Insights Into Diabetic Kidney Disease Using Urinary Proteomics and Bioinformatics

BRIEF REVIEW www.jasn.org

Insights into Diabetic Kidney Disease Using Urinary Proteomics and Bioinformatics

† † Julie A.D. Van,* James W. Scholey,* and Ana Konvalinka*

*Institute of Medical Science, University of Toronto, Toronto, Ontario, Canada; and †Department of Medicine, Division of Nephrology, University Health Network, Toronto, Ontario, Canada

ABSTRACT A number of proteomic and peptidomic analyses of urine from diabetic subjects changes in protein expression, deposi- have been published in the quest for a biomarker that predicts progression of ne- tion, or turnover in the diabetic kidney. phropathy. Less attention has been paid to the relationships between urinary pro- We reviewed the literature on urinary teins and the underlying biological processes revealed by the analyses. In this proteomics/peptidomics, biomarkers, review, we focus on the biological processes identified by studying urinary proteins and diabetic kidney disease in humans and protein-protein interactions at each stage of diabetic nephropathy to provide (Figure 1A, Supplemental Table 1). We an overview of the events underlying progression of kidney disease reflected in the selected the most robust candidate urine. In uncomplicated diabetes, proteomic/peptidomic analyses indicate that markers at each stage of diabetic kidney early activation of fibrotic pathways in the kidney occurs before the onset of micro- disease and then highlighted their roles albuminuria. In incipient nephropathy, when albumin excretion rates are abnormal, in biological processes that may contrib- proteomic/peptidomic analyses suggest that changes in glomerular permselectivity ute to progression from uncomplicated and tubular reabsorption account, at least in part, for the proteins and peptides that diabetes to incipient diabetic nephropa- appear in the urine. Finally, overt nephropathy is characterized by proteins involved thy to overt diabetic nephropathy (Table in wound healing, ongoing fibrosis, and inflammation. These findings suggest that 1). Although reviews on similar topics there is a spectrum of biological processes in the diabetic kidney and that assessing exist, none integrate findings across protein networks may be more informative than individual markers with respect to studies and assess their biological impli- the stage of disease and the risk of progression. cations on mechanisms underlying diabetic kidney disease progression. J Am Soc Nephrol 28: 1050–1061, 2017. doi: 10.1681/ASN.2016091018

INTEGRATING DATA WITH Microalbuminuria is currently the most Advances in mass spectrometry have BIOINFORMATICS reliable predictor of diabetic nephrop- enhanced our ability to identify thou- – athy.1 3 However, recent evidence chal- sands of proteins and peptides in urine Proteins and peptides are interconnected lenges this notion. The Joslin Study of in a single analysis—some of which may in large networks, which can be altered in Natural History of Microalbuminuria serve as new markers. This large-scale disease. Expression proﬁles vary accord- demonstrated that the likelihood of re- study of proteins is termed proteomics, ing to the type of cell, tissue, organ, or gression from microalbuminuria to whereas the study of naturally occurring specimen, which can shed light on normal urinary albumin excretion out- peptides generated by endogenous pro- the location of injury.11,12 Further- weighs the likelihood of progression tease activity is termed peptidomics. more, multiple datasets of potential to overt proteinuria.4 Other longitudi- Urinary proteomics and peptidomics biological markers can be more useful nal, observational studies have repli- add different dimensions to the investi- cated these results in children and gation of underlying biology. Their ap- Published online ahead of print. Publication date adults, in type 1 and type 2 diabetes, plication in urine has important clinical available at www.jasn.org. and in individuals from North Amer- implications for diabetic kidney disease, ica, Asia, and Europe.5–9 Taken to- given that urine can be collected nonin- Correspondence: Ms. Julie Van, Division of Nephrology, Toronto General Hospital, University gether, these studies provide strong vasively with relative ease and is directly Health Network, 11-PMB-189, Toronto, ON M5G evidence that microalbuminuria may produced by the kidneys.10 As such, 2N2, Canada. Email: [email protected] not be the ideal marker of progression changes in the relative abundance of uri- Copyright © 2017 by the American Society of after all. nary proteins and peptides may reﬂect Nephrology

1050 ISSN : 1046-6673/2804-1050 JAmSocNephrol28: 1050–1061, 2017 www.jasn.org BRIEF REVIEW

Figure 1. Summary of literature search and bioinformatic analyses. (A) Out of the 155 search results, 34 studies were screened and 31 relevant urinary proteomic/peptidomics studies in diabetes were included into our review. Reasons for study exclusion are shown for the initial screening and final inclusion. (B) The protein datasets, bioinformatic tools and software, and output figures are outlined for each of the bioinformatic analyses. when considered together, as opposed to in the progression of diabetic kidney glomerular basement membrane thick- in isolation, so that the most robust and disease using the Biological Networks ening, and podocyte loss.16,19 These consistent network of biological func- Gene Ontology13 and Enrichment Map14 functional and structural abnormalities tions may be identified. This “mechanis- plug-ins of Cytoscape software for the en- distinguish the diabetic kidney from a tic” approach can be used to explore and tire set of proteins associated with each healthykidneybeforetheonsetofmi- address the current gaps in knowledge stage of diabetic kidney disease (Figures 3 croalbuminuria. However, current uri- underlying the pathophysiology and and 4). For the purposes of this review, nary markers are not sensitive enough progression of diabetic nephropathy. we compared sequential stages of dia- to confidently detect these changes. We thus investigated the biological betic kidney disease to better understand Hence, there is a pressing need for accu- implications of differentially excreted transitions and progression, although we rate and reliable markers of early dia- urinary proteins in diabetic kidney dis- realize that this construct is somewhat betic kidney injury. ease by taking advantage of published, artificial. Furthermore, we constructed In healthy adolescents with type 1 di- relevant data (Figure 1B). Candidate protein-protein interaction networks abetes, Meier and colleagues identified a markers with differential excretion be- with the Search Tool for the Retrieval of urinary proteome profile, which was dif- tween cases and controls were extracted Interacting Genes/Proteins (Figure 5).15 ferent from that of healthy, age-matched from all 31 studies (Supplemental Table Overall, these analyses have enabled us to controls.20 Without the use of tandem 1). All candidates were weighted equally comprehensively identify common and mass spectrometry, differences could when inputted into the network analy- stage-specific biological processes in di- not be assigned to specific proteins. Nev- ses. However, we selected a handful of abetic kidney disease by including all ertheless, this study demonstrated that promising candidates for each stage proteins associated with a particular differences could be detected early in di- of diabetic kidney disease on the basis stage of disease. abetic kidney disease, setting the stage of (1) between-group differential excre- for later studies. In the largest urinary tion rates in at least two studies, (2) sup- peptidome study of diabetes to date, port from verification and validation, UNCOMPLICATED DIABETES Maahs and colleagues demonstrated and/or (3) high enrichment and/or spec- lower urinary levels of collagen and uro- ificity in the kidney (Table 1). For the Uncomplicated diabetes is the earliest modulin in diabetic cases with normal localization analysis, these promising stage of diabetic kidney disease. Some renal function compared with controls.21 candidates were mapped onto specific maladaptive changes such as renal hyper- Zhang and colleagues found two urinary nephron segments on the basis of their filtration and hypertrophy are often pre- fragments of fibrinogen a chain and renal expression in normal tissues using sent at time of diagnosis of diabetes.16 prothrombin—two important and closely the Human Protein Atlas11 and inferred Renal hyperfiltration has been linked to related players in coagulation—which site of injury using the existing literature the initiation and progression of diabetic were associated with type 2 diabetes (par- (Figure 2). For the functional analysis, we nephropathy.17,18 Many patients develop ticularly in those with high glycated hemo- deduced biological processes implicated lesionssuchasmesangialexpansion, globin, HbA1c).22 In total, 35 protein

J Am Soc Nephrol 28: 1050–1061, 2017 Urinary Proteomics in Diabetes 1051 1052 RE REVIEW BRIEF ora fteAeia oit fNephrology of Society American the of Journal

Table 1. Characteristics of promising markers for each stage of diabetic kidney disease www.jasn.org Stage of Diabetic Direction of Nephron Protein Description Biological Processes Validated Kidney Disease Urinary Excretiona Segment

