Extracellular Matrix Assembly: a Multiscale Deconstruction

Home , COL24A1, Extracellular space, Lysyl oxidase, Type V collagen

FOCUS ON the extracellular matrix REVIEWS

Extracellular matrix assembly: a multiscale deconstruction

Janna K. Mouw1, Guanqing Ou1,2 and Valerie M. Weaver1,3–6 Abstract | The biochemical and biophysical properties of the extracellular matrix (ECM) dictate tissue-specific cell behaviour. The molecules that are associated with the ECM of each tissue, including collagens, proteoglycans, laminins and fibronectin, and the manner in which they are assembled determine the structure and the organization of the resultant ECM. The product is a specific ECM signature that is comprised of unique compositional and topographical features that both reflect and facilitate the functional requirements of the tissue.

The extracellular matrix (ECM) has important roles in can function both as support for cells and as active regulating the development, function and homeostasis of participants in signalling to control cell behaviour9,10. all eukaryotic cells1–3. In addition to providing physical In all cases, each class of ECM molecule has evolved the support for cells, the ECM actively participates in the ability to interact with another class to produce unique establishment, separation and maintenance of differen- physical and signalling properties that support tissue tiated tissues and organs by regulating the abundance structure, growth and function1. Growing insight into

1Center for Bioengineering of growth factors and receptors, the level of hydration the biological importance of the ECM in development, 1,4 and Tissue Regeneration, and the pH of the local environment . These diverse pathophysiology and the normal function of tissues Department of Surgery, functions are achieved through its complex chemical throughout the body has confirmed the importance of University of California, composition and organization. Although it is primarily elucidating the mechanisms of ECM assembly. San Francisco. 2University of California composed of water, proteins and polysaccharides, This Review provides a general overview of our cur- San Francisco and the ECM shows exquisite tissue specificity as a result rent understanding of the ‘structural principles’ that University of California of the unique ECM compositions and topographies that underlie the assembly of the ECM at several levels, high- Berkeley Joint Graduate are generated through a dynamic biochemical and bio- lighting key molecular components and their organiza- Group in Bioengineering, physical interplay between the various cells in each tissue tion into an interlocking network. We also consider San Francisco, California 4 94143, USA. and the evolving microenvironment . The mature ECM the biochemical and mechanical cues that affect ECM 3Department of Anatomy, can also undergo dynamic remodelling in response to assembly and how its organization directly relates to University of California, environmental stimuli, such as applied force or injury, its tissue-specific functions, including as a structural San Francisco. which enables the tissue to maintain homeostasis and support, as a barrier between different tissues and as a 4Department of Bioengineering and to respond to physiological challenges and stresses, source of both chemical and physical cues. We highlight 2,5–8 Therapeutic Sciences, including disease . unanswered questions that require further investigation University of California, Understanding cellular differentiation, tissue and the techniques and model systems that may facilitate San Francisco. morphogenesis and physiological remodelling requires this research. 5 Eli and Edythe Broad Center an understanding of the ECM components that are of Regeneration Medicine and Stem Cell Research, produced by cells, as well as the assembly of those Key molecular players in the ECM University of California, macromolecules into a functional three-dimensional Our understanding of the composition, structure and San Francisco. structure2,3. The macromolecules that constitute the function of the ECM is continuously evolving, aided by 6UCSF Helen Diller ECM have evolved structural and chemical properties the discovery of novel ECM molecules, the mapping Comprehensive Cancer Center, University of that are particularly suited to their specific biological of internal sites within ECM proteins that are crucial California, San Francisco, functions in their respective tissues. Small, modular for self-assembly and for interactions with other ECM California 94143, USA. subunits form homopolymers and heteropolymers that components, the characterization of the proteases and Correspondence to V.M.W. become supramolecular assemblies with highly special- the protease inhibitors that are responsible for ECM e‑mail: Valerie.Weaver@ ized organizations1–3. These assemblies contain binding degradation and turnover, and the identification of ucsfmedctr.org doi:10.1038/nrm3902 domains for growth factors and chemokines, establish novel receptors and signalling mechanisms that medi- 1–3 Published online complex adhesion surfaces and form diffusion barriers ate cellular responses to the ECM . The architecture 5 November 2014 between different cellular layers; thus, ECM molecules of the ECM is highly organized as a result of the innate

