Figure 3—figure supplement 1A-H). Thpogfp/gfp mice also had significantly decreased responses to irradiation in bone marrow cellularity, LSK and HSC frequencies relative to wild-type controls (Figure 3—figure supplement 1I-N). The functional impact of these differences was demonstrated by a significant increase in mortality among Thpogfp/gfp mice following irradiation (Figure 3—figure supplement 1O). Compared with baseline levels, neither wild-type nor Thpogfp/gfp mice had notable differences in spleen cellularity, or spleen LSK frequency in response to irradiation (Figure 3—figure supplement 1P and R). However, Thpogfp/gfp mice had diminished regeneration of spleen HSC frequency and total HSC number compared with controls (Figure 3—figure supplement 1P-W). Thus, Thpo is required for hematopoietic and HSC recovery after irradiation.

Hepatic but not bone marrow THPO is required for HSC regeneration after ionizing radiation

It is not clear what cells are the major source of THPO after irradiation. Bone marrow stromal cells (CD45- Ter119-) did not have significantly upregulated Thpo transcripts 14 days after irradiation (Figure 1—figure supplement 2A). To assess the translation of THPO protein following irradiation, ThpocreER; Rosa26LSL-ZsGreen translational reporter mice were irradiated and immediately dosed with a 10-day course of tamoxifen treatment. Although robust recombination of the Rosa26LSL-ZsGreen allele was observed in the liver (Figure 3A), no ZsGreen was observed in the bone marrow (Figure 3B). Hepatocytes were the major cell type that expressed THPO in the liver after irradiation (Figure 3C and D). This finding suggests that hepatocytes are the major source of THPO with no appreciable THPO expression in the bone marrow following radioablation.

Figure 3 with 2 supplements see all
Bone marrow thrombopoietin (THPO) is not required for hematopoietic recovery from irradiation.

(A) Confocal images of liver sections from tamoxifen-treated ThpocreER; Rosa26LSL-ZsGreen mice 10 days after irradiation. (B) Confocal images of femur sections from tamoxifen-treated ThpocreER; Rosa26LSL-ZsGreen mice 10 days after irradiation. (C and D) Confocal images showing immunostaining of HNF4α on liver sections from tamoxifen-treated ThpocreER; Rosa26LSL-ZsGreen mice 10 days after irradiation. (D) Enlarged image of the region denoted in C. Arrows point to individual HNF4α+ hepatocytes. (E–H) Blood counts of mice with Thpo conditionally deleted from osteoblasts (Col1a1-cre), bone marrow stromal cells (Lepr-cre), or both (Prrx1-cre), and controls after irradiation. n = 3–10 mice. (I–L) Cellularity and frequencies of hematopoietic stem cells (HSCs) in the bone marrow and spleens from mice with Thpo conditionally deleted from osteoblasts, bone marrow stromal cells, or both, and controls after irradiation. n = 3–11 mice. (M) Total numbers of HSCs in the bone marrow and spleens from mice with Thpo conditionally deleted from osteoblasts, bone marrow stromal cells, or both, and controls after irradiation. n = 3–11 mice. Con: Thpogfp/+ control mice. Col1a1-cre: Col1a1-cre; Thpofl/gfp mice. Lepr-cre: Leprcre; Thpofl/gfp mice. Prrx1-cre: Prrx1-cre; Thpofl/gfp mice. Data represent mean ± SEM. Statistical significance was assessed with one-way ANOVA with Dunnett’s test. Each dot represents one independent mouse in E–M.

Figure 3—source data 1

Numerical values of the data plotted in panels E-M.

https://cdn.elifesciences.org/articles/69894/elife-69894-fig3-data1-v2.xlsx

To functionally test whether osteoblasts and bone marrow mesenchymal cells are sources of THPO for hematopoietic recovery and HSC regeneration after irradiation, we gave Col1a1-cre; Thpofl/gfp, Leprcre; Thpofl/gfp or Prrx1-cre; Thpofl/gfp mice, along with littermate controls, a single 5.5 Gy dose of radiation and analyzed them 4 weeks later. Compared with irradiated controls, Col1a1-cre; Thpofl/gfp, Leprcre; Thpofl/gfp or Prrx1-cre; Thpofl/gfp mice showed no significant differences in leukocyte, neutrophil, and platelet counts, as well as reticulocyte frequency (Figure 3E–H). Bone marrow cellularity, LSK and HSC frequencies were also normal (Figure 3I and J, Figure 3—figure supplement 2A). These mice also had normal spleen cellularity, LSK and HSC frequencies, as well as total HSC numbers (Figure 3K–M and Figure 3—figure supplement 2B). These data suggest that osteoblasts and bone marrow mesenchymal stromal cells are dispensable sources of THPO for HSC and hematopoietic recovery from radioablative challenges.

To test the role of hepatocyte-derived THPO, we then irradiated Thpofl/fl mice treated with AAV8-TBG-cre virus vector. Compared to controls, these mice showed normal responses to irradiation in leukocyte, neutrophil, and platelet counts as well as reticulocyte frequency (Figure 4—figure supplement 1A–H). Although the bone marrow cellularity response to irradiation was similar between Thpofl/fl mice treated with AAV and control, there was a significant reduction in bone marrow HSC but not LSK frequency recovery in AAV-treated Thpofl/fl mice relative to controls after irradiation (Figure 4A–D and Figure 4—figure supplement 1I and J). Compared with controls, spleen cellularity, LSK and HSC frequencies from Thpofl/fl mice treated with AAV had normal responses after irradiation (Figure 4E–H and Figure 4—figure supplement 1K and L). Overall, Thpofl/fl mice treated with AAV had a significant reduction in total HSC number recovery after irradiation (Figure 4I and J). Importantly, bone marrow cells from these irradiated AAV-treated mice showed functional defects in reconstituting irradiated recipients (Figure 4K). These data suggest that hepatic THPO is required for proper functional recovery of HSCs following irradiation-mediated myeloablation.

Figure 4 with 1 supplement see all
Hepatic thrombopoietin (THPO) is required for effective hematopoietic stem cell (HSC) recovery from ionizing radiation.

