The gene expression and protein profiles of ADAMTS and TIMP in human chondrosarcoma cell lines induced by insulin: The potential mechanisms for skeletal and articular abnormalities in diabetes

Background:The delay in wound healing, decrease in the long bones resilience to fracture, and delay in fracture healing are among common complications DM patients, and they still remain as challenging issues to be solved. The mechanism has not been fully understood yet, but high sugar and/or insulin deficiency or unresponsiveness to insulin in blood are potential causes to blame. Extracellular matrix degradation/remodeling is one of the important mechanisms whereby cell differentiation, bone remodeling and wound repair can be regulated. ADAMTS proteins play important roles in cartilage/bone metabolism. This study aimed to determine whether ADAMTS/TIMP proteins were affected by insulin application in OUMS-27 cells.Material and methods:OUMS-27 cells were induced by 10μg/mL insulin for 1,3,7,and 11days. Cells were harvested, mRNA and protein extractions were performed. Total mRNA and cDNA levels were measured by qRT-PCR and protein levels were detected by WB.Results:ADAMTS1,5, and 7 levels were significantly decreased, while TIMP-3 levels were detected increased (mRNA/protein concentrations).Conclusions:Pathologies and disturbances of cartilage/bone metabolism, delayed fracture healing in particular, in patients with DM may result from insulin deficiency. ADAMTS genes that play a role in healing process are increased during insulin deficiency, which consequently interrupts healing process by causing cartilage ECM degradation.


INTRODUCTION
Glucose is an essential nutrient for metabolism and structural needs for the growth centers of the hyaline cartilage and the bone growth center. When chondrocytes are left in an anaerobic environment, it does not only use glucose as a primary substrate for the production of ATP (1,2), but it is also the source of glucosamine sulfate which is important for the development, protection, reparation and remodeling of the cartilage. Wound healing, bone resilience, healing of fractured bones and in procedures done before tooth implant surgery to change the bone structure are abnormal in patients with type-2 diabetes mellitus (DM). experiment has ended, the samples were run and visualized on an agarose gel (2 %) to assess the results. The optimal annealing temperature and primer combination was given to the band with the highest intensity (yield) with no nonspecific products. The calculated Tm values by gradient PCR were given in Table 1. Protein isolation, Immunoblotting and Antibodies: Anti-ADAMTS1, Anti-ADAMTS4, Anti-ADAMTS5, Anti-ADAMTS7, Anti-ADAMTS15, Anti-ADAMTS18, Anti-TIMP-3, Anti-TIMP-4, and Anti-GAPDH primary antibodies were purchased from Santa Cruz Biotechnology, Inc., CA and used at a 1:1000 dilution. Cross-reactivity was confirmed before the study to match with data that described on the manufacturer's data sheet. The cells were washed once with phosphate-buffered saline and scraped from the plates. Cells were solubilized in 300 μL of CelLytic M (Sigma Aldrich, St. Louis, MO) with a protease/phosphatase inhibitor (Pierce Protease and Phosphatase Inhibitor Mini Tablets, EDTA Free, Thermo Fisher Sci.). After incubation in at 4 °C for 15 min, the samples were centrifuged (15 min, 12,000xg). The protein concentration of cell extracts was determined using Thermo Scientific Bradford Assay kit and BSA as standard. Protein samples were boiled at 95 °C in Laemmli sample buffer (BioRad) and β-mercaptoethanol for 8 min. Ten micrograms of total protein were run in Western blotting (WB) tanks. Briefly, 10μL of each sample including protein marker (BioRad Precision Plus Protein Western C Standard) were loaded to WB gel (BioRad Mini PROTEAN TGX Stain Free Gels, 4-15 %, 15-well comb, 15 μL) in BioRad 1XTris/Glycerine/SDS running buffer and run at 250 V for 20 min. After electrophoresis, proteins were transferred onto PVDF membranes (BioRad Trans-Blot Turbo Transfer Pack, 0.22 μM) (BioRad, Singapore). Membranes were blocked for one hour in 2.5 % nonfat dried skim milk in Tris-buffered saline containing of 0.05 % Tween-20. The membranes were incubated overnight with primary antibodies ( Table 2) diluted in blocking buffer. After washing with Tris-buffered saline/Tween 20 (thrice) for 8 min. each at room temperature, the membranes were incubated for one hour with the secondary antibodies ( Table 2). Following three successive washes, immunoreactive bands were visualized using enhanced chemiluminescence system (BioRad Immun-Star Western C kit) for 90s. Signals were detected with BioRad ChemiDoc MP Imaging System (Singapore), and the densitometry was performed with Image-J software and normalized to the signal intensity of GAPDH for equal protein loading control of each sample in each experiment. This quantification was performed with the linear range of the standard curve defined by the standard GAPDH, for all densitometry analysis.
