Summary
The study was aimed at identification by proteomics and validation by enzyme‐linked immunosorbent assay (ELISA) of potential urinary biomarkers for lupus nephritis. Study subjects comprised 88 systemic lupus erythematosus (SLE) patients and 60 controls (rheumatoid arthritis, diabetes mellitus and healthy individuals). Based on the SLE disease activity index (SLEDAI), patients were classified as active renal (AR), active non‐renal (ANR) or inactive disease (ID). Urinary proteins from a group of patients with AR or ID were resolved by two‐dimensional gel electrophoresis and identified by matrix‐assisted laser desorption ionization–time of flight–mass spectrometry (MALDI‐TOF‐MS/MS). The selected biomarkers were validated by ELISA using samples from all patients and controls. AR patients were followed‐up for 12 months after start of therapy. Three urinary proteins, alpha‐1 anti‐chymotrypsin (ACT), haptoglobin (HAP) and retinol binding protein (RBP), were detected in patients with AR and not ID. Upon validation, ACT levels were higher in AR patients than the other groups (P < 0·001) and showed good correlation with renal SLEDAI (r = 0·577, P < 0·001) as well as SLEDAI (r = 0·461, P < 0·001). Similarly, HAP levels were > 10‐fold higher in AR than other groups (P < 0·001) and correlated well with renal SLEDAI (r = 0·594, P < 0·001) and SLEDAI (r = 0·371, P < 0·01). RBP levels were also higher in AR patients than in other groups (P < 0·05), except diabetes, and showed moderate correlation with renal SLEDAI (r = 0·284, P < 0·008) and SLEDAI (r = 0·316, P < 0·003). Upon follow‐up with treatment, levels of all three proteins declined at 6 and 12 months (P < 0·01). Multiple logistic regression identified ACT as the best marker to differentiate AR from ANR. Urinary HAP, ACT and RBP are potential biomarkers for lupus nephritis activity.
Keywords: biomarkers, cyclophosphamide, glomerulonephritis, systemic lupus erythematosus
Introduction
Lupus nephritis (LN) is a key determinant of the treatment outcome in systemic lupus erythematosus (SLE). Patients with LN require aggressive immunosuppression and, in the absence of any better option, the use of existing drugs in a patient‐tailored manner remains the only way to improve outcomes 1. In this situation, biomarkers could help in diagnosis, treatment and assessment of response to the treatment. A number of serum and urinary biomarkers for LN have been pursued in the past with the hope of finding some which could reflect disease activity in a robust and reliable manner 2. Of late, realizing that the hypothesis‐driven approaches have not succeeded, researchers are trying hypothesis‐free approaches such as proteome or metabolome analyses 2. Kidney activity and damage is likely to be reflected more accurately by biomarkers present in the urine rather than serum. Being easier to obtain in a non‐invasive manner, urine is also an ideal biological sample for diseases requiring repetitive sampling 3.
Surface‐enhanced laser desorption/ionization–time of flight–mass spectrometry (SELDI‐TOF‐MS) has been used frequently as a tool to study the urine proteome, even though it identifies ‘proteomic signatures’ rather than individual proteins 4. None the less, in some instances proteins could be identified by integrating on‐chip sequencing or tandem MS (MS/MS). In one such study 5, a proteomic signature comprising eight proteins was identified as a marker for active LN in children. While four of the proteins turned out to be albumin, the remaining four were identified as transferrin, ceruloplasmin, alpha‐1 acid glycoprotein and lipocalin‐type prostaglandin D synthase. In their subsequent study 6, however, the authors observed that a combination of biomarkers rather than individual proteins could be a better predictor of LN activity. In another SELDI‐TOF‐based study, Zhang et al. 7 identified the signature proteins as hepcIN, alpha‐1 anti‐trypsin and a fragment of albumin. An alternative approach for proteome analysis involving two‐ dimensional gel electrophoresis (2‐DE) and MS/MS is capable of identifying not only proteomic signatures, but also individual proteins. Using this approach, Somparn et al. 8 identified 16 urinary proteins that were expressed differentially between active and inactive LN patients. After validation, they proposed prostaglandin D synthase as a potential biomarker for active LN. Using a combination of 2‐DE and artificial neural network analysis, Varghese et al. 9 identified 11 urinary proteins which could differentiate between LN and certain other nephropathies. In another 2‐DE‐based study, Wu et al. 10 identified 71 proteins from the urine of mice with nephritis induced by injecting preformed anti‐glomerular basement membrane antibodies.
The candidate urinary biomarkers for LN discovered so far have not come into clinical use, as they lack the sensitivity and specificity required for a ‘stand‐alone’ clinical test. Thus, to determine if any new potential urinary biomarkers can be identified that could recognize patients with lupus nephritis and reflect LN activity, we planned to study the urine proteome in SLE using 2‐DE and mass spectrometry. In this endeavour, we laid emphasis upon proteins that were expressed prominently in active nephritis, and not in inactive disease. The identified biomarkers were validated in a larger group of patients and compared with healthy controls, inflammatory disease controls (rheumatoid arthritis patients) and non‐inflammatory proteinuric controls (diabetic nephropathy). To evaluate whether their levels reflect response to treatment, patients with active nephritis were followed longitudinally for 1 year after initiation of induction therapy.
