Pharmacoproteomic study of the effects of chondroitin and glucosamine sulfate on human articular chondrocytes

Osteoarthritis (OA) is becoming increasingly prevalent worldwide because of the combination of an aging population and growing levels of obesity. Despite the increasing number of OA patients, treatments to manage this disease are limited to controlling pain and improving function and quality of life while limiting adverse events [1]. Effective therapies to regenerate damaged cartilage or to slow its degeneration have not been developed.

The failure of conventional treatments (analgesics or non-steroidal anti-inflammatory drugs) to satisfactorily control OA progression, combined with their frequent adverse side effects, may explain the increasing use of such SYSADOA (SYmptomatic Slow-Acting Drugs for Osteoarthritis) therapies as glucosamine sulfate (GS) and chondroitin sulfate (CS). Different clinical trials have proved that GS [2‐4] and CS [5, 6] are effective in relieving the symptoms of OA [7], probably due to their anti-inflammatory properties. However, although these reports were intended to resolve and clarify the clinical effectiveness of these supplements regarding OA, they leave doubts among the scientific community and fuel the controversy [8]. The recently published results of the Glucosamine/chondroitin Arthritis Intervention Trial (GAIT) showed that, in the overall group of patients with osteoarthritis of the knee, GS and CS alone or in combination did not reduce pain effectively [9]. For a subset of participants with moderate-to-severe knee pain, however, GS combined with CS provide statistically significant pain relief compared with placebo. One possible explanation for this discrepancy may be the relative participation of inflammatory cytokines in different subpopulations; and it is also hypothesized that the effects of GS and CS are better realized in patients with more severe OA, which have greater involvement of interleukin-1beta (IL-1β) [10].

With the aim to describe more clearly the effects of GS and CS on cartilage biology and characterize their mechanism of action, we performed proteomic analyses of articular chondrocytes treated with exogenous GS and/or CS. Most previous studies have evaluated single proteins, but have not addressed the total chondrocyte proteome. With the introduction of proteomics, it has become possible to simultaneously analyze changes in multiple proteins. Proteomics is a powerful technique for investigating protein expression profiles in biological systems and their modifications in response to stimuli or particular physiological or pathophysiological conditions. It has proven to be a technique of choice for study of modes of drug action, side-effects, toxicity and resistance, and is also a valuable approach for the discovery of new drug targets. These proteomic applications to pharmacological issues have been dubbed pharmacoproteomics [11]. Currently, many proteomic studies use two-dimensional electrophoresis (2-DE) to separate proteins [12]; we have recently used this proteomic approach to describe the cellular proteome of normal and osteoarthritic human chondrocytes in basal conditions [13, 14] and also under IL-1β stimulation [15].

To more clearly define the effects of GS and CS on cartilage biology, we performed proteomic analyses of articular chondrocytes treated with exogenous GS and/or CS. Because the treatment efficacy of these compounds appears to vary with the pathological severity of OA, we used an in vitro model employing normal human chondrocyte cultures stimulated with IL-1β, a proinflammatory cytokine that acts as a mediator to drive the key pathways associated with OA pathogenesis [16].

To assess the influence of GS and CS on the intracellular pathways of human particularcz chondrocytes, we compared five different conditions: cells before treatment (basal), IL-1β-treated cells (control), IL-1β + GS-treated cells, IL-1β + CS-treated cells and IL-1β + GS + CS-treated cells. Two-dimensional electrophoresis (2-DE) gels of each condition were obtained from three healthy donors (a representative image of them is shown in Figure 1). The 15 digitalized images of these gels were analyzed using PDQuest analysis software. The program was able to detect more than 650 protein spots on each gel. The matched spots (540) were analyzed for their differential abundance. After data normalization, 48 protein spots were found to be altered more than 1.5-fold in the GS- and CS-treated samples (both increased and decreased compared to control condition), considering only those with a significance level above 95% by the Student's t-test (P < 0.05). These spots were excised from the gels and analyzed by MALDI-TOF and MALDI-TOF/TOF MS. The resulting protein identifications led to the recognition of 35 spots corresponding to 31 different proteins that were modulated by GS- or CS- treatment. Interestingly, some of these proteins, such as heat shock protein beta-1 (HSPB1) or alpha enolase (ENOA) were present in more than one spot, indicating that they undergo posttranslational modifications, such as glycosylation or phosphorylation. Table 1 summarizes the differentially expressed proteins identified in this proteomic analysis.

