TGF-β Regulates Sclerostin Expression via the ECR5 enhancer

Abstract

Wnt signaling is critical for skeletal development and homeostasis. Sclerostin (Sost) has emerged as a potent inhibitor of Wnt signaling and, thereby, bone formation. Thus, strategies to reduce Sclerostin expression may be used to treat osteoporosis or non-union fractures. Transforming growth factor-beta (TGF-β) elicits various effects upon the skeleton both in vitro and in vivo depending on the duration and timing of administration. In vitro and in vivo studies demonstrate that TGF-β increases osteoprogenitor differentiation but decreases matrix mineralization of committed osteoblasts. Because Sclerostin decreases matrix mineralization, this study aimed to examine whether TGF-β achieves such inhibitory effects via transcriptional modulation of Sost. Using the UMR106.01 mature osteoblast cell line, we demonstrated that TGF-β TGF-β₁, -β₂, -β₃ and Activin A increase Sost transcript expression. Pharmacologic inhibition of Alk4/5/7 in vitro and in vivo decreased endogenous Sost expression, and siRNA against Alk4 and Alk5 demonstrated their requirement for endogenous Sost expression. TGF-β₁ targeted the Sost bone enhancer ECR5 and did not affect the transcriptional activity of the endogenous Sost promoter. These results indicate that TGF-β₁ controls Sost transcription in mature osteoblasts, suggesting that Sclerostin may mediate the inhibitory effect of TGF-β upon osteoblast differentiation.

Introduction

Since the discovery in the past decade that the Wnt glycoprotein co-receptor Lrp5 regulates bone mass [1, 2], tremendous efforts have attempted to elucidate the mechanisms involved. Wnts are ligands for Lrp4, 5, and 6, and a subset of Wnts increase the osteogenic commitment of bone marrow stem cells, enhance matrix formation, and decrease apoptosis of osteoblasts and osteocytes [3, 4]. Regulation of Wnt signaling occurs via secreted decoy receptors (secreted frizzled-related protein [sFRP]) or antagonists (Sclerostin [SOST], Dickkopf [DKK]) that bind to Lrp4 - 6 to prevent Wnt-Lrp interactions, and subsequent signal transduction [5, 6]. As activating mutations in LRP4-6 promote high bone mass (HBM) phenotypes [1, 2, 7, 8], complementary phenotypes emerge from deletion of Lrp4/5/6 antagonists [9–15]: deletion of sFRPs increases trabecular bone [15] and bone mineral density [14], and deletion of DKK isoforms or SOST increases markers of bone formation and bone mass [9, 10].

The influence of SOST on skeletal formation and function is phenotypically observed via loss of Sclerostin protein, which is achieved by two distinct genetic mechanisms. One set of mutations occur within the SOST transcript and comprise either nonsense mutations in exon 2 or aberrant splice sites resulting in null alleles [10]. These mutations cause sclerosteosis in humans (MIM 269500), which is characterized by generalized cortical hyperostosis accompanied by occasional syndactyly of the digits [10]. A highly similar bone mineral density phenotype is observed in van Buchem disease patients (MIM 239100) who also have severe skeletal hyperplasia, but carry no mutations in the SOST gene. Instead, van Buchem results from the deletion of a 52kb non-coding region (also refered to as van Buchem deletion region) that is 35kb downstream of SOST [11]; this van Buchem deletion region functions in cis to enhance SOST transcription in bone. We have previously demonstrated that an evolutionarily-conserved region present within the van Buchem deletion region, termed ECR5, is sufficient to drive reporter assays in bone cells, in vitro and in vivo [13], and confers responsiveness to parathyroid hormone (PTH) [16].

