Glomerular expression of matrix metalloproteinases in systemic lupus erythematosus in association with activity index and renal function

Abstract Purpose: The aim of this study was to examine the expression of matrix metalloproteinases (MMPs) MMP-1, MMP-2, MMP-3, MMP-9, and their specific tissue inhibitor TIMP-1 in kidney biopsies of patients with lupus nephritis (LN) and to investigate the relationship between MMPs, activity index, and renal function at the time of kidney biopsy. Methods: We performed immunohistochemistry with monoclonal antibodies against MMP-1, MMP-2, MMP-3, MMP-9, and TIMP-1 in 58 kidney-biopsy specimens with LN (according to the 2004 ISN/RPS classification) and eight specimens from normal kidney tissue. We used clinical data of 36 patients at the time of kidney biopsy to evaluate the association between MMPs expression and renal function. Results: We found increased MMP-1, MMP-2, and MMP-3 expression in LN glomeruli and a significant correlation with the activity features, with higher activity index score and worse renal function (p < .001). In particular, we have noticed a significant correlation of MMP-1 with leukocyte influx (OR:16.5 95%CI 4.3–62.5 p < .001), and MMP-3 with glomerular hypercellularity (OR:18.6 95%CI 4.8–72.8 p < .001). Moreover, we found a strong correlation of MMP-2 expression with fibrinoid necrosis and cellular crescents formation (OR:17.1 95%CI 4.3–67.7 p < .001). Conclusions: MMP expression in renal biopsy of patients with LN is increased and directly related to a highly active inflammatory response. Moreover, stronger MMP expression is associated with higher activity index and a more profound renal dysfunction.


Introduction
Lupus nephritis (LN) is a serious manifestation of systemic lupus erythematosus (SLE) that contributes significantly to patient morbidity and mortality [1]. Despite the overall improvement in the care of SLE, the prognosis of LN remains unsatisfactory. To improve the outcome of LN further, newer treatment strategies with better efficacy, along with more accurate diagnostic tools is necessary [2]. A newer 2004 approach of LN has led to the International Society of Nephrology (ISN)/ Renal Pathology Society (RPS) classification. Apart from the ISN/RPS class of LN, features of reversible (active) or irreversible (chronic) damage on biopsy can predict the prognosis of LN patients. Those with a higher activity index or chronicity index score were more likely to progress to ESRD [3]. Glomerular sclerosis and interstitial fibrosis of the renal parenchyma plays a key role in this unfavorable outcome. While increased synthesis of extracellular matrix (ECM) certainly plays an important role, recent studies have focused on the role of degradative systems. The major physiologic regulators of ECM degradation in the glomerulus are matrix metalloproteinases (MMPs) [4,5].
MMPs are a large family of zinc-dependent matrixdegrading enzymes [5] which include interstitial collagenases (MMP-1, -8, and -13), gelatinases (MMP-2 and -9), stromelysins (MMP-3, -7, and -10) and others. Their proteolytic activity mainly involves the degradation of the ECM components. New evidence indicates that beyond ECM remodeling, MMPs play a key role in inflammation, cellular proliferation, and vascular damage. Therefore, an excess of MMPs may contribute to an inflammatory reaction and result in fibroblast activation, thereby leading to an acceleration of renal fibrosis [5,6].
Several lines of evidence indicate an association between MMPs and many kidney disorders. The type of MMP and the pattern of expression in glomerulus depend on the disease [4].
The activity of MMP is normally inhibited by endogenous specific tissue inhibitors of metalloproteinases (TIMPs) [7], which also have an important role in maintaining mesangial homeostasis [8].
The data on the role of MMPs and TIMPs in the pathogenesis and evolution of renal damage in patients with LN remain very limited. The aim of this study was to examine the expression of MMPs and TIMPs in kidney biopsies of patients with LN and to investigate the relationship between MMPs and renal function at the time of kidney biopsy.

