Proteomic analyses of kidney glomerular
extracellular matrix in health and disease
A thesis submitted to the University of Manchester for the degree of Doctor of Philosophy in the Faculty of Medical and
Human Sciences
2014
Michael Randles
School of Medicine
Institute of Human Development 2
Contents
List of figures ...... 7
List of tables ...... 9
Abstract ...... 10
Declaration ...... 11
Copyright statement ...... 11
Acknowledgments ...... 12
Rationale for an alternative thesis format ...... 14
Key Abbreviations ...... 15
1 General introduction ...... 16
1.1 Overview of the thesis ...... 16
1.2 The kidney ...... 17
1.3 Glomerular filtration ...... 19
1.3.1 The glomerular filtration barrier ...... 19
1.3.2 Size and charge selective glomerular filtration ...... 22
1.3.3 The albumin retrieval hypothesis ...... 22
1.3.4 The electrokinetic model ...... 23
1.4 Microalbuminuria ...... 24
1.5 Cellular components of the glomerulus ...... 25
1.5.1 The glomerular endothelium ...... 25
1.5.2 Podocytes ...... 26
1.5.3 Mesangial cells ...... 28
1.5.4 Parietal epithelial cells ...... 28
1.6 Glomerular extracellular matrix ...... 31 3
1.6.1 The glomerular basement membrane ...... 31
1.6.2 The mesangial extracellular matrix and Bowman's capsule ...... 36
1.6.3 Regulation of the extracellular matrix by transforming growth factor beta ...... 36
1.6.4 The Matrisome ...... 38
1.7 Podocyte adhesion ...... 38
1.7.1 Integrin biology ...... 39
1.7.2 Integrin α3β1...... 41
1.7.3 CD151 ...... 41
1.7.4 Syndecans ...... 42
1.7.5 Dystroglycan ...... 42
1.7.6 Integrin mediated adhesion complexes ...... 43
1.7.7 The adhesome ...... 46
1.7.8 Slit diaphragm ...... 48
1.8 Mass spectrometry-based proteomics ...... 53
1.9 Thesis aims ...... 55
2 Defining the glomerular matrisome ...... 58
2.1 Introduction ...... 58
2.2 Statements ...... 59
2.3 Global analysis reveals the complexity of the human glomerular extracellular matrix ...... 60
2.3.1 Abstract ...... 61
2.3.2 Introduction ...... 62
2.3.3 Results ...... 64
2.3.4 Materials and Methods ...... 76
2.3.5 Acknowledgments ...... 81
2.3.6 Statement of competing financial interests ...... 81
2.3.7 Supplementary data ...... 82 4
3 Glomerular cell-derived matrices ...... 91
3.1 Introduction ...... 91
3.2 Statements ...... 92
3.3 Glomerular cell crosstalk influences the composition and assembly of extracellular matrix93
3.3.1 Abstract ...... 94
3.3.2 Introduction ...... 95
3.3.3 Results ...... 96
3.3.4 Discussion ...... 110
3.3.5 Methods ...... 112
3.3.6 Acknowledgments ...... 119
3.3.7 Statement of competing financial interests ...... 120
3.3.8 Supplementary data ...... 121
4 Genetic background alters glomerular matrix ...... 141
4.1 Introduction ...... 141
4.2 Statements ...... 142
4.3 Genetic background is a key determinant of glomerular extracellular matrix composition and
organisation ...... 143
4.3.1 Abstract ...... 144
4.3.2 Introduction ...... 145
4.3.3 Results ...... 146
4.3.4 Methods ...... 160
4.3.5 Acknowledgements ...... 166
4.3.6 Conflict of interest ...... 167
4.3.7 Supplementary data ...... 168
5 Podocyte adhesion complexes ...... 177
5.1 Introduction ...... 177 5
5.2 Statements ...... 178
5.3 The influence of extracellular matrix ligand on the assembly of podocyte adhesion complexes
179
5.3.1 Abstract ...... 180
5.3.2 Introduction ...... 181
5.3.3 Results ...... 183
5.3.4 Discussion ...... 203
5.3.5 Materials and Methods ...... 205
5.3.6 Acknowledgements ...... 209
5.3.7 Conflict of interest ...... 209
6 Nephrin associated protein complexes ...... 210
6.1 Introduction ...... 210
6.2 Statements ...... 211
6.3 The nephrin interactome ...... 212
6.3.1 Abstract ...... 213
6.3.2 Introduction ...... 213
6.3.3 Results ...... 214
6.3.4 Discussion ...... 231
6.3.5 Materials and Methods ...... 232
6.3.6 Acknowledgements ...... 238
6.3.7 Conflict of interest ...... 238
7 General discussion ...... 239
7.1 Summary of findings ...... 239
7.2 The glomerular ECM ...... 241
7.2.1 The human glomerular ECM ...... 241
7.2.2 Cell derived ECMs ...... 242 6
7.2.3 Genetic background, sex and the glomerular ECM ...... 242
7.2.4 The use of mass spectrometry to study extracellular matrix ...... 244
7.3 Adhesion signalling complexes ...... 245
7.3.1 Cell-ECM adhesion complexes ...... 245
7.3.2 Cell-cell adhesion complexes ...... 246
7.3.3 The use of mass spectrometry to study adhesion complexes ...... 247
7.4 Conclusion ...... 248
8 General materials and methods ...... 249
8.1 General reagents and equipment ...... 249
8.2 Cell culture media, reagents and plasticware ...... 249
8.3 General Buffers ...... 250
8.4 Western blotting ...... 250
8.5 Image processing ...... 252
8.6 Mass Spectrometry ...... 252
8.6.1 Mass spectrometry reagents and equipment ...... 252
8.6.2 MS data acquisition ...... 253
8.6.3 In-gel proteolytic digestion ...... 253
8.6.4 Offline peptide desalting ...... 254
9 References ...... 256
CD containing movies ...... 278
Word count: 73, 317 7
List of figures
Figure 1.1 Overview of the kidney and the nephron ...... 18
Figure 1.2 Schematic cross section of the glomerulus and the glomerular filtration barrier ...... 21
Figure 1.3 Podocyte slit diaphragm and adhesion genes mutated in nephrotic syndrome ...... 27
Figure 1.4 Common histological changes which occur in glomerular disease ...... 30
Figure 1.5 Glomerular basement membrane components ...... 34
Figure 1.6 Laminin and collagen binding integrins ...... 40
Figure 1.7 Integrin mediated adhesion complex ...... 47
Figure 1.8 Overview of the podocyte slit diaphragm ...... 52
Figure 2.1 Isolation of enriched glomerular ECM...... 65
Figure 2.2 MS analysis of enriched glomerular ECM fractions...... 66
Figure 2.3 Comparison of the glomerular ECM proteome to published glomerular proteomic datasets
...... 68
Figure 2.4 Interaction network analysis of human glomerular ECM...... 70
Figure 2.5 Localisation of glomerular ECM proteins in the Human Protein Atlas (HPA) database. .71
Figure 2.6 Co-localisation of novel and known glomerular ECM proteins...... 73
Figure S2.7 Gene ontology (GO) enrichment analysis of the full MS dataset...... 88
Figure S2.8 Glomerular ECM interaction network analysis...... 89
Figure S2.9 The most connected proteins in the ECM interaction network...... 90
Figure 3.1 Glomerular cells synthesise and organise ECM in vitro ...... 97
Figure 3.2 GEC and podocytes in vitro have differential ECM composition...... 99
Figure 3.3 GEC and podocyte ECM interaction networks...... 101
Figure 3.4 Glomerular ECM in vitro resembles a developmental phenotype...... 103
Figure 3.5 Glomerular cell coculture is associated with altered ECM organisation...... 104
Figure 3.6 Coculture ECM interaction network...... 106
Figure 3.7 The composition of coculture ECM resembles glomerular ECM in vivo ...... 107
Figure 3.8 Model of the glomerular ECM interactome...... 109
Figure 3.9 GEC ECM interaction network analysis...... 136
Figure 3.10 Podocyte ECM interaction network analysis...... 137 8
Figure 3.11 Podocytes and GEC are viable in coculture...... 138
Figure 3.12 The effects of coculture on ECM deposition and cell-cell junctions...... 139
Figure 3.13 Coculture ECM interaction network analysis...... 140
Figure 4.1 Glomerular transcriptomics...... 147
Figure 4.2 Genetic background is a regulator of ECM composition...... 149
Figure 4.3 Topological network analysis of glomerular ECM protein interactions...... 151
Figure 4.4 Network analysis of enriched glomerular ECM proteins...... 152
Figure 4.5 Validation of glomerular ECM proteins regulated by genetic background or sex...... 154
Figure 4.6 Structural abnormalities in glomerular ECM in disease-susceptible FVB mice...... 155
Figure 4.7 GO enrichment map and pathway analysis...... 157
Figure 4.8 Workflow of glomerular ECM enrichment...... 172
Figure 4.9 Unconnected glomerular ECM proteins with enrichment due to strain or sex...... 173
Figure 4.10 Known GBM components do not change in a robust strain- or sex-dependant manner
...... 174
Figure 4.11 GO enrichment analysis of proteomic data-set ...... 175
Figure 5.1 Podocyte attachment to ECM ligands ...... 184
Figure 5.2 Laminin and collagen IV adhesion influence podocyte morphology ...... 186
Figure 5.3 Collagen IV and laminin influence podocyte adhesion complex size and number ...... 188
Figure 5.4 Isolation of adhesion complexes ...... 190
Figure 5.5 Podocyte adhesion complex composition ...... 196
Figure 5.6 GO enrichment analysis ...... 200
Figure 5.7 Podocyte barrier function assessed by electrical cell-substrate impedance sensing (ECIS)
...... 202
Figure 6.1 In silico analysis of the nephrin interactome ...... 216
Figure 6.2 Generation of Nephrin-FLAG and N ∆CT-FLAG cell lines ...... 221
Figure 6.3 Development of nephrin complex isolation assay ...... 224
Figure 6.4 A nephrin signalling complex identified by mass spectrometry ...... 225
Figure 6.5 Nephrin integrin β1 merged interactome ...... 230 9
List of tables
Table 2.1 Human glomerular ECM proteome ...... 82
Table 3.1 Glomerular endothelial cell ECM proteome ...... 121
Table 3.2 Podocyte ECM proteome ...... 126
Table 3.3 Coculture ECM proteome ...... 132
Table 4.1 Mouse glomerular ECM proteome ...... 168
Table 5.1 Adhesome proteins identified by MS ...... 192
Table 5.2 Non-adhesome proteins identified by MS ...... 197
Table 6.1 The nephrin interactome ...... 217
Table 6.2 Novel nephrin signalling proteins ...... 226
Table 8.1 List of antibodies ...... 251
Table 8.2 Chapter specific methods ...... 254 10
Abstract
University of Manchester
Name: Michael Randles
Degree title: Doctor of Philosophy
Thesis title: Proteomic analyses of kidney glomerular extracellular matrix in health and disease
Date: 2014
Glomerular filtration is a vital physiological process removing waste products from the circulation and this process occurs across the glomerular filtration barrier (GFB). The cells and extracellular matrix (ECM), which form this barrier, are exposed to forces during ultrafiltration and special adaptation is required to withstand these forces. Dysfunction in cellular adhesion machinery or ECM assembly within the GFB causes loss of selective glomerular filtration, however, the mechanisms governing these processes are poorly understood. To this end we sought to characterise the glomerular ECM and adhesion machinery using high throughput mass spectrometry (MS)-based proteomics.
MS of human glomerular ECM identified a highly complex extracellular niche, revealing the potential involvement of novel ECM proteins in glomerular development and disease processes. Furthermore we identified that glomerular cells in culture had distinct ECM proteomes and interestingly, coculture experiments demonstrated that the ECM proteome was influenced by cellular crosstalk and had a closer resemblance to glomerular ECM in vivo . Protein network analyses of in vivo and in vitro ECM datasets revealed a common core of highly connected structural ECM proteins that may be important for glomerular ECM assembly. To understand how this ECM proteome altered in disease, we studied mice with mild glomerular dysfunction. Here, transcriptomic and proteomic analyses identified alterations in ECM composition and 3D electron microscopy revealed striking ultrastructural changes in glomerular ECM.
MS-based proteomics was next applied to the analysis of glomerular podocyte adhesion complexes, leading to the discovery that the actin cytoskeletal regulators and trafficking machinery are recruited to adhesions sites in an ECM-ligand dependent manner. Furthermore, these differences functionally altered cell shape and adhesion strength. These same analyses were applied to podocyte cell-cell junctions, revealing an unexpected overlap of cell-ECM and cell-cell adhesion machinery.
Overall, these findings demonstrate for the first time the complexity of the glomerular ECM and adhesion signalling complexes and reinforce the benefits of global, unbiased experimental approaches. In addition the results suggest that glomerular ECM composition, organisation and adhesion signalling are context dependent, and therefore, represent potential therapeutic targets. 11
Declaration
No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning .
Copyright statement
The following four notes on copyright and the ownership of intellectual property rights must be included as written below: i. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and s/he has given The University of
Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the
University has from time to time. This page must form part of any such copies made. iii. The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy
(see http://www.campus.manchester.ac.uk/medialibrary/policies/intellectual-property.pdf ), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s 12
regulations (see http://www.manchester.ac.uk/library/aboutus/regulations ) and in The University’s policy on presentation of Theses . Acknowledgments
Foremost, I would like to thank my supervisors, Rachel Lennon and Adrian Woolf, for giving me the fantastic opportunity to work on this project. Rachel has an infectious enthusiasm which has helped me enormously throughout my PhD. Furthermore, Rachel has given me numerous opportunities to present my work at national and international conferences, which has been integral to my development. Adrian has had a key contribution to my development as a scientist, both in terms of planning experiments and in communicating ideas. Both Rachel and Adrian have been instrumental in directing this project and giving me good advice throughout my PhD.
Next, I would like to thank Hellyeh Hamidi, who taught me a great deal. Martin Humphries and Patrick
Caswell, who have also given me fantastic advice throughout my PhD. Adam Byron and Jonathan
Humphries (funniest man alive) who were both instrumental in the developing the mass spectrometry approaches that I have used over the course of my PhD. I would also like to thank David Long for his collaborations and for accommodating me in his lab.
I would like to thank Maryline Fresquet for teaching me the art of protein purification. All the members of the Lennon Lab for fantastic input during data meetings: James McCaffrey, Salman Hosawi and
Stephanie Murphy. All members of the Caswell and Humphries lab old and new: Mr Morgan, Guilluame
Jacquemet, Janet Askari, Matthew Jones, Sue Craig, Ewa Koper, Nikki Howay Paul, Joe Robertson,
Edward Horton, Ruth McDowall, Eve Blumson, Angelique Stuani, Dave The Green Machine, Rebecca
Bridgewater, Joe Hetmanski and David Bradley. I would also like to thank Kids Kidney research for funding. I would like to thank Scott Cowan and co for keeping me sane during writing through the medium of football.
I would like to thank my family: Christopher, Shelley, Ethan, Nan, Grandad, Dad and most of all my
Mum who have always supported me.
.
Finally, I would like to thank my wonderful girlfriend Liz Scott. 13
14
Rationale for an alternative thesis format
There are several reasons for presenting the thesis in an alternative format. Firstly, the alternative format is good practice to prepare data into paper format, which is a vital skill for early career researchers. Secondly, because data is organised as papers, work can be published more quickly after the completion of my PhD. Finally, my project covers distinct areas of research that are linked by the approach used (i.e using proteomics to study extracellular matrix, cell-extracellular matrix adhesion and cell-cell adhesion), therefore dividing up each area into a separate papers prepares the reader for these different areas of research. 15
Key Abbreviations
ACE Angiotensin converting enzyme ARB Angiotensin receptor blocker B6 C57BL/ 6JOlaHsd BM Basement membrane CRAPome Contaminant Repository for Affinity Purification ECIS Electrical cell-substrate impedance sensing ECM Extracellular Matrix EM Electron microscopy FAK Focal adhesion kinase FGF2 Fibroblast growth factor 2 FSGS Focal segmental glomerulosclerosis FVB FVB/NHanHsd GAG Glycosaminoglycan GAP GTPase activating protein GBM Glomerular basement membrane GDI Guanine nucleotide dissociation inhibitor GEC Glomerular Endothelial cell GEF Guanine-nucleotide exchange factor GEnC Glomerular endothelial cell GFB Glomerular filtration barrier GFR Glomerular filtration rate GO Gene Ontology HPA Human Protein Atlas HSPG Heparan sulphate proteoglycan ILK Integrin linked kinase MMP Matrix metalloproteinase MS Mass spectrometry PAN Puromycin aminonucleoside PCA Principal component analysis PEC Parietal epithelial cell PRIDE PRoteomics IDEntifications SBF-SEM Serial block face scanning electron microscopy TEM Transmission electron microscopy
16
1 General introduction
1.1 Overview of the thesis
The human glomerular capillaries are fascinating; within the kidney, glomeruli perform an essential filtration process, removing waste products from the circulation, whilst retaining essential plasma components within the circulation. As a result, glomeruli have adapted unique cellular and extracellular matrix (ECM) structures. There is an enormous interest in understanding the biology of the glomerulus, as loss of glomerular function has become a major worldwide healthcare problem, for which there are currently no targeted therapies. Glomerular filtration occurs across a composite three layered capillary wall, termed the glomerular filtration barrier (GFB). Within this structure, adhesion of the two cellular layers to the extracellular niche is crucial; defects in the glomerular adhesion machinery, or ECM, results in loss of glomerular function. The ECM and mechanisms of adhesion within the glomerulus have been studied using candidate based approaches. However, holistic understanding of ECM and adhesion signalling mechanisms within the glomerulus is lacking. In order to address this, global unbiased approaches, such as proteomics, have enormous potential.
The specific proteome of a tissue, cell type, organelle or ECM is a major determinant of its function.
Consequently, mass spectrometry (MS)-based proteomics can generate large data-sets that contain important protein level information about the cell type or organelle studied. Extracting this information from these large data-sets poses a major challenge. Applying bioinformatic approaches, such as hierarchical clustering and protein-protein interaction network analysis can provide holistic understanding of the biological system studied. Furthermore, these analyses can direct hypothesis generation and identify key proteins or signalling networks, and as a result reveal unexpected biology.
Excitingly, approaches to isolate ECMs from tissues and cells 1-3 and adhesion complexes from cells 4-6 are rapidly improving, enabling targeted proteomic studies of ECM and adhesion. The primary focus of this thesis is the use of proteomic approaches to study ECM and adhesion signalling within the glomerulus, with the aim of providing a more comprehensive understanding of glomerular ECM and adhesion signalling mechanisms. 17
1.2 The kidney
Human kidneys have essential roles in homeostasis, including the removal of metabolic waste products from the circulation, the control of intravascular volume and the regulation of extracellular pH.
Additionally, the kidneys secrete the enzyme renin and a number of hormones including erythropoietin and calcitriol. Renin is involved in the renin-angiotensin-aldosterone signalling axis, which controls blood pressure. Erythropoietin is essential for the differentiation of hematopoietic stem cells down the erythroid linage and calcitriol is involved in reabsorption of calcium and bone metabolism. Loss of kidney function with disease compromises these vital biological processes. Chronic kidney disease is a significant worldwide healthcare problem, costing the National Health Service an estimated £1.4 billion per year in the United Kingdom, which is more than lung, skin, colon and breast cancer combined. 7
The individual functional units of the kidney are the nephrons. Nephrons are intricate structures composed of: a capillary network, called a glomerulus; Bowman’s capsule, which encapsulates the glomerular tuft; and a tubular network, where urine is modified and concentrated (Figure 1.1). A healthy adult human kidney contains between 1 million and 1.5 million nephrons. 8 Blood enters the kidney via the paired renal arteries, which branch and supply the glomerular capillaries via the afferent arterioles.
In the glomerular capillaries the blood is filtered, producing an ultrafiltrate that enters Bowman’s space.
Filtered blood exits the glomerulus via the efferent arteriole and the kidney via the renal vein. Under normal physiological conditions the electrolyte composition of the ultrafiltrate is similar to plasma, however, large macromolecules and proteins are absent. This is because the glomerular capillary wall is freely permeable to water and small solutes, but impermeable to larger molecules, such as albumin.
In a 70 kg adult human, glomerular filtration produces approximately 180 litres of ultrafiltrate every day.
In the proximal tubule most of the water, glucose, salts and amino acids are reabsorbed from the ultrafiltrate into the peritubular capillaries, whilst some toxins are actively secreted into the ultrafiltrate.
Ultimately, the human kidneys produce between 1 and 1.5 litres of urine each day.