(A–H) Cellularity and frequencies of HSCs in the bone marrow and spleens from mice with Thpo conditionally deleted from hepatocytes (AAV) and controls with and without irradiation. Normalized fold changes after irradiation (relative to no irradiation baseline) are also shown. n = 6–7 mice. (I and J) Total numbers of HSCs in the bone marrow and spleens from mice with Thpo conditionally deleted from hepatocytes (AAV) and controls with and without irradiation. Normalized fold changes after irradiation (relative to no irradiation baseline) are also shown. n = 6–7 mice. (K) Numbers of total recipients and long-term multilineage reconstituted recipients after transplantation of 1,000,000 bone marrow cells from PBS- or AAV-treated Thpofl/fl and irradiated donor mice along with 1,000,000 irradiated competitor bone marrow cells. Recipients were scored as reconstituted if they showed donor-derived myeloid, B and T cells in the peripheral blood 16 weeks after transplantation. Data were from recipients of two donor pairs from two independent experiments. PBS: PBS-treated Thpofl/fl or wild-type control mice. AAV: AAV8-TBG-cre-treated Thpofl/fl mice. Data represent mean ± SEM. *p < 0.05, **p < 0.01, ****p < 0.0001. Statistical significance was assessed with two-way ANOVA with Turkey’s test (A, C, E, G, I), Student’s t-test (B, D, F, H, J), or Fisher’s exact test (K). Each dot represents one independent mouse in A–J.

Figure 4—source data 1

Numerical values of the data plotted in panels A-J.

https://cdn.elifesciences.org/articles/69894/elife-69894-fig4-data1-v2.xlsx

Discussion

Our data show that the bone marrow does not meaningfully upregulate THPO protein after 5-FU treatment or irradiation. Although we did not exhaustively investigate all models of ablative conditioning, it appears that the bone marrow is not a functional source of THPO in hematopoietic recovery from acute myeloablative stress. However, our data do not rule out the possibility that bone marrow may serve as a source of THPO in certain specific stress conditions, such as idiopathic thrombocytopenia purpura (Sungaran et al., 1997). Future investigation into additional stress conditions is needed. We found that hepatocytes are an important source of THPO for stress hematopoiesis. Although several other studies have investigated the role of THPO on hematopoietic recovery after stress, the source of THPO and its impact on HSCs (rather than hematopoietic progenitors) have not been determined. By examining HSCs and hematopoietic progenitors, we show that HSCs are more sensitive to acute THPO perturbation than hematopoietic progenitors (LSKs), suggesting the major role of THPO in myeloablation is to promote HSC regeneration, at least under the conditions we studied. To our knowledge, THPO is the first systemic factor required for HSC regeneration and hematopoietic recovery identified to date. The presence of a stem cell hormone, like THPO, raises the possibility of other endocrine regulators of HSC regeneration and hematopoietic recovery.

Numerous connections between hepatic and hematopoietic pathophysiology have long been appreciated. Aplastic anemia is a known sequela of pediatric liver failure, and thrombocytopenia is a well-characterized feature of hepatic disease (Hadzić et al., 2008; Peck-Radosavljevic, 2017; Tung et al., 2000). While mechanisms such as metabolic toxicity and splenic sequestration may play a role in mediating these effects, it seems likely that disruption of the THPO signaling is also involved. Indeed, several recent clinical studies have shown that newly developed THPO agonists can reduce the need for platelet transfusions in patients with chronic liver disease (Peck-Radosavljevic et al., 2019; Xu and Cai, 2019). While early trials of THPO mimetics in primary and acquired aplastic anemia have been promising, to our knowledge, no group has specifically studied their use in liver failure-associated aplastic anemia (Desmond et al., 2014; Gill et al., 2017). Given that hepatic Thpo mRNA and serum THPO levels are significantly downregulated in liver failure (Wolber et al., 1999), further studies are warranted. Our data suggest that cancer patients with underlying liver disease may have greater sensitivity to myeloablative conditioning and supplementing THPO mimetics may help alleviate the adverse effects associated with myeloablative conditioning in these patients. On the other hand, THPO may also have effects on cells beyond the hematopoietic system. For example, a recent report suggests that in hepatocellular carcinoma, THPO may facilitate tumor progression by promoting VEGF signaling in hepatocytes (Vizio et al., 2021). Thus, strategies aimed at manipulating the THPO pathway need careful evaluation.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Genetic reagent (Mus musculus)Prrx1-crePMID:12112875JAX stock (005584)
Genetic reagent (Mus musculus)LeprcrePMID:11283374JAX stock (008320)
Genetic reagent (Mus musculus)Rosa26LSL-ZsGreenPMID:20023653JAX stock (007906)
Genetic reagent (Mus musculus)Col1a1-crePMID:15470637
Genetic reagent (Mus musculus)ThpogfpPMID:29622652
Genetic reagent (Mus musculus)ThpocreERPMID:29622652
Genetic reagent (Mus musculus)ThpoflPMID:29622652
Recombinant DNA reagentAAV8-TBG-crePenn Vector core or AddgeneCat# 107787-AAV8
Chemical compound, drug5-FluorouracilFreseniusKabiCat# 101,710150 mg/kg IP
Chemical compound, drugTamoxifenSigmaCat# T5648
Chemical compound, drugCollagenase, Type IVWorthingtonCat# LS004188
Chemical compound, drugDNase ISigmaCat# D4527
Antibody(Rat monoclonal) anti-CD2 (RM2-5)BiolegendFlow cytometry (1:200)
Antibody(Rat monoclonal) anti-CD3 (17A2)BiolegendFlow cytometry (1:200)
Antibody(Rat monoclonal) anti-CD5 (53–7.3)BiolegendFlow cytometry (1:400)
Antibody(Rat monoclonal) anti-CD8a (53–6.7)BiolegendFlow cytometry (1:400)
Antibody(Rat monoclonal) anti-B220 (6B2)BiolegendFlow cytometry (1:400)
Antibody(Rat monoclonal) anti-Gr1 (8C5)BiolegendFlow cytometry (1:400)
Antibody(Rat monoclonal) anti-Ter119BiolegendFlow cytometry (1:200)
Antibody(Rat monoclonal) anti-Sca1 (E13-161.7)BiolegendFlow cytometry (1:200)
Antibody(Rat monoclonal) anti-cKit (2B8)BiolegendFlow cytometry (1:200)
AntibodyArmenian (Hamster monoclonal) anti-CD48 (HM48-1)BiolegendFlow cytometry (1:200)
Antibody(Rat monoclonal) anti-CD150 (TC15-12F12.2)BiolegendFlow cytometry (1:200)
Antibody(Rat monoclonal) anti-CD45.2 (104)BiolegendFlow cytometry (1:400)
Antibody(Rat monoclonal) anti-CD45.1 (A20)BiolegendFlow cytometry (1:400)
Antibody(Rat monoclonal) anti-Mac1 (M1/70)BiolegendFlow cytometry (1:400)
Antibody(Rat monoclonal) anti-CD45 (30 F-11)BiolegendFlow cytometry (1:400)
Antibody(Rabbit monoclonal) anti-HNF4α (EPR16885-99)AbCamCat# Ab201460IF (1:10)
Sequence-based reagentOLD815IDT DNACCACCACCATGCCTAACTCT
Sequence-based reagentOLD816IDT DNAGTTCTCCTCCACGTCTCCAG
Sequence-based reagentOLD817IDT DNATCGCTAGCTGCTCTGATGAA
Sequence-based reagentZsGreen FIDT DNAGGCATTAAAGCAGCGTATCC
Sequence-based reagentZsGreen RIDT DNAAACCAGAAGTGGCACCTGAC
Sequence-based reagentOLD581IDT DNACATCTCGCTGCTCTTAGCAGGG
Sequence-based reagentOLD582IDT DNAGAGCTGTTTGTGTTCCAACTGG
Sequence-based reagentOLD292IDT DNACGGACACGCTGAACTTGTGG
Sequence-based reagentOLD528IDT DNAACTTATTCTCAGGTGGTGACTC
Sequence-based reagentOLD653IDT DNAAGGGAGCCACTTCAGTTAGAC
Sequence-based reagentOLD434IDT DNACATTGTATGGGATCTGATCTGG
Sequence-based reagentOLD435IDT DNAGGCAAATTTTGGTGTACGGTC
Sequence-based reagentOLD338IDT DNAGCATTTCTGGGGATTGCTTA
Sequence-based reagentOLD339IDT DNAATTCTCCCACCGTCAGTACG
Sequence-based reagentOLD390IDT DNACCTTTGTCTATCCCTGTTCTGC
Sequence-based reagentOLD391IDT DNAACTGCCCCTAGAATGTCCTGT
Sequence-based reagentOLD27IDT DNAGCTCTTTTCCAGCCTTCCTT
Sequence-based reagentOLD28IDT DNACTTCTGCATCCTGTCAGCAA
Software, algorithmFacsDivaBD
Software, algorithmFlowJoFlowJo
Software, algorithmPrismGraphPad
OtherSuperScript IIIThermoFisherCat# 18080093
OtherProtoScript IINEBCat# M0368S
OtherPELCO Cryo-Embedding compoundTed Pella, IncCat# 27,300

