Renal expression of matrix metalloproteinases in human ANCA-associated glomerulonephritis

Background. Expression of matrix metalloproteinases (MMPs) by inﬁltrating and intrinsic renal cells is increased in inﬂammatory conditions, and may correlate with disease activity of glomerulonephritis. We analysed renal expression of MMPs, tissue inhibitor of metalloproteinase-1 (TIMP-1) and markers of neutrophil and monocyte inﬁltration in renal biopsies of patients with active anti-neutrophil cytoplasmic antibody (ANCA)-associated glomerulonephritis. Methods. Immunohistochemical expression of MMP-2, -3, -9, TIMP-1, the neutrophil- and monocyte-derived MMP activators cathepsin G, neutrophil elastase and myeloperoxidase (MPO), and the monocyte marker CD14 was determined in renal biopsies of active proteinase 3 (PR3)-ANCA ( n ¼ 7) and MPO-ANCA ( n ¼ 6) associated glomerulonephritis, and in normal renal tissue ( n ¼ 4). Double labelling experiments of MMPs and TIMP-1 were performed with MPO and CD68, labelling neutrophils and macrophages. Results. and TIMP-1-positive cells were detected in ANCA-associated glomerulonephritis in glomeruli with active inﬂammation (cellular crescents or ﬁbrinoid necrosis), only occasionally in normal appearing glomeruli, and not in sclerotic glomeruli and positive cells were found in the tubulo-interstitium. MMPs and TIMP-1 were expressed predominantly by MPO-and CD68-positive cells. In normal renal tissue, no expression was detected, with the exception of weak mesangial staining for MMP-2. In ANCA-associated glomerulonephritis, glomerular MMP-2, -9 and TIMP-1 correlated with glomerular cathepsin G expression, while the number of MMP-9-expressing cells per glomerulus correlated with the percentage of crescentic glomeruli. Tubulo-interstitial expression of MMPs correlated with all markers of neutrophil and monocyte inﬁltration, and interstitial MMP-9 and TIMP-1 expression correlated with renal function at the time of renal biopsy. Conclusions. Expression of glomerular and interstitial MMP-2, -3, -9 and TIMP-1 is increased in active ANCA-associated glomerulonephritis and correlates with inﬂammatory activity. and MPO-ANCA-positive patients. Using this test, renal expression of MMPs, TIMP-1 and markers of neutrophil and monocyte inﬁltration were compared. Correlations were tested using Spearman’s correlation coefﬁcient. Statistical signiﬁcance was deﬁned as a two-sided P <0.05.


Introduction
Anti-neutrophil cytoplasmic antibodies (ANCAs) directed against proteinase 3 (PR3) and myeloperoxidase (MPO) are the hallmark of a group of autoimmune small-vessel vasculitic disorders, such as Wegener's granulomatosis, microscopic polyangiitis and renal limited vasculitis [1,2]. Many patients with these vasculitic disorders develop renal involvement with necrotizing crescentic glomerulonephritis, characterized by fibrinoid capillary necrosis and cellular crescents. Subsequently, these cellular, inflammatory crescents evolve and become acellular and fibrotic, leading to permanent loss of functional glomerular volume. The development of crescents is the result of interaction between infiltrating leukocytes, local proliferating cells and changes in the extracellular matrix.
Matrix metalloproteinases (MMPs), the main matrix-degrading enzymes, have been implicated in both the inflammatory and fibrotic phase of crescent formation and are upregulated during inflammation and physiological remodelling processes [3]. MMP gene expression and subsequent production as pro-enzymes is tightly regulated by various controlling mechanisms [4]. The extracellular activation of pro-MMPs is predominantly a two-step process with initial cleavage by a protease, followed by a final cleavage, usually by another MMP [5,6]. Specialized tissue inhibitors of matrix metalloproteinases (TIMPs) and various other molecules, including a 2 -macroglobulin, are able to inhibit activated MMPs [7]. During inflammation, activated MMPs degrade the extracellular matrix hence promoting trans-basement membrane migration of leukocytes and release of pro-inflammatory products from the extracellular matrix. Thus MMPs are likely to have an important pathogenic role in the acute phase of human crescentic glomerulonephritis. To study whether MMPs reflect renal inflammation in active renal ANCA-associated vasculitis, we determined intrarenal expression of MMPs, their activators and a specific MMP inhibitor, TIMP-1, and correlated this with histological and functional markers of renal disease.

