Identification of urinary proteins potentially associated with diabetic kidney disease

Selection of patients

We recruited three groups of T2DM patients from Department of Nephrology at Nizam's Institute of Medical Sciences at Hyderabad, India [Table 1]. This included 10 patients each with normoalbuminuria, micro- and macro-albuminuria (urinary albumin = 3–30, 30–300, and >300 mg/dl, respectively). All patients met inclusion criteria of T2DM with and without history of proteinuria. The exclusion criteria were type 1 diabetes, history of hematuria, pregnant or lactating women, malignancy, and liver diseases. Ten healthy individuals from the same population, matched for age, sex, and socioeconomic status and in steady state conditions (defined as the lack of clinical events during the past 6 months to 1 year), were included as controls. They had normal renal function and blood pressure, no proteinuria, and normal urinary sediment and none of them were under any medication. All subjects signed written informed consent. Institutional Ethics Committees of both participating institutions approved the study.

Table 1 Clinical and biochemical characteristics of the study population

Collection, purification, and concentration of urine samples

Approximately, 50 ml of second-morning urine samples were collected using consistent procedures for sample collection, storage, and transportation (www.molmeth.org/protocols/MF43XA). The samples were centrifuged, and supernatant was stored at −80°C until further processing. Desalting and concentration of urinary proteins were done using 3 kDa molecular weight cut-off spin column (Amicon ultra-15 device, Millipore, Billerica, macroalbuminuria) by centrifugation at 4000 × g at 4°C. The protein content was estimated by the Bradford method.

Immunodepletion of urine samples of six abundant proteins

Equal amount of proteins from the concentrated urine samples from healthy individuals and patients from each group was pooled and processed for depletion as per manufacturer's instructions. Urine samples were pooled to avoid the heterogeneity of samples between the patients and ignore the individual differences. We pooled the samples to differentiating each group of patients rather than individuals. Briefly, human six immunoaffinity multiple affinity removal system (MARS column, 4.6 mm × 100 mm, Agilent Technologies, Santa Clara, CA, USA) was connected to a high-performance liquid chromatography system and equilibrated with Agilent buffer-A. We loaded 4 mg of urinary protein from each group onto the column to deplete six abundant proteins (albumin, IgG, IgA, antitrypsin, transferrin, and haptoglobin). Flow-through fractions were collected, concentrated and desalted using 5 kDa molecular weight cut-off spin column (Agilent, Palo Alto, CA). We determined the protein concentration using Bradford assay.

Sodium dodecyl sulfate polyacrylamide gel electrophoresis prefractionation and tryptic digestion

Urinary proteins were reduced with 10 mM dithiothreitol at 60°C for 45 min, followed by alkylation with 50 mM iodoacetamide in 25 mM ammonium bicarbonate at room temperature for 1 h. Urinary proteins were prefractionated on 4–20% gradient precast sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gels (Tris-Glycine midi gels, Invitrogen, CA). Based on molecular weight, the gel was divided into seven parts and sliced manually and destained with destaining solution (50% acetonitrile [ACN] in 50 mM ammonium bicarbonate). Gel pieces were dehydrated with 100% ACN, lyophilized, and digested with trypsin solution (Promega, Madison, WI, USA; 20 ng/μl) at 37°C for 16 h. Tryptic peptide mixtures were extracted using 0.3% trifluoroacetic acid in 50% ACN solution and lyophilized. A total of seven peptide fractions were made each for patient and control groups.

