Introduction
Osteoarthritis (OA) is considered a systemic and heterogeneous disease. The endpoint of OA is cartilage degradation, which is associated with and/or preceded by subchondral bone alterations [
1] and synovial membrane inflammation [
2]. The chronological events of these phenomena are still being debated in the literature, although recent developments emphasize the pathophysiological role of subchondral bone [
3]. In OA, it would appear that the extracellular matrix of both cartilage and subchondral bone are altered [
4],[
5]. Early-stage OA increases remodelling and bone loss, whereas late-stage OA decreases remodelling and increases subchondral densification. Both stages are important components of the pathogenic process that leads to OA lesions [
3]. In regard to subchondral bone changes, modifications of vascularization parameters have been suspected. The subchondral region of long bones is particularly vascularised, indicating high nutrient supply [
6] not only for bone cells but also for delivering nutrients to the chondrocytes of the deep layers of articular cartilage [
7]. Therefore, the epiphyses are at risk of circulatory insufficiency. Although vascularization of the subchondral bone plate seems to increase during OA progression [
8],[
9] and there are occurrences of numerous microchannels through the osteochondral plate [
10], the literature contains evidence of blood flow disturbances in this region. For instance, venous stasis [
11] was noted a long time ago and more recently was confirmed with dynamic contrast-enhanced magnetic resonance imaging (MRI) [
12] and positron electron transfer [
13]. Venous stasis sometimes leads to intraosseous hypertension in patients with OA [
11]. Furthermore, similar blood flow disturbances were demonstrated in the subchondral bone medial plateau of guinea pigs, which develop spontaneous knee OA [
14]. In this model, the observation of outflow obstruction preceding bone and cartilage OA lesions lends support to a key role of abnormal vascularization of OA bone tissues. These perfusion abnormalities have also been associated with osteonecrosis and bone marrow lesions [
12], which are dynamic episodes visualised by MRI often predictive of progressive OA for these patients [
15]-[
18]. Hence, this altered OA subchondral bone vascularization leads to ischemic episodes associated with hypoxic conditions [
11],[
19],[
20]. Interestingly, the presence of carbonic anhydrase IX, a recognized hypoxia biomarker, has recently been demonstrated in osteoarthritic bone [
21]. Hypoxia has been shown to drive several pathways in bone, such as growth [
22], regeneration and disease [
23],[
24], as well as the maintenance of haematopoiesis [
25]. Recently, Chang et al. demonstrated that hypoxia induced vascular endothelial growth factor (VEGF) expression as well as other genes involved in bone vasculature in OA osteoblasts (Obs) [
21]. In addition, they suggested that hypoxia modifies the OA Ob phenotype. Key mediators—namely, hypoxia-inducible transcription factors (Hifs)—play a vital role during these hypoxia. Hif heterodimers are composed of an oxygen-sensitive α-subunit and a constitutively expressed β-subunit. Hif-1α and Hif-2α are the best characterized α isoforms [
26]. During hypoxia, an accumulated Hif-α protein translocates to the nucleus and forms a dimer with Hif-β.
Among several factors potentially involved in OA pathogenesis, leptin and dickkopf-related protein 2 (DKK2) are good candidates [
27]-[
31], and their local production by Obs isolated from the subchondral bone seems to be important [
29],[
31],[
32]. Leptin levels in the synovial fluid are greater in OA patients than that in healthy people, and this adipokine exerts a proinflammatory effect in synovial fibroblasts [
33]. The level of leptin also increases in cartilage with the severity of OA as well as in subchondral bone [
29],[
34]. In fact, leptin might act as a proinflammatory factor on cartilage metabolism and exerts a catabolic effect on OA joints [
35]. Interestingly, members of our collaborative group previously reported that DKK2 is increased fourfold in OA Obs and that it also inhibits Wnt/β-catenin signalling, suggesting that DKK2 participates in the altered phenotype of OA Obs and to the reduced mineralization [
31]. The regulation of leptin gene by hypoxia seems to be cell-specific [
36]-[
38] or cell environment–dependent [
39],[
40], whereas DKK2 has not yet been described as a potential target of Hifs. Therefore, in the present study, we investigated this hypothesis in experiments using human OA Obs isolated from the subchondral bone plate.