Uncomplicated a1-antitrypsin (SERPINA1) 40-kD serine protease inhibitor ↑ (9 of 9) PT Coagulation, inflammation ELISA, SRM Diabetes Clusterin (CLU) 50-kD nuclear form ↑ (3 of 4) PT Complement activation, ELISA 75-kD secretory form inflammation, lipid metabolism Type I collagen (COL1A1/ .130-kD extracellular matrix protein ↓ (8 of 8) NS Extracellular matrix organization None COL1A2) Heparan sulfate proteoglycan 469-kD basement membrane protein ↓ (2 of 2) NS Blood vessel development, None (HSPG2) extracellular matrix organization Osteopontin (SPP1) 35-kD secreted phosphoprotein ↑ (1 of 1) LoH, DCT* Adhesion, tissue development None Uromodulin (UMOD) 85-kD GPI-anchored glycoprotein ↓ (11 of 11) LoH, DCT* Tissue development ELISA, SRM Incipient Diabetic a1-acid glycoprotein 1 (ORM1) 24-kD positive acute-phase reactant ↑ (5 of 5) G Inflammation, immune system, ELISA, immuno- Nephropathy transport turbidimetry Cubilin (CUBN) 399-kD endocytic receptor ↑ (1 of 1) PT* Endocytosis, lipid metabolism None Haptoglobin (HP) 45-kD positive acute-phase reactant ↑ (2 of 4) NS Immune system response, ELISA, SRM inflammation Megalin (LRP2) 522-kD endocytic receptor ↑ (1 of 1) PT* Endocytosis, tissue development None Mannan-binding lectin serine 76-kD serine protease ↑ (1 of 2) NS Complement activation, immune SRM protease 2 (MASP2) system response Transferrin (TF) 77-kD plasma carrier of iron ↑ (2 of 2) G Iron ion homeostasis, transport SRM Overt Diabetic a2-HS-glycoprotein (AHSG) 39-kD plasma carrier ↑ (5 of 5) G, PT Endocytosis, inflammation, tissue None Nephropathy development b2-microglobulin (B2M) 14-kD MHC class I component ↑ (5 of 5) G, PT Ion homeostasis, immune system None response Hemopexin (HPX) 59-kD plasma carrier of heme ↑ (1 of 1) G Iron ion homeostasis, transport None Retinol-binding protein 4 (RBP4) 23-kD plasma carrier of retinol ↑ (5 of 6) PT* Immune system response, tissue ELISA

mScNephrol Soc Am J development, transport Transthyretin (TTR) 55-kD homotetrameric plasma carrier ↓ (4 of 7) PT Extracellular matrix organization, ELISA of RBP4 and thyroxine transport The asterisk (*) denotes high enrichment in and/or specificity to the kidney. ↑, increased urinary excretion in cases compared to controls; PT, proximal tubules; SRM, selected reaction monitoring; ↓, decreased urinary excretion in cases compared to controls; NS, not specific to the kidneys or a particular segment; LoH, loop of Henle; DCT, distal convoluted tubules; G, glomerulus. aFraction of studies that reported the specific direction of urinary excretion (in cases versus controls) out of all studies that identified differential excretion of the protein (shown in brackets). 28: 1050 – 01 2017 1061, www.jasn.org BRIEF REVIEW

matrix regulation and metabolism, macromolecular organization, regulation of wound healing and coagulation, regulation of cholesterols and lipids, re- sponses to chemical stimulus and stress, and (“anatomic,”“blood vessel,”“carti- lage,” and “skin”)tissuedevelopment. Processes appear as clustered nodes in the enrichment map comparing uncomplicated diabetes and incipient diabetic nephropathy (Figure 3). The majority of processes in uncomplicated diabetes were also enriched in incipient diabetic nephropathy, except for macromolecular organization. Coagulation, tissue development, and extracellular matrix regulation and metabolism were among the most significantly enriched processes. Interestingly, these biological processes play important roles within the broader biological processes of wound healing. In fact, coagulation is one of the first processes activated in wound healing, fa- cilitating leukocyte entry into the site of injury.28 Increased urinary excretion of thrombin and fibrinogen in uncomplicated diabetes21,22 suggests that coagulation may be activated in early diabetes (Figure 5A). Experimental studies have shown that increased activation of coag- Figure 2. Localization of the most promising urinary markers of diabetic kidney disease in ulation in diabetes impaired glomerular different nephron segments. Proteins were mapped according to their normal and induced permselectivity and induced apoptosis 12 expression in renal tissues using the Human Protein Atlas and the existing literature. The of endothelial cells and podocytes.29 asterisk (*) denotes high protein enrichment and/or specificity in the segment. AHSG, a -HS- 2 Tissue development, in the context of glycoprotein; B2M, b2-microglobulin; CLU, clusterin; CUBN, cubilin; HPX, hemopexin; HSPG2, heparan sulfate proteoglycan; LRP2, megalin; MASP2, mannan-binding lectin serine wound healing, relies on normal extra- protease 2; ORM1, a -acid glycoprotein 1; RBP4, retinol-binding protein 4; SERPINA1, cellular matrix organization. Platelets re- 1 b a1-antitrypsin; SPP1, osteopontin; TF, transferrin; TTR, transthyretin; UMOD, uromodulin. lease TGF to stimulate the deposition and suppress the degradation of extra- candidates have been identified as potential development and progression of dia- cellular matrix,30 which replaces the fi- markers of early, uncomplicated diabetes. betic nephropathy.25,26 In fact, regres- brin clot and acts as a scaffold for healing sion of microalbuminuria in type 1 tissues. As the wound heals, TGFb levels Early Diabetes: a Tubular diabetes was significantly associated subside, and the deposited matrix is de- Pathology? with lower urinary levels of tubular in- graded, restoring homeostasis.30 Chronic Diabetic kidney disease is often described jury markers: kidney injury molecule-1 injury, however, overrides the tight con- as a glomerular pathology, wherein in- and N-acetyl-b-D-glucosaminidase.26,27 trol of TGFb. The subsequent accumula- jury to the glomerulus not only precedes Targeted attempts to identify and treat tion of extracellular matrix, as evidenced but typically outweighs progressive in- early tubular injury may prevent or de- by significant reductions of urinary fi- jury to the tubulo-interstitium.19,23,24 lay diabetic nephropathy. brillar collagens (e.g., type I and III) and However, the differential excretion of tu- heparan sulfate proteoglycan in compli- bular proteins such as uromodulin and Coagulation and Fibrosis May Be cated diabetes compared with healthy osteopontin suggests that the tubular Activated in Chronic Hyperglycemia controls,21,31,32 hinders cellular migration compartment is an important site of early In our functional analysis, the 35 differ- and proper tissue development, culminat- injury (Figure 2). Other lines of evidence entially excreted proteins in uncompli- ing in scar tissue and fibrosis. Interest- also support the notion that early changes cated diabetes were associated with cell ingly, urinary excretion of nonfibrillar in the tubulointerstitium contribute to the adhesion, coagulation, extracellular type VIII collagen was also diminished

J Am Soc Nephrol 28: 1050–1061, 2017 Urinary Proteomics in Diabetes 1053 BRIEF REVIEW www.jasn.org

Figure 3. Comparison of enriched biological processes in uncomplicated diabetes and incipient diabetic nephropathy. Significantly enriched biological processes were identified for each stage using Biological Networks Gene Ontology with Benjamini and Hochberg multiple testing correction (P,0.05) and then run on Enrichment Map with Jaccard coefficient of 0.5 (P value cut-off = 0.001; false dis- covery rate Q-value cut-off = 0.05). Each node represents an enriched biological process. Red node colors correspond to high enrichment, whereas gray node colors correspond to no enrichment. As shown in the figure legend, the outer circle color corresponds to the level of enrichment in uncomplicated diabetes, whereas the inner circle color corresponds to that in incipient diabetic nephropathy. Edge thickness denotes the amount of overlapping markers between two connected nodes within uncomplicated diabetes (blue) and within incipient diabetic nephropathy (green). DM, diabetes mellitus; DN, diabetic nephropathy. in uncomplicated diabetes,21 possibly as a diabetes seems plausible (Figure 5A). develops after 10–15 years of type 1 diabe- result of impaired turnover and increased Moreover, mesangial matrix expansion tes, but may already be present when type accumulation in the glomerulus. Studies has been well documented at this stage 2 diabetes is diagnosed.16 have shown that its upregulated expres- of diabetic kidney disease.16,38,39 In their comparison of patients with sion in the kidney may be specifictodi- incipient diabetic nephropathy and con- abetes and may contribute to profibrotic trols with uncomplicated diabetes, Jin TGFb-driven pathophysiologic processes INCIPIENT DIABETIC and colleagues performed isobaric tag such as the proliferation of renal fibro- NEPHROPATHY for relative and absolute quantitation blasts and mesangial cells.33,34 Blood ves- to identify and multiple reaction moni- sel development is also greatly impaired Between 20% and 30% of patients with toring to verify differential excretion of 35 by fibrosis, contributing to the develop- diabetes advance to incipient diabetic ne- a1-antitrypsin, transferrin, ceruloplasmin, ment and progression of renal lesions in phropathy,5,40,41 defined by the onset of albumin, haptoglobin, vitamin D–binding 36,37 diabetic animals. persistent microalbuminuria. Although protein, a1-acid glycoprotein 1, and Although renal fibrosis is a hallmark of hyperfiltration may persist, some patients prostate stem cell antigen.43 Thrailkill latediabeticnephropathy,extracellularma- experience a progressive decline in GFR.42 and colleagues posited that increased uri- trix accumulation in early, uncomplicated Incipient diabetic nephropathy usually nary excretion of megalin and cubilin