Mice

All mice were 8–16 weeks of age and maintained on a C57BL/6 background. Prrx1-cre (stock # 005584) recombining in bone marrow mesenchymal cells of the long bones, Leprcre (stock # 008320) recombining in bone marrow mesenchymal stromal cells, and Rosa26LSL-ZsGreen (stock # 007906) mice were obtained from the Jackson Laboratory. Col1a1-cre mice (with 2.3 kb promoter), recombining in mature osteoblasts, were described previously (Liu et al., 2004). The generation of Thpogfp, ThpocreER, and Thpofl mice was described previously (Decker et al., 2018). Thpogfp is a null allele of Thpo. ThpocreER knockin mice allows tracing of cells that translate THPO protein. All mice were housed in specific pathogen-free, Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC)-approved facilities at the Columbia University Medical Center. All protocols were approved by the Institute Animal Care and Use Committee of Columbia University.

Genotyping PCR

Request a detailed protocol

Primers for genotyping ThpocreER: OLD815, 5’-CCACCACCATGCCTAACTCT-3’; OLD816, 5’-GTTCTCCTCCACGTCTCCAG-3’; and OLD817, 5’-TCGCTAGCTGCTCTGATGAA-3’.

Primers for genotyping Rosa26LSL-ZsGreen: GGCATTAAAGCAGCGTATCC and AACCAGAAGTGGCACCTGAC.

Primers for genotyping Thpofl: OLD581, 5’-CATCTCGCTGCTCTTAGCAGGG-3’ and OLD582, 5’-GAGCTGTTTGTGTTCCAACTGG-3’.

Primers for genotyping Thpogfp: OLD292, 5’-CGGACACGCTGAACTTGTGG-3’; OLD528 5’-ACTTATTCTCAGGTGGTGACTC-3’ and OLD653 5’-AGGGAGCCACTTCAGTTAGAC-3’.

Primers for genotyping Leprcre: OLD434 5’-CATTGTATGGGATCTGATCTGG-3’ and OLD435 5’-GGCAAATTTTGGTGTACGGTC-3’.

Primers for genotyping cre: OLD338, 5’-GCATTTCTGGGGATTGCTTA-3’ and OLD339, 5’-ATTCTCCCACCGTCAGTACG-3’.

Chemoablative challenge

Request a detailed protocol

5-FU (150 mg/kg) was administered to mice via a single intraperitoneal injection. Ten days post-injection, mice were euthanized for analysis.

Radioablative challenge

Request a detailed protocol

Mice were irradiated by a Cesium 137 Irradiator (JL Shepherd and Associates) at 300 rad/min with one dose of 550 rad. Four weeks after irradiation, mice were euthanized for analysis.

Tamoxifen administration

Request a detailed protocol

Tamoxifen (Sigma) was dissolved in corn oil for a final concentration of 20 mg/mL. Every other day for 10 days, 50 µL of the solution was administered by oral gavage. Mice were analyzed 2–4 days after the final tamoxifen administration.

Viral vector infections

Request a detailed protocol

Replication-incompetent AAV8-TBG-cre was obtained from the Penn Vector Core or Addgene. AAV8-TBG-cre vector carries cre recombinase gene under the regulatory control of hepatocyte-specific thyroid-binding globulin (TBG) promoter. Efficient recombination was achieved at a dose of 2.5 × 1011 viral particles diluted in sterile 1× PBS. AAV8-TBG-cre was administered to mice via retro-orbital venous sinus injection.

Complete blood count

Request a detailed protocol

Peripheral blood was collected to EDTA tubes (Microvette CB300, Sarstedt) to prevent clogging. Blood samples were then run on a complete blood count analyzer (Genesis, Oxford Science).

