Early prediction markers of diabetic nephropathy

Abstract

The present invention concerns a method for the in vitro detection of an increased risk of diabetic nephropathy in a subject suffering from diabetes and being normoalbuminuric. Another aspect of the invention pertains to a method for the in vitro identification of a marker for prediction of diabetic nephropathy. Finally, the invention concerns a kit comprising means for detecting at least two proteins selected from the group consisting of heparan sulfate proteoglycan core protein or fragments thereof, carbonic anhydrase 1, prothrombin or fragments thereof, tetranectin, CD59 glycoprotein, plasma serine protease inhibitor, mannan-binding lectin serine protease 2 or isoforms thereof, antithrombin-III, alpha-1-antitrypsin, collagen alpha-1(I) chain, alpha-enolase, histone H2B type 1-O, glutaminyl-peptide cyclotransferase, protein AMBP and zinc-alpha-2-glycoprotein.

Claims

1. A method for identifying risk of developing diabetic nephropathy in a normoalbuminuric diabetic and treating comprising: (a) obtaining a urine sample from a normoalbuminuric diabetic subject, wherein the subject is capable of developing microalbuminuria after a physical exercise test, (b) measuring the protein expression level of carbonic anhydrase 1 in the urine sample, (c) identifying the normoalbuminuric diabetic subject of step (a) at increased risk of developing diabetic nephropathy when the protein expression level of carbonic anhydrase 1 is at least 10% higher than a pre-determined value, wherein the predetermined value is a level of the carbonic anhydrase 1 protein measured in a urine sample obtained from a control normoalbuminuric diabetic subject suffering from diabetes and determined not to be at risk for diabetic nephropathy by either absence of microalbuminuria after the physical exercise test or absence of clinical symptoms for at least 2 years; and (d) administering drugs that lower blood pressure to the identified subject of (c), wherein the drugs are selected from angiotensin-converting enzyme (ACE) inhibitors or angiotensin II receptor blockers (ARBs).

2. The method according to claim 1, wherein said normoalbuminuric diabetic subject has type-1 or type-2 diabetes.

3. The method according to claim 1, wherein said measuring the expression level of step (b) comprises gel electrophoresis, 2D gel electrophoresis, mass spectrometry, or an immunological assay to determine the expression level of carbonic anhydrase 1.

4. The method according to claim 3, wherein said mass spectroscopy is targeted mass spectroscopy or LC/MS-MS.

5. The method according to claim 3, wherein said immunological assay is an ELISA, a multiplex immunoassay or an antibody array.

6. The method according to claim 1, said method further comprising: obtaining a second urine sample from the normoalbuminuric diabetic subject, said urine sample being taken from the subject after a physical exercise test.

7. The method according to claim 6, wherein the exercise test comprises a cycle ergometer.

Description

BRIEF DESCRIPTION OF THE FIGURES

(1) FIG. 1: Definition of the four groups of urine samples of diabetic patients.

(2) FIG. 2: Quality assessment of the series of 2D-GE experiments by a dispersion tree approach.

(3) FIGS. 3A-3B: Quantification of the Western-blot signals obtained for the analyzed biomarkers in native urine samples of patients at risk of developing diabetic nephropathy (G4) and “control” patients (G2).

(4) FIGS. 4A-4B: Quantification of the Western-blot signals obtained for the analyzed biomarkers in urine samples of healthy subjects and diabetic patients suffering from nephropathy.

(5) FIG. 5: ROC curves of CA1+CD59 combination (mROC method) for G1 versus G3 and G2 versus G4 comparison.

EXAMPLES

Example 1: Materials and Methods

(6) 1.1. Recruitment of Patients

(7) Twenty-six type 1 diabetic patients with a good glycemic control (all of them were equipped with an insulin pump) were recruited from patients of AMTIM in the Lapeyronie Hospital at Montpellier. The inclusion criteria were the following: type 1 diabetic patient with an average duration of diabetes of more than 5 years, no MA (defined as albumin/creatinine ratio (ACR)<30 mg/g). All patients gave written informed consent to participate in the study. The study was approved by regulatory authorities (approval of the Mediterranean Committee for the Protection of Persons, No. 2006-A00162-49).

(8) 1.2. Exercise Test and Urine Sample Collection

(9) Patients were directed to the Central Service of Clinical Physiology, unit of metabolic exploration (CERAMM), Lapeyronie Hospital, Montpellier, France. At their arrival, mainly in the morning, the patients were given 200 ml of water every 20 min starting 1 h before the exercise test. The morning midstream clean-catch urine sample of each patient was collected just before starting the test. Then, all participants performed an exercise test on a cycle ergometer. To individualize the increment of exercise intensity during exercise test, the workload of each step was calculated from the theoretical maximal aerobic power (W.sub.max), i.e., power corresponding to the theoretical maximal oxygen consumption (VO.sub.2max). In consequence, the subjects underwent a test with the same relative incremental workload and were compared at the same percentage of their W.sub.max. The test consisted of six-minute steady-state workloads at 20, 40, 60 and 80% of W.sub.max. At the end of 24 minutes of exercise, all the patients rested for 1 hour and drank water. After this resting phase, the urine sample of each patient was collected. All urine samples were collected in sterile cups containing 1 mM sodium orthovanadate, 10 mM sodium fluoride, 20 mM glycerol 2-phosphate disodium (Sigma Aldrich, St. Louis, Mo.) and protease inhibitors (Roche Diagnostics, Meylan, France), and stored at −80° C. until use. Collection of urine samples was performed in a reproducible manner and in the same period.

(10) 1.3. Study Design

(11) Urinary albumin/creatinine ratio (UACR) was determined for all samples before and after exercise test. Depending on UACR in samples collected after exercise test, the patients were divided into two cohorts: 1) a control cohort consisting of 14 patients whose urine remain negative for microalbuminuria (MA−) (UACR<30 mg/g) after the exercise test; and 2) a cohort considered at risk for developing DN consisting of 12 patients becoming positive for microalbuminuria (MA+) (30<UACR<300 mg/g) after the exercise test. The clinical characteristics of the patients are shown in Table 2.