Methods
Patients and clinical parameters
Tibial plateaus were obtained from patients with OA who were undergoing total knee replacement surgery. A total of 14 patients (age 69.1 ± 8.7 years; BMI 31.4 ± 9.6; 10 women and 4 men) who had OA according to the recognized clinical criteria of the American College of Rheumatology were included [
41]. All specimen collection and all procedures were approved by the ethics committee of the Nancy University Hospital (CHU) (agreement UF 9607-CPRC 2005) and conducted in conformity with the Declaration of Helsinki principles. Written informed consent was obtained from all participants as mentioned elsewhere [
42].
Preparation of primary subchondral bone cell cultures
Isolation of the subchondral bone plate and preparation of Ob cultures were performed using medial and lateral tibial plateaus, where bone sclerosis is observed as described previously [
43]. Cells were passaged once at 25,000 cells/cm
2 and grown for 5 days in Dulbecco’s modified Eagle’s medium/F12 containing 10% foetal bovine serum (FBS) and a 1% penicillin-streptomycin mixture. Confluent cells were then preconditioned for 24 hours with the same medium containing 0.5% FBS and then incubated in the presence of 50 nM 1,25-dihydroxyvitamin D
3 (VitD
3; Sigma-Aldrich, St-Quentin Fallavier, France) or of the vehicle (ethanol 96%, 1 μl/ml) under normoxia (20% O
2) or hypoxia (2% O
2) for different periods of time as specified in the experiments. Supernatants were collected at the end of the incubation period and kept at −20°C prior to assays.
RT-PCR assays
Total RNA was extracted from Obs using the RNeasy Mini Kit (QIAGEN, Courtaboeuf, France) according to the manufacturer’s instructions. Gene expression was analysed by quantitative real-time PCR (StepOnePlus Real-Time PCR System; Applied Biosystems, St-Aubin, France) using iTaq Universal SYBR Green Supermix (Bio-Rad Laboratories, Marnes la Coquette, France) according to the manufacturer’s protocol. The gene-specific primer pairs are given in Table
1.
RP29 served as a housekeeping gene.
Table 1
Oligonucleotides used for real-time RT-PCR
a
RP29
| S | 5′-GGG TCA CCA GCA GCT CGA GA-3′ | NM_001032.3 |
AS | 5′-CAG ACA CGA CAA GAG CGA GA-3′ |
Leptin
| S | 5′-GGC TTT GGC CCT ATC TTT TC-3′ | NM_000230.2 |
AS | 5′-GGA TAA GGT CAG GAT GGG GT-3′ |
DKK2
| S | 5′-GGG TTT TGC TGT GCT CGT-3′ | NM_014421.2 |
AS | 5′-TGG CTT TGG AGG AGT AGG TG-3′ |
Hif-1
| S | 5′-GAA AGC GCA AGT CCT CAA AG-3′ | NM_181054.2 |
AS | 5′-TGG GTA GGA CAT GGA GAT GC-3′ |
Hif-2
| S | 5′-TTG-ATG-TGG-AAA-CGG-ATG-AA-3′ | NM_001430.4 |
AS | 5′-GGA-ACC-TGC-TCT-TGC-TGT-TC-3′ |
Osteocalcin
| S | 5′-CAT-GAG-AGC-CCT-CAC-A-3′ | NM_199173.4 |
AS | 5′-AGA-GCG-ACA-CCC-TAG-AC-3′ |
VEGFα
| S | 5′-TGG-GCC-TTG-CTC-AGA-GCG-GA-3′ | NM_001171629.1 |
AS | 5′-GCC-TTG-CAA-CGC-GAG-TCTGT-3′ |
TGFβ1
| S | 5′-GCG-TGC-TAA-TGG-TGG-AAA-C-3′ | NM_000660.5 |
AS | 5′-GCT-GAG-GTA-TCG-CCA-GGA-A-3′ |
Sirt1
| S | 5′-GCT-GGA-ACA-GGT-TGC-GGG-AA-3′ | NM_012238.4 |
AS | 5′-GGG-CAC-CTA-GGA-CAT-CGA-GGA-3′ |
Western blot analysis
Cell extracts were loaded onto 4–20% SDS-PAGE precast gels (Bio-Rad Laboratories), and Western immunoblot analysis and signal detection were performed as previously described [
44]. As primary antibodies, we used a mouse anti-Hif-2α monoclonal antibody at a dilution of 1:1,000 (Thermo Scientific, Rockford, IL, USA), a rabbit anti-Hif-1α polyclonal antibody at a dilution of 1:1,000 (Novus Biologicals Interchim, Montluçon, France) and a rabbit anti-actin antibody at a dilution of 1:8,000 (Sigma-Aldrich). Anti-mouse and anti-rabbit horseradish peroxidase–conjugated immunoglobulin G secondary antibodies were used at a dilution of 1:10,000 (Thermo Scientific).
