Objective To analyse the influence of mitochondrial DNA haplogroups, as well as the radiographic grade, on serum levels of proteolytic enzymes in patients with osteoarthritis (OA).
Methods Serum levels of metalloproteinase-1 (MMP-1), MMP-3, MMP-13, myeloperoxidase and cathepsin K were analysed in 73 patients with OA and 77 healthy controls carrying the haplogroups J, U and H, by ELISA. Knee and hip radiographs were classified according to Kellgren and Lawrence (K/L) scoring from grade 0 to grade IV. Non-parametric and multiple regression analyses were performed to test the effects of clinical variables, including gender, age, smoking status, diagnosis, haplogroups and radiological K/L grade on serum levels of these enzymes.
Results A significant influence of the haplogroups on the serum levels of MMP-3 and MMP-13 was detected (p=0.027 and p=0.035, respectively). Patients with OA with haplogroup H showed higher serum levels of MMP-3 than healthy controls. Serum levels of MMP-13 were significantly higher in patients with OA (p<0.001), and carriers of the haplogroup J showed lower levels than H carriers. Besides, levels of MMP-13 were proportionally higher in radiological groups B (K/L grade II and III) and C (K/L grade IV) than in group A (K/L grade 0 and I) (p=0.005).
Conclusions This study shows that haplogroups have a significant influence on serum levels of MMP-3 and MMP-13. The influence of the haplogroups on serum levels of MMP-3 is clearly dependent on the diagnosis, whereas the influence of the haplogroups on serum levels of MMP-13 is independent of diagnosis.
Statistics from Altmetric.com
Osteoarthritis (OA), the most common joint disease related to ageing, is characterised by deterioration of articular cartilage leading to joint destruction and severe impairment of mobility.1 As a consequence of its late diagnosis because physical examinations and radiographic studies cannot detect early disease,2 there are few efficacious treatments for OA.
Currently, the best established method for assessing the progression of cartilage loss in OA is plain radiography and arthroscopic evaluation of the joint.3 To identify patients at risk for destructive OA and for monitoring drug efficacy, a non-invasive method that is repeatable and has improved sensitivity over the current diagnostic methods is needed.4 To accomplish this, molecular markers have been developed with the aim of detecting the progression of OA with more reliability and sensitivity, preferably early in the disease process.4 ,5 Because of their greater sensitivity compared with radiographs, several molecular markers for bone, cartilage and synovium have been described as useful for the early identification of OA and of patients at high risk for progression, for monitoring disease progression and for assessing therapeutic response.4 ,6 ,7
Studies have delineated the role of mitochondria in the osteoarthritic process.8,–,10 Restriction fragment length polymorphism studies have revealed a number of stable polymorphic sites in mitochondrial DNA (mtDNA) coding regions that define related groups of mtDNAs called haplogroups.11 ,12 A number of studies have shown a correlation of the mtDNA haplogroups with multifactorial diseases13 and ageing,14 and it is also known that the mtDNA haplogroups modulate the activity of the respiratory complexes of the mitochondrial respiratory chain (MRC).15 Additionally, our group has recently shown that people carrying the mtDNA haplogroup J are at lower risk for developing knee and hip OA, and that patients with OA who have this haplogroup may have less severe progression of the disease, meanwhile OA in patients with the mtDNA haplogroup U is associated with higher severity, based on the Kellgren/Lawrence (K/L) score.16 ,17 These results were reinforced by another study by our group which showed that mtDNA haplogroups modulate the serum levels of classical type II collagen molecular markers, in which carriers of the mtDNA haplogroup H showed higher levels of catabolic markers than J carriers.18
Among the important factors that mediate OA, oxidative stress, which leads to structural and functional cartilage damage like cell death and matrix degradation,19 and deleterious proteinases released from articular chondrocytes, such as the aggrecanases and matrix metalloproteinases (MMP-1, -3 and -13), are noteworthy. Interestingly, mitochondrial dysfunction induced by stimulation of cultured chondrocytes with inhibitors of the MRC, modulates the mRNA expression of these MMPs.20
Immunoassays have been developed for some MMPs, including MMP-1, -3 and -13. An assay has also been developed to evaluate the oxidative stress that occurs in OA by measuring the serum levels of myeloperoxidase (MPO),21 a specific enzyme of neutrophilic azurophilic granules involved in cartilage destruction by degrading extracellular matrix constituents, such as glycosaminoglycans.22
In consideration of this evidence, our study was carried out to evaluate the effects of mtDNA haplogroups, as well as possible interactions between (i) haplogroups and diagnosis and (ii) haplogroups and radiographic K/L grade, on serum levels of MMP-1, MMP-3, MMP-13, MPO and cathepsin K in a population of patients with OA and healthy controls from the north of Spain. For this study we analysed the serum levels of five enzymes in 73 OA subjects and 77 healthy controls carrying the mtDNA haplogroups, J, U or H.