Patients
After reviewing the files of the nephrology department and the archive of the renal pathology department, we obtained a total of 64 cases with a diagnosis of LN. Of the above cases, six were excluded because the biopsy material was inappropriate (lack of glomeruli or generally inadequate biopsy material). Thus, 58 cases (11 males [19%] and 47 females [81%], mean age 34.7 years [SD ± 13.14]) were finally obtained with available formalin-fixed paraffin-embedded tissue suitable for immunohistochemistry, from the original diagnostic renal-biopsy specimen. Four patients (6.9%) had class II LN, 22 patients (37.9%) had class III LN, and 32 patients (55.2%) had class IV LN. 27 patients of them, had also features of membranous LN (class V). The diagnosis of LN was based on tissue studies from light and immunofluorescence microscopy according to the 2004 ISN/RPS classification. Although a revision of ISN/ RPS classification has been proposed [9], it is not yet widely applied. However, we have adopted some of the new definitions (e.g., endocapillary hypercellularity instead of endocapillary proliferation). All the biopsies which were performed before 2004 (n ¼ 44), were reassessed and reclassified by two different renal-expert pathologists, using the 2004 ISN/RPS classification. Particularly, the individual features of activity and chronicity were reassessed and recorded in detail. Eight histologically normal kidney specimens obtained from eight patients who underwent nephrectomy because of renal carcinoma or renal trauma were used as normal control tissues.

Antibodies
The immunohistochemical staining was performed on 2.5 lm sections using purified rabbit polyclonal antibodies to human MMP-1, MMP-2, MMP-3, MMP-9, and TIMP-1 which were purchased from Zytomed systems GmbH (Anhaltinerstr. 22, Berlin 14163, Germany). To examine the degree of inflammation, tissue sections were also stained with an antibody against CD68 antigen (DAKO, Carpinteria, CA, USA) as a marker of macrophages.

Grading of MMPs and TIMP-1 immunostains
All of the glomeruli (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) in each section were examined by light microscopy. All the immunostains for MMPs and TIMP-1 were scored without knowledge of the initial histological diagnosis, regarding the ISN/RPS class and the activity and chronicity index or patient's clinical features. Immunostains were graded on the basis of the intensity of brown staining of the kidney tissue on a scale of 0-3þ. A grade of 0-1þ corresponded to absent or weak staining, respectively. A grade of 2þ or 3þ corresponded to increased or significantly increased staining, respectively. For the purposes of comparison, grades 0-1þ were defined as negative and grades 2þ-3þ were defined as positive. To assess the variability between MMP grades determined with this technique, grading of the same biopsy samples was repeated on two different days.

Statistical analysis
Data were expressed as mean ± SD for quantitative variables and as percentages for qualitative variables. The Kolmogorov-Smirnov test was utilized for normality analysis of the quantitative variables. Bivariate analyses were made by using the independent samples t-test or Mann-Whitney test in case of violation of normality and Chi-square test, Fisher's exact test to analyze the relation between the variables of interest (MMPs) (normal [0-1] vs. abnormal [2][3] for MMP1, MMP2, MMP3, TIMP1, and CD68) and the quantitative and qualitative outcome variables (activity and chronicity index and renal function) respectively. Multifactorial logistic and linear regression models were used to identify independent predictors (MMP1, MMP2, MMP3, TIMP1, and CD68) of the qualitative and quantitative outcome variables. All assumptions of linear regression analysis (homoscedasticity, linearity, normality, and independence of error terms, as well as multicollinearity of independent variables) were examined. All tests were two-sided, a p value <.05 was used to denote statistical significance. All analyses were carried out using the statistical package SPSS version 17.00 (Statistical Package for the Social Sciences, SPSS Inc., Chicago, IL).

Results
On staining for MMP-9, we found no difference between LN and normal glomeruli. We observed a distinct staining pattern for MMP-9, which was consistently expressed in the glomerular basement membrane (GBM), in both normal and LN kidneys. Regarding the TIMP-1, generally we found no significant expression in both normal and LN glomeruli. In some LN glomeruli with stronger expression of TIMP-1, we were unable to show any correlation with activity or chronicity features (Figure 1(4a,4b)).

MMP-1
We found that MMP-1 was strongly expressed in LN glomeruli. Moreover, we found a significant correlation with all activity features ( Figure 1(1a,1b)).

MMP-3
MMP-3 was also strongly expressed in LN glomeruli. In addition, we found a significant correlation with all activity features except for subendothelial hyaline deposits (  p=.018]). Similarly, strong MMP-3 expression in LN glomeruli was correlated with higher activity index compared with the glomeruli with absent or weak staining (activity index 9.67 ± 2.78 vs. 3.96 ± 2.75, p < .001) ( Table 3 and Figure 1(3a,3b)).
Conversely, we found no significant correlation between any of MMP-1, MMP-2, or MMP-3 expression with any of the glomerular lesions that define chronicity (glomerular sclerosis, fibrous crescents, tubular atrophy, and interstitial fibrosis).

