Resistin as a potential marker of renal disease in lupus nephritis

Abstract

Systemic lupus erythematosus (SLE) and lupus nephritis (LN) have strong concomitance with cardiovascular disease that cannot be explained fully by typical risk factors. We examined the possibility that serum or urine expression of adipokines may act as biomarkers for LN, as these proteins have been associated previously with cardiovascular disease as well as SLE. Antibody arrays were performed on serum and urine from lupus patients and matched controls using a cross-sectional study design. From the initial array-based screening data of 15 adipokines, adiponectin, leptin and resistin were selected for validation by enzyme-linked immunosorbent assay (ELISA). Correlations were determined between adipokine expression levels and measures of disease activity or lupus nephritis. The expression of adiponectin and resistin was increased in both sera and urine from LN patients, while leptin was increased in LN patient sera, compared to matched controls. Serum resistin, but not urine resistin, was correlated with measures of renal dysfunction in LN. Serum resistin expression may be useful as a marker of renal dysfunction in patients with LN, although longitudinal studies are warranted. Further studies are necessary to determine if resistin has functional consequences in LN.

Introduction

Systemic lupus erythematosus (SLE) is a polygenic chronic inflammatory disease which affects multiple organ systems. While renal and dermatological manifestations are often considered hallmarks of the disease, patients with SLE also have as much as a 50-fold increased risk of developing cardiovascular complications compared to age-matched healthy controls 1. Significant body composition changes 2,3, increased incidence of metabolic syndrome mediated by insulin resistance 4–7 and obesity 8,9 have all been associated previously with SLE. Adipose tissue-derived proteins called adipokines are critical mediators of obesity, metabolic syndrome and cardiovascular disease development 10,11. Taken together, these data may suggest a contributing role for adipokines in the development of SLE and lupus nephritis (LN).

Energy storage was believed to be the primary function of white adipose tissue (WAT) until the discovery of adipsin 12 and leptin 13. The investigation of adipose-derived molecules has since expanded exponentially, with WAT becoming an increasingly popular target for investigating immunological dysfunction related to obesity and metabolism. WAT has since been linked to at least 50 bioactive proteins 14. Many of these adipokines [including interleukin (IL)-1, IL-6, IL-10, CCL2, CCL5, CXCL8 and CXCL10] are expressed primarily by resident non-adipocytes (including macrophages) in the stromal vascular fraction of adipose tissue 11,15. Others, including adiponectin and leptin, are produced and secreted predominantly or exclusively by adipocytes 16. The source of resistin is still in question; however, it is now widely accepted that resistin can be produced by other cells aside from adipocytes in humans 17,18. Many adipokines, including leptin and resistin, have been found to have proinflammatory effects beyond their roles in metabolism. While adiponectin is typically regarded as having anti-inflammatory properties, multiple studies have reported increased adiponectin levels in connection with several disease conditions. This may be related to differing isoforms of adiponectin, as high molecular weight adiponectin has been shown previously to have proinflammatory effects 19,20. A role for adiponectin in protein-energy wasting has also been proposed to explain this discrepancy in chronic kidney disease 21. Aside from the proinflammatory, atherogenic nature of many adipokines, several adipokines, including adiponectin, leptin and resistin, have also been linked previously to renal diseases 22,23. Most previous studies have examined one or a few adipokines at a time, with little effort to study all adipokines comprehensively. In this study we first undertake a comprehensive array-based screen to interrogate the level of multiple adipokines in serum and urine from patients with SLE. Several differentially expressed adipokines were then validated by enzyme-linked immunosorbent assay (ELISA). From the examined adipokines, serum resistin emerges as a promising marker of renal dysfunction in SLE.

Methods

Patients

Lupus nephritis (LN) patients were recruited from the renal clinic at Parkland Hospital, an affiliated hospital of the University of Texas Southwestern Medical Center, and all patient-related procedures were performed strictly following the institutionally approved Institutional Review Board (IRB) protocol (UT Southwestern, #STU 082010-119). All patient-informed consents were obtained prior to sample collection. For the initial array-based screen of serum, five LN patients [mean age, 33·4; five females, three African American + two Hispanic, mean Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) = 6·8] and three healthy individuals (mean age, 52; three females, two African American + one Hispanic) were investigated. The study population for the initial array-based screen of urine comprised three healthy individuals (mean age, 35; three females) and five patients with LN (mean age, 38·3; four females + one male; mean SLEDAI = 19·4). For the validation studies, samples from 38 LN patients were studied using an orthogonal method. Of these patients, 36·8% had inactive LN (rSLEDAI = 0) and 63·2% had active LN (rSLEDAI ≥ 4). There were no patients with intermediate SLEDAI values (0∼3). LN was diagnosed and classified based upon ISN/RPS 2003 classification. Inclusion criteria included LN patients with biopsy-proven LN. Exclusion criteria were patients with end-stage renal disease. Clinical data were gathered by chart review and SLEDAI was calculated based on chart review. Detailed information pertaining to these patients is provided in Table 1.