18
Figure 1.1 Overview of the kidney and the nephron The renal artery supplies unfiltered blood to the kidney. This blood enters the glomeruli via afferent arterioles where it is filtered producing an ultrafiltrate or 'primary urine' which enters Bowman's space. Filtered blood exits the glomerulus via the efferent arteriole and kidney via the renal vein. The primary urine produced by ultrafiltration is processed by the proximal tubule where most of the water and salts are reabsorbed. In the loop of Henle the urine is concentrated by a countercurrent multiplier system. The urine then passes through the distal tubule which links to the collecting duct system, here K+, Na+, Ca++ and pH undergo additional regulation. Finally, the urine is collected and is propelled out of the kidney into the urinary bladder via the ureter. 19
1.3 Glomerular filtration
Glomerular filtration is the process of producing virtually protein free ultrafiltrate, which is further processed in the tubules to produce urine. Loss of glomerular filtration is common to a range of kidney diseases. Deciphering the mechanism of normal glomerular filtration is prerequisite to understanding the pathogenesis of diseases that affect the glomerulus. The process of glomerular filtration requires the generation of forces within the glomerulus, which the cells and ECM are specially adapted to cope with. As a result, adhesion and ECM within the glomerulus are important topics of research.
1.3.1 The glomerular filtration barrier
Glomerular filtration occurs across the glomerular capillary wall, also known as the GFB, which is the highly specialised interface between the vasculature and the urinary space. This barrier is composed of three major components: the glomerular endothelium, the glomerular basement membrane (GBM) and the glomerular visceral epithelial cells (podocytes) (Figure 1.2). Glomerular filtration is determined by the following parameters: the permeability of the glomerular filtration surface (K), the surface area available for filtration (S), the hydrostatic pressure in the glomerular capillaries (P GC ), the hydrostatic pressure in Bowman’s capsule (P T), the colloid oncotic pressure in the glomerular capillaries ( ∏GC ) and the colloid oncotic pressure in Bowman’s space ( ∏T). The glomerular filtration rate (GFR) can be defined as:
9 GFR = K S [(P GC – P T) – ( ∏GC -∏T)].
In humans, the normal rate of glomerular filtration is > 90 mL/minute. Blood enters and exits the glomerulus via arterioles; therefore, the pressure in the glomerular capillaries is higher than in typical capillaries. Normally, renal autoregulation prevents systemic increases in blood pressure from perturbing GFR. Renal autoregulation is intrinsic to the kidney, leading to vasoconstriction and vasodilatation of the arterioles to maintain GFR. 10, 11 However, a decrease in the surface area available for glomerular filtration, for example with reduced numbers of nephrons or sclerosis of glomeruli, causes a reduction in GFR. In addition, perturbation of the permeability properties of the GFB leads to leakage of plasma proteins, such as albumin, into the urine referred to as proteinuria or albuminuria respectively. 20
The GFB performs a dual role whereby it not only filters blood to remove metabolic waste products from the circulation, but also retains large molecules within the circulation. Water, small solutes and proteins with molecular weights < 40 kDa pass freely into the ultrafiltrate; in fact the permeability of the
GFB to water is orders of magnitude higher than in other capillary beds. 12 In contrast, macromolecules with molecular weights > 40 kDa are highly restricted, for example the albumin (molecular weight 69 kDa) concentration in plasma is 40, 000 mg/L, but the albumin concentration in urine is only 4 mg/L. 13
Reduction in GFR and an increase in proteinuria are risk factors for both renal and cardiovascular disease. 14 It is therefore critical to build an understanding of the mechanisms of glomerular filtration and GFB permeability properties, in order to appreciate the causes of loss of GFB functionality.
21
Figure 1.2 Schematic cross section of the glomerulus and the glomerular filtration barrier The glomerular capillaries are lined on the inside with fenestrated endothelial cells. Attached to the outside of the glomerular capillaries, with their large cell bodies inside Bowman's space , are the glomerular podocytes. Podocyte foot processes interdigitate to form a highly ordered structure, with specialised cell junctions called slit diaphragms in-between. Both glomerular endothelial cells (GECs) and podocytes attach to the glomerular basement membrane (GBM), a thick extracellular matrix (ECM) that is sandwiched in-between the two cellular layers . Mesangial cells and associated ECM hold capillary loops together by interacting with the GBM and with the glomerular endothelium. Parietal epithelial cells (PECs) and their associated basement membrane form Bowman's capsule. The glomerular filtration barrier (GFB) comprises three major layers: the glomerular endothelium (and glycocalyx), the GBM and podocyte foot processes.
22
1.3.2 Size and charge selective glomerular filtration
Classically, the mechanism of selective glomerular filtration is thought to be due to the size and charge of filtered molecules. Size selectivity of the barrier is well established; for neutral solutes, glomerular sieving coefficients become smaller with increasing molecular size. 13, 15-17 Electron microscopy of the glomerular podocyte led to the discovery of the podocyte slit diaphragm, an electron dense cell-cell junction. 18, 19 Researchers then hypothesised that podocyte slit diaphragms could act as the size excluding layer of the GFB. Consistent with this hypothesis, the cross-sectional dimension of slit diaphragm pores is approximately the size of an albumin molecule. 18, 19 Furthermore, mutations in podocyte slit diaphragm genes cause proteinuria. 20-22 However, proteinuria can occur in the absence of podocyte morphological changes. 23-25 Indeed, proteinuria can occur regardless of which layer of the
GFB is damaged. 13, 26
Selectivity of the GFB based on charge is supported by tracer studies, which demonstrate that negatively charged molecules are impeded more than neutral molecules when crossing the GFB. 15, 27
However, the exact location of the putative charge barrier is controversial; the GBM, 28, 29 endothelial glycocalyx 30, 31 and the podocyte coating protein podocalyxin 32 have all been proposed as charge barriers. Interestingly, when renal plasma flow is stopped, albumin can rapidly pass into the Bowman's space by diffusion,33 which suggests that flow is required for permselectivity. Another interesting property of the GFB is that despite exquisite permselectivity, preventing the passage of an estimated
250,000 kg of plasma proteins in a human lifetime, 34 the GFB does not clog. Several anti-clogging mechanisms have been proposed including: uptake of plasma proteins by mesangial cells 35 or podocytes, 36 charge repulsion by the GBM, 37 tangential-flow filtration 38 and the gel permeation hypothesis.39, 40
1.3.3 The albumin retrieval hypothesis
Some researchers have proposed that the GFB is more permeable to albumin than is generally accepted and that albumin is in fact reabsorbed in the proximal tubule. This hypothesis is supported by measurements of albumin filtration taken using 2-photon microscopy, which suggest that non- proteinuric rats filter much larger amounts of albumin into Bowman's space than recorded by 23
micropuncture experiments. 41 If this were true , albuminuria would have to be reinterpreted as a marker of tubular dysfunction rather than glomerular dysfunction. However, subsequent studies using 2-photon microscopy found that the GFB was indeed highly impermeable to albumin. 42, 43 In addition, only small
amounts of albumin are detected in the urine when tubular protein adsorption is inactivated ,44, 45
whereas when glomerular structure is compromised large amounts of albumin escape into the urine. 20-
22 23-25
1.3.4 The electrokinetic model
An attractive new model of glomerular filtration, the electrokinetic model, has recently been proposed. 34
This hypothesis is supported by the measurement of an electric field of –0.045 mV per 10 cm H 2O across the GFB of the common mudpuppy. 46 This electric field is thought to be a 'streaming potential' generated by the forced filtration of ionic solutes [Na +, K +, Ca 2+ , Cl – and HCO3–] across the GFB.
Streaming potentials are phenomena produced by the tangential motion of a fluid containing small ions along the charged surface of a filter. The glomerular endothelium coat, or glycocalyx, represents a negatively charged surface, as does the GBM, and both of these structures could be important for the induction of streaming potentials across the GFB. This potential difference may cause electrophoresis of negatively charged molecules, forcing them towards the capillary lumen. This offers an anti-clogging mechanism in the form of active electrophoresis and an explanation for the loss of selectivity when plasma flow is absent.
Although, further studies are required to refine and unify the proposed mechanisms of selective glomerular filtration, numerous studies have now established that all three layers of the GFB are critical for permselectivity and that no one layer is the definitive 'glomerular filter'. It is therefore crucial to study all three layers of the GFB to understand the pathologies associated with proteinuria. 24
1.4 Microalbuminuria
The GFB must remain almost completely impermeable to plasma proteins throughout a human lifetime.
Leakage of plasma proteins into the urine is referred to as proteinuria. In humans this specifically means the escape of > 300 mg/day of plasma proteins into the urine. Proteinuria is a clinical manifestation of glomerular disease, in addition to haematuria and reduced GFR. The combination of these clinical features specifies a variety of clinical syndromes; nephritic syndrome, nephrotic syndrome and chronic renal failure. Nephritic syndrome is characterised by proteinuria, haematuria, oedema and hypertension. Nephrotic syndrome, however, is characterised by severe proteinuria > 3.5 g/day in adults (detectable via dipstick urinalysis), hypoalbuminemia, peripheral oedema and hypercholesterolaemia.47
Some small loss of plasma proteins, for example albumin, does occur in healthy individuals. Typically,
< 30 mg/day of albumin escapes into the urine of a healthy individual. However, this figure is variable, and elevated excretion of albumin between 30 - 300 mg/day, is referred to as microalbuminuria. In individuals with diabetes mellitus microalbuminuria precedes, and may also predict, progression to diabetic nephropathy. 48, 49 In addition, microalbuminuria is an independent risk factor for cardiovascular disease. 14
A series of observations support genetic background as a determinant of albumin excretion. Firstly, in the United States of America, non-Hispanic blacks and Mexican-Americans have a greater incidence of microalbuminuria compared with whites. 50 Secondly, normotensive black individuals excrete lager amounts of albumin compared with white individuals.51 Moreover, these increases in albumin excretion are associated with an increased risk of end-stage renal disease in black individuals compared with
Hispanic and white individuals. 52-55 In addition to racial background, sex also influences the degree of albumin excretion. In both the Netherlands and the United Kingdom it has been reported that men have a higher albumin excretion rate than women. 22, 23 These data suggest that the composition of the GFB may be dependent on racial background and sex. If this is the case, understanding these intrinsic 25
differences may provide insights into the determinants of GFB permeability and facilitate more targeted personalised treatment of glomerular disease. 1.5 Cellular components of the glomerulus
Classically, the glomerulus is thought to contain four major cell types: glomerular endothelial cells
(GECs), podocytes, mesangial cells and parietal epithelial cells (PECs). All of these cells produce, organise and adhere to ECM to form the unique structure of the glomerulus (Figure 1.2). Tight regulation of the ECM and adhesion by glomerular cells is required for normal glomerular function.
1.5.1 The glomerular endothelium
The glomerular endothelium is the first barrier to the passage of fluid and macromolecules during glomerular filtration. It is a continuous endothelium within which each GEC contacts adjacent GECs via cell-cell junctions. The glomerular endothelium contains transcellular cytoplasmic holes, called fenestrae, that are 70-100 nm in diameter and do not possess diaphragms. 56 Fenestrae make up 20-
50% of the glomerular endothelial surface (Figure 1.2), 57, 58 an adaptation that enables high water permeability, which is essential for glomerular filtration. 59 The diameter of these fenestrae is much greater than the diameter of albumin, which suggests that fenestrae have a limited role in the size restriction of albumin.
The glomerular endothelium has an apical cell surface glycocalyx comprising transmembrane proteoglycans, glycoproteins, sialic acids, 60 hyaluronan and adsorbed plasma proteins (Figure 1.2). 26 ,
30, 60, 61 This extracellular structure is a 200 nm thick meshwork that confers negative charge to the glomerular endothelium and has been shown to contribute to the selective passage of albumin across the GFB. 60, 62 This is consistent with either a charge selective role, or in generating streaming potentials as suggested by the electrokinetic model. 30, 60 The endothelial glycocalyx could also be important in glomerular disease; the glycocalyx volume is reduced in individuals with type 1 and type 2 diabetes mellitus, suggesting a possible role for the glycocalyx in the glomerular complications of these diseases. 63-65
26
1.5.2 Podocytes
The second cellular layer of the GFB is the podocyte. Podocytes are terminally differentiated visceral epithelial cells that reside within the urinary space. They have large cell bodies that are detached from the underlying GBM; instead, podocyte adhesion to the GBM is mediated by long microtubule based primary foot processes and actin rich secondary foot processes (Figure 1.2). Podocyte foot processes interlink and form specialised cell-cell junctions called slit diaphragms. These junctions have a width of
30-45 nm. 18 In 1998 the discovery of nephrin led to major interest in the biology of the glomerular podocyte. NPHS1 , the gene encoding nephrin, is mutated in a severe form of congenital nephrotic syndrome. 66 Nephrin is a 180 kDa transmembrane immunoglobulin superfamily member, which is localised to podocyte slit diaphragms. 67-69 Further studies have led to the discovery that numerous genes encoding slit diaphragm proteins are mutated in genetic forms of nephrotic syndrome (Figure
1.3).
Loss of slit diaphragms and podocyte foot process architecture are major histological features of glomerular diseases. Indeed, minimal change disease is associated with nephrotic syndrome, but a normal appearance of glomeruli at light microscopy. However, with electron microscopy, loss of podocyte slit diaphragms and foot processes can be appreciated (Figure 1.4). Loss of podocyte foot processes may in fact be a mechanism to protect the glomerulus from the loss of podocytes.
Podocytes are unique in that their cell bodies are detached from the underlying GBM, only their foot processes mediate adhesion. When podocytes are stressed they may be stimulated to adhere more tightly to the GBM to prevent detachment. As a consequence, specialised podocyte morphology and function is lost. A plethora of evidence supports podocyte foot processes and slit diaphragms as essential structures for selective glomerular filtration. Therefore, the GBM and podocyte adhesion to the GBM are clearly important areas of research. 27
Figure 1.3 Podocyte slit diaphragm and adhesion genes mutated in nephrotic syndrome List of podocyte genes (in bold) that cause glomerular abnormalities when mutated in humans and/or mice. Genes are sorted into slit diaphragm, ECM and adhesion or actin regulators. Gene names represented as proteins in schematic. Slit diaphragm: NPHS1, nephrin; NPHS2, podocin; TRPC6, short transient receptor potential channel 6; CD2AP, CD2-associated protein; PLCE1, 1-phosphatidylinositol 4,5-bisphosphate phosphodiesterase ε1. Actin regulators: MYH9, myosin heavy chain 9; MYO1E, unconventional myosin Ie; ARHGDIA, Rho GDP-dissociation inhibitor 1; ARHGAP24, Rho-GTPase- activating protein 24; ACTN4, α-actinin-4; INF2, inverted formin-2. ECM and adhesion: ILK, integrin linked kinase; TLN1, talin1; ITGA3, integrin α3; COL4A3, COL4A4, COL4A5, type IV collagen a3, a4 and a5 respectively; LAMB2, laminin β2.
28
1.5.3 Mesangial cells
In addition to the glomerular cells that form the GFB, there are two other types of resident glomerular cells. Mesangial cells are structural pericyte-like cells, which help maintain the integrity of the GFB.
They sit between glomerular capillary loops directly contacting GECs (Figure 1.2). Their overall contribution to permselectivity is unknown, but thought to be small. 13 Although GFR is controlled predominately at the level of the afferent and efferent arterioles, mesangial cells are involved in fine tuning GFR through contraction and growth factor secretion. 70-73
Mesangial cells are the target for the immune mediated disease IgA nephropathy. IgA nephropathy is caused by the deposition of aberrantly glycosylated IgA in the mesangium. 74, 75 As a result, mesangial cells are activated to proliferate and secrete increased amounts of ECM, causing proteinuria.
Proteinuria is usually associated with dysfunction in GFB components; therefore, proteinuria in IgA nephropathy may provide evidence for crosstalk between the mesangium and GFB components. 76 In a number of pathologies, including diabetic nephropathy, proliferation and expansion of mesangial cells and associated ECM contribute to glomerulosclerosis and reduce GFR by decreasing the surface area available for glomerular filtration (Figure 1.4). 77
1.5.4 Parietal epithelial cells
PECs are resident glomerular cells which together with their basement membrane form Bowman's capsule (Figure 1.2). 78 Podocytes and PECs arise from the same progenitor metanephric mesenchyme population during development. 79 Eventually, PEC and podocyte phenotypes diverge and they begin to express different marker proteins; podocytes express nephrin 67 and WT1 80 , and PECs express Pax-281,
82 and Claudin-1. 83, 84 No glomerular disease has thus far been found to be a PEC specific disease. 85
However, in some glomerular diseases, PECs can become injured or activated causing them to migrate more and produce more ECM (Figure 1.4). PEC activation can lead to a lesion where the glomerular tuft fuses to Bowman’s capsule. In this situation PECs attach to GBM when left bare, for example by podocyte detachment, which could be a common starting point of glomerulosclerosis in a wide range of glomerulopathies. 86, 87 29
Podocytes do not proliferate under normal conditions or in glomerular diseases where podocytes are lost from the glomerular tuft. 88 However, under certain circumstances podocytes can be regenerated, even in the absence of podocyte proliferation, which suggests the existence of a reservoir of podocyte progenitors. 89 PECs have been touted as possible progenitor cells, in addition to cells of the renin linage, 90 because PECs express progenitor cell markers such as a glycosylated isoform of CD133 and
CD24. 91 Moreover, cells lining Bowman's basement membrane can express podocyte markers or show podocyte ultrastructure in normal human glomeruli, these cells are referred to as parietal podocytes 92,
93 or ectopic podocytes. 94 In some cases these cells coexpress markers of differentiated podocytes and
PECs such as WT1, nephrin, Pax-2 and Claudin-1. 95 However, evidence from experimental animal models suggests that these cells are podocytes that have migrated away from the glomerular tuft onto
Bowman's basement membrane rather than PECs differentiating into podocytes. 95-97 30
Figure 1.4 Common histological changes which occur in glomerular disease Histological changes that are common to a range of glomerular diseases include: thickening of the GBM, mesangial proliferation and ECM expansion, podocyte foot process effacement, detachment of podocytes, sclerosis and overall glomerular hypotrophy. In combination these histological changes can result in a complete loss of function within the glomerulus. ECM, extracellular matrix; GBM, glomerular basement membrane; PEC, parietal epithelial cell. 31
1.6 Glomerular extracellular matrix
The ECM is essential for multicellular life providing a structural scaffold with appropriate mechanical properties to support adjacent cells. 98 It comprises a complex network of glycosaminoglycans and fibrous proteins, which are synthesised and secreted by cells. Podocytes and GECs adhere to the
GBM via cell surface receptors. This cell-ECM interface forms a signalling platform that controls all aspects of cell fate decisions including shape, growth, differentiation and survival. 98, 99 In addition to this signalling platform, the ECM modulates cell-cell signalling by sequestering secreted growth factors and cytokines, forming reservoirs for appropriate spatiotemporal release. 98 Thus the composition of ECM within the glomerulus is fundamental, and studying the glomerular ECM is required for a comprehensive understanding of glomerular biology.
1.6.1 The glomerular basement membrane
Basement membranes are thin sheets of ECM with a supramolecular assembly structured around laminin and collagen IV networks. In the glomerulus ECM is organised as the GBM of the capillary walls, the basement membrane of Bowman’s capsule and the loosely associated mesangial ECM between capillary loops (Figure 1.2). The mature GBM is thicker than most basement membranes
(300-350 nm in humans) and is formed by the fusion of two distinct basement membranes, one produced by podocytes the other by GECs .100 The study of human glomerular disease led to the discovery of tissue-restricted isoforms of laminin and collagen IV in the mature GBM , and these are key components of this specialised ECM .101, 102 In a wide range of glomerulopathies the GBM becomes thickened and irregular (Figure 1.4); these changes in the GBM remain poorly understood.
1.6.1.1 Laminins
Laminins are self-assembling heterotrimeric glycoproteins ,103, 104 which are absolutely required for basement membrane formation. 105, 106 All laminin heterotrimers contain one α, one β and one γ chain.
