The hepcidin regulator erythroferrone is a new member of the erythropoiesis-iron-bone circuitry

Background: Erythroblast erythroferrone (ERFE) secretion inhibits hepcidin expression by sequestering several bone morphogenetic protein (BMP) family members to increase iron availability for erythropoiesis. Methods: To address whether ERFE functions also in bone and whether the mechanism of ERFE action in bone involves BMPs, we utilize the Erfe-/- mouse model as well as β–thalassemic (Hbbth3/+) mice with systemic loss of ERFE expression. In additional, we employ comprehensive skeletal phenotyping analyses as well as functional assays in vitro to address mechanistically the function of ERFE in bone. Results: We report that ERFE expression in osteoblasts is higher compared with erythroblasts, is independent of erythropoietin, and functional in suppressing hepatocyte hepcidin expression. Erfe-/- mice display low–bone–mass arising from increased bone resorption despite a concomitant increase in bone formation. Consistently, Erfe-/- osteoblasts exhibit enhanced mineralization, Sost and Rankl expression, and BMP–mediated signaling ex vivo. The ERFE effect on osteoclasts is mediated through increased osteoblastic RANKL and sclerostin expression, increasing osteoclastogenesis in Erfe-/- mice. Importantly, Erfe loss in Hbbth3/+mice, a disease model with increased ERFE expression, triggers profound osteoclastic bone resorption and bone loss. Conclusions: Together, ERFE exerts an osteoprotective effect by modulating BMP signaling in osteoblasts, decreasing RANKL production to limit osteoclastogenesis, and prevents excessive bone loss during expanded erythropoiesis in β–thalassemia. Funding: YZG acknowledges the support of the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (R01 DK107670 to YZG and DK095112 to RF, SR, and YZG). MZ acknowledges the support of the National Institute on Aging (U19 AG60917) and NIDDK (R01 DK113627). TY acknowledges the support of the National Institute on Aging (R01 AG71870). SR acknowledges the support of NIDDK (R01 DK090554) and Commonwealth Universal Research Enhancement (CURE) Program Pennsylvania.


ABSTRACT
Erythroblast erythroferrone (ERFE) secretion inhibits hepcidin expression by sequestering several bone morphogenetic protein (BMP) family members to increase iron availability for erythropoiesis. We report that ERFE expression in osteoblasts is higher compared with erythroblasts, is independent of erythropoietin, and functional in suppressing hepatocyte hepcidin expression. Erfe -/mice display low-bone-mass arising from increased bone resorption despite a concomitant increase in bone formation. Consistently, Erfe -/osteoblasts exhibit enhanced mineralization, Sost and Rankl expression, and BMP-mediated signaling ex vivo. The ERFE effect on osteoclasts is mediated through increased osteoblastic RANKL and sclerostin expression, increasing osteoclastogenesis in Erfe -/mice. Importantly, Erfe loss in β-thalassemic (Hbb th3/+ ) mice, a disease model with increased ERFE expression, triggers profound osteoclastic bone resorption and bone loss. Together, ERFE exerts an osteoprotective effect by modulating BMP signaling in osteoblasts, decreasing RANKL production to limit osteoclastogenesis, and prevents excessive bone loss during expanded erythropoiesis in β-thalassemia.