Table 1.

Demographics and clinical characteristics

Total no. of subjects	38
Subjects with serum samples	28
Subjects with urine samples	27
Subjects with both serum and urine	16
Female, no. (%)	32 (84·2)
Age, mean ± s.e., years	35·1 ± 1·9
BMI	29·5 ± 1·6
Ethnicity, African American/Hispanic/Caucasian, no.	20/16/2
SLEDAI, median (interquartile)	10 (4–16)
Renal SLEDAI, median (interquartile)	4 (4–8)
No. of patients with renal SLEDAI = 0 (%)	8 (21)
Protein : creatinine ratio, mg/mg, mean ± s.e.	1·08 ± 0·17
Serum Cr, mg/dl, mean ± s.e.	1·35 ± 0·18
Non-renal manifestations and lupus serology, no. (%) Malar rash	16 (42·1)
Photosensitivity	6 (15·8)
Oral ulcers	6 (15·8)
Non-erosive arthritis	17 (44·7)
Pleuritis or pericarditis	8 (21)
Neurological disorder	5 (13)
Haematological disorder	16 (42·1)
Immunological disorder	33 (86·8)
Positive ANA	33 (86·8)
Comorbidities, no. (%)
Diabetes mellitus	4 (10·5)
Hypertension	28 (73·7)
Dyslipidaemia	16 (42·1)
Cardiovascular disease	4 (10·5)
Anaemia	23 (60·5)
Anti-phospholipid syndrome	4 (10·5)
Venous thromboembolism	4 (10·5)
Others	24 (63·1)
Current medications, no. (%)
Prednisone	29 (74·4)
Mycophenolic acid	10 (25·6)
Cyclophosphamide	3 (7·7)
Azathioprine/MTX	6 (15·4)
Cyclosporin/tacrolimus	2 (5·1)
Hydrochloroquine	17 (43·6)
Angiotensin blocking agents	23 (59·0)

The healthy controls had the following characteristics: mean age: 36; ethnicity: African American (77·8%) and Hispanic (22·2%); gender: female (44·4%) and male (55·6%). ANA = anti-nuclear antibodies; BMI = body mass index; MTX = methotrexate; SLEDAI = Systemic Lupus Erythematosus Disease Activity Index; s.e. = standard error.

Protein analyte array

All serum and urine samples were hybridized to glass slide arrays that interrogate the level of multiple analytes, including 15 adipokines (AAH-CYT-G4000-8; Raybiotech, Norcross, GA, USA). For each adipokine surveyed, the mean of the healthy and LN groups was calculated and a mean ratio was computed representing the ratio of the mean signal of the LN samples to the mean of the healthy control samples. These mean ratios were then log₂-transformed and plotted as bar graphs to demonstrate the differential adipokine expression in serum and urine. With this transformation, values centred on 0 imply equal means and a twofold increase results in a transformed ratio of 1, and a 50% decrease results in a transformed ratio of −1.

Pathway analysis

Expression data (fold-change in LN) pertaining to the 15 adipokines interrogated in the urine antibody array screen were entered into Ingenuity Pathway Analysis software (Ingenuity Systems, Redwood City, CA, USA). Data were queried for direct and indirect molecular relationships in the context of all available data sources, species and cell types.

Validation assays

Serum and urine samples obtained from the renal clinic at Parkland Hospital, Dallas, TX were aliquoted prior to storage at −80°C. One aliquot was retrieved for each assay to avoid multiple freeze/thaw cycles. Urine and serum adiponectin, leptin and resistin were measured using ELISA kits from Raybiotech Inc. Urinary adipokines were normalized against urine creatinine. Patient characteristics for the validation ELISAs are described in Table 1.

Statistics

Statistical analysis was performed using Medcalc version 13 and Graphpad Prism version 6. For all statistical analyses, an alpha level of 0·05 was considered significant. The data were applied to the Shapiro–Wilk normality test to determine distribution. Based on these normality results, Mann–Whitney U-tests were used to compare data from the different groups and correlation was determined using the Spearman's rank order method with a two-tailed test.