Trimeric laminin has a cruciform shape with one long arm and three short arms (Figure 1.5). The short 32
arms contain the N-terminal (LN) domains of the laminin heterotrimer and the long arms contain the C- terminal globular laminin G-like (LG) domains (Figure 1.5) .107-111 The short arms form the nodes within the laminin network via a three-arm interaction mode; interactions between the N-terminal regions of one α, one β and one γ chain are required for formation of ternary nodes in the network (Figure 1.5) .112-
114 The cell surface receptor binding sites within laminin trimers are mostly localised within the long arm
LG domains, especially LG1-5 of the α chain .107-111
Laminin isoforms within the GBM undergo developmental transitions. The primitive GBM contains laminin α5β1γ1 (laminin-511), the mature GBM however, comprises predominately laminin-521. 101 A complete laminin-521 network is required for a functional GBM; mutations in LAMB2 , the gene encoding the laminin β2 chain, cause Pierson syndrome in humans. Affected individuals have a spectrum of pathology with genotype-phenotype correlation. Truncating mutations cause congenital nephrotic syndrome, microcoria, muscular hypotonia and neurodevelopmental deficit. 115, 116 Lamb2 mutations in mice are also associated with glomerular dysfunction. Mice with null mutations die after 3 weeks of age with severe proteinuria and neuromuscular defects. 117 Interestingly, these mice develop proteinuria before podocyte morphological defects occur, which supports a role of the GBM in restricting macromolecules directly, rather than via modulating cellular phenotype. Over time however, podocyte morphological changes do occur and proteinuria becomes more severe. Furthermore, these animals have accumulation of ectopic laminin chains in the GBM, including α1, α2, α3, β1, β3, and γ2, however, these chains do not compensate for the loss of the β2 chain, possibly due to low expression or the absence of a complete laminin network. 25 The theory that insufficient expression of laminin chains accounts for the observed lack of compensation is supported by the finding that podocyte overexpression of Lamb1 in Lamb2 null mice ameliorates proteinuria. 118
1.6.1.2 Collagen IV
Unlike the laminin network, the collagen IV network is dispensable for basement membrane formation , however , the basement membranes formed lack strength and stability. 119 Collagen IV forms heterotrimers composed of three alpha chain combinations ( α1α1α2, α3α4α5 or α5α5α6). Each alpha chain contains three distinct domains: an N-terminal 7S domain rich in cysteines and lysines, which enables tetramer formation and inter-chain cross -linking; a long collagenous repeat domain, around 33
1400 amino acids in length; and a carboxy terminal non-collagenous domain (NC1), which forms dimers (Figure 1.5) .120 In combination, these interactions lead to the formation of a ‘chicken wire’ meshwork of collagen IV.
A novel chemical bond, not previously identified in biomolecules, the sulfilimine bond (-S=N-), was recently discovered in collagen IV. 121 This bond cross -links lysine/hydroxylysine-211 and methionine-
93 of adjoining protomers in the NC1 domains of both collagen IV α1α1α2 and α3α4α5. This bond may provide additional resistance of the network to mechanical strain. Furthermore , peroxidasin, an enzyme found in basement membranes, catalyses the formation of the sulfilimine bond .122 Ionic bromide is a cofactor required for peroxidasin-catalysed formation of the sulfilimine cross -links in collagen IV networks, 123 which is the first known essential function for ionic bromide in animals.
As with laminin a developmental transition in collagen IV isoforms is critical for GBM maturation. The primitive GBM contains predominantly the ubiquitous collagen IV α1α1α2, whereas the mature GBM contains collagen IV α3α4α5. 102 Mutations leading to a reduction or absence of the α3α4α5 networks cause human Alport syndrome, characterised by a renal phenotype of haematuria, proteinuria and progressive renal failure. 120, 124 The GBM in Alport syndrome has increased collagen IV α1α1α2, but this isoform is unable to compensate for the lack of the α3α4α5 network. As a consequence the GBM develops splits and a typical basket-weave appearance, leading to speculation that mechanical strain cannot be tolerated by the GBM in the absence of collagen IV α3α4α5. Potentially, the reduced numbers of disulphide bonds in the α1α1α2 relative to α3α4α5 network could explain their differing capacities in coping with mechanical strain. This concept is further supported by the observation that reducing mechanical strain in the glomerulus with angiotensin-converting-enzyme (ACE) inhibitors, which lower blood pressure as well as transcapillary filtration pressure, significantly delays disease progression in Alport syndrome. 125-127 34
Figure 1.5 Glomerular basement membrane components Panels display schematics of: laminin heterotrimer, laminin network, collagen IV heterotrimer, collagen IV network and the basic assembly of the glomerular basement membrane (GBM). The laminin-521 network in the mature GBM is located close to the podocyte and glomerular endothelial cell (GEC) surfaces. The collagen IV α3α4α5 network is located deeper within the GBM. Agrin is located close to the podocyte and GEC surfaces and interacts with cell surface receptors and the laminin network. Nidogen and perlecan likely occupy the region between the laminin and collagen IV and act to crosslink these two networks.
35
1.6.1.3 Additional glomerular basement membrane components
Evidence suggests that the laminin and collagen IV networks are indirectly linked via nidogens 128 and the heparan sulphate proteoglycans, perlecan, 129, 130 and agrin (Figure 1.5) .131 Podocyte specific deletion of agrin from the GBM results in a significant reduction in the negative charge associated with the GBM, however , alone or combined with knockout of perlecan, agrin deletion does not cause proteinuria. The lack of proteinuria in this context suggests the GBM has a limited role in charge selection during glomerular filtration .132 Nidogen 1 and 2 are dumbbell shaped proteins that reportedly bind to both laminin and collagen IV. Mice with knockout of nidogen 1 or 2 are viable and have normal basement membranes. Deletion of both isoforms, however, causes perinatal lethality. 133 This is consistent with a degree of redundancy in their ability to bind collagen IV and laminin. Surprisingly, the
GBM has a normal appearance even in the double ( Nid-1, Nid-2) knockout. This suggests that nidogen is dispensable for the formation of the GBM, but it may be required for the GBM to resist mechanical strain.
1.6.1.4 Orientation of proteins in the glomerular basement membrane
To elucidate the position of ECM proteins within the GBM a systematic analysis of ECM components within basement membranes, with respect to each other and to their receptors, was performed in vivo using super resolution microscopy .134 This investigation found two separate laminin networks, one adjacent to the podocytes the other adjacent to the glomerular endothelium (Figure 1.5). The collagen
IV α3α4α5 network was distributed along the centre of the GBM alongside nidogen, consistent with the putative cross -linking function of nidogens (Figure 1.5). The human GBM is approximately two-fold thicker than the mouse GBM and interestingly this study found increased thickness of the human collagen IV α3α4α5 network, and potentially, an additional layer of laminin-521 closer to the centre of the GBM. 134 36
1.6.2 The mesangial extracellular matrix and Bowman's capsule
In addition to the GBM, mesangial cells produce ECM that holds capillary loops together and PECs synthesise the basement membrane that encapsulates the glomerular tuft. The ECM produced by mesangial cells is composed of mostly basement membrane components, but is not organised into a basement membrane structure. Compositionally, mesangial ECM contains predominately collagen IV
α1α1α2 rather than collagen IV α3α4α5. As a result these collagen chains can be used to distinguish the GBM and mesangial ECM compartments. In addition, the mesangial ECM contains laminins, fibronectin, nidogen, perlecan and decorin. 135 Mesangial cells also interact with the GBM; interactions between mesangial cells and laminin α5LG3-5 in the GBM are required for organisation of glomerular capillary loops. 136
PECs produce the Bowman's basement membrane, which is multilayered and contains collagen IV
α5α5α6, in addition to the ubiquitously expressed collagen IV α1α1α2. 137 The basement membrane of
Bowman's capsule has a defective appearance in hereditary angiopathy with nephropathy, aneurysms, and muscle cramps (HANAC) syndrome. 138 This disease is caused by glycine mutations in COL4A1 the gene that encodes collagen IV α1. Individuals with this disease have a number of extrarenal pathologies which reflect the ubiquitous expression of collagen IV α1α1α2. This disease demonstrates the importance of Bowman's basement membrane and the collagen IV α1α1α2 network within it.
Conversely, the glomerular basement membrane is largely unaffected in individuals with HANAC syndrome, which emphasises that the collagen IV α3α4α5 is the important collagen IV network within the GBM.
1.6.3 Regulation of the extracellular matrix by transforming growth factor
beta
Named originally for its ability to induce a transformed phenotype in cultured cells, 139, 140 transforming growth factor beta (TGF-β) is a dimeric cytokine. TGF-β comprises three isoforms (-β1, -β2 and -β3), that are part of a larger family of 33 proteins, 141 which include bone morphogenetic proteins, growth 37
and differentiation factors, activin and nodal. TGF-β signalling can stimulate or inhibit growth in a context dependent manner and is required for normal development. 142, 143 Moreover, TGF-β is involved in diverse cellular processes including differentiation, wound healing and angiogenesis. In addition to its role in normal processes, increases and decreases in production and/or activation of TFG-β lead to numerous disease sates including cancers, atherosclerosis and fibrotic disease of the kidney, liver, and lung. Fibrosis is defined as an excessive deposition of extracellular matrix (ECM), which leads to the destruction of organ architecture and impairment of organ function. 144
TGF-β proteins have widespread expression; nearly every cell in the human body secretes TGF-β and expresses receptors for it, however, TGF-β activity is tightly regulated. TGF-β is generated and secreted in a latent form and activation is localized to sites where TGF-β is released from latency.
Large aminoterminal prodomains aid folding and dimerisation, furin cleaves this prodomain, but even after secretion it remains associated with TGF-β through non-covalent interactions conferring latency to
TGF-β. In addition, this latency associated peptide (LAP) targets TGF-β for storage in the extracellular matrix, either in complex with latent TGF binding proteins (LTBPs) or fibrillins. 145, 146 TGF-β activation is regulated by thrombospondin-1 147, 148 and by binding of αV integrins to the prodomain leading to exertion of force upon this domain. 149 _ENREF_9
Once activated TGF-β binds to serine/threonine type II receptors; at the cell surface two type II receptors and two type I receptors form an active TGF-β family receptor. 150 Different type I/II receptor combinations allow TGF-β ligands to induce different signalling pathways, with different ligands having preference for different receptor combinations. The signalling complexity is enhanced by a plethora of accessory binding proteins which can enhance or modify ligand binding specificity. Activation of type
I/II heterodimers by TGF-β ligands causes the recruitment and phosphorylation of Smad proteins, in vertebrates there are eight Smad proteins. Smad2 and Smad3 are receptor Smads, which are phosphorylated through an interaction with activated type I receptors. These proteins bind to Smad4 and this complex tanslocates into the nucleus where it regulates gene expression. In contrast, Smad6 and Smad7 are inhibitory Smads which perform an autoregulation mechanism, controlling ligand induced signalling. 151 38
TGF-β is one of the most potent regulators of the ECM. TGF-β stimulates fibroblasts and other cell types to produce fibronectin, collagens I,III,V,VI, Proteoglycans, tenascin, osteonectin, osteopontin and thrombospondin, in addition to inhibitors of ECM degrading enzymes such as plasminogen-activator inhibitor type 1 and tissue inhibitor of metalloprotease. Moreover, it also represses the expression of
ECM degrading enzymes: collagenase, heparinase, and stromelysin. 152 Overall, active TGF-β increases ECM deposition and reduces ECM turnover, hence when dysregulated TGF-β is a driver of fibrosis in a number of disease states. As such TGF-β is potentially a major therapeutic target in fibrotic diseases such as scleroderma, 153 diabetic nephropathy 154 and renal fibrosis. 155
1.6.4 The Matrisome
Recent predictions suggest that the human ECM proteome (matrisome) contains 1065 components, expressed in a tissue dependent manner. 1 MS-based proteomic analyses of aorta, lung and colon tissue have revealed tissue restricted matrisomes containing 103-143 proteins. 1, 156 A small number of
ECM proteins have been identified as key determinants of glomerular function, however, it remains possible that many more may contribute. Global analysis of glomerular ECM will build upon the previously studied proteins and may provide unexpected insights to glomerular biology.
1.7 Podocyte adhesion
For most cells, including the cells of the glomerulus, adhesion is absolutely required for cell survival.
Adhesion sites do not act as mere anchoring points for cell attachment, rather they form a signalling nexus that controls all aspects of cell fate decisions. 157 Podocytes and endothelial cells express, like all adherent cells, transmembrane adhesion receptors. Adhesion receptors contain extracellular domains, which bind to specific ECM ligands or other cell surface receptors, and cytoplasmic domains that recruit effector proteins. Recruitment of proteins to the cytoplasmic domains of adhesion receptors leads to the formation of macromolecular signalling complexes. It is through these complexes that the
ECM controls cellular fate. 98, 99 Both cell-ECM and cell-cell contacts regulate the actin cytoskeleton through direct mechanical linkage and the recruitment of effector proteins. 157, 158 Exquisite control of the actin cytoskeleton is required to produce and maintain the elegant morphology of podocyte foot 39
processes. 159 Thus to understand the mechanisms of diseases which affect podocyte morphology, a comprehensive understanding of both the glomerular ECM and adhesion is required.
1.7.1 Integrin biology
A major family of proteins responsible for cell-ECM adhesion are the integrins . Integrins are αβ - heterodimers that propagate signals from outside the cell to the cell interior, whilst also transducing intracellular signals to the extracellular environment. Integrins form 24 different αβ combinations, 99 which have different affinities for ECM ligands and recruit different proteins to their cytoplasmic domains. 160 All integrins link to the actin cytoskeleton, with the exception of α6β4 which links to intermediate filaments. 99 Laminin binding integrins include: α1β1, α2β1, α3β1, α6β1, α10 β1, α7β1 and
α6β4, whereas the collagen binding integrins are: α1β1, α2β1, α10 β1, α11 β1 and αXβ2 (Figure 1.6) .161
Furthermore, integrin α3β1, α6β4 have a preference for laminin-511 and 521, which are present in the
GBM. 107 In contrast, α1β1 and α2β1 have higher affinities for collagen IV and collagen I respectively. 162,
163
Upon integrin engagement of the ECM, integrins cluster and become activated. Conformational changes in integrins are central to the regulation of their activity. Integrins adopt either a low affinity bent conformation, a primed, active high affinity extended conformation or a ligand occupied state. 164
Integrins lack intrinsic enzymatic activity; therefore, propagation of signals by integrins occurs via the recruitment of a number of adaptor and effector proteins into sites known as focal adhesions. At least
232 protein components are recruited to adhesion complexes in a cell type and context dependent manner, demonstrating the potential for adhesion signalling to bring about different cellular outcomes. 165 Furthermore , global analysis of adhesion complexes using MS suggest an even larger number of proteins may be recruited to focal adhesions. 4-6 These proteins include: scaffolds, adaptors, actin remodelling proteins, signalling proteins, guanine-nucleotide exchange factors (GEFs) , GTPase activating proteins (GAPs), Guanine nucleotide dissociation inhibitors (GDIs), serine threonine and tyrosine kinases and phosphatases. Furthermore, integrin association and crosstalk with other transmembrane receptors , such as syndecans and growth factor receptors, increase the potential for cellular regulation by adhesion complexes .166, 167 40
Figure 1.6 Laminin and collagen binding integrins The α1, α2, α10 and α11 integrin subunits all contain an inserted A-domain. This family of integrins bind collagen and laminin. α1β1 and α2β1 have preference for collagen IV and collagen I respectively. Non A-domain containing laminin binding integrins include: α6β4, α 3β1, α 6β1 and α 7β1.
41
1.7.2 Integrin ααα3βββ1
The laminin binding integrin α3β1 heterodimer is the most highly expressed integrin on the podocyte cell surface, and it is an indispensible link between podocyte foot processes and the GBM. 168-170 In humans homozygous mutations in ITGA3, the gene encoding integrin α3, leads to congenital nephrotic syndrome with defects in the GBM , interstitial lung disease and epidermolysis bullosa. 171 In addition, a mutation in ITGA3 causing a gain of glycosylation which prevents α3β1 dimer formation causes fatal interstitial lung disease and congenital nephrotic syndrome. 172 This phenotype is recapitulated in mice lacking the integrin α3 subunit, which die within the first day of life due to developmental defects in the lungs and kidneys. 173 These defects include loss of podocyte foot processes and disorganised thickened GBMs. The renal phenotype caused by mutations ITGA3 is probably due to loss of integrin
α3β1 from the podocyte cell surface, rather than other glomerular cells, because podocyte specific deletion of Itga3 in mice results in the same GBM and podocyte phenotype as the total knock out .174
Human mutations affecting integrin-β1 have not been described to date and this may be due to embryonic lethality since integrin-β1 forms at least 12 heterodimers. However , the role of this integrin has been studied with podocyte specific deletion of Itgb1 in the mouse. This resulted in a severe phenotype of proteinuria from birth and renal failure by 3 weeks featuring both glomerular and tubular pathologies. 175, 176
1.7.3 CD151
Humans with mutations in CD151 develop haematuria and proteinuria progressing to end stage kidney disease , in addition to pretibial epidermolysis bullosa, sensorineural deafness, and beta thalassaemia minor. 177 CD151 is a tetraspanin that binds tightly to integrin α3β1.178 In mice, deletion of Cd151, both globally and specifically in podocytes, causes early proteinuria, abnormalities of the GBM , loss of podocyte foot processes, glomerulosclerosis and renal failure. This phenotype, however, is dependent on the genetic background of the mice, with Cd151 -knockout mice on the FVB background displaying the pathological phenotype. 174, 179, 180 In contrast, Cd151 -knockout mice on the C57BL/6 background do not spontaneously develop renal failure , but when challenged with induced hypertension they develop 42
significant proteinuria. Furthermore, treatment of the susceptible Cd151 -knockout FVB strain with ACE inhibitors ameliorates progression of renal failure. In addition to in vivo experiments, in vitro experiments show that podocytes lacking CD151 lose their resistance to shear stress when cultured on laminin. 179 It is therefore likely that CD151 increases the strength of podocyte integrin α3β1-laminin-
521 adhesion complexes. This evidence supports a crucial role for integrin α3β1 as a major adhesion receptor, and in combination with CD151, a complex necessary for podocytes to withstand mechanical forces within the glomerulus.
1.7.4 Syndecans
In addition to the integrin family of adhesion receptors, transmembrane heparan sulphate proteoglycan receptors, such as the syndecan family, are key regulators of cell-ECM interactions. 167 Cooperation of integrins and syndecans in adhesion formation has been shown on a variety of ECM ligands including fibronectin, vitronectin and laminin. 181-184 Syndecans regulate integrin trafficking to the cell surface, 185 a process used by cells to regulate adhesion formation and disassembly. 186-188 In addition to modulating integrin dynamics, syndecans facilitate growth factor binding to their receptors. 189, 190 In podocytes null for EXT1 , a key molecule required for heparan sulphate glycosaminoglycan assembly, adhesion complexes are reduced in size, the actin cytoskeleton is rearranged and cell surface syndecan 4 is upregulated. 191 However, mice null for Ext1 specifically in podocytes do not develop significant proteinuria, despite some podocyte abnormalities, including a degree of foot process effacement. 192 In podocytes, autocrine signalling by the soluble vascular endothelial growth factor receptor, sFLT1, causes actin rearrangements in podocytes and this is associated with phosphorylation of both syndecan 1 and 4. 193 Thus , there is accumulating evidence that syndecans contribute to cell-matrix adhesion and signalling in podocytes.
1.7.5 Dystroglycan
Dystroglycan is a cell surface adhesion receptor that comprises a highly glycosylated extracellular
α− dystroglycan subunit and a non-covalently linked intracellular β− dystroglycan subunit . The
α− dystroglycan subunit binds to laminins and β− dystroglycan links to the actin cytoskeleton via an 43
interaction with utrophin. 194, 195 Dystroglycan is expressed on podocytes 196 and the expression pattern is altered in glomerular pathologies. 197, 198 As a result it seemed likely that dystroglycan was important for podocyte adhesion, however, defective glycosylation of α− dystroglycan which abrogates
α− dystroglycan-laminin interactions does not cause proteinuria, only mild podocyte foot process effacement .199 Furthermore, podocyte specific deletion of dystroglycan in mice causes only mild GBM thickening. 200 These data suggest that dystroglycan is not a critical adhesion receptor in podocytes.
1.7.6 Integrin mediated adhesion complexes
1.7.6.1 Talin 1
A number of proteins link integrins to the actin cytoskeleton. One such protein is talin-1, a 270 kDa protein comprising an N-terminal globular head and flexible rod domain. The head domain contains binding sites for integrin-β subunit cytoplasmic tails, F-actin, focal adhesion kinase (FAK) and
PIPK1 γ90. The rod domain contains additional binding sites for integrin, actin , the Rap1 effector RIAM and multiple vinculin binding sites. 201 Finally , the C-terminal domain contains helices responsible for talin-1 dimerisation. Binding of talin-1 to the cytoplasmic tail of β-integrins triggers a conformational change in the extracellular domain of integrins, which amplifies the affinity of the integrin for the ECM.
Talin-1 dependent recruitment of further proteins to active integrin triggers the formation of focal adhesions. 202
Talin-1 expression in podocytes is required for the specialised actin morphology of foot processes.