(12) TABLE-US-00002 TABLE 2 Clinical characteristics of type 1 diabetic patients. Cohort Clinical data Control At risk Number (n) 14 12 Age (years) .sup. 48 ± 10.9 43.9 ± 10.4 Sex (women/men) 8/6 8/4 Duration of diabetes (years) 20.8 ± 7   19.6 ± 12.sup.  Body mass index (kg/m2) 25.9 ± 4.1  23.06 ± 2.5  HbA1C (%) 7.6 ± 1.2 7.1 ± 0.8 Systolic Before exercise 124 ± 13  116 ± 13  blood After exercise  .sup. 171 ± 17 .sup.(b)  .sup. 182 ± 20 .sup.(b) pressure (mm Hg) Diastolic Before exercise 86 ± 35 69 ± 9  blood After exercise 76 ± 10 75 ± 9  pressure (mm Hg) Creatinuria Before exercise 0.46 ± 0.26 0.40 ± 0.32 (g/L) After exercise 0.31 ± 0.25 0.34 ± 0.29 Albuminuria Before exercise 3.7 ± 2.2 4.5 ± 3.8 (mg/l) After exercise 4.4 ± 6.5   .sup.  48 ± 26.1 .sup.(a) (b) UACR Before exercise .sup. 9 ± 4.6 13. ± 7.1 (mg/g) After exercise 12.9 ± 7.2    .sup. 178.6 ± 117.3 .sup.(a) (b) Values are expressed as mean ± S.D. .sup.(a) P < 0.05 compared with control cohort; .sup.(b) P < 0.05 compared with the same cohort after exercise test.

(13) Thus, 4 different groups of urine samples were defined (FIG. 1): before the exercise test, the G1 group corresponds to samples collected from the control cohort and the G3 group collected from the cohort at risk; and after the exercise test, the G2 group corresponds to samples collected from the control cohort and the G4 group collected from the cohort at risk.

(14) 1.4. Urine Sample Preparation

(15) Urine was centrifuged at 4° C. for 30 min at 11,000×g and the supernatant was dialyzed against 18 MΩ.cm water at 4° C. for 48 hours. Then, the sample was concentrated using 5000 Da cut-off centrifugal tube (Millipore, Bedford, Mass.) at 4° C. to approximately 1/40 of the initial volume. The concentrated urine was lyophilized and then solubilized in lysis buffer containing 8 M urea, 2 M thiourea, 4% w/v CHAPS, 65 mM DTE, 40 mM Tris-base and protease inhibitors for 2 h at room temperature on a rotating wheel. Protein amount was estimated using RCDC protein assay kit (Bio-Rad, Hercules, Calif.).

(16) 1.5. 2D-GE

(17) 18 cm long precast IPG strips with nonlinear immobilized pH 3-10 gradient were rehydrated with 170 μg of protein sample overnight in a solution containing 8 M urea, 2 Mthiourea, 4% w/v CHAPS, 65 mM DTE, 0.0025% v/v bromophenol blue and 1% v/v IPG buffer (3-10). Isoelectric focusing was carried out on the Ettan™ IPGphor™ isoelectric focusing system at 20° C. to a total amount of 50 kVh. After the first dimensional run, the proteins were reduced (6 M urea, 50 mM Tris-HCl, pH 8.8, 30% v/v glycerol, 2% w/v SDS, 0.001 v/v bromophenol blue and 65 mM DTT) and alkylated for 10 min in the same buffer containing 135 mM iodoacetamide instead of DTT. Then, proteins were separated in the second dimension on homemade 12% SDS-polyacrylamide gels using an ISO-DALT electrophoresis unit at a constant voltage of 120 V overnight at 10° C. The gels were stained with Sypro Ruby fluorescent dye (Bio-Rad, Hercules, Calif.).

(18) 1.6. Image Analysis

(19) Gel images were digitalized individually with a Typhoon 9200™ scanner (GE Healthcare, Uppsala, Sweden) at 50 μm resolution with the photo multiplier tube (PMT) voltage adjusted for maximum range without signal saturation. Gel images were analyzed using Progenesis SameSpot software v3.0 (Nonlinear Dynamics, Durham, UK). Gels were warped to align images and protein spots were automatically detected. This software is based on the concept of recursive gel matching, which means that each gel of a matching set is recursively used as “reference gel” once during the matching process. The quality of the automatic match was critically evaluated in each case, and, if necessary, corrections were done manually.

(20) 1.7. Analysis of Data Quality

(21) The Phylopuce method (Copois, Bibeau et al. 2007) was used to estimate the homogeneity of 2D-GE experiments so as to determine those of eventual bad quality. Briefly, each gel is represented by an expression vector of dimension n (n being the number of spots). The Euclidian distance between vectors (representing all experiments and their normalized spot intensities) was calculated. The resulting distance matrix was used to construct a phylogenetic tree using the “Kitsch algorithm”. To visualize the obtained tree, the “Drawtree algorithm” which gives a graphical representation of an unrooted tree was used. This approach of classification based on distances between two dimensional gels establishes the overall homogeneity between gels and eventually pinpoints an experimental bias or the particular behavior of a sample.

(22) 1.8. Statistical Analysis

(23) All statistics were computed with the “R/Bioconductor” statistical open source software (Gentleman, Carey et al. 2004). The differential intensity levels of protein spots between groups were analyzed using different statistical tests: Wlcoxon's test (Multtest package), Welch's test (Multtest package), VarMixt method (VarMixt package) and SAM method (siggenes package). With the multiple testing methodologies, it is important to adjust the p-value of each protein spot to control the False Discovery Rate (FDR). The Benjamini and Hochberg procedure (Benjamini 1995) was applied on all statistical tests and an adjusted p value less than 0.05 was considered as statistically significant. The AUC (area under the curve) ROC (receiver operating characteristic) was also calculated with the ROC package and an AUC ROC value greater than 0.75 was considered as significant.

(24) For each significantly differential spot between two patient groups with one of the statistical tests used, a value was assigned to this spot according to Table 3. Thus all spots had a total score for all statistical tests between 0.5 and 5. Only spots with a score greater than or equal to 2 were included in the differential analysis.