Inhibition of Hif-1α and Hif-2α expression by RNA silencing
Silencing RNAs (siRNAs) were obtained from Eurogentec (Liège, Belgium), and we followed the manufacturer’s directions for their preparation. OA Obs were split into 24-well plates at 25,000 cells/cm2 to obtain 50% confluence at the time of transfection. siHif-2α, siHif-1α or scramble siRNA (siScr; basal condition) was added to OA Obs at a final concentration of 10 nM in the presence of 2 μl of INTERFERin™ reagent (Polyplus Transfection, Strasbourg, France). After 24 hours, cells were starved with medium containing 0.5% FBS for 24 hours and then treated with 50 nM VitD3 or ethanol under hypoxic conditions for 6, 24 or 48 hours. Conditioned media were collected at the end of the incubation period and kept at −20°C prior to assays. Cells were harvested in either Buffer RLT Plus lysis buffer with β-mercaptoethanol (QIAGEN) for RNA extraction or in Laemmli 1× buffer for protein extraction. These experiments were reproduced with Obs from 5 different patients; (n= 5). Experiments with siRNAs targeting leptin or DKK2 were performed for a 48-hour period under similar hypoxic conditions. To evaluate Hif-1 activity, OA Obs split in 24-well plates were transfected with 500 ng of a Hif-1 response element Renilla luciferase reporter construct (SwitchGear Genomics, Carlsbad, CA, USA) in the presence of a polycationic transfection reagent (ExGen 500; Euromedex, Strasbourg, France) according to the manufacturer’s recommendations (in triplicates for each condition; n =4). Renilla luciferase activity was measured after 36-hour incubation of Obs in hypoxic conditions.
Inhibition of p40/p42 MAPK, p38 MAPK and phosphoinositide 3-kinase
OA Obs (n =5) were treated with p40/p42 inhibitor (10 μM PD98059; Calbiochem, Darmstadt, Germany), p38 inhibitor (5 μM SB203580; Calbiochem) or phosphoinositide 3-kinase (PI3K) inhibitor (5 μM LY294002; Calbiochem), or with vehicle (1 μl/1,000 μl dimethyl sulphoxide), under normoxia 1 hour before addition of 50 nM VitD3 or ethanol and incubation under hypoxia for 48 hours. Conditioned media were collected at the end of the incubation period for evaluation of leptin production (see next subsection), and cells were lysed in 0.5% SDS for protein quantitation by bicinchoninic acid assay.
Evaluation of leptin production
Leptin was evaluated in Obs-conditioned media using a human Quantikine immunoassay (R&D Systems, Minneapolis, MN, USA). The sensitivity of the assay is 7.8 pg/ml, and the intra-assay precision is 3.2 ± 0.2%.
Statistical analysis
Data are expressed as mean ± SEM. Statistical analyses were performed using either one-way analysis of variance followed by Dunnett’s multiple comparisons post hoc test or a t-test using GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA). P-values <0.05 between different experimental groups were considered significant.