Podocyte specific Tln1 -knockout mice develop proteinuria and die within 10 weeks. 203 These mice, however, do not have major defects in integrin β1 activation or podocyte adhesion. Nevertheless, the actin cytoskeleton is perturbed and podocyte foot processes are lost. These data show that talin-1, a protein known to be important in focal adhesion formation and linkage to the actin cytoskeleton in vitro , is important for relaying signals from integrins to the actin cytoskeleton in podocytes in vivo .203
44
1.7.6.2 Vinculin
Another key adaptor protein involved in integrin mediated adhesion complex formation is vinculin.
Vinculin is a 123 kDa protein that is recruited by talin to focal adhesions and binds to the actin cytoskeleton. 204 Vinculin comprises an N-terminal head, proline rich neck and a C-terminal tail domain. 205, 206 Cytoplasmic vinculin assumes an autoinhibited inactive conformation ,207 but following recruitment to adhesion sites by talin, vinculin is extended by forces applied through the actin cytoskeleton, revealing an open active state. 208 This conformational change allows vinculin to directly interact with a number of proteins including α-actinin, Arp2/3, actin and paxillin. 209, 210 Considering vinculin is an important link between integrins and the actin machinery this protein may have a key role in force sensing by podocytes via integrin α3β1.
1.7.6.3 Paxillin
Paxillin is another component of focal adhesions that acts as a scaffolding protein. It contains multiple protein-binding modules, many of which are regulated by phosphorylation. Paxillin localises to focal adhesions when phosphorylated, demonstrating the importance of phosphorylation in controlling adhesion signalling .211, 212 Paxillin is an important molecular adaptor; its N-terminus controls most of its signalling activity and provides docking sites for vinculin, FAK, Src and Crk1/2. Moreover, paxillin brings about spatiotemporal control of Rho family small GTPases by recruiting numerous GEFs and
GAPs. 213 Although paxillin is known to be an important focal adhesion protein, it has yet to be studied in the glomerulus.
1.7.6.4 Focal adhesion kinase
Another highly studied focal adhesion protein is FAK, a non-receptor tyrosine kinase, which is recruited to focal adhesions by talin and paxillin. 214 FAK has a number of roles at focal adhesion sites including recruitment of p130Cas, Crk1/2 and Src-family kinases. 215 Global deletion of FAK in mice is lethal in embryogenesis, causing a profound migration defect.216 The importance of FAK in podocytes was highlighted by the observation that FAK is phosphorylated upon podocyte injury. 217 Surprisingly , podocyte-specific deletion of FAK in mice leads to a normal phenotype; however, these mice are 45
protected from proteinuria and podocyte injury after experimental podocyte insult. 217 Additionally, FAK inhibition reduces podocyte injury in a mouse model of glomerular injury. 217
A role for FAK has also been found in Alport syndrome where ectopic laminins, α1 and α2, accumulate in the GBM. Laminin α2 causes phosphorylation of FAK at Y397 and this phosphorylation associates with the upregulation of the matrix metalloproteinases (MMPs) 9 and 10 , in addition to GBM defects. 218
FAK inhibition reduced proteinuria, MMP levels and GBM defects in this context .218 These data support a role for FAK in glomerular dysfunction.
1.7.6.5 Integrin linked kinase/PINCH/parvin
Integrin linked kinase (ILK) binds directly to the integrin β1 cytoplasmic tail and is important for signal transduction at adhesion sites. 219 ILK was originally identified as a kinase, but increasing evidence suggests that it is a pseudokinase. 220-225 In fact, the C-terminal kinase homology domain of ILK mediates multiple protein–protein interactions at adhesion sites, including interactions with α/β/γ -
Parvin. 226, 227 ILK also contains 5 ankyrin domains that mediate interactions with PINCH-1/2. 228-230
Kindlin 2 is another ILK interacting protein, which is expressed in podocytes. Kindlin 2 localises to focal adhesions and through ILK/PINCH/parvin , or migfilin–filamin binds to the actin cytoskeleton. 231-233 It is through this scaffolding role that ILK orchestrates focal adhesion signalling. The ILK/PINCH/parvin complex influences the actin cytoskeleton, 234 in addition to negatively regulating cell contractility. 235
A number of studies have shown that the ILK/PINCH/parvin axis is vital during development; total loss of ILK or PINCH in mice is lethal in embryogenesis, due to failure in epiblast polarisation .214, 216 Indeed, this axis is important for adhesion signalling in the glomerulus. The interaction between ILK and α- parvin is required for kidney development; mutations in ILK in the α-parvin binding site causes renal agenesis. 236 Furthermore, a similar phenotype is observed when α-parvin is genetically deleted in mice. 236 Podocyte-specific loss of ILK in mice causes GBM defects, loss of slit diaphragms and podocyte foot process effacement .237, 238 In addition to the role of ILK at cell-ECM adhesion sites, ILK interacts with nephrin at podocyte slit diaphragms.238 Therefore, ILK represents a potential link between cell-cell and cell-ECM adhesion signalling. Finally, increased expression of ILK is observed in 46
a variety of glomerular diseases. 239, 240 This evidence strongly supports a role for ILK in adhesion signalling in podocytes.
1.7.7 The adhesome
Overall , cell adhesion to the GBM occurs at a complex cell-matrix interface (Figure 1.7). A wide range of proteins localise to focal adhesions. Through the recruitment of these proteins integrins transmit information regarding the extracellular environment to the cell interior. A process that is undoubtedly vital for glomerular cells. The cellular adhesome has been predominantly investigated in the context of fibroblast-fibronectin adhesion complexes, but less so for epithelial cells .165 Similar analyses in glomerular cells will help to build our understanding about the key cellular components that are involved in cell-matrix adhesion in the glomerulus. 47
Figure 1.7 Integrin mediated adhesion complex The ECM and the cell cytoskeleton are physically linked through integrin mediated adhesion complexes. Top panel; simplified adhesion complex view displaying two major linkages from integrins to the actin cytoskeleton: talin/vinculin/paxillin and ILK/PINCH/parvin. Bottom panel; a predicated ECM- 48
adhesion complex protein-protein interaction network. Nodes (circles) represent proteins and edges (grey lines) represent reported protein-protein interactions. ILK, Integrin linked kinase. 1.7.8 Slit diaphragm
In addition to cell-ECM adhesion, cell-cell adhesion is also critical for the GFB. Slit diaphragms are the unique cell-cell junctions that connect the entire length of adjacent podocyte foot processes, forming a zipper-like substructure containing pores .18 The calculated cross-sectional dimensions of these pores is approximately the size of an albumin molecule; therefore, consistent with the observations from tracer studies , using ferritin and dextrans , that the slit diaphragm contributes to the retention of macromolecules within the circulation. 241 35 Interestingly, these pores increase in size with proteinuria, perhaps providing an explanation for the increased transit of proteins across the GFB in pathologies where foot processes appear normal .19
Podocytes change dramatically in human glomerular disease; their actin-rich foot processes flatten and slit diaphragms are lost. 159 Similarly , these changes are observed in animal studies of puromycin aminonucleoside (PAN) induced nephrotic syndrome. 242 Whilst these dramatic morphological changes are associated with a profound GFB defect, remarkably these changes seem to completely reverse , for example in the subset of patients with nephrotic syndrome who respond to treatment with glucocorticosteroids.
The first junctions to form in podocytes are apical and have been described as tight junctions. 243 During glomerular development the junctional complexes descend toward the GBM and widen to become the mature slit diaphragm. These are highly specialised and unique junctions and many studies have investigated the components that localise to slit diaphragms . The tight junction components: zona occludens protein (ZO-1) ,244 JAM-A, occludin and cingulin are associated with slit diaphragms and PAN nephrosis increases the expression of these tight junction components .245 Additionally, classical components of adherens junctions including cadherin-3 and α-, β-, γ-catenins 246 and the gap junction protein connexin-43 247 localise to the podocyte slit diaphragm . Moreover, connexin-43 is upregulated the early phase of PAN nephrosis. 247 Taken together, these findings suggest, not surprisingly, that there is context dependent composition of slit diaphragms.
49
1.7.8.1 Neph family proteins
The slit diaphragm is a hybrid of adherens and tight junction components, but it also contains unique proteins (Figure 1.8). Nephrin and the homologues Neph-1, Neph-2 and Neph-3 are members of the immunoglobulin superfamily of cell adhesion receptors that are involved in the development of specialised junctions in neurons and at the slit diaphragm. 248 Orthologs of these proteins are expressed in Drosophila nephrocytes, 249 which have diaphragm structures with very similar composition to the mammalian slit diaphragm. 250 Further back in evolution , C. Elegans express orthologs of NEPH1 and nephrin, called SYG-1 and SYG-2 respectively. SYG-1 and SYG-2 are required for synapse formation and specificity. Investigation of the crystal structures of these orthologs revealed SYG-1 homodimers with a conserved binding interface and an unusual, angled geometry in the heterophilic SYG1/2 complex. 251 The crystal structures of nephrin and Neph1-3 remain unresolved, however, there is evidence for these proteins interacting in both a heterophilic and homophilic manner (Figure 1.8). 252-255
Nephrin and Neph-1 are required for slit diaphragm formation. Mutations in nephrin cause human congenital nephrotic syndrome , with a high incidence in Finland , although many mutations have now been described in individuals with later onset of disease and from diverse ethnic backgrounds .256
Infants require albumin infusions to maintain intravascular volume and ultimately proceed to removal of their kidneys prior to dialysis and transplantation. This disease phenotype is mimicked in the mouse , where deletion of Nphs1 leads to early proteinuria and death within 24 hours. These mice do not form slit diaphragms or normal podocyte foot processes .257 Neph1 deletion in mice is also associated with perinatal lethality with proteinuria and podocyte foot process effacement. 258 As yet no human mutations in NEPH1 have been described , but it is more widely expressed than nephrin; therefore , mutations may be incompatible with life.
1.7.8.2 Slit diaphragm signalling
Neph family of proteins are thought of as the signature slit diaphragm proteins, but there are additional proteins that are vital for slit diaphragm signalling including podocin, CD2AP and FAT atypical cadherin-1. Podocin is a stomatin family protein that is mutated in patients with early onset nephrotic syndrome .259 Moreover, podocin is important for the recruitment of proteins to the slit diaphragm complex and for facilitating signalling .260 CD2AP is an adaptor protein that is required for maintaining 50
the integrity of the GFB in mice 261 and humans, 262 in addition to the balance of receptor tyrosine kinase signalling in podocytes. 263 FAT atypical cadherin-1 also regulates GFB formation and mice lacking this component have significant glomerular defects, in addition to eye and brain abnormalities. 244
Phosphorylation of nephrin and Neph-family proteins by Src family kinases is key to signal transduction at the slit diaphragm. Indeed, deletion of the Src family kinase member, Fyn , results in GFB dysfunction. 264 Phosphorylation of nephrin by Fyn leads to the recruitment of a number of proteins including the adaptor proteins Nck1/2, 265, 266 Crk1/2, 267 CrkL ,268 Grb2 269, 270 and phosphoinositide 3-OH kinase (PI3-kinase) .271, 272 Following recruitment, Nck1/2 binds phosphorylated nephrin which initiates actin reorganisation via the actin nucleation factor N-WASP. 273 The receptor Robo2 also links to
Nck1/2 and is expressed in podocytes. 274 Robo2 inhibits actin reorganisation and it appears to negatively regulate nephrin and Nck1/2 signalling. Crk1/2 is recruited to phosphorylated nephrin via p130Cas. Deletion of Crk1/2 reduces podocyte foot process effacement in a glomerular injury model. 267
The p85 regulatory subunit of PI3-kinase also interacts with nephrin leading to downstream activation of Akt 271 and subsequently to actin reorganisation. 272 Emphasising the importance of signalling via Akt, deletion of the Akt2 causes podocyte dysfunction .275
Regulation of the podocyte actin cytoskeleton is key to maintaining barrier integrity. Actin reorganisation in podocytes is likely to relate directly to the dramatic podocyte foot process effacement, which is seen across the spectrum of glomerular disease. The actin cross -linking protein alpha-actinin-
4 has been associated with human nephrotic syndrome. Moreover, mutations in ACTN4 are associated with adult onset focal segmental glomerulosclerosis (FSGS) .276 These are gain of function mutations with the mutant alpha-actinin-4 protein binding filamentous actin more tightly than the wild-type protein.
This indicates that balanced actin regulation is important for normal podocyte function. Another class of actin regulators are the GTPases, which are in turn regulated by GEFs, GAPs and GDIs. Podocyte specific deletion of the GTPase RhoA does not cause barrier defects. 277 However , RhoA is activated in humans with mutations in the formin INF2, a common cause of adult onset FSGS. 278 The small
GTPase Rac1, which is required for lamellipodia formation, has a role in podocyte biology. Rac1 is not essential for glomerular development , but over expression leads to GFB dysfunction either with constitutive activation of Rac1 279 or RhoGDI-alpha knockout. 280 However , a number of studies have also shown that Rac1 activation protects the GFB from injury. 268, 281 Thus, it is likely that a fine balance 51
of Rac1 activity is required for GFB stability. The GTPase CDC42 is linked to the formation of filopodia and its absence results in early barrier dysfunction .277, 282 This may be due to links with apical-basal polarity proteins, which are also required for slit diaphragm formation. 283
The dynamic regulation of this specialised cell junction undoubtedly requires quality control and recycling of components . In rat glomeruli turnover rates of slit diaphragm proteins is high and regulated by atypical protein kinase C (aPKC). 284 Accordingly, the endocytic pathway components dynamin, synaptojanin and endophilin have been shown to be important for maintaining barrier function. 285 The correct localisation of proteins is also key; Neph1 and nephrin localisation at the plasma membrane requires the endocytic protein unconventional myosin-Ic. 286 The long tailed myosin, unconventional myosin-Ie may also contribute to endocytosis in podocytes. 287, 288 Finally, mutations in the transient receptor potential cation channel TRPC6 have been associated with adult onset FSGS, indicating the importance of calcium signalling at the slit diaphragm .289, 290
The slit diaphragm includes components that make the connections between adjacent podocyte foot processes extracellularly, in addition to the more dynamic network of proteins that assemble intracellularly (Figure 1.8). To identify novel components of this complex, unbiased approaches with
MS have led to the discovery of proteins including IQGAP , MAGI-2, CASK, spectrins and alpha- actinin .291 As methods to isolate and analyse silt diaphragms improve, more unexpected components may be identified. A recent bioinformatic analysis of the cadherin adhesome has predicted an assembly of 170 components, which provides an indication of the probable scale of nephrin signalling complexes .292 52
Figure 1.8 Overview of the podocyte slit diaphragm Slit diaphragms form a physical link between adjacent podocyte foot processes. A number of adaptor, signalling and cytoskeleton proteins are recruited to slit diaphragms. Top panel; proteins that are known to localise to podocyte slit diaphragms. Bottom panel; protein-protein interaction network view of known slit diaphragm components. The network is annotated with processes that slit diaphragm proteins are known to regulate. Nodes (circles) represent proteins and edges (grey lines) represent reported protein-protein interactions.
53
1.8 Mass spectrometry-based proteomics
To date candidate based approaches have advanced our understanding of glomerular biology, with laminins, nidogens, perlecan, agrin and type XVIII collagen being the most well studied glomerular
ECM components. Understanding the function of these molecules has been advanced through studies of genetic mutations in humans and animals in addition to targeted studies of these proteins at the level of tissue expression, single molecule structure and analysis of interactions with other biological molecules. These are irreplaceable scientific approaches, however, systems biology attempts to systematically study all concurrent processes in a tissue by the global measurement of differentially perturbed states and this approach has enormous potential to complement targeted studies.
Immunoaffinity-based approaches such as immunofluorescence microscopy, flow cytometry, enzyme- linked immunosorbent assay (ELISAs) and immunoblotting are central to the study of proteins in biological systems. One caveat of these approaches is that they rely on antibody affinity, antigen accessibility and the availability of reagents. Furthermore, these approaches do not enable the identification and quantification of hundreds of proteins in a single experiment or the discovery of novel proteins as the target protein is predetermined. As a result alternative techniques are required to enable the field of systems biology.
Analysis of the transcriptome with microarrays and RNA sequencing experiments enable global quantification of gene expression, and these approaches are employed for systems level analyses.
These analyses are frequently utilized as surrogates for protein abundance. However, although powerful, these techniques do not represent the true presence and abundance of proteins within a biological sample because protein expression is regulated through post-transcriptional mechanisms, such as synthesis and degradation. 293 In contrast proteomics allows the unbiased study of protein composition and can therefore be used to examine multiple concurrent processes occurring within a biological system.
54
Ideally, researchers would like to identify and quantify all proteins in biological sample under different experimental or disease conditions. However, the desire to achieve maximal coverage including low abundance proteins, and to probe the huge array of protein modifications necessitates highly sensitive analytical techniques. Mass spectrometry (MS) enables the analysis of proteins in a global manner. 294
The analytical instrument, a mass spectrometer, contains: an ion source, which generates charged gaseous analytes; a mass analyzer, which measures the mass-to-charge ratio (m/z) of analytes; and a detector that records the number of ions at each m/z value.
For analysis of proteins, the process of generating ions in the gas phase is most commonly achieved by either electrospray ionization (ESI) 295 or matrix-assisted laser desorption/ionization (MALDI). 296
These processes overcome the propensity of large molecules (such as proteins/peptides) to fragment when ionized, and are referred to as 'soft ionization techniques. ESI ionizes the analytes out of a solution and is therefore readily coupled to prior liquid-based separation tools, for example, chromatography. MALDI sublimates and ionizes proteins or peptides out of a dry, crystalline matrix via laser pulses. MALDI-MS is normally used to analyze relatively simple protein/peptide mixtures, whereas integrated liquid-chromatography ESI-MS systems (LC-MS) are preferred for the analysis of complex samples. An additional benefit of ESI is the production of multiply charged ions extending the m/z range of the mass analyser. The development of ESI in 1989 led to John B. Fenn sharing the 2002
Nobel prize in chemistry.
Following the generation of ions in the gaseous phase, ions are captured in mass analyzers. These devices operate by measuring the trajectories of ions in an electric field or by trapping ions for further manipulation. Examples of mass analyzers include quadrupole mass filters, time of flight instruments, fourier transform ion cyclotron analyzers and the Orbitrap. These analyzers can be stand alone or, more commonly, assembled in tandem to take advantage of their relative strengths.
Biological samples are complex and frequently contain thousands of proteins. In order to detect, and ideally quantify, all of the proteins present in a biological sample, proteins are digested to peptides
(frequently using enzymatic trypsin digestion) and these peptides are analyzed. By breaking down a 55
complex protein into its building blocks there is more chance of assigning a unique identification (ID).
An analogy here is the accurate weighing of all staff at a University. It would not be possible to give a unique ID based on their total body weight alone due to the limitation of accuracy and precision of the device used to weigh each individual and the large number of individuals. However, if each individual was asked to remove their coat, shoes and bag and each of these items were weighed separately, then the probability of assigning a unique ID to each member of staff would increase. To assign a unique ID to peptides, tandem MS (MS/MS), employs two mass analyzers in series. In the first analyzer precursor peptides are selected and analyzed. The exact mass of this precursor peptide is determined and then it is then fragmented along its backbone, usually by collision-induced dissociation
(CID) with an inert gas. These fragment ions are then detected in the second analyzer, giving rise to characteristic spectra. The fragment and precursor m/z measurements can be compared to protein database search engines. These databases have libraries of peptides predicted in silico from theoretical digests of the complete genome sequence of the organism of interest and therefore the identity of a peptide is thus inferred. The next challenge is to determine the original proteins from which the identified peptides have arisen. This is trivial in the case of peptides that are unique to a given protein, but pose a complex problem for peptides with sequences that are shared between proteins. As a result a range of software are now available for processing MS proteomic data and the most popular include; Mascot, 297 SEQUEST 298 and X!Tandem. 299
1.9 Thesis aims
In health the podocyte coordinates signals from the slit diaphragm and cell-ECM interactions. As our understanding of mechanisms that control cell adhesion in the glomerulus develops, then insight into the effects of disease will improve. The ultimate goal will be to develop targeted therapies to prevent and repair defects in the GFB and to restore glomerular function. Currently the glomerular ECM and podocyte adhesion have not been studied in a global manner.
The principal aim of this thesis is to identify, validate and characterise components of the glomerular
ECM and podocyte adhesion complexes in health and disease. This will be achieved by using mass spectrometry based proteomics and imaging techniques to:
56
1. Catalogue the components of the human glomerular ECM
Prior knowledge of the normal composition of the human glomerular ECM is prerequisite to understanding the changes that occur in the glomerular ECM during disease. Although glomeruli have been studied with MS, I hypothesise that an ECM enrichment strategy coupled with MS-based proteomics will enable the identification of a more comprehensive glomerular specific ECM proteome.