(25) TABLE-US-00003 TABLE 3 Value assigned to each statistical test and the total score for the spot Non adjusted test Adjusted test Other Total Wilcoxon Welch Wilcoxon Welch VarMixt SAM AUC score 0.5 0.5 0.5 0.5 1 1 1 5

(26) 1.9. In-Gel Digestion

(27) Spots were excised from gels with a Propic robot (Perkin-Elmer, Wellesley, Mass.). All subsequent steps were done automatically using a Multiprobe II robot (Perkin-Elmer, Wellesley, Mass.). Spots were first washed with 300 μl of water and then 300 μl of 25 mM NH4HCO3. Destaining was performed twice in the presence of 300 μl of 50% acetonitrile in 25 mM NH4HCO3. Gel pieces were then dehydrated twice by 300 μl of 100% CH3CN, and finally dried at 37° C. for 1 h. Eight microliters of a trypsin solution (Sequencing Grade Modified Trypsin, Promega, Madison, Wis., USA), at a concentration of 0.0125 μg/μl in 25 mM NH4HCO3, was added to every spot and the samples were kept for 15 min first on ice and then at room temperature. Digestion was performed overnight at 37° C. and was stopped by addition of 2 μl of 2% formic acid. Digests were sonicated in an ultrasonic bath for 10 min and supernatants were transferred into HPLC polypropylene tubes.

(28) 1.10. Mass Spectrometry

(29) The protein digests were analysed using a High Capacity ion trap mass spectrometer (Esquire HCT; Bruker Daltonik GmbH, Bremen, Germany), interfaced with a nano-HPLC Chip-Cube system (Agilent Technologies, Santa Clara, Calif., USA). The chips contained both the pre-column and the column (Zorbax 300SB-C18; Agilent Technologies, Santa Clara, Calif., USA). Samples were first loaded onto the 4 mm enrichment cartridge at a flow rate of 4 μl/min using 0.1% formic acid. After pre-concentration, peptides were separated on the column (75 μm diameter, 43 mm length) at a flow rate of 0.3 μl/min using a 15 min linear gradient from 3% to 80% acetonitrile in 0.1% formic acid, and eluted into the mass spectrometer. A capillary voltage of 1.8-2.1 kV in the positive ion mode was used together with a dry gas flow rate of 4.5 l/min at 250° C. A first full-scan mass spectrum was measured in the 310 m/z to 1800 m/z range, followed by a second scan at higher resolution to measure precisely the mass of the three major ions in the previous scan. Finally, a third scan was performed to acquire the collision-induced MS/MS spectra of the selected ions. MS/MS raw data were analyzed using Data Analysis software (Bruker Daltonik GmbH, Bremen, Germany) to generate the peak lists. The NCBI non-redundant database (NCBInr, release 20101018) was queried locally using the Mascot search engine (v. 2.2.04; Matrix Science, London, U.K.) with the following parameters: Homo Sapiens for the taxonomy, trypsin as enzyme, 1 missed cleavage allowed, carbamidomethylation of Cysteine as fixed modification, oxidation of Methionine as variable modification, and 0.6 Da mass accuracy in both MS and MS/MS. Under these conditions, individual ion scores above 40 indicated identity or extensive homology (p<0.05) and proteins were validated once they showed at least one peptide over this threshold.

(30) 1.11. Validation by Western Blotting

(31) For Western blotting, 30 μg of each urine samples from the G1, G2, G3 and G4 groups (n=4, in each group) were resolved with SDS-PAGE at 160 V for approximately 2 h using SE260 mini-Vertical Electrophoresis Unit (GE Healthcare, Uppsala, Sweden). Then, proteins were transferred onto a nitrocellulose membrane and non-specific binding was blocked with 5% (w/v) skim milk in PBS-Tween 0.1% at 4° C. overnight. The membranes were then incubated with the appropriate primary antibodies (Table 4) at room temperature for 2 h. After washing, the membranes were further incubated with appropriate secondary antibodies anti-whole molecule IgG conjugated with horseradish peroxidase (Table 4) at room temperature for 1 h. Reactive protein bands were detected by enhanced chemiluminescence (ECL) (GE Healthcare, Uppsala, Sweden) using an autoradiogram.

(32) TABLE-US-00004 TABLE 4 List of primary and secondary antibodies used for Western blotting HRP Primary Provider conjugated Provider antibody primary Concen- secondary secondary anti- antibody Reference tration antibody antibody Dilution CA1 Sigma HPA006558 0.1 μg/ml anti-rabbit Sigma 1:150,000 IgG CD59 Sigma HPA026494 0.1 μg/ml anti-rabbit Sigma 1:150,000 IgG CLEC3B R&D AF5170 0.2 μg/ml anti-sheep Santa Cruz 1:40,000 Systems IgG Biotech- nology COL1A1 Santa Cruz sc-28657 1 μg/ml anti-rabbit Sigma 1:150,000 Biotech- IgG nology HSPG2 R&D AF2364 0.2 μg/ml anti-goat Sigma 1:300,000 Systems IgG F2 Santa Cruz sc-33769 1 μg/ml anti-rabbit Sigma 1:150,000 Biotech- IgG nology MASP2 Sigma SAB1401534 2 μg/ml anti-rabbit Sigma 1:150,000 IgG SERPIN Sigma HPA001292 0.02 μg/ml anti-rabbit Sigma 1:150,000 A1 IgG SERPIN R&D AF1266 0.2 μg/ml anti-goat Sigma 1:300,000 A5 Systems IgG SERPIN Sigma HPA001816 1 μg/ml anti-rabbit Sigma 1:150,000 C1 IgG

Example 2: Clinical Characteristics of Diabetic Patients and Proteomic Data Quality

(33) Based on the assumption that the appearance of microalbuminuria after exercise test is predictive of the onset of persistent MA after 10 years (O'Brien, Watts et al. 1995), the inventors have used a controlled exercise test to classify patients with type 1 diabetes in two cohorts: a cohort of control patients (14 patients remaining negative for MA after the exercise) and a cohort of patients at risk for DN (12 patients who became positive for MA after the exercise). For each cohort, urine samples were collected before (G1 and G3) and after (G2 and G4) exercise test (FIG. 1).