Patients and methods
Patients
SLE patients satisfying the 1997 American College of Rheumatology (ACR) classification criteria 11 seen during a 1·5‐year period and who gave consent were included in the study. Patients with infection, pregnancy or diagnosed cancer were excluded. Based on their disease activity [measured as SLE Disease Activity Index (SLEDAI)] and renal involvement (measured by the four renal components of SLEDAI and expressed as rSLEDAI), they were classified into three groups: active renal disease (AR, with SLEDAI > 4 and rSLEDAI > 0), active non‐renal disease (ANR, with SLEDAI > 4 and rSLEDAI = 0) and inactive disease (ID, with SLEDAI ≤ 4 and rSLEDAI = 0). Urine and serum samples were collected from all patients at baseline. Patients with AR were treated according to the ACR 2012 guidelines 12. At 6 and 12 months’ follow‐up, repeat urine samples were collected from patients with AR disease. Twenty subjects each of rheumatoid arthritis (RA), diabetic nephropathy without renal failure (DM) and healthy controls were enrolled as controls.
Urine collection and protein isolation
The standard protocol for urine collection 13, recommended by the ‘Human Urine and Kidney Proteome Project (HUKPP)’ and ‘European Kidney and Urine Proteomics COST Action (EuroKUP)’ networks, was adopted. Mid‐stream urine specimens, after centrifugation (1000g for 15 min at room temperature) to remove the cells and debris, were divided into 9‐ml aliquots and stored at −80°C.
Urinary proteins were isolated by precipitation using the trichloroacetic acid (TCA)/acetone method, adopting the previously described protocols 10, 13, 14. In brief, 1 ml TCA solution (Sigma, St Louis, MO, USA) was added to each 9‐ml aliquot of urine while vortexing (final TCA concentration = 10%). After overnight incubation (4°C), the tubes were centrifuged (10 000 g for 30 min). Supernatant was removed and the sediment was washed twice, using suspension and centrifugation with chilled acetone, and air‐dried. The sediment in each tube was reconstituted with water (1 ml) and all tubes corresponding to a particular study subject were pooled. After determining protein concentrations 15, the samples were divided into aliquots of 200 µg protein and lyophilized.
Two‐dimensional gel electrophoresis (2‐DE)
A previously described protocol 16 was used, with some modifications. Aliquots (200 µg) of lyophilized urinary protein were dissolved in 200 µl of sample solubilization medium (7 M urea, 2 M thiourea, 4% CHAPS, 0·5% carrier ampholytes, 1% dithiothreitol, 10% isopropanol and 5% glycerol). The solubilized proteins were applied onto immobilized pH gradient (IPG) strips (7 cm, pH 3–10 non‐linear; Bio‐Rad Laboratories, Hercules, CA, USA) in a rehydration tray and left overnight. Isoelectric focusing (IEF) was performed using a three‐step gradient, with the following parameters: 0–250 V in 2 h, 250–2500 V in 4 h and 2500 V for a maximum of 8000 Vh (which led to a ‘steady state’). Maximum current was 50 µA/strip. Later, the IPG strips were equilibrated (10 min each) in solutions ‘A’ [0·05 M Tris‐HCl, pH 8·8 containing 6 M urea, 30% glycerol, 2% sodium dodecyl sulphide (SDS) and 1% dithiothreitol] and ‘B’ (solution ‘A’ without dithiothreitol, but with 4% iodoacetamide and 0·005% bromophenol blue). The strips were loaded onto SDS‐polyacrylamide gel slabs (12% gel) and electrophoresis was performed at a constant current of 15 mA. The gels were stained with Coomassie Brilliant Blue R250 [sensitivity ≤ 50 ng bovine serum albumin (BSA)]. Molecular weight markers (Sigma‐Aldrich, St Louis, MO, USA) of 15–250 kDa range were used.
Gel imaging and image analysis
Gel imaging was performed on the Proteomic Imaging System (Bio Rad Laboratories). PD Quest‐2D software (Bio Rad Laboratories) was used for spot detection and image analysis. To ascertain reproducibility of corresponding spot patterns, repeat 2‐DE was performed for each urine sample. Spots of interest were excised and processed for protein identification.
Protein identification
Proteins (in gel plugs) were processed according to standard techniques 17. In brief, samples were washed, in‐gel reduced, S‐alkylated and digested overnight with trypsin. The extracted peptides were vacuum‐dried, resolubilized and desalted with ZipTips. Matrix‐assisted laser desorption ionization–time of flight–mass spectrometry (MALDI TOF MS/MS) was performed on a 5800 Proteomics Analyzer (Applied Biosystems, Proteomics International, Perth, Australia) using a previously described procedure 18. Peptides were mixed with matrix solution (alpha‐cyano 4‐hydroxy cinnamic acid) spotted on a 384‐well Opti‐TOF plate and analysed using a first run of standard TOF MS. The second run (MS/MS) focused on the 15 most intense peaks of the first run (excluding peaks of trypsin). The laser was set to fire 400 times per spot in MS mode and 2000 times per spot in MS/MS mode. Laser intensity was 2800 J (for MS) and 3900 J (for MS/MS). A mass range of 400–4000 amu with a focus mass of 2100 amu was used. Spectra were analysed using Mascot sequence matching software (Matrix Science, London, UK) and the MSPnr100 database, with taxonomy set to Homo sapiens. Protein identifications were based on the molecular weight search (MOWSE) scoring algorithm, using a confidence level of 95% (P < 0·05). Search parameters were: peptide tolerance ± 0.4; MS/MS tolerance ± 0.4; peptide charge,+1; mass, monoisotopic; missed cleavage, 1.