Materials and methods
The population analysed in this study has been described previously.18 A total of 73 unrelated patients diagnosed with knee or hip OA according to the American College of Rheumatology (ACR) criteria.23 (52 women; 21 men; mean age 67.74±8.96 years (range 51–95); 25 carrying haplogroup J, 25 carrying haplogroup U and 23 carrying haplogroup H) were included in this study. Knee and hip radiographs from 148 subjects were classified according to K/L scoring from grade 0 to grade IV.24 Of the 77 subjects who met the inclusion criteria for normal subjects, 25 carried haplogroup J, 25 haplogroup U and 27 haplogroup H. These control subjects included 39 women and 38 men, mean age 66.01±11.88 years (range: 42–94), who did not meet the ACR criteria for knee or hip OA. Details of clinical variables, such as gender, age and smoking status, were also collected for all subjects (table 1). In all cases, informed consent and the agreement of the ethical committee from the Galician Health Administration were obtained.
mtDNA haplogroups genotyping
The samples obtained for the study were haplogroup-typed using a previously described assay.16 For this study, only subjects carrying mtDNA haplogroups J, H or U were included.
Fasting blood samples were collected from each subject in plain tubes containing separation gel. These were allowed to stand for 20 min, then centrifuged for 10 min at 800 g. The serum was then divided into aliquots and stored at −80ºC pending assay.
For this study, five serum proteolytic enzymes were measured: MMP-1 (interstitial collagenase, pro-enzyme), MMP-3 (stromelysin 1, active enzyme), MMP-13 (collagenase 3, active enzyme), MPO and cathepsin K, a cysteine protease that also cleaves the triple helix of type II collagen. Determination of MPO serum levels was performed at the Bone and Cartilage Research Unit of the University of Liege (Belgium), using a previously described assay.21 The remaining biomarkers were measured in our facilities using ELISAs, according to the manufacturer's recommendations. Serum MMP-1 and MMP-3 levels were measured using kits from R&D Systems (Minneapolis, Minnesota, USA), MMP-13 was measured using a kit from Bender MedSystems (Vienna, Austria) and cathepsin K was measured using a kit from Biomedica Medizinprodukte (GhbH & Co KG Vienna, Austria).
Determination of the serum levels of the proteolytic enzymes was performed by simultaneously assaying patients with OA and healthy controls, regardless of the mtDNA haplogroup.
Statistical analyses were performed using SPSS software, release 17 (Chicago, Illinois, USA). In univariate analysis, a non-parametric study was performed using the Mann–Whitney U test to compare serum molecular marker concentrations between patients with OA and healthy controls. The Kruskal–Wallis test was used to compare molecular marker concentrations among the three haplogroups (J, U and H) regardless of the diagnosis. In both cases, the Bonferroni correction for multiple comparisons was applied. Thus, p values were multiplied by the number of outcomes (k=5) tested.