Renal biopsy specimens
Renal biopsy specimens were studied from 13 patients with active renal ANCA-associated vasculitis. Samples of normal kidney tissue from patients with renal cell carcinoma (n ¼ 4) were used as control specimens. Blinded to the clinical status and renal function of the patient at the time of biopsy, renal biopsies were evaluated by light microscopy. Normal, sclerosed, crescentic glomeruli and those with fibrinoid necrosis were counted. Results were expressed as a percentage of the total number of glomeruli. Interstitial fibrosis, tubulo-interstitial inflammation and atrophy were scored semi-quantitatively from 0 to 4 (0 ¼ no interstitial fibrosis/ inflammation/atrophy; 1 ¼ <10%; 2 ¼ 10-25%; 3 ¼ 25-50%; 4 ¼ >50%). Age, ANCA status, serum creatinine, proteinuria and C-reactive protein (CRP) were recorded from all patients at the time of biopsy. Additionally, data on creatinine clearance at biopsy, and 1, 3 and 12 months after renal biopsy were collected.

Immunohistochemistry
Frozen 4 mm sections were fixed at room temperature (for 10 min in 100% acetone) and washed for 5 min in phosphatebuffered saline (PBS; pH 7.4). The sections were incubated for 1 h at room temperature with the primary antibody diluted in 1% bovine serum albumin (BSA) in PBS (the source, specificity and dilution of the primary antibodies are given in Table 1). Next, the slides were rinsed with PBS for 5 min. Endogenous peroxidase activity was blocked by incubating the slides in 0.075% hydrogen peroxide/PBS for 30 min. Following three washes with PBS, the slides were incubated for 30 min with peroxidase-conjugated secondary antibodies (all from Dako, Glosstrup, Denmark; specific combinations are given in Table 1), supplemented with 1% human AB serum. The sections were washed in PBS and peroxidase activity was developed in a freshly prepared anti-rabbit antibody, RaG Po ¼ peroxidase-conjugated rabbit anti-goat antibody, GaM-IgG3 Po ¼ peroxidase-conjugated goat anti-mouse IgG3 antibody, GaM-IgG2a Po ¼ peroxidase-conjugated goat anti-mouse IgG2a antibody, SaR Po ¼ peroxidase conjugated swine anti-rabbit antibody, GaM-IgG1 FITC ¼ FITC-labelled goat anti-mouse IgG1 antibody.
Additional double labelling experiments were performed on cryostat sections. Briefly, after blocking of endogenous peroxidase, sections were incubated with primary antibodies, monoclonal mouse anti-human CD68, polyclonal rabbit antihuman MPO and mouse anti-human a-smooth muscle actin (a-SMA), respectively. In contrast to CD14, which predominantly labels monocytes, CD68 is a pan-macrophage marker.

Statistical analysis
The non-parametric Mann-Whitney test was used to compare clinical and histological data between PR3-and MPO-ANCA-positive patients. Using this test, renal expression of MMPs, TIMP-1 and markers of neutrophil and monocyte infiltration were compared. Correlations were tested using Spearman's correlation coefficient. Statistical significance was defined as a two-sided P<0.05.

Clinical and histological characteristics
Clinical and histological features of the included patients are shown in Table 2   0.06-0.78)] (P ¼ 0.051). Tubulo-interstitial inflammation did not differ between the two groups.
Immunolocalization of MMP-2, -3, -9 and TIMP-1 in normal and nephritic glomeruli Immunohistochemical staining in renal biopsies of patients with ANCA-associated glomerulonephritis revealed MMP-2-and -3-positive cells in cellular cresents. In addition, MMP-2 and -3 were detected in interstitial, glomerular and periglomerular cells (Figure 1). MMP-9-positive cells, in contrast to MMP-2 and MMP-3, were only present in glomeruli with active crescents, and in the interstitium. In control specimens, only weak MMP-2 staining was found in some mesangial cells, whereas MMP-3 and MMP-9 were absent. TIMP-1 protein was expressed in glomeruli with active crescents and in smooth muscle cells of larger arteries. The latter was also seen in control kidneys (Figure 1). Cathepsin G, neutrophil elastase and MPO were found in active crescents and were also present in interstitial inflammatory cells. Cathepsin G, neutrophil elastase and MPO were absent in control kidneys ( Figure 1). In fibrous crescents, staining for MMPs, TIMP-1, cathepsin G, neutrophil elastase and MPO was virtually absent. Double labelling experiments were performed to clarify which cells produced MMP-2, MMP-9 and TIMP-1 in the crescentic glomeruli. MMP-9-and TIMP-1-expressing cells were predominantly MPOpositive, while only a fraction of MMP-2 expressing cells were MPO-positive ( Figure 2). Additionally, some CD68-positive macrophages produced MMP-2, MMP-9 and TIMP-1 (Figure 2). Part of the MMP-2positive cells were neither MPO-nor CD68-positive, and these cells had the morphological characteristics of glomerular mesangial cells (not shown). In additional double labelling experiments, a substantial part of the MMP-2-positive cells were found to be a-SMA-positive (Figure 2).
In general, expression of all enzymes tended to be higher in PR3-as compared with MPO-ANCAassociated cases. The differences did, however, only reach statistical significance for glomerular neutrophil elastase and MMP-9 ( Table 3). The number of positive cells per glomerulus and the interstitial scores for MMP-2, -3, -9 and TIMP-1 are shown in Table 3 for controls and patients, categorized according to their ANCA specificity. . MMP-2 was weakly stained in the glomerular mesangium from control kidneys (C). Increased expression was observed in glomerular crescents (outlined), interstitial and glomerular cells from patients with necrotizing glomerulonephritis (D). MMP-3 and MMP-9 were absent in control kidneys (E and G) and markedly upregulated in glomerular cresents (arrows) from patients with necrotizing glomerulonephritis (F and H). TIMP-1 was expressed in smooth muscle cells (arrows), but not in glomeruli and in the interstitium from control kidneys (I), whereas TIMP-1 was expressed in glomeruli (arrows) and in the interstitium from patients with necrotizing glomerulonephritis (J).