Flow cytometry

Request a detailed protocol

Bone marrow cells were isolated by flushing the long bones or by crushing the long bones, pelvis, and vertebrae with mortar and pestle in Ca2+- and Mg2+-free HBSS with 2% heat-inactivated bovine serum. Spleen cells were obtained by crushing the spleens between two glass slides. The cells were passed through a 25 G needle several times and filtered with a 70 μm nylon mesh. The following antibodies were used to perform HSC staining: lineage markers (anti-CD2 [RM2-5], anti-CD3 [17A2], anti-CD5 [53–7.3], anti-CD8a [53–6.7], anti-B220 [6B2], anti-Gr-1 [8C5], anti-Ter119), anti-Sca-1 (E13-161.7), anti-c-kit (2B8), anti-CD48 (HM48-1), anti-CD150 (TC15-12F12.2).

For flow cytometric sorting of bone marrow stromal cells, bone marrow plugs were flushed as described above, then digested with collagenase IV (200 U/mL) and DNase1 (200 U/mL) at 37°C for 20 min. Samples were then stained with anti-CD45 (30 F-11) and anti-Ter119 antibodies and sorted on a flow cytometer.

Quantitative real-time PCR

Request a detailed protocol

Cells were lysed in Trizol. Total RNA was extracted according to manufacturer’s instructions. Total RNA was subjected to reverse transcription using SuperScript III (Invitrogen) or ProtoScript II (NEB). Quantitative real-time PCR was run using SYBR green on a CFX Connect system (Biorad). β-Actin (Actb) was used to normalize the RNA content of samples. Primers used were: Thpo: OLD390, CCTTTGTCTATCCCTGTTCTGC and OLD391, ACTGCCCCTAGAATGTCCTGT; β-actin: OLD27, 5’-GCTCTTTTCCAGCCTTCCTT-3’ and OLD28, 5’-CTTCTGCATCCTGTCAGCAA-3’.

Bone and liver sectioning

Request a detailed protocol

Freshly disassociated long bones were fixed for 3 hr in a solution of 4% paraformaldehyde, 7% picric acid, and 10% sucrose (W/V). The bones were then embedded in 8% gelatin, immediately snap-frozen in liquid N2, and stored at –80°C. Bones were sectioned using a CryoJane system (Instrumedics). For liver, cardiac perfusion with formalin was performed immediately after mouse sacrifice, and perfused liver tissue was dehydrated in a 30% sucrose solution overnight at 4°C. Liver tissue was then placed in PELCO Cryo-Embedding compound (Ted Pella, Inc), frozen on dry ice, and stored at –80°C. Liver tissue was sectioned and directly transferred onto microscope slides. Both bone and liver sections were dried overnight at room temperature and stored at –80°C. Sections were rehydrated in PBS for 5 min, stained with DAPI for 15 min, then mounted with Vectashield Hardset Antifade Mounting Medium (Vector Laboratories). An rabbit anti-HNF4α antibody (AbCam, ab201460) was used to stain hepatocytes. Images were acquired on a Nikon Eclipse Ti confocal microscope (Nikon Instruments) or a Leica SP8 confocal microscope (Leica Microsystems).

Long-term competitive reconstitution assay

Request a detailed protocol

Adult recipient mice were lethally irradiated by a Cesium 137 Irradiator (JL Shepherd and Associates) at 300 rad/min with two doses of 550 rad (total 1100 rad) delivered at least 2 hr apart. Cells were transplanted by retro-orbital venous sinus injection of anesthetized mice. Donor bone marrow cells were transplanted along with recipient bone marrow cells into lethally irradiated recipient mice. For irradiation-induced myeloablation, the competing recipient bone marrow cells were from mice that were irradiated similarly as the donor mice. Mice were maintained on antibiotic water (Baytril 0.17 g/L) for 14 days, then switched to regular water. Recipient mice were periodically bled to assess the level of donor-derived blood lineages, including myeloid, B, and T cells for at least 16 weeks. Blood was subjected to ammonium chloride potassium red cell lysis before antibody staining. Antibodies including anti-CD45.2 (104), anti-CD45.1 (A20), anti-CD3 (17A2), anti-B220 (6B2), anti-Gr-1 (8C5), and anti-Mac-1 (M1/70) were used to stain cells. Mice with presence of donor-derived myeloid, B, and T cells for 16 weeks were considered as long-term multilineage reconstituted.

Statistical analysis

Request a detailed protocol

All analyses were done using GraphPad Prism 7.0. In any comparison between a pooled control cohort and multiple experimental conditions, we used one-way ANOVA with Dunnett’s test. In any comparison between control and mutant in multiple experimental conditions, we used two-way ANOVA with Turkey’s test. For surviving curve comparison, we used Log-rank (Mantel-Cox) test. For transplantation experiments, we used Fisher’s exact test. For all other comparisons, we used unpaired t-test. In all figures, error bars represent standard error of mean.

Data availability

All data were presented with individual data points from each mouse. Source data files have been provided. No sequencing or diffraction data are generated.

References

Decision letter

  1. Hossein Ardehali
    Reviewing Editor; Northwestern University, United States
  2. Mone Zaidi
    Senior Editor; Icahn School of Medicine at Mount Sinai, United States
  3. Hossein Ardehali
    Reviewer; Northwestern University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This paper shows that hepatic TPO is responsible for HSC regeneration after 5FU and irradiation. The paper is novel and will shift the paradigm in the field. The studies are conducted carefully and the experimental approach is sound. The authors have been responsive to the previous comments and have modified the manuscript significantly.

Decision letter after peer review:

Thank you for submitting your article "Thrombopoietin from hepatocytes promotes hematopoietic stem cell regeneration after myeloablation" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Hossein Ardehali as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Reviewing Editor and Mone Zaidi as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) The authors only assessed the transcriptional regulation of TPO after 5FU in the bone marrow. They need to also assess the transcript level in the liver.

2) What is the role of TPO expression from osteoblasts and mesenchymal stem cells if it is not involved in HSC regeneration after myeloablation?

3) Although the studies are focused on the number of hematopoietic cells after 5-FU and irradiation, it would be helpful to see if the difference in the numbers of these cells is associated with any disorders, like increased inflammation or bleeding. Also, it would be helpful to determine whether deletion of TPO in the liver is associated with anemia in mice.

4) What it the rational to move some of the data to the supplements, while others are kept in the main figures? e.g Figure 1 showing no effects in the deleted mice throughout the figure, in Figure 2 the leukocytes, neutrophiles counts etc. moved to S3?

5) AAV stands for adeno-associated viral vector, not adenovirus and also not adeno-associated virus, as the AAV used in the study is replication incompetent and should be addressed as vector.

6) How was the reticulocyte count determined?