Discussion
In the present study, we show that human OA Obs produced higher amounts of leptin under hypoxia than under normoxia. Production leptin was further increased in the presence of VitD3. In addition, DKK2, another gene with disturbed expression in OA, was upregulated under hypoxia. The regulation of leptin was mainly Hif-2α-dependent, because the knock-down of Hif-1α expression failed to affect leptin expression. Hypoxic regulation of leptin is also under the control of p38 MAPK and PI3K signalling pathways. Unlike leptin, the direct regulation of DKK2 seemed to be independent of either Hif-1α or Hif-2α. Also, under hypoxia, leptin did not alter the expression of DKK2, whereas, reciprocally, DKK2 did not alter the expression of leptin. This indicates that hypoxia per se was responsible for their respective increase and that neither leptin nor DKK2 can regulate each other.
Vascularization was reported to be increased in the subchondral bone plate during early to late stages of OA [
7],[
8]. In other studies, researchers found that bone hypertension and vascular stasis generated hypoxia in OA [
11],[
12],[
20]. Interestingly, in the spontaneous OA guinea pig model, changes in perfusion outflow of gadolinium with diethylenetriaminepentaacetic acid at 6 to 9 months temporally preceded and spatially colocalized with both cartilage degradation and bone remodelling in the medial tibial plateau [
12]. Measuring real partial oxygen pressure in bone is difficult to assess, and concentrations of 3% to 7% have been described [
45],[
46]. Meanwhile, Hif signalling has been demonstrated to intervene during both normal bone physiology and pathologies, indicating that values under 3% can be reached [
23]. These hypoxic conditions could thus explain why OA Obs produce higher levels of VEGF [
47] and hepatocyte growth factor [
48], particularly in the sclerotic zones [
49]. Consistent with these findings, our results show that hypoxia directly regulates VEGF expression by OA Obs (Figure
1A). Nevertheless, hypoxia could modify the phenotype of OA Obs [
21], and, as described previously, Hif-1α and Hif-2α are the main effectors in the modulation of gene transcription under hypoxic conditions [
23].
Regulation of the activity of these factors is complicated because their transcriptional regulation is not fully predictive of their protein level. For example, in the present study, 24-hour incubation with VitD
3 induced Hif-2α expression by almost twofold under normoxia (data not shown); however, protein stabilization was almost undetectable in immunoblot analysis (Figure
4A). However, after 6 hours of hypoxia, Hif-2α protein was strongly increased in the presence of VitD
3. Similarly, we observed that VitD
3 increased Hif-1α expression. It was not involved in leptin regulation, however. VitD
3 has been reported to be involved in the regulation of Hif-1α in breast epithelial cells [
50], but, to our knowledge, the present study is the first one demonstrating the regulation of Hif-2α by VitD
3. Among potential genes regulated by hypoxia, we paid attention to leptin, which plays a role in the pathogenesis of OA [
29],[
33]. In preliminary experiments, we found that leptin expression was similar at 20% and 12% oxygen tension, followed by a noticeable increase at 7% and a huge stimulation at 3% and 2% (data not shown). Leptin production was also synergistically regulated by hypoxia and VitD
3, and Hif-2α was the main regulator of leptin production under hypoxia. Conversely, Hif-1α played almost no role in leptin regulation in OA Obs. This situation has also been observed in fibroblast-like synoviocytes in rheumatoid arthritis [
37]. However, the alternate regulation by either Hif-1α or Hif-2α factors might be species-dependent, because leptin is regulated mainly by Hif-1α under hypoxic conditions in zebrafish [
51].
With regard to the leptin promoter, we found a hypoxia-responsive element (HRE) site juxtaposed to a potential vitamin D–responsive element (VDRE) site, which allows hypothesizing a theoretical cooperative interaction between VDRE and Hif-2α in a VDRE partnering model, as proposed by Carlberg and Campbell [
52]. Therefore, VitD
3 could be involved at two levels of leptin regulation, first by stabilizing Hif-2α protein levels and second by directly interacting with the VDRE present in the leptin promoter to synergistically stimulate leptin transcription. However, the presence of HRE and VDRE in the leptin promoter is probably necessary but not sufficient to presume a potential activation of transcription, because no synergy of transcriptional activity was noticed in other cell types [
36].
