Single-nephron proteomes connect morphology and function in proteinuric kidney disease

In diseases of many parenchymatous organs, heterogeneous deterioration of individual functional units determines the clinical prognosis. However, the molecular characterization at the level of such individual subunits remains a technological challenge that needs to be addressed in order to better understand pathological mechanisms. Proteinuric glomerular kidney diseases are frequent and assorted diseases affecting a fraction of glomeruli and their draining tubules to variable extents, and for which no specific treatment exists. Here, we developed and applied a mass spectrometry-based methodology to investigate heterogeneity of proteomes from individually isolated nephron segments from mice with proteinuric kidney disease. In single glomeruli from two different mouse models of sclerotic glomerular disease, we identified a coherent protein expression module consisting of extracellular matrix protein deposition (reflecting glomerular sclerosis), glomerular albumin (reflecting proteinuria) and LAMP1, a lysosomal protein. This module was associated with a loss of podocyte marker proteins while genetic ablation of LAMP1-correlated lysosomal proteases could ameliorate glomerular damage in vivo. Furthermore, proteomic analyses of individual glomeruli from patients with genetic sclerotic and non-sclerotic proteinuric diseases revealed increased abundance of lysosomal proteins, in combination with a decreased abundance of mutated gene products. Thus, altered protein homeostasis (proteostasis) is a conserved key mechanism in proteinuric kidney diseases. Moreover, our technology can capture intra-individual variability in diseases of the kidney and other tissues at a sub-biopsy scale. In diseases of many parenchymatous organs, heterogeneous deterioration of individual functional units determines the clinical prognosis. However, the molecular characterization at the level of such individual subunits remains a technological challenge that needs to be addressed in order to better understand pathological mechanisms. Proteinuric glomerular kidney diseases are frequent and assorted diseases affecting a fraction of glomeruli and their draining tubules to variable extents, and for which no specific treatment exists. Here, we developed and applied a mass spectrometry-based methodology to investigate heterogeneity of proteomes from individually isolated nephron segments from mice with proteinuric kidney disease. In single glomeruli from two different mouse models of sclerotic glomerular disease, we identified a coherent protein expression module consisting of extracellular matrix protein deposition (reflecting glomerular sclerosis), glomerular albumin (reflecting proteinuria) and LAMP1, a lysosomal protein. This module was associated with a loss of podocyte marker proteins while genetic ablation of LAMP1-correlated lysosomal proteases could ameliorate glomerular damage in vivo. Furthermore, proteomic analyses of individual glomeruli from patients with genetic sclerotic and non-sclerotic proteinuric diseases revealed increased abundance of lysosomal proteins, in combination with a decreased abundance of mutated gene products. Thus, altered protein homeostasis (proteostasis) is a conserved key mechanism in proteinuric kidney diseases. Moreover, our technology can capture intra-individual variability in diseases of the kidney and other tissues at a sub-biopsy scale. Many mammalian organs consist of repetitive functional subunits at the scale of a few micrometers or hundreds of cells. Among these are islets of the pancreas, liver lobules, or kidney nephrons. In a variety of diseases, including kidney disease, decay in individual unit function and morphology determines organ function and clinical prognosis (for examples, see references 1Aguayo-Mazzucato C. van Haaren M. Mruk M. et al.β cell aging markers have heterogeneous distribution and are induced by insulin resistance.Cell Metab. 2017; 25: 898-910.e5Abstract Full Text Full Text PDF PubMed Scopus (108) Google Scholar, 2Weening J.J. D'Agati V.D. Schwartz M.M. et al.The classification of glomerulonephritis in systemic lupus erythematosus revisited.J Am Soc Nephrol. 