CD68
Further immunohistochemical analysis showed an increased number of CD68þ cells in cases with higher activity index score (p < .001) (( Figure 1(5a,5b)).   It represents the presence of CD68þ cells and is particularly associated with higher endocapillary hypercellularity and more intense leukocyte infiltration. PGM-1 staining was not correlated with chronicity indices. N: number of patients, or: odds ratio.

Correlation of MMP expression with renal function and other clinical parameters
Unfortunately, because of the inadequacy of laboratory data and clinical information in the older patients' records, we were able to obtain laboratory data only from 36 patients (

Discussion
Although important progress has been made regarding the pathogenesis of SLE and LN over the past years, several points about the pathogenic processes that underlie the progressive decay in renal function in LN remain obscure. There are some studies supporting an altered glomerular MMP activity in glomerulopathies and particularly in LN [10]. The role of MMPs is quite different in noninflammatory glomerulopathies and glomerulonephritis. In general terms, a downregulation of MMPs has been associated with progression of noninflammatory diseases such as hypertensive glomerulosclerosis and diabetic nephropathy [4]. On the contrary, in inflammatory glomerular diseases, the increased levels of MMP are generally associated with disease activity and influx of inflammatory cells, and both the level and the duration of MMP elevation determine the extent of glomerular damage [11]. We explored this possibility by studying the expression of MMPs (MMP-1, MMP-2, MMP-3, and MMP-9) in the glomeruli of LN kidneys. In addition, given the predictive value of activity and chronicity index regarding the prognosis of LN patients, we studied the correlation of these MMPs with the activity and chronicity features.
Regarding the MMP-9, although there is some evidence supporting that MMP-9 may be involved in the pathogenesis of human glomerulonephritis [10], we did not find any difference in MMP-9 expression between normal and LN glomeruli. MMP-9 was consistently expressed in the glomerular basement membrane in Table 5. Clinical and parameters of patients (N ¼ 36) in comparison with MMP-1, MMP-2, and MMP-3 expression, at the time of renal biopsy. Strong expression of MMP-1, MMP-2, and MMP-3 was correlated with worse renal function (eGFR) at the time of kidney biopsy. All the patients of the study where white Caucasians (Greeks) and all the biopsies came from patients who were identified with LN for first time. None of the patients was given immunosuppressive therapy with cyclophosphamide or mycophenolate acid prior to biopsy. Only nine patients were given corticosteroids and/or hydroxychloroquine before the biopsy, because of the extrarenal manifestations of SLE, but it was not correlated with the intensity of the MMP expression.
both normal and LN kidneys. This result might suggest that MMP-9 plays a physiological role in basement membrane turnover. However, a significant expression of MMP-1, MMP-2, and MMP-3 in LN glomeruli emerged in our study. Moreover, we found a strong correlation of MMP-1, MMP-2, and MMP-3 expression with the various activity features of LN and a higher activity index score.
Regarding MMP-3, the results are quite similar. In particular, we have noticed a strong correlation with glomerular hypercellularity (p < .001).
These findings suggest that high MMP-1 and MMP-3 expression in LN glomeruli is more compatible with an exudative form of glomerulonephritis in which leukocyte accumulation lead to endothelial cell injury and capillary wall destruction. MMP-1 and MMP-3 may be secreted by activated infiltrating cells (monocytes/macrophages and neutrophils) [12] through activation of the Toll-like receptor 9 signaling pathway [13,14]. This is in accordance with our finding of increased number of CD68þ cells within the glomeruli with higher activity index score (hypercellularity and leukocyte influx) and strong MMP expression.
It is known that MMP-1 is involved in the breakdown of the main collagens (types I-III) of the interstitial connective tissue as well as laminin and tenascin [15][16][17] and MMP-3 have a broad substrate specificity including collagen IV, proteoglycans, fibronectin, and laminin [17,18]. Since collagen IV, laminin, and fibronectin are the main constituents of mesangial matrix and glomerular basement membrane, it appears that MMP-1 and MMP-3 are capable of mediating changes in glomerular basement membrane by cleavage of particular collagen fragments and lead to the exposure of highly immunoreactive epitopes, increasing this way the production of autoantibodies against matrix structures [19,20].
The gelatinase MMP-2 is also involved in the catabolism of collagen IV and laminin, the most important components of mesangial matrix and glomerular basement membrane. Strong MMP-2 expression showed a particular correlation with fibrinoid necrosis (p < .001) and the presence of cellular crescents (OR:17.1 95%CI 4.3-67.7 and p < .001). Among the patients with strong MMP-2 staining (n ¼ 22), 15 patients (68.2%) had fibrinoid necrosis and cellular crescents, while only seven patients (31.8%) had no fibrinoid necrosis or cellular crescents.
