Elevated hepatocyte growth factor levels in osteoarthritis osteoblasts contribute to their altered response to bone morphogenetic protein-2 and reduced mineralization capacity

(1)

Original Full Length Article

Elevated hepatocyte growth factor levels in osteoarthritis osteoblasts contribute to their altered response to bone morphogenetic protein-2 and reduced mineralization capacity

E. Abed

^a

, B. Bouvard

^b,c

, X. Martineau

^a

, J.-Y. Jouzeau

^b,d

, P. Reboul

^b

, D. Lajeunesse

^a,

⁎

aUnité de recherche en Arthrose, Centre de Recherche du Centre Hospitalier de l'Université de Montréal (CRCHUM), Montréal, Québec, Canada

bUMR7365 IMoPA, Université de Lorraine/CNRS, Vandœuvre lès Nancy, 54505, France

cService de Rhumatologie, Centre Hospitalier Universitaire (CHU), Angers 49933, France

dService de Pharmacologie Clinique et de Toxicologie, Centre Hospitalier Universitaire (CHU), Nancy 54023, France

a b s t r a c t a r t i c l e i n f o

Article history:

Received 24 October 2014 Revised 24 January 2015 Accepted 1 February 2015 Available online 7 February 2015 Edited by J. Aubin

Keywords:

Osteoarthritis

Hepatocyte growth factor Bone morphogenetic protein-2 Osteoblasts

Wnt signaling

Purpose:Clinical andin vitrostudies suggest that subchondral bone sclerosis due to abnormal osteoblasts is involved in the progression of osteoarthritis (OA). Human osteoblasts isolated from sclerotic subchondral OA bone tissue show an altered phenotype, a decreased canonical Wnt/ß-catenin pathway, and a reduced mineral- izationin vitroas well asin vivo.These alterations were linked with an abnormal response to BMP-2. OA osteoblasts release factors such as the hepatocyte growth factor (HGF) that contribute to cartilage loss whereas chondrocytes do not express HGF. HGF can stimulate BMP-2 expression in human osteoblasts, however, the role of HGF and its effect in OA osteoblasts remains unknown. Here we investigated whether elevated endogenous HGF levels in OA osteoblasts are responsible for their altered response to BMP-2.

Methods:We prepared primary human subchondral osteoblasts using the sclerotic medial portion of the tibial plateaus of OA patients undergoing total knee arthroplasty, or from tibial plateaus of normal individuals obtained at autopsy. The expression of HGF was evaluated by qRT-PCR and the protein production by western blot analysis.

HGF expression was reduced with siRNA technique whereas its activity was inhibited using the selective inhibitor PHA665752. Alkaline phosphatase activity (ALPase) and osteocalcin release were measured by substrate hydrolysis and EIA respectively. Canonical Wnt/β-catenin signaling (cWnt) was evaluated both by target gene expression using the TOPﬂash TCF/lef luciferase reporter assay and western blot analysis ofβ-catenin levels in response to Wnt3a stimulation. Mineralization in response to BMP-2 was evaluated by alizarin red staining.

Results:The expression of HGF was increased in OA osteoblasts compared to normal osteoblasts and was maintained during theirin vitrodifferentiation. OA osteoblasts released more HGF than normal osteoblasts as assessed by western blot analysis. HGF stimulated the expression of TGF-β1. BMP-2 dose-dependently (1 to 100 ng/ml) stimulated both ALPase and osteocalcin in normal osteoblasts whereas, it inhibited them in OA osteoblasts.

HGF–siRNA treatments reversed this response in OA osteoblasts and restored the BMP-2 response. cWnt is reduced in OA osteoblasts compared to normal, and HGF–siRNA treatments increased cWnt in OA osteoblasts al- most to normal. Smad1/5/8 phosphorylation in response to BMP-2, which is reduced in OA osteoblasts, was corrected when these cells were treated with PHA665752. The BMP-2-dependent mineralization of OA osteoblasts, which is also reduced compared to normal, was only partially restored by PHA665752 treatment whereas 28 days treatment with HGF reduced the mineralization of normal osteoblasts.

Conclusion:OA osteoblasts expressed more HGF than normal osteoblasts. Increased endogenous HGF production in OA osteoblasts stimulated the expression of TGF-β1 and reduced their response to BMP-2. Inhibiting HGF expression or HGF signaling restored the response to BMP-2 and Smad1/5/8 signaling. In addition, decreased HGF signaling partly corrects the abnormal mineralization of OA osteoblasts while increased HGF prevents the normal mineralization of normal osteoblasts. In summary, we hypothesize that sustained elevated HGF levels in OA osteoblasts drive their abnormal phenotype and is implicated in OA pathophysiology.

Introduction

Osteoblasts from human osteoarthritis (OA) patients express an altered phenotype in bone tissuein situ[1–3], a situation which can still be observedin vitro. Indeed, OA osteoblasts have an abnormally

⁎ Corresponding author at: Unité de recherche en Arthrose, Centre de recherche du Centre Hospitalier de l'Université de Montréal (CRCHUM), 900, rue Saint-Denis, Montréal, Québec H2X 0A9, Canada. Fax: +1 514 412 7583.

E-mail address:daniel.lajeunesse@umontreal.ca(D. Lajeunesse).

Contents lists available atScienceDirect

Bone

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / b o n e

(2)

high alkaline phosphatase activity (ALPase), show an increased osteocalcin secretion, yet express normal levels of speciﬁc cell surface makers CD73 and CD105[4–7]. Bone tissue from OA patients has been described as undermineralized by a number of research groups [8–11]. More importantly, although presenting high expression of markers of osteoblast/osteocyte-like cells, OA osteoblasts fail to miner- alize normally in response to BMP-2 stimulation[7]. We determined this was due to elevated levels of transforming growth factorβ-1 (TGF-β1) in human OA osteoblasts[7], and we and others demonstrated previously that this was linked with an abnormal regulation of the ratio of type 1 collagenα1 toα2 chains bothin vivo[8,9]andin vitro[7].