In turn this data-set will provide the framework for a more comprehensive understanding of glomerular
ECM biology.
2. Investigate the relative contribution of glomerular cells to the ECM
The glomerular basement membrane is a crucial structure within the glomerulus, but isolation of the
GBM is currently technically unfeasible. To study the GBM, GEC and podocyte cell-derived ECMs
(CDMs) are an alternative option. Proteomic analysis of GEC and podocyte CDMs has the potential to identify cell specific differences in ECM production, reveal the origin of GBM components, and therefore provide new insights into the regulation of the GBM.
3. Characterise the influence of genetic background, sex and degree of microalbuminuria on the
composition and organisation of glomerular ECM
Genetic background and sex determine the degree of albuminuria exhibited by an individual, but the reasons for this remain poorly understood. This project will utilise electron microscopy and proteomics to explore the ECM as a possible cause of sex- and genetic-dependent variation in microalbuminuria. I hypothesise that this analysis will identify structural and compositional differences within the glomerular
ECM, which may contribute to microalbuminuria.
4. Investigate podocyte cell-matrix adhesion complexes
The sophisticated function of glomerular filtration relies on podocyte adhesion at the interface with the
ECM. Adhesion underpins the regulation of the actin cytoskeleton, which in turn generates the labyrinthine podocyte foot process architecture. Using proteomics to identify differences between adhesion complexes formed by podocytes attached to GBM ligands, collagen IV and laminin, may provide important insights into normal podocyte adhesion signalling. Moreover, these analysis may discern possible mechanism pertaining to the pathogenesis of Alport and Pierson syndromes. I 57
hypothesise that this analysis will pinpoint fundamental differences in adhesion signalling on different
ECM ligands.
5. Investigate podocyte cell-cell adhesion complexes
In addition to cell-ECM adhesion, cell-cell adhesion is also critical within the glomerulus. Podocyte slit diaphragms are unique cell-cell junctions that are crucial for podocyte function. Nephrin is a key component of the slit diaphragm, but both low expression and biophysical properties have made this molecule challenging to study. The generation of cell lines expressing different nephrin constructs, coupled with proteomic approaches will be applied to the study of nephrin induced signalling complexes. Furthermore, these analyses may identify novel signalling pathways regulated by slit diaphragms. 58
2 Defining the glomerular matrisome
2.1 Introduction
The extracellular matrix (ECM) is a scaffold that creates the complex structure of the glomerulus, in addition to forming a signalling platform that orchestrates cell behaviour. Indeed, the glomerular ECM is critical for glomerular function as it is frequently altered in glomerular disease. Candidate based studies of the glomerular ECM have revealed tissue restricted isoforms of laminin ( α5β2γ1) and collagen IV
(α3α4α5), which are crucial components of the glomerular basement membrane (GBM). The ECM in the glomerulus can be segregated on the basis of collagen IV chain expression: collagen IV α3α4α5 predominates within the GBM, collagen IV α1α1α2 is highly expressed within the mesangium and collagen IV α5α5α6 is expressed within Bowman's capsule. Furthermore, a handful of other ECM proteins including agrin, nidogen and perlecan have been studied in detail in the glomerulus. However, recent data have revealed that the human genome encodes 1065 ECM proteins, 1 which suggests the
ECM may be far more complex than previously appreciated. A more detailed understanding of the composition of the human glomerular ECM will facilitate a better understanding of glomerular disease processes. 59
2.2 Statements
Author contributions to data generation and analysis presented as figures of this paper are indicated in figure legends.
Rachel Lennon planned the study and designed experiments. Adam Byron, Jonathan D. Humphries and Michael J. Randles contributed to the study design and analysis. Paul E. Brenchley, Roy Zent and
Martin J. Humphries contributed to the study design.
Michael J. Randles performed Gene Ontology enrichment analysis of the human glomerular ECM proteomic dataset, performed in gel-proteolytic digestion, offline peptide desalting of human glomerular
ECM samples, analysis of peptide intensity for relative quantification of glomerular ECM proteins by mass spectrometry, and evaluated the effectiveness of the ECM enrichment approach used in this study.
Rachel Lennon isolated enriched glomerular ECM fractions, performed Western blotting and immunofluorescence imaging and cross-referenced the proteomic dataset with the Human Protein
Atlas. Adam Byron built the database of human ECM protein-protein interactions and performed network topological analysis. Alex Carisey performed colocalisation analyses of immunofluorescence images. David Knight aided with mass spectrometry data acquisition.
The manuscript is written in the style of a Journal of the American Society of Nephrology research article by Rachel Lennon, Adam Byron, Jonathan D. Humphries and Michael J. Randles. Roy Zent and
Martin J. Humphries contributed to the preparation of the manuscript and the manuscript was critically assessed by all authors . Author guidelines restrict the size of the article to 3000 words or less
(excluding title page, methods, figure legends, tables, and references) .
60
2.3 Global analysis reveals the complexity of the human
glomerular extracellular matrix
Rachel Lennon, 1,2 Adam Byron, 1, * Jonathan D. Humphries, 1 Michael J. Randles, 1,2 Alex Carisey, 1
Stephanie Murphy, 1,2 David Knight, 3 Paul E. Brenchley, 2 Roy Zent, 4,5 and Martin J. Humphries. 1
1Wellcome Trust Centre for Cell-Matrix Research, Faculty of Life Sciences, University of Manchester,
Manchester, UK; 2Faculty of Medical and Human Sciences, University of Manchester, Manchester, UK;
3Biological Mass Spectrometry Core Facility, Faculty of Life Sciences, University of Manchester,
Manchester, UK; 4Division of Nephrology, Department of Medicine, Vanderbilt University Medical Center,
Nashville, TN, USA; and 5Veterans Affairs Hospital, Nashville, TN, USA.
*Present address: Edinburgh Cancer Research UK Centre, Institute of Genetics and Molecular Medicine,
University of Edinburgh, Edinburgh, UK.
Running title: Proteome of the glomerular matrix
Corresponding author:
Dr Rachel Lennon, Wellcome Trust Centre for Cell-Matrix Research, Michael Smith Building, University of Manchester, Manchester M13 9PT, UK.
Phone: 0044 (0) 161 2755498. Fax: 0044 (0) 161 2755082.
Email: [email protected] 61
2.3.1 Abstract
The glomerulus contains unique cellular and extracellular matrix (ECM) components, which are required for intact barrier function. Studies of the cellular components have helped to build understanding of glomerular disease; however, the full composition and regulation of glomerular ECM remains poorly understood. Here, we employed mass spectrometry–based proteomics of enriched ECM extracts for a global analysis of human glomerular ECM in vivo and identified a tissue-specific proteome of 144 structural and regulatory ECM proteins. This catalogue includes all previously identified glomerular components, plus many new and abundant components. Relative protein quantification demonstrated a dominance of collagen IV, collagen I and laminin isoforms in the glomerular ECM together with abundant collagen VI and TINAGL1. Protein network analysis enabled the creation of a glomerular ECM interactome, which revealed a core of highly connected structural components. More than half of the glomerular ECM proteome was validated using co-localisation studies and data from the Human Protein
Atlas. This study yields the greatest number of ECM proteins, relative to previous investigations of whole glomerular extracts, highlighting the importance of sample enrichment. It also demonstrates that the composition of glomerular ECM is far more complex than previously appreciated and suggests that many more ECM components may contribute to glomerular development and disease processes. 62
2.3.2 Introduction
The glomerulus is a sophisticated organelle comprising unique cellular and extracellular matrix (ECM) components. Fenestrated capillary endothelial cells and overlying podocytes are separated by a specialised glomerular basement membrane (GBM), and these three components together form the filtration barrier. Mesangial cells and their associated ECM, the mesangial matrix, exist between adjacent capillary loops and maintain the three-dimensional organisation of the capillary bundle. In turn the parietal epithelial cells and ECM of Bowman’s capsule enclose this network of capillaries. Cells adhere to ECM proteins via adhesion receptors, and these interactions are required to maintain intact barrier function of the glomerulus. 300, 301
In addition to operating as a signalling platform, ECM provides a structural scaffold for adjacent cells and has a tissue-specific molecular composition. 302, 303 Candidate-based investigations of glomerular
ECM have focused on the GBM and have shown it resembles the typical basal lamina found in multicellular organisms, containing a core of glycoproteins (collagen IV, laminins and nidogens) and heparan sulphate proteoglycans (agrin, perlecan and collagen XVIII). 304 Mesangial and parietal cell
ECMs have been less well investigated nonetheless, they are also thought to contain similar core components in addition to other glycoproteins, including fibronectin. 305, 306 Thus the glomerulus consists of a combination of condensed ECM within the GBM and Bowman’s capsule and loose ECM supporting the mesangial cells.
The ECM compartments in the glomerulus are thought to be distinct and to exhibit different functional roles. The GBM is integral to the capillary wall and therefore functionally linked to glomerular filtration. 304 Mutations of tissue-restricted isoforms of collagen IV ( COL4A3, COL4A4 and COL4A5 ) and laminin ( LAMB2 ), which are found in the GBM, cause significant barrier dysfunction and ultimately renal failure. 307, 308 Less is understood about the functions of mesangial and parietal cell ECMs, although expansion of the mesangial compartment is a histological pattern seen across the spectrum of glomerular disease. 309
63
Compositional investigation of the distinct glomerular ECM compartments is limited by the technical difficulties of separation. Early investigations of GBM constituents used the relative insolubility of ECM proteins to facilitate separation from cellular proteins in the glomerulus, but did not separate the GBM from mesangial and parietal cells ECMs. 310, 311 Recently studies incorporating laser microdissection of glomerular sections have been coupled with proteomic analyses. 312, 313 These studies report both cellular and ECM components and typically require pooled material from glomerular sections to improve protein identification. The ability of laser microdissection to separate glomerular ECM compartments has not yet been tested, however this approach will be limited by the amount of protein it is possible to retrieve. To achieve good coverage of ECM proteins within a tissue, proteomic studies need to combine a reduction in sample complexity with maximal protein quantity. Currently the inability to separate glomerular ECM compartments in sufficient quantity is a limitation that prohibits proteomic studies of these structures, however for other tissues proteomic analysis of ECM has been achieved by enrichment of ECM combined with sample fractionation. 314
Whilst the composition of the ECM in other tissues has been addressed using proteomic approaches, 314 studies of glomerular ECM to date have employed candidate-based technologies.
These studies have identified key molecular changes during development and disease and highlighted the compositional and organisational dynamics of glomerular ECM. Nonetheless, the extracellular environment within the glomerulus is the setting for a complex series of interactions between both structural ECM proteins and ECM-associated proteins, such as growth factors 315-317 and proteases, 318 which together provide a specialised niche to support glomerular cell function. Therefore, to interrogate this complexity effectively, a systems-level understanding of glomerular ECM is required. To address the need for a global analysis of the extracellular environment within the glomerulus, we used mass spectrometry (MS)–based proteomics of glomerular ECM fractions to define the human glomerular
ECM proteome. 64
2.3.3 Results
2.3.3.1 Isolation of glomerular ECM
Using purified isolates of human glomeruli (Figure 2.1A), we developed a fractionation approach to collect glomerular ECM proteins (Figure 2.1B). Prior to homogenisation and solubilisation, extracted glomeruli appeared acellular, suggestive of successful ECM enrichment (Figure 2.1A). Western blotting confirmed enrichment of ECM proteins (collagen IV, laminin) and depletion of cytoplasmic and nuclear proteins (nephrin, actin, lamin B1) in the ECM fraction (Figure 2.1C). Glomerular ECM fractions from three adult male human kidneys were then extracted and analysed by MS.
2.3.3.2 Gene ontology enrichment analysis of ECM fractions
All proteins identified by MS were allocated to categories according to their gene ontology (GO) assignment, and the enrichment of GO terms in the dataset was assessed using GO enrichment analysis. The network of enriched GO terms clustered into three sub-networks, and the largest, most confidently identified sub-network was for ECM-related GO terms (Figure 2.2A). A large majority of other proteins were included in two additional sub-networks representing mitochondrial and cytoskeletal GO terms, and the presence of these proteins in the ECM fractions is likely to be due to the strength of their inter-molecular interactions with ECM proteins. Spectral counting was used to determine the enrichment of ECM in each of the four fractions collected from isolated human glomeruli. There was 38% enrichment of extracellular proteins in the glomerular ECM fractions (Figure
2.2B), and this compares favourably with the 12–30% enrichment of ECM proteins reported in comparable proteomic studies. 314 65
Figure 2.1 Isolation of enriched glomerular ECM. A: Human glomeruli were isolated by differential sieving, yielding >95% purity. Prior to homogenisation, glomeruli appeared acellular (right panel). B: A proteomic workflow for the isolation of enriched glomerular ECM by fractionation (see methods for details). C: Coomassie staining and Western blotting (WB) of fractions 1–3 and the ECM fraction probing for the extracellular proteins with pan-collagen IV and pan-laminin probes and the intracellular proteins nephrin, actin and lamin B1. M, molecular weight marker. Rachel Lennon generated data for this figure. 66
Figure 2.2 MS analysis of enriched glomerular ECM fractions. A: Gene ontology (GO) enrichment analysis of the full MS dataset. Nodes (circles) represent enriched GO terms and edges (grey lines) represent overlap of proteins between GO terms. Node colour indicates the significance of GO term enrichment; node diameter is proportional to the number of proteins assigned to each GO term; edge weight is proportional to the number of proteins shared between connected GO terms. The full list of GO terms are detailed in Figure S2.7. B: All four protein fractions were analysed by MS, and spectral counting was used to determine the enrichment of ECM proteins (as identified by GO analysis). The mean ECM enrichment was 38% from three biological replicates. C: Relative quantification for the ten most abundant ECM proteins detected by MS. Relative protein abundance was calculated using peptide intensity as described in the Methods. Gene names are shown for clarity, and collagen IV and laminin isoforms are combined as one value. D: Western blotting (WB) confirmed enrichment of TINAGL1 and collagen VI in glomerular ECM. Michael Randles generated data parts A, B and C. David Knight contributed to part C. Rachel Lennon generated data part D. 67
2.3.3.3 Defining the glomerular ECM proteome
Using the GO classification of extracellular region proteins and cross-referencing with the human matrisome project ,302 we identified 144 extracellular proteins in the glomerular ECM, including all previously known glomerular ECM components (Table S2.1). Only proteins identified in at least two of the three biological replicate analyses were included in the dataset, and identified ECM proteins were further categorised as basement membrane, other structural ECM proteins or ECM-associated proteins. There are a number of published glomerular proteomic studies 319-323 and a comparison was made to protein identifications in our dataset (Figure 2.3). Of the two published studies for which complete MS datasets were available, 320, 323 neither study enriched for glomerular ECM or analysed more than one biological replicate. We detected all of the 15 ECM proteins reported by Yoshida et al. 320 (which represents 8% of ECM proteins detected in this study) and 76% of the 91 ECM proteins reported by Cui et al. 323 (which represents 39% of ECM proteins detected in this study), based on identification in one biological replicate, as reported in these published studies. Furthermore, we identified 12 times as many ECM proteins as Yoshida et al. 320 and twice as many ECM proteins as Cui et al. 323 Therefore, our study yielded the greatest number of ECM proteins from glomeruli to date, likely owing to the use of effective ECM enrichment prior to state-of-the-art MS analysis. 68
Figure 2.3 Comparison of the glomerular ECM proteome to published glomerular proteomic datasets. The glomerular ECM proteome identified in this study was compared to other glomerular proteomic studies for which full datasets were available (Cui et al. ; Yoshida et al.). Numbers of proteins in each intersection set of the area proportional Euler diagram are in bold italics. ECM proteins were categorised as basement membrane, other structural ECM or ECM-associated proteins and were coloured and arranged accordingly. Nodes (circles) are labelled with gene names for clarity. ECM proteins detected in any of the three biological replicates reported in this study were included in the comparison with other proteomic datasets; these published datasets each reported one biological replicate. Large node size indicates proteins detected in at least two biological replicates in this study. Michael Randles performed this analysis and generated this figure. 69
2.3.3.4 Creation of a protein interaction network
To visualise the components of glomerular ECM as a network of interacting proteins, the identified proteins were mapped onto a curated protein interaction database to generate an interaction network
(Figure 2.4A). Interestingly, basement membrane and structural ECM proteins were involved in more interactions with other proteins in the network than were ECM-associated proteins (Figure 2.4B).
Topological network analysis confirmed that basement membrane and other structural ECM proteins formed a highly connected “core” sub-network, whereas ECM-associated proteins were less clustered in the network (Figure S2.8, S2.9). These data suggest that structural ECM proteins mediate multiple sets of protein–protein interactions and thus have important roles in the assembly and organisation of glomerular ECM.
2.3.3.5 Localisation of glomerular ECM proteins
Validation of protein expression was performed by searching the Human Protein Atlas (HPA) database 324 for the 144 glomerular ECM proteins identified in this study (Figure 2.5A, 2.5B).
Glomerular immunostaining was not available for 20 proteins, including the known glomerular ECM proteins collagen IV α3, α4, α5, and laminin α5. Protein expression was confirmed for 78 ECM components (63% of proteins with available HPA data) and was reported as negative for 46 ECM components, although there were notable false negatives, including collagen IV α2, collagen XVIII and laminin β1. Immunohistochemistry relies upon antibody specificity, and therefore combining expression data from antibody-based investigations with MS data has the potential to significantly increase the number of protein identifications. Indeed, our proteomic dataset increased the number of ECM proteins detected in the glomerulus by 59% compared to HPA data alone. We further evaluated the HPA database to determine the pattern of immunostaining as glomerular basement membrane (GBM), mesangial matrix, Bowman’s capsule or a combination of compartments. The majority of components were present in more than one ECM compartment (Figure 2.5B), suggesting a common core of protein components between the glomerular ECM compartments.
70
Figure 2.4 Interaction network analysis of human glomerular ECM. A: Protein interaction network constructed from enriched glomerular ECM proteins identified by MS. Nodes (circles) represent proteins and edges (grey lines) represent reported protein–protein interactions. ECM proteins were categorised as basement membrane, other structural ECM or ECM- associated proteins and were coloured and arranged accordingly. Nodes are labelled with gene names for clarity. B: Distribution of degree (number of protein–protein interactions per protein) for basement membrane, other structural ECM or ECM-associated proteins. Data points are shown as circles; 71
outliers are shown as diamonds. **, P < 0.01; NS, P ≥ 0.05. Adam Byron performed this analysis and generated this figure.
Figure 2.5 Localisation of glomerular ECM proteins in the Human Protein Atlas (HPA) database. A: The HPA was searched for glomerular ECM proteins identified in at least two biological replicates in this study. Glomerular immunostaining was reviewed (+, detected; −, not detected; N/A, data not available in the HPA; left panel) and localisation was determined as glomerular basement membrane (GBM), mesangial matrix (MM), Bowman’s capsule (BC) or a combination of these ECM compartments (right panel). B: ECM proteins were categorised as basement membrane, other structural ECM or ECM-associated proteins and were coloured and arranged accordingly. Proteins not detected or without data in glomeruli in the HPA are shown separately (right panel). Nodes (circles) are labelled with gene names for clarity. Node diameter (proteins localised in glomeruli in the HPA only; left panel) is proportional to the intensity of glomerular immunostaining in the HPA. Rachel Lennon and Michael Randles contributed data to this figure parts A and B. 72
2.3.3.6 Co-localisation of novel glomerular ECM proteins
In order to confirm the expression of new glomerular ECM proteins, which were either abundant in our
MS analysis or not detected in the HPA database, we conducted co-localisation studies. Collagen VI and TINAGL1 (also known as TIN -Ag-RP, lipocalin-7, oxidized LDL-responsive gene 2 and androgen- regulated gene 1) were amongst the most abundant proteins detected in this study (Figure 2.2C), and these were validated by Western blotting (Figure 2.2D). Collagen VI was also present in the highly connected sub-network of ECM proteins (Figure S1.8, S1.9). Immunohistochemical and correlation intensity analysis of human renal cortex revealed co-localisation of TINAGL1 with collagen IV α1, which has a mesangial pattern of immunostaining, whereas collagen VI overlapped both with collagen IV α1 and with collagen IV α3 and laminin, which both have a GBM pattern of immunostaining (Figure 2.6A,
C). MS analysis also detected nephronectin and vitronectin. Correlation intensity analysis demonstrated that nephronectin localised in the GBM and mesangial matrix, whereas vitronectin localised in the mesangial matrix alone (Figure 2.6B, D). This objective and quantitative method of co- localisation allows protein expression to be correlated with reliable markers predominating in distinct
ECM compartments. With antibodies of suitable specificity, this approach could be extended to map the full proteome into glomerular ECM compartments. 73
Figure 2.6 Co-localisation of novel and known glomerular ECM proteins. A and B: Immunohistochemistry of human renal cortex was used to examine the co-localisation of collagen VI, TINAGL1, nephronectin and vitronectin with laminin, collagen IV α3 (A) and collagen IV α1 (B). C and D: Bar charts show intensity correlation quotients calculated from immunohistochemistry images (n = 6–10 images for each analysis), demonstrating co-localisation of laminin, collagen VI and nephronectin with collagen IV α3 (C) and co-localisation of collagen VI, TINAGL1, nephronectin and vitronectin with collagen IV α1 (D). Rachel Lennon generated data parts A and B. Alex Carisey generated parts C and D. 74
2.3.3.7 Discussion
We employed an ECM enrichment strategy coupled with an unbiased proteomics approach to confirm the presence of all known glomerular ECM proteins, in addition to many potentially novel components.