Validation of identified proteins as biomarkers
Commercially available enzyme‐linked immunosorbent assay (ELISA) kits were used to measure urine and serum levels of retinol binding protein (RBP; Abcam, MA, USA), alpha‐1 anti‐chymotrypsin (ACT; Abcam) and haptoglobin (HAP; R&D Systems, Minneapolis, MN, USA) in SLE patients (AR, ANR and ID) at the baseline. The screening dilution of urine was 1 : 7 for ACT, 1 : 10 for RBP and undiluted for HAP. The samples which were out of range were diluted further. Similarly, 6 and 12 months’ follow‐up samples from AR patients were also tested to assess the change in their levels. The levels of target proteins were correlated with the presence or absence of nephritis, active and inactive lupus nephritis and treatment response. Urinary biomarker protein levels were normalized for urinary creatinine excretion. As controls, urine samples from healthy subjects, patients with rheumatoid arthritis (disease control) and diabetic nephropathy (proteinuric control) were also analysed and values normalized for urinary creatinine excretion.
Statistical analysis
Variables were expressed as median (range). Mann–Whitney U‐test and χ2 tests were used for analysis between two groups and the Kruskal–Wallis test was used for analysis among three or more groups. Spearman's rank correlation was used to assess the correlation between different parameters. A P‐value < 0·05 was considered significant. Multiple logistic regression was performed to identify independent predictors of AR among these three biomarkers. All analysis was performed using spss software.
Results
Definition of the study population
A total of 88 patients of SLE were enrolled into the study. Of these, 44 were classified as AR, 20 as ID and 24 as ANR. The female : male ratio was 10 : 1 and the mean age was 27·5 ± 9·0 years (Table 1). Among 44 patients in the AR group, 35 (80%) underwent renal biopsy before commencement of treatment. Of these 35 patients, three had class II nephritis, 13 each had classes III and IV and six had class V nephritis. The remaining nine patients could not have a renal biopsy for various reasons, such as thrombocytopenia, ascites, etc. However, they were treated according to the most likely histological class (eight as proliferative and one as class II nephritis) based on their urine examination, C3, C4, creatinine and anti‐dsDNA antibody report, and were included into the final analysis. Patients were treated according to the ACR 2012 guidelines 10. The complete response rates at 6 and 12 months were 53 and 66%, respectively. One patient developed chronic kidney disease (CKD) at 6 months and three had relapse of LN at 6, 11 and 12 months, respectively.
Table 1.
Baseline characteristics of systemic lupus erythematosus (SLE) patients in the three categories
| Active renal (AR) | Active non‐renal (ANR) | Inactive disease (ID) | P‐value | |
|---|---|---|---|---|
| Number | 44 | 24 | 20 | |
| F : M | 42:2 | 23:1 | 15:5 | |
| Mean age (± s.d.) | 26·8 (± 8·4) | 29·7 (± 9·0) | 28·9 (± 10) | 0·557 |
| C3 (mg/dl) | 44·9 (16·9–156) | 66·7 (17·3–163) | 120·5 (57·5–194) | < 0·001 |
| Low C3 | 37 | 11 | 1 | |
| C4 (mg/dl) | 7·1 (5·3–55·7) | 11·1 (5·3–32·2) | 22·2 (5·6–44·7) | < 0·001 |
| Low C4 | 38 | 13 | 6 | |
| Anti‐ds DNA (IU/ml) | 200 (13·1–300) | 185 (6·25–300) | 53·6 (6·25–200) | < 0·001 |
| rSLEDAI | 8 (4–16) | 0 | 0 | < 0·001 |
| SLEDAI | 18·5 (5–28) | 10 (5–20) | 1 (0–4) | < 0·001 |
| UPr/UCr ratio | 3·2 (0·4–8·04) | 0·3 (0–1·4) | 0·07 (0–0·8) | < 0·001 |
| Serum creatinine (mg/dl) | 0·9 (0·5–2·9) | 0·8 (0·6–1·4) | 0·8 (0·6–1) | 0·653 |
SLEDAI = SLE Disease Activity Index; rSLEDAI = renal SLEDAI; s.d. = standard deviation; F : M = female : male.
AR and ID patients showed typical urinary protein profiles
Depending on yields following precipitation, urinary proteins of 31 AR and 12 ID patients could be subjected to 2‐DE. In samples from ANR patients and controls, the protein yields were insufficient for proteomic analysis. Representative 2‐DE patterns of AR and ID patients are shown in Fig. 1. While there were certain obvious similarities between the two profiles, some of the spots, falling particularly within two gel areas (boxed in Fig. 1a, b), appeared to be markedly different. A total of 26 protein spots (10 from AR, marked in Fig. 1a and 16 from ID, marked in Fig. 1b) were selected for identification. The selected spots were present either exclusively in one of the samples (A‐1, A‐2, A‐7, A‐9, A‐10, I‐1, I‐2, I‐6, I‐8, I‐10–I‐13) or shared by both (A‐3–A‐6, A‐8, I‐3–I‐5, I‐7, I‐9, I‐14–I‐16).
Figure 1.