Statistical Analysis: Statistical Package for Social Studies (SPSS) version 16.0 was used for statistical tests. Nonparametric Kruskal Wallis Test was applied. The relationships between the variables were tested by Mann Whitney-U test. P<0.05 was accepted as significant.

RESULTS
In order to assess the effects of insulin on ADAMTS gene expression in cultured OUMS-27 cells, qRT-PCR was performed on mRNA eluted from cell lysate. Quantitation and standard curves as well as other data of qRT-PCR instrument were regular and flawless. On the other hand, in order to achieve a better understanding of the mechanism of insulin on cartilage matrix degradation, we analyzed ADAMTS proteins in cultured OUMS-27 cells in vitro. WB analysis of ADAMTS proteins using antibodies, which recognize specific enzyme proteins of ADAMTS, was performed. In the graphics of WB analysis, x-axis shows insulin application time intervals (in days), y-axis shows band densities of ADAMTS and TIMP proteins normalized by GAPDH densities. There is no numeric value of this ratio. ADAMTS1/GAPDH graphic is shown in Figure 1A. From quantitative nucleic acid concentration graphics, cDNA concentration of D1, D7 and D11 groups was rather low compared to that of control group P= 0.047, 0.009, 0.009, and 0.028, respectively). It generally means that insulin application lead to a decrease in amplicon levels of ADAMTS1. There was also a statistically significant difference in amplicon concentrations between D1 and D7 groups (P= 0.028) ( Figure  1A). It seems that ADAMTS1 protein production was decreased upon insulin induction in OUMS-27 cells. Immunoreaction between antibodies and proteins attached to the membranes revealed two bands: 110 kDa and 70 kDa. It was suggested that similar bands could be found after protein degradation. According to the manufacturer's instruction and the related scientific publications, the second band was being expected around 85 kDa. The bands shown on 110 kDa and 70 kDa were identified as preform (inactive) and proform (active), respectively. Protein levels at first day after insulin induction were the same compared to control group. There was a dramatic decrease on D3, then the decrease continues up to D7. Although ADAMTS1 level increases on D11, the level was still about one fourth of the beginning level ( Figure 1B).
The results of ADAMTS4 nucleic acid concentration are given in Figure 2A. Nucleic acid concentrations were decreased in D1 group and increased in D3, D7, and D11 groups compared to control group. However, as seen from high-rise SD bars, there were no significant differences between study and control groups (Figure 2A).

Figure 1: (A) qRT-PCR test results of A disintegrin and metalloproteinases with thrombospondin type 1 motif 1 (ADAMTS1) in OUMS-27 chondrosarcoma cells. Values were standardized and normalized by proportioned with GAPDH values. Bars and error bars represent mean values and standard deviations, respectively. There were statistically significant differences between Control-Day 1 study group (C-D1), C-D7, C-D11, and D1-D7 (P = 0.047, 0.009, 0.009, and 0.028 respectively). (B) The electrophoretic image of ADAMTS1 protein of OUMS-27 cells by Western blotting technique and their graphics (two different bands) obtained by the calculation of band densities by CCD camera and ImageJ program. The x-axis shows insulin application time intervals (in days) and y-axis shows band densities of ADAMTS1. The densities for ADAMTS1 were normalized with GAPDH densities.