The TGF-β superfamily is composed of more than 40 structurally and functionally related cytokines that regulate a variety of biological processes including morphogenesis, proliferation, stem cell differentiation, apoptosis, and epithelial-to-mesenchymal transition [17]. The superfamily clusters into the subfamilies TGF-β, bone morphogenetic proteins (BMPs), growth and differentiation factors (GDFs), activins and inhibins, and Mullerian inhibiting factor (MIF) [18]. The TGF-β subfamily contains three distinct proteins—TGF-β₁, -β₂, and -β₃— which exert pleiotropic effects upon cells responsible for maintaining or altering skeletal architecture. Indeed, the TGF-β subfamily demonstrates chemotactic effects on osteoprogenitors during endochondral condensation [19], promotes proliferation and differentiation of early osteoprogenitors, yet it also decreases matrix formation in fully-differentiated osteoblasts (reviewed in Janssens et al. [20] and Bonewald et al. [21], amongst others). TGF-β_1–3 can interact with osteotropic factors like PTH [22] or prostaglandin E₂ [23] to enhance bone formation. Conversely, factors like BMPs [24–26], PTH [16, 27, 28], and prostaglandin E₂ [29] regulate Wnt signaling via manipulation of Wnt or Lrp5/6 antagonist expression. BMP signaling through BMPR1A increases Sost expression and decreases Wnt signaling [30], but the influence of other TGF-β superfamily members on Sclerostin expression has not yet been explored. Provided the evidence for a biphasic influence of TGF-β upon osteogenic differentiation, we hypothesized that TGF-β increases Sost transcription in mature osteoblasts, and sought the intracellular mechanisms involved.

Materials and methods

Cell culture

UMR106.01 cells were kindly provided by Dr. Alexander Robling (Indiana University School of Medicine). Cells were cultured in MEM with Earle’s Salts (Invitrogen), which was supplemented with 10% fetal bovine serum (FBS; Invitrogen) and 1% penicillin-streptomycin (Invitrogen). Cells were maintained in a standard humidified incubator at 37°C/95% air / 5% CO₂, and were routinely sub-cultured with 0.05% trypsin when 75–90% confluent. Unless otherwise indicated, cells were seeded for experiments at 5k/cm², and growth factor supplements were added two days later.

Reagents

TGF-β₁, TGF-β₂, TGF- β₃, and Activin A (R&D Systems) and human parathyroid hormone(1–34) (hPTH(1–34); Bachem) were dissolved in 0.1% BSA in PBS and stored at −20°C. Cycloheximide was purchased from Sigma; SIS3 and SB431542 were from Calbiochem. Purified RNA from adult murine calvariae or femora (Zyagen), used to examine Alk4/5/7 expression were purchased from Zyagen.

Quantitative PCR (qPCR)

At indicated time points, cell culture samples were washed with PBS, collected in RLT buffer with 2-mercaptoethanol (Qiagen), from which RNA was purified using RNeasy Kit (Qiagen). RNA purity was tested by measuring the absorbance at 260 and 280nm. One microgram of RNA was reverse-transcribed with QuantiTect Reverse Transcription kit (Qiagen), which includes genomic DNA elimination. qPCR was performed using primers listed in Table 1 and either QuantiFast Probe PCR Kit (Qiagen) or QuantiFast SYBR Green PCR Kit (Qiagen). Cycling conditions were 95°C for 3 minutes (5 minutes for SYBR), followed by 40 cycles of 95°C for 3 seconds (10 seconds for SYBR) and 30 seconds at 60°C. qPCR results were calculated relative to internal control (Rpl32 or Tbp; 2^−ΔCt) with the exception of Figure 3A and 3B, results were further normalized to control, time-matched conditions (2^−ΔΔCt) [31].

TABLE 1.

qPCR primers

Target	Source	Assay no.	Species	Chemistry
Sost	Applied Biosystems	Rn0057791_m1	Rat	TaqMan
Rpl32	Applied Biosystems	Rn00820748_g1	Rat	TaqMan
Alk4 (Acvr1b)	Qiagen	Mm_Acvr1b_1_SG	Mouse	SYBR Green
Alk5 (Tgfbr1)	Qiagen	Mm_Tgfbr1_1_SG	Mouse	SYBR Green
Alk7 (Acvr1c)	Qiagen	Mm_Acvr1c_2_SG	Mouse	SYBR Green
Tbp	Qiagen	Mm_Tbp_1_SG	Mouse	SYBR Green

Figure 3. Alk4 and Alk5 regulate Sost expression.