1054 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1050–1061, 2017 www.jasn.org BRIEF REVIEW

Figure 4. Comparison of enriched biological processes in incipient diabetic nephropathy and overt diabetic nephropathy. Significantly enriched biological processes were identified for each stage using Biological Networks Gene Ontology with Benjamini and Hochberg multiple testing correction (P,0.05) and then run on Enrichment Map with Jaccard coefficient of 0.5 (P value cut-off = 0.001; false dis- covery rate Q-value cut-off = 0.05). As shown in the figure legend, the outer circle color corresponds to the level of enrichment in incipient diabetic nephropathy, whereas the inner circle color corresponds to that in overt diabetic nephropathy. Edge thickness denotes the amount of overlapping markers between two connected nodes within incipient diabetic nephropathy (blue) and within overt diabetic nephropathy (green). DM, diabetes mellitus; DN, diabetic nephropathy. contributed to the increased urinary ex- function decline. We identified a total of thickening of the glomerulus and proxi- cretion of other proteins, such as albu- 47 differentially excreted proteins in in- mal tubules.16,46 Although a healthy glo- min, transferrin, and vitamin D–binding cipient diabetic nephropathy, 36% of merular filtration barrier with normal protein, in urine.44 Focusing on peptido- which were also identified in uncompli- permselectivity limits entry of large macro- mics, Merchant and colleagues per- cated diabetes. molecules into Bowman’sspace,somepro- formed a case-control study, comparing teins are filtered, including trace amounts “decliner” patients whose GFRs progres- Urinary Markers Reflect Defects in of albumin.47 However, nearly all of these sively declined by 3.3% per year during the Glomerulus and Proximal filtered proteins are reabsorbed into the follow-up and “nondecliner” controls Tubules proximal tubules via megalin/cubilin- whose GFRs remained relatively stable.45 Localization of markers within the glo- mediated endocytosis.48 Interestingly, Differentially excreted peptides derived merulus and proximal tubules may reflect endocytosis was among the enriched bi- from extracellular matrix elements (e.g., impaired glomerular permselectivity and ological processes in incipient diabetic types IV and V collagen, tenascin-X), cell proximal tubular reabsorption (Figure 2). nephropathy (Figure 3). Several of the dif- adhesion molecules (e.g., zona occludens Within the nephron segments, reduced ferentially excreted proteins (e.g., albumin, 3, protocadherin FAT 2), and the enzyme urinary excretion of type IV collagen clusterin, apos, vitamin D–binding pro- inositol pentakisphosphate 2-kinase and and heparan sulfate proteoglycan likely tein, and transferrin) are known ligands were associated with progressive renal reflects progressive basement membrane of megalin and cubilin,49 suggesting

J Am Soc Nephrol 28: 1050–1061, 2017 Urinary Proteomics in Diabetes 1055 BRIEF REVIEW www.jasn.org

Figure 5. Protein-protein interaction networks for differentially excreted proteins in diabetic kidney disease using the Search Tool for the Retrieval of Interacting Genes/Proteins v10 database. Each node represents a candidate marker. Node colors illustrate a protein’sin- volvement in the four significantly enriched biological processes (identified by Biological Networks Gene Ontology) for each stage. Gray nodes are not associated with any of the four significantly enriched biological processes. Square nodes indicate high enrichment in and/or specificity to the kidney. Some proteins were not found in the Search Tool for the Retrieval of Interacting Genes/Proteins database and were labeled in red (e.g., PLPP3). The edges represent protein-protein interactions, and the nature of the protein-protein interactions are

1056 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1050–1061, 2017 www.jasn.org BRIEF REVIEW

Figure 5. Continued. tubular dysfunction before development signaling cascades (e.g.,ofEGFand markers of inflammation include several of overt proteinuria. thrombin). Fibrinolysis refers to the break- acute-phase reactant proteins such as down to fibrin clots, which are formed a1-acid glycoprotein 1, haptoglobin, Inflammation, Cholesterol and Lipid from fibrinogen by thrombin during coag- clusterin, a2-HS-glycoprotein, and Dysregulation, and Ion Imbalance ulation, which were discussed in the un- mannan-binding lectin serine protease Underlie Incipient Diabetic complicated diabetes section. 2 (Figure 5B). Nephropathy Inflammation, similar to coagulation The transport and metabolism of cho- Despite the overlap between uncompli- and the regulation of extracellular matrix lesterols and lipids are regulated, in part, cated diabetes and incipient diabetic ne- in uncomplicated diabetes, is linked to by the apo family.51 Several members phropathy, regulation of cholesterols and the wound healing pathways. In wound (e.g., apo A, E, H, clusterin/J) were differ- lipids, inflammation, and ion homeostasis healing, macrophages infiltrate the site of entially excreted throughout the progres- emerged as the top biological processes in injury during inflammation to phagocy- sion of diabetic kidney disease (Figure 5) incipient diabetic nephropathy (Figures 3 tose cellular debris, which in turn facili- and significantly contributed to the en- and 4). Endocytosis, regulation of phos- tates the migration and proliferation of richment of cholesterols and lipids in phorylation, and regulation of fibrinolysis other cells.28 Eventually, inflammation is diabetic nephropathy (Figure 4). Stud- were exclusively enriched in this stage of turned “off” to allow for the resolution ies have shown that diabetes impairs disease. Endocytosis is the primary mech- of injury through tissue proliferation insulin- and leptin-mediated regulation anism for reabsorption of proteins via and remodeling.28 However, this inflam- of cholesterol and lipid, which increases megalin/cubilin, as previously discussed. matory state is sustained by chronic hy- the risk of cardiovascular disease.52,53 In- Phosphorylation is a common post- perglycemia, predisposing to progressive terestingly, impaired cholesterol esterifi- translational modification used in several diabetic kidney disease.50 Urinary cation and efflux have been linked to

color-coded as indicated in the figure legend (e.g., binding interactions in blue and unspecified interactions in gray). (A) Uncomplicated diabetes was characterized by coagulation, extracellular matrix, and tissue development, which are also activated in wound healing (n=35). (B) Incipient diabetic nephropathy was characterized by regulation of cholesterols and lipids, inflammation, ion homeostasis, and wound healing (n=47). (C) Overt diabetic nephropathy was characterized by extracellular matrix, inflammation, immune system, and wound healing (n=50).