2004; 15: 241-250Crossref PubMed Scopus (1462) Google Scholar, 3Ishak K. Baptista A. Bianchi L. et al.Histological grading and staging of chronic hepatitis.J Hepatol. 1995; 22: 696-699Abstract Full Text PDF PubMed Scopus (4142) Google Scholar, 4Loupy A. Haas M. Solez K. et al.The Banff 2015 Kidney Meeting Report: current challenges in rejection classification and prospects for adopting molecular pathology.Am J Transplant. 2017; 17: 28-41Crossref PubMed Scopus (454) Google Scholar). Because morphologies (as obtained in the histology of biopsy specimens) are frequently inconclusive, molecular patterns in native biomaterial could probably provide a "molecular diagnosis," thereby revealing treatment options and a personalized prognosis. In fact, in many of these diseases, transcriptomic patterns have been analyzed and yielded preliminary insights into the classification and mechanisms of diseases ("integrative genomics").5Ju W. Smith S. Kretzler M. Genomic biomarkers for chronic kidney disease.Transl Res J Lab Clin Med. 2012; 159: 290-302Abstract Full Text Full Text PDF Scopus (30) Google Scholar, 6Sampson M.G. Hodgin J.B. Kretzler M. Defining nephrotic syndrome from an integrative genomics perspective.Pediatr Nephrol. 2015; 30 (quiz 59): 51-63Crossref PubMed Scopus (17) Google Scholar, 7Susztak K. Understanding the epigenetic syntax for the genetic alphabet in the kidney.J Am Soc Nephrol. 2014; 25: 10-17Crossref PubMed Scopus (52) Google Scholar, 8Genovese G. Friedman D.J. Ross M.D. et al.Association of trypanolytic ApoL1 variants with kidney disease in African Americans.Science. 2010; 329: 841-845Crossref PubMed Scopus (1402) Google Scholar However, transcripts can only partially explain protein abundance, underlining the importance of proteome data, especially in its targeted form.9Liu Y. Beyer A. Aebersold R. On the dependency of cellular protein levels on mRNA abundance.Cell. 2016; 165: 535-550Abstract Full Text Full Text PDF PubMed Scopus (1386) Google Scholar Unfortunately, only a few highly specialized analytical pipelines analyzing pathophysiological processes at the single functional unit level have been developed, and they have not been applied to patient samples so far, mainly because the mass spectrometer sensitivity is commonly considered to be insufficient.10Waanders L.F. Chwalek K. Monetti M. et al.Quantitative proteomic analysis of single pancreatic islets.Proc Natl Acad Sci U S A. 2009; 106: 18902-18907Crossref PubMed Scopus (180) Google Scholar Immunostainings are an alternative approach. However, antibodies have a number of other limitations, among these are recognition of unspecific epitopes, limited multiplexing capabilities, and inaccurate quantification of signal intensities caused by background signals and signal saturation.11Couchman J.R. Commercial Antibodies: The good, bad, and really ugly.J Histochem Cytochem. 2009; 57: 7-8Crossref PubMed Scopus (80) Google Scholar, 12Uhlen M. Bandrowski A. Carr S. et al.A proposal for validation of antibodies.Nat Methods. 2016; 13: 823-827Crossref PubMed Scopus (344) Google Scholar Other means to assess heterogeneity are single-cell approaches. Although they are powerful in assessing heterogeneity, they disrupt the initial tissue structure and allow only limited insights into extracellular matrix protein abundance, a strong pathologic criterion in a variety of fibrotic and chronic diseases. The human kidney is one of the most complex parenchymatous organs and comprises ∼1 million functional units, the nephrons. Each nephron consists of a renal glomerulus and a draining tubule. The glomerulus is the site of primary urine production. It harbors a 3-layered filtration barrier that consists of podocytes, endothelial cells, and the glomerular basement membrane. The primary urine is passed to a draining tubule that consists of various segments with defined functions, including metabolite reabsorption. Decay of individual nephron function is a characteristic feature of chronic disease. Chronic kidney disease (CKD) is very frequent (1 in 7 Americans13Chronic Kidney Disease (CKD) Surveillance Project. Available at: https://nccd.cdc.gov/ckd/. Accessed October 2017.Google Scholar) and a severe risk factor for cardiovascular events and stroke.14Meguid El Nahas A. Bello A.K. Chronic kidney disease: the global challenge.Lancet. 