Urinary proteins from patients with active renal (AR, Fig. 1a) or inactive disease (ID, Fig. 1b) resolved by two‐dimensional gel electrophoresis (2‐DE). The two samples showed marked difference in pattern of spots falling within two areas of the gels (boxed in the Figure). Ten protein spots from AR (A‐1–A‐10) and 16 from ID (I‐1–I‐16) were processed for protein identification. Spots corresponding to four ID proteins (I‐5, 14–16, Fig. 1b) were also present in the AR sample (indicated by un‐numbered arrows in Fig. 1a). [Colour figure can be viewed at wileyonlinelibrary.com]
Albumin and immunoglobulin light chains were major constituents of the urine proteome
The results of protein identification are shown in Table 2. Constituent protein in two (of 26) spots could not be identified, as one was contaminated with keratin and the other failed to show a significant peptide‐mass match (MOWSE score). Albumin and its degradation products emerged as the major constituents of urine proteome, accounting for nine of the 24 proteins identified. Apart from albumin spots, which were shared by both AR and ID samples, some of the apparently unique spots of the ID sample (I‐1, I‐2, I‐6, I‐10 and I‐11) also turned out to be albumin fragments. Immunoglobulin light chains (A‐4, A‐5 and I‐4) and alpha‐1 anti‐trypsin (A‐9, I‐12 and I‐13) accounted for three spots each, with none being specifically present in either of the samples. Spots corresponding to I‐5 (alpha‐1 microglobulin/bikunin precursor or AMBP) and I‐16 (serotransferrin) were also present in the AR samples (Fig. 1a). AMBP is known to be processed proteolytically to produce alpha‐1‐microglobulin, which helps to regulate the inflammatory processes, and bikunin, which is a urinary trypsin inhibitor. The alpha‐1 acid glycoprotein (spots A‐8 and I‐9) was also shared by both the samples.
Table 2.
Urinary proteins of patients with active renal and inactive disease resolved by two‐dimensional gel electrophoresis (2‐DE) and identified by matrix‐assisted laser desorption ionization–time of flight–mass spectrometry (MALDI‐TOF‐MS/MS)
| Sr. no. | Spot no. | Spot density | Protein | Mascot score* | Sequence coverage (%) | NCBI Accession no. |
|---|---|---|---|---|---|---|
| 1 | A‐1 | 8758 | Haptoglobin | 115 | 9 | P00738 |
| 2 | A‐2 | 12496 | Retinol‐binding protein | 83 | 10 | P02753 |
| A‐4 | 21016 | Immunoglobulin κ‐chain | 76 | 24 | P01620 | |
| 4 | A‐5 | 18982 | Immunoglobulin light chain | 202 | 28 | 149673887 |
| 5 | A‐7 | 25007 | Haptoglobin | 240 | 13 | P00738 |
| 6 | A‐8 | 34642 | Alpha‐1‐acid glycoprotein 1 | 354 | 32 | P02763 |
| 7 | A‐9 | 13362 | Alpha‐1 anti‐trypsin | 574 | 21 | P01009 |
| 8 | A‐10 | 13151 | Alpha‐1 anti‐chymotrypsin | 233 | 10 | P01011 |
| 9 | I‐1 | 9580 | Albumin | 57 | 3 | P02768 |
| 10 | I‐2 | 7392 | Albumin | 94 | 3 | P02768 |
| 11 | I‐3 | 10358 | Albumin | 92 | 4 | P02768 |
| 12 | I‐4 | 4127 | Immunoglobulin light chain | 186 | 28 | 149673887 |
| 13 | I‐5 | 8346 | Alpha‐1 microglobulin/bikunin precursor (AMBP) | 117 | 15 | P02760 |
| 14 | I‐6 | 8187 | Albumin | 232 | 8 | P02768 |
| 15 | I‐7 | 8495 | Albumin | 282 | 11 | P02768 |
| 16 | I‐8 | 27165 | Zinc‐alpha‐2‐glycoprotein | 678 | 34 | P25311 |
| 17 | I‐9 | 43888 | Alpha‐1‐acid glycoprotein 1 | 277 | 16 | P02763 |
| 18 | I‐10 | 9570 | Albumin | 304 | 14 | P02768 |
| 19 | I‐11 | 12390 | Albumin | 213 | 5 | P02768 |
| 20 | I‐12 | 10702 | Alpha‐1 anti‐trypsin | 322 | 17 | P01009 |
| 21 | I‐13 | 20334 | Alpha‐1 anti‐trypsin | 480 | 21 | P01009 |
| 22 | I‐14 | 17776 | Albumin | 439 | 12 | P02768 |
| 23 | I‐15 | 60622 | Albumin | 806 | 15 | P02768 |
| 24 | I‐16 | 23476 | Serotransferrin | 390 | 10 | P02787 |
Spots A‐1–A‐10 were from active renal and I‐1–I‐15 from inactive disease. *Mascot scores > 35 denote significant match (P < 0·05) with protein. Spot A‐3 did not produce a significant score. Spot A‐6 was contaminated with keratin.
Haptoglobin, retinol binding protein and alpha‐1 anti‐chymotrypsin were present exclusively in the AR samples
The proteins HAP (represented by spots A‐1 and A‐7), RBP (spot A‐2) and ACT (spot A‐10) were seen in the AR but not in ID samples (Table 2). Spots corresponding to at least one of these proteins were seen in most (> 80%) of the AR samples which were analysed (Supporting information, Fig. S1a). Consequently, all three proteins were selected for detailed evaluation as potential biomarkers for AR. Conversely, the protein zinc alpha‐2 glycoprotein (ZA2G) was prominently observed only in the ID samples (spot I‐8 in Fig. 1b; also shown in Supporting information, Fig. S1b).
Validation of HAP, RBP and ACT as potential biomarkers for AR
At baseline, median urinary levels of RBP (uRBP) in the AR group were significantly higher than the ID, ANR, HC and RA groups, but they were not different from the DM group. Median urinary ACT (uACT) levels were significantly higher in the AR group compared to all other groups (P < 0·001 for all). Similarly, median urinary HAP (uHAP) was also significantly higher in AR group than the other groups (Table 3, Fig. 2).
Table 3.
