Elevated macrophage migration inhibitory factor (MIF) levels in the urine of patients with focal glomerular sclerosis

Abstract

The pathogenesis of focal glomerular sclerosis (FGS) is poorly understood. Macrophage migration inhibitory factor (MIF) is a potent pro-inflammatory cytokine released from T cells and macrophages, and is a key molecule in inflammation. To examine further the possible role of MIF in FGS, we measured MIF levels in the urine. The purpose of the present study was to evaluate the involvement of MIF in FGS. Urine samples were obtained from 20 FGS patients. The disease controls included 40 patients with minimal-change nephrotic syndrome (MCNS) and membranous nephropathy (MN). A group of healthy subjects also served as controls. Biopsies were performed in all patients prior to entry to the study. The samples were assayed for MIF protein by a sandwich enzyme-linked immunosorbent assay (ELISA). The levels of MIF in the urine of FGS patients were significantly higher than those of the normal controls and patients with MCNS and MN. In contrast, the levels of urinary MIF (uMIF) in patients with MCNS and MN did not differ significantly from normal values. In the present study, attention also focused on the relationship between uMIF levels and pathological features. Among the patients with FGS, uMIF levels were significantly correlated with the grade of mesangial matrix increase and that of interstitial fibrosis. There was also a significant correlation between uMIF levels and the number of both intraglomerular and interstitial macrophages. Although the underlying mechanisms remain to be determined, our study presents evidence that urinary excretion of MIF is increased in FGS patients with active renal lesions.

Introduction

Macrophage migration inhibitory factor (MIF) is a cytokine that plays a critical role in the regulation of macrophage effector functions and T cell activation [1–3]. MIF has also been demonstrated to be a crucial cofactor in T cell activation [1]. MIF may function as an autocrine growth factor in certain cell systems [1–4]. MIF induces both systemic and local acute-phase responses, including the secretion of proinflammatory cytokines and growth factors [4,5]. Recent evidence suggests that mesangial cells can produce MIF in vivo and in vitro[6]. The pathological importance of excess MIF is evident in rat experimental glomerulonephritis (GN); the number of MIF-secreting cells is locally increased in the diseased glomeruli of rat anti-Thy1 nephritis [6]. In the kidney, MIF is produced not only by mesangial cells (and infiltrating macrophages) but also by glomerular epithelial cells and glomerular endothelial cells [7] and by tubular cells [8,9]. The comparative overproduction of MIF in rat crescentic GN [7] further substantiates the postulation that MIF is a key upstream regulator of the glomerular inflammation. These findings indicate an important role for MIF-mediated immune responses in inflammatory conditions of the glomeruli.

The pathogenesis of focal glomerular sclerosis (FGS) is still unknown. Its aetiology is complex and seems to be multifactorial. There is increasing evidence that cytokines play a critical role in the development and perpetuation of FGS [10]. The available data suggest that a disturbed balance of pro- and anti-inflammatory cytokines is important mechanism of chronic inflammation in FGS [11–13]. Although several lines of research have implicated MIF in the pathogenesis of experimental GN [6,14], very little is known of the role that MIF may play in the pathological complication of human FGS. As far as we aware there has no previous attempt to measure urinary MIF (uMIF) in a patient population of adults with nephrotic syndrome (NS). We designed this study to expand and confirm our early observations suggesting that patients with some types of NS had high incidence of significant MIF to various antigens [15].

To examine further the possible role of MIF in FGS, in the present study we measured MIF levels in urine in the different clinicopathological stages of FGS. Furthermore, uMIF levels were also measured during longitudinal testing and the effect of steroids on uMIF levels was also examined. To our knowledge, this is the first study to investigate the role of MIF in human FGS. The aim of the present study was to examine the profile of uMIF in FGS patients.