Results

Antibody arrays reveal altered adipokine profiles in SLE sera and urine

Antibody arrays were performed on several serum and urine samples isolated from LN patients and matched healthy controls, as detailed in Methods. The expression of a number of previously described adipokines was found to be altered in LN patients compared to controls. In patient sera, adipsin, epithelial neutrophil-activating peptide (ENA-78) (CXCL5), leptin (P < 0·05), pulmonary and activation-regulated chemokine (PARC) (CCL18), platelet-derived growth factor-BB (PDGF-BB) (P < 0·05), regulated on activation, normal T cell expressed and secreted (RANTES) (CCL5), plasminogen activator inhibitor-1 (PAI-1) (serpinE1), resistin and serum amyloid A (SAA) expression were increased, while adiponectin (P < 0·05), angiopoietin-like -4 (ANGPTL4) (P < 0·05) and IL-8 expression were decreased compared to the healthy controls (Fig. 1a). Only the molecules indicated reached statistical significance at P < 0·05. If corrected for multiple testing correction (n = 15 tests), none of these differences reached statistical significance, most probably because of the limited sample size. In urine, expression of adiponectin, adipsin, angiopietin-1, ANGPTL4, ENA-78, IL-8, PAI-1, PARC, PDGF-BB, RANTES, resistin and SAA were all increased in SLE compared to healthy controls. However, none of these increases in urine reached statistical significance, possibly because of the limited number of subjects used in the pilot screening study (Fig. 1b). It is interesting that the altered adipokines were related to each other functionally, as determined by pathway analysis (Fig. 1c).

Fig 1.

Adipokine expression is altered in patients with systemic lupus erythematosus (SLE) or lupus nephritis. Comprehensive screening was performed using antibody arrays and serum (a) isolated from patients with lupus nephritis (LN) (n = 5) or normal controls (n = 3), or urine (b) isolated from patients with LN (n = 5) or normal controls (n = 3). Bars represent log₂ transformation of the mean ratio (LN : healthy), *P-value less than 0·05 resulting from Mann–Whitney U-tests of non-transformed raw data. (c) A molecular adipokine network including molecules identified from the antibody arrays as being expressed differentially in lupus or lupus nephritis patients. The depicted network was generated by Ingenuity Pathway Analysis. Solid lines represent direct interactions, dotted lines represent indirect interactions. Red represents up-regulation in LN serum or urine compared to the normal controls.

Adipokines in urine correlate with markers of inflammation

Because IL-6 and tumour necrosis factor (TNF)-α are proinflammatory cytokines which have been described independently in SLE 24 and also as adipokines 25–27, we next explored whether the expression of the other adipokines also correlated with these measures of inflammation. In serum, levels of most adipokines were generally unrelated to IL-6 and TNF-α expression in the same serum sample, although PARC correlated negatively (r = −0·74, P < 0·05) with IL-6 (Supporting information, Fig. S1a). While serum levels of adipokines were only mildly associated with markers of inflammation, urine levels of IL-6 and TNF-α were correlated positively with the levels of urinary adiponectin (r = 0·71, P = 0·05 versus TNF-α), adipsin (r = 0·81, P < 0·05 versus IL-6; r = 0·71, P = 0·05 versus TNF-α), PAI-1 (r = 0·81, P < 0·05 versus IL-6), PARC (r = 0·88, P < 0·01 versus IL-6) and resistin (r = 0·81, P < 0·05 versus IL-6; r = 0·71 P = 0·05 versus TNF-α; Supporting information, Fig. S1b).

Adipokine validation by ELISA

ELISAs were performed on available samples to validate the observed differences in selected adipokines. Adiponectin and leptin were chosen for validation based on their biological relevance, their well-documented role as adipokines and because they were altered significantly in the serum array data (Fig. 1a). Despite falling short of statistical significance (P = 0·06), resistin was also validated in our cohort based on previous descriptions of its role as an adipokine, with adipocyte-specific contributions and its association with autoimmunity 28–32. Our interest in this molecule was also supported by the observation that resistin deomonstrated the highest fold change in the serum array data (Fig. 1a). Serum and urine from patients and controls which had not been used previously for the antibody arrays were examined for expression of adiponectin, leptin and resistin by ELISA. In serum, adiponectin (3·5-fold, P < 0·001), leptin (1·6-fold, P < 0·05) and resistin (2·1-fold, P < 0·01) were all increased significantly in LN patients (n = 28) compared to normal controls (n = 9, Fig. 2). In urine, adiponectin (4·1-fold) and resistin (3·3-fold, P < 0·05) expression was increased in patients with lupus nephritis (n = 27) compared to normal controls (n = 8). Given that adiponectin, leptin and resistin were all up-regulated in patient sera compared to control sera, correlations between these adipokines were tested. All tested correlations trended in the positive direction and demonstrated a moderate correlation (Fig. 3a–c), as determined by Spearman's rank-order correlation (P ≤ 0·05). A three-dimensional plot of adiponectin, leptin and resistin underscores the moderately positive relationship between the serum levels of these three adipokines (Fig. 3d).