Median urinary levels of the three biomarkers in systemic lupus erythematosus (SLE) patients and control groups
| AR (n = 44) | ANR (n = 24) | ID (n = 20) | HC (n = 20) | RA (n = 20) | DM (n = 20) | |
|---|---|---|---|---|---|---|
| URBP/UCr (×100 ng/mg) | 9·65 (0–1612·9) | 1·66 (0–210·8)* | 1·7 (0–297·9)* | 0·14 (0–116·9)** | 0 (0–21·1)** | 75·9 (0–1391·3)n·s· |
| UACT/UCr (×100 ng/mg) | 8 (1·37–32·3) | 1·63 (0–13·23)*** | 0·49 (0·25–8·74)*** | 0·53 (0·22–5·1)*** | 1·8 (0·5–6·87)*** | 1·39 (0·31–3·96)*** |
| UHAP/UCr (×100 ng/mg) | 2·6 (0·02–338·5) | 0·14 (0–5·43)*** | 0·17 (0·01–20·62)** | 0·11 (0–0·71)*** | 0·12 (0·04–1·37)*** | 0·55 (0–4·99)* |
U = urinary; RBP = retinol binding protein; ACT = alpha‐1 anti‐chymotrypsin; HAP = haptoglobin; Cr = creatinine; AR = active renal; ANR = active non‐renal; ID = inactive disease; HC = healthy control; RA = rheumatoid arthritis; DM = diabetes mellitus. *P < 0·05; **P < 0·01; ***P < 0·001 compared to AR group; n.s. = P > 0·05.
Figure 2.
Scatterplots showing baseline median urinary (a) retinol binding protein (RBP); (b) alpha‐1 anti‐chymotrypsin (ACT) and (c) haptoglobin (HAP) values in the three groups of systemic lupus erythematosus (SLE) patients and controls. AR = active renal; ID = inactive disease; ANR = active non‐renal; HC = healthy control; RA = rheumatoid arthritis; DM = diabetes mellitus (*P < 0·05; **P < 0·01; ***P < 0·001; n.s. = non‐significant).
Among the AR patients, baseline uRBP showed modest correlation with spot urinary protein creatinine (PC) ratio (r = 0·39; P < 0·01), rSLEDAI (r = 0·31; P < 0·01) and SLEDAI (r = 0·31; P < 0·01) and poor correlation with anti‐ds DNA antibodies [r = 0·2; P = not significant (n.s.)]. Baseline uRBP did not correlate with baseline serum RBP (sRBP) (r = 0.223; P = n.s.). Baseline uACT showed a modest correlation with PC ratio (r = 0·54), anti‐ds DNA antibodies (r = 0·31), rSLEDAI (r = 0·59) and SLEDAI (r = 0·47) (P < 0·01 for all) and did not correlate with serum ACT (sACT) (r = 0·37; P = n.s.). Baseline uHAP showed modest correlation with PC ratio (r = 0.53; P < 0·001), rSLEDAI (r = 0·6; P < 0·001) and SLEDAI (r = 0·38; P < 0·01) but not with anti‐ds DNA antibodies (r = 0·26; P < 0·05) and baseline serum HAP (sHAP) (r = 0·07; P = n.s.).
Receiver operating characteristics (ROC) analysis for differentiation between AR and ANR revealed that, apart from spot urinary PC ratio, uACT and uHAP were the best markers when compared with the conventional ones. The area under the curve (AUC) for spot urinary PC ratio was 0·93 (0·86–0·99, P < 0·001), for anti‐dsDNA antibodies was 0·61 (0·45–0·77, P = n.s.), for serum C3 was 0·68 (0·55–0·82, P < 0·05), for serum C4 was 0·66 (0·51–0·82, P < 0·05 = n.s.), for uRBP was 0·7 (0·52–0·79, P < 0·05), for uACT was 0·86 (0·76–0·97, P < 0·001) and for uHAP was 0·87 (0·78–0·96, P < 0·001) (Fig. 3).
Figure 3.
Receiver operating characteristic (ROC) curves comparing traditional biomarkers of lupus nephritis and urinary retinol binding protein (RBP), alpha‐1 anti‐chymotrypsin (ACT) and haptoglobin (HAP).
Multiple logistic regression identified ACT as the only independent predictor of AR disease; addition of either HAP or RBP did not improve the predictive model significantly.
All three markers – uRBP, uACT and uHAP – decreased significantly in the AR group at 6 and 12 months of follow‐up with treatment along with conventional markers of disease activity (Table 4 and Supporting information, Fig. S2).
Table 4.
Change in different disease activity parameters and median urinary RBP, ACT and HAP in the active nephritis group with treatment over 1 year
| Baseline | 6 months | 12 months | |
|---|---|---|---|
| Number | 44 | 44 | 44 |
| C3 (mg/dl) | 44·9 (16·9–156) | 95·15 (48·7–148) | 102 (35·9–166) |
| Low C3 | 37 | 4 | 2 |
| C4 (mg/dl) | 7·1 (5·3–55·7) | 19·5 (5·6–61·1) | 19·2 (5·6–79·3) |
| Low C4 | 38 | 12 | 11 |
| Anti‐ds DNA (IU/ml) | 200 (13·1–300) | 45·3 (6·25–300) | 45 (7·9–300) |
| rSLEDAI | 8 (4–16) | 0 (0–4) | 0 (0–8) |
| SLEDAI | 18·5 (5–28) | 2 (0–8) | 2 (0–15) |
| UPr/UCr ratio | 3·2 (0·4–8·04) | 0·39 (0–8·6) | 0·24 (0–6·2) |
| Serum creatinine (mg/dl) | 0·9 (0·5–2·9) | 0·8 (0·46–1·27) | 0·8 (0·4–2·8) |
| URBP/UCr (×100 mcg/mg) | 0·0097(0–1·61) | 0 (0–76·3) | 0 (0–0·29) |
| UACT/UCr (×100 ng/mg) | 8 (1·37–32·3) | 2·4 (0·15–22·7) | 2·6 (0·14–137·44) |
| AP/UCr (×100 ng/mg) | 2·6 (0·02–338·5) | 0·28 (0–48·4) | 0·11 (0–80) |
RBP = retinol binding protein; ACT = alpha‐1 anti‐chymotrypsin; HAP = haptoglobin; SLEDAI = SLE Disease Activity Index; rSLEDAI = renal SLEDAI.