Matrix metalloproteinases in ANCA-associated glomerulonephritis 1415
Correlation between MMPs and TIMP-1 vs cathepsin G, neutrophil elastase, MPO and CD14 The relationship between expression of MMPs and their neutrophil-derived activators, cathepsin G, neutrophil elastase and MPO, was studied. The interstitial expression of MMP-3 and MMP-9 correlated significantly (P<0.01) with CD14 as marker of monocyte infiltration and all neutrophil-derived proteinases ( Table 4). The interstitial expression of MMP-2 correlated with MPO (P<0.01), cathepsin G (P<0.05) and CD14 (P<0.01), but not with the number of cells expressing neutrophil elastase. In the glomeruli, the number of MMP-expressing cells did not correlate with neutrophil elastase and MPO (see Table 4), CD14 was not detected in the glomeruli. Only MMP-2 correlated significantly (P<0.05) with the glomerular expression of cathepsin G. In contrast, the number of TIMP-1-expressing cells in the glomeruli correlated with all markers of neutrophil infiltration in the glomerulus (P<0.01).

Correlation between MMPs, TIMP-1 and renal histology
To study whether expression of MMPs and TIMP-1 reflected active inflammation, correlation with renal histology was tested. The number of MMP-9-positive cells in the glomeruli correlated significantly (r ¼ 0.60; P<0.05) with the percentage of crescentic glomeruli. In PR3-ANCA-positive patients, the number of crescentic glomeruli tended to be higher (P ¼ 0.051), and significantly more MMP-9-positive cells were present (P ¼ 0.007) than in MPO-ANCA-positive patients. There were no correlations between numbers of positive cells for MMP-2, -3 and TIMP-1 in the glomeruli and glomerular damage, as reflected by the percentage of normal, sclerosed and crescentic glomeruli, and glomeruli with fibrinoid necrosis. The interstitial expression of MMP-2, -3, -9 and TIMP-1 did not correlate with the histologically classified grades of interstitial fibrosis, and tubular atrophy (data not shown). In contrast, interstitial expression of MMP-9 (r ¼ 0.87, P<0.0001) and TIMP-1 (r ¼ 0.66, P<0.05) correlated significantly with tubulo-interstitial inflammation, whereas interstitial expression of MMP-2 (r ¼ 0.43, P ¼ 0.17) and MMP-3 (r ¼ 0.57, P ¼ 0.06) did not.