Although the antibody used for ADAMTS5 protein showed two bands at 105 kDa/75 kDa, the second one could not be obtained properly. Thus, only 105 kDa band was interpreted. The production of ADAMTS5 protein was decreased upon insulin induction in OUMS-27 cells. Protein levels were decreased gradually during the experiments up to D7. Band density was increased to some extent on D11 ( Figure 3B). Figure 4A shows the quantitation of nucleic acid concentration of ADAMTS7. Concentrations of D1 and D3 groups are quite similar to those of control group. Although there is an increase upon insulin induction in D1 group, it was not expected to be significant because of high SD. However, nucleic acid concentrations of D7 and D11 groups were significantly decreased (P= 0.009). It probably means that insulin application affected mRNA levels of ADAMTS7 proteins at least in two groups. It is not valid for D1 and D3 groups of cells. In addition to that, there were statistically significant differences in cDNA concentrations between D1-D7, D1-D11, D3-D11 (P= 0.016, 0.009, and P= 0.050, respectively) ( Figure 4A).

(B) The electrophoretic image of ADAMTS5 protein of OUMS-27 cells by Western blotting technique and its graphic (one band) obtained by the calculation of band density by CCD camera and ImageJ program. The x-axis shows insulin application time intervals (in days
) and y-axis shows band densities of ADAMTS5. The density for ADAMTS5 was normalized with GAPDH densities.
ADAMTS7 protein level was decreased dramatically during the experiment in OUMS-27 cells (there was a 40 % decrease in protein amount of 70 kDa band and 60 % decrease for 55 kDa at D1 of experiments compared to controls). There was a gradual decrease in both bands up to around 90 % in D11 ( Figure 4B).
The qRT-PCR results of ADAMTS15 is presented in Figure 5A. Although there were some differences in quantitative nucleic concentrations of groups D1 (increase), D3 (increase), D7 (decrease), and D11 (decrease) compared to control, they were statistically insignificant. As a conclusion, insulin application did not affect ADAMTS15 mRNA concentrations. There were two bands for ADAMTS15. The first band was anticipated at around 103 kDa according to prospectus whereas it was found at around 70 kDa. Two possibilities arise from this discrepancy: This band may originate from degradation of ADAMTS15 or it may be accepted as a nonspecific band. ADAMTS15 protein levels were affected by insulin in OUMS-27 cells. At the first and subsequent days, protein amounts were decreased gradually up to D7. After that day, 55 kDa band continued to decrease, but 70 kDa band showed an increase up to those levels of D1 ( Figure 5B).
There was a gradual increase in ADAMTS18 nucleic acid concentrations of groups D1, D3, D7, and D11 in OUMS-27 cells induced by insulin ( Figure 6A). Because there were great differences between the individual results, statistical analyses gave no difference between study groups and control group. The antibodies used for detection of ADAMTS18 in OUMS-27 cells gave three bands in WB: two (100 kDa and 75 kDa) of them were considered to be nonspecific bands. The smallest band, around 45-50 kDa, was considered to be the active form of enzyme. The decrease in the protein amount of ADAMTS18 upon insulin induction was found to be negligible when compared to other ADAMTS proteins. There was a 30% decrease on the first day of insulin induction, and then it remained stable around this percentage during 11 days of experiment ( Figure 6B).
Although OUMS-27 cells produced 6 times more TIMP-3 protein in D1 after insulin induction compared to controls, it was not statistically significant because of a huge gap between lowest and highest values of instrument analyses. There was serious decrease in TIMP-3 cDNA levels after D3. The result of D3 was not statistically significant although a prominent (20%) decrease was observed compared to control group. However, the decrease in nucleic acid concentration of groups D7 and D11 compared to control group was statistically significant (P= 0.017 and 0.047, respectively). Therefore, the insulin application led to the decrease in TIMP-3 mRNA levels in OUMS-27 cells. This is not valid for D1 and D3 groups. Besides, there were significant differences in cDNA concentrations between D1-D7, D1-D11, and D3-D7 groups (P= 0.016, 0.021, and P= 0.028, respectively) ( Figure 7A).

Figure 6: (A) qRT-PCR test results of A disintegrin and metalloproteinases with thrombospondin type 1 motif 18 (ADAMTS18) in OUMS-27 chondrosarcoma cells. Values were standardized and normalized by proportioned with GAPDH values. Bars and error bars represent mean values and standard deviations, respectively. There were no statistically significant differences between the groups. (B) The electrophoretic image of ADAMTS18 protein of OUMS-27 cells by Western blotting technique and its graphic (one band) obtained by the calculation of band density by CCD camera and ImageJ program. The x-axis shows insulin application time intervals (in days) and y-axis shows band densities of ADAMTS18. The density for ADAMTS18 was normalized with GAPDH densities.