Abnormal production of growth factors has been described in OA osteoblasts, such as for the insulin-like growth factor-1 (IGF-1)[12], TGF- β1[6,13], and the hepatocyte growth factor (HGF)[14]. Both TGF-β1 and HGF may be important for the initiation and progression of OA. Re- cent studies have indicated that overproduction of TGF-β1 speciﬁcally in mouse osteoblasts leads to OA-like features whereas inhibition of TGF-β1 in subchondral bone mesenchymal stem cells signaling reduced OA progression[15]. HGF, which contributes to the increased expression of metalloproteinase-13 (MMP-13) in human OA cartilage samples [16], is not produced by articular cartilage chondrocytes but likely orig- inates from subchondral osteoblasts[14].

Interactions between HGF, TGF-β1 and BMP-2 have been observed in different cell systems. Indeed, TGF-β1 can stimulate HGF production in some cell types whereas the reverse has also been reported, suggesting that the interaction between TGF-β1 and HGF could be cell-speciﬁc [17–20]. TGF-β1 can inhibit the effect of BMP-2, which has a beneﬁcial role on bone tissue homeostasis,viathe SMAD signaling pathway[21], while HGF has also been described to play a similar role in the murine myeloid cell line C2C12[22].

In general, TGF-β1 activates the ALK5/Smad2/3 pathway [23]

whereas BMP-2 actsviathe ALK1/Smad1/5/8 pathway[24]. The delete- rious effect of TGF-β1 on the action of BMP-2 in bone tissue involves the ALK5 receptor pathway which triggers SMAD2/3 and inhibits the ALK1 receptor pathway activation[25]. Of note, recent studies have indicated that altered SMAD2/3 activity could be detrimental to the cartilage[26], and this triggers MMP-13 activity which is a key player for the loss of articular cartilage. A shift in ALK5 and ALK1 activity can also be observed during aging and in OA pathogenesis[27]. However, a role for HGF on the phenotype of osteoblasts in osteoarthritis has never been described, nor if there is a link between HGF and altered ALK1/Smad1/5/8 signaling in OA osteoblasts.

TGF-β1 and HGF also play important immunoregulatory role on OA mesenchymal stem cells (MSC)[28–30]. In addition, it is of note that cytokines and growth factors play key role in osteophyte formation[31], and that elevated TGF-β1[6,13]and HGF[14]levels are observed in OA osteoblasts[32]. Osteophyte formation may be considered a repair response to stabilize the damaged joints, and it requires the local recruitment of speciﬁc MSC[33].

Hence, the present study was performed to unravel the potential role of HGF to alter the phenotype of OA osteoblasts,viatheir response to BMP-2 on ALPase activity and osteocalcin secretion, alterations of the cWnt and Smad1/5/8 pathways, and its role in bone mineralization in these cells.

Material and methods Patients and clinical parameters

Tibial plateaus were obtained from OA patients undergoing knee replacement surgery and prepared as previously described[4–6]. A total of 37 patients (69 ± 9.4 years old, mean ± SD; 11 males/26 females) classiﬁed as having OA according to the criteria of the American College of Rheumatology were used[34]. No patients received medication that could interfere with bone metabolism. Specimens from 12 normal individuals (68.1 ± 15.7 years old, mean ± SD; 6 males/6

females) were obtained at autopsy within 12 h of death. Ethical approv- al was obtained for the use of all human material following a signed agreement by the patients undergoing knee surgery and for the autopsy specimens by their relatives, in accordance with the CHUM ethical com- mittee guidelines.

Preparation of primary subchondral bone osteoblasts

Isolation of subchondral bone plate and the cell cultures was prepared as previously described[4]. Conﬂuent normal and OA osteoblasts were maintained in culture media containing 10% FBS to promote their differentiation until they were switched to media containing 2% FBS for the last 48 h of culture. For the determination of bio- markers conﬂuent cells were incubated in Ham's F12/DMEM culture media with 2% fetal bovine serum (FBS) for their last 48 h of culture in the presence of 1,25(OH)2D3(50 nM). In some cases, cells were stimulated with increasing doses of human recombinant BMP-2 (rhBMP-2), 1 to 100 ng/ml, or the vehicle for their last 48 h of culture in the presence of 1,25(OH)2D3. Supernatants were collected at the end of the incubation for the determination of osteocalcin levels.

Cells were prepared in either ALPase buffer for the phenotypic determination of ALPase activity, RIPA buffer for western blot analysis, or in TRIzol^TMfor qRT-PCR experiments. Protein determination was performed by the bicinchoninic acid method[35]. In some experiments, cells were maintained in culture 20 days post-conﬂuence to determine the expression of HGF at speciﬁed time points.

Phenotypic characterization of human subchondral osteoblast cell cultures

ALPase activity was determined by substrate hydrolysis using p-nitrophenylphosphate of whole cell lysates whereas, osteocalcin release in cell supernatants was evaluated using an enzyme immunoassay (EIA) as previously described[4,6]. Determinations were performed in duplicate for each specimen.

Preparation of Wnt3a conditioned media (Wnt3a-CM)

Conditioned media (CM) were prepared from Murine L cell lines transfected with either an empty vector (Parental) or with Wnt3a (Wnt3a) obtained from the American Type Culture Collection (Cedarlane Laboratories, Ontario) as described[13]. CM were added to cells at a 20%ﬁnal concentration.

Evaluation of mineralization

Conﬂuent cells were incubated in BGJb media containing 10% fetal bovine serum (FBS), 50 μg/ml ascorbic acid, and 50 μg/ml β- glycerophosphate. This medium was changed every two days until day 28. Mineralization of cell cultures was measured by quantiﬁcation of alizarin red staining (ARS) with the procedure of Gregory et al.[36].

Inhibition of HGF expression in OA osteoblasts by siRNA

HGF expression was inhibited in OA osteoblasts by specific siRNA using previously described methods[7,13,37,38]. Briefly, OA osteoblasts were split at 100,000 cells/ml. HGF–siRNA preparations (4 different siRNA constructs are provided by the manufacturer within the same sample) were obtained at Dharmacon (Lafayette, CO) as well as scrambled RNA, and they were added to OA osteoblasts at afinal concentration of 100 ng/ml with 4.5μl Hi-perfect (Quiagen, Ontario) per 100μl total volume in BGJb media without serum for 1 h on day 0 and day 3.

Controls were performed using scrambled RNA (siSCR) preparations provided by the same manufacturer, and cells were treated as per siHGF. The inhibition of HGF expression was followed using qRT-PCR.