Patients and methods

Clinical characteristics

Two groups of patients with a biopsy-proven diagnosis were studied. The first group included 20 patients with FGS. They were 12 men and 8 women with a mean age of 27 years. Renal biopsy was performed in all FGS patients, and the diagnosis was made by light, immunofluorescence, and electron microscopy. The histological diagnosis of FGS was based on the criteria of Schwartz and Korbet [16]. FGS required the presence of one or more glomeruli in which sclerosis was present in a segmental distribution. The lesion was focal in that some glomeruli were affected, whereas others were normal. Immunofluorescence studies showed positive fluorescence for IgM and C3 in affected lobules, and electron microscopic studies revealed electron-dense deposits in the same areas. The mesangial matrix increase was semiquantitatively graded (mild 1+, moderate 2+ and marked 3+) upon light microscopic observation. The interstitial fibrosis was also graded (absent or minimal 0, mild 1+, moderate and marked 2–3+) in tissues from patients with FGS. In addition, a portion of each renal biopsy was processed for immunoperoxidase staining with monoclonal antibodies [11]. Anti-CD68 (KPI; Dako, Copenhagen, Denmark) was used as a macrophage marker on frozen sections. For each biopsy the number of labelled cells was counted in each glomerulus and expressed as the number of cells per glomerular cross section. The number of CD68 positive macrophages in the interstitium was counted using an eye piece graticule to identify 10 microscopic fields, each 0·02 mm², and hence a total area of 0·2 mm² was counted. The numbers were then expressed as cells per mm².

The second group was composed of NS caused by other types of glomerular disease. The diagnosis was established by renal biopsy in each case. There was 20 patients with minimal–change nephrotic syndrome (MCNS) and 20 with membranous nephropathy (MN). Their mean age was 36 years; 20 had NS, 20 were in remission. On light microscopy, the MCNS glomeruli appeared normal or showed only minor changes. Immunofluorescence revealed lack of deposition of immunoglobulins and complement in the glomeruli. The criteria described by Mallick et al. [17] were used to characterize the MN: NS and subepithelial electron-dense deposits on renal biopsy. They had a creatinine clearance of more than 80 ml/min. Systemic lupus erythematosus and hepatic diseases were excluded on the basis of clinical history, the results of examination, and the laboratory data. None of these patients had received corticosteroids or immunosuppressive agents before the study, and had no evidence of bacterial or viral infection at the time of the study.

Twenty-one additional patients with renal disease (3 patients with acute poststreptococcal GN (AGN), 3 with polycystic kidney, 1 with horseshoe kidney, 5 with nephrolithiasis, 6 with diabetic nephropathy and 3 with amyloidosis) were studied as a disease control. Twenty healthy adults (11 men, 9 women) with a mean age of 29 years and with normal urine analysis composed the control group. The clinical characteristics of the patients are summarized in Table 1.

Table 1.

Clinical and laboratory features of the patients with renal disease.

FGS							MCNS							MN
Patient No.	Age years	Sex	Urinary protein g/day	Serum albumin g/dl	Serum TC mg/dl	Creatinine clearance ml/min	Patient No.	Age years	Sex	Urinary protein g/day	Serum albumin g/dl	Serum TC mg/dl	Creatinine clearance ml/min	Patient No.	Age years	Sex	Urinary protein g/day	Serum albumin g/dl	Serum TC mg/dl	Creatinine clearance ml/min
1	24	M	4.6	2.9	302	89	1	23	M	6.1	2.3	323	103	1	40	M	5.1	2.8	303	93
2	25	M	5.1	3.2	279	90	2	22	F	6.0	2.6	308	101	2	42	F	5.3	2.8	312	116
3	20	F	4.6	2.3	292	104	3	26	M	6.4	2.8	329	110	3	53	M	5.3	2.9	309	100
4	30	F	3.9	3.4	298	91	4	24	M	5.2	2.5	279	102	4	49	M	4.2	2.9	270	90
5	36	F	4.2	3.1	269	103	5	21	F	4.7	2.7	292	109	5	54	F	4.0	2.7	296	101
6	26	F	5.1	3.4	292	93	6	24	F	4.4	3.1	306	109	6	53	F	5.4	3.0	296	99
7	22	M	4.1	2.8	301	82	7	26	F	4.9	2.9	300	111	7	36	F	3.9	2.6	260	110
8	22	F	6.7	3.1	280	89	8	24	M	6.0	2.4	329	109	8	44	M	5.2	2.5	322	119
9	30	M	3.8	3.8	262	100	9	21	M	5.4	2.9	351	112	9	57	M	4.3	2.5	364	117
10	23	M	4.6	3.3	284	98	10	25	F	5.0	3.1	290	103	10	40	F	4.9	2.9	298	89
11	29	M	2.7	3.9	241	107	11	31	M	0.2	4.0	222	112	11	49	M	0.7	3.1	229	122
12	30	M	1.1	4.0	228	101	12	30	F	0.2	4.1	221	106	12	46	F	0.8	3.6	231	100
13	34	F	1.9	4.5	230	89	13	20	M	0.2	4.1	220	99	13	38	M	0.2	4.0	217	93
14	29	F	2.7	4.4	231	92	14	20	M	0.2	4.0	192	98	14	45	M	0.2	4.2	222	99
15	30	M	0.7	4.9	229	99	15	31	F	0.1	4.1	208	109	15	55	F	0.2	4.8	200	103
16	26	M	0.5	4.8	212	94	16	32	F	0.1	4.6	214	109	16	48	F	0.1	4.3	212	102
17	29	M	0.3	5.0	207	86	17	34	M	0.1	4.3	202	109	17	39	M	0.1	4.7	219	101
18	26	M	0.9	4.6	179	101	18	33	M	0.1	4.4	213	110	18	33	M	0.1	4.3	210	101
19	28	F	0.4	4.6	179	104	19	22	F	0.1	4.9	182	100	19	44	F	0.1	4.2	198	104
20	30	M	0.8	4.6	199	108	20	28	M	0.1	4.7	221	112	20	48	M	0.1	4.0	211	114