The bands of TIMP-3 were detected around 25 kDa. Insulin induction led to a marked increase in TIMP-3 protein amount in OUMS-27 cells. When compared to control group, protein levels were decreased 49% on D1, 63% on D3, and 22% on D7 and finally protein amount suddenly showed a 2.75 fold increase compared to control values ( Figure 7B).
TIMP-4 bands were detected around 22 kDa. The effect of insulin on TIMP-4 proteins in OUMS-27 cells was limited. Insulin induction during all 11 days led to an increase of merely 10 %. The levels of TIMP-4 protein on other harvesting days were almost the same as those of control group (Figure 8). 0.016, 0.047, 0.016, 0.021, and 0.028 respectively)

. (B) The electrophoretic image of TIMP-3 protein of OUMS-27 cells by Western blotting technique and its graphic (one band) obtained by the calculation of band density by CCD camera and ImageJ program. The x-axis shows insulin application time intervals (in days) and y-axis shows band densities of TIMP-3 protein. The density for TIMP-3 was normalized with GAPDH densities.
At a glance, the WB and qRT-PCR results overlap at some points and show differences in others. For example, while mRNA and protein measurements show parallelism in terms of ADAMTS1, 5 and 7 concentrations, the mRNA and protein levels show differences in terms of TIMP-3. It has been observed that some changes in the protein levels of other studied ADAMTS enzymes have not occurred in mRNA levels. For example, ADAMTS4 protein decreased gradually after the application of the insulin. This gradual reduction is present in both observed bands. If we only compare the control and D11 groups, we can see a half reduction on the first and a nearly 3-fold reduction on the second band. No difference is found in the qRT-PCR results between the control and insulin-induced group. It is the same for ADAMTS15. In fact, the results are more dramatic here. The reduction on the 55 kDa band is more than 10 times in the D11 insulin-induced group. A 40% reduction in the D1 group, a 3-fold reduction in the D2 group and a 5-fold reduction in the D7 group are present. Although not as severe as that in 70 kDa band, when the D11 group is considered, the two-fold reduction attracts attention. If we compare the qRT-PCR results of ADAMTS15 of the control and insulin groups, the differences appear to be statistically insignificant. In other words, no significant difference is observed in the mRNA levels. In addition to this, the ADAMTS18 results of the WB and qRT-PCR analyses have again revealed to be parallel. In both methods, no significant differences have been found between the control and insulin groups. Because TIMP-4 has only been examined with the WB method, it has not been possible to compare the results. As for the differences of ADAMTS4 and 15 between the two methods; the cells have maintained their mRNA levels in the insulin and non-insulin groups up to the protein phase, but after that the protein stayed intact in the insulin-free environment whereas the level decreased due to digestion of the protein because of insulin induction. This result actually brings a different aspect of the original agenda: Insulin may be acting as a controlling regulator of the digestion of these proteins.

DISCUSSION
In this experimental study, we investigated the effects of insulin in chondrosarcoma cells in vitro. Since the study was performed in an immortal cancerous cell line in vitro, it is essential to refrain from overinterpretations and to be cautious when conducting and extrapolating these findings in vivo and disease conditions. However, it is also essential to connect the basic findings with real disease conditions in order to give an objective estimation to the readers. The obtained positive and/or negative findings of PCR and WB are important in several aspects and might be qualified as original: 1) Because our experiment was conducted in a way that the induction of insulin to the cells was investigated and the absence/ineffectiveness of insulin in DM was considered as a precondition, the exact opposite of our findings should be considered. In this case, there should be two changes; the intracellular matrix degrading enzymes would be activated and their inhibitory enzymes would become passive. These two findings make this study unique, because molecular causes of fracture healing problems in DM have a potential for further research. There might be some clues in our findings to reveal the molecular mechanisms of the fracture healing problems in DM.

Figure 8: The electrophoretic image of the tissue inhibitors of metalloproteinases-4 (TIMP-4) protein of OUMS-27 chondrosarcoma cells by Western blotting technique and its graphic (one band) obtained by the calculation of band density by CCD camera and ImageJ program. The x-axis shows insulin application time intervals (in days) and y-axis shows band densities of TIMP-4 protein. The density for TIMP-4 was normalized with GAPDH densities.