2005; 365: 331-340Abstract Full Text Full Text PDF PubMed Scopus (890) Google Scholar Decay in glomerular function is diagnosed frequently as proteinuric kidney disease, such as in diabetic nephropathy and focal segmental glomerulosclerosis (FSGS). FSGS is triggered by a variety of causes, including genetic mutations (WT1, NPHS1), and chemical and inflammatory stimuli.15Bierzynska A. Soderquest K. Koziell A. Genes and podocytes–new insights into mechanisms of podocytopathy.Front Endocrinol. 2014; 5: 226PubMed Google Scholar These insults lead to the loss of podocyte function and sclerosis, typically in a heterogeneous pattern. Here, we introduce a scalable method to directly extract quantitative proteomic information from single units of the kidney (i.e., single glomeruli and single tubule segments). With this technology, we are able to quantify the proteomes in kidney nephron segments—glomeruli and tubules—consisting of as few as 200 cells16Garg L.C. Knepper M.A. Burg M.B. Mineralocorticoid effects on Na-K-ATPase in individual nephron segments.Am J Physiol. 1981; 240: F536-F544PubMed Google Scholar and monitor podocyte marker proteins from as few as 80 podocytes per glomerulus independently of antibodies.17Puelles V.G. van der Wolde J.W. Schulze K.E. et al.Validation of a three-dimensional method for counting and sizing podocytes in whole glomeruli.J Am Soc Nephrol. 2016; 27: 3093-3104Crossref PubMed Scopus (45) Google Scholar We here describe an analytical pipeline to integrate causative disease mechanisms in heterogeneous nephron populations at a subbiopsy scale. To analyze the proteome of a single glomerulus, we modified a protocol that was designed for very small sample amounts.18Hughes C.S. Foehr S. Garfield D.A. et al.Ultrasensitive proteome analysis using paramagnetic bead technology.Mol Syst Biol. 2014; 10: 757Crossref PubMed Scopus (513) Google Scholar Commonly used C18-based sample preparation protocols use relatively large volumes and have a suboptimal sample recovery rate. In contrast, this protocol minimizes sample loss by tight binding of proteins and peptides to carboxylated magnetic beads during protein purification (Figure 1a), making it compatible with most comprehensive lysis buffers (up to 10% sodium dodecylsulfate). We microdissected wild-type mouse glomeruli and subjected them, one at a time, to proteomic analysis (Figure 1b). The ultrasensitive sample preparation method largely outperformed the C18-based, "standard" sample preparation (stage tips, Supplementary Figure S1). Proteomic analysis by liquid chromatography (LC)–tandem mass spectrometry (MS)/MS showed a substantial ion current signal from a single mouse glomerulus (Figure 2a). As expected, the method was even more successful for larger human glomeruli (Figure 2b). The single glomerulus datasets contained core podocyte proteins such as nephrin, ACTN4, podocin, and CD2AP, proteins of the basement membrane such as collagen type 4 and markers of mesangial cells such as desmin. Next, we analyzed the proteome of anatomically defined, microdissected mouse tubule segments, which is, to our knowledge, the first proteomic analysis of these structures that complements recent transcriptome acquisitions.19Lee J.W. Chou C.-L. Knepper M.A. Deep sequencing in microdissected renal tubules identifies nephron segment-specific transcriptomes.J Am Soc Nephrol. 2015; 26: 2669-2677Crossref PubMed Scopus (344) Google Scholar Again, shotgun proteomic analysis could clearly distinguish single proximal tubules (S1 segments), thick ascending limbs, and cortical collecting ducts by means of known marker protein expression (Figure 2c) and on the global proteome level (Supplementary Figure S2A and B). As expected, thicker segments yielded more peptides than thinner segments, such as a thick ascending loop of Henle (Supplementary Figure S2C). In proximal tubules, the data covered >1500 proteins, and the abundance spanned 4 orders of magnitude in single tubules (Supplementary Figure S2D). In a similar fashion, we also determined the proteome of single proximal tubules from human samples (Supplementary Figure S2E and F).Figure 2Development of a single-segment proteomic method