(3)

Inhibition of HGF signaling in OA Ob

In some experiments, HGF signaling was blunted using a selective HGF receptor blocker, PHA665752. In brief, OA osteoblasts were grown to conﬂuence under normal culture conditions as described above, and PHA665752 was added at aﬁnal concentration of 0.5μM for an hour prior to assay. Cells were then used either for western blot analysis of Smad1/5/8 signaling or mineralization assays.

Protein determination by western blotting

Cell extracts were prepared for western blotting as previously described[39], and immunoblotting was performed as described in the ECL Plus western blotting detection system's manual (Amersham pharmacia biotech, UK, England). Rabbit anti-human HGF (1:1000, R&D Systems, MN), rabbit anti-humanβ-catenin (1:2000, Cell Signaling Technology, MA), rabbit anti-smad1/5/8 and rabbit anti-p-smad1/5/8 (1:1000, Cell Signaling Technology), and rabbit anti-human actin (1:10,000, Sigma-Aldrich, ON) were used as primary antibodies, whereas a HRP-conjugated goat anti-rabbit IgG (1:10,000, Upstate Biotechnol- ogy, NY) was used as the secondary antibody. Densitometry analysis of western blotﬁlms was performed using the public domain NIH Image program developed with the Scion Image 1.63 program[40].

qRT-PCR assays

RT reactions were primed with random hexamers with 1μg of RNA followed by q-PCR amplification with the Rotor-Gene 6® RG-3000A (Corbett Research, Australia) as described[5,39,41]using 20 pmol of specific primer pairs: HGF, F: ATGCATCCAAGGTCAAGGAG, R: TTCCAT GTTCTTGTCCCACA; TGF-β1, F: GCGTGCTAATGGTGGAAAC, R: GCTGAG GTATCGCCAGGAA; GAPDH, F: CAGAACATCATCCCTGCCTCT, R: GCTTGA CAAAGTGGTCGTTGAG, added at afinal concentration of 200 nM. The data were processed with the GeneAmp 5700 SDS software and given as threshold cycle (Ct). Ct values were converted to number of target gene molecules and values expressed as the ratio to GAPDH.

TOPﬂash dual luciferase reporter assays

OA osteoblasts were plated in 24-well plates at a density of 1.5 × 10⁵cells/well containing 10% FBS in BGJb media and left overnight after their siHGF or siSCR treatment as described above. Plasmid mixtures containing 2μg TOPﬂash luciferase construct (Upstate Biotechnology, NY) and 0.05 μg Renilla luciferase driven by the SV40 promoter (Promega, WI) were transfected into cells overnight using the FuGENE 6 transfection Reagent (Roche, ON) according to the manufacturer's protocol. Media were changed and cells were left to recover from transfection for 6 h prior to incubation for 24 h with Wnt3a-CM or Parental- CM. At the end of the incubation, cells were lyzed before the evaluation of the Wnt/β-catenin signaling pathway (cWnt) by measuring the luciferase activity using the dual luciferase assay kit (Promega, WI). Values for TOPﬂash luciferase activity were normalized with Renilla activity.

Statistical analysis

Quantitative data are expressed as mean ± SEM. The data were analyzed by the Student'sttest or an ANOVA, as appropriate. P valuesb0.05 were considered statistically signiﬁcant between subgroups.

Results

Phenotype alteration of OA osteoblasts

Previous reports have shown that the phenotype of OA osteoblasts is altered compared to normal Ob[4,6,13]. Similar observations were made herein for ALPase activity (Fig. 1A), and osteocalcin secretion

(Fig. 1B) which were increased in OA osteoblasts compared to normal osteoblasts. Furthermore, when testing the response of normal and OA osteoblasts to BMP-2, we observed an altered behavior of OA osteoblasts. Indeed, as shown inFig. 1C, normal Ob display a normal dose– response increase in ALPase activity in response to BMP-2 stimulation (pb0.04 by ANOVA) whereas, OA osteoblasts have a decreased ALPase activity in response to BMP-2 challenge (p b 0.001 by ANOVA). Likewise, BMP-2 dose-dependently stimulated osteocalcin secretion in normal osteoblasts (pb0.01 by ANOVA) yet, induced a decrease of osteocalcin secretion in OA osteoblasts (Fig. 1D, pb0.001 by ANOVA).

Production of HGF by OA Ob

As we previously reported that OA osteoblasts produced more HGF than normal osteoblasts and since sustained elevated levels of HGF have been shown to alter osteoblast markers in osteoblast-like cell lines[22,42,43], we set out our goals to:ﬁrst, study the regulation of HGF expression in OA osteoblasts and second, determine if abnormal HGF production could play a role in the reduced response to BMP-2 in these cells. As shown inFig. 2A, HGF expression was increased ~3-fold in OA osteoblasts compared to normal osteoblasts, and was maintained during the differentiation of osteoblasts up to 20 days post-conﬂuence.

This increase in HGF expression was also reﬂected by an increased production of HGF by OA osteoblasts as visualized using western blot analysis compared to normal osteoblasts (Fig. 2B). As OA osteoblasts also show elevated levels of other growth factors such as IGF-1[4]and TGF-β1[6], and since HGF has been reported to regulate TGF-β1 expression, we next tested the hypothesis that HGF could regulate TGF-β1 in our OA osteoblasts cell system. First, we measured the expression of TGF-β1 and showed that OA osteoblasts expressed more TGF-β1 than normal osteoblasts (Fig. 2C). Moreover, HGF directly stimulated the expression of TGF-β1 in OA osteoblasts (Fig. 2D), hence suggesting that elevated TGF-β1 levels observed here, and as previously reported [7,13,37,38], is linked, in part, with elevated HGF levels in these cells.

Contribution of high endogenous HGF to the altered BMP-2-induced phenotype markers in OA osteoblasts

HGF and TGF-β1 can both regulate osteoblast cell markers and cell proliferation[21–24]. We previously showed that OA osteoblasts prolif- erate faster than normal osteoblasts[41]. Here, we tested whether the presence of HGF could alter the phenotype of OA osteoblasts and if it could also alter the response of these cells to BMP-2 as previously reported in osteoblast-like cell models[25]. We used RNA silencing to reduce HGF expression in OA osteoblasts. As shown inFig. 3A, scrambled RNA (siSCR) did not alter signiﬁcantly the HGF levels compared to control cells without siRNA, while siHGF reduced HGF expression about 65%. Using this approach, we demonstrated that siHGF treatments sig- niﬁcantly reversed the inhibitory effect of BMP-2 on ALPase activity in OA osteoblasts (Fig. 3B). Likewise, siHGF treatments also reversed the inhibitory effect of BMP-2 on osteocalcin secretion in OA osteoblasts, whatever the BMP-2 concentration used (Fig. 3C).