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 771

a Standard collagen molecule properties of its constituent molecules and their inter N terminal C terminal actions, as well as the activities of resident cells. The propeptide propeptide interactions between cells and the environment that 1 Gly-X-Y repeats are created by the ECM in turn have important roles in directing processes in development, homeostasis and 3 α chains (2 α1,1 α2) pathogenesis. In this Review, we define the ECM as all secreted molecules that are immobilized outside a cell, including growth factors, cytokines and cell adhesion molecules, although we primarily focus on the macromolecules that are mainly responsible for tissue-type C-terminal propeptide 2 initiates triple helix formation specific extracellular architecture. in the rough ER The protein and non-protein constituents of the ECM vary not only in terms of their functional roles but also in terms of their structure. The polypeptide chains of ECM macromolecules often consist of many Procollagen is secreted individual domains, and homologous domains between into the extracellular space 3 Procollagen macromolecules can have sequence and structural differences that impart different functions. Individual ECM macromolecules rarely exist in isolation but Enzymes cleave propeptides instead often integrate into supramolecular structures to form collagen 4 Collagen containing different molecular species that differ in both their identity and their relative abundance. The ECM is primarily composed of two main classes of macromolecules: fibrous proteins (including collagens and elastin) and glycoproteins (including fibronectin, Topographies b FACIT and related collagens proteoglycans (PGs) and laminin)1. The three-dimensional qualities Gly-X-Y interruption of surfaces or structures, including contours and relief. Fibrillar collagens. Collagens are major proteins of the In the context of the ECM this Non-collagen ECM and are arguably the most dominant. Decades of includes features such as peaks Collagen research have uncovered 28 different collagen types, and valleys, changes in and each type is comprised of homotrimers and hetero roughness and geometric features. trimers that are formed by three polypeptide chains (known as α-chains) (FIG. 1; TABLE 1). More than 40 Morphogenesis distinct α-chains have been identified in humans as Figure 1 | Collagen structure. a | The standard fibrillar The process of cell movement well as multiple other proteins containing collagen- collagen molecule isNature characterized Reviews by| Molecular amino- and Cell Biology during embryonic development like domains11,12. The structural hallmark of all colla- that controls the size, the carboxy-terminal propeptide sequences, which flank a shape and the patterning of series of Gly‑X‑Y repeats (where X and Y represent any gens is their triple helix — a tight right-handed helix tissues and organs. amino acids but are frequently proline and hydroxyproline) of three α-chains, each of which contains one or more (step 1). These form the central triple helical structure of regions characterized by the repeating amino acid motif Proteoglycans procollagen and collagen. Three α‑chains (the illustration Gly‑X‑Y, where X and Y can be any amino acid13 (FIG. 1). (PGs). Glycoproteins that shows two 1-chains and one 2-chain, which is consist of a core protein to α α Some collagen molecules assemble as homotrimers, which one or more representative of type I collagen) are intracellularly whereas others assemble as heterotrimers that are com- glycosaminoglycan chain is assembled into the triple helix following initiation of this prised of two or three distinguishable α-chain types. The attached. process by the C‑terminal domain (step 2). Procollagen is large family of collagens and proteins with collagenous secreted by cells into the extracellular space (step 3) and domains has revealed that the collagen triple helix is a Glycosaminoglycans converted into collagen by the removal of the N- and (GAGs). Long, linear, charged C‑propeptides via metalloproteinase enzymes (step 4). basic motif that can be adopted by proteins that have polysaccharides that comprise b | Fibril-associated collagens with interrupted triple helices different functions. The rod-like domain has the poten- repeating pairs of sugars, of (FACIT) and related collagens have a different structure to tial to engage in various modes of self-association and which one is an amino sugar. standard fibrillar collagen; they contain non-collagenous the capacity to bind to cell surface receptors, other glycosaminoglycans Fibrillar collagen regions — that is, non-triple helical sequences. These lead proteins, (GAGs) and nucleic acids. Polymerized, supramolecular to kinks in the resulting macromolecular structure that A detailed description of the structural organizations of collagen that has been straighten under small strains. Figure part a is modified, the different types of collagens can be found in REF. 14. organized into fibrils; collagen with permission, from REF 135 © 2012 Fan et al.; licensee For the purposes of this Review, we specifically focus on types I, II and III form fibrils. BioMed Central Ltd, and from REF. 136, Klug, William S.; the assembly of fibrillar collagen, which comprises multiple Cummings, Michael R., Concepts of Genetics, 5th Edition, © complex intracellular and extracellular post-translational Tensile forces 1997. Reprinted by permission of Pearson Education, Inc., The forces required to exert a Upper Saddle River, NJ. Figure part b: this figure was processes from the translational product to a fibrillar certain amount of tension originally published in Biochem. J. Jäälinoja, J., Ylöstalo, J., structure that is capable of withstanding tensile forces. (which is a type of normal The unique mechanical properties of fibrillar collagen stress) on a one-dimensional Beckett, W., Hulmes, D. J. S. & Ala-Kokko, L., Trimerization of object. In this context, a pulling collagen IX alpha-chains does not require the presence of are mainly controlled by its structure, which shows force on a cable tending to the COL1 and NC1 domains. Biochem. J. 2008; 409: 545–554 the important relationship between three-dimensional cause extension of the cable. © the Biochemical Society. protein structure and the function of the resultant ECM6.