J Am Soc Nephrol 28: 1050–1061, 2017 Urinary Proteomics in Diabetes 1057 BRIEF REVIEW www.jasn.org podocyte injury.54 Cholesterols also play decreased levels of ubiquitin-60S ribo- included extracellular matrix regulation an important role in cell membrane in- somal protein L40 from 38 patients with and metabolism, inflammation, and reg- tegrity. In diabetes, overexpression of diabetic nephropathy, compared with 45 ulation of the immune response (Figure cholesterol in platelet membranes im- healthy controls.63 These proteins are 4). Interestingly, the number of proteins paired fluidity and in turn heightened ubiquitously expressed in the body and involved in the regulation of the im- platelet sensitivity to thrombin,55 per- likely derived from the circulation. As mune response greatly outnumbered haps contributing to increased throm- such, their presence in the urine may re- those involved in wound healing, al- bosis and vascular injury. There could flect impaired glomerular permselec- though there is significant overlap be- be an interesting interplay between coag- tivity. Sharma and colleagues identified tween both processes (Figure 5C). The ulation, blood vessel development, and a1-antitrypsin as the primary marker immune system plays an integral role in cholesterol and lipid regulation. How- between cases and controls, but were successful wound healing, promoting ever, unlike other processes, the enrich- limited by a small sample size of eight the infiltration of immune cells into 64 ment of cholesterol and lipid regulation patients. The presence of a1-antitrypsin the site of injury and regulating inflam- waslargelydrivenbyaposaswellasother in urine is consequently not specifictoa mation.28 For the most part, these biolog- high-abundance plasma proteins. Although single stage of diabetic kidney disease. ical processes (e.g., extracellular matrix these biological processes may reflect sys- Rao and colleagues focused on low- regulation and metabolism, inflamma- temic changes, these plasma proteins may abundance proteins by immunodeplet- tion, regulation of the immune response, be detected in the urine as a result of im- ing albumin, immunoglobin G and A, and cell adhesion) were also present in the paired glomerular permselectivity and a1-antitrypsin, transferrin, and hapto- earlier stages (Figures 3 and 4), underlin- 65 tubular reabsorption. globin. Transthyretin and a2-HS- ing their persistent role in the progression Maintaining ion homeostasis is an glycoprotein among others were identified of diabetic kidney disease. important function of the kidney, espe- as potential markers. Otu and colleagues cially the proximal tubule. These seg- characterized the urinary proteome for ments are largely responsible for the diabetic nephropathy in Pima Indians of DIABETIC KIDNEY DISEASE AS A bulk reabsorption of a variety of electro- Southern Arizona.66 This particular de- WHOLE lytes including sodium, chloride, bicar- mographic has been extensively studied bonate, and phosphate. Interestingly, for its predisposition to type 2 diabetes Overall, the majority of enriched biolog- iron ion homeostasis emerged as one of and vascular complications.67,68 Limita- ical processes were identified at all stages the most enriched individual biological tions of this proteomic study include the of diabetic kidney disease. These processes processes within the broader cluster of lack of tandem mass spectrometry, sur- include coagulation, inflammation, cell ion homeostasis, a finding that might im- vivor bias, and the questionable stability adhesion, regulation of the immune sys- plicate oxidative stress in incipient dia- of urine samples stored for .10 years. tem, extracellular matrix metabolism, betic nephropathy.56,57 Iron catalyzes the Although 50 proteins were differentially tissue and blood vessel development, en- formation of hydroxyl radical, a reactive excreted in overt diabetic nephropathy, dothelial cell proliferation, and cell ad- oxygen species, via the Haber–Weiss re- only nine were exclusive to this stage. hesion (Figures 3 and 4). Interestingly, action as a catalyzing agent.58,59 Iron Consequently, we hypothesize that the these processes all fall within the spec- metabolism and transport are regulated biological processes underlying overt di- trum of wound healing—the umbrella by haptoglobin and transferrin (Figure abetic nephropathy were also enriched term of complex biological processes in- 5B), which were elevated in the urines in previous stages. volved in the body’s response to injury of patients with diabetic kidney disease. and subsequent repair.28 The identifica- Glomerular Dysfunction tion of wound healing processes in early, Predominantly Manifests in Late uncomplicated diabetes indicates a de- OVERT DIABETIC NEPHROPATHY Disease gree of renal injury long before the onset Defects in permselectivity and reabsorption of microalbuminuria. Even though uri- The onset of proteinuria, or macroalbu- continuetoprevailinovertdiabeticnephrop- nary markers were identified in different minuria, marks late-stage diabetic kidney athy, as exemplified by the increased urinary studies, they were highly connected on disease.16 Functionally, urinary albumin presence of several carrier proteins (e.g., the basis of our analysis of protein-protein excretion continues to rise, whereas GFR a2-HS-glycoprotein, hemopexin, trans- interactions (Figure 5). For example, the 60,61 declines. Structurally, there is evidence thyretin) from the plasma compartment. presence of a1-antitrypsin alone in urine of glomerulosclerosis and tubular atrophy. would not have made a compelling argu- Nearly 10% of patients with diabetes prog- Underlying Processes in Overt ment for coagulation in early diabetes. ress further to ESRD62 and require RRT Diabetic Nephropathy Were Likely However, we found supporting evidence such as dialysis and transplantation. Active in Previous Stages in that thrombin, antithrombin, fibrino- Dihazi and colleagues found increased In overt diabetic nephropathy, the sig- gen, and kininogen were also identified urinary levels of b2-microglobulin and nificantly enriched biological processes in urine—all of which are involved in

1058 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1050–1061, 2017 www.jasn.org BRIEF REVIEW coagulation.69,70 As such, stage-specific was predominantly characterized by im- 5. Schultz CJ, Konopelska-Bahu T, Dalton RN, networks of proteins may be more infor- pairedglomerular permselectivityandtu- Carroll TA, Stratton I, Gale EA, Neil A, fi fl Dunger DB; Oxford Regional Prospective mative than individually quanti ed pro- bular reabsorption via in ammation and Study Group: Microalbuminuria prevalence teins, due to the complexity of biology. ion dysregulation. These processes contin- varies with age, sex, and puberty in children There are several limitations to our ued to play a significant role in overt di- with type 1 diabetes followed from diagnosis analyses. First, we assumed that the pro- abetic nephropathy, highlighted by wound in a longitudinal study. Diabetes Care 22: – teins measured in the urine reflected bi- healing, progressive fibrosis, and chronic 495 502, 1999 fl 6. Giorgino F, Laviola L, Cavallo Perin P, Solnica ological processes in the diabetic kidney. in ammation. Overall, we demonstrate B, Fuller J, Chaturvedi N: Factors associated Given the protein networks enriched in how proteomic/peptidomic datasets may with progression to macroalbuminuria in mi- protein-protein interactions, it seems be combined and integrated to highlight croalbuminuric Type 1 diabetic patients: The unlikely that the proteins were found in the most robust markers and to character- EURODIAB Prospective Complications Study. – the urine by chance. In a study on the ize the biological context for the proteins. Diabetologia 47: 1020 1028, 2004 7. Araki S, Haneda M, Sugimoto T, Isono M, characterization of the normal urinary Isshiki K, Kashiwagi A, Koya D: Factors asso- proteome, nearly 70% of urinary pro- ciated with frequent remission of micro- teins were likely derived from the kid- ACKNOWLEDGMENTS albuminuria in patients with type 2 diabetes. ney,71 supporting the use of urine as a Diabetes 54: 2983–2987, 2005 surrogate biospecimen for kidney tissue. 8. Roy MS, Affouf M, Roy A: Six-year incidence J.A.D.V. is supported by the Banting & Best of proteinuria in type 1 diabetic African Second, our bioinformatic analyses did Diabetes Centre–Novo Nordisk Scholarship. Americans. Diabetes Care 30: 1807–1812, not include quantitative results as these A.K. is supported by the Kidney Foundation 2007 were not available for all reviewed studies. of Canada operating grant and the Kidney 9. Son MK, Yoo HY, Kwak BO, Park HW, Kim KS, Chung S, Chae HW, Kim HS, Kim DH: Re- Given the differences in methodology, it Research Scientist Core Education and Na- would also be difficult to normalize data gression and progression of microalbuminuria tional Training program/Canadian Institutes in adolescents with childhood onset diabetes across studies. Third, our bioinformatic of Health Research (CIHR) salary and in- mellitus. Ann Pediatr Endocrinol Metab 20: analyses was largely tailored to proteomic frastructure support for new investigators. 13–20, 2015 data. Differentially excreted peptides J.W.S. is supported by operating grants from 10. Konvalinka A, Scholey JW, Diamandis EP: were incorporated into protein networks Searching for new biomarkers of renal dis- the Heart and Stroke Foundation of Canada eases through proteomics. Clin Chem 58: as their precursor proteins. Although the and the CIHR. – fi “ ” 353 365, 2012 eld of mechanistic peptidomics is rel- J.A.D.V. wrote the manuscript. All authors 11. Pontén F, Jirström K, Uhlen M: The human atively new, there are tools currently contributed equally to researching data for protein atlas–a tool for pathology. J Pathol – available for the prediction of proteases the article, and reviewing and editing the 216: 387 393, 2008 12. Uhlen M, Fagerberg L, Hallström BM, involved in the generation of naturally manuscript before submission. occurring peptides (e.g.,Proteasix72) Lindskog C, Oksvold P, Mardinoglu A, Sivertsson Å, Kampf C, Sjöstedt E, Asplund and the mapping of peptides onto their A, Olsson I, Edlund K, Lundberg E, Navani S, native protein sequence (e.g., Peptide Ex- DISCLOSURES Szigyarto CA, Odeberg J, Djureinovic D, tractor73). Such analyses will enhance Takanen JO, Hober S, Alm T, Edqvist PH, None. our understanding of the proteolytic Berling H, Tegel H, Mulder J, Rockberg J, processes that may be altered in diabetic Nilsson P, Schwenk JM, Hamsten M, von Feilitzen K, Forsberg M, Persson L, kidney disease. Nevertheless, our findings REFERENCES fi Johansson F, Zwahlen M, von Heijne G, identi ed major biological processes that Nielsen J, Pontén F: Proteomics. Tissue- develop early, persist throughout, and be- 1. Viberti GC, Hill RD, Jarrett RJ, Argyropoulos based map of the human proteome. Sci- come dominant at different stages in the A, Mahmud U, Keen H: Microalbuminuria as a ence 347: 1260419, 2015 progression of diabetic kidney disease. predictor of clinical nephropathy in insulin- 13. Maere S, Heymans K, Kuiper M: BiNGO: A dependent diabetes mellitus. Lancet 1: Cytoscape plugin to assess overrepresentation Over 75differentiallyexcreted urinary – fi 1430 1432, 1982 of gene ontology categories in biological proteins were identi ed by studies using 2. Parving HH, Oxenbøll B, Svendsen PA, networks. Bioinformatics 21: 3448–3449, urinary proteomic and peptidomic ap- Christiansen JS, Andersen AR: Early de- 2005 proaches, but only a minority of candi- tection of patients at risk of developing di- 14. Merico D, Isserlin R, Stueker O, Emili A, dates have been verified and validated. abetic nephropathy. A longitudinal study of Bader GD: Enrichment map: A network- urinary albumin excretion. Acta Endocrinol based method for gene-set enrichment vi- Our bioinformatic analysis of the biolog- (Copenh) 100: 550–555, 1982 sualization and interpretation. PLoS One 5: ical implications of the candidate markers 3. Mogensen CE: Microalbuminuria predicts e13984, 2010 uncovered stage-specific and overarching clinical proteinuria and early mortality in 15. Szklarczyk D, Franceschini A, Wyder S, mechanisms potentially acting on the di- maturity-onset diabetes. NEnglJMed310: Forslund K, Heller D, Huerta-Cepas J, – abetic kidney. Urinary markers in uncom- 356 360, 1984 Simonovic M, Roth A, Santos A, Tsafou KP, 4. Perkins BA, Ficociello LH, Silva KH, Kuhn M, Bork P, Jensen LJ, von Mering C: plicated diabetes indicated early activation fi Finkelstein DM, Warram JH, Krolewski AS: STRING v10: Protein-protein interaction of renal brosis and the importance of tu- Regression of microalbuminuria in type 1 di- networks, integrated over the tree of life. bular injury.Incipientdiabeticnephropathy abetes. N Engl J Med 348: 2285–2293, 2003 Nucleic Acids Res 43: D447–D452, 2015

