The HIF signaling pathway in osteoblasts directly modulates erythropoiesis through the production of EPO

Abstract

Osteoblasts are an important component of the hematopoietic microenvironment in bone. However, the mechanisms by which osteoblasts control hematopoiesis remain unknown. We show that augmented HIF signaling in osteoprogenitors results in HSC niche expansion associated with selective expansion of the erythroid lineage. Increased red blood cell production occurred in an EPO-dependent manner with increased EPO expression in bone and suppressed EPO expression in the kidney. In contrast, inactivation of HIF in osteoprogenitors reduced EPO expression in bone. Importantly, augmented HIF activity in osteoprogenitors protected mice from stress-induced anemia. Pharmacologic or genetic inhibition of prolyl hydroxylases1/2/3 in osteoprogenitors elevated EPO expression in bone and increased hematocrit. These data reveal an unexpected role for osteoblasts in the production of EPO and modulation of erythropoiesis. Furthermore, these studies demonstrate a molecular role for osteoblastic PHD/VHL/HIF signaling that can be targeted to elevate both HSCs and erythroid progenitors in the local hematopoietic microenvironment.

Figure 1. Generation of mice deficient for VHL or HIF-1 and HIF-2 in osteoblasts

A–C. ROSA-LacZ reporter expression of Osterix-Cre (A–B) in 8-week-old mouse tibia. Arrows in (A) point to osteoblasts at the endosteal bone surface and embedded in cortical bone with B-galactosidase activity. B. Shows the absence of detectable B-galactosidase activity in ROSA-LacZ mice without OSX-Cre. C. Arrows point to subset of hypertrophic chondrocytes expressing B-galactosidase activity. D. Efficient recombination of the VHL, HIF-1, and HIF-2 alleles in the bone of Osterix-Cre mutant mice. PCR analysis of genomic DNA isolated from bone of 1) OSX-Cre control, 2) OSX-VHL, 3) OSX-Cre control, and 4) OSX-HIF-1/HIF-2 mice. Abbreviations: conditional allele, 2-lox; recombined allele, 1-lox; and wild-type allele, WT. E. Immunohistochemical analysis of HIF-1 (top) and HIF-2 (bottom) expression in OSX-Cre tibias. Arrows point to osteoblasts expressing HIF-1 or HIF-2.

Figure 2. Augmented HIF activity in osteoblasts through VHL deletion expands the HSC niche

A. Histological analysis of H&E stained OSX-Cre tibias at 8-weeks of age. B–C. Histomorphometric analysis of trabecular bone volume (BV/TV) and trabecular number (Tb.N) in tibia (n=6). D. H&E stained tibia at 8-weeks of age. E–F. Histomorphometric analysis of trabecular bone volume (BV/TV) and trabecular number (Tb.N) in OSX-Cre tibias (n=6). G. Flow cytometric analysis (FACS) of c-Kit⁺ Lineage^low Sca-1⁺ (KLS) progenitors in OSX-Cre bone marrow. H. Frequency of KLS in OSX-Cre mutant bone marrow shown as percentage of total bone marrow cells at 8-weeks-of age (K, n = 7 in each group). I. FACS analysis of multipotent progenitor (MPP) and KLS-SLAM (HSC) of 8-week-old OSX-Cre bone marrow. J. Frequency of MPPs in OSX-Cre bone marrow shown as percentage of total bone marrow cells. For each group analyzed n = 4. K. Frequency of HSCs in OSX-Cre bone marrow shown as percentage of total bone marrow cells. For each group analyzed n = 4. FACS experiments shown are a representative experiment in which littermate controls were directly compared to OSX-mutant mice. All experiments were independently performed at least three times. See also Figure S1. All data are represented as mean +/− SEM.

Figure 3. Selective expansion of the erythroid lineage in OSX-VHL mice leads to the development of HIF-dependent polycythemia

A. CBC analysis of peripheral blood from 8-week-old OSX-Cre mutant mice (n = 8 for all groups expect n = 6 for OSX-VHL/HIF-1/HIF-2). B. Photograph of 8-week-old OSX-VHL and littermate control (OSX-CNTRL) paws. C. Photograph of spleens collected from 8-week-old OSX-CNTRL (left) and OSX-VHL (right) mice. D. Histological analysis of spleen from OSX-Cre mice at 40X (top) and 200X (bottom). E. Average hematocrit (HCT) of OSX-Cre mice at 8-weeks-of age ( CNTRL, n = 8; VHL, n = 8; VHL/HIF-1, VHL/HIF-2, VHL/HIF-1/HIF-2, n = 6). F–I. Frequency of myeloid progenitor KLS− (cKit^high Lineage^low Sca1⁻); granulocyte-macrophage progenitors (GMP, KLS− FcgRII/III^high CD150−); Pre-megakaryocyte-erythroid, Pre-erythroid, megakaryocyte (Pre MegE/MkP/Pre CFU-E, KLS− FcgRII/III^low CD150+) and mature megakaryocyte (Mk, KLS− CD41+) cells within OSX-CNTRL and OSX-VHL bone marrow ( n = 4 for all groups). J. FACS analysis of erythroid lineage CD71+ Ter119+ in bone marrow and spleen of OSX-Cre mice. K–L. Significant increase in the frequency of erythroid progenitors in OSX-VHL mutant bone marrow (F, n = 4 in each group) and spleen (G, n = 3 in each group) as determined by FACS analysis. M–N. Quantification of CFU-E ( n = 5) and BFU-E ( n = 3) colonies in the bone marrow of OSX-mice. FACS experiments shown are a representative experiment in which littermate controls were directly compared to OSX-mutant mice. All experiments were independently performed at least three times. Data are represented as mean +/− SEM.