and application to renal segments of the WT1 heterozygous mouse. (a) Total ion current during liquid chromatography tandem mass spectrometery acquisition of samples from 0 (vehicle solution only) and 1 mouse glomerulus. (b) Comparison of identified peptides from mouse and human glomeruli as determined by ultrasensitive proteomics and MaxQuant analysis (12 single glomeruli from 3 mice and 18 single glomeruli from 3 humans). +, mean; − (horizontal lines on box plot) median. *P < 0.01 in a 2-tailed t test. Boxes indicate 25% to 75% percentiles. Outliers beyond the 95% percentile are marked with filled circles. (c) Analysis of single microdissected tubules from a mouse kidney cortex. S1 proximal tubules, cortical thick ascending limb (TAL), and cortical collecting ducts (CCDs) are clearly discernible by proteomic analysis compared with "empty" (0) samples. A selection of proteins is clustered with their gene symbol. The proteins in the first 3 rows are nonsegment-specific tubular proteins such as Na/K ATPase (Atp1b1). (d) A volcano plot of single glomeruli. Proteomics analysis of glomeruli obtained from a WT1 heterozygous knockout mouse (Wt1het) and wild-type mice as controls. Negative log (P value) of a 2-tailed t test is plotted against a log2 fold change of the Wt1het/wild type. Quantification is based on 20 single glomeruli from 3 different Wt1het animals and 20 single glomeruli from 3 different wild-type controls. (e) Hierarchical clustering of correlation coefficients across samples. (f) Volcano plot quantification of single S1 tubules microdissected from wild-type versus control mice. Amino acid transporters are marked with blue, and other transporters are indicated in red. (g) Selective aminoaciduria of proteinuric Wt1het mice for proline (a substrate of SLC6A20), lysine (a substrate of SLC3A1), and glutamate (a lower affinity substrate of SLC13A3). N = 6, *P < 0.05 2-tailed t test.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Because proteomes from single functional units could be resolved, we studied a disease model of WT1 haploinsufficiency. WT1 is a podocyte-specific transcription factor controlling expression of various podocyte-specific genes; Wt1 heterozygous (Wt1het) mice develop proteinuria and partially sclerotic lesions at the age of 14 weeks resembling FSGS.20Schumacher V.A. Schlötzer-Schrehardt U. Karumanchi S.A. et al.WT1-dependent sulfatase expression maintains the normal glomerular filtration barrier.J Am Soc Nephrol. 2011; 22: 1286-1296Crossref PubMed Scopus (51) Google Scholar Shotgun proteomics applications, as used for initial testing of the method, are inherently limited due to stochastic acquisition and undersampling.21Aebersold R. Mann M. Mass-spectrometric exploration of proteome structure and function.Nature. 2016; 537: 347-355Crossref PubMed Scopus (1105) Google Scholar To address heterogeneity and protein expression wiring independently of stochastic acquisition, we set up a targeted proteomics assay, specifically parallel reaction monitoring (PRM). We developed a "podocyte function sentinel assay" that comprises 20 proteins (41 peptides) that were selected based on the known, genetically verified importance for podocyte function and that were all detected in the global glomerular proteomes (Supplementary Table S1). We quantified these proteins using the assay in single glomeruli from mice heterozygous for the transcription factor WT1 (Wt1het) and wild-type mice (3 mice per group, 7 single glomeruli/mouse) (Supplementary Table S2). We found that albumin as well as the extracellular matrix proteins collagen type IV and laminin were significantly increased in single glomeruli from Wt1het mice (Figure 2d), a finding consistent with FSGS and proteinuria in these animals (Supplementary Figure S3A and B). Yet a large variation between different glomeruli was observed (Supplementary Figure S3C illustrates some of the individual measurements for proteins). To illustrate this heterogeneity, we calculated the correlation of all proteins across all glomeruli (correlations with each other) and performed hierarchical clustering of the correlation coefficients. As an example, a strong correlation occurred between laminin 5 and the lysosomal marker LAMP1 (Figure 2e). LAMP1 and ACTN1/4 correlated negatively (Figure 2e). For some proteins, there was no correlation (i.e., between nephrin and ACTN1/4) (Figure 2e). Next, we analyzed the proximal tubule of the same proteinuric mice. Proximal tubules alter their physiological function due to proteinuria and overwhelming albumin uptake.22Eshbach M.L. Weisz O.A. Receptor-mediated endocytosis in the proximal tubule.Annu Rev Physiol. 