Liquid chromatography-tandem mass spectrometry analysis

We performed nanoflow electrospray ionization and tandem mass spectrometric analysis of peptide samples using linear trap quadrupole (LTQ)-Orbitrap Velos (Thermo Scientific, Bremen, Germany) interfaced with nanoflow liquid chromatography (LC) system. Each peptide fraction was separated on a Bio Basic C18 PicoFrit nanocapillary column (75 μm × 10 cm; New Objective, MA, USA) using a 90 min linear gradient of the mobile phase (5% ACN containing 0.1% formic acid [buffer-A] and 95% ACN containing 0.1% formic acid [buffer-B]) at a flow rate of 200 nl/min and analyzed on LTQ Orbitrap mass spectrometer. The voltage applied for ionization was 1.7 kV. Full scan MS spectra (from m/z 300–2000) were acquired after accumulation to a target value of 10⁻⁶ in the Fourier transform (FT). Resolution in the FT was set to r = 60,000 at MS level. The 20 most intense peptide ions with charge states ≥2 were sequentially isolated to a target value of 5000 and fragmented in linear ion-trap (IT) by collision-induced dissociation (CID) with normalized collision energy of 35% and wideband-activation enabled. The resulting fragment ions were scanned out in the low-pressure IT at the “normal scan rate” (33,333 amu/s) and recorded with the secondary electron multipliers. Ion selection threshold was set at 500 counts for MS/MS and the maximum allowed ion accumulation times were 500 ms (mille second) for full scans and 25 ms for CID-MS/MS measurements in the LTQ-Orbitrap Velos. An activation Q = 0.25 and activation time of 10 ms were used. Each fraction was run in duplicate resulting in generation of 56 raw files from all the four groups, which were searched against protein database.

Identification of proteins and analysis of urine proteome dataset in different groups

We analyzed the data files using SEQUEST search algorithm in Proteome Discoverer (Thermo Fisher Scientific, Beta Version 1.3.0.339, Bremen, Germany) against the human International Protein Index database (version ipiHUMANv378fasta). Search parameters included trypsin as the enzyme with two missed cleavages allowed, oxidation of methionine as a dynamic modification and carbamidomethylation of cysteine as static modification. Precursor and fragment mass tolerance were set to 10 ppm and 0.8 Da respectively. The peptide and protein identifications were obtained using high peptide confidence and top one peptide rank filters. The false discovery rate (FDR) was calculated by enabling the peptide sequence analysis using a decoy database and high confidence peptide identifications were obtained by setting a target FDR threshold of 1% at the peptide level. The proteins thus obtained were analyzed for its gene ontological distribution such as biological processes, molecular function, cellular localization, and associated molecular pathways and cellular networks using GeneGo metacore software analysis (www.genego.com).

Clinical data

The clinical and biochemical characteristics of the study subjects are given in Table 1. The age and sex distribution of patients were comparable in all the groups. Diabetic patients (normoalbuminuria and microalbuminuria) had higher body mass index and waist to hip ratio as well as higher systolic and diastolic blood pressure compared to normal healthy subjects. We noted a progressive increase in serum creatinine, 24 h urinary total proteins, urinary albumin and a reduction in estimated glomerular filtration rate (all P < 0.001) from normoalbuminuria to micro- and macro-albuminuria group of patients indicating progressive decline in renal function with advanced disease. The glycemic control as indicated by glycosylated hemoglobin (Hb) and serum lipid levels also worsened from normoalbuminuria to micro- and macro-albuminuria group of T2DM patients. Thus, all patient groups exhibited metabolic abnormalities as well as evidence of progressive development of nephropathy from normoalbuminuria group to micro- and macro-albuminuria groups.

Proteome profiling of urine samples from diabetic and control groups

All urine samples were efficiently depleted of the six abundant proteins [

Cellular and functional classification of identified urinary proteins

We classified the identified proteins according to Gene Ontology, which allowed annotation of close to 75% proteins [Figure 1]. Based on sub-cellular localization, the majority of the proteins in the macroalbuminuria group were extracellular and membrane-related [43%; [Figure 1a]. Classification on the basis of molecular functions revealed that one-third fraction (33%) were proteins that bind sugar, polysaccharide, glycosaminoglycan, nucleic acids, etc., and a near equal fraction (28%) was involved in catalytic and enzyme regulatory activity [Figure 1b]. Based on biological activity, the majority of the proteins (34%) were noted to be involved in metabolic processes such as carbohydrate and lipid metabolism and others were involved in cellular and developmental processes, biological regulation, and immune response [Figure 1c]. Similar results were observed in other two diabetic groups [Figure 2]. Thus, the majority of the identified proteins belonged to membrane and extracellular class were functionally binding proteins and biologically involved in various metabolic processes.