INTRODUCTION
It has become increasingly clear that both erythropoiesis and skeletal homeostasis are susceptible to changes in iron metabolism, especially during stress or ineffective erythropoiesis.
Both TDT and NTDT generally present with anemia and iron overload, requiring iron chelation therapy. Surprisingly, however, TDT patients show more marked decrements in bone mineral density (BMD) compared with NTDT, despite chronic RBC transfusion that suppresses expanded and ineffective erythropoiesis. Optimization of RBC transfusion has reduced the frequency of overt bone disease, such as frontal bossing, maxillary hyperplasia and limb deformities, and importantly, has enabled prolonged survival 8 . Nonetheless, growth patterns have not significantly improved 9 , and low bone mass remains a frequent, significant, and poorly understood complication even in optimally-treated patients. As such, β-thalassemia-induced bone disease has warranted formal guidelines for management 10 .
Proposed mechanisms of bone loss in β-thalassemia include direct effects of abnormal erythroid proliferation 11,12 , increased circulating erythropoietin (Epo) 13 , iron toxicity 14 , oxidative stress 15 , inflammation 16 , and changes in bone marrow adiposity 17 . Strong negative correlations between BMD and systemic iron concentrations 18 and the profound bone loss noted in patients with hereditary hemochromatosis 19 underscore the premise that BMD and iron homeostasis may be associated causally. However, mice lacking the transferrin receptor, TFR2, display increased rather than decreased bone mass and mineralization despite iron overload 20 . These latter findings prompted us to take a fresh look at the mechanisms underpinning bone loss in diseases of iron dysregulation. Erythroferrone (ERFE), a protein secreted by bone marrow erythroblasts, is a potent negative regulator of hepcidin 21 , which, in turn, inhibits iron absorption and recycling 22 . Hepcidin suppression enables an increase in iron availability during stress erythropoiesis. Very recently, ERFE has been shown to bind and sequester certain members of the bone morphogenetic protein (BMP) family, prominently BMP2, BMP6 and the BMP2/6 heterodimer 23,24 . BMPs stimulate bone formation by osteoblasts during skeletal development, modeling, and ongoing remodeling 25 . We thus hypothesized that, by modifying BMP availability, ERFE may be a key player in the newly discovered erythropoiesis-iron-bone circuitry. As a result, ERFE may also be an important link between altered iron metabolism, abnormal erythropoiesis, and bone loss in β-thalassemia.
Here, we demonstrate that global deletion of Erfe in mice results in a low-bone-mass phenotype, which is phenocopied in β-thalassemic mice lacking ERFE (i.e., Hbb th3/+ ;Erfe -/mice). Despite the osteopenic phenotype, we found that ERFE loss stimulated mineralization in cell culture. The net loss of bone in Erfe -/mice in the face of a pro-osteoblastic action could therefore only be attributed to a parallel increase in bone resorption, which we found was the case in both Erfe -/and Hbb th3/+ ;Erfe -/mice. Furthermore, the increase in osteoclastogenesis was osteoblastmediated exerted via an increased expression of Sost and Tnfsf11, the gene encoding RANKL.
Together, our data provide compelling evidence that ERFE loss induces BMP-mediated osteoblast differentiation, but upregulates Sost and Tnfsf11 to increase osteoclastogenesis with net bone loss. Therefore, high ERFE levels in β-thalassemia are osteoprotective and prevent the bone loss when erythropoiesis is expanded.

Mouse Lines
C57BL/6 and β-thalassemic (Hbb th3/+ ) mice 33 were originally purchased from Jackson Laboratories. Erfe -/mice were a generous gift from Tomas Ganz (UCLA) 21 . Progeny of Erfe -/mice crossed with Hbb th3/+ yielded Hbb th3/+ ;Erfe -/mice. The mice have been backcrossed onto a C57BL/6 background for more than 11 generations. All mice had ad libitum access to food and water and were bred and housed in the animal facility under AAALAC guidelines. Experimental protocols were approved by the Institutional Animal Care and Use Committee at Icahn School of Medicine at Mount Sinai.
Images were analyzed by TrapHisto and OsteoidHisto 35 . On the day of sacrifice, BMD was also measured in intact mice 36 . Frozen bone sections were incubated for 4 min at room temperature in Alkaline Phosphatase Substrate Solution ImmPACT Vector Red (Vector Laboratories). After washing with buffer, the sections were counterstained with hematoxylin (Vector Laboratories) and mounted with VectaMount AQ Mounting Medium (Vector Laboratories). Sections were visualized using Olympus BH-2 Microscope and images obtained with OMAX A35180U3 Camera were analyzed by ImageJ.

Isolation and Culture of Bone Marrow Cells
Erythroblasts were isolated from bone marrow and purified using CD45 beads, as

Primary Hepatocyte Culture
Hepatocytes were isolated by perfusion with collagenase and liver digestion, as described previously 40 . Briefly, 0.025% (w/v) collagenase type IV (Gibco) and 5 mM CaCl2 was added to

Quantitative PCR
RNA was purified from osteoblasts, osteoclasts, erythroblasts, and hepatocytes using PureLink RNA (Sigma) and analyzed with SuperScript III Platinum SYBR Green One-Step (Invitrogen). As previously described 41,42 , ΔΔCT values were used to calculate fold increases relative to β-actin, α-tubulin, and RLP4. Primers are listed in Table I.