2017; 79: 425-448Crossref PubMed Scopus (80) Google Scholar To assess this systematically, we again developed a proteomics sentinel assay to survey proximal tubule status. In total, we monitored 74 proteins that determine key physiological proximal tubule function (such as metabolite transporters, the Na+/K+-ATPase, and key signaling molecules [Supplementary Table S3]). We isolated proximal tubules from the same Wt1het knockout mice as described previously and applied the tubule sentinel PRM assay (3 mice per group, 7 tubules per mouse). In isolated S1 tubular segments, initial correlational analysis revealed distinct clusters of proteins highly associated with each other. One of the key tasks of the proximal tubule is the reabsorption of metabolites, such as amino acids, and disruption of this function results in aminoaciduria, which is occasionally observed in nephrotic patients.23Opienska-Blauth J. Kowalska H. Aminoaciduria in the nephrotic syndrome in children.Clin Chim Acta Int. 1961; 6: 805-813Crossref PubMed Scopus (2) Google Scholar, 24Praga M. Andres A. Hernandez E. et al.Tubular dysfunction in nephrotic syndrome: incidence and prognostic implications.Nephrol Dial Transplant. 1991; 6: 683-688Crossref PubMed Scopus (6) Google Scholar Quantitative analysis (Wt1het vs. wild type) revealed that the amino acid carriers SLC3A1, SLC13A3, and SLC6A20a/b were decreased in single Wt1het mouse proximal tubules (Figure 2f). Interestingly, only a few tubules regulated these transporters: those particular tubules that expressed low abundance of these transporters contained high expression of other amino acid transporters with an overlapping amino acid transport spectrum (Supplementary Figure S4A). Furthermore, these tubules also showed increased abundance of albumin and collagen type IV (Figure 2f, Supplementary Figure S4A). Consistent with the stochastic decrease in these 3 transporters, we found moderately increased levels of urinary lysine (one of the substrates of SLC3A125Lee W.S. Wells R.G. Sabbag R.V. et al.Cloning and chromosomal localization of a human kidney cDNA involved in cystine, dibasic, and neutral amino acid transport.J Clin Invest. 1993; 91: 1959-1963Crossref PubMed Scopus (117) Google Scholar) and proline (one of the substrates of SLC6A2026Takanaga H. Mackenzie B. Suzuki Y. et al.Identification of mammalian proline transporter SIT1 (SLC6A20) with characteristics of classical system imino.J Biol Chem. 2005; 280: 8974-8984Crossref PubMed Scopus (124) Google Scholar), and glutamate (one of the low-affinity substrates of SLC13A327Chen X. Tsukaguchi H. Chen X.Z. et al.Molecular and functional analysis of SDCT2, a novel rat sodium-dependent dicarboxylate transporter.J Clin Invest. 1999; 103: 1159-1168Crossref PubMed Scopus (92) Google Scholar) in the urine of proteinuric mice (Figure 2g); changes in the other amino acids were not significant (Supplementary Figure S4B). These data demonstrate that single-segment proteomics can reflect pathomechanisms in proteinuric kidney disease. We decided to follow up on the glomerular function in a second FSGS model using single glomerular proteomics analysis and the same sentinel assay. We used the model of doxorubicin- (trade name: Adriamycin) induced FSGS and proteinuria (Supplementary Figure S5). We found that in contrast to the Wt1het FSGS model, there was a significant decrease in nephrin (gene symbol Nphs1) in these glomeruli (Figure 3a, Supplementary Table S4). To determine similarities of the Wt1het and the doxorubicin damage models, we compared fold changes of the observed proteins between both datasets, unraveling that albumin, laminin, collagen, and LAMP1 were upregulated in both models, and ACTN1/4 and CD2AP were decreased in both models (Figure 3b). We plotted correlation coefficients between all protein pairs in both the doxorubicin