The local Ethical Committee approval was received for the studies and the informed consent of all participating subjects was obtained.

MIF assay

Urine samples were collected from patients before steroid therapy. Fresh urine samples collected into 50 ml centrifuge tubes were centrifuged at 1000 g for 10 min at 4°C, passed through a 0·45 µm filter and dialysed twice against phosphate-buffered saline for 4 h and once against RPMI 1640 (Gibco, Grand Island, NY, USA) for 4 h at 4°C. Samples were harvested from the dialysing tubes and frozen at −80°C until use. The concentration of MIF in the urine was measured by an enzyme-linked immunosorbent assay (ELISA) method [18] using commercial antibodies purchased from R & D Systems (Minneapolis, Minn., USA). When 1 ml of phosphate-buffered saline (PBS) was used, the antibody concentration was 500 µg/ml. Immunoplates (Nunc, Roskilde, Denmark) were coated overnight at 4°C with mouse antihuman MIF monoclonal antibody (Cat. No. MAB289; R & D Systems) at a final concentration of 2 µg/ml in 0·1 m NaHCO3 buffer, pH 8·2. Active binding sites were then blocked by PBS containing 0·5% Tween 20 and 3% bovine serum albumin for 2 h at room temperature. Urine samples or recombinant human MIF (Cat. no. 289-MF-002, R & D Systems) used as standard were added and incubated overnight at 4°C. Bounded MIF was detected using 1 µg/ml biotinylated antihuman MIF detection antibody (Cat. No. BAF289, R & D Systems). All assays were revealed by incubation with 0·5 µg/ml of streptavidin-peroxidase (Amersham, Aylesbury, UK) for 30 min at 37°C, then with OPD (0·4 mg/ml, Merck, Darmstadt, Germany) and 0·01% H2O2 in 0·1 m citrate buffer pH 0·5. The absorbance of each sample was recorded at 492 nm in an automated plate reader (Model 3550 UV Microplate Reader; Bio-Rad, Hercules, CA, USA). The urine samples from each patient were assayed in triplicate on the same plate. The levels of MIF were determined by comparison with a standard curve obtained using recombinant human MIF. Creatinine levels were measured in aliquots of the urine samples using an enzymatic/spectrophotometric method. The MIF concentrations were then corrected for creatinine, with final concentration values reported in pg/ml/mg creatinine. The sensitivity of the assay was 10 pg/ml of MIF. There was no cross-reactivity in this ELISA with other cytokines and related molecules such as interleukin (IL)-1, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, IL-13, IL-15, Il-17, IL-18, tumour necrosis factor-α (TNF-α), macrophage colony stimulating factor (M-CSF), macrophage-derived chemokine (MDC), macrophage inflammatory protein-1α (MIP-1α) and monocyte chemotactic protein-1 (MCP-1/MCAF).

Immunohistochemistory

Immunohistochemical analysis was performed to further evaluate MIF localization in biopsy specimens. Cryostat sections (4 µm) were labelled by a four-layer heterologous peroxidase-antiperoxidase (PAP) technique using an antihuman MIF monoclonal antibody (R & D Systems) as previously described [19]. Slides were counterstained with Harris haematoxylin solution.