Close

Figure 1

Gene ontology-based classification of urinary proteins identified in macroalbuminuria group. (a) Sub-cellular localization, (b) molecular function, and (c) biological process

Export to PPT

Close

Figure 2

Comparison of gene ontology-based classification of urinary proteins among different groups. (a) Sub-cellular localization, (b) molecular function, and (c) biological process

Export to PPT

Comparison of urine proteome profile of diabetic patient groups with the control group and among individual diabetic groups

On comparison of the proteome from each patient group independently with the controls, we detected 315, 294, and 218 proteins unique to normoalbuminuria, micro- and macro-albuminuria groups, respectively [Figure 3]. Further comparison among three diabetic groups identified 169, 129, and 132 proteins, respectively in normoalbuminuria, micro- and macro-albuminuria groups only [Figure 4 and Supplementary Tables

Close

Figure 3

Venn diagram representing comparison of urinary proteins identified in various diabetic groups with the control group. Numbers in parentheses represent proteins identified in each group, whereas those within circle represent proteins unique to the specific group. The intersect region of the circles shows proteins common to two comparison groups

Export to PPT

Close

Figure 4

Venn diagram representing comparison of urinary proteome studies of all four groups. Numbers in parentheses represent proteins identified in each group, whereas those within circle represent proteins unique to the specific group. The intersect region of the circles shows proteins common to all diabetic groups

Export to PPT

We detected the majority of proteins commonly reported in the urine of macroalbuminuria patients. This includes proteins such as hemopexin, zinc-α₂ -glycoprotein, α-1-antichymotrypsin, α-1-microglobulin, α-2-HS-glycoprotein, afamin, apolipoprotein (apo) A-I, β-2-microglobulin, and Vitamin D binding protein. In addition, several novel proteins including extracellular matrix protein 1 (ECM1), complement factor H, myoglobin, kininogens, and galectin-3 were also identified. We detected more than 200 low molecular weight (LMW) proteins, of which many have so far not been reported in the urine of macroalbuminuria patients [

We identified several proteins in microalbuminuria group including α-1-acid glycoprotein, α-1B-glycoprotein, α-2-HS-glycoprotein, basement membrane-specific heparan sulfate proteoglycan core protein, β-2-microglobulin, cystatin-B, and transthyretin, which are similar to a large fraction of proteins identified by Andersen et al.[15] In addition, 129 unique proteins such as calreticulin, cartilage intermediate layer protein, CD2-associated protein, guanylate cyclase activator 2B, tubulin β-2C chain, acyl-CoA-binding protein, fatty acid binding protein 1, gastricsin, and desmoglein-2 were also observed in the microalbuminuria group [

We identified 169 unique proteins in normoalbuminuria and several of them such as superoxide dismutase, α-L-fucosidase, complement factor B, galectin-3-binding protein, primary amine oxidase, matrix metalloproteinase-9, isoform five of prominin-1, mucin-6, tenascin XB, and annexin A5 have not been reported earlier [

In addition to the unique proteins, we also observed proteins not present in healthy control group but present in all three diabetic groups and found that 53 proteins were common among three patient groups and were absent in the control group [Figure 4]. We observed that several proteins such as zinc-α₂ -glycoprotein, α-1-microglobulin, α-2-HS-glycoprotein, β-2-microglobulin, transthyretin, apo A-I, and ceruloplasmin are excreted in all patients groups [Table 2]. This may be attributed to loss of negative charge and hyperfiltration of glomerular basement membrane in T2DM. These proteins might have some predictive values and need to study further to qualify for diagnosis of DN. Until that time, this is a descriptive study.

Table 2 List of total proteins identified in all type 2 diabetes groups

Furthermore, we compared normoalbuminuric with proteinuric (micro- and macro-albuminuria) individuals and found 303 proteins were unique to proteinuric group [

We have also done an independent urinary proteomic study on nondiabetic chronic kidney disease (minimal change disease, membranous glomerulonephritis, focal segmental glomerulosclerosis, and lupus nephritis) and compared the DN proteome data. Those observations are very different and found 96 proteins unique to DN only [

Network and pathway analysis

We could map close to 80% of the identified proteome for different molecular and biological network pathways. Alternate complement and blood coagulation network pathways characterized by 25 and 20 proteins, respectively, were most significantly associated with the urine proteome dataset [