772 | DECEMBER 2014 | VOLUME 15 www.nature.com/reviews/molcellbio

Table 1 | Primary types of collagens* Classes Type Genes encoding proteins typical Description of collagen composition Fibrillar I COL1A1 and COL1A2 Quarter-stagger packed fibrils that are primarily found in fibrous stromal matrices, such as skin, bone, II COL2A1 tendons and ligaments III COL3A1 V COL5A1, COL5A2 and COL5A3 XI COL11A1, COL11A2 and COL11A3 XXIV COL24A1 XXVII COL27A1 FACIT IX COL9A1, COL9A2 and COL9A3 Fibril-associated molecular bridges that are associated with type I (XII, XVI, XIX, XXI) and type II XII COL12A1 (IX, XVI, XIX) collagen fibrils XIV COL14A1 XVI COL16A1 XIX COL19A1 XX COL20A1 XXI COL21A1 XXII COL22A1 Basement IV COL4A1, COL4A2, COL4A3, Network structure comprised of laminins and membrane COL4A4, COL4A5 and COL4A6 basement membrane proteins Long chain VII COL7A13 Anchoring fibrils that are associated with the basement membrane Filamentous VI COL6A1, COL6A2, COL6A3, Beaded microfibrils COL6A5 and COL6A6 Short chain VIII COL8A1 Hexagonal lattice structure X COL10A1 These collagens both regulate, and are regulated by, hypertrophy in cartilage Multiplexins XV COL15A1 Multiple triple helix domains with interruptions containing chondroitin sulphate and heparin XVIII COL18A1 sulphate glycosaminoglycans MACIT XIII COL13A1 Cell surface molecules with extracellular, (transmembrane membrane-spanning and intracellular domains domain) XVII COL17A1 XXIII COL23A1 *Collagen types are classified on the basis of domain structure homology and suprastructural assembly. FACIT, fibril-associated collagens with interrupted triple helices; MACIT, membrane-associated collagens with interrupted triple helices.