J Am Soc Nephrol 28: 1050–1061, 2017 Urinary Proteomics in Diabetes 1059 BRIEF REVIEW www.jasn.org

16. Mogensen CE, Christensen CK, Vittinghus E: and N-acetyl-b-D-glucosaminidase. Kidney complications. Health Technol Assess 9: iii–vi, The stages in diabetic renal disease. With Int 79: 464–470, 2011 xiii–163, 2005 emphasis on the stage of incipient diabetic 28. Gurtner GC, Werner S, Barrandon Y, Longaker 42. Krolewski AS, Niewczas MA, Skupien J, nephropathy. Diabetes 32[Suppl 2]: 64–78, MT: Wound repair and regeneration. Nature Gohda T, Smiles A, Eckfeldt JH, Doria A, 1983 453: 314–321, 2008 Warram JH: Early progressive renal decline 17. Ruggenenti P, Porrini EL, Gaspari F, 29. Gilbert RE, Marsden PA: Activated protein C precedes the onset of microalbuminuria and Motterlini N, Cannata A, Carrara F, Cella C, and diabetic nephropathy. NEnglJMed its progression to macroalbuminuria. Di- Ferrari S, Stucchi N, Parvanova A, Iliev I, 358: 1628–1630, 2008 abetes Care 37: 226–234, 2014 Dodesini AR, Trevisan R, Bossi A, Zaletel J, 30. Border WA, Noble NA: Transforming growth 43. Jin J, Ku YH, Kim Y, Kim Y, Kim K, Lee JY, Cho Remuzzi G; GFR Study Investigators: Glo- factor beta in tissue fibrosis. NEnglJMed YM, Lee HK, Park KS, Kim Y: Differential merular hyperfiltration and renal disease 331: 1286–1292, 1994 proteome profiling using iTRAQ in micro- progression in type 2 diabetes. Diabetes 31. Rossing K, Mischak H, Dakna M, Zürbig P, albuminuric and normoalbuminuric type 2 Care 35: 2061–2068, 2012 Novak J, Julian BA, Good DM, Coon JJ, diabetic patients. Exp Diabetes Res 2012: 18. Magee GM, Bilous RW, Cardwell CR, Hunter Tarnow L, Rossing P; PREDICTIONS Net- 168602, 2012 SJ, Kee F, Fogarty DG: Is hyperfiltration as- work: Urinary proteomics in diabetes and 44. Thrailkill KM, Nimmo T, Bunn RC, Cockrell sociated with the future risk of developing CKD. J Am Soc Nephrol 19: 1283–1290, GE, Moreau CS, Mackintosh S, Edmondson diabetic nephropathy? A meta-analysis. Di- 2008 RD, Fowlkes JL: Microalbuminuria in type 1 abetologia 52: 691–697, 2009 32. Lewandowicz A, Bakun M, Kohutnicki R, diabetes is associated with enhanced excre- 19. Pagtalunan ME, Miller PL, Jumping-Eagle S, Fabijanska A, Kistowski M, Imiela J, Dadlez tion of the endocytic multiligand receptors Nelson RG, Myers BD, Rennke HG, Coplon M: Changes in urine proteome accompany- megalin and cubilin. Diabetes Care 32: NS, Sun L, Meyer TW: Podocyte loss and ing diabetic nephropathy progression. Pol 1266–1268, 2009 progressive glomerular injury in type II di- Arch Med Wewn 125: 27–38, 2015 45. Merchant ML, Perkins BA, Boratyn GM, abetes. J Clin Invest 99: 342–348, 1997 33. Gerth J, Cohen CD, Hopfer U, Lindenmeyer Ficociello LH, Wilkey DW, Barati MT, Bertram 20. Meier M, Kaiser T, Herrmann A, Knueppel S, MT,SommerM,GröneHJ,WolfG:Collagen CC, Page GP, Rovin BH, Warram JH, Hillmann M, Koester P, Danne T, Haller H, type VIII expression in human diabetic ne- Krolewski AS, Klein JB: Urinary peptidome Fliser D, Mischak H: Identification of urinary phropathy. Eur J Clin Invest 37: 767–773, may predict renal function decline in type 1 protein pattern in type 1 diabetic adoles- 2007 diabetes and microalbuminuria. JAmSoc cents with early diabetic nephropathy by a 34. Loeffler I, Hopfer U, Koczan D, Wolf G: Type Nephrol 20: 2065–2074, 2009 novel combined proteome analysis. JDiabetes VIII collagen modulates TGF-b1-induced 46. Brito PL, Fioretto P, Drummond K, Kim Y, Complications 19: 223–232, 2005 proliferation of mesangial cells. JAmSoc Steffes MW, Basgen JM, Sisson-Ross S, 21. Maahs DM, Siwy J, Argilés A, Cerna M, Delles Nephrol 22: 649–663, 2011 Mauer M: Proximal tubular basement mem- C, Dominiczak AF, Gayrard N, Iphöfer A, 35. Nakagawa T, Kosugi T, Haneda M, Rivard CJ, brane width in insulin-dependent diabetes Jänsch L, Jerums G, Medek K, Mischak H, Long DA: Abnormal angiogenesis in diabetic mellitus. Kidney Int 53: 754–761, 1998 Navis GJ, Roob JM, Rossing K, Rossing P, nephropathy. Diabetes 58: 1471–1478, 2009 47. Pollock CA, Poronnik P: Albumin transport Rychlík I, Schiffer E, Schmieder RE, Wascher 36. Yamamoto Y, Maeshima Y, Kitayama H, and processing by the proximal tubule: TC, Winklhofer-Roob BM, Zimmerli LU, Kitamura S, Takazawa Y, Sugiyama H, Physiology and pathophysiology. Curr Opin Zürbig P, Snell-Bergeon JK: Urinary collagen Yamasaki Y, Makino H: Tumstatin peptide, an Nephrol Hypertens 16: 359–364, 2007 fragments are significantly altered in di- inhibitor of angiogenesis, prevents glomer- 48. Hryciw DH, Lee EM, Pollock CA, Poronnik P: abetes: A link to pathophysiology. PLoS One ular hypertrophy in the early stage of diabetic Molecular changes in proximal tubule func- 5: e13051, 2010 nephropathy. Diabetes 53: 1831–1840, 2004 tion in diabetes mellitus. Clin Exp Pharmacol 22. Zhang M, Fu G, Lei T: Two urinary peptides 37. Ichinose K, Maeshima Y, Yamamoto Y, Physiol 31: 372–379, 2004 associated closely with type 2 diabetes mel- Kitayama H, Takazawa Y, Hirokoshi K, 49. Christensen EI, Birn H: Megalin and cubilin: litus. PLoS One 10: e0122950, 2015 Sugiyama H, Yamasaki Y, Eguchi K, Makino Multifunctional endocytic receptors. Nat Rev 23. Mauer SM: Structural-functional correlations H: Antiangiogenic endostatin peptide ame- Mol Cell Biol 3: 256–266, 2002 of diabetic nephropathy. Kidney Int 45: 612– liorates renal alterations in the early stage 50. Navarro-González JF, Mora-Fernández C: 622, 1994 of a type 1 diabetic nephropathy model. Di- The role of inflammatory cytokines in di- 24. Chavers BM, Bilous RW, Ellis EN, Steffes MW, abetes 54: 2891–2903, 2005 abetic nephropathy. J Am Soc Nephrol 19: Mauer SM: Glomerular lesions and urinary 38. Mauer SM, Steffes MW, Ellis EN, Sutherland 433–442, 2008 albumin excretion in type I diabetes without DE, Brown DM, Goetz FC: Structural-func- 51. Phillips MC: Molecular mechanisms of cellular overt proteinuria. N Engl J Med 320: 966– tional relationships in diabetic nephropathy. cholesterol efflux. JBiolChem289: 24020– 970, 1989 J Clin Invest 74: 1143–1155, 1984 24029, 2014 25. Liu F, Brezniceanu ML, Wei CC, Chénier I, 39. Steffes MW, Osterby R, Chavers B, Mauer 52. O’Rourke L, Yeaman SJ, Shepherd PR: Insulin Sachetelli S, Zhang SL, Filep JG, Ingelfinger SM: Mesangial expansion as a central and leptin acutely regulate cholesterol ester JR, Chan JS: Overexpression of angio- mechanism for loss of kidney function in metabolism in macrophages by novel sig- tensinogen increases tubular apoptosis in diabetic patients. Diabetes 38: 1077– naling pathways. Diabetes 50: 955–961, diabetes. J Am Soc Nephrol 19: 269–280, 1081, 1989 2001 2008 40. Orchard TJ, Dorman JS, Maser RE, Becker 53. O’Rourke L, Gronning LM, Yeaman SJ, 26. Bonventre JV: Can we target tubular damage DJ, Drash AL, Ellis D, LaPorte RE, Kuller LH: Shepherd PR: Glucose-dependent regula- to prevent renal function decline in diabetes? Prevalence of complications in IDDM by sex tion of cholesterol ester metabolism in mac- Semin Nephrol 32: 452–462, 2012 and duration. Pittsburgh Epidemiology of rophages by insulin and leptin. JBiolChem 27. Vaidya VS, Niewczas MA, Ficociello LH, Diabetes Complications Study II. Diabetes 277: 42557–42562, 2002 Johnson AC, Collings FB, Warram JH, 39: 1116–1124, 1990 54. Pedigo CE, Ducasa GM, Leclercq F, Sloan A, Krolewski AS, Bonventre JV: Regression of 41. Newman DJ, Mattock MB, Dawnay AB, Mitrofanova A, Hashmi T, Molina-David J, Ge microalbuminuria in type 1 diabetes is asso- KerryS,McGuireA,YaqoobM,Hitman M, Lassenius MI, Forsblom C, Lehto M, ciated with lower levels of urinary tubular GA, Hawke C: Systematic review on urine al- Groop PH, Kretzler M, Eddy S, Martini S, injury biomarkers, kidney injury molecule-1, bumin testing for early detection of diabetic Reich H, Wahl P, Ghiggeri G, Faul C, Burke