Western Immunoblotting and ELISA
For Western immunoblotting, differentiated cells at day 3 were lysed in ice cold SDS page lysis buffer (2% SDS, 50 mM Tris-HCl, pH 7.4, 10 mM EDTA) with protease and phosphatase inhibitors. 20 µg of heat-denatured protein was loaded onto a 10% gel, run, and transferred onto a 0.4 µm nitrocellulose membrane (Thermo Scientific). After blocking with 5% BSA in Trisbuffered saline with 1% Tween-20 (TBS-T), the membranes were incubated with primary antibodies to signaling proteins (Table II) overnight at 4°C, washed, and incubated with the corresponding HRP-conjugated secondary antibodies at room temperature. Proteins were visualized using the ImageQuant LAS 4010 and quantified using Image J. Osteoblast supernatants from wild type and Erfe -/mice were collected and centrifuged for 10 min at 10,000 x g, and BMP2 (Abnova) and RANKL (R&D) concentrations were measured by ELISAs. Serum BMP2 concentration was determined using mouse BMP2 ELISA (abnova, KA0542), per manufacturers instructions. ERFE concentration in conditioned media was determined as described 43 with the substitution of DELFIA europium-conjugated streptavidin for horseradishperoxidase-conjugated streptavidin. Fluorescence was measured by CLARIOstar plate reader.

Complete Blood Counts
Peripheral blood (100 µL from each mouse) was collected from the retro-orbital vein in EDTA-coated tubes and analyzed by IDEXX Procyte Hematology Analyzer.

Statistical Analyses
Data are reported as means ± SEM. Unpaired Student's t-test was used to determine if differences between groups were significant at P<0.05.

RESULTS
To understand if ERFE has a role in regulating skeletal integrity in health, we first studied the effect of ERFE loss on BMD and bone remodeling in adult Erfe -/mice, as well as in compound mutant mice in which the Erfe gene was deleted on a β-thalassemia Hbb th3/+ background.
Compared with wild type littermates, both 6-week-old and 5-month-old male Erfe -/mice showed significant reductions in whole body BMD, and BMD at mainly-cortical (femur and tibia) sites ( Fig.   1A and 1B). However, in contrast to young mice, the older Erfe -/mice did not show a difference in lumbar spine BMD compared with wild type littermates. Interestingly, unlike hypogonadal bone loss, which is predominantly trabecular, the sustained reduction in femur and tibia BMD is consistent with prominent cortical loss seen in patients with β-thalassemia 2,3 .
Bone resorption and bone formation are tightly coupled to maintain bone mass during each remodeling cycle 44 . Bone is lost when either both processes are increased-with resorption exceeding formation, as in hypogonadism-or when there is uncoupling in which formation decreases while resorption rises, as in glucocorticoid excess 44 . To differentiate between relative increases and uncoupling, we measured both formation and resorption in intact bone. Dynamic histomorphometry performed after the sequential injections of calcein and xylenol orange, which yielded dual fluorescent labels, allowed us to derive parameters of bone formation. We observed that mineralizing surface (MS), mineral apposition rate (MAR) and bone formation rates (BFR) were all increased in young Erfe -/mice, consistent with the pro-osteoblastic (anabolic) action of ERFE deficiency (see below) ( Fig. 1C and 1D). No differences in MS, MAR, and BFR were noted in 5-month-old mice (Fig. 1D). We also analyzed alkaline phosphatase stained sections of femurs to find no difference in osteoblast surfaces (Ob.S) or osteoblast number (N.Ob) per bone surface (BS) in 5-week-old Erfe -/relative to wild type littermates (Fig. 1E).
Finally, to study whether an increase in osteoclastic bone resorption caused the notable reduction in BMD in Erfe -/mice, we measured TRAP-positive osteoclast surfaces (Oc.S) and number (N.Oc) per bone surface (BS). Both Oc.S/BS and N.Oc/BS were increased significantly in Erfe -/compared with wild type bones in older mice, and to a lesser extent, in younger mice (Fig. 1F). Thus, the overall low-bone-mass phenotype in Erfe -/mice primarily resulted from a relative increase in osteoclastic bone resorption over osteoblastic bone formation, suggesting that ERFE has a function in preventing skeletal loss. To confirm that decreased BMD in Erfe -/mice did not result from changes in erythropoiesis, we measured circulating red blood cells (RBCs) and reticulocytes, and bone marrow erythroblasts. We also measured spleen weight given the ubiquity of compensatory erythropoiesis that results in splenomegaly. Our results show no differences between 6-week-old wild type and Erfe -/mice ( Supplementary Fig. 1), consistent with what has been previously reported in Erfe -/mice 21 .
To probe the mechanism of action of ERFE on osteoblastic bone formation and osteoclastic bone resorption, we first asked which cells in bone marrow produce ERFE, and whether secreted ERFE was functional. Intriguingly, time course studies in differentiating osteoblasts revealed that Erfe expression was 10-and 2-fold higher at 3 and 21 days of culture, respectively, compared with cultured erythroblasts--the only previously known source of ERFE in bone marrow ( Fig. 2A). Furthermore, Erfe expression in mature osteoclasts was similar to cultured erythroblasts, with little expression in immature osteoclasts (Fig. 2B). Likewise, conditioned media from osteoblast cultures revealed increased ERFE concentration at 3 days with no differences in conditioned media from osteoclast cultures (Fig. 2C).
To determine whether osteoblast-derived ERFE was functional, we established a bioassay based on the known inhibitory action of ERFE on hepcidin (Hamp) expression. For this, wild type hepatocytes were exposed to supernatants from differentiating wild type or Erfe -/osteoblasts. Hamp expression was suppressed with Erfe -/osteoblast supernatants, but importantly, this suppression was significantly greater with wild type supernatants (Fig. 2D)