Four distinct stages define collagen fibril assembly to promote the stability and glycosylation of the triple after the transcription and translation of the procollagen helices following formation17. At the same time, a lim- α-chains: import into the rough endoplasmic reticulum, ited number of lysine hydroxyl groups are glycosylated. where the α-chains are modified so that they can form These proline and hydroxyproline amino acids can triple-helical procollagen; modification of procollagen comprise 20% of the molecule, contributing to triple in the Golgi apparatus and its packaging into secretory helix stabilization. Although the exact mechanism that vesicles; cleavage of the procollagen to form the colla- controls the stabilization of the triple helix has gener- gen molecule in the extracellular space; and lysyl oxidase- ated much controversy, the current consensus suggests Procollagen catalysed crosslinking between collagen molecules to that increased stabilization is partly due to its high A trimeric collagen precursor stabilize the supramolecular collagen structure6,11,12,15,16 content of amino acids, increased interchain hydrogen molecule containing large (FIGS 1,2a). All fibrillar collagens are initially synthesized bonds through hydration networks and electrostatic amino- and carboxy‑terminal 13,18,19 propeptide domains. as precursor molecules that contain large amino- and interactions between lysine residues . Following carboxy-terminal propeptides and a signal recognition lysyl and prolyl hydroxylation and O‑linked glycosyla- Lysyl oxidase sequence, which promotes targeting to the rough endo- tion, α-chains associate to form procollagen, and triple An enzyme that participates plasmic reticulum, where post-translational modifica- helix trimerization is initiated by the association of the in the formation of collagen tions take place, leading to the assembly of procollagen C‑propeptide domains through both specific chain polymers by facilitating 17 oxidative deamination of molecules . Within the endoplasmic reticulum, particu- recognition and the formation of a stable nucleus, which 15,20–23 peptidyl lysine residues on lar lysine and proline residues within the propeptides are favours triple-helical assembly . Recent work eluci- collagen monomers. hydroxylated by lysyl and prolyl hydroxylase enzymes dating the crystal structure of the C‑propeptide domain

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 773

a Fibrillar collagen Procollagen

Procollagen peptidase Collagen

Fibril formation Microﬁbril

Fibril Aldol and aldol-histidine crosslinks Fibre via lysyl oxidase Stained

Regular patterning Fibre bundle (subfascicle) b

Type V

Fascicle

Tertiary ﬁbre bundle

Type IX, XII, XIV (FACIT)

Tendon Figure 2 | Fibrillar collagen assembly. a | After procollagen is secreted into the extracellular space, collagen type-specific metalloproteinase enzymes remove the amino- and carboxy-propeptides.Nature Collagen Reviews is| Molecular assembled Cell into Biology cross-striated microfibrils that occur in the extracellular matrix of connective tissues. This can be observed via electron microscopy as regularly spaced bands, as represented by the grey, striped bar. The position of non-fibrillar collagens in this structure is shown in similar diagrams in part b. Short microfibrils merge into mature fibrils through longitudinal and axial growth. To form mature fibres, lysyl oxidase catalyses the formation of intramolecular and intermolecular covalent crosslinks between collagen molecules. Fibres are bundled together within a connective tissue and stabilized via interactions with fibril-associated collagens with interrupted triple helices (FACIT) collagens and small leucine-rich repeat proteoglycans (SLRPs) into linear structures capable of transmitting tensile forces. Specifically in the case of tendons, fibres ranging in size from 1 μm to 20 μm are bundled into 15–400 µm diameter triangular subfascicles, subfascicles are bundled into 150–1000 µm fascicles and fascicles are grouped into tertiary bundles ranging from 1000 μm to 3000 µm. Finally, multiple tertiary bundles are enclosed by the epitenon connective tissue, which contains the vascular, lymphatic and nerve sources for the tendon tissue. b | In addition to fibrillar collagen, other collagen subtypes, such as type V and FACIT collagens, are incorporated into the fibril structure. Type V collagen is inserted between strands of the microfibril, and FACIT collagens cling to the surface of the microfibril and work with SLRPs to stabilize higher order structures. Figure part a is modified, with permission, from REF. 135 © 2012 Fan et al.; licensee BioMed Central Ltd, from Klug, William S.; Cummings, Michael R., Concepts of Genetics, 5th Edition, © 1997 (REF. 136). Reprinted by permission of Pearson Education, Inc., Upper Saddle River, NJ, and from REF. 138, reprinted from Trends Biotechnol., 26, Liu, Y., Ramanath, H. S. & Wang, D.-A., Tendon tissue engineering using scaffold enhancing strategies, 201–209, Copyright (2008), with permission from Elsevier. Figure part b: this figure was originally published in Biochem. J. Jäälinoja, J., Ylöstalo, J., Beckett, W., Hulmes, D. J. S. & Ala-Kokko, L., Trimerization of collagen IX alpha-chains does not require the presence of the COL1 and NC1 domains. Biochem. J. 2008; 409: 545–554 © the Biochemical Society.