We observed that leptin production is strongly regulated via p38 MAPK and PI3K, although we have not discriminated whether this regulation directly affected Hif-2α phosphorylation or targeted other factors. The role of leptin in bone metabolism is controversial [
53],[
54]. However, in the context of hypoxia, which is known to induce osteocyte apoptosis within a few hours [
55],[
56], the need for bone remodelling is crucial to maintaining bone health. Therefore, the coactivation of leptin expression by hypoxia and VitD
3 should be considered a warning signal (that is, Obs sense oxygen tension), which, when oxygen tension decreases too drastically, alarms Obs to activate genes able to promote bone and vascular remodelling. In this context, leptin, which has been shown to increase Ob number and activity [
29],[
57], as well as osteoclast activity [
54],[
58], could play a role in remodelling. This type of scenario could happen, given the presence of defective capillaries within the subchondral bone plate, as reflected by the presence of stasis or microemboli, and could potentially lead to OA if the vasculature is not fully repaired. This would explain why Obs produce high levels of VEGF in response to hypoxia [
47], as we show in the present study, leading to a compensatory hypervascularization of the subchondral bone that might not, however, result in fully mature functional vessels. If we assume that vasculature dysfunction plays a role in OA [
20], this could explain the high prevalence of OA in patients who show signs of the metabolic syndrome, in whom the vascular system is often affected. If the vasculature fails to be well repaired, sustained amount of leptin could then be produced in the vicinity of the defect, which would lead over time to impairments in leptin signalling, as observed in obese patients [
59]. Therefore, the effects of both the anabolic and catabolic properties of leptin on bone cannot be fully achieved, leading to (1) an uncoupling of bone remodelling and, possibly, to sclerosis of the subchondral bone plate and (2) an abnormal OA Ob phenotype, as previously described [
29]. In addition to the overproduction of leptin, a combination of hypoxia and VitD
3 increased DKK2 expression, which has also been related to the abnormal phenotype of OA Obs [
31]. Unlike leptin, the expression of DKK2 was not significantly repressed by either Hif-1α or Hif-2α. Hence, as neither siHif-1α nor siHif-2α altered the expression of DKK2, which was stimulated under hypoxia, yet another mechanism must be set in place to regulate the expression of DKK2 under hypoxia. However, we and others previously demonstrated that transforming growth factor β1 (TGF-β1) can promote the expression of leptin in human Obs and rat bone tissue [
29],[
30], as well as the expression of DKK2 in human primary Obs [
31]. We also observed that hypoxia was able to induce TGF-β1 expression in OA Obs, which could explain, in part, the links between hypoxia, TGF-β1, leptin and DKK2 and why the expression of DKK2 could not be directly modulated by siHifs. However, the intricate relationship between these factors remains to be fully investigated.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BB contributed to the preparation of osteoblasts, participated in silencing and immunoblotting experiment design, performed experiments, analysed the data concerning leptin and DKK2 and drafted the manuscript. EA contributed to the preparation of osteoblasts, participated in transfection experiment design, performed experiments, analysed the data concerning DKK2 crosstalk between DKK2 and leptin and drafted the manuscript. MYO participated in the real-time RT-PCR experiments and drafted the manuscript. AB participated in the design of signalling pathway experiments, analysed the corresponding data, identified the potential VDRE response element adjacent to Hif-1 in leptin promoter and revised the manuscript. PN and DM performed statistical analyses and helped to revise the manuscript. DL and JYJ participated in the study design and coordination and helped to draft the manuscript. PR conceived of the study, participated in its design and experimental assays and helped to draft the manuscript. All authors read and approved the final manuscript. All authors agree to be accountable for all aspects of the work to ensure that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.