(34) Clinical characteristics of the two cohorts of type 1 diabetic patients, control cohort and cohort at risk of developing DN, are described in Table 2. No significant differences were observed for age, duration of diabetes, systolic and diastolic blood pressure before exercise test, creatinuria and HbA1c among the two cohorts of patients. Compared with diabetic controls, diabetic subjects at risk of developing DN have higher levels of albuminuria and UACR after exercise test. The average level of systolic blood pressure was significantly higher after exercise than before for both cohorts.

(35) The protein profiles of each urinary sample were studied by 2D-GE. After image analysis, 768 protein spots were visualized on each gel and the relative abundance of the protein spots was determined. The inter-experiment reproducibility was also assessed (average CV=19%).

(36) To evaluate the global quality of the 2D-GE data, the Phylopuce method was used to analyze the dispersion of experiments. This allows representing the series of experiments graphically in the form of an unrooted tree (FIG. 2). Ideally, a circle passing through all branches of the tree would reflect a perfect homogeneity of 2D gels. An analysis of the first results by Phylopuce showed that some gels were lying out from the remainder of the other gels (data not shown). By exploring the images of these distant gels, we noted a low quality due to bad migration of proteins with distortion and/or the presence of protein streaks. The samples corresponding to these gels were migrated again in 2D gels, which yielded to satisfactory quality 2D gel. The resulting unrooted tree in FIG. 2 shows an acceptable dispersion of the gels. Three of them behave, however, slightly differently, probably due to particular intrinsic properties of these samples and not to experimental problems since examination of these gels showed that they did not present apparent defects. It was decided to use all gels in further differential analysis.

Example 3: Comparative Analysis of Protein Expression

(37) Urine samples were collected before and after exercise test from all patients from the two cohorts: control cohort (MA negative before and after exercise test) and at risk cohort (MA negative before test and MA positive after exercise test) yielding to four groups of urine, as described above.

(38) Each group of urine samples (G1-G4) matched to a proteome of particular interest (Table 5) which can be revealed by differential analysis. Thus, the proteome profile of the G1 group (urine samples of control diabetic patients collected before the exercise test) contains all proteins shared between control type 1 diabetic patients. The proteome profile of the G2 group (urine samples of control diabetic patients collected after the exercise test) reflects all proteins whose urinary excretion was increased or decreased under the effect of physical activity among control patients. A set of proteins so-called “exercise proteome of control patients” is revealed by the comparison of protein profiles from urine samples of control patients collected before the exercise test (G1 group) with those collected from same patients after exercise (G2 group), i.e. the G1G2 comparison.

(39) The G3 group (urine samples of diabetic patients at risk collected before exercise test) may include the candidate biomarkers for the early diagnosis of DN revealed without any physical activity. These biomarkers could be identified by comparing protein profiles from urine samples collected before the exercise test from control patients (G1 group) with those from at risk patients (G3 group), i.e. the G1G3 comparison.

(40) The G4 group (urine samples of diabetic patients at risk for developing DN collected after exercise test) reflects biomarkers which are not differential before exercise test but become differential versus control after the exercise test. These biomarkers are revealed by the comparison of protein profiles from urine samples collected after the exercise test from control patients (G2 group) with those from at risk patients (G4 group), i.e. G2G4 comparison. The intersection of G2G4 comparison with G1G3 comparison reflects the candidate biomarkers for the early diagnosis of DN which are differential before and after exercise.

(41) Furthermore, since it is collected after the exercise test, G4 group reflects “the exercise proteome of at risk patients”. This proteome is highlighted by comparing the protein profiles of the urine samples collected from at risk patients before the exercise test (G3 group) with those collected from the same patients after exercise (G4 group), i.e. the G3G4 comparison.

(42) TABLE-US-00005 TABLE 5 Each group of urine samples (G1-G4) and comparisons between them reflect a particular proteome and revealed biomarkers of DN Group Reflection of Comparison G1 Proteins shared between control type 1 diabetic — patients G2 The exercise proteome of control patients G1G2 G3 Candidate biomarkers of DN before exercise G1G3 test G4 Candidate biomarkers of DN only after exercise G2G4 test Candidate biomarkers of DN before and after G2G4 ∩ G1G3 exercise test The exercise proteome of at risk patients G3G4

(43) These four comparisons, G1G3, G2G4, G1G2 and G3G4, were performed to select potentially informative differences in protein expression. Several stringent criteria to select protein spots for further analysis were imposed. First, differences in protein expression were considered statistically significant if the total score calculated for each protein spot (representing the contribution of seven statistical tests) was greater than or equal to 2 (see Statistical analysis in the Materials and Methods section). Second, differential protein spots should be present in at least 50% of gels from one group compared to the other. Third, protein spots corresponding to albumin were eliminated (in G2G4 and G3G4 comparisons).

(44) These analyses resulted in a total of 177 protein spots showing differential intensity. These protein spots derived from different comparisons, as follows.

(45) For the detection of early DN biomarkers by comparing control cohort with at risk cohort:

(46) before exercise test i.e. G1G3 comparison: 14 spots, including 8 up-regulated and 6 down-regulated in G3; and after exercise test i.e. G2G4 comparison: 156 spots, including 104 up-regulated and 52 down-regulated in G4.
To identify the exercise proteome of controls and at risk diabetic patients, by comparing samples from the same cohort before and after exercise test: in control cohort i.e. G1G2 comparison: 5 spots, including 2 up-regulated and 3 down-regulated in G2; and in at risk cohort i.e. G3G4 comparison: 101 spots, including 72 up-regulated and 29 down-regulated in G4.

(47) Overall, the number of differential protein spots between control cohort and at risk cohort was higher after exercise test (G2G4 comparison) than before (G1G3 comparison), suggesting that exercise has different effects on the protein urinary excretion in diabetic patients at risk for DN than in control patients. The G1G3 comparison reveals early candidate biomarkers DN between normoalbuminuric patients without performing the exercise test. Only three protein spots were specific of the G1G3 comparison; others were also found in G2G4, G3G4 and G1G2 comparisons. The exercise proteome of control patients appeared small (G1G2 comparison) with only 5 protein spots differentially expressed. At the opposite, the exercise proteome of at risk patients is more important with 101 differential spots expressed (G3G4 comparison).