Discussion
In the present study, MMP-2, MMP-3, MMP-9 and TIMP-1 were detected in glomeruli and tubulointerstitium in renal biopsies of patients with active ANCA-associated necrotizing crescentic glomerulonephritis, while expression was virtually absent in controls. In addition, expression of MMPs and TIMP-1 was found mainly in glomeruli with cellular Table 3. Glomerular and tubulo-interstitial expression of MMP-2, MMP-3, MMP-9, TIMP-1, cathepsin G, MPO and neutrophil elastase in renal biopsies of 13 patients with active ANCA-associated glomerulonephritis   Matrix metalloproteinases in ANCA-associated glomerulonephritis 1417 crescents and/or fibrinoid necrosis, but not in unaffected or sclerotic glomeruli, and expression correlated with markers of neutrophil and monocyte infiltration. Also, tubulo-interstitial expression of MMPs and TIMP-1 correlated with markers of neutrophil and monocyte infiltration, and correlated with renal function at the moment of renal biopsy. These findings suggest that expression of these MMPs and TIMP-1 represents active inflammation in ANCA-associated necrotizing crescentic glomerulonephritis and can potentially be useful in assessing the activity of the inflammatory process. Double staining experiments showed that MPO-positive infiltrated neutrophils produced MMP-2, MMP-9 and TIMP-1; in contrast, fewer CD68-positive macrophages were found to produce MMPs and TIMP-1. These results correspond to the findings of Urushihara et al. who also showed that many neutrophils, but not macrophages, produced MMP-9 in various glomerulonephritides [8].
The MMPs have been firmly linked to inflammation. Pro-inflammatory cytokines such as interleukin-1 and tumour necrosis factor-a have been shown to induce transcription and expression of several MMPs in both inflammatory cells such as neutrophils, monocytes and intrinsic renal cells, while the anti-inflammatory and profibrotic cytokine transforming growth factor-b (TGF-b) is an important downregulator of MMP expression and enhances expression of MMP inhibitors such as TIMP-1 [4,9,10]. In addition, an essential step in the activation of MMPs, which are secreted as inactive pro-enzymes, is the cleavage by proteases derived from activated inflammatory cells, especially neutrophils and monocytes. In experimental glomerulonephritis, increased expression of MMPs is associated with disease activity and infiltration by inflammatory cells. Recently, gene expression of MMP-2 and the membrane-bound MT1-MMP, which is an important activator of MMPs, was shown to correlate in different stages with inflammation and renal damage in a longitudinal murine anti-glomerular basement membrane (GBM) glomerulonephritis model [11]. Moreover, in another model, treatment with an MMP inhibitor targeting elevated levels of MMP-2 and MMP-9 expression in anti-Thy1.1 nephritis diminished histological damage and proteinuria [12]. However, in an accelerated model of anti-GBM nephritis, MMP-9deficient mice showed more severe disease than control mice [13]. This suggests that MMP-9 not only causes glomerular damage, but might also protect glomeruli against injury, or promote glomerular recovery, after initial damage.
In human renal disease, increased glomerular expression of MMP-9, but not of MMP-2, has been reported in IgA nephropathy, mesangial proliferative glomerulonephritis and lupus nephritis [8]. In addition, elevated plasma levels of MMP-2, MMP-3, MMP-9 and TIMP-1 have been found in some of these diseases, while elevated urinary levels of MMP-2 and MMP-9 have been found in diabetic nephropathy [14][15][16]. In human rheumatoid arthritis, elevated plasma levels of different MMPs, most often MMP-3, have been found and correlated with disease activity [17]. Data on expression of MMPs and TIMP-1 in human crescentic glomerulonephritis have, to our knowledge, not been published. As our study involved only a limited number of patients, all with active glomerulonephritis at the time of renal biopsy, but none with previous, but currently inactive, renal disease, we cannot be certain that expression is not increased in the latter situation. However, expression of MMPs and TIMP-1 clearly colocalized with histological markers of active glomerulonephritis, i.e. in crescentic glomeruli and within the tubulo-interstitial infiltrates, while normal or sclerotic glomeruli were negative.
As has been suggested by others based on morphological data, our data, despite the limited number of biopsies, do suggest a difference in inflammation according to ANCA antigenic specificity [18]. In line with these morphological data, higher expression of inflammation-associated proteins in PR3-as compared with MPO-ANCA-associated glomerulonephritis was found, although only for glomerular neutrophil elastase and MMP-9 did this difference reach statistical significance. Since our data suggest that expression of MMPs and TIMP-1 is limited to active lesions of ANCA-associated glomerulonephritis and may reflect inflammatory activity, determination of plasma or urinary levels may be of potential value in assessing renal disease activity in these diseases and should be studied. Whether expression of MMPs or TIMPs in the acute or recovery stage of active ANCA-associated glomerulonephritis or other forms of glomerulonephritis may also have prognostic significance on long-term follow up has to be studied. Given the important role of TGF-b, and probably other pro-fibrotic cytokines such as CTGF, in regulating expression and activation of MMPs and TIMPs, longitudinal assessment may be of value in predicting long-term outcome in glomerulonephritis and other progressive renal diseases. Increased serum and urinary levels of TIMP-1 and the matrix protein tenascin have been described in patients with renal function impairment due to different diseases, but have not been correlated with renal inflammation and fibrosis or changes of renal function during follow-up [19]. Likewise, cross-sectional data in patients with a kidney transplant have correlated renal TIMP-1 mRNA expression with interstitial fibrosis 6 months post-transplant, but not with serum and urinary levels of TIMP-1 or renal function during follow-up [8,20].
In conclusion, increased renal MMP-2, -3, -9 and TIMP-1 expression reflected the inflammatory process in ANCA-associated glomerulonephritis, in particular MMP-9 expression correlated with active glomerular lesions and tubulo-interstitial inflammation. Establishing a link between the increased inflammation-related renal expression and elevated serum or urinary levels of MMPs and TIMP-1 will be the first step required to test whether monitoring of MMP and TIMP-1 levels has potential value as a non-invasive marker of renal disease activity in ANCA-associated and other forms of glomerulonephritis.