After these preliminary analyses, an analysis of covariance (ANCOVA) was used to evaluate the effects on enzyme serum levels (dependent variable) of haplogroup and diagnosis, adjusting for the confounder effects of gender, smoking status and covariate age. When a significant effect was found either in haplogroups or in diagnosis, the conservative Bonferroni post hoc multiple comparisons test was performed to compare group means. If the interaction between diagnosis and haplogroup was statistically significant main effects should not be interpreted, so differences among the 2×3=6 possible combinations of diagnosis and haplogroups were analysed by using the same Bonferroni post hoc multiple comparisons test. Before the multivariate analysis, a distribution analysis using the Kolmogorov–Smirnov test showed that most of these dependent variables were not normally distributed; therefore, the concentrations were log, square root or inverse transformed to obtain a normal distribution. ANCOVA assumptions were tested based on analysis of the residuals of the adjusted models. Normality of the residuals was tested by normal probability plots and the Shapiro–Wilk test. Homogeneity of variance was tested by scatter plots of the residuals versus fitted and covariate values.
The influence of radiographic grade on the serum levels of the molecular markers was tested using the non-parametric Jonckheere–Terpstra test for ordered groups. For this approach we obtained the radiographic grade of 148 subjects of the 150 who were included in the study. Using K/L scores, subjects were divided into three radiographic groups: group A included 77 healthy controls (K/L grade 0 and grade I), group B consisted of 47 patients with OA with K/L grades II and III, and group C included 24 patients with OA with K/L grade IV.
Non-parametric analysis of serum levels of the molecular markers
We first compared the serum levels of the enzymes assayed between patients with OA and healthy controls. The results showed that levels of both MMP-13 and cathepsin K were significantly increased in patients with OA (p=0.005 and p=0.05, respectively) (figure 1). When we compared the three groups based on the K/L score, MMP-13 showed a significant trend towards proportionally higher serum levels in both groups B (K/L grade II and III) and C (K/L grade IV) than in group A (K/L grade 0 and I) (p=0.005); similarly, median serum levels of cathepsin K were also higher in groups B and C than in group A, though did not reach statistical significance after Bonferroni correction (table 1). When the serum levels of these enzymes were compared among the three haplogroups, regardless of disease status, we found no statistically significant differences, although carriers of haplogroup J showed a non-significant trend towards lower median levels of MMP-13 than non-J carriers (online supplementary table 1). Finally, Spearman's r showed a low positive correlation between MPO and MMP-3 (correlation coefficient 0.233, p=0.0052), and a very weak positive correlation between cathepsin K and MMP-13 (correlation coefficient 0.167, p=0.041). No other significant correlations were found.
Multiple regression analysis
We then performed a multiple regression analysis to assess the effects of the mtDNA haplogroups and clinical variables, including gender, age, smoking status and diagnosis, on the serum levels of the markers. To carry out the analysis, ANCOVA models were adjusted after transformation of dependent variables. ANCOVA assumptions were tested in each case and no violations were found.
The analysis did not reveal statistical differences in mtDNA haplogroups and clinical variables in the serum levels of MMP-1.
A statistically significant interaction between diagnosis and mtDNA haplogroups was found for MMP-3 serum levels (p=0.027) (table 2). Patients with OA with haplogroup H showed higher levels of MMP-3 than both healthy controls carrying the haplogroup H and patients with OA carrying haplogroup J or haplogroup U (p<0.020). On the other hand, healthy controls carrying the mtDNA haplogroup J had significantly higher values than patients with OA with both haplogroups J (p=0.011) and U (p=0.002) (figure 2A, table 3). Additionally, an influence of gender and smoking status emerged from this analysis: men had higher serum levels of MMP-3 than women (p<0.001) and smokers had higher values than non-smokers (p=0.046) (table 2).
Results of the multiple regression analysis (table 2) showed that patients with OA had significantly higher serum levels of MMP-13 than healthy controls (p<0.001). For this proteolytic enzyme, a mtDNA haplogroup effect was detected (p=0.035), showing a trend for people carrying the mtDNA haplogroup J to have lower serum levels of MMP-13 than carriers of the mtDNA haplogroup H (figure 2B and supplementary table 1).