Discussion
A biomarker which can predict nephritis activity in SLE patients will be a valuable tool in the hands of the clinician. In our quest for a robust biomarker, the characteristic urine proteome profiles of AR and ID patients prompted us to focus upon proteins that were present exclusively in either category of patients. Consequently, we selected RBP, ACT and HAP, which were seen in the urine of AR but not ID patients. When validated in a cohort comprising patients and controls, the urinary levels of all three markers were found to be significantly higher in AR compared to ID and ANR patients or the controls. During a 1‐year follow‐up period with treatment, the levels of all three markers decreased significantly in the AR patients.
Two of the three markers identified in this study, RBP and HAP, had also been identified by a few earlier studies, although they were not pursued as potential biomarkers. In a 2‐DE‐based study conducted by Somparn et al. 8, the urine proteome profiles of AR and ID patients were not as distinct as seen by us, due probably to differences in the method for protein extraction or criteria for patient selection. None the less, the authors had noted significantly higher amounts of RBP and HAP in their AR patients. Varghese et al. 9 had also identified a set of proteins, including RBP and HAP, which could differentiate LN from other nephropathies. The urinary proteome of a mouse model of LN comprised a large number of proteins, including HAP 10. A serum proteome study had also shown HAP to be up‐regulated in SLE compared to healthy controls 19.
RBP mediates transfer of vitamin A, and by affecting the intracellular vitamin A level may have an effect on inflammation 20. In circulation, it can remain either free or complexed with pre‐albumin. The bulkier complexed form cannot cross the glomerular membrane. However, the low molecular weight free protein is filtered easily, but most of it is reabsorbed by proximal tubules 21. As even a minor defect in tubular function can enhance excretion of RBP, uRBP has been proposed as the most sensitive marker for loss of proximal renal tubule function in humans 21. Tubular dysfunction is known to occur during flare of the renal disease 22, which could explain the high uRBP levels in our AR patients. It can also explain high uRBP levels in patients of diabetic nephropathy (used as a control group), where tubular dysfunction is a well‐known phenomenon. Some earlier studies have also reported uRBP as a marker of LN disease activity 9, 23, 24. Marks et al. 24 observed that the paediatric lupus patients with ANR disease, who had elevated uRBP levels, went on to develop AR. Consequently, a decline in uRBP following treatment may reflect improvement in tubular function and can be used as a criterion to monitor therapy.
HAP is a glycoprotein which can be bound to haemoglobin in order to prevent loss of iron through the kidney, and also to prevent oxidative damage to the kidney 25. The protein comprises two subunits, alpha and beta, with the alpha subunit having two major allelic forms, alpha‐1 and alpha‐2. The level of the alpha‐2 form has been reported to be higher than alpha‐1 in sera of SLE patients 26. HAP is synthesized mainly in the liver, but may also be expressed in renal tubules in response to either injury or oxidative stress. In mice, HAP is expressed in the kidney following lipopolysaccharide (LPS) treatment 27. We observed higher uHAP levels in AR patients compared to ID and control groups. The lack of correlation between serum and urine levels suggests that the source for uHAP could be kidney. Tubular injury may also cause leakage of HAP from peritubular capillaries into the tubules.
To our knowledge, ACT has not been reported so far as a urinary biomarker for AR disease, even though its enhanced plasma levels have been reported in LN 28. The source of uACT in our AR patients could be either the protein filtered from plasma or local synthesis within the kidney. The lack of correlation between serum and uACT levels, however, indicates towards its local synthesis. Indeed, expression of ACT has been seen in proximal tubular cells and monocyte/macrophage lineage cells within the glomerulus 29. In a study of patients with various nephropathies, an intense ACT staining of glomeruli in renal biopsies was seen in certain types of nephropathies 30. Enhanced synthesis of ACT by the kidney may be viewed as a protective mechanism against renal damage. In its absence, the defence mechanisms geared towards protection of glomerular structures from damage by proteolytic enzymes (in this case chymotrypsin) could suffer a setback.
We found ZA2G to be present exclusively and prominently in the urine of patients with inactive disease. ZA2G is known to be a secreted protein synthesized by epithelial cells of various organs and tissues including the proximal and distal tubules of kidney 31. One of the suggested roles for ZA2G is regulation of fibrosis. Mice with the genetic ZA2G deletion develop significantly more kidney fibrosis, which can be rescued by injection of the recombinant protein 32. It is thus possible that enhanced production of ZA2G by kidney, as suggested by our results, could have an ameliorating effect on AR disease. This observation also finds support from the study by Somparn et al. 8, who concluded that ZA2G is not a marker for AR disease.
Two major urinary proteins in both AR and ID patients were albumin and alpha‐1 anti‐trypsin. They have also been shown as major proteins in the urine of patients with nephrotic syndrome, where at least 50 isoforms of both were characterized by proteome analysis 33. While most isoforms were derived from plasma, the low molecular weight ones were detected only in the urine, suggesting their local formation. Albumin adducts harbouring both carboxy and amino terminal parts of the protein were also detected. In another study, it was shown that the proteolytic activity against albumin and the anti‐proteolytic activity of alpha‐1 anti‐trypsin are probably linked 16. These findings point towards certain specific structural and functional aspects of proteinuria, which may reflect pathogenesis of LN.