Statistical analysis

Spearman's rank correlation test was performed for a two-factor correlation. A comparison between two groups was analysed using the Mann–Whitney U-test. When comparing more than two groups, we first performed a nonparametric analysis of variance using the Kruskal–Wallis test followed by the Tukey-Kramer multiple comparison test. These statistical analyses were performed using software programs for a personal computer (StatView, Ver.4·5, Abacus Concepts, Berkeley, CA., USA, and SPSS, Ver.4·0, SPSS, Chicago, IL., USA). P < 0·05 was considered to be statistically significant.

Results

Concentrations of MIF in urine specimens from patients with renal disease

Our article deals with the quantification of MIF antigen levels in urine of patients with FGS as compared with MCNS, MN and renal diseases of nonimmune aetiology. Urinary levels of MIF in normal controls and in patients with renal disease are shown in Fig. 1. The levels of uMIF are given as median and range for each group. Urine from the patients was sampled in the morning of the day of renal biopsy. As can be seen from Fig. 1 in normal control subjects tested, low levels of MIF could be detected in the urine. The MIF excretion was significantly higher in the urine from patients with FGS compared with normal controls and patients with MCNS and MN. In contrast, the levels of uMIF in MCNS and MN patients were not significantly different from normal controls. The results presented in Fig. 1 demonstrate that in FGS patients with NS, uMIF is significantly higher than in both MCNS and MN patients with NS (P < 0·005 for each comparison) and that in FGS patients without NS, uMIF is significantly higher than in both MCNS and MN patients without NS (P < 0·05 for each comparison). As shown in Table 1, the non-nephrotic patients with FGS have more proteinuria than non-nephrotic patients with MCNS or MN.

Fig. 1.

Urinary levels of MIF in patients with renal disease and normal controls. Urine from the patients was sampled in the morning of the day of renal biopsy. The results demonstrate that in FGS patients with nephrotic syndrome (NS), uMIF is significantly higher than in both MCNS and MN patients with NS (**P < 0·005 for each comparison) and that in FGS patients without NS, uMIF is significantly higher than in both MCNS and MN patients without NS (*P < 0·05 for each comparison). The levels of uMIF in MCNS and MN patients with or without NS were not significantly different from normal controls. No patient had received corticosteroid therapy at the time of the study. MIF is expressed as the ratio to the urine creatinine concentration. Diagnostic criteria for nephrotic syndrome (NS) were massive proteinuria (>3·5 g/day) and hypoalbuminaemia (<4·0 g/100 ml) with or without oedema. Each data point represents data from each patient or control. Each bar represents median in each group.

Without parallel studies in other diseases with glomerular inflammation such as postinfectious GN it is not possible to interpret these findings in confidence to indicate that uMIF is a marker of glomerular inflammation. To see if uMIF is truly associated with glomerular inflammation, urine samples were obtained from three patients with AGN. Their renal biopsy showed mesangial hypercellularity, some polymorphonuclear leucocytes and macrophages in the glomerular capillaries. Significant urinary levels of MIF were detected in the patients (median 182·5 pg/mg creatinine).

Nine additional patients with nonglomerular disease (3 with polycystic kidney, 1 with horseshoe kidney and 5 with nephrolithiasis) were studied as a disease control. Each of these patients had normal MIF excretion, and the median of uMIF for all these patients was 35·1 pg/mg creatinine (range 19–41 pg/mg creatinine), which did not differ from the normal subjects. We further determined the uMIF levels in 9 patients with other types of glomerulopathies (6 with diabetic nephropathy and 3 with amyloidosis) in whom T cell abnormalities were not observed. The patients excreted approximately equal amounts of MIF (median 39·8, range 21–54 pg/mg creatinine) and more than that from urine samples of healthy subjects, but differences were not statistically significant.

Relationship between urinary levels of MIF and clinicopathological activity

As shown in Fig. 2a, urinary levels of MIF correlated well with the grade of mesangial matrix increase in glomeruli in FGS patients (P < 0·01). We have also examined the interstitial changes other than mesangial matrix increase. As shown in Fig. 2b, the uMIF levels were associated with the degree of interstitial fibrosis (P < 0·01).

Fig. 2.