Comparison with the published literature

Several studies have used various proteomic techniques to generate urinary proteome in DN but due to their limited sensitivity, the number of identified proteins is small.[8910] To our knowledge, this is one of the high throughput proteomic studies that has identified large number of urinary proteins in well-defined clinical samples of T2DM patients with and without kidney complications. In earlier studies, Sharma et al. identified 99 significantly altered spots in the urine of DN patients by employing two-dimensional differential gel electrophoresis (2DE).[5] Using 2DE, matrix-assisted laser desorption/ionization time-of-flight MS (TOF) technique, Jain et al., identified five altered proteins in the urine of diabetic patients compared to control subjects.[7] In relatively recent studies on macroalbuminuria patients, Rao et al. identified 62 proteins using difference gel electrophoresis and LC-MS/MS approach[13] while Bellei et al. reported 54 proteins by 2DE and electrospray ionization-Q-TOF MS/MS analysis.[23] Compared to their urinary protein profile, we detected 273 proteins that were not reported by either of them [

Table 3 Details of urinary proteins identified in MA patients and common to earlier studies

Identification of reported proteins such as α₂-HS-glycoprotein precursor (fetuin A), zinc-α₂ -glycoprotein, hemopexin, and Vitamin D binding protein in different diabetic groups adds strength to the results from this study. Fetuin A is the most potent circulating inhibitor of calcium phosphorus precipitation and thus contributes to vascular calcification in DN.[24] It is also an important mediator of insulin resistance, thus it is possible that in the presence of nephropathy, fetuin-A exacerbates insulin resistance and thus, worsens the pro-atherogenic milieu among T2DM patients. Zinc-α₂ -glycoprotein has been reported as one of the most abundant urinary proteins in macroalbuminuria and also associated with glomerular lesions.[13232526] Higher levels of plasma hemopexin have been reported in type 2 diabetes, and they are postulated to cause glomerular inflammation and proteinuria by altering glomerular filtration barrier.[2728] Vitamin D binding protein has been consistently associated with type 2 diabetes, glucose intolerance, insulin secretion and/or resistance, and pregnancy associated-diabetes.[2930313233] It is reported to be responsible for immunopathogenesis and progression of the DN.[13]

Identification of novel urinary proteins and their possible role

We also identified large number of novel proteins whose function is still not well understood. However, the majority of them belonged to the category of binding proteins and were involved in biological activities related to carbohydrate and lipid metabolism. Several unique proteins were exclusively detected in different diabetic groups. One such interesting protein identified only in macroalbuminuria was ECM1 that has been implicated in cell proliferation, angiogenesis, and differentiation. Hyperglycemia is known to stimulate ECM1 synthesis in mesangial and endothelial cells through generation of reactive oxygen species, and DN is characterized by excessive accumulation of extracellular matrix in the kidney.[3435] Its relevance is further strengthened with identification of complement factor H, which is a member of the regulators of complement activation family and is associated with obesity and metabolic disorders.[36] Factor H along with adrenomedullin is known to inhibit glucose-induced over-expression of ECM1 in mesangial cells induced, and thus discharge renoprotective functions. Thus, ECM1 may act as a marker of glomerular and tubulointerstitial damage.

We also detected myoglobin in macroalbuminuria patients, which is known to circulate freely in diabetic patients. Uncontrolled diabetes may induce its glycation, which is known to stimulate increased generation of free radicals and consequent renal damage.[37] We identified both high and LMW kininogens, which are established cofactors for coagulation and inflammation and components of the kallikrein-kinin system (KKS). The KKS regulates podocyte apoptosis, and thus glomerular hemodynamics and tubular function.[38] Other proteins such as immunoglobulin k-light chains (KLC) and β-2-microglobulin increase in urine as a result of proximal tubular dysfunction.1939] Similarly, collagen fragments identified in early renal damage in T2DM patients may help in the prognosis and monitoring of DN.[1040] Overall, it may not be unreasonable to speculate that urinary levels of LMW proteins may help in early diagnosis and work better than urine albumin levels in predicting risk of macroalbuminuria.