Multiple regression analysis for serum levels of MPO revealed a weak significant interaction between diagnosis and mtDNA haplogroups (p=0.05) (table 2). A subsequent Bonferroni post hoc test showed a non-significant trend towards higher levels of MPO in patients with OA carrying the mtDNA haplogroup H than in healthy controls with the same haplogroup (table 3).
Contrary to the results observed in the non-parametric assay, the multiple regression analysis did not show statistical differences between patients with OA and healthy controls in the serum levels of cathepsin K for mtDNA haplogroups and clinical variables.
In this study we analysed whether the serum levels of several proteolytic enzymes involved in the complex OA process are related to mtDNA haplogroups. Although a number of previous studies have been performed to determine significant proteinases levels in OA joints,25,–,27 this is the first study to correlate mtDNA haplogroups with serum levels of OA-related proteolytic enzymes. Haplogroups J, U and H were chosen for this study as a result of the different relationships that our group had found between haplogroups J and U and the prevalence and severity of knee16 and hip OA17 and because of the association of haplogroups H and J with serum levels of type II collagen molecular markers.18
The results obtained in this work point out the influence of the haplogroups on serum levels of MMP-3 and MMP-13, though in a different manner. The preliminary non-parametric study showed no significant differences between serum levels of MMP-3 in healthy controls and those in patients with OA, contrary to results observed by other authors,28 ,29 but in agreement with other studies.30 Interestingly, significant interactions between haplogroups and diagnosis for this metalloproteinase were detected. Examination of the results obtained in those carriers of haplogroup H shows that patients with OA have higher levels of MMP-3 than healthy controls, in agreement with previous reports.28 ,29 On the contrary, healthy controls carrying the haplogroup J showed significantly higher serum levels of MMP-3 than patients with OA with the same haplogroup (table 3, figure 2A). Because haplogroup H is the most prevalent in European populations (∼45% vs 7–12% for haplogroup J and 12–22% for haplogroup U), it is assumed that previous European studies included more samples carrying the mtDNA haplogroup H. In contrast, in this study the frequency of the three haplogroups was equal (33.3% each) and the possible effects of the mtDNA haplogroup H would be diluted. Taking this into account, our results are in agreement with those studies showing increased serum levels of MMP-3 in patients with OA which included a higher prevalence of H carriers. This would strengthen the influence of the haplogroups and diagnosis in the serum levels of MMP-3. These results also suggest that mtDNA haplogroups could define different OA phenotypes. Finally, our results also showed that men had higher serum levels of MMP-3, as previously described,31 and smokers had higher values than non-smokers.
The results derived from the analysis of serum levels of MMP-13 showed that patients with OA had higher levels than healthy controls, in agreement with previous studies.32 Interestingly, people carrying the mtDNA haplogroup J had lower values than those with the mtDNA haplogroup H, regardless of their diagnosis. This strengthens the proposed protective role of mtDNA haplogroup J in the development of OA since haplogroup J carriers would have less destruction of articular cartilage by MMP-13. This supposition would agree with previous findings obtained by analysis of the mtDNA haplogroups in patients with both knee and hip OA from the north of Spain.16 ,17 One explanation for this interaction might be the incidence of mitochondrial dysfunction in the expression of MMPs,10 ,20 and even the oxidative phosphorylation system (OXPHOS) performance of the different haplogroups,15 ,33 indicating a clear influence of the mitochondria on the role of MMP-13 in the OA process. These results are of special interest since MMP-13 is thought to be the most important of the classical collagenases for the degradation of collagen within cartilage because of its preferential digestion of collagen type II over collagen types I and III.34 To summarise, since no significant interactions between haplogroups and diagnosis were detected, the influence of the haplogroups on the serum levels of MMP-13 is independent of diagnosis, contrary to the results obtained for MMP-3.