Moreover, using immunohistochemistry, we have confirmed the MS identification of TINAGL1, collagen
VI, nephronectin and vitronectin within specific glomerular compartments. These findings demonstrate that the composition of glomerular ECM is far more complex than previously appreciated and imply that many more ECM components may contribute to glomerular development and disease processes.
This unbiased, global approach to define the glomerular ECM proteome demonstrated that this specialised ECM has a core of highly connected extracellular components. In adult human glomeruli,
144 ECM proteins, including all previously described components, were identified. In addition, we found many more structural and regulatory ECM proteins, revealing the complexity of the glomerular
ECM. This proteome is comparable in size to ECM profiles recently reported for vasculature-, lung- and bowel-derived ECMs. 302, 325, 326 In our analysis, several novel glomerular proteins were highly abundant, including collagen VI and TINAGL1. Collagen VI is a structural component that forms microfibrils and is important for muscle function 327, 328 but its role in the glomerulus has not been investigated. Our study demonstrated that collagen VI is present in a highly connected core network of proteins in the glomerular ECM, and it is localised within both the GBM and mesangial matrix. TINAGL1 was also abundantly expressed in glomerular ECM, where it predominantly localised to the mesangial matrix.
TINAGL1 is a glycoprotein and structurally related to TINAG, a tubular basement membrane component, which is the antigenic target in autoimmune anti-TBM disease. 329 TINAGL1 has a proteolytically inactive cathepsin domain, 330 and it has been shown to have a role in angiogenesis; 331 however, its function within the glomerulus is undefined. Both of these highly abundant proteins may have roles in barrier function or glomerular disease.
To assess the relative abundance of ECM proteins, we used two distinct methods of quantification, peptide intensity analysis 332 and spectral counting, and both approaches gave very similar results.
Peptide intensity analysis revealed collagen IV and laminin isoforms to be the most abundant proteins in the glomerular ECM, and both analyses revealed abundant collagen VI and TINAGL1. There were 75
some differences between the methods for the quantification of collagen isoforms, but the stoichiometry of different proteins is difficult to determine by any global proteomic methodology, including antibody-based techniques, which rely on probe immunoaffinity. To perform absolute protein quantification by MS, labelled peptide standards for each protein could be used; however, this would be a significant undertaking for a large protein dataset and would not permit the discovery of unknown
ECM proteins. Therefore, the findings presented in this study provide relative patterns of glomerular
ECM protein enrichment and pave the way for further in-depth study of the ECM components identified.
In comparison to other large-scale glomerular proteomic studies, which have identified up to a total of
1800 proteins, 320, 323 this investigation had the greatest yield of ECM proteins, thus highlighting the importance of sample fractionation for improving protein identification. However, the large-scale requirement for the analysis prohibited the separation of glomerular ECM components, and therefore protein localisation was determined by immunohistochemistry. Whilst laser microdissection studies have the ability to precisely define the anatomical structure for analysis, the total number of proteins identified in the most recent of these studies, using a variety of tissue samples, ranged from 114 to
340, which is fewer proteins than the large-scale glomerular studies and with significantly fewer ECM proteins compared to this study. 312, 333, 334 Future developments in technologies, which couple tissue imaging with improved sensitivity of molecular analysis, may allow direct compositional analysis of
ECM within distinct tissue compartments. 335, 336
The profile of glomerular ECM we have described is likely to represent a snapshot of a highly dynamic extracellular environment, changing under the influence of cellular, physical and chemical environmental cues. Nonetheless, the datasets provide a valuable resource for further investigation of the composition and complexity of glomerular ECM. Our methodology can now be applied to the comparison of glomerular ECMs in vivo in the context of development or disease. Systems -level analysis will be integral to the downstream interrogation of data in order to identify informative, predictive networks of ECM composition and function. Combined with the methodologies and datasets described herein, such analyses will enable the construction of a dynamic glomerular ECM interactome and help to build understanding of how this network may alter during glomerular development and disease.
76
2.3.4 Materials and Methods
2.3.4.1 Antibodies
Monoclonal antibodies used were against actin (clone AC-40; Sigma-Aldrich, Poole, UK), nephronectin
(ab64419; Abcam, Cambridge, UK) TINAGL1 (ab69036; Abcam), vitronectin (ab11591; Abcam) and collagen IV chain-specific antibodies (provided by B. Hudson, Vanderbilt University Medical Center,
Nashville, TN, USA). Polyclonal antibodies used were against pan–collagen IV (ab6586; Abcam), pan- laminin (ab11575; Abcam), pan-collagen VI (ab6588; Abcam), lamin B1 (ab16048; Abcam) and nephrin
(ab58968; Abcam). Secondary antibodies against rabbit IgG conjugated to TRITC and mouse or rat
IgG conjugated to FITC (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) were used for immunofluorescence; secondary antibodies conjugated to Alexa Fluor 680 (Life Technologies,
Paisley, UK) or IRDye 800 (Rockland Immunochemicals, Glibertsville, PA, USA) were used for
Western blotting.
2.3.4.2 Western blotting
See General Materials and Methods.
2.3.4.3 Isolation of human glomeruli
Normal renal cortex from human donor kidneys technically unsuitable for transplantation was used with full ethical approval (reference 06/Q1406/38). Three adult male donors between 37 and 63 years were chosen to reduce the influence of age and sex. Normal renal function was determined by pre- nephrectomy serum creatinine values. At 4 oC, renal cortex (2.5 g) was finely diced and pressed onto a
250- m sieve (Endecotts, London, UK) using a 5-ml syringe plunger. Glomeruli were rinsed through sieves (250- m and 150- m) with cold PBS and separated from tubular fragments by collection on both 150 and 100- m sieves. Retained glomeruli were retrieved into 10 ml PBS and washed a further three times with PBS and interval centrifugation. Glomerular purity was consistently >95% as determined by counting whole glomeruli and non-glomerular fragments using phase-contrast light microscopy.
77
2.3.4.4 Isolation of enriched glomerular ECM
This was adapted from previously published methods 337 and utilised to reduce the complexity of protein samples for MS analysis by removing cellular components and enriching for ECM proteins. All steps were carried out at 4 oC to minimise proteolysis. Pure glomerular isolates from three human kidneys were incubated for 30 minutes in extraction buffer (10 mM Tris, 150 mM NaCl, 1% (v/v) Triton X-100,
25 mM EDTA, 25 g/ml leupeptin, 25 g/ml aprotinin and 0.5 mM AEBSF) to solubilise cellular proteins, and samples were then centrifuged at 14000 × g for 10 minutes to yield fraction 1. The remaining pellet was incubated for 30 minutes in alkaline detergent buffer (20 mM NH 4OH and 0.5%
(v/v) Triton X-100 in PBS) to further solubilise cellular proteins and to disrupt cell–ECM interactions.
Samples were then centrifuged at 14000 × g for 10 minutes to yield fraction 2. The remaining pellet was incubated for 30 minutes in a deoxyribonuclease (DNase) buffer (10 g/ml DNase I (Roche,
Burgess Hill, UK) in PBS) to degrade DNA. The sample was centrifuged at 14000 × g for 10 minutes to yield fraction 3, and the final pellet was re-suspended in reducing sample buffer (50 mM Tris-HCl, pH
6.8, 10% (w/v) glycerol, 4% (w/v) sodium dodecyl sulfate (SDS), 0.004% (w/v) bromophenol blue, 8%
(v/v) β-mercaptoethanol) to yield the ECM fraction. Samples were heat denatured at 70 oC for 20 minutes.
2.3.4.5 MS data acquisition
See General Materials and Methods.
2.3.4.6 MS data analysis
Tandem mass spectra were extracted using extract_msn (Thermo Fisher Scientific) executed in
Mascot Daemon (version 2.2.2; Matrix Science, London, UK). Peak list files were searched against a modified version of the IPI Human database (version 3.70; release date, 4 March 2010) , containing ten additional contaminant and reagent sequences of non-human origin , using Mascot (version 2.2.03;
Matrix Science). 297 Carbamidomethylation of cysteine was set as a fixed modification; oxidation of methionine and hydroxylation of proline and lysine were allowed as variable modifications. Only tryptic peptides were considered, with up to one missed cleavage permitted. Monoisotopic precursor mass values were used, and only doubly and triply charged precursor ions were considered. Mass tolerances 78
for precursor and fragment ions were 0.4 Da and 0.5 Da, respectively. MS datasets were validated using rigorous statistical algorithms at both the peptide and protein level 338, 339 implemented in Scaffold
(version 3.00.06; Proteome Software, Portland, OR, USA). Protein identifications were accepted upon assignment of at least two unique validated peptides with ≥ 90% probability, resulting in ≥ 99% probability at the protein level. These acceptance criteria resulted in an estimated protein false discovery rate of 0.1% for all datasets.
2.3.4.7 MS data access
The MS proteomics data have been deposited to the ProteomeXchange Consortium
(http://proteomecentral.proteomexchange.org ) through the PRoteomics IDEntifications partner
340 repository with the dataset identifier PXD000456.
2.3.4.8 MS data quantification and statistical analysis
MS quantification and proteomic data analyses were performed as previously 341, 342 with modifications as described below. Relative protein abundance was calculated using the unweighted spectral count of a given protein normalised to the total number of spectra observed in the entire sample and to the molecular weight of that protein (normalised spectral count). Mean normalised spectral counts were calculated using data from enriched glomerular ECM samples from three human kidneys. Peptide intensity was used as an alternative method of protein quantification, as described by Silva et al. 343
Orbitrap MS data were entered into Progenesis LCMS (Non Linear Dynamics Ltd, Newcastle upon
Tyne, UK) and aligned. Spectra were extracted using extract_msn (Thermo Fisher Scientific, Waltham,
MA, USA) executed in Mascot Daemon (version 2.4; Matrix Science, London, UK) and imported back into Progenesis to acquire intensity data. Peptides assigned to proteins with fewer than three unique peptides per protein were removed, and median normalised intensity was calculated for all peptides.
Peptides with multiple entries were reduced to the most intense median peptide, and the three peptides with the most intense median values were retained and summed. The sum value is used to compare approximate stoichiometries between proteins in the same sample. 343
79
2.3.4.9 Functional annotation and enrichment analysis
Proteins identified in at least two of the three biological replicates were included for further analysis.
344, 345 Gene ontology (GO) annotations were downloaded using the online resource DAVID. The GO cellular compartment annotation chart (GOTERM_CC_FAT) was selected and the protein list was
346 347 imported into Cytoscape (version 2.8.1) and analysed using the Enrichment Map plugin. The following criteria were used to generate the network: P value < 0.005, FDR (Benjamini–Hochberg) cut- off < 0.01, similarity cut-off > 0.6. The network was then analysed using the Markov Cluster Algorithm
(MCL) to generate distinct clusters. Proteins annotated in the extracellular region cluster were further
1 cross-referenced with the human matrisome project, and cytoplasmic proteins annotated as extracellular region (ACTN1, ACTN2, ACTN4, CALM1, CSNK2B, FLNA, GLIPR2, HINT2, HSPD1,
RNH1, TLN1, TTN, TUB4A4, and VCL) were excluded. The final list of 144 proteins was termed the glomerular ECM proteome, and this was used for further analysis.
2.3.4.10 Protein interaction network analysis
346 Protein interaction network analysis was performed using Cytoscape (version 2.8.1). Proteins identified in at least two biological replicates were mapped onto a merged human interactome built from the Protein Interaction Network Analysis platform Homo sapiens network (release date, 28 June
348 2011) and Mus musculus network (release date, 28 June 2011), the ECM interactions database
MatrixDB (release date, 26 August 2010), 10 and a literature-curated database of integrin-based
349 adhesion–associated proteins. Topological parameters were computed using the NetworkAnalyzer plug-in. 350
2.3.4.11 Immunohistochemistry and image analysis
Formalin-fixed, paraffin-embedded tissue blocks were sectioned at 5 m. Sections were dewaxed and treated with recombinant proteinase K (Roche Diagnostics, IN, USA) for 15 minutes. Sections were blocked with 5% (v/v) donkey serum (Sigma) and 1.5% (v/v) BSA (Sigma) for 30 minutes and with primary antibodies overnight at 4 oC. Sections were washed three times with PBS, incubated with secondary antibodies, mounted with polyvinyl alcohol mounting medium (Fluka 10981, Sigma) and imaged using a BX51 upright microscope (Olympus, Southend on Sea, UK) equipped with a 20× UPlan 80
Fln 0.50 objective and controlled through MetaVue software (Molecular Devices, Wokingham, UK).
Images were collected using a CoolSnap HQ camera (Photometrics, Tucson, AZ, USA) and separate
DAPI/FITC/Cy3 filters (U-MWU2, 41001, 41007a, respectively; Chroma, Olching, Germany) to minimise bleed-through between the different channels. Images were processed and analysed using
Fiji/ImageJ software (version 1.46r; National Institutes of Health, Bethesda, MD, USA). Raw images were subjected to signal re-scaling using linear transformation for display in the figures. For calculation of the Pearson’s correlation coefficient, a region of interest was drawn and a threshold was set to restrict analysis to single glomeruli. The coefficient was measured using the Intensity Correlation
Analysis (ICA) plugin for Fiji/ImageJ 351 and the subsequent plotting step were performed using
MATLAB (version R2012a; MathWorks, Natick, MA, USA).
2.3.4.12 Analysis of Human Protein Atlas immunohistochemistry data
The Human Protein Atlas (HPA) (version 11.0) was searched for the 144 ECM proteins identified in this study, and glomerular immunohistochemistry was reviewed. Proteins were categorised as being positive (strong, moderate, weak), negative or not available according to the HPA report. Proteins with positive immunostaining were further allocated to one or more ECM compartments (GBM, mesangial matrix, Bowman’s capsule) according to our assessment of the histological pattern of staining.
Negative immunostaining controls were not present for analysis in the HPA database, limiting our independent evaluation of protein expression. In addition, the HPA report indicated protein expression in cells of the glomerulus, rather than ECM compartments. Both of these factors could influence false negative or false positive reporting.
2.3.4.13 Statistical analysis
All measurements are shown as mean ± standard error of the mean. Box plots indicate 25 th and 75 th percentiles (lower and upper bounds, respectively), 1.5× interquartile range (whiskers) and median
(black line). Numbers of protein–protein interactions were compared using Kruskal–Wallis one-way analysis of variance tests with post-hoc Bonferroni correction. GO enrichment analyses were compared using modified Fisher’s exact tests with Benjamini–Hochberg correction. P values <0.05 were deemed significant. 81
2.3.5 Acknowledgments
This work was supported by a Wellcome Trust Intermediate Fellowship award to R.L. (ref: 090006) and
Wellcome Trust grant 092015 to M.J.H. R.Z. is supported by VA Merit Award 1I01BX002196-01,
DK075594, DK069221, DK083187 and an American Heart Association Established Investigator Award.
The mass spectrometer and microscopes used in this study were purchased with grants from the
Biotechnology and Biological Sciences Research Council, Wellcome Trust and the University of
Manchester Strategic Fund. Mass spectrometry was performed in the Biological Mass Spectrometry
Core Facility, Faculty of Life Sciences, University of Manchester, and we thank Stacey Warwood for advice and technical support. We thank Julian Selley for bioinformatic support. Microscopy was performed in the Bioimaging Core Facility, Faculty of Life Sciences, University of Manchester. The collagen IV chain–specific antibodies were kindly provided by the Billy Hudson research group,
Department of Medicine, Vanderbilt Medical Center.