774 | DECEMBER 2014 | VOLUME 15 www.nature.com/reviews/molcellbio

from procollagen III suggests that there is a structural After cleavage of procollagen into collagen, initial mechanism by which α-chains are selectively recognized, fibril formation events occur at the cell surface38. Colla- with the chain recognition sequence (CRS) of one chain gen organization in fibrils has been intensely studied by interacting with the CRS of a neighbouring chain during electron microscopy and X‑ray diffraction39–44. Collagen intracellular trimerization of procollagen to ensure cor- self-assembles to form microfibrils with a quarter stag- rect homotrimeric and heterotrimeric chain stoichiome- ger axial D periodicity of approximately 67 nm to create try20. In addition, protein disulphide isomerase catalyses the characteristic ‘striation’ that was recognized in initial the formation and the rearrangement of intramolecular electron microscopy observations of collagen-containing and intermolecular disulphide bonds24. Many other pro- tissues16,45 (FIG. 2a). Intermediate-sized microfibrils are teins influence and guide triple helix formation during thought to form via a multistage process that includes procollagen assembly, including prolyl 4‑hydroxylase, the nucleation, organization and unilateral elongation 78 kDa glucose-regulated protein (GRP78) and GRP94 of short primary fibrils. Current models suggest that (REF. 15). Importantly, the tight packing of the three microfibril assembly is nucleated and modulated by the α-chains near the common axis places steric constraints inclusion of collagens V and/or XI, which, as discussed on every third amino acid position, with only the small- above, retain portions of their N‑terminal propeptide to est amino acid, glycine, accommodated without chain co‑assemble into fibrils with types I, II and III (FIGS 1,2b). distortion; this generates the Gly‑X‑Y repeating motif The N‑terminal propeptides of collagens V and/or XI are and imparts further stability to the triple helix13. Detailed thought to project outwards through the gap between reviews discussing the collagen triple helix structure and adjacent collagen molecules to interact with the fibril assembly can be found in REFS 13,15,25,26. surface, which limits lateral growth via steric hindrance In the second step, after assembly in the endoplasmic and charge interactions38 (FIG. 2b). Although collagen V is reticulum, procollagen is packaged in the Golgi appa- a quantitatively minor component relative to collagen I ratus for export into the extracellular matrix. For the and comprises less than 5% of the total collagen con- third step of collagen fibril synthesis, evidence suggests tent in most tissues, it is considered to be a dominant that procollagen processing, exocytosis and initial fibril regulator of fibrillogenesis, as deletion of collagen V in assembly are closely associated with membrane-enclosed the mouse is associated with a lack of fibril assembly38. compartments15. Early models proposed that cells exert After collagen microfibrils have been assembled, considerable spatial control over this pericellular space, short microfibrils may merge or grow into fibrils, with procollagen excreted by exocytosis and modified increasing both longitudinally and axially; these transi- within deep, narrow recesses between cellular protru- tions are regulated by various molecules38,46 (FIG. 2). For sions close to the cell body27,28. Although procollagen example, the large body of evidence gained using in vitro Pericellular space processing was originally viewed as a solely extracellular techniques and in vivo knockout transgenic experiments The space surrounding a cell. event (with the second and third steps of collagen fibril shows that small leucine-rich repeat proteoglycans (SLRPs) synthesis being distinct and mutually exclusive), later with collagen-binding properties, including decorin, Fibripositors work provided evidence that procollagen processing and biglycan, fibromodulin and lumican, can affect fibril (Also known as fibropositors). 