Fig 2.

Validation of serum and urine adipokines. Adipokine array data were validated using serum (a) or urine (b) samples by enzyme-linked immunsorbent assay. Serum and urine were collected from lupus nephritis (n = 27–28) patients or normal controls (n = 8–9) as described in the Materials and methods. Sera or urine was assayed for adiponectin, leptin or resistin concentrations as recommended by the manufacturer (RayBiotech, Norcross, GA, USA). Each dot represents an independent sample and the horizontal lines represent the median and interquartile range. All data were compared using the Mann–Whitney U-test. *P < 0·05; **P < 0·01; ***P < 0·005.

Fig 3.

Correlation of serum adipokines. Correlation of serum adipokines was determined by comparing expression of the serum levels of (a) adiponectin and leptin, (b) adiponectin and resistin or (c) leptin and resistin. Correlations were determined by Spearman's rank-order method. To determine the level to which these three adipokines were correlated, (d) a three-dimensional graph of adipokine expression was created using SigmaPlot. Each dot represents an independent sample. Red dots represent lupus nephritis (LN) patients (n = 28), blue dots represent controls (n = 9).

Serum adipokines correlate with renal damage

While resistin was not correlated significantly with disease activity (although a positive trend was observed, Fig. 4a), it demonstrated a moderate correlation with markers of renal dysfunction and glomerular filtration rate (GFR), including serum creatinine (Fig. 4b, r = 0·45, P = 0·01), blood urea nitrogen (BUN, Fig. 4c, r = 0·54, P < 0·005) and urine protein/urine creatinine ratio (Fig. 4d, r = 0·50, P = 0·01). Serum leptin (Fig. 4e,f) did not demonstrate a significant correlation with either disease activity or markers of renal damage. Serum adiponectin levels generally decreased as disease activity increased (Fig. 4g), and also demonstrated a significant correlation with serum creatinine levels (Fig. 4h), but not other markers of renal dysfunction (data not shown).

Fig 4.

Correlation of serum adipokines to measures of disease. Serum levels of (a–d) resistin, (e,f) leptin or (g,h) adiponectin were compared to various disease parameters as indicated. Correlations were determined by Spearman's rank-order method. Each dot represents an independent sample. Only patients with both urine and serum samples were included in this analysis (n = 16).

As urinary adiponectin and resistin levels were also increased significantly in patients compared to normal controls (Fig. 2b), the potential correlations between serum adiponectin or resistin and urine adiponectin or resistin (normalized against urine creatinine) were next examined in LN patients from whom both sets of data were available (n = 16). Serum adiponectin and urine adiponectin were not correlated significantly (data not shown). Although serum and urine resistin levels were not correlated significantly with each other (Fig. 5a), the data trended in the positive direction suggesting that, in general, increased serum resistin levels are associated with higher urinary resistin levels. Interestingly, despite the significant correlation between serum resistin levels and multiple markers of renal damage, no significant correlation existed between urine resistin and these disease markers, including the urine protein : creatinine ratio (Fig. 5b–d).

Fig 5.

Correlation of urine resistin expression with (a) serum resistin, (b) Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) or (c,d) measures of lupus nephritis. Correlations were determined by the Spearman's rank-order method. Each dot represents an independent sample. Only patients with both urine and serum samples were included in this analysis (n = 16).