The association between uMIF and the grade of mesangial matrix increase (a) or interstitial fibrosis (b) in patients with FGS. Urinary levels of MIF were significantly correlated with the grade of mesangial matrix increase (Kruskal–Wallis test, P < 0·01) and that of interstitial fibrosis (P < 0·01).

Figure 3 shows the correlation between the number of intraglomerular or interstitial macrophages and the concentration of uMIF in FGS patients. We found that there was a strong correlation between the severity of intraglomerular and interstitial macrophage numbers and uMIF in the patients (r = 0·602, P < 0·001; r = 0·631, P < 0·001, respectively).

Fig. 3.

Correlation between the number of intraglomerular (a) or interstitial (b) macrophages and the concentration of uMIF in patients with FGS. A significant, positive correlation was observed between the two parameters (r = 0·602, P < 0·001; r = 0·631, P < 0·001, respectively).

To characterize further the cellular source of MIF, we examined the renal biopsy specimens by immunohistological methods [19]. Eight cases of FGS were studied. In the present study we have found strong staining in some mesangial lesions and in most damaged tubules, together with many infiltrating macrophages (Fig. 4).

Fig. 4.

(a) Glomerulus from a patient with early lesions of FGS stained by the immunoperoxidase method using a monoclonal antibody to MIF. MIF expression was increased in the glomerulus. Original magnification × 250. (b) Immunohistochemical study showing essentially negative results in a disease control of MCNS. Original magnification × 250.

As seen in Fig. 5a, there was a significant correlation between the levels of uMIF and those of urinary protein in FGS patients (r = 0·612, P < 0·001). In contrast, there was no correlation for the control diseases of MCNS and MN (Fig. 5b). To clarify some aspects of non-T cell mediated renal diseases with proteinuria (diabetic nephropathy, amyloidosis and polycystic kidney), we have attempted to correlate the uMIF with proteinuria. There was no significant correlation between the levels of uMIF and the degree of proteinuria in the disease control group.

Fig. 5.

(a) Relationship between urinary levels of MIF and the amount of proteinuria in patients with FGS. There was a positive correlation between the uMIF levels and urinary protein excretion (r = 0·612, P < 0·001). NS = nephrotic syndrome. (b) In contrast, there was no correlation for the control diseases of MCNS and MN (r = 0·179, P = 0·294).

We next analysed the correlation between serum MIF content and urinary levels of MIF in patients with FGS. We found no significant correlation between the two parameters in the patients (Fig. 6). The data suggest that increased uMIF excretion is not simply a reflection of increased clearance of serum MIF.

Fig. 6.

Correlation between serum levels of MIF and uMIF in patients with FGS. There was no correlation between the two parameters (r = 0·192, P = 0·301).

Longitudinal analysis of uMIF levels in seven patients with FGS

The correlation with disease activity was reflected by the seven individual FGS patients studied at seven different time points. The change in uMIF correlated with disease activity. The improved correlation may be due to a closer relationship between uMIF levels and disease activity amongst the longitudinal data than the cross-sectional data. Figure 7 shows seven cases of FGS in which seven urine samples could be measured for uMIF. The patients had an acute exacerbation of FGS with heavy proteinuria. Their renal electron micrograph revealed part of a glomerulus showing capillary collapse, mesangial sclerosis and small electron dense deposits. The podocytes are detached from the basement membrane and the space is filled with layers of new basement membrane material. The immunofluorescence micrograph showed irregular segmental staining for IgM and C3 corresponding to a site of segmental sclerosis. Prednisolone therapy was started after the third assay was performed. They had high levels of uMIF particularly in the beginning of acute exacerbation; then uMIF levels decreased sharply about 2 weeks after the steroid therapy, and then decreased to base line levels with the improvement of their proteinuria. The effect of steroid therapy on uMIF and proteinuria was similar at the same time points for these seven patients. There has been no subsequent exacerbation of the disease to date.

Fig. 7.

Longitudinal analysis of uMIF levels in seven patients with FGS. The clinical and laboratory data and urinary samples were obtained at the same time. In these cases, the urinary levels of MIF decreased in parallel with their disease activity. The effect of steroid therapy on uMIF and proteinuria was similar at the same time points for these seven patients. The degree of proteinuria is shown as median and range.