7) There are various Lepr-Cre-deleter mice at Jackson lab (same for the other lines). It is not sufficient to name them Lepr-Cre or Prx1-Cre in the Materials and methods part, please give the correct name of the mouse line.

8) Figure legends for the supplementary Figures are missing.

9) In respect to the discussion: The systemic effect of Thpo regulation has been discussed by deGraaf et al., 2010.

10) It would be of advantage for the reader, to mention and explain the results from their previous study (the tissue-specific deletion of Thpo under steady state conditions in the mouse) in the introduction.

11) Platelet counts in PBS treated mice (Figure 2A) are lower than in untreated control mice (Figure 1G). What is the major difference between these controls?

12) The paper would benefit to directly compare the data from the osteoblast and stromal cell deleted mice with the AAV treated mice in one graphic. Except for the difference in technology used to delete Thpo there is no obvious reason why the data are split up in the different figures.

13) The response in platelet counts after 5-FU administration in Mpl-/- mice has been studied (Levin et al., Blood 2001). These mice can respond in the absence of intact Thpo/Mpl signaling. These data are contradictory to what is shown here. What could be the explanation for that?

14) The major difference between using Cre-deleter mice in contrast to the AAV mediated application is that in the latter case Thpo is deleted in an adult mouse while in the other case, mice have develop under the condition that no Thpo is supplied from the bone marrow niche (osteoblasts or mesenchymal cells, accordingly). This questions whether these two scenarios can be safely compared. Similar to this, the inhibition of Thpo/Mpl signaling by a dominant-negative Mpl receptor in the adult mouse causes severe and lethal aplasia (Wicke et al., Mol Ther 2010), while the germ line knockout of Mpl or Thpo is in general well tolerated in the mouse and HSC defects get only more obvious after transplantation (Alexander et al., Blood 1996; Kimura et al., PNAS 1998; deSauvage et al., J Ex Med 1996; Yoshihara et al., CSC 2007, Qian et al., CSC 2007). What was the rational to use the AAV8 Cre delivery in contrast to using a deleter mouse with liver-specific Cre expression?

15) The spread in the data after AAV application to mice may be caused by differences in the transduction efficiencies in the liver. How efficient was the deletion in the liver after AAV application?

16) In the transplantation experiment, the donor mice were treated with AAV-Cre. To address the role of hepatic Thpo in the recovery after transplantation, it would be of interest to delete Thpo in the recipients.

17) The transplantation of irradiated BM cells should in general not lead to a successful engraftment of these cells. What is the aim of this experiment?

18) What is the threshold of donor contribution that is judged as positive engraftment?

19) As mentioned above, the study shows that, similarly to what occurs in bone marrow HSC maintenance in steady-state conditions, hepatic TPO appears to be a major functional trigger for hematopoietic recovery and HSC regeneration after myelosuppression caused by radio-chemotherapy injury.

To avoid misleading conclusions and thus contribute to clinical medicine more precisely, the experimental design of the study must relate to a tumor context, since radio-chemotherapies play an established role in the treatment of cancer patients. The genetically modified animal models must have pathophysiological characteristics as comparable as possible to those that occur in humans (for animal-to-human extrapolation).

We thus recommend that the reported experiments be flanked by others run upon a murine tumor model.

20) Since it has been shown that TPO is an important and nonredundant regulator of VEGF-A production in HSCs, and that is activates a VEGF-A internal autocrine loop, participating in the favorable TPO effects on HSC self-renewal and expansion (Kirito K et al., Blood. 2005; 105:4258) it would be interesting to investigate the contribution of this mechanism to the efficacy of hepatic TPO in hematopoietic recovery in stress conditions in genetically altered mouse models. This would be of interest in particular because VEGF-A, notoriously elevated in a tumor context, could also be counted among the endocrine factors which, together with TPO, contribute to bone marrow recovery, as suggested in the discussion.

21) In the discussion, in the light of their data, the authors assume that cancer patients with underlying liver disease, such as HCC, may have enhanced sensitivity to myeloablative conditioning because of liver-failure-related hepatic downregulated TPO levels. Since TPO mimetic drugs, such as romiplostim and eltrombopag, have been shown to improve recovery after ablative challenge, the authors suggest that these agents could help alleviate the adverse effects associated with myeloablative conditioning in such patients.

However, it has recently been reported that, in the HCC context (Vizio B et al., Int. J. Mol. Sci. 2021, 22, 1818), the increased hepatic TPO level appears to be a novel triggering signal to promote tumor progression. Therefore, curative strategies aimed at manipulating THPOR in HCC patients should be evaluated with extreme care.

The possibility of tumor conditioning should be pointed out, and recent literature in this connection included in the discussion.

https://doi.org/10.7554/eLife.69894.sa1

Author response

Essential revisions:

1) The authors only assessed the transcriptional regulation of TPO after 5FU in the bone marrow. They need to also assess the transcript level in the liver.

It has been reported that Thpo transcripts in the liver are expressed at constant

levels (Blood 1995 86:3668). Because protein production is the ultimate readout of THPO production and Thpo is under profound translational control (Blood 1998 92:4023), we focused on assessing THPO translation using our translational reporter mice. Nonetheless, per the request by the reviewer, we performed qPCR analysis (see Author response image 1). Consistent with the prior report, Thpo transcript levels were not changed in the liver after 5-FU treatment.

Author response image 1
5-FU treatment does not change Thpo transcript levels in the liver.

qPCR analysis showing the relative expression levels of Thpo transcripts in controls and 5-FU-treated mice. n = 4 mice for each condition.

2) What is the role of TPO expression from osteoblasts and mesenchymal stem cells if it is not involved in HSC regeneration after myeloablation?

This is a great question that intrigues us as well. We initially hypothesized that THPO from the local bone marrow niche is for HSC regeneration after myeloablation, but this does not turn out to be true. Because only 5-FU and irradiation conditions were tested, we can not completely rule out the possibility that locally produced THPO plays roles for HSC regeneration in other stress conditions. We have addressed this in the Discussion section.

3) Although the studies are focused on the number of hematopoietic cells after 5-FU and irradiation, it would be helpful to see if the difference in the numbers of these cells is associated with any disorders, like increased inflammation or bleeding. Also, it would be helpful to determine whether deletion of TPO in the liver is associated with anemia in mice.