RNA from UMR106.01 cells (A) and (B) murine calvaria and long bone was analyzed by qPCR for expression of TGF-β receptors Alk4, Alk5, and Alk7 relative to Tbp. (C) UMR106.01 cells were transfected with custom siRNA against Alk4, Alk5, or Alk7, or scrambled control. 24 hours later, RNA was isolated for qPCR analysis of Alk4, Alk5, or Alk7 relative to Gapdh. (D) RNA from (C) was analyzed for Sost transcript by qPCR relative to Gapdh. * indicates p < 0.05 compared to Vehicle, ** indicates p < 0.01 compared to Vehicle, *** indicates p < 0.001 compared to Vehicle.

siRNA transfection

siRNA (Qiagen) were designed against murine Sost, Gapdh, Alk4, Alk5, and Alk7; specificity was confirmed with BLAST. Cells were seeded at a density of 6,000 cells per well in 48-well plates. One hour later, 10nM siRNA and Fugene 6 (Roche) were diluted into Opti-MEM (Invitrogen) and were gently added to the culture plate. Samples were collected 48 hours later and were processed for qPCR analysis of target knock-down.

TGF-βrI kinase inhibitor treatment in vivo

Eight-week old male C57BL/6 mice were treated for 24 hours with vehicle (HBSS), or SD-208 (60 mg/kg twice daily, one injection every 12 hours; Tocris Biosciences) by intraperitoneal delivery (IP). No adverse effects of SD-208 on mouse health were detected during the study. At 24 hours after the first injection animals were euthanized humanely. Femoral shafts were scraped of soft tissue and skeletal muscle, flushed with ice-cold HBSS with a 25 gauge needle to remove the bone marrow before placing into RNAlater (Qiagen) and stored at 4°C. For RNA isolation, samples were removed from RNAlater and homogenized in 1 mL Qiazol and purified using Rneasy mini-kit using manufacture’s guidelines (Qiagen). Approvals for work conducted on the mice used in this study were granted by Lawrence Livermore National Laboratory Institutional Animal Care and Use Committee, under application no. 168. Animals were treated humanely; housing and experimental procedures followed the guidelines outlined in the National Institute of Health 'Principles of Laboratory Care'.

Reporter gene assays

pGL3-based reporter plasmids (Promega) containing ECR5 upstream of the human SOST or the SV40 promoter were previously described [13, 16]. A putative SMAD site was predicted within a multiple sequence alignment of human and mouse ECR5 sequences using power weight matrices available from TRANSFAC and utilized by MultiTF (http://multitf.dcode.org/). ECR5 was PCR cloned into the EcoR1 site of pGL3-promoter vector, as well as in a pGL3 vector where the SV40 promoter has been replaced by a 2kb fragment of the human sclerostin promoter. Subsequently the SMAD or MEF2 site was deleted using site-directed mutagenesis according to the manufacturer's instructions (Quickchange Site-directed Mutagenesis kit; Stratagene). UMR106.01 cells were seeded at 20k/well into 48-well plates. On the following day, media was removed, replaced with Opti-MEM, and transfected with Fugene 6, the reporter of interest (250ng/well), and pRL-TK (50ng/well; Promega) as a transfectant control. 24 hour later, media was removed and replaced with TGF-β₁. Samples were analyzed 24 hours later using Dual-Luciferase Reporter System (Promega) and TD-20/20 luminometer (Turner Systems).

Statistical analysis

Each experiment was performed a minimum of 3 times, each time in duplicate or triplicate. Unless otherwise noted, data are presented as mean ± standard error of the mean. Statistical significance was assessed by two-tailed Student’s t test or ANOVA for non-repeated measurements followed by a Dunnet post-hoc analysis compared to control (vehicle). p < 0.05 was considered statistically significant.

Results

TGF-β superfamily members modulate Sost expression in mature osteoblastic cells

A described mode of modulating Wnt signaling involves altering the expression levels of Lrp5/6 antagonists, as both deletion of Dkk1 and Sost have been shown to promote aggressive bone overgrowth characterized by hyperactive osteoblast activity due to elevated Wnt signaling. We evaluated whether the expression of Sost was influenced by members of the TGF-β superfamily. UMR106.01 cells were cultured in the presence of 10ng/mL of TGF-β₁, -β₂, -β₃, Activin A, or Nodal for 6 hours; these growth factors were chosen because they differentially activate signaling through the Alk4, Alk5, or Alk7 TGF-β type I receptors. Alternately, cells were cultured with 100nM hPTH(1–34), which decreases Sost expression [27–29]. Each TGF-β isoform increased Sost expression nearly 3-fold compared to vehicle control (Figure 1A); Activin A was less effective than TGF-β_1–3, and Nodal did not affect Sost expression at the dose examined.