Consequence of HGF modulation on the BMP-2-induced mineralization process

As we observed that the response to BMP-2 was altered for ALPase activity and osteocalcin secretion, but could be corrected by inhibiting HGF expression, we next evaluated if: i) HGF could curb the mineralization of normal osteoblasts and, ii) if inhibiting HGF expression and/

or inhibiting HGF signaling in OA osteoblasts could correct the poor mineralization observed in these cells[7]. Indeed, as OA osteoblasts show a reduced mineralization in response to BMP-2[7]and a reduction in bone nodule formation[37], situations that have been demonstrated in the presence of HGF treatment in osteoblast-like cells[43],

(4)

A B

C D

Fig. 1.Alkaline phosphatase activity and osteocalcin secretion in normal and OA osteoblasts. Conﬂuent normal and OA osteoblasts were treated with increasing doses of BMP-2 for 48 h in the presence of 1,25(OH)2D3. Cell culture medium was collected for the determination of osteocalcin secretion and cell lysates were used for the determination of alkaline phosphatase activity (ALPase). A) ALPase in normal (n = 12) and OA (n = 37) osteoblasts; B) osteocalcin release by normal (n = 12) and OA (n = 37) osteoblasts; C–D) conﬂuent normal (n = 4) and OA osteoblasts (n = 12) were stimulated with increasing doses of BMP-2 for 48 h in the presence of 1,25(OH)2D3. C) ALPase activity; D) osteocalcin release.

A B

C D

Fig. 2.Expression of HGF and TGF-β1 in normal and OA osteoblasts and effect of HGF on TGF-β1 expression. Confluent normal and OA osteoblasts were maintained in culture media containing 10% FBS to promote their differentiation until they were switched to media containing 2% FBS for the last 48 h of culture. Confluent cells were lyzed in either RIPA buffer for western blot analysis, or in TRIzol^TMfor qRT-PCR analysis at the specified time point post-confluence. A) Time-dependent expression of HGF mRNA levels in post-confluent normal (n = 7) and OA osteoblasts (n = 13); B) representative western blot analysis of HGF release by confluent normal (n = 4) and OA (n = 8) osteoblasts; C) TGF-β1 mRNA level in normal (n = 7) and OA osteoblasts (n = 16); D) TGF-β1 mRNA level in OA osteoblasts in the absence or in the presence of 10 ng/ml rhHGF (n = 5).

(5)

we next tested if reducing HGF expression or signaling in OA osteoblasts would correct the low mineralization of these cells. First, Fig. 4A shows the reduction in alizarin red staining in OA osteoblasts compared to normal osteoblasts. Second, we show that BMP-2, one of the most potent stimulator of mineralization, induced a progres- sive mineralization of normal osteoblasts as a function of time over 28 days, as measured by alizarin red staining. However, exposure of normal osteoblasts to HGF (10 ng/ml) during the 28 day mineralization period blunted the mineralization of these cells (Fig. 4B). The continuous inhibition of HGF expression by RNA silencing in OA osteoblasts led to abnormal cell growth and behavior, including mineralization (not illustrated). Hence, we used an alternative approach by selectively blocking HGF receptors using PHA665752 at afinal concentration of 0.5μM.Fig. 4C shows that chronic exposure of OA osteoblasts to PHA665752 increased the mineralization of these cells after 28 days of incubation. In addition, this increased mineral- izing capacity of OA osteoblasts was linked with a concomitant increase in the number and the size of bone nodules observed at high magnification (Fig. 4D). Quantification of alizarin red staining of OA osteoblasts showed a slight increase of about 20% upon PHA665752 treatments (Fig. 4E).

Contribution of high endogenous HGF to the altered cell signaling in OA osteoblasts

Since high endogenous HGF expression and production in OA osteoblasts altered their capacity to respond to BMP-2, we next evaluated whether this could be supported by an alteration of the BMP-2 and cWnt signaling pathways, both involved in the regulation of osteoblasts phenotype markers and mineralization. First, as shown inFig. 5A,

exposure of OA osteoblasts to the HGF receptor antagonist PHA665752 increased the phosphorylation of Smad1/5/8 in response to 10 ng/ml BMP-2 by about 50%. Second, since we previously showed that the cWnt signaling pathway is reduced in OA osteoblasts compared to normal and that TGF-β1 regulated Wnt/β-catenin signaling in these cells[13], and as we show herein that the expression of TGF-β1 is stimulated by HGF, we next tested if HGF could also affect cWnt signaling.

Indeed, reducing endogenous HGF expression by siRNA stimulated the Wnt3a-dependent canonical Wnt signaling pathway (Fig. 5B) whereas, acute HGF addition could not further alter the response of these cells.

The increase in TOPﬂash activity in response to siHGF was paralleled by the increase of freeβ-catenin levels as measured by western blot analysis (Fig. 5C). Indeed, siHGF treatment increasedβ-catenin levels about 2.2-fold in response to Wnt3a (pb0.025 vs siScr) whereas, this increase was only about 1.6-fold in cells transfected with siSCR (Fig. 5D).

Discussion

The exact cause of OA remains elusive. However, key observations have led to the idea that abnormal tissue remodeling and disruption of tissue homeostasis are involved in structural alterations observed in all tissues of the OA joint. Inasmuch as reports have shown abnormal bone and cartilage tissue remodeling takes place in joints of OA patients and that this is linked with abnormal osteoblasts cell behavior, this raises the hypothesis that OA is a metabolic disease in which systemic and/or local factors are involved[44–46]. Indeed, because OA osteoblasts show increased osteocalcin secretion and ALPase activityin vitro [4,6,8,47]that reﬂectin situobservations[1–3], this suggests that changes in abnormal cellular metabolism, not systemic changesper se, are responsible for these bone tissue alterations. Thus, abnormal local

A

B C

Fig. 3.Regulation of alkaline phosphatase activity and osteocalcin release by inhibition of HGF expression in OA osteoblasts. Conﬂuent normal and OA osteoblasts were treated with either 10 and 100 ng/ml BMP-2 in the presence of 1,25(OH)2D3. In addition, OA osteoblasts were exposed either to siSCR or siHGF. Conditioned media were collected for the determination of OC secretion and cells were lyzed either with ALPase buffer for the determination of alkaline phosphatase activity or in TRIzol^TMfor qRT-PCR analysis. A) HGF mRNA level in OA osteoblasts under control conditions, siSCR or siHGF treatment (n = 4); B) regulation of alkaline phosphatase activity in normal (n = 4) and OA osteoblasts (n = 10); C) regulation of osteocalcin secretion in normal (n = 4) and OA osteoblasts (n = 10).