Thank you for raising these points. Inflammation or bleeding is not considered as myeloablation. However, as the reviewer pointed out, inflammation and bleeding are also interesting conditions to look at. This is something we plan to work on in the future and out of the scope of the current study. Please note that, regardless what the results of these future experiments may be, they will not change the main conclusion of the current manuscript, which focuses on the major acute myeloablation conditions. The impact of liver-derived THPO on reticulocytes depends on the treatment condition. Under the steady-state condition, deletion of Thpo from the liver did not lead to a significant alteration of reticulocyte frequencies (Figure 2-figure supplement 1 H, I and Figure 4—figure supplement 1 G, H). Under 5-FU treatment, deletion of Thpo from the liver led to a trend in reduction of reticulocyte frequency compared with controls (Figure 2—figure supplement 1 H, I). Under irradiation condition, deletion of Thpo from the liver did not lead to a significant difference of reticulocyte frequencies (Figure 4—figure supplement 1 G, H).

4) What it the rational to move some of the data to the supplements, while others are kept in the main figures? e.g Figure 1 showing no effects in the deleted mice throughout the figure, in Figure 2 the leukocytes, neutrophiles counts etc. moved to S3?

We apologize for the confusion. We organized the data based on myeloablative conditions (5-FU vs irradiation) and where Thpo is deleted (local bone marrow vs the liver) in the original version of the manuscript. Because deletion of Thpo from the liver leads to phenotypes in the hematopoietic system under the steady-state condition, we need to show normalized phenotypes, such as in Figure 2A. As a result, the data from mice with Thpo deleted from the liver are more complicated. Thus we moved the data showing no differences to Supplementary Figures in the original version of the manuscript. We agree with the reviewer that the data in the figures can be better organized to be consistent throughout. We modified the figures by moving all peripheral blood data in Figures 2 and 4 into Supplementary Figures in the revised manscript, as suggested by the reviewer.

5) AAV stands for adeno-associated viral vector, not adenovirus and also not adeno-associated virus, as the AAV used in the study is replication incompetent and should be addressed as vector.

Thank you for pointing out the confusion. We agree with reviewer and have made this clear by modifying the text to ‘replication incompetent hepatotropic AAV8-TBG-cre viral vector (AAV)’ in the revised manuscript.

6) How was the reticulocyte count determined?

The reticulocyte percentages were determined by a Complete Blood Count analyzer (Genesis, Oxford Science). We have added these details to the method section in the revised

manuscript.

7) There are various Lepr-Cre-deleter mice at Jackson lab (same for the other lines). It is not sufficient to name them Lepr-Cre or Prx1-Cre in the Materials and methods part, please give the correct name of the mouse line.

We have added the Jackson Lab stock number information in the Materials and methods section in the revised manuscript.

8) Figure legends for the supplementary Figures are missing.

We sincerely apologize for this omission. The legends for the supplementary figures have been added in the revised manuscript.

9) In respect to the discussion: The systemic effect of Thpo regulation has been discussed by deGraaf et al., 2010.

We have cited the paper and added discussion in the introduction section of the revised manuscript.

10) It would be of advantage for the reader, to mention and explain the results from their previous study (the tissue-specific deletion of Thpo under steady state conditions in the mouse) in the introduction.

We have added this information to the introduction section of the revised manuscript.

11) Platelet counts in PBS treated mice (Figure 2A) are lower than in untreated control mice (Figure 1G). What is the major difference between these controls?

Controls in Figure 1G are genotyping negative control mice treated with 5-FU.

Controls in the original Figure 2A (new Figure 2—figure supplement 1 F) are wild-type mice treated with PBS. 5-FU treatment increases platelet counts.

12) The paper would benefit to directly compare the data from the osteoblast and stromal cell deleted mice with the AAV treated mice in one graphic. Except for the difference in technology used to delete Thpo there is no obvious reason why the data are split up in the different figures.

We have considered the possibility of a direct comparison. Mice with Thpo deleted from the bone marrow did not have any phenotypes under the steady-state condition. A direct comparison among these mice under 5-FU or irradiation conditions would be appropriate (e.g. Figure 1). However, mice with Thpo deleted from the hepatocytes (AAV-treated mice) have hematopoietic phenotypes under the steady-state condition. We need to normalize the data under 5-FU and irradiation conditions to the baseline (mice with same genotype under the steady-state condition). It would be inappropriate and confusing to directly compare the data from the mice with Thpo deleted from the bone marrow with those with Thpo deleted from the liver (AAV-treated mice). Therefore, we separated the data of mice with Thpo deleted from the bone marrow from the data of mice with Thpo deleted from the liver (AAV-treated mice) into different figures.

13) The response in platelet counts after 5-FU administration in Mpl-/- mice has been studied (Levin et al., Blood 2001). These mice can respond in the absence of intact Thpo/Mpl signaling. These data are contradictory to what is shown here. What could be the explanation for that?

Thank you for raising this point. 5-FU induces the strongest platelet rebound at around 10 days after its administration (Experimental Hematology 2013 41:635, Blood 1992 80:904). We thus analyzed mice at day 10 after 5-FU administration. Note that Levin et al., (Blood 2001 98:1019) show that at day 10 after 5-FU treatment, Mpl-/- mice have significant defects in platelet recovery (see Figure 1A therein). At day 20-25, Mpl-/- mice can catch up and have relatively normal platelet counts compared to controls. Subsequently, their platelet levels decrease again. It appears that Mpl-/- mice can transiently produce platelet in respond to 5-FU through THPO-independent mechanisms, although the peak response is much lower and later than wild-type mice. Nonetheless, their day 10 data are consistent with ours in that efficient platelet rebound at this stage is Thpo/Mpl-dependent.

14) The major difference between using Cre-deleter mice in contrast to the AAV mediated application is that in the latter case Thpo is deleted in an adult mouse while in the other case, mice have develop under the condition that no Thpo is supplied from the bone marrow niche (osteoblasts or mesenchymal cells, accordingly). This questions whether these two scenarios can be safely compared. Similar to this, the inhibition of Thpo/Mpl signaling by a dominant-negative Mpl receptor in the adult mouse causes severe and lethal aplasia (Wicke et al., Mol Ther 2010), while the germ line knockout of Mpl or Thpo is in general well tolerated in the mouse and HSC defects get only more obvious after transplantation (Alexander et al., Blood 1996; Kimura et al., PNAS 1998; deSauvage et al., J Ex Med 1996; Yoshihara et al., CSC 2007, Qian et al., CSC 2007). What was the rational to use the AAV8 Cre delivery in contrast to using a deleter mouse with liver-specific Cre expression?