. No
Hamp suppression was evident with wild type or Erfe -/osteoclast supernatants. This latter suggests that osteoblast-but not osteoclast-derived ERFE is functional. However, as Erfe -/supernatants also suppressed Hamp expression, other yet unknown osteoblast-derived factors likely function in hepcidin regulation. Finally, unlike in erythroblasts, Erfe expression in mature osteoblasts or osteoclasts was not responsive to Epo (Fig. 2E).
Given that osteoblasts secrete ERFE that is known to inhibit hepcidin 21 by sequestering BMPs 23,24,45 that are skeletal anabolics 25 , we measured serum BMP2 concentration to find elevated BMP2 levels in Erfe -/relative to wild type mice (Fig. 3A). Given the specific importance of BMP2 in bone remodeling 26 3C).
To further understand how ERFE impacts BMP2-mediated signaling, we evaluated the effect of BMP2 on wild type and Erfe -/osteoblasts in vitro. Treatment with BMP2 (50 ng/ml) in osteoblast cultures showed that pSmad1/5/8 and pERK signaling was not further induced in Erfe -/relative to wild type osteoblasts ( Fig. 3C and 3D). In all, the data establish that increased BMP2 in Erfe -/mice leads to maximal induction of BMP signaling that remains unaffected by the further addition of BMP2. This supports the hypothesis that ERFE functions in bone by sequestering BMP2, thus, attenuating downstream signaling.
We studied whether the stimulation of bone formation in Erfe -/mice was due to a cellautonomous action of ERFE on osteoblasts. For this, we compared the ability of wild type and Erfe -/bone marrow stromal cells ex vivo to differentiate into mature mineralizing colony forming units-osteoblastoid (Cfu-ob). Stromal cells from 5-month-old Erfe -/mice showed enhanced von Kossa staining of mineralizing Cfu-ob colonies (Fig. 4A). This mineralizing phenotype was associated with enhanced expression of the osteoblast transcription factors Runx2 and Sp7, and downstream genes Sost and Tnfsf11 46,47 , increased supernatant RANKL levels, and suppressed expression of Opg (Figs. 4B, 4C). Enhanced RANKL profoundly increases osteoclastogenesis, as noted below, while sclerostin, encoded by the Sost gene, reduces the production of OPG, hence further increasing osteoclast formation.
Given that Erfe is expressed in osteoclasts, and that Erfe -/mice display a pro-resorptive phenotype, we questioned whether ERFE directly affected the osteoclast, or whether the action resulted via a primary osteoblastic effect. Erfe -/bone marrow cell cultures derived from 5-monthold mice showed no difference in TRAP-positive osteoclast number compared to wild type cultures (Fig. 4D). Consistent with this, the program of osteoclast gene expression remained unchanged in these 5-day cultures (Fig. 4E). The data collectively suggest that the absence of ERFE results in the de-sequestration of BMP2, stimulates the osteoblast to upregulate RANKL and sclerostin, and thus enhances osteoclastic bone resorption indirectly.
To confirm that decreased BMD in Hbb th3/+ ;Erfe -/mice did not result from further expanded erythropoiesis, we measured circulating red blood cells (RBCs) and reticulocytes, bone marrow erythroblasts, and spleen weight. Our results demonstrate a mildly decreased RBC count and hemoglobin, but no differences in spleen weight or bone marrow erythroblasts between 6-weekold Hbb th3/+ and Hbb th3/+ ;Erfe -/mice ( Supplementary Fig. 2). This is consistent with what has been previously reported in Hbb th3/+ ;Erfe -/mice 43 .