(6)

autocrine/paracrine regulation of osteoblasts and overlying articular chondrocytes could lead to these cellular changes observed in OA tissues.

Likely candidates that could alter tissue remodeling and serve as autocrine/paracrine messengers are growth factors and adipokines which are released in abundance by osteoblasts. Indeed, IGFs, TGF-β1, HGF and leptin could all be likely contributors to the altered local cell metabolism and tissue remodeling observed in OA. We previously demonstrated the key roles played by IGF-1, TGF-β1 and leptin in OA osteoblasts[4,6,7,41], where they altered a number of intracellular cell signaling pathways, namely Erk1/2, P38, JNK and cWnt. In addition, our group previously reported that elevated HGF levels in OA articular cartilage is linked with alterations in MMP-13 activity[16]. Moreover, Dankbar et al.[48]sug- gested that HGF could trigger osteophyte formation in OA patients by promoting monocyte chemoattractant protein-1 expression that medi- ates the entry of monocytes/macrophages into the OA affected joints.

Another potential role of HGF could be linked with its capacity to enhance BMP receptors and promote fracture healing[42]. Indeed, as previous reports have shown that OA bone tissue presents indices of multiple callus indicating attempts to repair microfractures which never seem to fully heal[49,50], this could suggest that alterations in local HGF levels could promote continuous tissue modeling in OA joint tissues. However, although being present in OA articular cartilage in

more abundance than in normal cartilage, HGF was not expressed by human chondrocytes yet it was abundantly produced by OA osteoblasts [14]. Hence, HGF could also be a potential candidate that directly alters OA osteoblasts metabolismviaan autocrine/paracrine pathway.

Here, we demonstrate that HGF is expressed and released by OA osteoblasts more than by normal Ob, that this capacity is maintained during cell differentiation in culture, and that the release of HGF drives the expression of TGF-β1 which is also elevated in OA osteoblasts[6,13,37, 38,51]. As previous reports using osteoblast-like cells demonstrated a key role for HGF to alter the expression of markers of osteoblasts, and since OA osteoblasts show abnormal ALPase activity, release of osteocalcin, and mineralization, all potentially regulated by BMPs, we questioned if the abnormally high HGF levels in OA osteoblasts would contribute to alter the response to BMP-2 in these cells. We demonstrate that BMP-2 stimulates ALPase activity and osteocalcin secretion in normal osteoblasts whereas it inhibited ALPase activity and osteocalcin secretion in OA osteoblasts. These alterations were both reversed in the presence of HGF-siRNA, demonstrating that reducing HGF expression and levels in OA osteoblasts normalized the response to BMP-2 in these cells. Previous reports from our laboratory[7,13,38]

and other investigators[8,9,52]have shown a reduction in mineralization in OA osteoblasts and OA bone tissue respectively. HGF can stimulate mineralization under acute conditions, however, HGF exposure for

A B

C

E

D

Fig. 4.Regulation of mineralization by HGF in normal and OA osteoblasts. Normal and OA osteoblasts were stimulated with 10 ng/ml BMP-2 for up to 28 days. Normal osteoblasts were also exposed to either the vehicle or 10 ng/ml rh HGF for 28 days. OA osteoblasts were incubated additionally in the presence or absence of the HGF receptor antagonist PHA665752 (0.5μM) for 28 days. A) Quantification of alizarin red staining in normal (n = 7) and OA osteoblasts (n = 20); B) representative alizarin red staining of normal osteoblasts as a function of time and chronic effect of HGF exposure; C) representative effect of PHA665752 on the mineralization of OA osteoblasts; D) representative microscopic views at higher magnification of OA osteoblasts treated with or without PHA665752 as in C). E) Quantification of alizarin red staining of OA osteoblasts (n = 4) treated with or without PHA665752.

(7)

28 days curbs the mineralization of bone tissue, and bone mineral density is reduced in cancer patients in whom elevated HGF levels have been detected[22]. Indeed, prolonged exposure to HGF of normal osteoblasts reduced their mineralization capacity, whereas, inhibiting HGF signaling in long-term culture of OA osteoblasts underin vitromineral- izing conditions increased their global alizarin red staining and the size of mineralized foci. This suggests that HGF contributes to alter expression of the osteoblast phenotype markers in OA and their regulation, and contributes to reduce the mineralization of OA bone tissue.

Chronic elevated HGF levels acted upon different signaling pathways to convey its inhibitory action of osteoblasts markers and mineralization. The BMP-2 signaling pathway is reduced in OA osteoblasts as a consequence of their elevated endogenous HGF levels, as reﬂected by the correction of Smad1/5/8 phosphorylation in response to the inhibition of HGF signaling. Indeed, an antagonist of HGF receptors prompted an increase of about 50% in Smad1/5/8 phosphorylation in OA osteoblasts. A similar alteration of Smad1/5/8 phosphorylation in response to BMP-6 has been observed in hepatocytes[53], whereas the same HGF blockade also induced elevations of Erk1/2 signaling. Our own studies have previously demonstrated that OA osteoblasts have increased phospho Erk1/2 signaling that was attributed to elevated endogenous IGF-1 levels in these cells[39], whereas a role of HGF has not been investigated in these cells. HGF can also modulate the Wnt signaling pathway by acting upon the response of Wnt agonist. Wnt signaling is crucial for normal skeletal tissue homeostasis and function. Interest- ingly, osteoblast functions which are altered in OA bone tissue and in cell culture have all been shown to be regulated by the cWnt signaling pathway under normal conditions. We previously reported that the abnormal phenotypic markers and reduced mineralization of OA

osteoblasts are linked with a reduction of the cWnt signaling pathway viaincreased expression of the antagonists DKK2[13]and sclerostin [38], and the inhibition of the expression of the non-Wnt family mem- ber yet Wnt agonist, R-spondin 2[37]. These alterations in OA osteoblasts have already been linked with elevated TGF-β1 levels in these cells. Indeed, the endogenous elevated production of TGF-β1 in OA osteoblasts has already been linked with alterations of the phenotype of these cells[13], and a reduction of their mineralization capacity[7].