1060 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1050–1061, 2017 www.jasn.org BRIEF REVIEW

GW 3rd, Kretz O, Huber TB, Mendez AJ, renal disease in patients with type 1 diabetes. (non-insulin-dependent) diabetes mellitus in Merscher S, Fornoni A: Local TNF causes JAMA 294: 1782–1787, 2005 Pima Indians. Diabetologia 31: 730–736, NFATc1-dependent cholesterol-mediated 63. Dihazi H, Müller GA, Lindner S, Meyer M, Asif 1988 podocyte injury. J Clin Invest 126: 3336– AR, Oellerich M, Strutz F: Characterization of 69. Furie B, Furie BC: The molecular basis of 3350, 2016 diabetic nephropathy by urinary proteomic blood coagulation. Cell 53: 505–518, 1988 55. Winocour PD, Watala C, Perry DW, Kinlough- analysis: Identification of a processed ubiq- 70.LongAT,KenneE,JungR,FuchsTA,Renné Rathbone RL: Decreased platelet membrane uitin form as a differentially excreted protein T: Contact system revisited: An interface fluidity due to glycation or acetylation of in diabetic nephropathy patients. Clin Chem between inflammation, coagulation, and in- membrane proteins. Thromb Haemost 68: 53: 1636–1645, 2007 nate immunity. JThrombHaemost14: 427– 577–582, 1992 64. Sharma K, Lee S, Han S, Lee S, Francos B, McCue 437, 2016 56. Swaminathan S, Fonseca VA, Alam MG, Shah P, Wassell R, Shaw MA, RamachandraRao SP: 71. Pieper R, Gatlin CL, McGrath AM, Makusky AJ, SV: The role of iron in diabetes and its com- Two-dimensional fluorescence difference Mondal M, Seonarain M, Field E, Schatz CR, plications. Diabetes Care 30: 1926–1933, gel electrophoresis analysis of the urine Estock MA, Ahmed N, Anderson NG, Steiner 2007 proteome in human diabetic nephropathy. S: Characterization of the human urinary pro- 57. Dixon SJ, Stockwell BR: The role of iron and Proteomics 5: 2648–2655, 2005 teome: A method for high-resolution display of reactive oxygen species in cell death. Nat 65. Rao PV, Lu X, Standley M, Pattee P, Neelima urinary proteins on two-dimensional electro- Chem Biol 10: 9–17, 2014 G, Girisesh G, Dakshinamurthy KV, Roberts phoresis gels with a yield of nearly 1400 dis- 58. Haber F, Weiss J: The catalytic de- CT Jr., Nagalla SR: Proteomic identification tinct protein spots. Proteomics 4: 1159–1174, composition of hydrogen peroxide by iron of urinary biomarkers of diabetic nephropa- 2004 salts. Proc R Soc A Math Phys Eng Sci 147: thy. Diabetes Care 30: 629–637, 2007 72. Klein J, Eales J, Zürbig P, Vlahou A, Mischak 332–351, 1934 66.OtuHH,CanH,SpentzosD,NelsonRG, H, Stevens R: Proteasix: A tool for automated 59. Kehrer JP: The Haber-Weiss reaction and Hanson RL, Looker HC, Knowler WC, Monroy and large-scale prediction of proteases in- mechanisms of toxicity. Toxicology 149: 43– M, Libermann TA, Karumanchi SA, Thadhani volved in naturally occurring peptide gener- 50, 2000 R: Prediction of diabetic nephropathy using ation. Proteomics 13: 1077–1082, 2013 60. Perkins BA, Ficociello LH, Ostrander BE, Silva urine proteomic profiling 10 years prior to 73. Guerrero A, Dallas DC, Contreras S, Chee S, KH, Weinberg J, Warram JH, Krolewski AS: development of nephropathy. Diabetes Parker EA, Sun X, Dimapasoc L, Barile D, Microalbuminuria and the risk for early pro- Care 30: 638–643, 2007 German JB, Lebrilla CB: Mechanistic peptido- gressive renal function decline in type 1 di- 67. Knowler WC, Bennett PH, Hamman RF, mics: Factors that dictate specificity in the for- abetes. J Am Soc Nephrol 18: 1353–1361, Miller M: Diabetes incidence and prevalence mation of endogenous peptides in human milk. 2007 in Pima Indians: A 19-fold greater incidence MolCellProteomics13: 3343–3351, 2014 61. Pavkov ME, Knowler WC, Lemley KV, Mason than in Rochester, Minnesota. Am J Epi- CC, Myers BD, Nelson RG: Early renal func- demiol 108: 497–505, 1978 tiondeclineintype2diabetes.Clin J Am Soc 68. Nelson RG, Newman JM, Knowler WC, – Nephrol 7: 78 84, 2012 Sievers ML, Kunzelman CL, Pettitt DJ, This article contains supplemental material online 62. Finne P, Reunanen A, Stenman S, Groop PH, Moffett CD, Teutsch SM, Bennett PH: In- at http://jasn.asnjournals.org/lookup/suppl/doi:10. Grönhagen-Riska C: Incidence of end-stage cidence of end-stage renal disease in type 2 1681/ASN.2016091018/-/DCSupplemental.