(48) There were some spots that overlap between the different comparisons. Four sets of differential spots which drew more attention were: Set of 12 spots shared between the G1G3 and G2G4 comparisons. Therefore, these spots were differentially expressed between cohorts of control patients and at risk patients, before and after the exercise. Set of 58 spots specific for the G2G4 comparison; they were differentially expressed between control patients and at risk patients after the exercise test. Set of 38 spots specific for the G3G4 comparison; they were differentially expressed between the urine samples collected from at risk patients before and after the exercise test.

(49) Graphs of statistical AUC ROC values of these sets of spots in the different comparisons show that most of the spots had AUC greater than 0.75. Interestingly, two protein spots were specific of G1G3 comparison, as two protein spots were specific of G1G2 comparison. Two other spots were shared between the G1G3, G2G4 and G3G4 comparisons. These spots as the set of 80 spots were part of candidate biomarkers as well as the exercise proteome of at risk patients, which explains their potential diagnostic interest.

Example 4: Identification of Differentially Expressed Proteins

(50) The majority of spots that were significantly differential in all the comparisons were extracted from gels, digested with trypsin, and prepared for mass spectrometric (MS) analysis. These spots were identified by Nano LC-MS/MS. The total number of identified proteins was 73. All of these proteins were present in the DUP database constructed by the inventors (see Worldwide Website: sysdiag.cnrs.fr//index.php?page=dup), which contains more than 3000 non redundant urinary proteins described in the proteomic analysis of human normal urine in 19 publications (release 20110210).

(51) From the statistical and functional analysis, 38 protein spots were selected which correspond to 24 different proteins identified by mass spectrometry. The expression level (up or down-regulated) and the molecular weight on 2D-GE of these 24 proteins are summarized in Table 6. All proteins are differentially expressed in post-exercise urine between at risk patients (G4 group) and that of control patients (G2 group). Carbonic anhydrase-1 (CA1), plasma protease C1 inhibitor (SERPING1), glutaminyl-peptide cyclotransferase (QPCT), protein AMBP, zinc-alpha-2-glycoprotein (AZGP1) and CD59 glycoprotein (CD59) are also differentially expressed in pre-exercise urine (G1G3 comparison). Endothelial protein C receptor (EPCR) is the only protein differentially expressed between pre-exercise urine of at risk patients (G3 group) and that of control patients (G1 group) (Table 6).

(52) TABLE-US-00006 TABLE 6 Differentially expressed proteins in the 2D-GE comparison between control patients and at risk patients. Each protein was identified from one or several spots corresponding to a fragment or the full length of the protein and which are differential in one or more comparisons. Theorical MW on MW 2D-GE Comparison/up or down-regulated Gene name (kDa) Spots (kDa) G1G3/in G3 G2G4/in G4 G3G4/in G4 EPCR 27 717 36 Down CA1 29 817 31.5 Up Up 822 31.5 Up Up 825 31.5 Up Up 831 31.5 Up Up SERPING1 55 817 31.5 Up Up CD59 14.2 1009 21 Down 994 21 Down Down 1644 21 Down Down 1641 21 Down Down 1413 21 Down Down Down CLEC3B 23 935 23 Down COL1A1 138 1331 60 Up F2 70/31 830 31 Down KLK3 29 784 30 Down Down 1551 22 Down SERPINA1 47 403 65 Up 414 65 Up Up 428 65 Up 433 65 Up 489 56 Up 1328 56 Up 1330 56 Up 1331 56 Up 1371 35 Up 1043 20 Down SERPINA5 46 537 55 Down SERPINC1 53 403 65 Up KLK1 30 592 45 Up Up MASP2 76 378 75 Up Up 966 22 Down 990 22 Down 1167 15 Down PLG 91 1152 15 Down S100A8 11 1572 14.5 Down Down 1573 14.5 Down CD44 81.5 671 38 Up HSPG2 78 904 23.5 Down 1185 14.7 Down GNS 62 499 55 Up 784 30 Down Down NID1 136 671 38 Up QPCT 40 613 43.5 Up 630 42.4 673 40 Down 1589 37.8 1607 42 Down 1683 38.4 Down Down AMBP 39 977 22 Down Down 1642 20.7 Down 994 21 Down 1004 20.7 Down 1009 20.6 Down Down 1413 21 Down 1456 35 Down 1457 35.4 Down 1461 34.5 Up 1641 21 Down ENO1 47 521 53.8 Down 537 52.8 Down AZGP1 34 410 65.5 Up Up 629 42.4 Up Up 674 40 Up 677 39 Up 1596 43.6 Up Up 1597 43.6 Up 411 65.3 Up Up 418 64.7 Up Up 550 50 Up Up 592 45.3 Up Up 616 43.4 Up 624 42.7 Up 626 42.6 Up Up 628 42.6 Up HIST1H2BO 13 1440 17.9 Down

Example 5: Confirmation of Proteomic Analysis by Western-Blot Experiments on Treated and Concentrated Urine

(53) In order to validate the results obtained by 2D-GE approach and to verify the identities of proteins deduced from the results of LC-MS/MS analysis, the expression levels of 19 potential DN markers were analyzed by Western blotting.

(54) Protein C inhibitor (SerpinA5), CD59 glycoprotein (CD59), tetranectin (CLEC3B) were decreased in G4 group in comparison with G2, confirming the proteomic data. Prothrombin or Coagulation factor II (F2) which has a molecular weight of 70 kDa is activated by a proteolytic cleavage, leading to formation of activation peptide fragments 1 and 2 (F1.2) (31 kDa), and active thrombin (F2a) (37 kDa). F1.2 is an index of in vivo thrombin generation, one molecule of F1.2 being released with the generation of each thrombin molecule. The antibody used was raised against amino-acids mapping near the N-terminus of F1.2. The 70 kDa of the prothrombin was increased in G4 group in comparison with G2, unlike the 31 kDa of cleaved F1.2 which was reduced in the same comparison. This suggests that the active thrombin is decreased in G4 group. Protein abundances of alpha-1-antitrypsin (SerpinA1), antithrombin-III (SerpinC1), carbonic anhydrase (CA1) and collagen alpha-1(I) chain (COL1A1) were higher in G4 group than in the G2 group. CA1, which has been also identified as differential between G1 and G3 groups, has not been confirmed by Western blotting since it was found in only 2 out of 4 samples of the G3 group. MASP2 and HSPG2 were later validated. A MASP2 fragment of around 20 kDa was decreased in G4 group in comparison with G2. HSPG2 is cleaved in 2 chains: Endorepellin and LG3 peptide. Endorepellin was higher in G4 group than in the G2 group and LG3 peptide was decreased in G4 group in comparison with G3 group.