Figure 1. TGF-β superfamily members regulate expression of Sost.

(A) UMR106.01 mature osteoblasts were cultured for 6 hours with vehicle or 10ng/mL of indicated ligand; 100nM hPTH(1–34) was used as a positive control for Sost suppression. RNA was isolated for qPCR analysis of Sost expression relative to Rpl32. (B) UMR 106.01 mature osteoblasts were cultured for 3, 6, or 24h in the presence of 10ng/mL growth factor. For the sake of clarity, only the influence of TGF-β₁ and -β₃ upon Sost are shown. Data are mean ± SEM of 5 independent experiments. * indicates p < 0.05 compared to Vehicle, ** indicates p < 0.01 compared to Vehicle, *** indicates p < 0.001 compared to Vehicle, and # indicates p < 0.05 compared to TGF-β₃.

Dose- and time-dependent effects of TGF-β family members upon Sost expression

To investigate dose-dependence of Sost expression in response to TGF-β₁, -β₂, -β₃, and Activin A, UMR106.01 cells were cultured in the presence of increasing concentrations of growth factor (1–10,000 pg/mL) for 6 hours. TGF-β₁, -β₂, -β₃, effective at Sost induction (257 – 301% increase) at 10ng/mL, whereas Activin A was about half as effective (TABLE 2). There were differences in potency amongst the four proteins: TGF-β₁ and -β₃ demonstrated a similar EC₅₀ for Sost induction (29 and 34pg/mL, respectively), whereas TGF-β₂ (186 pg/mL) and Activin A (1371 pg/mL) were less potent. These data indicate that TGF-β₁, -β₂, -β₃ and Activin A, members of the TGF-β superfamily, are capable of inducing Sost expression.

TABLE 2.

TGF-β isoform effects upon Sost

Ligand	Sost EC50 (ng/mL)	Maximum induction (percent of vehicle)
TGF-β1	29	257
TGF-β2	186	301
TGF-β3	34	297
Activin A	1371	175

The temporal nature of TGF-β-induced Sost expression was examined by treating cells with 10ng/mL of each growth factor, after which RNA was collected 3, 6, or 24h later. TGF-β₁ exerted a rapid increase upon Sost expression after 3 hours of culture (Figure 1B); there was a trend (p<.12) for decreased Sost induction after 6 hours of culture, although expression remained significantly elevated after 6 or 24 hours of culture compared to vehicle control. TGF-β₂ exerted a gradual increase in Sost induction over the time course examined (data not shown). The effect of TGF-β₃ upon Sost expression was transient, as it was significantly elevated after 3 or 6 hours of treatment, but regressed toward baseline after 24 hours of treatment (Figure 1B). The effect of Activin A mimicked that of TGF-β₃ (data not shown).

Sost expression is inhibited by antagonists of Alk4/5/7 in vitro and in vivo

TGF-β and Activin A activate intracellular signaling cascades once bound to type I and type II receptors. TGF-β type I receptors include Alk4, Alk5, and Alk7. The addition of the Alk4/5/7 antagonist SB431542 to UMR 106.01 cells decreased endogenous Sost expression (Figure 2A). We next examined whether inhibition of Alk4/5/7 similarly influenced Sost expression, in vivo. Wild-type mice treated with a related Alk4/5/7 kinase inhibitor, SD-208 (60mg/kg), for 24 hours [32] exhibited a 63% decrease in Sost expression compared to PBS-treated controls (Figure 2B).

Figure 2. Sost expression is inhibited by Alk4/5/7 antagonists.