J Am Soc Nephrol 28: 1050–1061, 2017 Urinary Proteomics in Diabetes 1061

Supplemental Table 1. Summary of study characteristics.

Study Population Technique Significant Findings Mischak et Case: 112 T2D adults with varying UAE CE-MS/MS; >800 Da Associated with renal as indicated by al., 20041 46 normal (< 20 mg/L) cut-off with ≥ 2 charge increased UAE 45 low grade (> 20 mg/L) states; MosaiquesVisu - ↑ ALB 21 high grade (> 100 mg/L) and Mascot software - ↓ INSL3, UMOD Control: 39 healthy non-smoker adults All: had baseline SCr concentrations of < 1.3 mg/dl; no heart failure, cancer, and chronic inflammatory disease Meier et al., Case: 44 T1D adolescents CE-MS; >800 Da cut-off Identified a urinary protein pattern for 20052 Control: 9 healthy age-matched volunteers with ≥2 charge states; early nephropathy that was different from MosaiquesVisu software that of controls Jain et al., Case: 100 T2D adults with microalbuminuria 2DE and MALDI- T2D with microalbuminuria v. controls 20053 Control: 20 healthy adults MS/MS; 5-mL cut-off; - ↑ AMBP, AZGP1, ORM 18 with diabetes and normoalbuminuria ‘Ettan MALDI Software’ 1 without diabetes and with with ‘proteo Metrics microalbuminuria LLC’ search engine 1 without microalbuminuria and diabetes Sharma et Case: 3 adults with macroalbuminuria and Fluorescence-based Diabetic nephropathy v. controls al., 20054 impaired renal function DIGE and SELDI- - ↑ SERPINA1 2 T1D and 1 T2D MS/MS; Profound search Validated SERPINA1 via ELISA Control: 5 healthy adults engine and Mascot software Dihazi et al., Case: 117 adults with varying disease GE and SELDI-MS/MS; Diabetic nephropathy v. controls 20075 38 with T2D nephropathy >1kDa cut-off; MSDB - ↑ B2M 45 with uncomplicated T2D and SwissProt dabatases; - ↓ UBA52 34 with non-diabetic nephropathy ProteinChip array Control: 45 healthy adults

Study Population Technique Significant Findings Otu et al., Case: 31 T2D adults with nephropathy SELDI-MS; 2-40 kDa Identified a urinary protein pattern for T2D 20076 Control: 31 T2D adults with range with ≥2 signal-to- nephropathy cases that was different from normoalbuminuria; matched for diabetes noise ratio Ciphergen that of controls duration, age, sex, and BMI Biomarker Wizard Survivor bias due to case-control design; All: had baseline SCr concentrations of ≤ 1.2 software unknown stability of urine samples stored mg/dl; Pima Indians for 10 years Rao et al., Case: 33 T2D adults with varying UAE DIGE-MS/MS; T1D with macroalbuminuria v. 20077 10 with normoalbuminuria immunodepletion of normoalbuminuria 13 with microalbuminuria abundant serum proteins; - ↑ A1BG, AHSG, AZGP1, GC, HPX, 10 with macroalbuminuria Protein-Lynx Global LRG1, SERPINA1, S100A9 Control: 9 healthy adults Server software using - ↓ AMBP, APOA1, RBP, TTR SwissProt database Bellei et al., Case: 24 T2D adults with varying UAE 2DE and LC-MS/MS; T2D nephropathy v. T2D and controls 20088 10 with normoalbuminuria Mascot software using - ↑ APOH, B2M, CAH, IGKC, RBP, 13 with micro- or macroalbuminuria SwissProt Database TTR Control: 12 healthy adults - ↓ PPAP, RNA3, KLK3 Rossing et Case: 89 T1D adults with varying UAE CE-MS and LC-MALDI- T1D with normoalbuminuria v. controls al., 20089 30 with normoalbuminuria MS/MS; <30 kDa cut- - ↓ COL1A1, UMOD 29 with microalbuminuria off; MosaiquesVisu T1D with macroalbuminuria v. 30 with macroalbuminuria software microalbuminuria Control: 30 healthy adults - ↑ ALB, AHSG, B2M, SERPINA1, TTR - ↓ COL (1A1, 1A2, 3A1), FGB, PGRMC1, PSORS1C2, UMOD Lapolla et Case: 30 adults with varying diseases LC-MS/MS; <30 kDa T2D nephropathy v. T2D and controls al., 200910 10 T2D without nephropathy cut-off; ProteinPilot v2.1 - ↓ COL1A1, UMOD 10 T2D with nephropathy using Uniprot Database 10 non-diabetic nephropathy Control: 10 healthy adults

Study Population Technique Significant Findings Jiang et al., Case: 162 T2D adults with varying UAE Fluorescence-based T1D with micro- or macroalbuminuria v. 200911 54 with normoalbuminuria DIGE and MALDI- controls 54 with microalbuminuria MS/MS; Mascot - ↑ ORM 54 with macroalbuminuria or nephropathy software using Swiss- Validated ORM via immunoturbidimetry: Control: 82 healthy adults; matched for age and Prot/TrEMBL protein increased urinary ORM levels was sex database associated with worsening diabetic kidney All: Of Chinese Han descent disease Jiang et al., Case: 12 adults with nephropathy 2DE and MALDI- Diabetic nephropathy v. controls 200912 6 T1D and 6 T2D MS/MS; Mascot - ↑ ALB, AZGP1, ECAD, IGKC, KNG, Control: 6 healthy adults; matched for age and software using Swiss- PTGDS, ORM, RBP sex Prot/TrEMBL protein - ↓ AMBP, HP, TTR, UMOD All: Of Chinese Han descent database; Validated ECAD via ELISA: increased urinary ECAD levels was associated with worsening diabetic kidney disease Merchant et Case: 21 T1D progressors with declining renal LC-MALDI-MS and Progressors v. non-progressors al., 200913 function MALDI-MS/MS; <10 - ↑ FAT2, IPPK, TJP3 Control: 40 stable T1D non-progressors kDa cut-off; Mascot - ↓ COL4A1, COL5A1, TNX All: from the Joslin Study of the Natural Software using NCBInr History of Microalbuminuria in Type 1 20060712 database Diabetes Snell- Case: 19 adults with coronary artery disease CE-MS; <20 kDa cut-off - Validated urinary protein patterns Bergeon et 12 with and 4 without T1D MosaiquesVisu software identified by and described in Rossing et al., 200914a Control: 19 adults without coronary artery al 2008 for T1D and T1D nephropathy disease; matched for age, diabetes status and duration, and sex All: from the Coronary Artery Calcification in Type I Diabetes (CACTI) study Thrailkill et Case: 24 T1D adults with varying UAE Fluorescence-based T1D with microalbuminuria v. other al., 200915 12 normoalbuminuria DIGE and LC-MS/MS; groups 12 microalbuminuria >3 kDa; Mascot and - ↑ CLU, CUBN, GC, LRP2, RBP4, EGF, Control: 12 healthy adults Scaffold software TF, ALB

