Introduction
Osteoarthritis (OA) represents a major cause of disability, particularly among the aging population; indeed, it is the most common form of joint disease. OA is a multifactorial disease characterized by loss of articular cartilage and subchondral plate thickening [
1]. As the loss of articular cartilage is believed to be the initial event responsible for joint destruction, numerous investigations have focused their efforts on understanding cartilage homeostasis. Therefore, biochemical analysis of the underlying bone tissue has received little attention, despite several reports of abnormal subchondral bone metabolism in OA.
Radin and coworkers were the first to study subchondral bone changes in patients with early degenerative joint disease, and to propose the participation of subchondral bone in the initiation and progression of cartilage damage [
2,
3]. They proposed that the thickening of the subchondral bone plate, resulting from repeated healing of microfractures, could be a key initiation factor in cartilage degradation. Other groups also reported abnormal cancellous bone collagen metabolism in OA, demonstrating that type I collagen, the most abundant bone matrix protein, is abnormal in OA [
4,
5]. Recently, we demonstrated that osteoblast (Ob) cells from human subchondral OA bone demonstrate an altered phenotype
in vitro. Our results showed increased alkaline phosphatase activity, release of osteocalcin, an increased activity of urokinase plasminogen activator (uPA) with no changes in plasminogen activator inhibitor-1 (PAI-1) abundance, and increases in insulin-like growth factor (IGF)-1 release compared to normal Obs [
6‐
8]. As IGF-1 production is increased in OA Obs compared to normal Obs, it is a likely candidate to promote bone remodeling and sclerosis in OA. Interestingly, our laboratory also demonstrated the presence of abnormal uPA regulation by IGF-1 in human OA Obs [
6]. These results suggest that IGF-1 signaling could be altered in these cells [
6]. The increased remodeling in OA bone could possibly account for the observation of hypomineralization of the subchondral bone tissues in established OA [
9‐
11]. Not only the bone matrix is altered in OA but recent studies have demonstrated that a putative factor(s) produced by OA subchondral bone cells can influence cartilage metabolism [
12]. This could possibly explain why increased subchondral bone activity can predict cartilage loss [
13‐
15].
After binding of IGF-1 to its specific surface receptor, the IGF-1 receptor (IGF-1R) kinase undergoes tyrosine phosphorylation of its α-subunit and kinase activation. This involves the phosphorylation of tyrosine residues of substrate adaptor proteins, principally the insulin receptor substrate (IRS)-1. Other targets have also been identified, such as Shc, IRS-2, IRS-3 and IRS-4 and GAB1 [
16,
17]. These proteins contain insulin/IGF-1R-specific tyrosine phosphorylation sites responsible for their association with various SH2 domain-containing downstream effector molecules. In the case of IRS-1, these include binding sites for phosphatidylinositol 3-kinase (PI3K), protein tyrosine-specific phosphatase Syp, 14.3.3 proteins, and the small adaptor protein Grb2, which is responsible for the activation of Ras and the MAPK pathway [
18,
19].
Thus, as the response to IGF-1 in human OA Obs is abnormal, we investigated IGF-1 signaling in OA Obs. Data revelead an abnormal interaction of phospho-Syp with IRS-1, possibly leading to decreased IRS-1 activity. Moreover, the interaction of Grb2 with IRS-1 was abnormal in OA Obs, possibly leading to altered downstream signaling. These data suggest that an abnormal response to IGF-1 is linked to abnormal intracellular signaling, affecting multiple pathways in OA osteoblasts.
Discussion
As an increasing amount of literature is demonstrating the involvement of bone tissue in the initiation/progression of OA, a better understanding of this tissue is clearly of utmost importance to better understand the etiology of this pathology. IGF-1 is one of the leading growth factors implicated in bone remodeling [
31]. Interestingly, IGF-1 expression is increased in OA Obs [
32] and these cells present an abnormal response to this growth factor [
6,
7]. Hence, we wanted to know if this is related to an alteration of its signaling pathway. We conducted a series of experiments to evaluate the IGF-1 signaling pathway in OA Obs and data revealed that the interaction of IRS-1 with Syp and Grb2 was modified in these cells in response to IGF-1 stimulation.
IGF-1R levels as well as phosphorylated IGF-1R levels following IGF-1 stimulation were similar in normal and OA Obs. However, IRS-1, the major IGF-1R docking protein, presented a reduced phosphorylation level in OA Obs, albeit the total protein level was similar between normal and OA Obs. In contrast, the IGF-1R docking protein Shc had similar protein and phosphorylation levels in normal and OA Obs. As IRS-1 was the only IGF-1R docking protein showing an abnormal modulation in OA Obs, we pursued our investigation with this factor and looked for molecules that could regulate IRS-1 phosphorylation.