2) The main reason for increased fragility of long bones in patients with uncontrolled DM may be absence and/or ineffectiveness of insulin. The reducing and/or inhibitory effects of insulin on ADAMTS proteins, ensure that the cartilage and bone tissues prefer to go to anabolic direction. In insulin absence, it can be anticipated that an increase in ADAMTS activity may accelerate the destruction of ECM in cartilage and bone tissues, leading consequently to a reduction in the resilience to bone fractures.
Little is known about the effect of DM on the cartilage phase of fracture healing. This phase plays an important role in the reparation process because the cartilage acts as a template for the subsequent bone formation (24). Our study has proven the reducing effect of insulin on three ADAMTS and the increasing effect on TIMP protein, and also sheds light on the molecular basis of deterioration of cartilage and bone repair in diabetes. Previous studies have shown that diabetic animals create a fracture callus, but the volume compared to that in normal animals is much smaller (25). Clinical and case studies have reported that diabetes delays fracture healing and dramatically increases recovery time (26). Animal studies have shown that DM causes the formation of small calluses resulting in the reduction of bone and cartilage development, reduces the proliferation and differentiation of osteoblastic cells and chondrocytes and that the mechanical strength is reduced two times during fracture reparation (9,16). An important decrease of collagen content in the callus has been identified in diabetic animals compared to normoglycemic animals (16). In an experimental study, during the healing of a tibial fracture, an unidentified catabolic effect and accelerated cartilage loss in the diabetic group was determined (27).
When insulin is induced into SRC chondrocyte cells, the insulin receptors have increased and this is not interpreted as an increase in the de novo receptor synthesis or a translocation of other intracellular receptors located in the cell compartments to the cell surface, but most likely as a decrease in the rate of receptor degradation (28). Chondrocytes synthesize and secrete the macromolecules forming the ECM of cartilage cells. SRC chondrocytes respond in a very special way to physiological concentrations of porcine insulin at in vitro conditions and produce plenty of cartilage-like proteoglycans (29), type-II-collagen (30), hyaluronic acid (31) and other secreted proteins (32). Rat chondrosarcoma chondrocytes are insulin-and IGF-1-dependent tumors. This tumor, shrinks significantly in hypophysectomized rats and also shows a distinct decrease in rats with streptozotocin-induced diabetes (33).
ADAMTS4/5 undertakes important roles in the degradation of the cartilage matrix proteoglycan (34). These enzyme levels drastically increase in animals with DM (35) and proteoglycans are removed very rapidly from the ECM. However, this process is slowed down with insulin therapy. Parallel to that, when insulin has been applied into chondrocyte cells as in our study, the levels/activity of these proteins have decreased and the removal of the proteoglycan groups have slowed down. We can provide a similar effect by directly introducing insulin into chondrosarcoma cells, so when viewed in terms of protein, we see that ADAMTS4 and 5 decrease drastically.
To understand the disorders occurring during increased fragility of long bones and delay in healing of fractured bones, the endochondral ossification process needs to be well analyzed. The cartilage matrix is digested during endochondral ossification. Studies in recent years related to the proteolytic enzymes, which could be responsible for the digestion, are based on two important main structures of the cartilage matrix proteins, type-II collagen and aggrecan digesting proteinases. Fibrillary collagen and aggrecan digesting MMP13 inside the growing cartilage are especially expressed in hypertrophic chondrocytes (36). The chondrocytes undergo regular hypertrophy in MMP13 knockout mice, but the end of hypertrophic chondrocytes expands and the leading ossification invasion delays (37). The extinction of the MMP13 protein as a result of a mutation in the MMP13, results in a growth defect called Missouri-type spondylometaphyseal dysplasia (38). Contrary to MMP13, MMP9 does not digest native fibrillar collagen, but can digest denatured collagen and aggrecan (36). The deletion of both MMP9 and MMP13 genes in mice results in huge delay during ossification compared to the situation in which only one deletion in either MMP9 or MMP13 genes takes place (37). ADAMTS1, 4 and 5 knockout mice can grow up normal and do not show any defect in the cartilage (39,40). Because these are the most important aggrecanases found in the cartilage, none of them takes part alone in the removal of cartilage aggrecan in a significant amount. Various enzymes belonging to the MMP and ADAMTS families are able to function interchangeably, thus if any single ADAMTS gene undergo knockout, another can fill its place to avoid growth defects. An alternative interpretation is the following: for a normal invasion premise of ossification, the removal of the fibrillar collagen is required, but the aggrecan removal is not important to this process.