Hence, as HGF drives TGF-β1 expression in OA osteoblasts, it was therefore not surprising that correcting HGF levels in OA osteoblasts,viaa HGF-siRNA technique, leads to a correction of Wnt3a-dependent Wnt signaling in these cells. Conversely, adding additional HGF to OA osteoblasts could no longer alter their response to Wnt3a, which could be viewed as indicating that the inhibition of this signaling pathway reached its maximal potential, and that it could not be reduced any further. However, inconsistent results using recombinant HGF in osteoblast-like cell culturesin vitrohave been linked with either abnormal levels of HGF and/or timing of HGF addition. Indeed, Kawasaki et al.

[43]showed that time of addition and duration of exposure can promote opposite responses to HGF in osteoblast-like cell models, therefore implying that local alterations in HGF release in OA osteoblasts could trigger abnormal responses to this growth factor as observed herein.

However, our data revealed that the expression of HGF was always increased in OA osteoblasts compared to normal osteoblasts during in vitrodifferentiation. OA osteoblasts failed to respond to exogenous HGF addition since their endogenous production was already increased.

Indeed, OA cells are exposed to continuous high levels of HGF, whereas normal osteoblasts did respond to HGF during continuous exposure up to 28 days of culture post-conﬂuence during the mineralization study.

A B

C D

Fig. 5.Effect of HGF inhibition on Smad1/5/8 and Wnt/β-catenin signaling activity in OA osteoblasts. In some cases, confluent OA osteoblasts were treated or not with 100 ng/ml rhBMP-2 for 15 min, and cells were lyzed in RIPA buffer for western blot analysis. In other experiments, confluent OA osteoblasts were initially treated with siSCR of siHGF for 48 h prior to transfection with TOPflash TCF/lef luciferase reporter plasmid. Cells were then treated for 4 h with either Parental or Wnt3a conditioned media in the presence or not of 10 ng/ml rhHGF. A) A representative western blot analysis of OA osteoblasts treated or not with 100 ng/ml rhBMP-2 for 15 min in the presence or not of PHA665752.

Phosphorylated and non-phosphorylated Smad1/5/8, andβ-actin were detected by western blot analysis (n = 4 experiments); B) TOPﬂash activity in OA osteoblasts in response to Parental and Wnt3a conditioned media either in the presence or absence of siHGF or rhHGF treatment. Values are reported relative to values in parental siSCR samples (n = 4 experiments); C) western blot analysis ofβ-catenin expression in OA osteoblasts in response to Wnt3a in the presence of either siSCR or siHGF. Representative experiment of n = 4 experiments; D) quantiﬁcation ofβ-catenin levels as detected in C).

(8)

These different data suggest that it is not the state of differentiation of our osteoblasts that alters the expression, nor the response to HGF, but the endogenous level of HGF in these cells. However, the altered expression of HGF in OA osteoblasts is responsible of their altered response to BMP-2. Whether HGF-driven TGF-β1 expression in OA osteoblasts is solely responsible for the alteration of the canonical Wnt signaling pathway observed in these cells cannot be ascertain. Indeed, HGF has also been shown to directly modulate the canonical Wnt signaling pathway and could contribute directly to reduceβ-catenin levels in these cells, a situation that was not evaluated herein and must await further studies.

In conclusion, we demonstrated that endogenous elevated HGF production in OA osteoblasts is responsible, in part, for their altered response to BMP-2 stimulation, leading to alterations of ALPase activity, osteocalcin secretion, and mineralization. This is linked with an inhibition of the phosphorylation of the ALK1/Smad1/5/8 pathway in response to BMP-2 and of the canonical Wnt signaling pathway in response to Wnt3a.

Author's contribution

EA, BB and XM contributed toin vitrostudy, analyzed the data and drafted the manuscript. EA, DL, JY, and PR contributed to the study de- sign. EA, BB, DL, JY, and PR contributed to the revision of theﬁnal manuscript.

Acknowledgments

This work was supported by grant MOP-49501 from the Canadian Institutes for Health Research to D Lajeunesse and the Osteoarthritis Chair of Excellence (RNG02CHAIR) (CNRS/Université de Lorraine) and in part by an award from the «Fondation Médicale pour la Recherche and la Région Lorraine» to P Reboul. E Abed is recipient of a Post-Doctoral award from the FRQ-S. Jouzeau/Reboul's and Lajeunesse's laboratories were awarded a scientiﬁc program of inter- national collaboration from CNRS/CRCHUM. The authors wish to ac- knowledge the expert technical assistance of Mrs Aline Delalandre for her contribution to this work.

Competing interests

The authors declare that they have no competing interests.

References

[1]Gevers G, Dequeker J, Martens M, Van Audekercke R, Nyssen-Behets C, Dhem A. Bio- mechanical characteristics of iliac crest bone in elderly women, according to osteoarthritis grade at the hand joints. J Rheumatol 1989;16(89):660–3.

[2]Gevers G, Dequeker J, Geusens P, Nyssen-Behets C, Dhem A. Physical and histomorphological characteristics of iliac crest bone differ according to the grade of osteoarthritis at the hand. Bone 1989;10(89):173–7.

[3]Raymaekers G, Aerssens J, Van den Eynde R, Peeters J, Geusens P. Devos P, et al. Al- terations of the mineralization proﬁle and osteocalcin concentrations in osteoarthritic cortical iliac crest bone, Calcif Tissue Int 1992;51(92):269–75.

[4]Hilal G, Martel-Pelletier J, Pelletier JP, Ranger P, Lajeunesse D. Osteoblast-like cells from human subchondral osteoarthritic bone demonstrate an altered phenotype in vitro: Possible role in subchondral bone sclerosis. Arthritis Rheum 1998;41(98):

891–9.

[5]Hilal G, Massicotte F, Martel-Pelletier J, Fernandes JC, Pelletier JP, Lajeunesse D. En- dogenous prostaglandin E2 and insulin-like growth factor 1 can modulate the levels of parathyroid hormone receptor in human osteoarthritic osteoblasts. J Bone Miner Res 2001;16:713–21.