Database searches allowed us to classify these 35 proteins according to their subcellular localization and cellular function. Most of them (52%) were predicted to be cytoplasmic, while the remaining 48% were either associated with the cell membrane (20%), extracellular matrix (8%), or located in subcellular organelles, including the endoplasmic reticulum (10%), mitochondria (5%) or nucleus (5%) (Figure 2A). The predicted biological functions for these proteins fell into six major groups: 1) energy production; 2) signal transduction; 3) protein synthesis and folding; 4) redox process and stress response; 5) cellular organization; and 6) metabolism (Figure 2B).

In the present work, we examined the utility of a pharmacoproteomic approach for analyzing the putative intracellular targets of glucosamine (GS) and chondroitin sulphate (CS) in cartilage cells. Using proteomic techniques, we studied the influence of these compounds, both alone and in combination, on the molecular biology of chondrocytes challenged with the proinflammatory cytokine IL-1β.

The conditions used in this study represent supraphysiological levels of both drugs and cytokine. These concentrations, however, are included in the range of in vitro concentrations used by other laboratories, thus facilitating the comparison with other studies [27, 28]. In our work, we chose them according to the bibliography, where a very wide range of both glucosamine and chondroitin sulfate have been used on different cell types and tissues [29, 30]. We tested different concentrations of both drugs in the standardization step of the proteomic analysis (CS from 10 to 200 μg/ml and GS from 1 to 10 mM), and selected the highest concentrations in order to better unravel the molecular mechanisms that are modulated by these compounds. Moreover, in the case of glucosamine it is important to emphasize that its pharmacokinetic is modulated by the levels of glucose in the culture medium, as it utilizes glucose transporters to be taken up by the cells [31, 32]. Since our cells are grown under high levels of glucose (DMEM, containing 25 mmol/l glucose), it is necessary to use high concentrations of glucosamine in order to appreciate its effect in the presence of high glucose. The molecular mechanisms driven with these high amounts of both drugs might not be comparable to their classical oral administration, but they can mimic a direct delivery into the joint. In this sense, it has been recently proposed that intra-articular administration of CS may provide an immediate contact with the synoviocytes and chondrocytes, as is the case in cellular culture models [33]. Furthermore, a recent study performed on cartilage explants shows how cyclic preloading significantly increased tissue PG content and matrix modulus when they are directly supplemented with high concentrations of the combination of GS and CS (500 μg/ml and 250 μg/ml, respectively), resulting in a reduction of matrix damage and cell death following an acute overload [34].

Despite their limitations, several in vitro studies have previously shown how CS and GS could moderate some aspects of the deleterious response of chondrocytes to stimulation with IL-1β. In chondrocyte cultures, GS and CS diminish the IL-1β-mediated increase of metalloproteases, [35, 36] the expression of phospholipase A2 [37, 38] and cyclooxygenase-2, [39] and the concentrations of prostaglandin E₂ [40]. They also reduce the concentration of pro-inflammatory cytokines, such as tumor necrosis factor-α (TNF-α) and IL-1β, in joints, [41] and systemic and joint concentrations of nitric oxide [42] and reactive oxygen species (ROS) [43]. All these studies showed similar results for both molecules, mainly related to their anti-inflammatory effect, while the results obtained by our pharmacoproteomic approach highlight the different molecular mechanisms affected by GS or CS. It is essential to point out that our study has been performed with chondrocytic intracellular extracts. In this context, it is difficult to identify proteins that are known to be secreted by the chondrocytes, such as metalloproteinases, cytokines or aggrecanases, which have been the focus of a recent mRNA-based analysis [44], or hyaluronan synthases, which have been newly found to be increased by CS in synoviocytes [45]. All these were also described to be modulated by GS in a previous transcriptomic study [10]. However, detection of this type of proteins in intracellular fractions by shotgun proteomics is not easily achievable because they are mainly delivered to the extracellular space after their synthesis, being those small amounts that are retained inside the cells masked by other typical cytoplasmic proteins which are more abundant [13]. Given the high dynamic range of proteins in biological systems, this problem is inherent to global screening proteomic experiments, and is only solvable employing hypothesis-driven proteomics strategies (targeted proteomics).