(A) UMR106.01 cells cultured for 6 hours in the presence of the Alk4/5/7 antagonist SB431542. RNA was isolated for qPCR analysis of Sost expression relative to Rpl32. (B) Mice were injected with either PBS or 20mg/kg SD208. 24 hours later, femorae were extracted and processed by qPCR for Sost expression relative to Rpl32. Data are mean ± SEM of 3 independent experiments. * indicates p < 0.05 compared to Vehicle

Alk4 and Alk5 mediate induction of Sost

Expression of the three TGF-β type I receptors (Alk4, Alk5, and Alk7) was confirmed in vitro in UMR106.01 osteoblasts (Figure 3A) and in vivo from femur and calvariae RNA (Figure 3B). Alk4 (receptor for Activin A) and Alk5 (receptor for TGF-β₁, -β₂, -β₃) were consistently expressed at higher levels compared to Alk7 (receptor for Nodal). siRNA directed against Alk4, Alk5, or Alk7 decreased target gene expression by 72–77% relative to non-silencing RNA controls (Figure 3C). siRNA directed against Alk4 and Alk5 significantly decreased endogenous Sost production by 76 and 81%, respectively, while siRNA directed against Alk7 only modestly influenced Sost levels (Figure 3D). These data indicate that Sost expression is in part regulated by Alk5 and Alk4 receptors, respectively, whereas in these experiments Alk7 does not appear to play a significant role in the regulation of Sost expression.

TGF-β₁ induction of Sost involves Smad3 and is antagonized by PTH

Direct regulation of gene expression by TGF-β involves Smad2 or Smad3 transcription factors. SIS3 is a cell-permeant Smad3 inhibitor that reduces Smad3 phosphorylation and DNA binding, without influencing Smad2 [33]. The addition of 10µM SIS3 to cultures significantly reduced Sost expression in both vehicle and 10ng/mL TGF-β₁-treated cells (Figure 4A), indicating that both endogenous and growth factor-induced Sost expression is dependent on Smad3 phosphorylation and DNA binding.

Figure 4. TGF-β induction of Sost involves Smad3 and is abrogated by PTH.

(A) UMR 106.01 cells were pretreated for 1 hour in the absence or presence of the cell-permeant Smad3 inhibitor, SIS3 (10µM), after which 10ng/mL TGF-β₁ or vehicle control was added for 6 hours. RNA was isolated for qPCR analysis of Sost expression relative to Rpl32. (B) UMR 106.01 cells were cultured for 6 hours in the absence or presence of 100nM hPTH(1–34) with or without 10ng/mL TGF-β₁.

We have previously demonstrated that PTH directly inhibits Sost expression without the need for de novo protein synthesis [28]; thus, we examined the capacity for PTH to interfere with TGF-β₁-induced Sost expression. Cells treated with both 100nM hPTH (1–34) and 10ng/mL TGF-β₁ for 6 hours revealed no significant increase in Sost expression relative to vehicle controls, suggesting that they regulate Sost expression independently (Figure 4B), and blunted each other’s effect on Sost transcription.

TGF-β₁ targets the ECR5 region of the distal Sost enhancer

Regulation of Sost by TGF-β₁ could involve activation of the proximal promoter, distal enhancer, or a synergistic effect of both. We examined the responsiveness of Luciferase reporter constructs containing a minimal SV40 promoter, the human SOST promoter (hSOST), each with or without the ECR5 distal enhancer [13, 16]. Constructs containing either the minimal SV40 promoter or the hSOST promoter demonstrated no change in Luciferase expression when treated with 10ng/mL TGF-β₁, whereas constructs containing the ECR5 region increased the Luciferase activity by approximately 3-fold (Figure 5A), independent of promoter choice. There was no additive effect in plasmids containing ECR5 and hSOST, indicating that these transcriptional regulatory components do not synergize, and that the TGF-β₁ response is dependent on the distal enhancer, ECR5.

Figure 5. TGF-β₁ induces Sost via the ECR5 transcriptional regulatory element through Smad and Mef2.

(A) UMR 106.01 cells were transfected with Luciferase constructs containing minimal SV40 promoter, the human SOST promoter, each with or without the ECR5 distal enhancer and treated with vehicle or 10ng/mL TGF-β₁ for 24 hours, after which Luciferase activity was measured. (B) UMR 106.01 cells were transfected with a wild-type 1xECR5-hSOST Luciferase construct or construct with deletions of the Smad (ΔSmad) or Mef2 (ΔMef2) consensus binding site and treated with vehicle or 10ng/mL TGF-β₁ for 24 hours, after which Luciferase activity was measured. *** indicates < 0.001 compared to plasmids lacking 3xECR5, ** indicates p < 0.01 compared to plasmids with deletions.