[6]Massicotte F, Lajeunesse D, Benderdour M, Pelletier J-P, Hilal G, Duval N, et al. Can altered production of interleukin 1ß, interleukin-6, transforming growth factor-ß and prostaglandin E2 by isolated human subchondral osteoblasts identify two subgroups of osteoarthritic patients. Osteoarthritis Cartilage 2002;10:491–500.

[7]Couchourel D, Aubry I, Delalandre A, Lavigne M, Martel-Pelletier J, Pelletier J-P, et al.

Altered mineralization of human osteoarthritic osteoblasts is due to abnormal collagen type 1 production. Arthritis Rheum 2009;60(5):1438–50.

[8]Mansell JP, Bailey AJ. Abnormal cancellous bone collagen metabolism in osteoarthritis. J Clin Invest 1998;101(98):1596–603.

[9]Bailey AJ, Sims TJ, Knott L. Phenotypic expression of osteoblast collagen in osteoarthritic bone: production of type I homotrimer. Int J Biochem Cell Biol 2002;34:

176–82.

[10]Kougias P, Chai H, Lin PH, Yao Q, Lumsden AB, Chen C. Effects of adipocyte-derived cytokines on endothelial functions: implication of vascular disease. J Surg Res 2005;126:121–9.

[11]Brown SJ, Pollintine P, Powell DE, Davie MW, Sharp CA. Regional differences in mechanical and material properties of femoral head cancellous bone in health and osteoarthritis. Calcif Tissue Int 2002;71:227–34.

[12]Massicotte F, Fernandes JC, Martel-Pelletier J, Pelletier JP, Lajeunesse D. Modulation of insulin-like growth factor 1 levels in human osteoarthritic subchondral bone osteoblasts. Bone 2006;38:333–41.

[13]Chan TF, Couchourel D, Abed E, Delalandre A, Duval N, Lajeunesse D. Elevated Dickkopf-2 levels contribute to the abnormal phenotype of human osteoarthritic osteoblasts. J Bone Miner Res 2011;26:1399–410.

[14]Guévremont M, Martel-Pelletier J, Massicotte F, Tardif G, Pelletier J-P, Ranger P, et al.

Human adult chondrocytes express hepatocyte growth factor (HGF) isoforms but not HGF. Potential implication of osteoblasts for the HGF presence in cartilage. J Bone Miner Res 2003;18:1073–81.

[15]Zhen G, Wen C, Jia X, Li Y, Crane JL, Mears SC, et al. Inhibition of TGF-beta signaling in mesenchymal stem cells of subchondral bone attenuates osteoarthritis. Nat Med 2013;19:704–12.

[16]Reboul P, Pelletier JP, Tardif G, Benderdour M, Ranger P, Bottaro DP, et al. Hepatocyte growth factor induction of collagenase 3 production in human osteoarthritic cartilage: involvement of the stress-activated protein kinase/c-Jun N-terminal kinase pathway and a sensitive p38 mitogen-activated protein kinase inhibitor cascade. Ar- thritis Rheum 2001;44:73–84.

[17]Cheng N, Chytil A, Shyr Y, Joly A, Moses HL. Transforming growth factor-beta signaling-deﬁcientﬁbroblasts enhance hepatocyte growth factor signaling in mam- mary carcinoma cells to promote scattering and invasion. Mol Cancer Res 2008;6:

1521–33.

[18]Gupta DK, Singh N, Sahu DK. TGF-beta mediated crosstalk between malignant hepatocyte and tumor microenvironment in hepatocellular carcinoma. Cancer Growth Metastasis 2014;7:1–8.

[19]Inagaki Y, Higashi K, Kushida M, Hong YY, Nakao S, Higashiyama R, et al. Hepatocyte growth factor suppresses proﬁbrogenic signal transduction via nuclear export of Smad3 with galectin-7. Gastroenterology 2008;134:1180–90.

[20]Kwiecinski M, Noetel A, Elﬁmova N, Trebicka J, Schievenbusch S, Strack I, et al. Hepa- tocyte growth factor (HGF) inhibits collagen I and IV synthesis in hepatic stellate cells by miRNA-29 induction. PLoS One 2011;6:e24568.

[21]Kumarasinghe DD, Sullivan T, Kuliwaba JS, Fazzalari NL, Atkins GJ. Evidence for the dysregulated expression of TWIST1, TGFbeta1 and SMAD3 in differentiating osteoblasts from primary hip osteoarthritis patients. Osteoarthritis Cartilage 2012;20:

1357–66.

[22]Standal T, Abildgaard N, Fagerli UM, Stordal B, Hjertner O, Borset M, et al. HGF inhibits BMP-induced osteoblastogenesis: possible implications for the bone disease of multiple myeloma. Blood 2007;109:3024–30.

[23]Lin PS, Chang MC, Chan CP, Lee SY, Lee JJ, Tsai YL, et al. Transforming growth factor beta1 down-regulates Runx-2 and alkaline phosphatase activity of human dental pulp cells via ALK5/Smad2/3 signaling. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2011;111:394–400.

[24]Matsumoto Y, Otsuka F, Hino J, Miyoshi T, Takano M, Miyazato M, et al. Bone morphogenetic protein-3b (BMP-3b) inhibits osteoblast differentiation via Smad2/3 pathway by counteracting Smad1/5/8 signaling. Mol Cell Endocrinol 2012;350:78–86.

[25]Ehnert S, Zhao J, Pscherer S, Freude T, Dooley S, Kolk A, et al. Transforming growth factor beta1 inhibits bone morphogenic protein (BMP)-2 and BMP-7 signaling via upregulation of Ski-related novel protein N (SnoN): possible mechanism for the fail- ure of BMP therapy? BMC Med 2012;10:101.

[26]Chen CG, Thuillier D, Chin EN, Alliston T. Chondrocyte-intrinsic Smad3 represses Runx2-inducible matrix metalloproteinase 13 expression to maintain articular cartilage and prevent osteoarthritis. Arthritis Rheum 2012;64:3278–89.

[27]van den Berg WB. Osteoarthritis year 2010 in review: pathomechanisms. Osteoar- thritis Cartilage 2011;19:338–41.

[28]Di Nicola M, Carlo-Stella C, Magni M, Milanesi M, Longoni PD, Matteucci P, et al.