As mentioned before, this study is focused on the investigation of the intracellular mechanisms modulated by CS and GS, which are the background for ulterior putative changes of ECM turnover. In our work, 25% of the proteins modulated by GS are involved in signal transduction pathways, 15% in redox and stress response, and 25% in protein synthesis and folding processes, whereas CS affects mainly energy production (31%) and metabolic pathways (13%) by decreasing the expression levels of 10 proteins (Figure 1B). Bioinformatic analysis using Pathway Studio 6.1 software (Ariadne Genomics, Rockville, MD, USA) enabled the characterization of the biological association networks related to these differentially expressed chondrocytic proteins. A simplified picture of their interactions is showed in Figure 6. Using this analysis, we identified the biochemical pathways that may be altered when chondrocytes are treated with GS and CS.

Most of the proteins modulated by GS belong to the complex homeostatic signalling pathway known as the unfolded protein response (UPR). The UPR system is involved in balancing the load of newly synthesized proteins with the capacity of the ER to facilitate their maturation. Dysfunction of the UPR plays an important role in certain diseases, particularly those involving tissues like cartilage that are dedicated to extracellular protein synthesis. The effect of GS on molecular chaperones and the role of protein disulfide isomerases (PDIs) in the maturation of proteins related with cartilage ECM structure have been described [46]. PDIA3 (GRP58) is a protein on the ER that interacts with the lectin chaperones, calreticulin and calnexin, to modulate the folding of newly synthesized glycoproteins [47], whereas PDIA1 (prolyl 4-hydroxylase subunit beta) constitutes a structural subunit of prolyl 4-hydroxylase, an enzyme that is essential for procollagen maturation [48]. The marked GS-mediated increase of these proteins in chondrocytes points to an elevation in ECM protein synthesis, which might be also hypothesized by the detected increase in Type IV Collagen (COL6A1, essential for chondrocyte anchoring to the pericellular matrix [49]) synthesis caused by GS.

Finally, GS remarkably increases another UPR-related protein, GRP78 (BiP), a fact that we confirmed both at transcript and protein levels. This protein is localized in the ER, and has been previously identified as an RA autoantigen [50], which was subsequently characterized by its anti-inflammatory properties through the stimulation of an anti-inflammatory gene program from human monocytes and the development of T-cells that secrete regulatory cytokines such as IL-10 and IL-4 [51]. In a previous work, we found an increase of this protein in OA chondrocytes, which might be a consequence of heightened cellular stress [14]. A number of previous reports have described the positive modulation of GS on ER proteins, including GRP78 expression [52], but this is the first time that such modulation was found to arise from GS treatment in chondrocytes; thus interestingly suggesting an specific mechanism of action for the putative anti-inflammatory effect of GS in OA.

On the other hand, most proteins modulated by CS are proteins related to metabolism and energy production. It is remarkable that all except one (an enolase isoform) were decreased. In this group, we identified seven out of the 10 enzymes that directly participate in the glycolysis pathway (aldolase, triose phosphate isomerase, glyceraldehyde phosphate dehydrogenase, phosphoglycerate kinase, phosphoglyceromutase, enolase and pyruvate kinase). This suggests that, while IL-1β treatment tends to elevate glycolytic energy production ([15] and our observations), it is then lowered by CS (which reduces five of these enzymes) and by the combination of both drugs (which reduces all seven glycolytic enzymes). The decrease of Neutral alpha-glucosidase AB (or glucosidase II, GANAB), only caused by CS alone (Figure 2), and two other metabolism-related proteins (AK1C2 and UGDH), points also to a reduction of cellular metabolism. GANAB is an ER-enzyme that has profound effects on the early events of glycoprotein metabolism, and has been recently proposed as biomarker for detecting mild human knee osteoarthritis [53].

Interestingly, only four proteins were found to be modulated by GS and CS combination but not by either of the drugs alone, whereas we observed a quantitative synergistic effect of the combination in more than a half (55%) of the altered proteins (Table 1). One of the proteins whose decrease by both drugs alone was significant and furthermore powered by their combination is the redox-related protein SOD2. This protein, the mitochondrial superoxide dismutase, has substantial relevance in stress oxidative pathways and in cytokine-related diseases, such as OA [54]. We found SOD2 to be upregulated by IL-1β ([15] and our observations), and downregulated by GS and CS treatment, both at the transcriptional and protein levels (Real Time-PCR and Western blotting). Supporting our findings, other authors have recently reported the role of GS in counteracting the IL-1β-mediated increase of inducible nitric oxide synthase (iNOS) and the decrease of heme oxygenase, and indicated that the influence of GS and CS on oxidative stress is a possible mechanism of action for its protective effect on chondrocytes [55].