The ECR5 region contains a binding site for the myocyte enhancer family 2 (Mef2) family of transcription factors [34]. This sequence also contains a putative binding site for Smad2/3 [35], transcription factors activated by TGF-β₁. To test the role of the Mef2 and Smad binding sites in TGF-β-driven Luciferase activity, reporter constructs were generated that contained mutated binding sites for Mef2 or Smads. Deletion of the Smad binding site prevented Luciferase expression in response to TGF-β₁ (Figure 5B); unexpectedly, deletion of the Mef2 binding site also prevented TGF-β₁-driven Luciferase activity.

Discussion

Genetic analyses of the rare sclerosing bone dysplasias sclerosteosis and van Buchem disease that either target the SOST transcript, or its distal cis-regulatory elements have positioned SOST as a robust regulator of skeletal homeostasis. Its role on mediating bone accrual and deposition was further confirmed through the generation of Sost [36] or van Buchem region [13] transgenic mice, which are osteopenic compared to wild-type control mice. Sclerostin was initially characterized as a BMP antagonist [36], but was later shown to function through Lrp4-6 [8, 37]. Functionally, a sclerostin-inhibiting antibody enhances bone mineral density and strength [38, 39]. Mechanistically, Sclerostin exerts anti-anabolic effects by decreasing osteoprogenitor proliferation and differentiation, and promoting osteoblast apoptosis [40]. Provided the increasing direct medical costs for osteoporosis-related or other traumatic fractures, it is critical to understand the mechanisms involved in Sost expression because of its function in vivo as an inhibitor of bone formation.

The TGF-β family represents an important superfamily involved in regulating bone metabolism since a significant number of its members are produced locally within the skeleton [41] where they accumulate within the matrix [42] and are released by osteoclasts during remodeling [43]. Germline deletion of individual TGF-β genes yields multiple, non-overlapping developmental defects [44]. Of the three TGF-β proteins, TGF-β₁ is the isoform most abundantly expressed in bone [45]. It has opposing effects on osteoprogenitors compared to osteoblasts, inducing the migration, proliferation and differentiation of the former, but inhibiting matrix accumulation of the latter [20]. Little is known as to how TGF-β elicits these differential effects, although Alliston et al. have demonstrated that TGF-β promotes the degradation of the transcription factor essential for osteoblast differentiation, Runx2 [46]. Later, Balooch et al. used a series of transgenic mouse models with alterations in TGF-β signaling to show elevated levels of TGF-β in bone decreased mechanical properties, mineral concentration, and fracture toughness [47]; these trends were reversed in mice expressing a dominant-negative type II TGF-β receptor. Further, Mohammed et al. showed that inhibition of Alk4/5/7 with SD-208 in vivo had catabolic and anabolic effects on the mouse skeleton [48]. These effects included: increased increased bone mineral density and trabecular bone volume independent of cortical bone changes; increased osteoprogenitor differentiation; and decreased osteoclast number. Functionally, Alk4/5/7 inhibition inceased both bone mechanical and material properties, suggesting that Alk4/5/7 inhibition could be used for conditions of skeletal fragility like osteoporosis. These findings combined with the data presented in this manuscript provide compelling evidence that signaling the the Alk4/5/7 axis positively influences Sost expression. We can further speculate that Sclerostin potentially plays an important role in the processes described by Balooch et al. and Mohammed et al. [48].

TGF-β interacts with a variety of signaling pathways implicated in skeletal homeostasis. For example, PTH increases TGF-β₁ synthesis and secretion and Smad3 phosphorylation to decrease osteoblast apoptosis [22]. TGF-β generally decreases the capacity for BMPs to induce osteogenic differentiation of osteoprogenitors, but de Gorter et al. recently demonstrated that this inhibitory effect is very much context dependent with respect to such details as the composition of the culture medium used and the duration of growth factor co-culture [49]. Similarly, TGF-β can regulate expression and activity of Wnt signaling. In mesenchymal stem cells, TGF-β₁ induces proliferation via Smad3-dependent β-catenin nuclear translocation [50] and increases β-catenin expression [51]. Within, we demonstrate that TGF-β isoforms increase Sost expression in mature osteoblasts. Provided the inhibitory effect of Sclerostin upon matrix mineralization and differentiation [52], this may provide a mechanism to explain the similarly inhibitory effect of TGF-β₁ upon differentiation of mature osteoblasts, although further studies are required to fully confirm this.