It is known that besides ECM degradation, MMPs play a key role in inflammation and cellular proliferation and may also influence ECM turnover via the regulation of certain growth factors. Thus, our finding may suggest that MMP-2 could play an important role in crescents formation. Moreover, MMP-2 expression could contribute to prediction of renal outcome in crescenting LN, as the presence of cellular crescents and fibrinoid necrosis are associated with a worse renal prognosis. Triantafyllopoulou et al. found that macrophages directly contribute to the elevated MMP-2 expression in LN [21].
Increased expression of the MMPs is often found to be accompanied by a compensatory increase in the levels of one or more of the TIMPs. The net result of these opposing stimuli could determine the final effect on the ECM turnover. There is also additional data in experimental nephritis showing that MMP inhibition attenuates glomerular lesions [22], suggesting a role for the MMP inhibitors in the treatment of acute glomerulonephritis [22][23][24]. Regarding TIMP-1, although there is evidence supporting increased plasma and urinary concentrations in patients with renal disease, the data are inadequate. In our study, although we found a stronger expression of TIMP-1 in some LN glomeruli, we were unable to determine any significant correlation with the various features of disease activity. This could be due to the fact that TIMP-1 is the main inhibitor of MMP-9, which was equally expressed in normal and LN glomeruli in our study.
In addition to correlating the expression of MMPs with the activity index of LN, the association of stronger MMP-1, MMP-2, and MMP-3 expression with worse renal function at the time of kidney biopsy, indicates that these proteolytic enzymes play a pivotal role in mesangial matrix and glomerular basement membrane damage, thus, contributing to the impairment of renal function.
Our findings are in accordance with previous studies showing increased levels of MMP-1, MMP-2, and MMP-3 on mice with experimental LN. Data from human studies also showed that elevated MMP levels are associated with a highly active inflammatory response associated with glomerular damage [10]. To our knowledge, our study is the first study in LN including such a large number of biopsy samples, and also the first study in LN correlating the MMP expression with the various activity features and renal function.
There is evidence that the MMP activity may be different even in the different stages of the same renal disease. MMPs may contribute to GBM degradation in the early stages, while they play a pivotal role in the removal of ECM associated with scarring and fibrosis at later stages [25]. In our study, we found no particular correlation between the expression of MMP-1, MMP-2, MMP-3, and MMP-9 and the various classes of LN. In addition, the lack of any significant correlation between MMP activity and the chronicity features, could mean that although MMPs play a certain role in cell proliferation and cell infiltration, their role in the EC degradation and removal in the advanced stages of LN is limited.
To successfully manage LN, it is important to be able to accurately distinguish the onset of renal disease, to determine the disease activity and finally to distinguish active nephritis from chronic kidney damage. The identification of LN biomarkers could help in accurate assessment of disease activity and severity and enable the physician to individualize the therapy [26]. Although some studies reported some biomarkers that correlate with the histological findings in LN [27], no reliable biomarker which could represent a specific histological lesion and with a clinical value has been emerged yet [26]. In this manner, our study could provide additional data regarding the MMPs expression in LN and their significance as a potential histologic biomarker with diagnostic and prognostic value.
In conclusion, MMP expression in renal biopsy of patients with LN is increased and directly related to a highly active inflammatory response. Moreover, stronger MMP expression is associated with higher activity index and a more profound renal dysfunction at the time of renal biopsy. Particularly, strong MMP-2 expression appears to be highly correlated with the presence of cellular crescents and fibrinoid necrosis. Our findings may contribute to the interpretation of the pathogenesis of renal damage in LN and could provide a potential marker for the activity of the disease.

Limitations of the study
This is a cross-sectional study regarding the expression of MMPs in LN and their correlation with the activity and chronicity index and renal function at the time of renal biopsy. Many biopsy specimens were retrieved from the archive of the renal pathology department and concerned patients who underwent a renal biopsy many years ago. So unfortunately, information regarding their subsequent course was not sufficient. A prospective study is required to confirm the predictive value of MMPs (besides their correlation with a higher activity index and a worse renal function at the time of kidney biopsy) in the outcome of renal function.

Disclosure statement
No potential conflict of interest was reported by the authors.