Example 6: Confirmation of Proteomic Analysis by Western-Blot Experiments on Native Urine

(55) Following proteomic studies on 2D-GE, ten biomarkers were validated analytically (Western blot) on concentrated and treated urine.

(56) The inventors investigated the presence of these biomarkers in samples of native urines (i.e. without previous treatment and unconcentrated).

(57) First, 20 μL of each urinary sample were analyzed (total protein concentrations of the samples were variable (47.86 to 450.89 μg/ml)).

(58) Among the ten biomarkers of interest, only the proteins HSPG2 (heparan sulfate proteoglycan 2), Col1A1 (Collagen, type I, alpha 1), Serpin C1 and F2 (prothrombin) were detected in Western blots on native urine.

(59) A relative quantification of these four biomarkers was performed on the same samples. An identical quantity of total proteins (1.6 μg) was loaded on SDS-PAGE gels. A same differential as that observed on treated urine was obtained (FIGS. 3A-3B). The four proteins were overexpressed in patients at risk of developing diabetic nephropathy (G4) in comparison to “control” patients (G2) (FIGS. 3A-3B).

Example 7: Confirmation of Proteomic Analysis by Western Blot Experiments on Urinary Samples from Healthy Subjects and Patients Suffering from Diabetic Nephropathy

(60) These ten biomarkers were selected as being specific of nephropathy and not of diabetes. In order to validate this point, analytic validation of the ten biomarkers was done for two other populations of subjects: i) healthy subjects who should behave as diabetic patients of the “control” cohort (G1, G2) and ii) diabetic patients suffering from nephropathy who should behave as diabetic patients of the “at risk” cohort (G3, G4).

(61) Western blots were performed on urine from patients in these different groups.

(62) The proteins CD59, CLEC3B, Serpin A5, and MASP2 were under-expressed in urine of diabetic patients at risk of developing nephropathy and of diabetic patients suffering from nephropathy (microalbuminuric and macroalbuminuric patients) (FIGS. 4A-4B). Moreover, the protein MASP2 was not detected in urine from diabetic patients suffering from nephropathy (FIGS. 4A-4B). For certain biomarkers, the signals obtained for patients in the same group were not homogeneous. Indeed, concerning proteins Serpin A5 and MASP2, a healthy subject (different in both cases) appeared to have a lower quantity of analyzed biomarkers than the other studied healthy subjects.

(63) The proteins HSPG2 and F2 were overexpressed in urine of diabetic patients at risk of developing diabetic nephropathy and in diabetic patients suffering from nephropathy (FIGS. 4A-4B). The signal obtained for a diabetic patient suffering from nephropathy was lower than for the other patients of the same group.

(64) Six biomarkers were validated on two other cohorts of patients: CD59, CLEC3B, Serpin A5, MASP2, HSPG2 and F2.

(65) The variations of protein expression were comparable between healthy subjects and patients of the “control” cohort, and between patients of the “at risk” cohort and diabetic patients suffering from nephropathy, showing that these six biomarkers are specific of nephropathy and not of diabetes.

Example 8: Diagnostic Performances of Single Marker or of Two-Marker Combinations

(66) The diagnostic performance of selected proteins according to the invention in the 2D-GE comparison between control patients and at risk patients of developing diabetic nephropathy before physical exercises (G1 versus G3) was evaluated using a Receiving Operating Characteristics (ROC) analysis (Table 7). ROC curves are the graphical visualization of the reciprocal relation between the sensitivity (Se) and the specificity (Sp) of a test for various values.

(67) TABLE-US-00007 TABLE 7 Examples of diagnosis performances of single marker for G1 versus G3 comparison. Markers AUC ROC Threshold Sp (%) Se (%) VPP (%) VPN (%) CI 95% CA1 0.875 10933795 83.3 91.7 84.6 90.9 [0.713; 1.000] CD59 0.792 −138077543 91.7 75.0 90.0 78.6 [0.583; 1.000] AMBP 0.771 −14365141 91.7 66.7 88.9 73.3 [0.567; 0.975] QPCT 0.764 −5589612 83.3 75.0 81.8 76.9 [0.560; 0.968] AUC ROC: area under the ROC curve; Threshold: expressed in 2D-GE relative intensity and selected by Youden index; Se: sensibility; Sp: specificity; PPV: positive predictive value (measures the proportion of subjects with positive test results who are correctly diagnosed); NPV: negative predictive value (measures the proportion of subjects with negative test results who are correctly diagnosed); CI 95%: 95% confidence interval.

(68) Multivariate analysis with mROC approach improved significantly AUC when comparing control patients versus at risk patients of developing diabetic nephropathy before physical exercises. The marker combination associating for example CA1 to CD59, CA1 to AMBP, CA1 to QPCT, CD59 to QPCT, and CD59 to AMPBP has a predictive value for higher risk of developing diabetic nephropathy as reported by the higher sensitivity and specificity (Table 8). The statistical analysis combining two markers generated a series of decision rules; a new virtual marker (Z) was calculated for each combination as illustrated in Table 9 and FIG. 5. Based on the combination of two markers, the virtual marker, transposing markers from the multivariate conditions into a univariate setting, discriminated significantly control patients from at risk patients of developing diabetic nephropathy before physical exercises with p-values<0.01.