15,47,48 Plasma membrane protrusions collagen fibril assembly are initiated in compartments growth rate, size, morphology and content . Most projecting from the cell surface, between the Golgi apparatus and the plasma membrane heterotypic collagen ﬁbrils are composed of a mixture where procollagen can that are targeted to plasma membrane extrusions (known of fibrillar collagens, with the FACIT (fibril-associated potentially be processed, as fibripositors) that project from the cell surface. Recent collagens with interrupted triple helices) collagens assembled and organized electron microscopy studies confirmed the presence of integrating into the fibrils, selectively altering the sur- by individual cells. collagen fibrils in cell surface fibripositors29–32. Follow- face properties of the collagen ﬁbril and its interactions 37,49 Fibrillogenesis ing or during secretion of procollagen into the ECM, with additional ECM molecules such as the SLRPs . The development or formation proteolytic processing of the large procollagen NH2- and Throughout the process of fibril assembly and growth, of fibrils, used to refer to COOH-propeptide domains leads to the production of interactions among the assembling fibrils, accessory collagen- or fibronectin-rich structures. mature collagen molecules that then self-assemble into macromolecules and cell adhesion receptors actively fibrils. Enzymatic removal of the propeptide domains organize the maturing fibril. Small leucine-rich repeat is carried out by collagen type-specific metalloprotein- To form mature collagen fibres, further modifications proteoglycans ases from the a disintegrin and metalloproteinase with are made during the assembly of collagen fibril aggre- (SLRPs). A family of thrombospondin motif (ADAMTS) and bone morpho gates. In the final step of collagen biosynthesis, cova- proteoglycans that share a common leucine-rich-repeat genetic protein 1 (BMP1) and tolloid-like families, as lent crosslinks are introduced into the supramolecular 16,33–36 motif in their conserved well as by furin-like proprotein convertases (FIG. 1). assembly to provide stability and enhanced mechani- carboxy termini; they have If the C‑propeptide remains attached, procollagen solu- cal properties50–52. Extracellular lysyl oxidases catalyse been strongly implicated in bility in the extracellular space remains high, which and facilitate the oxidative deamination of targeted modulating fibrillar collagen assembly. inhibits premature fibril assembly; by contrast, the peptidyl lysine residues, which results in the formation persistence of the N‑propeptide influences fibril shape of aldols or β‑ketoamines by reacting with aldehydes or FACIT and diameter, possibly by increasing the surface area to amino groups on lysines between collagen monomers (Fibril-associated collagens volume ratio, without affecting fibril formation16,37. In to form intramolecular and intermolecular covalent with interrupted triple helices). collagens I, II and III, the N‑propeptides are completely crosslinks49,51–53. The N- and C-terminals of neighbour- A type of collagen that does 15 not form fibrils by itself but removed, leaving short N-propeptides . More detailed ing collagen monomers interact with each other and are that is associated with the information on the roles and processing of the C- and covalently crosslinked by lysyl oxidase, both within and surface of fibrillar collagen. N-propeptides can be found in REF. 16. between microfibres52–54. Overall, interactions between