The major strengths of our study are the identification of ACT as a new putative biomarker and validation of HAP and RBP as markers for active nephritis in a large number of patients, and the sensitivity of these markers to therapy. The three biomarkers identified reflect renal injury, especially tubular dysfunction, and are thus not specific for LN. They may serve as potential biomarkers of early recognition of nephritis in SLE and as markers to follow activity of LN.
The major limitation, however, is infrequent flares during the follow‐up period, making it difficult to assess the predictive value of biomarkers. Considering that any single biomarker may not suffice as a clinical test, biomarker panels will need to be developed. The three putative biomarkers identified in this study, along with those reported elsewhere, such as monocyte chemoattractant protein‐1 34, would need to be validated on a single platform using a large cohort of patients with adequate incidence of flares in order to identify the best combination for prediction of activity, damage and response to therapy.
Disclosure
The authors declare no disclosures.
Supporting information
Additional Supporting Information may be found in the online version of this article at the publisher's website.
Fig. S1. Representative two‐dimensional gel electrophoresis (2‐DE) patterns in active renal (AR) (Fig. S1a) and inactive disease (ID) (Fig. S1b) patients. In Fig. S1a, top arrow in panels a–c depicts spot of alpha‐1 anti‐chymotrypsin (ACT) and the bottom arrow (c, d) depicts spots of retinol binding protein (RBP). In Fig. S1b, arrow in panels a–d depicts spot of zinc alpha‐2 glycoprotein (ZA2G).
Fig. S2. Temporal trends of SLE Disease Activity Index (SLEDAI), renal SLEDAI (rSLEDAI) and urine protein/creatinine ratio along with (a) retinol binding protein (RBP), (b) alpha‐1 anti‐chymotrypsin (ACT), (c) haptoglobin (HCT).
Acknowledgements
Technical help from Mr A. Yadav and Mr S. Singh is gratefully acknowledged. The project was funded by a grant from Department of Biotechnology, Government of India (Grant no: BT/PR15397/Med/30/604/2011). R. G. was supported partly by Senior Research Associateship of Council of Scientific and Industrial Research and P. S. was supported by Senior Research Fellowship of University Grants Commission, India. This paper has CDRI communication no. 9371.
Contributor Information
A. Aggarwal, Email: aa.amita@gmail.com
S. Sinha, Email: sinha.sudhir@gmail.com
References
- 1. Ward MM. Changes in the incidence of end‐stage renal disease due to lupus nephritis in the United States, 1996–2004. J Rheumatol 2009; 36:63–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Misra R, Gupta R. Biomarkers in lupus nephritis. Int J Rheum Dis 2015; 18:219–32. [DOI] [PubMed] [Google Scholar]
- 3. Soliman S, Mohan C. Lupus nephritis biomarkers. Clin Immunol 2016; doi: 10.1016/j.clim.2016.08.001. [DOI] [PubMed] [Google Scholar]
- 4. Thongboonkerd V. Recent progress in urinary proteomics. Proteomics Clin Appl 2007; 1:780–91. [DOI] [PubMed] [Google Scholar]
- 5. Suzuki M, Wiers K, Brooks EB et al Initial validation of a novel protein biomarker panel for active pediatric lupus nephritis. Pediatr Res 2009; 65:530–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Brunner HI, Bennett MR, Mina R et al Association of noninvasively measured renal protein biomarkers with histologic features of lupus nephritis. Arthritis Rheum 2012; 64:2687–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Zhang X, Jin M, Wu H et al Biomarker discovery for lupus nephritis through longitudinal urine proteomics. Kidney Int 2008; 74:799–807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Somparn P, Hirankarn N, Leelahavanichkul A, Khovidhunkit W, Thongboonkerd V, Avihingsanong Y. Urinary proteomics revealed prostaglandin H2D‐isomerase, not Zn‐α2‐glycoprotein, as a biomarker for active lupus nephritis. J Proteomics 2012; 75:3240–7. [DOI] [PubMed] [Google Scholar]
- 9. Varghese SA, Powell TB, Budisavljevic MN et al Urine biomarkers predict the cause of glomerular disease. J Am Soc Nephrol 2007; 18:913–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Wu T, Fu Y, Brekken D et al Urine proteome scans uncover total urinary protease, prostaglandin D synthase, serum amyloid P, and superoxide dismutase as potential markers of lupus nephritis. J Immunol 2010; 184:2183–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Hochberg MC. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1997; 40:1725. [DOI] [PubMed] [Google Scholar]
- 12. Hahn BH, McMahon MA, Wilkinson A et al American College of Rheumatology guidelines for screening, treatment, and management of lupus nephritis. Arthritis Care Res 2012; 64:797–808. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Mischak H, Kolch W, Aivaliotis M et al Comprehensive human urine standards for comparability and standardization in clinical proteome analysis. Proteomics Clin Appl 2010; 4:464–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Magistroni R, Ligabue G, Lupo V et al Proteomic analysis of urine from proteinuric patients shows a proteolitic activity directed against albumin. Nephrol Dial Transplant 2009; 24:1672–81. [DOI] [PubMed] [Google Scholar]