(69) TABLE-US-00008 TABLE 8 Examples of diagnosis performances (mROC approach) of two- markers combination for G1 versus G3 comparison. Markers combinations AUC ROC Threshold Sp (%) Se (%) VPP (%) VPN (%) CI 95% CA1 + CD59 0.882 −0.2256 83.3 83.3 83.3 83.3 [0.738; 1.000] CA1 + AMBP 0.931 0.9028 100.0 75.0 100.0 80.0 [0.835; 1.000] CA1 + QPCT 0.917 −0.4629 75.0 100.0 80.0 100.0 [0.807; 1.000] CD59 + QPCT 0.847 −3.5075 83.3 75.0 81.8 76.9 [0.691; 1.000] CD59 + AMBP 0.833 −2.7376 66.7 100.0 75.0 100.0 [0.663; 1.000] AUC ROC: area under the ROCcurve; Threshold: expressed in 2D-GE relative intensity and selected by Youden index; Se: sensibility; Sp: specificity; PPV: positive predictive value (measures the proportion of subjects with positive test results who are correctly diagnosed); NPV: negative predictive value (measures the proportion of subjects with negative test results who are correctly diagnosed); CI 95%: 95% confidence interval.

(70) TABLE-US-00009 TABLE 9 Examples of a.sub.1 and a.sub.2 coefficients maximizing AUC of ROC curve for two-markers combination (mROC approach). Markers combinations Z = a.sub.1 x [Marker1] + a.sub.2 x [Marker2] CA1 + CD59 Z = +(1.56361513356411e−07)x[CA1] − (1.01677256583679e−08)x[CD59]  CA1 + AMBP Z = +(1.35230449215072e−07)x[CA1] − (4.18637841481153e−08)x[AMBP] .sup. CA1 + QPCT Z = +(1.24848530852367e−07)x[CA1] − (9.75979899375066e−07)x[QPCT] CD59 + QPCT  Z = −(1.36301781949143e−08)x[CD59] − (1.14127754739464e−06)x[QPCT] CD59 + AMBP Z = −(1.0745556660865e−08)x[CD59] − (3.39944153133327e−08)x[AMBP]

(71) The diagnostic performance of selected proteins according to the invention in the 2D-GE comparison between control patients and at risk patients of developing diabetic nephropathy after physical exercises (G2 verus G4) was evaluated using a Receiving Operating Characteristics (ROC) analysis (Table 10). ROC curves are the graphical visualization of the reciprocal relation between the sensitivity (Se) and the specificity (Sp) of a test for various values.

(72) TABLE-US-00010 TABLE 10 Examples of diagnostic performances of single markers for G2 versus G4 comparison. Markers AUC ROC Threshold Sp (%) Se (%) VPP (%) VPN CI 95% CD59 0.891 −116810836 92.3 91.7 91.7 92.3 [0.723; 1.000] AMBP 0.872 −9832649 100.0 83.3 100.0 86.7 [0.696; 1.000] AZGP1 0.872 22537171 92.3 83.3 90.9 85.7 [0.718; 1.000] CLEC3B 0.872 −4736752 69.2 100.0 75.0 100.0 [0.733; 1.000] HIST1H2BO 0.872 −12126916 92.3 83.3 90.9 85.7 [0.719; 1.000] F2 0.865 −11964063 76.9 83.3 76.9 83.3 [0.725; 1.000] ENO1 0.840 −1600077 69.2 91.7 73.3 90.0 [0.677; 1.000] MASP2 0.840 −7360249 76.9 91.7 78.6 90.9 [0.657; 1.000] SERPINA1 0.808 12377059 100.0 58.3 100.0 72.2 [0.633; 0.982] HSPG2 0.801 −11716038 84.6 75.0 81.8 78.6 [0.616; 0.987] QPCT 0.795 −1222240 69.2 83.3 71.4 81.8 [0.602; 0.988] CA1 0.776 15684748 92.3 75.0 90.0 80.0 [0.559; 0.993] SERPINA5 0.776 −2657487 76.9 75.0 75.0 76.9 [0.579; 0.973] COL1A1 0.756 2727091 92.3 66.7 88.9 75.0 [0.533; 0.980] SERPINC1 0.724 1403159 38.5 100.0 60.0 100.0 [0.521; 0.928] AUC ROC: area under the ROC curve; Threshold: expressed in 2D-GE relative intensity and selected by Youden index; Se: sensibility; Sp: specificity; PPV: positive predictive value (measures the proportion of subjects with positive test results who are correctly diagnosed); NPV: negative predictive value (measures the proportion of subjects with negative test results who are correctly diagnosed); CI 95%: 95% confidence interval.

(73) Multivariate analysis with mROC approach improved significantly AUC when comparing control patients versus at risk patients of developing diabetic nephropathy after physical exercises. The marker combination associating for example CLE3B to ENO1, SERPINA5 to AZGP1, SERPINA5 to SERPINA1, and CLEC3B to CD59 has a predictive value for higher risk of developing diabetic nephropathy as reported by the higher sensitivity and specificity (Table 11). The statistical analysis combining two markers generated a series of decision rules; a new virtual marker (Z) was calculated for each combination as illustrated in Table 12 and FIG. 5. Based on the combination of two markers, the virtual marker, transposing markers from the multivariate conditions into a univariate setting, discriminated significantly control patients from at risk patients of developing diabetic nephropathy after physical exercises with p-values<0.001.