Nonsense mutations in Sost cause sclerosteosis [9], and a 52kb deletion 35kb downstream of Sost is responsible for van Buchem disease [11]. Using cross-species sequence comparison and enhancer assays, we have previously identified a 255-bp evolutionarily conserved sequence within the van Buchem deletion, termed ECR5, that confers bone-specific expression of SOST [13]. Reporter assays containing the ECR5 enhancer, but not the proximal promoter, are responsive to PTH via Mef2 transcription factors [16]. Within this work, we demonstrate that, similar to PTH, TGF-β₁ targets the ECR5 enhancer but not the SOST promoter (Figure 5). Within the ECR5 enhancer, we identified consensus binding sites for Smad2/3, in addition to the previously identified Mef2 sites; surprisingly, both Smad2/3 and Mef2 sites contribute to TGF-β₁ transcriptional activation of Sost. We have not observed any changes in expression of Mef2c or Mef2d transcription in response to TGF-βs (data not shown), indicating that the regulation is likely post-transcriptional in nature. One candidate mechanism is direct physical interaction between Mef2 and Smad2. Indeed, Quinn et al. previously demonstrated direct interaction of Smad2 with Mef2 in vivo to enhance Mef2 transcriptional activity [53], as did Ishikawa et al. [54]. How TGF-β₁ regulates Mef2 transcriptional activity in osteoblastic cells is currently under investigation.

Similar to TGF-βs, we observed that Activin A, which signals through the Alk4 type I receptor, increases Sost expression, although with less potency and efficacy than TGF-β₁,-β₂, -β₃. Despite the decreased potency, siRNA directed against Alk4 decreased endogenous Sost expression to a similar degree as did siRNA against Alk5. Similar to TGF-β, the role of Activin A in skeletal homeostasis is conflicting, as various reports demonstrate its capacity to increase or decrease osteoblast- and osteoclastogenesis [55, 56]. Pathologically, Activin A and TGF-β are implicated in the progression of osteolytic metastases. Serum Activin A is elevated in patients with multiple myeloma [57], and a soluble Activin receptor IIA fusion protein decreases bone metastasis and resorption [58]. Interestingly, Dkk1, a Wnt antagonist like Sclerostin, is highly expressed in osteolytic cancers such as multiple myeloma [59]; whether Wnt signaling in osteolytic cancers is dependent upon TGF-β is a possibility worthy of investigation.

In conclusion, we demonstrate that members of the TGF-β superfamily beyond BMPs are also capable of inducing Sost expression, as TGF-βs and Activin A demonstrated graded effects on Sost transcription. This is not a general effect of the TGF-β superfamily, as Nodal (Alk7 agonist) had no effect on Sost expression at the concentration examined. Inhibition of Alk4/5/7 with two different antagonists decreased Sost expression in vitro and in vivo, indicating that use of a clonal mature osteoblastic cell line was not a confounding factor in regulation of Sost by TGF-βs. siRNA directed against Alk4 and Alk5 decreased Sost expression. The specific Smad3 inhitor SIS3 attenuated endogenous and TGF-β-stimulated Sost expression, suggesting that traditional TGF-β/Smad, rather than TGF-β/MAPK, signaling is involved in transcription of Sost. Luciferase reporter assays indicated that TGF-β₁ targets the ECR5 bone enhancer, and not the proximal promoter, through a mechanism involving both Smad2/3 and Mef2 transcription factors. These results lay the foundation for future work designed to examine the interplay between Alk4 and Alk5 function and Wnt signaling in vivo, and identification of protein intermediates required for TGF-β—dependent induction of Sost transcription.

Highlights.

• TGF-β isoforms and Activin A increase Sost expression via Smad3

• Inhibition of Alk4 and Alk5 both in vitro and in vivo reduces Sost expression

• TGF-β targets ECR5 to enhance Sost

Abbreviations

TGF-β

Transforming Growth Factor-beta

Lrp5

Low-density lipoprotein-related receptor 5

Wnt

Wingless int

ECR5

Evolutionarily Conserved Region 5

PTH

Parathyroid Hormone