First, 14.3.3 protein, which is known to bind IRS-1 and modulate its activation [
19], was not significantly different between normal and OA Obs, and thus is unable to explain the underphosphorylation of IRS-1 in OA. Second, as the phosphorylation of both Shc and IRS-1 is linked to IGF-1R kinase activation and the phosphorylation of Shc was similar between normal and OA Obs, we can not conclude that an abnormal IGF-1R kinase activation explains the reduced phosphorylation of IRS-1. Third, we looked for phosphatases able to modulate IRS-1 activity. The best characterized phosphatase that binds to and modulates IRS-1 is Syp (or SHP-2) [
33]. Syp protein levels were unaltered between normal and OA Obs. In contrast, its phosphorylation levels clearly demonstrated abnormal regulation in OA Obs. Indeed, OA Obs showed increased basal phosphorylation levels compared to normal, which was followed with a rapid decrease upon IGF-1 stimulation, in contrast to the situation for normal Obs. Moreover, the co-immunoprecipitation of Syp/IRS-1 also demonstrated an increased interaction in the basal state between IRS-1 and Syp in OA Obs, unlike in normal cells. Since there is an increase in IGF-1 production in OA Obs with a concomitant decrease of the major insulin-like growth factor binding proteins, namely BP-3, BP-4 and BP-5, OA Obs are likely to be more chronically stimulated by IGF-1 than normal Obs [
32]. This suggests that the dowregulation of IRS-1 in OA Obs, the major IGF-1 signaling pathway, is a feedback response to increased exposure to elevated endogenous IGF-1 levels.
The observed increase in ERK1/2 phosphorylation, while there was no significant increase in PI3K activity measured by Akt/PKB phosphorylation, should promote an increase in cell proliferation. In this respect, it is noteworthy that primary OA Ob cell cultures grow faster than normal Obs [
7], and we observed a reduction of the ratio of expression of Bax-α in OA Obs compared to normal. As Bax-α promotes apoptosis, a reduction in the Bax-α/Bcl2 ratio suggests an inhibition of apoptosis in OA Obs [
30]. In the present study we also observed that OA Obs can grow faster than normal Obs and that they respond to IGF-1 stimulation with a greater cellular proliferation rate. This response to IGF-1 was strictly ERK1/2 dependent since PD98059 was able to fully inhibit the effect of IGF-1 on OA Obs. Such a situation could then lead to more cells being available and prolonged cell life that would possibly lead to the laying down of more extracellular matrix, as reported in OA subchondral bone tissue [
4,
5]. This also agrees with the recent demonstration that Obs from OA patients show enhanced proliferation and collagen type I expression
in vitro compared to normal Obs [
34]. However, addition of exogenous IGF-1 to OA Obs failed to increase collagen type I levels, which are already higher in these cells than normal. In contrast, as IGF-1 promoted alkaline phosphatase production by Obs, it is noteworthy that it stimulated this activity better in OA Obs than in normal Obs, and that this was also dependent on ERK1/2 activity. Similar observations were previously reported for both activities in Ob-like cells [
35] and for cell proliferation alone in mesangial cells [
36]. Overall, these data would then suggest that the activation of the ERK1/2 pathway in OA Obs in response to IGF-1 is important for cell proliferation, retards apoptosis and affects alkaline phosphatase. This could also promote the production of collagen type I overall as more cells would synthesize it, resulting in more collagen being laid down
in vivo, although we could not show that more collagen was produced
per cell in vitro in response to IGF-1. This is reminiscent of observations made in other tissues where IGF-1 alone could not promote collagen type I production but, in combination with other growth factors or high glucose levels, could do so [
35‐
39].