Thank you for raising this important point. Wicke et al., (Mol Ther 2010 18:343) used an Mpl overexpression model but not a dominant-negative Mpl model in adult mice. The adult inducible Thpo knockout mice have similar phenotypes in the hematopoietic system to the germline Thpo knockout mice (Science 2018 360:106), suggesting the adaptation to Thpo depletion during development is unlikely. However, this important point needs some consideration. No THPO protein (as can be detected by our reporter mice) is generated in the bone marrow. And mice with Thpo deleted from the bone marrow niche have no phenotypes, either under the steady-state condition or after myeloablation. It is thus unlikely that HSCs and the hematopoietic system adapt to Thpo deletion in these knockout mice. The most plausible explanation for the lack of phenotype is that bone marrow-derived THPO (if any) is not required for HSC regeneration after myeloablation. We have used mouse models with Cre specifically expressed in the hepatocytes (Alb-cre) to delete Thpo. Alb-cre; Thpofl/fl mice have a severe depletion of HSCs (virtually complete depletion) under the steady-state condition (Science 2018 360:106). It is thus technically difficult to normalize the parameters in myeloablation conditions to the baseline levels which have nearly no HSCs in these mice. In addition, Alb-cre; Thpofl/fl mice do not survive the myeloablation well. Thus, we used acute inducible Thpo deletion model with AAV8-TBG-cre.

15) The spread in the data after AAV application to mice may be caused by differences in the transduction efficiencies in the liver. How efficient was the deletion in the liver after AAV application?

We have quantified the deletion efficiency in the liver after the AAV application. Over 99% Thpo transcripts were depleted. The new data have been added to Figure 2-figure supplement 1 A in the revised manuscript.

16) In the transplantation experiment, the donor mice were treated with AAV-Cre. To address the role of hepatic Thpo in the recovery after transplantation, it would be of interest to delete Thpo in the recipients.

HSCs are defined by their capacity to reconstitute lethally irradiated recipient mice after transplantation. The purpose of the transplantation experiments is to show that there is a reduction of HSC function in mice with Thpo deleted from the hepatocytes by AAV treatment, independent of the surface marker profiles assayed by flow cytometry. Our data suggest that there is a reduction of HSC function in AAV-treated mice after myeloablation compared with controls.

17) The transplantation of irradiated BM cells should in general not lead to a successful engraftment of these cells. What is the aim of this experiment?

See response to question # 16. The purpose of the transplantation experiments is to confirm functional HSC reduction. We have transplanted a large number of bone marrow cells (1 million together with 1 million irradiated competitor cells) and observed significant differences between AAV- and PBS-treated mice.

18) What is the threshold of donor contribution that is judged as positive engraftment?

We set up the threshold based on background staining in recipient-type mice without transplantation in multilineage (myeloid, B and T cells, typically <0.5%). Mice with longterm multilineage reconstitution were regarded as positive engraftment.

19) As mentioned above, the study shows that, similarly to what occurs in bone marrow HSC maintenance in steady-state conditions, hepatic TPO appears to be a major functional trigger for hematopoietic recovery and HSC regeneration after myelosuppression caused by radio-chemotherapy injury.

To avoid misleading conclusions and thus contribute to clinical medicine more precisely, the experimental design of the study must relate to a tumor context, since radio-chemotherapies play an established role in the treatment of cancer patients. The genetically modified animal models must have pathophysiological characteristics as comparable as possible to those that occur in humans (for animal-to-human extrapolation).

We thus recommend that the reported experiments be flanked by others run upon a murine tumor model.

As the reviewer can see, our study here focuses on the identification of the cellular source of THPO for HSC regeneration and hematopoietic recovery after myeloablation. Through analyzing conditional knockout mice under two major clinically relevant myeloablative conditions (5-FU and irradiation), we showed that THPO from hepatocytes was important for HSC regeneration after myeloablation. Myeloablation is typically associated with chemo- or radio-therapy. Therefore, we believe that our findings are clinically relevant. Introducing a tumor to our experimental system may make the interpretation of the results complicated as the tumor (depending on the tumor type) may directly or indirectly contribute to the regulation of HSC regeneration. In addition, performing careful in vivo experiments on a murine tumor model using our Thpo conditional knockout mice as suggested by the reviewer requires extensive time and efforts. Although this is something worth investigating in the future, importantly, regardless what the results of these Thpo conditional knockout mice on a murine tumor model will be, they will not change our major conclusion in the current manuscript.

20) Since it has been shown that TPO is an important and nonredundant regulator of VEGF-A production in HSCs, and that is activates a VEGF-A internal autocrine loop, participating in the favorable TPO effects on HSC self-renewal and expansion (Kirito K et al., Blood. 2005; 105:4258) it would be interesting to investigate the contribution of this mechanism to the efficacy of hepatic TPO in hematopoietic recovery in stress conditions in genetically altered mouse models. This would be of interest in particular because VEGF-A, notoriously elevated in a tumor context, could also be counted among the endocrine factors which, together with TPO, contribute to bone marrow recovery, as suggested in the discussion.

Thank you for raising this interesting point regarding how THPO may promote HSC self-renewal and expansion. Kirito K et al., (Blood 2005 105:4258) show that THPO promotes VEGF-A expression in hematopoietic progenitors (Sca1+cKit+Gr1-). The authors further show that inhibition of the VEGF signaling with SU5416 (a VEGFR inhibitor) reduces the survival and proliferation of hematopoietic progenitors in response to THPO in vitro. Consistent with an earlier report (Nature 2002 417:954), the authors conclude that THPO promotes VEGF production in hematopoietic progenitors/HSCs probably through an intracellular autocrine loop. That is, the effects of VEGF are HSC-intrinsic and do not seem to depend on its secretion outside of the cell, but rather rely on its intracellular roles, since treating HSCs with cell permeable VEGFR inhibitors, but not soluble VEGFR-IgG chimeric protein, leads to their death and compromised reconstitution (Nature 2002 417:954). The focus of our current study is to elucidate the cellular source of THPO for HSC regeneration after myeloablation. The VEGF link is interesting, but it appears to be a downstream event after HSCs receive THPO signaling and probably should not be counted as an endocrine factor, particularly given that VEGF seems to act as an intracellular signal. Investigating the role of VEGF in vivo requires careful and long-term mouse genetic experiments. However, no matter what the results of these experiments may be, these studies will not impact our conclusions here. Nonetheless, we have performed qPCR to measure Vegfa levels in HSCs after Thpo was conditionally deleted from hepatocytes. We found no significant differences of Vegfa expression in HSCs from the conditional knockout mice compared with controls (see Author response image 2). This result is somewhat surprising. The cell purity could be a major factor as Kirito K et al., (Blood 2005 105:4258) examined Sca1+cKit+Gr1- hematopoietic progenitors while we used LSKCD150+CD48- HSCs.