2.3.6 Statement of competing financial interests
No competing interests . 82
2.3.7 Supplementary data
Table 2.1 Human glomerular ECM proteome
Uniprot ID Gene name Alias Abundance (nSC) Category
O00468 AGRN Agrin 3.075 Glycoprotein
P39059 COL15A1 Collagen alpha-1 (XV) 0.052 Collagen chain
P39060 COL18A1 Collagen alpha-1 (XVIII) 5.23 Collagen chain
P02462 COL4A1 Collagen alpha-1 (IV) 12.515 Collagen chain
P08572 COL4A2 Collagen alpha-2 (IV) 17.709 Collagen chain
Q01955 COL4A3 Collagen alpha-3 (IV) 8.155 Collagen chain
P53420 COL4A4 Collagen alpha-4 (IV) 8.961 Collagen chain
P29400 COL4A5 Collagen alpha-5 (IV) 3.778 Collagen chain
Q14031 COL4A6 Collagen alpha-6 (IV) 0.589 Collagen chain
P23142 FBLN1 Fibulin-1 0.029 Glycoprotein
P35555 FBN1 Fibrillin-1 1.403 Glycoprotein
P02751 FN1 Fibronectin 1.893 Glycoprotein
Q86XX4 FRAS1 Extracellular matrix protein 0.039 Glycoprotein FRAS1
Q96RW7 HMCN1 Hemicentin-1 0.007 Glycoprotein
P98160 HSPG2 Perlecan 7.012 Proteoglycan
P24043 LAMA2 laminin subunit alpha-2 0.043 Glycoprotein
O15230 LAMA5 laminin subunit alpha-5 9.482 Glycoprotein
P07942 LAMB1 laminin subunit beta-1 0.977 Glycoprotein
P55268 LAMB2 laminin subunit beta-2 14.745 Glycoprotein
P11047 LAMC1 laminin subunit gamma-1 8.818 Glycoprotein
P14543 NID1 Nidogen-1 10.247 Glycoprotein
Q14112 NID2 Nidogen-2 1.249 Glycoprotein
Q0UJW2 TINAG Tubulointerstitial nephritis 12.873 Glycoprotein antigen 83
Q6PCB0 VWA1 von Willebrand factor A 1.171 Glycoprotein domain-containing protein 1
Q9BXN1 ASPN Asporin 1.199 Proteoglycan
P21810 BGN Biglycan 2.268 Proteoglycan
Q99715 COL12A1 Collagen alpha-1 (XII) 0.073 Collagen chain
P02452 COL1A1 Collagen alpha-1 (I) chain 0.253 Collagen
P08123 COL1A2 Collagen alpha-2 (I) chain 0.594 Collagen
P02461 COL3A1 Collagen alpha-1 (III) chain 0.143 Collagen
P12109 COL6A1 Collagen alpha-1 (VI) 12.685 Collagen chain
P12110 COL6A2 Collagen alpha-2 (VI) 5.889 Collagen chain
P12111 COL6A3 Collagen alpha-3 (VI) 11.145 Collagen chain
P07585 DCN Decorin 0.067 Proteoglycan
Q07507 DPT Dermatopontin 0.397 Glycoprotein
Q9Y6C2 EMILIN1 EMILIN-1 0.699 Glycoprotein
P02671 FGA Fibrinogen alpha chain 0.954 Glycoprotein
P02675 FGB Fibrinogen beta chain 1.965 Glycoprotein
P02679 FGG Fibrinogen gamma chain 2.002 Glycoprotein
P08493 MGP Matrix Gla protein 0.557 Glycoprotein
Q6UX19 NPNT Nephronectin 3.707 Glycoprotein
B1ALD8 POSTN Periostin, osteoblast 0.191 Glycoprotein specific factor
Q8IVN8 RPESP RPE-spondin 1.364 Glycoprotein
Q15582 TGFBI Transforming growth 0.369 Glycoprotein factor-beta-induced protein
Q9GZM7 TINAGL1 Tubulointerstitial nephritis 20.01 Glycoprotein antigen-like 1
P04004 VTN Vitronectin 9.015 Glycoprotein
Q9BVH8 VWA5B2 von Willebrand factor A 0.024 Glycoprotein domain-containing protein 5B2
E2QRD0 VWA8 von Willebrand factor A 0.076 Glycoprotein domain-containing protein 8
P01019 AGT Angiotensinogen 0.327 Protease inhibitor
P02768 ALB Serum albumin 11.448 Secreted
P02760 AMBP protein AMBP 8.382 Secreted 84
E9PDK5 ANXA11 Annexin A11 0.371 Annexin
P07355 ANXA2 Annexin A2 1.621 Annexin
P08758 ANXA5 Annexin A5 0.283 Annexin
P20073 ANXA7 Annexin A7 0.062 Annexin
P02743 APCS Serum amyloid P- 19.195 Secreted component
P02647 APOA1 Apolipoprotein A-I 4.906 Apolipoprotein
P06727 APOA4 Apolipoprotein A-IV 5.896 Apolipoprotein
Q6Q788 APOA5 Apolipoprotein A-V 0.055 Apolipoprotein
B0YIW2 APOC3 Apolipoprotein C-III variant 3.204 Apolipoprotein 1
P02649 APOE Apolipoprotein E 16.039 Apolipoprotein
P20160 AZU1 Azurocidin 0.194 Protease
P61769 B2M Beta-2-microglobulin 2.157 Secreted
Q6UXH0 C19orf80 Hepatocellular carcinoma- 0.256 Secreted associated gene TD26
P02747 C1QC Complement C1q 0.352 Complement subcomponent subunit C
P01024 C3 Complement C3 4.768 Complement (Fragment)
P01031 C5 Complement C5 0.559 Complement
P13671 C6 Complement component 6 0.458 Complement precursor
P10643 C7 Complement component 0.445 Complement C7
P07357 C8A Complement component 0.151 Complement C8 alpha chain
P07358 C8B Complement component 0.276 Complement C8 beta chain
P07360 C8G Complement component 6.669 Complement C8 gamma chain
Q8WUY1 C8orf55 UPF0670 protein C8orf55 0.134 Secreted
P02748 C9 Complement component 2.713 Complement C9
P00918 CA2 Carbonic anhydrase 2 1.506 Enzyme
P13987 CD59 CD59 glycoprotein 3.179 Cell surface
Q5VYL6 CFHR5 Complement factor H- 0.477 Complement related 5
P10909 CLU Clusterin 4.587 Secreted 85
P09543 CNP CNPI of 2',3'-cyclic- 0.743 Enzyme nucleotide 3'- phosphodiesterase
P53621 COPA Coatomer subunit alpha 0.128 Secreted
Q96IY4 CPB2 Carboxypeptidase B2 0.198 Enzyme
P01034 CST3 Cystatin-C 1.315 Protease inhibitor
P07339 CTSD Cathepsin D 0.11 Protease
P08311 CTSG Cathepsin G 1.188 Protease
A5JHP3 DCD Dermcidin isoform 2 0.075 Secreted
P59665 DEFA1 Neutrophil defensin 1 15.387 Secreted
Q03001 DST Dystonin 0.002 Plakin
P08246 ELANE Neutrophil elastase 0.779 Protease
P00740 F9 Coagulation factor IX 0.821 Coagulation
Q5D862 FLG2 Filaggrin-2 0.041 Cytoskeleton
P17948 FLT1 Flt1 of Vascular endothelial 0.037 Secreted growth factor receptor 1
Q5SZK8 FREM2 FRAS1-related 0.071 Cell surface extracellular matrix protein 2
Q92820 GGH Gamma-glutamyl 0.146 Enzyme hydrolase
Q5JWF2 GNAS Guanine nucleotide- 0.262 Cell surface binding protein G (s) subunit alpha
P06744 GPI Glucose-6-phosphate 0.482 Enzyme isomerase
P22352 GPX3 Glutathione peroxidase 3 8.232 Enzyme
P06396 GSN Gelsolin 1.039 Cytoskeleton
F5GXA6 HLAC HLA class I 0.267 Cell surface histocompatibility antigen, Cw-7 alpha chain
P00738 HP Haptoglobin 0.163 Secreted
P02790 HPX Hemopexin 1.231 Secreted
Q53GQ0 HSD17B12 Estradiol 17-beta- 0.17 Enzyme dehydrogenase 12
P01861 IGHG4 Ig gamma-4 chain C region 1.806 Secreted
P58166 INHBE Inhibin beta E chain 0.536 Secreted
P19827 ITIH1 Inter-alpha-trypsin inhibitor 0.032 Protease inhibitor heavy chain H1
Q06033 ITIH3 Inter-alpha-trypsin inhibitor 0.035 Protease inhibitor heavy chain H3 86
P01042 KNG1 Kininogen-1 0.674 Protease inhibitor
P09382 LGALS1 Galectin-1 4.983 Secreted
P17931 LGALS3 Galectin-3 2.965 Secreted
Q08380 LGAS3BP Galectin-3-binding protein 0.55 Secreted
P11150 LIPC Hepatic triacylglycerol 0.265 Enzyme lipase
P98164 LRP2 Low-density lipoprotein 0.101 Cell surface receptor-related protein 2 (megalin)
P16126 LYZ Lysozyme C 1.547 Secreted
P14780 MMP9 Matrix metalloprotein ase-9 0.111 Protease
P05164 MPO Myeloperoxidase 0.868 Enzyme
Q8WX17 MUC16 Mucin-16 0.001 Cell surface
P30086 PEBP1 Phosphatidylethanolamine- 0.506 Protease inhibitor binding protein 1
P62937 PPIA Peptidyl-prolyl cis-trans 4.879 Enzyme isomerase A
P15888 PRELP Prolargin 0.198 Secreted
Q92954 PRG4 Proteoglycan 4 0.016 Secreted
O95084 PRSS23 Serine protease 23 0.299 Protease
Q5SQ09 PTGDS Prostaglandin D2 synthase 0.564 Enzyme 21kDa
Q5VY30 RBP4 Retinol binding protein 4, 20.718 Secreted plasma
Q86YZ3 S100A18 Hornerin 0.012 Cytoplasmic
P06703 S100A6 Protein S100-A6 1.59 Cell surface
P05109 S100A8 Protein S100-A8 3.701 Secreted
P02735 SAA1 Serum amyloid A protein 0.675 Secreted
Q13214 SEMA3B Semaphorin-3B 0.032 Secreted
P01009 SERPINA1 Alpha-1-antitrypsin 5.59 Protease inhibitor
P01011 SERPINA3 Alpha-1-antichymotrypsin 0.481 Protease inhibitor
P05154 SERPINA5 Plasma serine protease 2.071 Protease inhibitor inhibitor
P01008 SERPINC1 Antithrombin-III 0.583 Protease inhibitor
P07093 SERPINE2 Glia-derived nexin 0.164 Protease inhibitor
P05155 SERPING1 Plasma protease C1 0.89 Protease inhibitor inhibitor
P50454 SERPINH1 Serpin H1 0.401 Protease inhibitor 87
Q9Y5W8 SNX13 Small inducible cytokine 0.782 Cytoplasmic B14 precursor
Q00796 SORD Sorbitol dehydrogenase 2.804 Enzyme
Q13103 SPP2 Secreted phosphoprotein 1.527 Secreted 24
P02787 TF Serotransferrin 0.096 Secreted
P21980 TGM2 protein -glutamine gamma- 1.302 Enzyme glutamyltransferase 2
P35625 TIMP3 Metalloprotein ase inhibitor 0.928 Protease inhibitor 3
O75888 TNFSF13 Gamma of Tumor necrosis 0.135 Secreted factor ligand superfamily member 13
P02766 TTR Transthyretin 3.88 Secreted
E9PEA4 UMOD Uromodulin 26.901 Secreted
A8K0G1 WNT7B Protein Wnt 0.058 Secreted 88
Figure S2.7 Gene ontology (GO) enrichment analysis of the full MS dataset. Nodes (circles) represent enriched GO terms and edges (grey lines) represent overlap of proteins between GO terms. Node colour indicates the significance of GO term enrichment ; node diameter is proportional to the number of proteins assigned to each GO term; edge weight is proportional to the number of proteins shared between connected GO terms. The network from Figure 2.2 is annotated (left panel); nodes are numbered according to their GO term (right panel). Michael Randles generated data for this figure. 89
Figure S2.8 Glomerular ECM interaction network analysis. Proteins identified by MS and classified as extracellular region according to Gene Ontology annotation were converted to a protein-protein interaction network model. The interaction network was clustered, and topological parameters were computed. Self-interactions were excluded from the analysis. Nodes are coloured according to their clustering coefficient, and node diameter is proportional to number of interaction partners (degree). Nodes are labelled with gene names. Proteins classified as basement membrane are displayed with red node borders; other structural ECM proteins are displayed with orange node borders; and ECM-associated proteins are displayed with green node borders. Adam Byron generated data for this figure. 90
Figure S2.9 The most connected proteins in the ECM interaction network. The ten proteins with the highest number of neighbours (degree) in the glomerular ECM interaction network are plotted as a bar chart. The most connected proteins (known as hubs), which incorporated the top 20% of the degree distribution, are indicated by dark blue bars. Self-interactions were excluded from the analysis. Proteins are labelled with gene names. Adam Byron generated data for this figure . 91
3 Glomerular cell-derived matrices
3.1 Introduction
The glomerular basement membrane (GBM) is a key extracellular matrix (ECM) structure within the glomerulus. Perturbation of GBM structure leads to loss of selective glomerular filtration. Currently, there is limited knowledge about how the GBM is formed during development, regulated in health and disrupted in disease. Examination of the GBM has been limited to candidate-based approaches, as it is currently technically unfeasible to separate the GBM and mesangial ECM compartments during ECM isolation from the glomerulus. It is established that the GBM is synthesised jointly by podocytes and glomerular endothelial cells (GECs), with both cell types contributing to the laminin networks. 352 In contrast, the mature collagen IV network ( α3α4α5) is produced solely by podocytes. 353 However, the
GBM contains more than just type IV collagen and laminins, and the relative contributions of podocytes and GECs to the plethora of other ECM proteins within the GBM are unknown. Establishing the origins of ECM proteins within the GBM in a global manner will enable better hypothesis generation regarding
ECM regulation within the glomerulus. 92
3.2 Statements
Author contributions to data generation and analysis presented as figures of this paper are indicated in figure legends.
Rachel Lennon, Adam Byron, Michael J. Randles and Jonathan D. Humphries planned the study and designed experiments. Roy Zent and Martin J. Humphries contributed to the study design.
Michael J. Randles characterised podocyte and glomerular endothelial cells with transmission electron microscopy, performed coculture experiments, assessed podocyte and glomerular endothelial cell viability in coculture, performed conditioned media experiments, quantified ECM deposition using transmission electron microscopy, and assessed ECM and cell junctions in coculture with immunofluorescence.
Rachel Lennon isolated enriched podocyte, GEC and coculture ECM fractions, performed Western blotting and immunofluorescence imaging and performed hierarchical clustering analysis. Adam Byron built the database of human ECM protein-protein interactions and performed network topology analysis. Jonathan D. Humphries performed QSpec analysis and Gene Ontology enrichment analysis.
Aleksandr Mironov aided with electron microscopy data acquisition. Hellyeh Hamidi generated the
GFP-podocyte cell line by lentiviral transduction. Shelley Harris performed RT-PCR and q-PCR. Peter
W. Mathieson , Moin A. Saleem and Simon C. Satchell provided human podocyte and glomerular endothelial cell lines.
The manuscript is written in the style of a Journal of the American Society of Nephrology research article by Rachel Lennon, Adam Byron, Michael J. Randles, and Jonathan D. Humphries. Roy Zent and
Martin J. Humphries contributed to the preparation of the manuscript and the manuscript was critically assessed by all authors. Author guidelines restrict the size of the article to 3000 words or less
(excluding title page, methods, figure legends, tables, and references) .
93
3.3 Glomerular cell crosstalk influences the composition and
assembly of extracellular matrix
† † Adam Byron ,1, *, Michael J. Randles ,1,2 , Jonathan D. Humphries ,1 Aleksandr Mironov ,1 Hellyeh
Hamidi, 1 Shelley Harris, 2 Peter W. Mathieson ,3 Moin A. Saleem ,3 Simon C. Satchell ,3 Roy Zent ,4,5
Martin J. Humphries ,1 and Rachel Lennon .1,2
1Wellcome Trust Centre for Cell-Matrix Research, Faculty of Life Sciences, University of Manchester,
Manchester, UK; 2Faculty of Medical and Human Sciences, University of Manchester, Manchester, UK;
3Academic Renal Unit, Faculty of Medicine and Dentistry, University of Bristol, Bristol, UK; 4Department of Medicine, Vanderbilt University Medical Center , Nashville, TN, USA; and 5Veterans Affairs Hospital,
Nashville, TN, USA .
*Present address: Edinburgh Cancer Research UK Centre, Institute of Genetics and Molecular
Medicine, University of Edinburgh, Edinburgh, UK.
† These authors contributed equally to this work.
Running title: Glomerular extracellular matrix
Corresponding author:
Dr Rachel Lennon, Wellcome Trust Centre for Cell-Matrix Research , Michael Smith Building, University of Manchester, Manchester M13 9PT, UK .
Phone: 0044 (0) 161 2755498. Fax: 0044 (0) 161 2755082 .
Email: [email protected] 94
3.3.1 Abstract
The glomerular basement membrane (GBM) is a specialised extracellular matrix (ECM) compartment within the glomerulus and is known to contain tissue-restricted isoforms of collagen IV and laminin. It is integral to the capillary wall and therefore functionally linked to glomerular filtration. While the composition of the GBM has been investigated with global and candidate-based approaches, the relative contributions of glomerular cell types to the production of ECM is not well understood. To enable the characterisation of specific cellular contributions to the GBM, we analysed ECM isolated from podocytes and glomerular endothelial cells in vitro using mass spectrometry–based proteomics.
These analyses identified cell type–specific differences in ECM composition, indicating distinct contributions to glomerular ECM assembly. Coculture of podocytes and endothelial cells resulted in an altered composition and organisation of ECM compared to monoculture ECMs from single cell types, suggesting a role for cell–cell crosstalk in the production of glomerular ECM. This was supported by the identification of basement membrane–like ECM deposition between cocultured cells using electron microscopy. Importantly, compared to monoculture, the coculture ECM proteome better resembled a tissue-derived glomerular ECM dataset, indicating its relevance to GBM in vivo . Protein network analyses revealed a common core of 35 highly connected structural ECM proteins that may be important for glomerular ECM assembly. Overall, these findings demonstrate the complexity of the glomerular ECM and suggest that both ECM composition and organisation are context dependent. 95
3.3.2 Introduction
The glomerular basement membrane (GBM) is integral to the glomerular capillary wall, and it directly contributes to barrier function .354, 355 The ultrastructural appearance of the GBM shows it is unusually thick (300–350 nm) as a result of two basement membranes fusing during glomerular development. 356
This specialised extracellular matrix (ECM) supports two distinct cell types: glomerular endothelial cells
(GEC) on the inner capillary wall and podocytes, which form a mesh-like covering of the outer surface.
However , there is limited understanding about the relative contribution of each cell type to the regulation of ECM synthesis and organisation in the GBM .
Some unique ECM isoforms, essential for normal barrier function, characterise the mature glomerulus.
For example, collagen IV networks are predominantly composed of α1 and α2 chains; however, at the capillary-loop stage of glomerular development there is an isoform switch to the expression of the collagen IV α3α4α5 network in the GBM. 357, 358 This network is thought to provide the tensile strength necessary to withstand filtration forces. 358, 359 Similar changes in laminin isoforms have been reported, switching from a combination of laminin α1β1γ1 (laminin-111) and 511 to a predominance of laminin-
521 in the mature GBM. 360 The importance of glomerular ECM composition for normal barrier function is highlighted by human genetic mutations, which result in the absence of the collagen IV α3α4α5 network in Alport syndrome 307 and the absence of laminin-521 in Pierson syndrome. 308 However, there is a much wider spectrum of inherited and acquired glomerular disease with associated abnormalities in the GBM, which are not explained by defects in collagen IV or laminin. 361, 362 Therefore, building understanding of ECM composition in an unbiased manner could help to identify additional genetic defects and to appreciate mechanisms of ECM regulation in the glomerulus.
These questions can be initially addressed with studies in vitro since cell-based investigations allow the dissection of molecular pathways in a manner not feasible in animal studies. Conditionally immortalised human glomerular cells have been derived to facilitate these investigations; 363, 364 however, their ability to synthesise and organise ECM has not been fully investigated. In vivo , these cells are highly specialised, and ex-vivo they may lose critical paracrine and environmental cues required to maintain a differentiated state. 365 To test this hypothesis and to inform mechanisms of ECM regulation in the 96
glomerulus, global investigation of cell-derived ECM represents a powerful, unbiased analytical approach.
Using an ECM enrichment strategy coupled to mass spectrometry (MS)–based proteomics, we recently described a glomerular ECM proteome derived from isolated human glomeruli. 366 To extend this global analysis, we analysed cell-derived ECM from GEC and podocytes to determine their relative contribution to ECM production and regulation in the glomerulus.
3.3.3 Results
3.3.3.1 Human glomerular cells synthesise and organise ECM in vitro
To investigate ECM production by glomerular cells either side of the GBM, we prepared monocultures of human GEC and podocytes in vitro . Both cell types synthesised and deposited ECM (Figure 3.1A), and this did not appear disrupted by the removal of cells prior to ECM extraction (Figure 3.1A; right panels). Ultrastructural analysis of both glomerular cell types revealed a thin, basal layer of ECM
(Figure 3.1B). In addition, GEC and podocytes exhibited distinct morphology (Figure 3.1B, C). GEC had more caveolae, which were visible as small (50–100-nm diameter) invaginations of the plasma membrane (Figure 3.1B, panel i), and which are known to be particularly abundant in endothelial cells. 367 GEC also had longer focal adhesions, observed as electron-dense sites of cell contact with the
ECM (Figure 3.1B, panel ii) and also had more glycogen granules (Figure 3.1B, panel iii). In contrast, podocytes had wider cell–cell junctions (29 ± 3.4 nm) (Figure 3.1B, panel v), more lysosomes (Figure
3.1B, panel vi), and greater cell width (6.1 ± 0.49 m) (Figure 3.1B, C). 97
Figure 3.1 Glomerular cells synthesise and organise ECM in vitro . A: Immunocytochemistry of human GEC and podocytes demonstrated expression of laminin and collagen IV in vitro . These ECM proteins remained detectable following removal of either cell type (denuded ECM; right panel). B: Ultrastructural analysis of GEC and podocytes revealed morphological differences. GEC had more caveolae (arrows; i) and glycogen deposits (arrows; iii), whereas podocytes had smaller focal adhesions (asterisks; iv, cf. ii), wider cell junctions (arrow; v), more lysosomes (asterisks; vi) and greater cell width. Both cell types deposited a thin basal layer of ECM adjacent to focal adhesions. C: Quantification of morphological differences between GEC (n = 11) and podocytes (n = 31 ). Graphs display mean ± SEM and results of statistical analysis using unpaired 98
Student’s t-tests. Rachel Lennon generated data for this figure part A and B. Michael Randles generated data for this figure part B and C. 3.3.3.2 Glomerular cells in vitro exhibit differential ECM protein synthesis
To collect ECM proteins for proteomic analysis, a fractionation workflow was adapted from previous studies (Figure 3.2A). 342, 368 Cell-derived ECMs were extracted from both GEC and podocyte monocultures and analysed by MS. We identified 127 extracellular proteins in GEC samples and 142 extracellular proteins in podocyte samples (Tables S3.1 and S3.2). There was considerable (55%) overlap between proteins detected in GEC and podocyte ECMs, but cell-type specific proteins were also identified (Figure 3.2B). Furthermore, 44% of proteins in the recently described in vivo glomerular
366 ECM proteome were identified in cell-derived ECMs (Figure 3.2B). Statistical analysis of all identified proteins in the cell-derived ECM samples revealed differential ECM protein synthesis between GEC and podocytes (Figure 3.2C). While Gene Ontology (GO) enrichment analysis highlighted cell-type specific differences between the ECMs. Specifically, there was significant enrichment of vascular development processes in GEC ECM, whereas cell adhesion processes were enriched in podocyte
ECM (Figure 3.2D). These data indicate a functional distinction between GEC and podocyte ECMs.
Moreover, the expected association of GEC with endothelial cellular functions suggests that this analysis has the capability to identify proteins with tissue-specific functions.
In contrast to the compositional differences highlighted by the GO analysis, previously identified glomerular ECM components were detected to a similar extent in ECMs from both cell types (Figure
3.2B), suggesting that the ECMs are qualitatively closely related. To examine putative interactions between the identified ECM proteins, interaction networks were generated from a curated protein interaction database. ECM protein interaction networks had similar topologies, with highly connected subnetworks of basement membrane and other structural ECM proteins (Figures 3.3, S3.9, S3.10).