Discussion

Our data show that levels of adipokines, including adiponectin, leptin and resistin, are altered in both the sera and urine of lupus patients compared to matched controls. While these data resonate well with previous reports, the present study investigates the association of adiponectin, leptin and resistin in sera and urine from the same LN patient cohort. Although contradictory reports exist 33,34, adiponectin 7,35,36 and leptin 7,35 have been shown to be elevated in SLE sera. Our data show a moderate correlation between serum adiponectin and serum creatinine levels. Despite the limited number of samples, these data are similar to previous reports indicating that adiponectin is increased in SLE patients with renal involvement 36. While some of the previous studies did not find any evidence that resistin levels were increased in patients with SLE 35,37–39, one study has demonstrated that resistin correlates with markers of inflammation and renal damage in SLE 40. Serum resistin was not correlated with rSLEDAI. Also, serum resistin was similar (15·0 ± 9·3 versus 15·8 ± 9·3, P = 0·43) in patients with rSLEDAI=0 compared to patients with rSLEDAI ≥ 4. Similar to that report, our data indicate a significant positive correlation between serum resistin levels and serum BUN, creatinine, (Fig. 4b,c) and urine protein : creatinine ratio (Fig. 4d), which represent currently used markers of renal damage in LN. Elevated urine resistin has not been reported previously in LN patients. Although serum resistin and urine resistin share a weak positive correlation in patients with LN (Fig. 5a), urine resistin levels did not correlate with markers of renal damage (Fig. 5b,c) or disease severity (Fig. 5d) or rSLEDAI in patients with LN. Overall, these findings suggest that most of the urine resistin may be serum-derived, and that serum resistin holds promise as a potential biomarker of lupus nephritis. Further studies in a larger patient cohort are necessary to discern the value of these observations.

Resistin was described initially as an adipocyte-limited protein associated with insulin resistance and obesity in mice 17. It is now widely accepted that the primary source of human resistin is bone marrow-derived mononuclear cells 41, although some circulating resistin may be derived from adipocytes 29. Furthermore, the function of resistin goes beyond its metabolic role in obesity and insulin resistance to include a more recently appreciated role in inflammation. Human resistin expression has been shown to be induced in macrophages following the initiation of a proinflammatory signalling cascade by TNF-α and IL-6 41. Multiple reports indicate that serum resistin correlates with markers of inflammation in diabetes as well as a number of other inflammatory conditions, including rheumatoid arthritis, Sjögren's syndrome, inflammatory bowel disease and sepsis 41–43. Resistin expression is also a marker of disease outcome in cardiovascular disease, breast cancer and asthma 41. Consistent with our findings, increased resistin levels have been demonstrated in patients with chronic kidney disease 44,45. Elevated resistin is also associated with decreased GFR in these patients 46,47, suggesting that elevated serum resistin levels may, in part, be caused by decreased filtration in the kidneys.

Previous reports have indicated that leptin is cleared rapidly from the circulation by glomerular filtration of the intact molecule and then degraded in the renal tubules 48,49. Multiple studies have indicated that urinary leptin is difficult to detect, and several have demonstrated that increased renal/urinary leptin is suggestive of end-stage renal failure. Human leptin is a 16 kDa protein with 167 amino acids. Leptin, unlike adiponectin and resistin, is not cleared through the kidney but is instead bound rapidly by megalin in the renal tubules and metabolized 49–51. Perhaps not surprisingly, it was undetectable in either patient or control urine. Human renal endothelium and glomerular shedding of adiponectin from endothelial surfaces by proteolytic cleavage could lead to the degradation of high-order complexes of adiponectin and subsequent appearance of the adiponectin monomer (∼28 kDa), dimer (∼56 kDa) and trimer (∼68 kDa) in the urine 52. Circulating resistin is a low molecular weight protein ∼12·5 kD. Significant negative correlation between resistin and eGFR have been reported 53,54, suggesting that resistin concentration depends upon renal function and is correlated with the severity of the renal disease.

There are some limitations to this study. First, the initial array-based screen for adipokines included only five LN samples. It is possible that a larger sample size for the primary screen may have uncovered additional adipokine differences in LN. Secondly, as the focus of these studies were on LN rather than SLE, future studies are warranted to determine if these adipokines are also elevated in LN-free SLE. Thirdly, in the present study, patients and subjects were not matched for body mass index (BMI). Given that adipokine levels can, potentially, be affected by body fat, future investigations need to examine this caveat closely. Finally, longitudinal studies are warranted to test the predictive value of serum resistin in LN.

Based on the emerging data implicating resistin in inflammation, the significance of increased serum resistin levels may go beyond that of a biomarker to include a functional role in the progression of lupus nephritis. Previous reports have shown that resistin can bind Toll-like receptor 4 (TLR-4) on human leucocytes leading to the activation of multiple proinflammatory cytokine cascades 55. Given that TLR-4 has been shown to play an important role in the development of SLE 56,57 and in promoting renal damage 58,59, it is possible that the increase in resistin may play a direct role in the pathogenesis of LN. Further studies are necessary to investigate this possibility fully.

Disclosure

The authors have no conflicting interests to report.

Supporting Information

Additional Supportingporting information may be found in the online version of this article at the publisher's web-site:

Fig. S1. Correlation of serum and urine adipokines with measures of inflammation.