- 15. Markwell MA, Haas SM, Bieber LL, Tolbert NE. A modification of the Lowry procedure to simplify protein determination in membrane and lipoprotein samples. Anal Biochem 1978; 87:206–10. [DOI] [PubMed] [Google Scholar]
- 16. Sinha S, Kosalai K, Arora S et al Immunogenic membrane‐associated proteins of Mycobacterium tuberculosis revealed by proteomics. Microbiology 2005; 151:2411–9. [DOI] [PubMed] [Google Scholar]
- 17. Shevchenko A, Wilm M, Vorm O, Mann M. Mass spectrometric sequencing of proteins from silver‐stained polyacrylamide gels. Anal Chem 1996; 68:850–8. [DOI] [PubMed] [Google Scholar]
- 18. Maurya VK, Singh K, Sinha S. Suppression of Eis and expression of Wag31 and GroES in Mycobacterium tuberculosis cytosol under anaerobic culture conditions. Indian J Exp Biol 2014; 52:773–80. [PubMed] [Google Scholar]
- 19. Kazemipour N, Qazizadeh H, Sepehrimanesh M, Salimi S. Biomarkers identified from serum proteomic analysis for the differential diagnosis of systemic lupus erythematosus. Lupus 2015; 24:582–7. [DOI] [PubMed] [Google Scholar]
- 20. Fattahi S, Kazemipour N, Hashemi M, Sepehrimanesh M. Alpha‐1 antitrypsin, retinol binding protein and keratin 10 alterations in patients with psoriasis vulgaris, a proteomic approach. Iran J Basic Med Sci 2014; 17:651–5. [PMC free article] [PubMed] [Google Scholar]
- 21. Norden AG, Lapsley M, Unwin RJ. Urine retinol‐binding protein 4: a functional biomarker of the proximal renal tubule. Adv Clin Chem 2014; 63:85–122. [DOI] [PubMed] [Google Scholar]
- 22. Yeung CK, Wong KL, Ng RP, Ng WL. Tubular dysfunction in systemic lupus erythematosus. Nephron 1984; 36:84–8. [DOI] [PubMed] [Google Scholar]
- 23. Sesso R, Rettori R, Nishida S, Sato E, Ajzen H, Pereira AB. Assessment of lupus nephritis activity using urinary retinol‐binding protein. Nephrol Dial Transplant 1994; 9:367–71. [PubMed] [Google Scholar]
- 24. Marks SD, Shah V, Pilkington C, Woo P, Dillon MJ. Renal tubular dysfunction in children with systemic lupus erythematosus. Pediatr Nephrol 2005; 20:141–8. [DOI] [PubMed] [Google Scholar]
- 25. Lim YK, Jenner A, Ali AB et al Haptoglobin reduces renal oxidative DNA and tissue damage during phenylhydrazine‐induced hemolysis. Kidney Int 2000; 58:1033–44. [DOI] [PubMed] [Google Scholar]
- 26. Pavón EJ, Muñoz P, Lario A et al Proteomic analysis of plasma from patients with systemic lupus erythematosus: increased presence of haptoglobin alpha2 polypeptide chains over the alpha1 isoforms. Proteomics 2006; S282–92. [DOI] [PubMed] [Google Scholar]
- 27. D'Armiento J, Dalal SS, Chada K. Tissue, temporal and inducible expression pattern of haptoglobin in mice. Gene 1997; 195:19–27. [DOI] [PubMed] [Google Scholar]
- 28. Sturfelt G, Sjöholm AG. Complement components, complement activation, and acute phase response in systemic lupus erythematosus. Int Arch Allergy Immunol 1984; 75:75–83. [DOI] [PubMed] [Google Scholar]
- 29. Fleming S, Gibson AA. Proteinase inhibitors in the kidney and its tumours. Histopathology 1986; 10:1303–13. [DOI] [PubMed] [Google Scholar]
- 30. Conz P, Bevilacqua PA, Ronco C et al Alpha‐1‐antichymotrypsin in renal biopsies. Nephron 1990; 56:387–90. [DOI] [PubMed] [Google Scholar]
- 31. Tada T, Ohkubo I, Niwa M, Sasaki M, Tateyama H, Eimoto T. Immunohistochemical localization of Zn‐alpha 2‐glycoprotein in normal human tissues. J Histochem Cytochem 1991; 39:1221–6. [DOI] [PubMed] [Google Scholar]
- 32. Sörensen‐Zender I, Bhayana S, Susnik N et al Zinc‐α2‐glycoprotein exerts antifibrotic effects in kidney and heart. J Am Soc Nephrol 2015; 26:2659–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Candiano G, Musante L, Bruschi M et al Repetitive fragmentation products of albumin and alpha‐1‐antitrypsin in glomerular diseases associated with nephrotic syndrome. J Am Soc Nephrol 2006; 17:3139–48. [DOI] [PubMed] [Google Scholar]
- 34. Gupta R, Yadav A, Aggarwal A. Longitudinal assessment of monocyte chemo‐attractant protein‐1 in lupus nephritis as a biomarker of disease activity. Clin Rheumatol 2016; 35:2707–14. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Additional Supporting Information may be found in the online version of this article at the publisher's website.
Fig. S1. Representative two‐dimensional gel electrophoresis (2‐DE) patterns in active renal (AR) (Fig. S1a) and inactive disease (ID) (Fig. S1b) patients. In Fig. S1a, top arrow in panels a–c depicts spot of alpha‐1 anti‐chymotrypsin (ACT) and the bottom arrow (c, d) depicts spots of retinol binding protein (RBP). In Fig. S1b, arrow in panels a–d depicts spot of zinc alpha‐2 glycoprotein (ZA2G).
Fig. S2. Temporal trends of SLE Disease Activity Index (SLEDAI), renal SLEDAI (rSLEDAI) and urine protein/creatinine ratio along with (a) retinol binding protein (RBP), (b) alpha‐1 anti‐chymotrypsin (ACT), (c) haptoglobin (HCT).