Despite the role that MMP-1 is assumed to play in the degradation of articular cartilage,28 ,35 ,36 several studies have shown no differences in the serum levels of MMP-1 between patients with OA and healthy controls,37 ,38 in agreement with our results.
In this work, a poorly significant interaction between mtDNA haplogroups and diagnosis on serum levels of MPO was found. Patients with OA carrying haplogroup H showed higher levels of MPO than healthy controls with the same haplogroup (table 2 and 3). Following the same explanation as for MMP-3 described above, these results would show a similar trend to those obtained by other authors,21 and reflect the existence of an MPO-related oxidative stress in OA chondrocytes. Indeed, median MPO levels were higher in those patients with OA belonging to radiographic group C (K/L grade IV), the most severe form of OA, than in the other groups (table 1). However, owing to the poor statistical significance detected, the results for the serum levels of MPO must be considered with caution.
Finally, little is known about the involvement of cathepsin K in the degradation of the extracellular matrix under physiological and pathological conditions. However, cathepsin K is known to be capable of cleaving collagen type II in articular cartilage.39 Increased levels of the C-terminal neoepitope C2K, generated by cathepsin K cleavage of type II collagen, have been found in OA and aged people in comparison with healthy and younger controls.39 The results obtained from the non-parametric analysis in our study agree with those from the studies cited above. Additionally, a non-significant trend towards proportionally increased median serum levels of cathepsin K in the different radiographic groups was detected (table 1), highlighting the possible role of this proteolytic enzyme as a potential severity biomarker for OA. In contrast, no effect of the mtDNA haplogroup was found for this enzyme. To our knowledge, no relationship between cathepsin K and mitochondrial dysfunction and/or activity has been demonstrated. Cathepsin K cleaves collagen type II at different sites and at different pH levels than the MMPs.39 Therefore, unlike the MMPs, it seems likely that mitochondrial dysfunction does not affect the functionality of cathepsin K, and that the mtDNA haplogroups are not related to the serum levels of cathepsin K.
In summary, the results obtained in this work show a significant influence of the mtDNA haplogroups on serum levels of MMP-3 and MMP-13, with the influence on serum levels of MMP-3 being dependent on diagnosis, whereas the influence on serum levels of MMP-13 is independent of diagnosis. This influence is probably due to differences in the OXPHOS performance,33 which defines different phenotypes of OA that respond to different energy behaviour for carriers of haplogroup H, on the one hand, and for people with haplogroups U and J, on the other, as previously described.15 ,40 Therefore we suggest that the effect of the mtDNA haplogroups on serum levels of these proteolytic enzymes must be considered. However, contrary to the results of our previous work,18 the influence of the radiographic grade and the interactions between radiographic grade and mtDNA haplogroup in the serum levels of these enzymes seem to be not entirely clear.
Thus, no significant differences were found in the K/L grade between men (grade IV: 42.1%) and women (grade IV: 30.8%) (p=0.371) (data not shown).
Despite the interesting results of this work, these results have not been replicated in another cohort of samples. However, we are assembling more patients with OA and healthy controls, regardless of their mtDNA haplogroup, to carry out a new prospective study to validate these preliminary results in a larger cohort of samples.
The authors are grateful to Ms Pilar Cal Purriños for her expert secretarial assistance. The authors express appreciation to Lourdes Sanjurjo and Maria Dolores Velo and to Department of Orthopaedic from the CHU A Coruña for providing cartilage samples.
Funding This study was supported by grants from Secretaria I+D+I Xunta Galicia (PGIDIT06PXIC916175PN); Fundación Española de Reumatologia (programa GEN-SER) and from Fondo Investigación Sanitaria (CIBER- CB06/01/0040)-Spain, Fondo Investigacion Sanitaria-PI 08/2028 Ministerio Ciencia en Innovacion PLE2009-0144, with participation of fundus from FEDER (European Community). IR-P was supported by Contrato de Apoyo a la Investigación-Fondo Investigación Sanitaria (CA06/01102).
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the ethical committee of the Galician Health Administration.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.