Of many novel proteins detected exclusively in microalbuminuria patients, CD2-associated protein (CD2AP) has interesting antecedents. A defective expression of CD2AP has been reported in DN and congenital nephrotic syndrome confirming its role in maintaining normal slit diaphragm function.[414243] Even urinary mRNA levels for CD2AP are reported to be significantly higher in DN and positively correlated with urinary albumin excretion, blood urea nitrogen, and serum creatinine.[44] Another protein is cartilage intermediate layer protein, which is known to participate in extracellular matrix structure and remodeling and thus influence insulin sensitivity.[45]

We also identified many novel proteins such as KLC, transthyretin, and retinol binding protein in urine of normoalbuminuric individuals. The current assumption is that release of these proteins may be due to disturbance of tubular function in diabetes, and hence their importance is a matter of speculation.

Network pathway analysis of proteome dataset identifies stage-specific proteins

Network and pathway analysis of the urinary proteome identified alternative complement and blood coagulation as two major pathways likely to be involved in the pathogenesis of renal complications of type 2 diabetes. Identification of common complement factors in micro- and macro-albuminuria indicates their potential role in initiation of renal complications in type 2 diabetes patients. CD59 is known to restrict MAC assembly by interacting with the terminal complement proteins C8 and C9 and thus protect kidney from complement-mediated attacks. Inactivation of CD59 due to glycation makes kidneys susceptible to increased MAC deposition and contributes to extensive microvascular damage, development of atherosclerosis, and diabetic complications.[4647] In fact, glycated human CD59 has been identified in urine of diabetic patients and correlates with the levels of glycated Hb.[48] Presence of CD59 in all the groups but absence of MAC in control group indicates loss of CD59 due to glycation and might predict development of kidney disease in diabetes. Subsequent progression to microalbuminuria or macroalbuminuria may depend on interaction of specific complement factors such as C5 and C8 groups, which control local pro-inflammatory response and MAC assembly.[49] These proteins and factor H were identified exclusively in macroalbuminuria patients, suggesting their importance as stage-specific markers. In addition to its influence on extracellular matrix, complement factor H is known to accelerate the destructive action of factor I, which is a regulator molecular of the alternative pathway for immune responses.

It was equally interesting to have similar observations in the blood coagulation pathway. Presence of fibrinogen beta, bradykinin, and alpha-1 antitrypsin indicates activated extrinsic and intrinsic coagulation pathways in diabetes. However, occurrence of clotting factors such as thrombin and fibrinogen in both micro- and macro-albuminuria suggests hypercoagulable state that is characteristic of renal complications of diabetes. Further progression to macroalbuminuria is related to dysregulation of balance between blood coagulation and fibrinolysis. This state is exemplified by exclusive expression of factor X that participates in the formation of prothrombinase complex, and that cleaves prekallikrein to kallikrein, which catalyzes bradykinin formation and of plasminogen, SERPINF2 which resulting in formation of protease-inhibitor complexes in the macroalbuminuric patients. Further quantitative analyses may help to understand whether their levels correlate with the intensity of renal damage.

This study has several strengths and few limitations. First, we used urine samples from well-characterized T2DM patients followed over last few years. Second, the analysis used the most sensitive tools, which allowed us to identify large number of proteins with high confidence. Finally, we identified several proteins, which were reported earlier in DN patients and thus built confidence in the overall urinary proteome dataset. One of the limitations is the absence of validation of these results and the lack of quantitative analysis of these proteins.

In spite of using very high throughput technology, we paved with few limitations. We pooled urine samples to overcome sample preparation and LC-MS/MS analysis costs. In the pooled samples, the biological variance between pools is reduced compared with individual samples, but practical consequence of this is that power could be reduced. As we did not validate our identifications, we cannot say that those proteins can be used as predictive biomarkers unless they are validated in large sample size.

In summary, this study reaffirms the role of several established proteins in the pathogenesis of DN and also identifies large number of novel proteins, specific to different stages of DN. We propose an important role of alternate complement and blood coagulation pathways, which fits with the currently existing hypothesis in the development of microvascular complications of diabetes. This study provides a resource useful for further exploration of these proteins as markers for various stages of diabetes.

Financial support and sponsorship

The funds for this study were provided by the Council of Scientific and Industrial Research (CSIR), Ministry of Science and Technology, Government of India, India under XII 5-year Plan grant THUNDER (to GRC).

Conflicts of interest

There are no conflicts of interest.