Figure 4. Increased hematocrit in OSX-VHL mice develops in an EPO-dependent manner associated with increased EPO expression in bone and decreased EPO expression in the kidney

A. Analysis of EPO protein levels in the serum of 2-month old OSX-mice determined by ELISA (CNTRL, VHL/HIF-1, VHL/HIF-2, and VHL/HIF-1/HIF-2, n = 6; VHL, n = 8). B. Soluble EPO-receptor therapy (EPOR) significantly reduces hematocrit in OSX-VHL mice. Average hematocrit of 6.5 week old OSX-Cre mice treated with adenovirus expressing FC control or soluble Epo-receptor (EPO-R). Statistical significant differences (p<0.05) in hematocrit were observed between VHL-FC and CNTRL-FC (*) treated mice as well as VHL-FC and VHL-EPOR (#) treated mice (n = 4 each group). C. Real time PCR analysis of EPO mRNA expression in tissues collected from 8-week-old OSX-Cre mice (n = 3 mice per group). D. Real time PCR analysis of EPO expression in bone marrow stromal cells isolated from OSX-Cre hindlimbs (n = 3 mice per group). Data are represented as mean +/− SEM. See also Figure S2.

Figure 5. HIF-1 and HIF-2 deletion in osteoblasts reduces EPO expression and erythroid progenitors in the bone but does not cause anemia

A. Real time PCR analysis of EPO expression in primary osteoblast cultures isolated from calvarie or bone marrow exposed to 21% or 2% oxygen (n = 5 each group). B. Real time PCR analysis of EPO expression in neonatal hindlimbs (without growth plate) from OSX-CNTRL and OSX-HIF-2 mice ( n = 4 each group). C–D Frequency of erythroid progenitors in 8-week-old OSX-CNTRL and OSX-HIF-1/HIF-2 (C) and OSX-CNTRL and OSX-HIF-2 (D) bone marrow (L, n = 4 each group). E–F. Frequency of erythroid progenitors in 8- week-old OSX-CNTRL and OSX-HIF-1/HIF-2 (E) and OSX-CNTRL and OSX-HIF-2 (F) spleen (L, n = 4 each group). G. CBC analysis of peripheral blood from 8-week-old OSX-Cre mutant mice. n = 8 in each group. H. OSX-HIF-1/HIF-2 mice have a normal response to erythropoietic stress. Mice (n = 5 in each group) were injected with phenylhydrazine on days 0 and 1. The proportion of reticulocytes in the red cell population was analyzed over 8 days. Corrected reticulocyte count (%) = reticulocyte count (%) X (hematocrit/45). Data are represented as mean +/− SEM. See also Figure S3.

Figure 6. Modulation of the PHD/VHL/HIF pathway is sufficient to induce EPO production in adult bone and protect mice from anemia

A. Augmented HIF signaling in osteoblasts protects mice from anemia. Percent hematocrit in 8-week-old OSX-Cre mice 48 hours after initial phenylhydrazine (PHZ) treatment ( n = 5 in each group). * indicate a statically difference compared to untreated OSX-Control mice as determined by Students t test (p<0.05). B. Pharmacologic inhibition of PHDs induces EPO expression in adult bone. EPO mRNA expression in 8-week-old bone 7 hours following intra-femoral injection of saline or DMOG ( n = 3 in saline and n = 3 in DMOG groups). C. Genetic inhibition of PHD1/2/3 activates Epo expression in bone. EPO mRNA expression in 8 week old OSX-Cre mutant mice ( n = 5 per group). D. Prolyl hydroxylase inhibition in osteoblasts increases hematocrit in 8-weekold mice ( n = 5 per group). Data are represented as mean +/− SEM.

Figure 7. The HIF signaling pathway in osteoblasts directly modulates erythropoiesis through the production of EPO and the number of HSCs

Osteoblasts produce EPO through a HIF dependent mechanism under physiologic and pathophysiologic conditions. Osteoblasts express HIF-1 and HIF-2 that transcriptionally activate EPO expression. In conditions of augmented HIF signaling (dashed line), osteoblasts produce both local and systemic levels of EPO that stimulate erythropoiesis in the bone marrow and spleen. Constitutive hyperactivation of HIF signaling leads to increased red blood cells (RBC) and oxygen in the blood that feeds back to suppress renal EPO production. Augmented HIF signaling in osteoblasts through VHL or PHD1/2/3 inhibition is sufficient to drive EPO expression in adult bone, elevate hematocrit, and protect mice from stress-induced anemia. Furthermore, pharmacologic inhibition of PHDs using prolyl hydroxylase inhibitors (PHI) is sufficient to activate EPO in adult bone. Additionally, augmented HIF activity in osteoblasts expands HSCs indicating that therapeutic manipulation of the PHD/VHL/HIF pathway in osteoblasts is a therapeutic strategy to elevate both EPO and HSCs in the local hematopoietic microenvironment.