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 775

the fibrillar and the FACIT collagens, as well as among common CSPGs in the neural ECM68,69 (FIG. 3a). This collagens, proteoglycans and crosslinking enzymes, can family consists of four members: aggrecan, neurocan, function as key regulators of fibrillar collagen organiza- veriscan and brevican61,69. The core protein is highly tion, resulting in immense tissue-, stage- and age-specific homologous across lecticans, which are characterized differences in both fibril composition and structure. by globular domains at both the N and the C termini. These variations are finely tuned to fulfil bespoke bio- Lecticans contain binding domains for hyaluronic chemical, structural and biomechanical roles25. Impor- acid (which is a common glycosaminoglycan in carti- tantly, the assembly of collagen fibrils is mainly driven lage and the nervous system), lectins and growth fac- by the unique chemical and physical interactions of the tors56,69,70. As with HSPGs, alternative splicing and dif- interacting components. This cell-driven self-assembly ferent densities of GAG chains on the core protein lead results in a multi-hierarchical collagen structure, which to considerable variations in the properties of these provides binding sites for other proteins and cells and is proteins56. crucial for the mechanical integrity of the ECM55. In addition to being dominant components of the ECM, proteoglycans can also function as accessory pro- Proteoglycans. In contrast to the predominantly fibril- teins in tissues rich in other matrix proteins. For exam- lar structure of collagens, proteoglycans form the basis ple, the SLRPs are a family of proteoglycans that have of higher order ECM structures around cells. The pri- been implicated in fibrillar collagen assembly48. This mary biological function of proteoglycans derives from family includes well-known members such as decorin, the biochemical and hydrodynamic characteristics biglycan and lumican, with new members still being dis- of the GAG components of the molecules, which bind covered and characterized. All SLRPs have two regions water to provide hydration and compressive resistance. — a variable N‑terminal domain that contains sulphated As such, proteoglycans are found in abundance in car- tyrosine or acidic amino acid residues and a conserved tilage and neural ECM56,57. Proteoglycans are character- carboxy terminus that contains the leucine-rich repeats ized by a core protein that is covalently linked to GAGs, (LRRs) for which the family was named48,71 (FIG. 3b). GAG which are long, negatively charged, linear chains of chains can then be added to these core proteins48. The disaccharide repeats . They are classified into subtypes LRRs contain collagen-binding regions and have been on the basis of the structure of these GAG carbohydrate well characterized as a result of this property; the func- chains, as well as on the distribution and the density of tion of the N‑terminal domains is less extensively under- these chains along the core protein58,59. Major GAGs stood48. Elucidation of the crystal structure of decorin include heparin sulphate, chondroitin sulphate, derma- suggests that these N‑terminal domains function as a tan sulphate, hyaluronan and keratin sulphate58,60–62 (the capping mechanism, whereby cysteine residues within structures of heparin sulphate and chondroitin sulphate this domain form disulphide bonds that link back to the are shown in FIG. 3a). C terminus via leucines in the consensus sequence72. Heparan sulphate proteoglycans (HSPGs) are a Five classes of SLRPs have so far been established, and major part of the basement membrane, which is a type of each class is distinguished by their homology on the pro- ECM structure (see below). Their GAG chains are com- tein and genomic levels, by the presence of N‑terminal prised of highly sulphated chains of hexuronic acid and cysteine-rich clusters and by chromosomal placement. d‑glucosamine disaccharide repeats. HSPGs can be cell Classes I–III are so‑called ‘canonical’ classes, which are surface-bound, such as the syndecans and CD44v3, glyco distinguished by the relative abundance of data describ- phosphatidylinositol (GPI)-linked, such as glypicans, or ing their function and localization within the body, secreted ECM proteins, such as agrin, collagen XVIII whereas classes IV–V have only been recently described and perlecan63. Variety in HSPGs is introduced through and there are many unanswered questions about their Basement membrane splice isoforms that control GAG chain lengths, the spac- function. A full listing and description of SLRPs is A thin, complex extracellular ing of GAG chain attachments along the protein core beyond the scope of this Review, but such information matrix that separates epimerization REFS 73,74 endothelial and epithelial cells and the extent of sulphation and of the sug- can be found in . The Class I SLRPs decorin 62,63 from their subjacent connective ars within the GAG chains . Resulting proteoglycans and biglycan have been extensively studied and partici- tissues. It is composed show different binding affinities for molecules within the pate in collagen fibrillogenesis57,75. Not only do these pro- of various collagens, ECM. For example, fibroblast growth factors (FGFs) are teins facilitate the formation of collagen fibrils but they proteoglycans and particularly sensitive to the spacing and sulphation of also regulate fibril width and growth through mecha- adhesive glycoproteins. GAG chains in their interactions with HSPGs64. HSPGs nisms that have not been fully elucidated74. In addition Syndecans have high negative charges that enable them to easily to their roles as structural proteins in ECM assembly, One of the two major families bind to other proteins, including growth factors such as SLRPs are considered to be ‘active’ members of the ECM of heparin sulphate FGF, chemokines, other ECM proteins, such as laminin and to participate in cell signalling. Binding sites for proteoglycans. and fibronectin, and cell surface receptors63,65,66. As such, growth factors, cytokines and ECM proteins other than 71 Glypicans they have been implicated in regulating many cellular collagen have recently been found on SLRPs . One of the two major families processes, including cell growth and migration63,67,68. of heparin sulphate Chondroitin sulphate proteoglycans (CSPGs) can be Connectors in the ECM proteoglycans. found in cartilage and neural ECM. CSPGs are functional- Collagen, proteoglycans, and hyaluronic acid repre- Epimerization ized with sulphated polysaccharides with repeating disac- sent major structural components within the ECM. 68 The formation of charides of glucuronic acid and N-acetylgalactosamine . They provide much of the supportive framework stereoisomers. Lecticans (also known as hyalectins) are the most within which other ECM components and cells interact.