Alteration of uMIF levels following steroid therapy in FGS patients

Twelve FGS patients were treated with 0·8 mg/kg per day of prednisolone for two months. The urinary samples were taken before and after 5 days of treatment with prednisolone. Our data revealed quantitative as well as temporal variations in the MIF levels following steroid therapy for FGS. We found that their levels were significantly decreased by steroid therapy (Fig. 8).

Fig. 8.

Urinary levels of MIF in 12 FGS patients directly before and 5 days after steroid therapy. The elevated uMIF levels in active FGS patients significantly (P < 0·001) decreased after corticosteroid therapy.

We have also examined eight FGS patients who did not respond to steroid therapy. The uMIF levels in the nonresponder (median 191·6, range 158–229 pg/mg creatinine) were significantly (P < 0·01) higher than those in eight steroid-sensitive controls (median 124·2, range 86–167 pg/mg creatinine). The levels of uMIF tended to increase when they did not respond to steroid therapy. Finally the patients developed steroid-resistant NS within the following 8 months. These findings suggest that uMIF may be useful as a marker of steroid-resistant NS.

Discussion

We have presented evidence that urinary excretion of MIF is increased in FGS patients and uMIF is higher in patients with active glomerular lesions in FGS. It is tempting to speculate that MIF is a pathogenic factor which precipitates and sustains the glomerular inflammation. However, the data do not imply the question of whether MIF plays a role as the initial or exacerbation mediator of FGS or in the progression of FGS. The pathological significance seems to be complex, but we believe that it may play a modest role in determining the activity of the disease. Further studies are needed to see if uMIF is truly associated with glomerular inflammation.

It has been shown that many proinflammatory cytokines such as TNF- α, IL-1, and chemokines are highly produced by glomerular cells, and contribute to glomerular inflammation [19–21]. Interestingly, it has been reported that mesangial cells can produce MIF in vivo and in vitro[6]. The interaction of MIF with other cytokines within the complex cytokine network is just being documented [1,4]. Indeed, MIF has been shown to promote TNF-α production from macrophage cell lines and peritoneal mouse macrophages [22]. Several preliminary experiments further illustrate the importance of MIF for the development of GN [6,14]. It has been demonstrated that administration of a neutralizing antibody to MIF dramatically suppressed an immunologically induced disease model of rapidly progressive crescentic GN [14]. This demonstrates a key regulatory role for MIF in the pathogenesis of immune-mediated GN. Of great importance is a recent study on immunohistochemical localization of MIF in human kidney [8]. The results suggest the possibility that MIF takes part in the mechanisms of inflammation and immunological events in the human kidney.

The cellular source of MIF was addressed in this study. We set out in this study to explore whether MIF is produced in the kidney (e.g. by inflammatory cells or mesangial cells) or elsewhere in the body. We also examined whether elevated uMIF levels resulted from local release of MIF at the site of renal involvement, or from the systemic effects of active FGS. New evidence in this study was the up-regulation of glomerular and tubular MIF expression in areas of macrophage accumulation. These findings suggest that uMIF reflects MIF expression within the injured kidney. Further supporting evidence is the demonstration that no correlation was found between the uMIF concentration and serum MIF levels, indicating that increased uMIF excretion is not simply a reflection of increased clearance of serum MIF. Taken together, our results strongly suggest the cellular source of MIF at sites of glomerular and interstitial inflammation. Immunocytochemical and in situ hybridization studies indicate local MIF production by resident glomerular cells [6]. In preliminary studies, Atkins et al. [23] have shown that MIF expression is up-regulated in human GN, with strong MIF expression in areas of macrophage infiltration. Macrophages are thought to be the major source of MIF in human GN [24,25]. MIF is also a potent inducer of macrophage activation [1]. Therefore, high levels of uMIF would reflect macrophage activation, which is considered to be involved in the development of the glomerular injury during the acute phase of FGS [25].