DISCUSSION
To date, the only known function of ERFE was on hepatocellular hepcidin expression exerted through the sequestration of BMPs 23,24 . Using genetically-modified mice and in vitro assays, we identify a new role for ERFE in skeletal protection. First, we show that Erfe expression is higher in osteoblasts comapred with erthyroblasts. Second, we find that ERFE is a potent down-regulator of BMP2-mediated signaling and RANK-L production by osteoblasts. Third, ERFE loss in vivo enhances bone formation, while also stimulating resorption by inducing the expression of osteoblastic Tnfsf11 and Sost. The net effect of these opposing changes is bone loss in both young and old mice (Fig. 6). Fourth, although also produced by osteoclasts, ERFE displays no cell-autonomous actions on osteoclast function. Taken together, and consistent with prior inferences [27][28][29][30][31][32]  We have also used the β-thalassemia mouse, Hbb th3/+ , as a relevant disease model to study a role for ERFE in β-thalassemia, a condition with known elevations in ERFE. Chronic erythroid expansion in β-thalassemia is associated with a thinning of cortical bone resulting in bone loss 2,3 . It is therefore surprising that patients with the more severe forms of β-thalassemia, namely TDT, in whom RBC transfusions lead to suppression of endogenous erythropoiesis, exhibit significantly greater decrements in BMD than NTDT patients 49 . We have previously shown that ERFE is suppressed post-transfusion in TDT patients, and is significantly higher in NTDT patients 50 . Our finding of a marked reduction of bone mass in Hbb th3/+ mice with geneticallydeleted Erfe (or Hbb th3/+ ;Erfe -/mice) compared with Hbb th3/+ mice provides strong evidence for a protective function of ERFE in preventing further worsening of the bone loss phenotype in βthalassemia.
Taken together, our findings uncover ERFE as a novel regulator of bone mass via its modulation of BMP signaling in osteoblasts. In addition, because RBC transfusion suppresses erythropoiesis and thus decreases ERFE in both mice 48 and patients 50 with β-thalassemia, a relative decrement of ERFE may explain the more severe bone disease in TDT than in NTDT patients. As a consequence, our findings identify ERFE as a promising new therapeutic target for hematologic diseases associated with bone loss, such as β-thalassemia.  Figure 6: Putative Osteoprotective Function of ERFE in Health and in β-Thalassemia. In conditions of elevated ERFE (A), such as β-thalassemia, more BMP2 and BMP6 is sequestered, decreasing signaling through the BMP/Smad and ERK pathways. This would results in decreased Sost and Rankl expression to decrease osteoclastogenesis and bone resorption. In contrast, when ERFE is low (B), increased BMP2 leads to increased BMP/Smad and ERK signaling, increased Sost and Rankl expression and thereby, Sclerostin and RANKL release-this results in a greater suppression of Wnt signaling and increased osteoclastogenesis with consequent decrease in bone formation. Abbreviations: ERFE = erythroferrone; BMP = bone morphogenetic protein; BMPR = BMP receptor; SOST = sclerostin; RANLKL = receptor activator of nuclear factor kappa-Β ligand; OPG = osteoprotegrin; LRP = lipoprotein receptor-related protein; Wnt = wingless-type MMTV integration site family.