Interestingly, basement membrane proteins had a higher number of protein–protein interactions than
ECM-associated proteins, suggesting that these molecules play key roles in glomerular ECM organisation (Figure 3.3B, 3.3D). These findings are very similar to the tissue-derived glomerular ECM interactome, which identified a core of highly connected structural components. 366 Overall, these data suggest that qualitative differences between GEC and podocyte ECMs do not affect the core organisation of the ECM. 99
Figure 3.2 GEC and podocytes in vitro have differential ECM composition. A: A proteomic workflow for the isolation of ECM from glomerular cells in vitro . B: Cell-derived ECM fractions were prepared from human GECs (n = 3) and human podocytes (n = 3) and analysed by MS. The Venn diagram compares identified ECM proteins with a glomerular ECM dataset, derived from isolated human glomeruli. Percentage overlaps of set pairs are indicated. C: QSpec statistical analysis of all GEC and podocyte proteins identified by MS demonstrates the differential relative abundance of ECM proteins between both datasets. Blue boxes indicate zones of significant differential relative abundance (false discovery rate <5%). D: Gene Ontology (GO) enrichment analysis of GEC and podocyte datasets revealed enrichment of distinct biological process terms. The top three biological processes for each cell type are shown. Rachel Lennon generated data for this figure parts A and B. Jonathan Humphries generated data for this figure parts C and D.
100
101
Figure 3.3 GEC and podocyte ECM interaction networks. A, C: Protein interaction networks constructed from GEC (A) or podocyte (C) ECM proteins identified by MS. ECM proteins were classified as basement membrane, other structural ECM or ECM- associated proteins, and were coloured and arranged accordingly. Nodes are labelled with gene names, and relative node positions are preserved in both interaction networks. B, D: Distributions of degree (number of protein -protein interactions per protein) for basement membrane, other structural ECM or ECM-associated proteins for the GEC (B) or podocyte (D) ECM interaction networks. Data points are shown as circles; outliers are shown as diamonds. **, P < 0.01; *, P < 0.05; NS ≥ 0.05. Adam Byron generated data for this figure.
3.3.3.3 Glomerular cell-derived ECM resembles a developmental phenotype
To assess quantitative differences between the compositions of the ECMs, predicted GBM components were analysed by unsupervised hierarchical clustering. A striking enrichment of collagen
IV chains was identified in glomerular ECM as compared to cell-derived ECMs (Figure 3.4A). We found chains of both isoforms in glomerular ECM ,366 with a predominance of α1 and α2 chains, which may reflect the mesangial ECM contribution. 369 In cell-derived ECMs, we identified α1 and α2 chains by MS, but minimal detection of α3, α4, or α5 chains (Figure 3.4A). These data were confirmed by Western blotting of ECM samples (Figure 3.4B). Taken together, these findings suggest that ECM derived from glomerular cells in culture resembles a developmental ECM, 370 and this concept is further supported by the low relative abundance of laminin β2, a predominant isoform in the mature glomerulus (Figure
3.4A). Whilst RT-PCR analysis revealed that both glomerular cell types expressed mRNA for COL4A1 ,
3, 4, and 5 (Figure 3.4C), q -PCR demonstrated the relative abundance of mRNA was significantly less than the amount isolated from human renal cortex (Figure 3.4D, 3.4E). It is conceivable that glomerular cells have the capacity to synthesise both the collagen IV α1 α 2 α 1 chains and the mature α3 α 4 α 5 chains , but require the appropriate environmental context. 102
103
Figure 3.4 Glomerular ECM in vitro resembles a developmental phenotype. A: Hierarchical clustering analysis of the abundance of basement membrane proteins identified in GEC, podocyte and glomerular ECMs identified a cluster of proteins with increased relative abundance in glomerular ECM. The heat map displays relative protein abundance (normalised spectral count), and the associated dendrogram displays clustering on the basis of uncentered Pearson correlation. Proteins are labelled with gene names for clarity. B: The low abundance of collagen IV α3, α4 and α5 in GEC and podocyte ECMs was confirmed by Western blotting (WB) using chain-specific antibodies to detect monomers (~25 kD) and dimers (~50 kD). C: RT-PCR of RNA pellets from GEC and podocytes confirmed expression of COL4A1, COL4A3, COL4A4 and COL4A5 transcripts. M, molecular weight marker. D, E: Quantitative PCR also confirmed the expression of COL4A1 and COL4A3 in podocytes and GECs. The relative mRNA abundance was calculated and compared with mRNA extracted from human fibroblasts and human renal cortex. With the lowest mRNA expression, human fibroblasts were used for normalisation, and glyceraldehyde-3-phosphate dehydrogenase was used as the endogenous mRNA control. These data represent n=3 replicates. Rachel Lennon generated data parts A and B. Shelley Harris generated data part C, D and E.
3.3.3.4 Glomerular cell coculture results in altered ECM organisation
We hypothesised that the developmental phenotype could be due to a lack of functional interaction between cell types. To investigate the role of cell–cell crosstalk in ECM synthesis, GEC and podocytes were grown in coculture. Cells were viable in coculture and over 14 days the proportion of each cell type did not alter significantly (Figure S3.11). We found that cells synthesised ECM (Figure 3.5A,
Figure S3.12) and ultrastructural analysis revealed a basal layer of ECM (Figure 3.5B, panel i), as previously seen in monoculture (Figure 3.1B). In addition, ECM was deposited between cells in coculture (Figure 3.5Bii–iv), and this organisation resembled a basement membrane, with parallel bundles of ECM proteins. Our phenotypic characterisation of cells in monoculture (Figure 3.1B, C) enabled the identification of ECM between adjacent GEC and podocytes as well as adjacent cells of the same type. Quantification of ECM thickness demonstrated significantly more ECM between cells in coculture compared to monoculture (Figure 3.5Bv). Furthermore, the thickness of the ECM between cocultured cells was in a range comparable to the thickness of the GBM in vivo .371, 372 To investigate whether GEC–podocyte contact or a soluble mediator was required for this basement membrane–like assembly of ECM, we cultured monocultures in conditioned medium from either the same or the other cell type (Figure 3.5C) . Interestingly, ECM was deposited in significant amounts between GEC monocultures in podocyte-conditioned medium (Figure 3.5Civ) , and this increase was significant compared to control-conditioned medium (Figure 3.5Cv). These data suggest that a secreted molecule released by podocytes contributes to the organisation of ECM between glomerular cells. 104
Figure 3.5 Glomerular cell coculture is associated with altered ECM organisation . A: GEC and podocytes were combined in cell suspension and seeded as cocultures. The ECM proteins collagen IV and laminin were detected by immunocytochemistry. B: Ultrastructural analysis of cocultured GEC and podocytes revealed a striking difference in ECM organisation compared to monocultured cells. Cocultured cells deposited basal ECM (i) and, unlike monoculture, deposited ECM between cells, as indicated by asterisks (i–iv). Based on characterisation of monoculture cells, this was not restricted to podocytes adjacent to GEC (ii) but was also evident with adjacent cells of the same type (iv). Quantification of ECM deposited between cells (v). C: Podocytes in GEC or control conditioned media and GECs in control conditioned media did not exhibit ECM deposition or organisation between cells (i–iii); GECs cultured in podocyte conditioned media deposited and organised ECM between cells (iv). Quantification of ECM deposited between cells (v). Pod CM, podocyte conditioned medium; GEC CM, GEC conditioned medium. Data points are shown as circles, n = 15-29. **, P < 0.0001; NS, P ≥ 0.05. Rachel Lennon generated data for this figure part A and B i-iv. Michael Randles generated data for this figure parts B v and C i-v. 105
3.3.3.5 Glomerular cell crosstalk alters ECM composition
To capture basal ECM in addition to the ECM deposited between cells, coculture ECM was extracted using a modified glomerular fractionation workflow. Samples were analysed by MS, which identified
123 extracellular proteins (Table S3.3). Once again, network analysis identified a highly connected subnetwork of basement membrane proteins (Figure 3.6, S3.13), indicating that the interconnected network topology of core structural components of the glomerular ECM was maintained in coculture.
Indeed basement membrane proteins were similarly identified in coculture and monoculture ECMs
(Figure 3.7A), but more differences were observed in other structural ECM proteins (Figure 3.7B) and
ECM-associated proteins (Figure 3.7C). We did not observe an increase in the abundance of mature collagen IV chains α3, α4, and α5 in ECM isolated from coculture (Figure 3.7A), a finding confirmed by
Western blotting (Figure 3.7D). However, collagen VI, which was abundant in tissue-derived glomerular
ECM 366 and absent in monoculture ECM, was enriched in coculture ECM (Figure 3.7B, 3.7D). In contrast, tenascin C, which was abundant in podocyte ECM, was less abundant in coculture and absent in glomerular ECM (Figure 3.7A, 3.7D), and CYR61, an inducer of angiogenesis, 373 was most abundant in GEC ECM, but less abundant in coculture and glomerular ECMs (Figure 3.7B, 3.7D). In summary, glomerular cell coculture results in altered ECM organisation and composition as compared to monoculture, regulating the synthesis and incorporation of ECM molecules in both a positive
(collagen VI) and negative (tenascin C, CYR61) manner. 106
Figure 3.6 Coculture ECM interaction network. A: Protein interaction network constructed from coculture ECM proteins identified by MS. ECM proteins were classified as basement membrane, other structural ECM or ECM-associated proteins, and were coloured and arranged accordingly. Nodes are labelled with gene names. ) B: Distributions of degree (number of protein -protein interactions per protein) for basement membrane, other structural ECM or ECM-associated proteins for the coculture ECM interaction network. Data points are shown as circles; outliers are shown as diamonds. **, P < 0.01; NS , P ≥ 0.05. Adam Byron generated data for this figure. 107
Figure 3.7 The composition of coculture ECM resembles glomerular ECM in vivo . A-C: Hierarchical clustering analysis of the abundance of basement membrane proteins (A), other structural ECM proteins (B) and ECM-associated proteins (C) identified in glomerular, coculture, podocyte and GEC ECMs. Heat maps display relative protein abundance (normalised spectral count), and associated dendrograms display clustering on the basis of uncentered Pearson correlation. Proteins are labelled with gene names for clarity. Arrowheads indicate candidates selected for Western blotting analysis. D: Western blotting (WB) confirmed the expression of selected structural ECM components identified by MS. Rachel Lennon generated data for this figure. 108
3.3.3.6 Building the glomerular ECM interactome
To investigate the molecular architecture of glomerular ECM in vivo and in vitro , a protein interaction network was generated to compare the monoculture and coculture ECM proteomes and the tissue- derived glomerular ECM proteome (JASN-2013-03-233). A total of 232 proteins were mapped to the interactome, which involved 511 reported protein–protein interactions (Figure 3.8). By clustering proteins according to their detection in different ECMs, we identified a core subnetwork of 35 proteins detected in cell- and tissue-derived ECM datasets. Intriguingly, these 35 proteins, which represented
15% of proteins in the network, were involved in 259 protein–protein interactions, which represented
51% of interactions in the network, demonstrating a substantial clustering of network connectivity in the core subnetwork (Figure 3.8). These findings suggest that these core proteins play key roles in mediating the majority of structural and functional interactions in glomerular ECM in vitro and in vivo .
Interestingly, the core ECM components included 50% of the basement membrane proteins (as compared to 9% of the ECM-associated proteins) in the interactome, supporting the central role of these structural components in ECM organisation. All other basement membrane proteins (apart from von Willebrand factor A domain–containing 1) were detected either in cell-derived ECM or tissue- derived glomerular ECM, but not in both, suggesting distinct structural ECM subnetworks in vitro and in vivo , which may reflect environmental context–dependent synthesis of glomerular ECM. 109
Figure 3.8 Model of the glomerular ECM interactome. All three in vitro ECM MS datasets and the in vivo glomerular ECM MS dataset were combined and mapped onto a database of human protein–protein interactions to create an interaction network model. The network comprised 232 glomerular ECM proteins (nodes; circles) and 511 protein–protein interactions (edges; grey lines). Proteins were clustered according to their identification in one or more ECM; the Venn diagram sets indicate in which ECMs each protein was identified. A core subnetwork of 35 proteins, which were detected in all four glomerular ECM preparations (central intersection set; inset), were involved in 51% (259) of protein–protein interactions in the network. ECM proteins were categorised as basement membrane, other structural ECM or ECM-associated proteins, and were coloured accordingly. Nodes are labelled with gene names for clarity. Adam Byron generated data for this figure. 110
3.3.4 Discussion
To investigate the relative cellular contribution to glomerular ECM synthesis, we focused on the cells either side of the GBM and characterised the ECM proteomes of GEC and podocytes in vitro . We found: (1) both GEC and podocytes synthesised and organised ECM in culture, the composition of which overlapped considerably with the tissue-derived glomerular ECM proteome; (2) the cell-derived
ECMs contained distinct components, with the enrichment of proteins involved in vascular development in GEC ECM and cell adhesion in podocyte ECM; (3) cell-derived ECM resembles a developmental phenotype; (4) glomerular cell crosstalk affects the composition of glomerular ECM; and (5) a core subnetwork of structural ECM proteins appears central to glomerular ECM assembly.
In addition to considerable overlap of ECM components, MS analysis of cell-derived ECMs revealed the enrichment of proteins involved in distinct cellular roles: vascular development in GEC ECM and cell adhesion in podocyte ECM. This enrichment of cell type–appropriate biological processes shows that the analysis of ECM isolated in vitro is a suitable approach for the investigation of the relative contribution of different cell types to glomerular ECM assembly.
There is growing evidence that cellular crosstalk within the glomerulus is critical for function. 315, 374 GEC and podocyte coculture was used to determine whether crosstalk altered the composition and organisation of cell-derived ECM. Ultrastructural analysis revealed the production of ECM between adjacent cells, which had a striking resemblance to a basement membrane. In addition, we observed
ECM between GEC in monoculture when these cells were cultured in medium conditioned by podocytes. Both observations provide evidence that the assembly and organisation of ECM requires crosstalk between glomerular cells and suggests that soluble mediators (eg. growth factors or ECM components) released by podocytes contribute to this effect.
The use of a coculture approach in vitro and hierarchical clustering analysis of MS data identified patterns of protein enrichment in common between the proteomes of the coculture ECM and the tissue- derived glomerular ECM (JASN-2013-03-233). Many of these similarities did not extend to monoculture-derived ECMs, suggesting that crosstalk between GEC and podocytes provides cues necessary to generate ECM that better resembles glomerular ECM in vivo . For example, components 111
such as collagen VI and TINAGL1 were enriched in coculture and glomerular ECM, whereas the expression of other components was reduced as compared to monoculture ECM. Interestingly, proteins associated with vascular development, such as CYR61 373 and HRTA1, 375 were downregulated in coculture ECM compared to GEC ECM. This suggests that paracrine factors may influence cell differentiation and fate. Tenascin C, which was highly expressed in podocyte ECM, was downregulated in coculture ECM and absent in glomerular ECM. This ECM ligand has roles in neurite guidance, 376 vascular morphogenesis and response to wounding. 377 The biological relevance of these molecules for glomerular function has yet to be established; however, our in vitro data suggest that these molecules could have temporally restricted expression in the glomerular ECM.
Among the basement membrane proteins detected, collagen IV chain α3, α4, and α5 and laminin β2 were present in low abundance in cell culture, indicating that cell-derived ECM resembles a developmental ECM phenotype. The expression of these ECM molecules was not substantially altered between monoculture and coculture, which contrasted strongly with the enrichment of these proteins to tissue-derived glomerular ECM (JASN-2013-03-233). Therefore, cell crosstalk between GEC and podocytes was not sufficient to recapitulate expression of the mature isoforms of these ECM molecules. We speculate that the developmental ECM phenotype could be due to the absence of environmental cues such as flow, tension, or specific cell–cell crosstalk in our existing cell culture system. Further investigation and incorporation of these environmental stimuli could lead to improved cell culture systems.
This unbiased, global investigation demonstrates the complexity of the glomerular ECM proteome.
Using a combination of strategies in vitro we were able to provide insight into the relative contribution of glomerular cells to ECM production and show that this may be regulated by cell–cell crosstalk. Network analyses revealed distinct and overlapping subnetworks of ECM proteins and a core set of components that may play key roles in the structure and function of glomerular ECM. These networks can now be extended to investigate ECM composition, organisation, and regulation in the context of glomerular development and disease.
112
3.3.5 Methods
3.3.5.1 Antibodies
Monoclonal antibodies used were against β-catenin (610154; BD Biosciences), cleaved caspase-3
(9664; Cell Signalling) PECAM1 (89C2; Cell Signalling), TINAGL1 (ab69036; Abcam, Cambridge, UK), tenascin C (ab108930; Abcam), and collagen IV chains (provided by B. Hudson, Vanderbilt University
Medical Center, Nashville, TN, USA). Polyclonal antibodies used were against pan-collagen VI
(ab6588; Abcam), CYR61 (ab24448; Abcam) and ZO1 (5406; Cell Signalling). Secondary antibodies against rabbit IgG conjugated to TRITC and mouse or rat IgG conjugated to FITC (Jackson
ImmunoResearch Laboratories, Inc., West Grove, PA, USA) were used for immunofluorescence; secondary antibodies conjugated to AlexaFluor 680 (Life Technologies, Paisley, UK) or IRDye 800
(Rockland Immunochemicals, Glibertsville, PA, USA) were used for Western blotting.
3.3.5.2 Western blotting
See general materials and methods.
3.3.5.3 Glomerular cell culture
Conditionally immortalised human podocytes 363 and GEC 364 were grown in monoculture or coculture on uncoated tissue culture plates. Podocytes between passage 10 and 16 were cultured for 14 days at
37 oC in RPMI-1640 medium with glutamine (R-8758; Sigma, St. Louis, MO, USA) supplemented with
10% (v/v) FCS (Life Technologies) and 5% (v/v) ITS (I-1184; Sigma; 1 ml/100 ml). GEC were grown in endothelial basal medium-2 (CC-3156; Lonza, Slough, UK) containing 5% (v/v) FCS and EGM-2
BulletKit growth factors (CC-4147; Lonza) , excluding VEGF, for 5 days at 37 oC. For coculture, proliferating cells in suspension were combined, seeded, and cultured in GEC medium for 14 days at
37 oC. For experiments with conditioned medium, feeder podocytes and GEC were grown in GEC medium (as above) for 12 days. Test podocytes and GEC were grown in parallel at 37 oC for 12 days.
Every 48 hours , medium from test cells was removed and replaced with 50% (v/v) fresh medium and
50% (v/v) conditioned medium from feeder cells. Before addition to test cells, conditioned medium was centrifuged at 1600 rpm for 4 minutes to remove cells and cell debris . 113
3.3.5.4 Non-glomerular cell culture
HEK 293T and human foreskin fibroblasts were cultured until confluent in Dulbecco's Modified Eagle
Medium (DMEM) supplemented with 10% foetal bovine serum (Gibco) .
3.3.5.5 Lentiviral production and transduction
Podocytes stably expressing GFP were produced by lentiviral transduction. HEK 293T cells were transfected with three plasmids obtained from Addgene (psPAX2 Addgene ID 12260, pMD2.G
Addgene ID 12259 and pWPXL Addgene ID 12257) using polyethyleneimine (Sigma-Aldrich).
Conditioned medium containing viruses was collected after 5 days following several media changes including an 8 hr incubation with 1M sodium butyrate-containing media to promote virus production.
Conditioned media was then used immediately to infect conditionally immortalised podocytes or stored at -80°C. To each well, one ml of normal culture media supplemented with 8 g.ml-1 of polybrene, and
1 ml of 293T conditioned media containing the virus was added. The day after the transduction, the medium was replaced by fresh media. Positive GFP-expressing podocytes were selected using fluorescence-activated cell sorting.
3.3.5.6 Isolation of cell-derived ECM from monocultures
ECM was isolated as previously described. 342, 368 Fifteen-centimeter plates (Corning BV Life Sciences,
Amsterdam, The Netherlands) were seeded with glomerular cells, and culture medium was exchanged alternate daily for fresh medium supplemented with 50 g/ml ascorbic acid to stabilise ECM fibrils.
Differentiated cells were washed with cold PBS, and ECM was denuded of cells by lysis with 20 mM
NH 4OH and 0.5% (v/v) Triton X-100 in PBS, followed by digestion with 10 g/ml DNase I (Roche) in
PBS containing 1 mM calcium and 0.5 mM magnesium. Denuded ECM was scraped into reducing sample buffer and heat denatured at 70 oC for 20 minutes. Three biological replicates were analysed for podocyte and GEC monocultures. 114
3.3.5.7 Isolation of cell-derived ECM from cocultures
To recover ECM deposited between cocultured cells, the method of ECM enrichment from tissue
(described above) was adapted. Differentiated cells were washed with cold PBS. Cells and ECM were scraped into extraction buffer (10 mM Tris, 150 mM NaCl, 1% (v/v) Triton X-100, 25 mM EDTA, 25