Study Population Technique Significant Findings Alkhalaf et Case: 64 T2D adults with nephropathy and CE-MS and CE-MS/MS; Diabetic nephropathy v. controls al., 201016b retinopathy <20 kDa cut-off; Mascot - ↑ AHSG, ALB, B2M, SERPINA1, TTR Control: 82 T2D adults with normoalbuminuria software using the - ↓ CD99, COL (1A1, 1A2, 3A1), UMOD All: from the Prevention of Diabetic SwissProt database Complications (PREDICTIONS) study; aged 35-75 years and had diabetes for ≥5 years Maahs et al., Case: 587 diabetic adults (299 T1D) CE-MS/MS; <20 kDa T1D and T2D v. controls 201017a 369 with and 218 without impaired renal cut-off; Mascot and - ↑ FG (A, B), SERPINA1, function (ACR >30 mg/g or GFR >60 MDSB Protein database - ↓ COL (1A1, 1A2, 2A1, 3A1, 8A2), ml/min) PGRMC1 Control: 315 healthy adults T2D v. T1D and controls All: from 10 different hospital centers in the - ↓ COL (1A1, 1A2) US, Europe and Australia T1D v. controls - ↓ UMOD Riaz et al., Case: 100 T2D patients SDS-PAGE and LC- T2D v. controls 201018 Control: 43 healthy adults; matched for age and MALDI-MS/MS; Mascot - ↑ ALB, AZGP1, ECAD, RBP4 sex software using the - ↓ AMBP, HP, TTR All: from a double-blind placebo-controlled SwissProt database; Verified all 7 candidates via ELISA randomized clinical trial in Lahore, Pakistan ELISA for candidate validation Wu et al., Case: 75 T2D adults with varying UAE ProteinChip H50 array; Identified 4 ion fragments with differential 201119 30 normoalbuminuria SELDI-MS; < 80 kDa; excretion between T2D adults with 25 microalbuminuria support vector machine microalbuminuria and normoalbuminuria 20 macroalbuminuria learning Control: 20 healthy sex-matched adults Jin et al., Case: 43 T2D adults with diabetic retinopathy LC-MS/MS; >3 kDa cut- T2D with microalbuminuria v. T2D 201220 and persistent microalbuminuria off; ProteinPilot v.2.0.1 controls Control: 43 T2D healthy adults; matched for and the Paragon - ↑ CP, GC, HP, PSCA, ORM1, age, sex, BMI, and diabetes duration algorithm; iTRAQ for SERPINA1, TF quantification; MRM for - ↓ FABP, HSPG2, MASP2 candidate validation Verified HP, ORM, PSCA, SERPINA1, and TF via MRM

Study Population Technique Significant Findings Schlatzer et Case: 13 T1D progressors who developed LC-MS/MS; unknown Progressors v. non-progressors al., 201221 micro- or macroalbuminuria cut-off; Proteomarker - ↑ ORM Control: 11 T1D non-progressors; matched for and Mascot softwares - ↓ CLU, GRN, UMOD diabetes duration and age Verified all 4 candidates via ELISA All: from CACTI study Soggiu et al., Case: 20 T1D adults with varying UAE MALDI-MS and LC- T1D (particularly with microalbuminuria) 201222 16 normoalbuminuria MS/MS; >10 kDa cut- v. controls 4 microalbuminuria off; Mascot software - ↑ A1BG, AMBP, AZGP1, RBP4 Control: 10 healthy adults - ↓ APO (A1, E), CD59, HMWK, UMOD Zürbig et al., Case: 15 diabetic progressors who developed CE-MS; <20 kDa cut-off; Progressors v. non-progressors 201223c macroalbuminuria MosaiquesVisu software - ↑ ALB, SERPINA1 6 T1D and 9 T2D - ↓ COL1A1, CD99, CLU, PIGR, UMOD Control: 20 diabetic non-progressors who remained normoalbuminuric 10 T1D and10 T2D Bhensdadia Case: 4 T2D progressors (UAE increased by LC-MS/MS; Mascot Progressors v. non-progressors et al., 201324 >60% by the end of follow-up) with worsening software and Scaffold; - ↑ AGRN, AGT, HP, MASP2 SCr selected reaction - ↓ LAMP Control: 4 T2D non-progressors with stable monitoring (SRM) and Verified the 5 markers above, UMOD, and SCr; matched for baseline SCr and UAE ELISA for validation NGAL in 30 adults via SRM; validated HP All: from Veterans Affairs Diabetes Trial via ELISA in 204 adults (VADT) Chu et al., Case: 28 uncomplicated T2D adults LC-MALDI-MS/MS; 10 T2D v. controls 201325 Control: 29 healthy adults kDa cut-off; Bioworks - ↓ CLU, EPRS, HINT1 Browser Roscioni et Case: 44 T2D progressors with worsening UAE CE-MS/MS; <20 kDa Progressors v. non-progressors al., 201326c Control: 44 T2D non-progressors cut-off; MosaiquesVisu - ↑ AHSG All: from the Prevention of Renal and Vascular software - ↓ COL1A1 End-stage Disease (PREVEND) study Associated with UAE - ↑ AHSG, ALB, SERPINA1 - ↓ COL1A1, UMOD

Study Population Technique Significant Findings Siwy et al., Case: 87 T2D adults with nephropathy CE-MS; <20 kDa cut-off; T2D nephropathy v. control 201427c Control: 78 T2D adults with normoalbuminuria MosaiquesVisu software - ↑ A1BG, AHSG, ALB, APOA1, B2M, All: from the Proteomic Prediction and Renin SERPIN (A1, C1), TTR Angiotensin Aldosterone System Inhibition - ↓ CD99, COL (1A1, 1A2), FXYD2, Prevention of Early Diabetic Nephropathy In FGA, PGRMC1, PIGR Type 2 diabetic adults with normoalbuminuria (PRIORITY) trial Lewandowicz Case: 72 T2D adults LC-MS/MS; unknown T2D nephropathy v. control et al., 201528 33 without retinopathy or nephropathy cut-off; iTRAQ - ↑ ALB, SERPINA1 15 with diabetic retinopathy quantitation; Mascot - ↓ COL1A1, HSPG2 24 with diabetic nephropathy (and Software T2D v. control retinopathy) - ↑ ALB, PTGDS, SPP1 Control: 27 healthy adults matched for age and - ↓ ACTB, APOA1, COL1A1, HSPG2 sex Suh et al., Case: 40 T1D children and adolescents SDS-PAGE and FASP; T1D v. control 201529 Control: 41 healthy age-matched siblings LC-MS/MS; > 30 kDa; - ↑ ENPEP, NAGA, MAN2B1, CTSC, Mascot Software and FUCA1, ASAH1, GNS, FUCA2. DPP7, Protein Prophet CPQ, HEXB, CTSB, LRG1, CST2, RBP, APOM, GAS6, GP5, TIMP1, SLC3A2, SELL, CDH5, MCAM, MSLN, PI16, VCAM1, COLEC12, ALCAM, ACE2, ERP44, HK3, GSN - ↓ LAMP, KNG1, ACY3, AMN, ERP MGAM, IGFLR1, MADCAM1, CPM, RNF149, TOLLIP, HRSP12, CTTN Verified 6 proteins (FUCA2, NAGA, COLEC-12, CD166, TIMP1, and APOM) via Western Blot. (These proteins were subsequently used into our bioinformatic analyses.)

Study Population Technique Significant Findings Zhang et al., Case: 49 T2D adults divided into groups based LC-MALDI-MS/MS; T2D v. control 201530 HbA1c levels [BM] and fasting plasma glucose <10 kDa cut-off; - ↑ FGA, F2 levels [GM] ClinProt software using T2D in high BM group v. low BM group Control: 29 healthy adults matched for age and the IPI human database - ↑ FGA, F2 sex Fu et al., Case: 28 T2D adults with microalbuminuria LC-MALDI-MS/MS; T2D with microalbuminuria v. T2D 201631 Control: 30 T2D adults with normoalbuminuria <10 kDa cut-off; controls ClinProt software using - ↑ F2, FGA, VTN the IPI human database - ↓ F2, FGA ITIH4, VTN Peptides of F2, FGA, and VTN were increased and decreased Verified via MALDI-TOF-MS/MS aStudies that also validated the diabetes7 model14 bStudies that also validated the Rossing peptidome9 cStudies that also validated the CKD273 classifier32

2DE, two-dimensional gel electrophoresis; CE, capillary electrophoresis; DIGE, difference gel electrophoresis; ELISA, enzyme-linked immunosorbent assay; GE, gel electrophoresis; LC, liquid chromatography; MALDI, Matrix-assisted laser desorption/ionization; MRM, multiple reaction monitoring; SRM, selected reaction monitoring; T1D, type 1 diabetes; T2D, type 2 diabetes; UAE, urinary albumin excretion.