One of the most important proteins in our study, the ADAMTS1 gene, has initially been cloned as a gene that is responsive to inflammation (41). It is not expressed in normal tissues but induced by lipopolysaccharide stimulation. It has been shown that ADAMTS4 together with ADAMTS1 play a key role in versican proteolysis (42). When compared with MMPs, ADAMTS proteases recognize substrates such as procollagen and proteoglycans with a broader specificity (42). It has also been reported that ADAMTS1 acts with a different function as angiogenesis inhibitor (43). One interesting fact is that the matrix around the cells does collide to allow chondrocyte hypertrophy. The fact that neither MMP nor ADAMTS proteins prevent this process of genetic manipulation, leads us to think that other proteinases are responsible for this process. Cathepsins and calpains are candidates for such a role because they are expressed by chondrocytes and have matrix destructive activities (36).
TIMPs are important for various physiological process controls. These can be classified as cell invasion, angiogenesis, digestion of articular cartilage, trophoblast implantation, inner fold formation of mammary gland, and wound healing (44,45). According to our results, the TIMP-3 levels were increased. qRT-PCR results do not support this finding. According to our WB results, the TIMP-3 level was increased. A significant decrease has been observed by D7 and 11. This increase in protein level presents a parallelism in the decrease of the ADAMTS level. TIMP-3 blocks the tissue proteinases, the cartilage tissue ECM cannot be cleaved, and the structure remains intact. No study was found about the levels of TIMP-3 and 4 after insulin induction to compare present results. In this respect, our findings are original. Upon insulin application, the TIMP-3 level will rise, trigger the anabolic processes in cartilage and healing bone tissue, contributing to tissue strength. TIMPs are natural inhibitors of MMPs. Factors like cytokines, hormones, and many other factors, including growth factors, affect the metalloproteinase activity (45). There are a lot of biological factors that affect metalloproteinase activities, including cytokines, hormones, growth factors, and many others (45).
Aggrecan proteoglycan gives the cartilage features, such as strength against the pressure force, dynamic weightbearing function and an osmotic feature (46). The aggrecan, which is an incredibly complex macromolecule, is specific to the cartilage and intervertebral disc, and is an extremely hydrophilic proteoglycan that consists of glycosaminoglycan chains, containing about 100-chondroitin sulfate and 30-keratin sulfate chains (47). The breaking down of the aggrecan is one of the important indicators of osteoarthritis. In the analysis of synovial fluid of arthritis patients, it has been shown that the aggrecan is pathologically cut from the aggrecanase cutting field (48). Many members of the ADAMTS family cut the molecule in vitro at this cutting point (49): most effective ones are ADAMTS4 and 5. Both of them are expressed in the normal and osteoarthritic cartilage and the synovium (50). Their inhibition prevents the collapse of aggrecan in vitro (51). ADAMTS4/5 knockout mice phenotypically differ from wild-type rats. The results obtained from this study have lots of data that shed light on mechanisms of many joint diseases, especially osteoarthritis. The loss of aggrecan from the cartilage in joint pathologies is mainly a proteolytic process conducted by aggrecanases. In vitro studies with human recombinant ADAMTS4 (52) and ADAMTS5 (53) enzymes, have shown that aggrecan was preferably cleaved from the CS-2 part. According to our findings, insulin behaves in a direction that might be classified totally as cartilage protective.
These effects of insulin on the chondrosarcoma cells, which we have been discussing in details, whether it is turning hyperglycemia to its normal state or its direct effect on cells, can be listed as following: accelerating fracture healing, improving the resilience to bone breakage/fractures, providing the growth and development of skeletal structures in children and ensuring success with pre-operational processes in dental implant surgery. To further investigate the subject and to clarify the exact mechanism(s), it would be helpful to investigate the ADAMTS and TIMP in terms of gene, mRNA, and protein phases in animals with experimentally induced DM.