model and the Wt1het model (Figure 3c). Interestingly, a common finding between both models was the positive correlation of the lysosomal marker LAMP1 with glomerular albumin and extracellular matrix proteins, both markers of glomerular damage (Figure 3c, upper right quadrant). Thus, for a number of proteins, coexpression is wired across single glomeruli and across damage, and LAMP1 expression was connected with the extracellular matrix and albumin amount in single glomeruli in 2 models, suggesting that an increase in LAMP1 is a common hallmark of the glomerular kidney diseases examined. LAMP1 is a structural protein of the lysosome and may thus not be suitable as a key therapeutic target in renal disease. Therefore, we focused on components of the lysosomes of higher accessibility that could be better druggable targets. Cathepsin L as well as other cathepsin proteases are expressed in cultured podocytes and possibly functional in podocytes based on degradomics data.28Rinschen M.M. Hoppe A.-K. Grahammer F. et al.N-degradomic analysis reveals a proteolytic network processing the podocyte cytoskeleton.J Am Soc Nephrol. 2017; 28: 2867-2878Crossref PubMed Scopus (32) Google Scholar, 29Garsen M. Rops A.L.W.M.M. Dijkman H. et al.Cathepsin L is crucial for the development of early experimental diabetic nephropathy.Kidney Int. 2016; 90: 1012-1022Abstract Full Text Full Text PDF PubMed Scopus (46) Google Scholar, 30Rinschen M.M. Schroeter C.B. Koehler S. et al.Quantitative deep mapping of the cultured podocyte proteome uncovers shifts in proteostatic mechanisms during differentiation.Am J Physiol Cell Physiol. 2016; 311: C404-C417Crossref PubMed Scopus (27) Google Scholar, 31Yaddanapudi S. Altintas M.M. Kistler A.D. et al.CD2AP in mouse and human podocytes controls a proteolytic program that regulates cytoskeletal structure and cellular survival.J Clin Invest. 2011; 121: 3965-3980Crossref PubMed Scopus (114) Google Scholar Because cathepsins can also act outside of lysosomes,32Boya P. Kroemer G. Lysosomal membrane permeabilization in cell death.Oncogene. 2008; 27: 6434-6451Crossref PubMed Scopus (1039) Google Scholar we investigated whether lysosomal (LAMP1) abundance is linked to cathepsin abundance. To this end, we designed a targeted proteomic assay to monitor the expression of the 3 proteases (cathepsin B, L, and Z) and LAMP1 (Supplementary Table S5). In single glomeruli of the Wt1het model of podocyte damage, there was a significant increase in cathepsin B and Z, whereas cathepsin L was increased, but not significantly (Figure 3d). Both cathepsin B and cathepsin L correlated significantly (P < 0.0001) with LAMP1 abundance (Figure 3e). Somewhat similar, in single glomeruli from proteinuric doxorubicin-treated mice, cathepsin B and Z were found to be increased (Figure 3f), and both cathepsin B and Z significantly correlated with LAMP1 abundance (Figure 3g). To systematically test whether LAMP1-correlated cathepsins may be causative for glomerular damage, we used constitutive knockout mice for cathepsin B, L, and Z. Under baseline conditions, the knockout animals were not proteinuric (Supplementary Figure S6A). However, all 3 knockout strains, especially cathepsin B knockouts, showed more resistance and faster recovery after glomerular damage or podocyte injury induced by nephrotoxic serum (Figure 3h, Supplementary Figure S6B). These findings demonstrate that single-unit proteomic analysis of glomeruli can resolve orchestrated glomerular cathepsin activity, thereby adding LAMP1-correlated cathepsin B as an important causal mediator in different modes of glomerular injury. Here, we found that in 2 independent models of glomerular disease, LAMP1 was part of a protein coexpression module also consisting of cathepsin proteases, albumin, and extracellular matrix proteins. Therefore, we asked whether this module of proteins could also be found to be disease driving in human proteinuric kidney diseases. We compared the proteome of single glomeruli from patients with nephrotic syndrome with those from tumor nephrectomy samples from adult patients and again measured the proteome of each single glomerulus separately. In single glomeruli from a nephrectomy sample from a 5-year-old patient with primary