(74) TABLE-US-00011 TABLE 11 Examples of diagnosis performances (mROC approach) of two- marker combination for G2 versus G4 comparison. Markers combinations AUC ROC Threshold Sp (%) Se (%) VPP (%) VPN (%) CI 95% CLEC3B + 0.994 −9.8628 92.3 100.0 92.3 100.0 [0.976; 1.000] ENO1 SERPINA5 + 0.981 −1.1267 100.0 91.7 100.0 92.9 [0.938; 1.000] AZGP1 SERPINA1 + 0.974 −2.0132 84.6 100.0 85.7 100.0 [0.926; 1.000] ENO1 SERPINA5 + 0.974 −7.0219 100.0 91.7 100.0 92.9 [0.920; 1.000] HIST1H2BO COL1A1 + 0.974 8.0182 92.3 91.7 91.7 92.3 [0.926; 1.000] AZGP1 CD59 + 0.974 0.0539 84.6 100.0 85.7 100.0 [0.926; 1.000] AZGP1 CD59 + ENO1 0.968 −5.8954 100.0 83.3 100.0 86.7 [0.910; 1.000] CLEC3B + 0.962 −4.2808 100.0 83.3 100.0 86.7 [0.897; 1.000] HIST1H2BO SERPINA5 + 0.955 −2.5367 76.9 100.0 80.0 100.0 [0.886; 1.000] SERPINA1 CA1 + ENO1 0.955 −0.9247 84.6 100.0 85.7 100.0 [0.883; 1.000] F2 + ENO1 0.955 −7.0566 84.6 91.7 84.6 91.7 [0.886; 1.000] CLEC3B + 0.949 −6.0067 84.6 100.0 85.7 100.0 [0.868; 1.000] SERPINA5 CLEC3B + 0.942 −3.8388 100.0 83.3 100.0 86.7 [0.854; 1.000] CD59 HSPG2 + 0.942 −4.5482 76.9 100.0 80.0 100.0 [0.858; 1.000] ENO1 SERPINA1 + 0.942 4.5949 84.6 100.0 85.7 100.0 [0.847; 1.000] AZGP1 CLEC3B + 0.942 −4.4092 100.0 83.3 100.0 86.7 [0.847; 1.000] QPCT CD59 + 0.942 −4.1036 100.0 91.7 100.0 92.9 [0.827; 1.000] AMBP SERPINA5 + 0.936 −5.1362 76.9 100.0 80.0 100.0 [0.847; 1.000] CD59 CA1 + CD59 0.936 −0.3404 84.6 91.7 84.6 91.7 [0.843; 1.000] CLEC3B + 0.936 −4.51 92.3 91.7 91.7 92.3 [0.840; 1.000] MASP2 AUC ROC: area under the ROC curve; Threshold: expressed in 2D-GE relative intensity and selected by Youden index; Se: sensibility; Sp: specificity; PPV: positive predictive value (measures the proportion of subjects with positive test results who are correctly diagnosed); NPV: negative predictive value (measures the proportion of subjects with negative test results who are correctly diagnosed); CI 95%: 95% confidence interval.

(75) TABLE-US-00012 TABLE 12 Examples of decision rules (mROC approach) for two-markers combination. Markers combinations Z = a.sub.1 x [Marker1] + a.sub.2 x [Marker2] ENO1 + Z = −(2.69376695675031e−06)x[ENO1] − CLEC3B (1.28313845176674e−06)x[CLEC3B] AZGP1 + Z = +(7.60396312049394e−08)x[AZGP1] − SERPINA5 (1.05893383881848e−06)x[SERPINA5] SERPINA1 + Z = +(1.05105404198974e−07)x[SERPINA1] − ENO1 (1.79201124674486e−06)x[ENO1] SERPINA5 + Z = −(1.34567451384797e−06)x[SERPINA5] − HIST1H2BO (2.84140535140061e−07)x[HIST1H2BO] AZGP1 + Z = +(2.30613768688269e−07)x[AZGP1] + COL1A1 (9.79281397114118e−07)x[COL1A1] AZGP1 + Z = +(1.7901623652807e−07)x[AZGP1] − CD59 (2.03071085055857e−08)x[CD59] ENO1 + Z = −(1.86395627696331e−06)x[ENO1] − CD59 (7.12337868208147e−08)x[CD59] CLEC3B + Z = −(6.50612939337824e−07)x[CLEC3B] − HIST1H2BO (1.41763210039085e−07)x[HIST1H2BO] SERPINA1 + Z = +(1.23367232406594e−07)x[SERPINA1] − SERPINA5 (1.19777770779838e−06)x[SERPINA5] ENO1 + Z = −(1.64889590005268e−06)x[ENO1] + CA1 (9.36482853645907e−08)x[CA1] ENO1 + Z = −(1.91161716055277e−06)x[ENO1] − F2 (2.67941736919358e−07)x[F2] SERPINA5 + Z = −(8.30972614812445e−07)x[SERPINA5] − CLEC3B (8.11124364413926e−07)x[CLEC3B] CLEC3B + Z = −(4.95333344427748e−07)x[CLEC3B] − CD59 (3.36932265360167e−08)x[CD59] ENO1 + Z = −(1.75495527888539e−06)x[ENO1] − HSPG2 (7.28595572223807e−08)x[HSPG2] AZGP1 + Z = +(2.04261743598041e−07)x[AZGP1] + SERPINA1 (3.2210138018648e−07)x[SERPINA1] CLEC3B + Z = −(7.04107016208075e−07)x[CLEC3B] − QPCT (1.41762801540533e−06)x[QPCT] AMBP + Z = −(−9.84720704127077e−09)x[AMBP] − CD59 (9.69931119857907e−08)x[CD59] SERPINA5 + Z = −(8.79594384945904e−07)x[SERPINA5] − CD59 (6.90048286568996e−08)x[CD59] CA1 + Z = +(7.02229904137993e−08)x[CA1] − CD59 (5.79940015439689e−08)x[CD59] CLEC3B + Z = −(7.61697614199002e−07)x[CLEC3B] − MASP2 (4.24259818124178e−08)x[MASP2]

Example 9: Discussion

(76) To search for early DN biomarker candidates in urine samples, 2D-GE analysis was performed, which is an accurate semi-quantitative comparison method to analyze differences between the urine proteomes from control and at risk patients. 14 and 156 protein spots that were differentially expressed before and after exercise test, respectively, were observed. The exercise proteomes of control and at risk diabetic patients were obtained by comparing the pre- and post-exercise test urine samples in each cohort (G1G2 and G3G4 comparisons, respectively). Greater variability of excreted urinary protein after the exercise in at risk patients (101 differential spots) than in control patients (5 differential spots) was observed.

(77) Among proteins identified as potential DN markers, ten were validated by Western blotting. Many differential spots in G2G4 comparison have a statistical total score equal to 5, which means that their differences in protein expression are statistically very significant (in the seven statistical tests). These proteins thus have a potential clinical diagnostic value.