One important question remains: if the main IGF-1R signaling pathway is dowregulated in OA Obs, how can we explain the increase in subchondral bone remodeling in OA bone tissue? Maybe the IGF-1 pathway is not implicated in this process. However, this seems unlikely since IGF-1 is a key regulator of bone remodeling and it increases uPA activity in OA Obs [
6]. Once stimulated, the tyrosine kinase activity of IGF-1R leads to its autophosphorylation as well as the phosphorylation of a number of intracellular proteins, such as IRS-1, Shc and Gab1. This gives rise to the activation of Ras and PI3K, thus resulting in the activation of MAPK and PKB [
40]. In our diseased cells, PI3K, MAPK and PKB protein levels were similar to normal, and we detected no significant increase in the activation of PKB while we observed a clear stimulation of the MAPK pathway. This implies the possibility of some complementary pathways for PKB activation that compensate for the lack of activation via IRS-1 in OA Obs. We recently demonstrated that transforming growth factor (TGF)-β production is increased in OA Obs [
20] and, since this growth factor can activate PKB in arthritis, TGF-β stimulation could be one such compensatory mechanism activating PKB. Interestingly, IGF-1 dependent p42/44 stimulation was significantly increased in OA Obs. Since the activation of the Ras/MAPK pathway can modulate the production of uPA following growth factor stimulation, a situation we already observed in OA Obs [
7], this suggests that the increased remodeling observed in OA subchondral bone could result from the upregulation of the p42/44 pathway following IGF-1 stimulation [
41].
Increased activation of the p42/44 pathway concomitant with the dowregulation of IRS-1 in OA Obs may seem contradictory. One possible mechanism could be competition between IRS-1 and Shc for Grb2 [
17]. In this model, IRS-1 and Shc compete for a limited cellular pool of Grb2, and the activation of the MAPK pathway would predominantly occur through the Shc-Grb2 signaling pathway. Grb2 is a small adaptor protein that can associate with IRS-1 and Shc via its SH2 domain and with the guanylnucleotide exchange factor for Ras, termed Son of Sevenless (SOS) via its SH3 domain. The association of the Grb2-SOS complex with tyrosine phosphorylated receptors and/or Shc have been directly implicated in the activation of the Ras signaling pathway. Since in OA Obs Grb2 levels were similar to normal, the downregulation of IRS-1 phosphorylation following IGF-1 stimulation in these diseased cells may result in an increased availability of Grb2 to the Shc pathway, leading to increased activity of the p42/44 pathway. However, as shown here, the interaction of Grb2 with IRS-1 was also increased in OA Obs, implying that the Grb2-Shc interaction should be reduced in these cells. On the other hand, the p42/44 kinase activity could also be activated directly by TGF-β via the Smad3 signaling pathway, as previously proposed by Sowa and colleagues [
42], a situation that overules Grb2-Shc signaling. Indeed, as OA Obs have elevated TGF-β levels [
20], this could directly activate the p42/44 pathway without the involvement of Grb2. Thus, the elevated endogenous TGF-β levels in OA Obs could then explain both the results for the PKB and p42/44 pathways observed here.
Taken together, these results could be interpreted as a general downregulation of the IGF-1R/IRS-1 pathways in OA Obs. However, dowstream signals were not actually reduced. The observed increase in Syp/IRS-1 interaction and increased Syp phosphorylation could actually promote IGF-1 signaling. A functional and highly phosphorylated SHP-2/Syp is necessary for sustained activation of ERK1/2 response to hepatocyte growth factor (HGF) stimulation in Madin-Darby canine kidney (MDCK) cells [
43] and in rat fibroblasts in response to insulin, IGF-1 or epidermal growth factor [
44]. Moreover, inactivating Syp antibodies [
44] or expression of a mutant phosphatase [
45] significantly reduces insulin, IGF-1 and epidermal growth factor signaling. Accordingly, as OA Obs showed high phosphorylated Syp levels and strong interaction with IRS-1, both under basal conditions and after IGF-1 stimulation, this could promote p42/44 activity in OA Obs, as observed in those studies. Moreover, a recent study indicated that functionally deficient SHP-1 mice are markedly glucose tolerant and insulin sensitive as a result of enhanced insulin receptor signaling to IRS-1 [
46], which suggests that elevated activity of SHP-1/Syp could reduce IGF-1 signaling to IRS-1, as observed in the present study. This would also suggest that, although IGF-1R-dependent IRS-1 phosphorylation is reduced in OA Obs, Syp phosphorylation and activity could compensate for this reduction. As Syp is central to other growth factors, such as HGF and epidermal growth factor [
43,
44], and since we recently showed a key role for HGF in OA Obs [
47] and possibly for the cross-talk between OA Obs and cartilage tissue [
48], the present results suggest that the HGF-dependent pathway could also be altered in OA Obs, a situation not investigated at present.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
FM performed most of the experiments and wrote the first draft of the manuscript. IA performed the experiments shown in Figure
4 and contributed to writing the manuscript. JM-P and J-PP contributed to writing the manuscript and discussion of the results. JCF provided the OA knee samples and contributed to discussion of the results. DL proposed original concepts, planned and performed some of the experiments, performed the statistical analyses, participated in the discussion and wrote the final version of the manuscript.