There were important studies [26] supporting the hypothesis that circulatory factors have important role in the pathogenesis of proteinuria and relapse of FGS. Therefore, the cellular source of MIF should be a major issue in this study. It is interesting to measure the circulatory concentration of MIF from the patients. We found that serum MIF levels were not correlated with uMIF levels in the patients. Furthermore, we provide additional data on the immunopathology of MIF in the patients. Macrophages were the majority of the interstitial inflammatory cells. Interestingly, tubular MIF expression and the number of interstitial MIF-positive cells were much more greater in the proteinuric patients. Podocyte injury is a major feature of FGS [17,27]. Discussion about a possible effect of MIF on podocytes may be considered. Two hypotheses could be proposed. First, increased MIF expression by the glomerular infiltrating cells may result in preferential binding of activated effector cells to the podocytes with subsequent injury of the cells. The ensuing proteinuria in itself may thereafter contribute to the cell damage and promote a cell-mediated glomerular immune reaction. The hypothesis may explain the increased frequency of podocyte injury in the proteinuric group. Second, MIF released from the glomerular inflammatory cells early in the course of the disease may induce reactive changes of the cells. Glomerular podocytes are capable of synthesizing MIF that could eventually trigger a cell-mediated immune reaction, thus promoting a perpetual cycle of injury.

The data on uMIF excretion in so- called control diseases with proteinuria such as MCNS and MN are presented in Fig. 5b. It is more convincing to see the uMIF excretion of these patients compared to their urinary protein levels to exclude the possibility that the increased uMIF excretion found in FGS is not just a reflection of the degree of proteinuria. As shown in Fig. 5b, there was no significant correlation between the levels of uMIF and the severity of proteinuria in the control diseases of MCNS and MN, suggesting that the urinary detectability of MIF reflects a specific involvement in the disease process of FGS.

Macrophage infiltration is not a regular feature of FGS in human [27]. We need to discuss the significance of the large number of macrophages in the glomeruli and interstitium of these patients shown in Fig. 3. We have found that interstitial macrophage infiltration is a feature of the more aggressive forms of human FGS. High levels of MIF in FGS may immobilize macrophages and facilitate their localization at the interstitium, leading to production of several inflammatory cytokines. Accumulating evidence suggests that infiltrating macrophages play a pivotal role in the progression to glomerulosclerosis [11]. Recent studies have revealed that neutralization of MIF by anti-MIF serum dramatically suppresses an immunologically induced disease model of rapidly progressive crescentic GN [14].

Macrophages have different state of activation and wide range of function [28].One cannot always assume that increased number implies more severe disease. Although we agreed that macrophage infiltration is not a regular feature of FGS in human [27], but further clarification of this cellular variants of FGS will be helpful. What is the significance of glomerular macrophages in FGS? Infiltration of activated macrophages expressing CCR5 antigen in biopsy kidney lesions has been reported [29]. It is still uncertain whether glomerular macrophages are activated in acute FGS, as some reports have provided evidence of macrophage activation [30], whereas others have not [31]. Saito et al. [30] reported that glomerular macrophage activation is evident during the active stage of FGS. Frosch et al. [32] demonstrated the functional differences of glomerular macrophages in nephritis, emphasizing the heterogeneity of renal macrophages. Glomerular and interstitial macrophage infiltration is a feature of all types of acute injury and chronic progressive renal disease [28]. Some reports suggest that interstitial macrophages correlate with more severe FGS [33]. Other studies [34] have pointed out an increased number of interstitial macrophages in patients with FGS, and a correlation of the number of interstitial cells and renal functional parameters in patients with FGS.

What is the biological significance of uMIF in patients with FGS? We believe that it is difficult to relate directly the increase of uMIF in this study to the progression of FGS, since this change is not recognized in every patient. However, it is possible that this increased MIF is indirectly related to the pathophysiology. We demonstrated that the levels of uMIF and proteinuria rose in parallel in this retrospective study of seven FGS patients, although the study of MIF before onset is difficult to establish and changes before disease onset are not known in human FGS. We believe that the change of MIF before flaring is not simply explained, and that the role of this lymphokine varies.

We monitored the urinary excretion of MIF before and after 5 days of treatment with glucocorticoids in patients with FGS. We found that their levels were significantly decreased by steroid therapy. We have also shown seven cases of FGS in which longitudinal analysis of uMIF concentration was performed. As shown in Fig. 7, the effect of steroid therapy on uMIF and proteinuria was similar at the same time points for these seven patients. As the final effects of the steroid treatment in FGS are not known, it is possible that the steroids in some way affect the MIF receptor expression by FGS cells. In addition, steroid may induce some inhibitory mediators which may down-regulate the production and activity of MIF.

In summary we first found that uMIF levels increased in patients with FGS, and reflected the clinicopathological characteristics of the disease. Further studies are necessary to fully characterize the significance of uMIF found in FGS patients.