Author response image 2
Deletion of Thpo from hepatocytes does not lead to changes in Vegfa expression in HSCs.

qPCR analysis showing that LSKCD150+CD48- HSCs from AAV-treated Thpofl/fl mice had similar Vegfa expression levels as PBS-treated controls. n = 3 mice for each condition.

21) In the discussion, in the light of their data, the authors assume that cancer patients with underlying liver disease, such as HCC, may have enhanced sensitivity to myeloablative conditioning because of liver-failure-related hepatic downregulated TPO levels. Since TPO mimetic drugs, such as romiplostim and eltrombopag, have been shown to improve recovery after ablative challenge, the authors suggest that these agents could help alleviate the adverse effects associated with myeloablative conditioning in such patients.

However, it has recently been reported that, in the HCC context (Vizio B et al., Int. J. Mol. Sci. 2021, 22, 1818), the increased hepatic TPO level appears to be a novel triggering signal to promote tumor progression. Therefore, curative strategies aimed at manipulating THPOR in HCC patients should be evaluated with extreme care.

The possibility of tumor conditioning should be pointed out, and recent literature in this connection included in the discussion.

Thank you for pointing out this potential issue. By no means do we suggest that promoting the THPO pathway will be universally beneficial. We have cited the paper and added discussion that strategies aimed at promoting the THPO pathway for myeloablation need careful evaluation as THPO could promote some other aspect of the pathology.

https://doi.org/10.7554/eLife.69894.sa2

Article and author information

Author details

  1. Longfei Gao

    1. Columbia Stem Cell Initiative, Columbia University Medical Center, New York, United States
    2. Department of Rehabilitation and Regenerative Medicine, Columbia University Medical Center, New York, United States
    3. Department of Microbiology and Immunology, Columbia University Medical Center, New York, United States, New York, United States
    Contribution
    Data curation, Formal analysis, Investigation, Writing - review and editing
    Contributed equally with
    Matthew Decker
    Competing interests
    none
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2331-8471
  2. Matthew Decker

    1. Columbia Stem Cell Initiative, Columbia University Medical Center, New York, United States
    2. Department of Rehabilitation and Regenerative Medicine, Columbia University Medical Center, New York, United States
    3. Department of Microbiology and Immunology, Columbia University Medical Center, New York, United States, New York, United States
    Contribution
    Data curation, Formal analysis, Investigation, Writing - original draft, Writing - review and editing
    Contributed equally with
    Longfei Gao
    Competing interests
    none
  3. Haidee Chen

    1. Columbia Stem Cell Initiative, Columbia University Medical Center, New York, United States
    2. Department of Rehabilitation and Regenerative Medicine, Columbia University Medical Center, New York, United States
    3. Department of Microbiology and Immunology, Columbia University Medical Center, New York, United States, New York, United States
    Contribution
    Investigation, Project administration
    Competing interests
    none
  4. Lei Ding

    1. Columbia Stem Cell Initiative, Columbia University Medical Center, New York, United States
    2. Department of Rehabilitation and Regenerative Medicine, Columbia University Medical Center, New York, United States
    3. Department of Microbiology and Immunology, Columbia University Medical Center, New York, United States, New York, United States
    Contribution
    Conceptualization, Funding acquisition, Project administration, Supervision, Writing - original draft, Writing - review and editing
    For correspondence
    ld2567@cumc.columbia.edu
    Competing interests
    none
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4869-8877

Funding

National Heart, Lung, and Blood Institute (R01HL132074)

  • Lei Ding

National Heart, Lung, and Blood Institute (R01HL153487)

  • Lei Ding

National Heart, Lung, and Blood Institute (R01HL155868)

  • Lei Ding

New York State Stem Cell Science (Training grant)

  • Longfei Gao

National Heart, Lung, and Blood Institute (1F30HL137323)

  • Matthew Decker

Rita Allen Foundation (Scholar Award)

  • Lei Ding

Irma T. Hirschl Trust (Research Awards)

  • Lei Ding

Leukemia and Lymphoma Society (Scholar Award)

  • Lei Ding

National Cancer Institute (P30CA013696)

  • Longfei Gao
  • Matthew Decker
  • Haidee Chen
  • Lei Ding

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by the National Heart, Lung and Blood Institute (R01HL132074). LG was supported by the NYSTEM Columbia training program in stem cell research and a Columbia Stem Cell Initiative Seed Grant. MD was supported by the Columbia Medical Scientist Training Program and the NIH (1F30HL137323). LD was supported by the Rita Allen Foundation, the Irma Hirschl Research Award, the Schaefer Research Scholar Program, and the Leukemia and Lymphoma Society Scholar award, R01HL153487 and R01HL155868. Images were collected in the Confocal and Specialized Microscopy Shared Resource of the Herbert Irving Comprehensive Cancer Center at Columbia University, supported by NIH grant P30CA013696 (National Cancer Institute). We thank M Kissner at the Columbia Stem Cell Initiative for help on flow cytometry.

Ethics

All mice were housed in specific pathogen-free, Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC)- approved facilities at the Columbia University Medical Center. All protocols were approved by the Institute Animal Care and Use Committee of Columbia University under AC-AAAZ9451.

Senior Editor

  1. Mone Zaidi, Icahn School of Medicine at Mount Sinai, United States

Reviewing Editor

  1. Hossein Ardehali, Northwestern University, United States

Reviewer

  1. Hossein Ardehali, Northwestern University, United States

Publication history

  1. Received: April 29, 2021
  2. Preprint posted: May 13, 2021 (view preprint)
  3. Accepted: August 27, 2021
  4. Accepted Manuscript published: August 31, 2021 (version 1)
  5. Version of Record published: September 22, 2021 (version 2)

Copyright

© 2021, Gao et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 930
    Page views
  • 159
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Longfei Gao
  2. Matthew Decker
  3. Haidee Chen
  4. Lei Ding
(2021)
Thrombopoietin from hepatocytes promotes hematopoietic stem cell regeneration after myeloablation
eLife 10:e69894.
https://doi.org/10.7554/eLife.69894