idiopathic steroid-resistant FSGS, abundance of LAMP1 and SCARB2, an alternative lysosomal marker protein, was significantly increased (Figure 4a). In addition, increased abundance of extracellular matrix proteins was detected compared with a control kidney (Figure 4a). We also microdissected and measured single glomeruli from nephrectomy samples from 2 patients with congenital nephrotic syndrome and minimal change disease caused by nephrin (NPHS1) mutation (see Supplementary Table S6 for the dataset, Supplementary Table S7 for details on the clinical presentation, and Supplementary Figure S7 for histology of patients and controls). We identified >2000 proteins in this dataset and quantified >1000 proteins in all samples. There was a strong reduction of nephrin in the samples obtained from the glomeruli with NPHS1 mutation (Figure 4b), and lysosomal marker LAMP1 was increased. These results could be confirmed when comparing single glomeruli of the second NPHS1 patient with those of a second control kidney (Figure 4c). As an ancillary finding, we found reduced amounts of mitochondria in all 3 datasets analyzed compared with control cells (Supplementary Figure S8). Furthermore, the mass spectrometry data were consistent with the histopathologic diagnosis of an enlarged mesangial matrix in 1 patient (Figure 4d), which translated into an increased fraction of collagen proteins in the single glomerular proteomes (Figure 4d). To find whether we could reduce the amount of tissue even more, we used laser dissection microscopy from 10-μm thick cryosections. We could obtain clear glomerular or tubular proteome patterns that included important mediators of podocyte signaling by proteomic analysis (Supplementary Figure S9). We successfully applied advanced sample preparation33Virant-Klun I. Leicht S. Hughes C. et al.Identification of maturation-specific proteins by single-cell proteomics of human oocytes.Mol Cell Proteomics. 2016; 15: 2616-2627Abstract Full Text Full Text PDF PubMed Scopus (117) Google Scholar to analyze the protein composition of kidney tissue at a single-unit level resolution. The "one-glomerulus, one-proteome" approach described here allows the study of correlated variation of molecular markers across nephrons and within a kidney and to functionally link them to both morphology and physiology (Figure 4e). A laser microdissection–coupled proteomics approach to glomerular disease has shown great success for identifying highly abundant extracellular proteins and thereby identifying novel disease mechanisms.34Sethi S. Vrana J.A. Theis J.D. et al.Laser microdissection and mass spectrometry-based proteomics aids the diagnosis and typing of renal amyloidosis.Kidney Int. 2012; 82: 226-234Abstract Full Text Full Text PDF PubMed Scopus (145) Google Scholar By improving sensitivity and utilizing targeted proteomics, we also obtain molecular information on lower-abundant intracellular signaling proteins from individual glomeruli, which allows definition of correlative modules. These could be targeted therapeutically, and we identified cathepsin B, LAMP1, extracellular matrix proteins, and albumin as key components of a disease-driving protein module. The composition and weight of these modules, however, may markedly depend on the time of disease progression. The method has been shown to sample representative proteomes from small sample amounts.18Hughes C.S. Foehr S. Garfield D.A. et al.Ultrasensitive proteome analysis using paramagnetic bead technology.Mol Syst Biol. 2014; 10: 757Crossref PubMed Scopus (513) Google Scholar The limitations of this method are chiefly defined by the properties of mass spectrometry–based proteomic acquisitions and include a certain bias toward highly abundant proteins. Also, the method does not yet allow for enrichment and detection of posttranslational modified proteins within renal tissue (e.g., phosphorylated, ubiquitylated, and proteolytically cleaved proteins) that can identify signaling networks regulating kidney function.28Rinschen M.M. Hoppe A.-K. Grahammer F. et al.N-degradomic analysis reveals a proteolytic network processing the podocyte cytoskeleton.J Am Soc Nephrol. 2017; 28: 2867-2878Crossref PubMed Scopus (32) Google Scholar, 35Rinschen M.M. Wu X. König T.