Human bone marrow stromal cells suppress T-lymphocyte proliferation induced by cellular or nonspeciﬁc mitogenic stimuli. Blood 2002;99:3838–43.

[29]Krampera M, Glennie S, Dyson J, Scott D, Laylor R, Simpson E, et al. Bone marrow mesenchymal stem cells inhibit the response of naive and memory antigen- speciﬁc T cells to their cognate peptide. Blood 2003;101:3722–9.

[30]Kuroiwa T, Kakishita E, Hamano T, Kataoka Y, Seto Y, Iwata N, et al. Hepatocyte growth factor ameliorates acute graft-versus-host disease and promotes hematopoi- etic function. J Clin Invest 2001;107:1365–73.

[31]Scharstuhl A, Vitters EL, van der Kraan PM, van den Berg WB. Reduction of osteophyte formation and synovial thickening by adenoviral overexpression of transforming growth factor beta/bone morphogenetic protein inhibitors during experimental osteoarthritis. Arthritis Rheum 2003;48:3442–51.

[32]Bau B, McKenna LA, Soeder S, Fan Z, Pecht A, Aigner T. Hepatocyte growth factor/scat- ter factor is not a potent regulator of anabolic and catabolic gene expression in adult human articular chondrocytes. Biochem Biophys Res Commun 2004;316:984–90.

[33]Singh S, Jones BJ, Crawford R, Xiao Y. Characterization of a mesenchymal-like stem cell population from osteophyte tissue. Stem Cells Dev 2008;17:245–54.

[34]Altman RD, Asch E, Bloch DA, Bole G, Borenstein D, Brandt KD, et al. Development of criteria for the classiﬁcation and reporting of osteoarthritis. Classiﬁcation of osteoarthritis of the knee. Arthritis Rheum 1986;29(86):1039–49.

(9)

[35]Smith PK, Krohn RI, Hermanson GT, Mallia AK, Gartner FH, Provenzano MD, et al. Measurement of protein using Bicinchoninic Acid. Anal Biochem 1985;

150(85):76–85.

[36]Gregory CA, Gunn WG, Peister A, Prockop DJ. An alizarin red-based assay of mineralization by adherent cells in culture: comparison with cetylpyridinium chloride ex- traction. Anal Biochem 2004;329:77–84.

[37]Abed E, Chan TF, Delalandre A, Martel-Pelletier J, Pelletier JP, Lajeunesse D. R- spondins are newly recognized players in osteoarthritis that regulate Wnt signaling in osteoblasts. Arthritis Rheum 2011;63:3865–75.

[38]Abed E, Couchourel D, Delalandre A, Duval N, Pelletier JP, Martel-Pelletier J, et al.

Low sirtuin 1 levels in human osteoarthritis subchondral osteoblasts lead to abnormal sclerostin expression which decreases Wnt/beta-catenin activity. Bone 2014;

59:28–36.

[39]Massicotte F, Aubry I, Martel-Pelletier J, Pelletier JP, Fernandes J, Lajeunesse D. Ab- normal insulin-like growth factor 1 signaling in human osteoarthritic subchondral bone osteoblasts. Arthritis Res Ther 2006;8:R177.

[40] Research Services Branch (RSB). Public domain NIH Image program (developed at the U.S. National Institutes of Health).http://rsb.info.nih.gov/nih-image/.

[41]Mutabaruka MS, Aoulad Aissa M, Delalandre A, Lavigne M, Lajeunesse D. Local leptin production in osteoarthritis subchondral osteoblasts may be responsible for their abnormal phenotypic expression. Arthritis Res Ther 2010;12:R20.

[42]Imai Y, Terai H, Nomura-Furuwatari C, Mizuno S, Matsumoto K, Nakamura T, et al.

Hepatocyte growth factor contributes to fracture repair by upregulating the expression of BMP receptors. J Bone Miner Res 2005;20:1723–30.

[43]Kawasaki T, Niki Y, Miyamoto T, Horiuchi K, Matsumoto M, Aizawa M, et al. The effect of timing in the administration of hepatocyte growth factor to modulate BMP-2- induced osteoblast differentiation. Biomaterials 2010;31:1191–8.

[44]Aspden RM, Scheven BAA, Hutchison JD. Osteoarthritis as a systemic disorder including stromal cell differentiation and lipid metabolism. Lancet 2001;357:1118–20.

[45]Lajeunesse D, Reboul P. Subchondral bone in osteoarthritis: a biologic link with articular cartilage leading to abnormal remodeling. Curr Opin Rheumatol 2003;15:

628–33.

[46]Aspden RM. Osteoarthritis: a problem of growth not decay? Rheumatology (Ox- ford) 2008;47:1452–60.

[47]Sanchez C, Deberg MA, Piccardi N, Msika P, Reginster JY, Henrotin YE. Osteoblasts from the sclerotic subchondral bone downregulate aggrecan but upregulate me- talloproteinases expression by chondrocytes. This effect is mimicked by interleukin-6, -1beta and oncostatin M pre-treated non-sclerotic osteoblasts. Os- teoarthritis Cartilage 2005;13:979–87.

[48]Dankbar B, Neugebauer K, Wunrau C, Tibesku CO, Skwara A, Pap T, et al. Hepatocyte growth factor induction of macrophage chemoattractant protein-1 and osteophyte- inducing factors in osteoarthritis. J Orthop Res 2007;25:569–77.

[49]Dequeker J, Aerssens J, Luyten FP. Osteoarthritis and osteoporosis: clinical and research evidence of inverse relationship. Aging Clin Exp Res 2003;15:426–39.

[50]Burr DB, Radin EL. Microfractures and microcracks in subchondral bone: are they relevant to osteoarthrosis? Rheum Dis Clin North Am 2003;29:675–85.

[51]Ferguson VL, Bushby AJ, Boyde A. Nanomechanical properties and mineral concentration in articular calciﬁed cartilage and subchondral bone. J Anat 2003;203:

191–202.

[52]Jaiprakash A, Prasadam I, Feng JQ, Liu Y, Crawford R, Xiao Y. Phenotypic characterization of osteoarthritic osteocytes from the sclerotic zones: a possible pathological role in subchondral bone sclerosis. Int J Biol Sci 2012;8:406–17.

[53]Goodnough JB, Ramos E, Nemeth E, Ganz T. Inhibition of hepcidin transcription by growth factors. Hepatology 2012;56:291–9.