776 | DECEMBER 2014 | VOLUME 15 www.nature.com/reviews/molcellbio

a Lecticans b SLRPs Hyaluronic GAG chains Tenascin-R acid binding binding Decorin S S S S H2N COOH Core protein NH COOH EGF 2 G1 G2 G3 LxxLxLxxNxLx Globular Globular domain Hyaluronic acid domains LRR Proteoglycan Link protein Biglycan – H/SO 3 S S S S NH2 COOH COO– O O O O HO O O HO O – NH H/SO 3 – Heparin sulphate COCH 3/SO3 /H H/SO – H/SO – 3 Lumican 3 O O S S S S NH2 COOH O O O HO O O OH NH O O– COCH Chondroitin sulphate 3 c Laminin d Fibronectin Fibrin, Gelatin, N terminal Coiled coil Globular heparin collagen FN Integrins Heparin Fibrin

α chain H2N COOH EGF-like FN assembly Variable regions Type I repeat Disulphide β chain bonds LM domain 4 Type II repeat γ chain LM four domain Type III repeat Fibronectin dimer N N Laminin 332 α chain Collagen Collagen binding β chain binding Folded ﬁbronectin dimer α3 β1 γ chain α helical region Cell binding

N N Integrin binding Plasma S S S S membrane S S S S Integrin C C C C Figure 3 | Non-collagenous molecules of the ECM. Lecticans (aggrecan molecule forms a dimer through disulphide bonds on its C terminus. is shown; part a) and small leucine-rich repeat proteoglycans (SLRPs; The folded fibronectin moleculeNature forms via Reviews ionic interactions | Molecular betweenCell Biology decorin, biglycan and lumican are shown; part b) are major proteoglycans type III domains of neighbouring molecules and is deformed by (PGs). The SLRPs are characterized by differences in their amino terminus mechanical force to reveal cryptic binding sites for other fibronectin structure and highly conserved leucine rich repeats (LRR) in the core molecules and cell surface receptors when interacting with cells. molecule. Lecticans have a core protein with binding domains for Figure part a: this image originally appeared in Diamond Light Source glycosaminoglycan (GAG) chains flanked by globular domains that Ltd’s Annual Report 2010 (p72–73) and has been reproduced with interact with hyaluronic acid (at the N terminus) and tenascin R (at the permission from Diamond Light Source Ltd, the University of Liverpool, T. carboxy terminus). Common GAGs include chondroitin sulphate and R. Rudd, M. A. Skidmore, S. E. Guimond, R. Xu, R. Hussain, D. G. Fernig, G. heparan sulphate, the chemical structures of which are depicted. Siligardi and E. A. Yates http://www.diamond.ac.uk/Science/Research/‌ The assembled bottle-brush-like aggrecan–link protein–hyaluronic acid Highlights/study2.html. Figure part b is modified from REF. 139, structure is shown. Laminins are formed by the incorporation of α, β and Nature Publishing Group. Figure part c is modified, with permission, from γ chains into a cruciform, Y‑shaped or rod-like structure (part c). These Yurchenco, P. D. Basement membranes: cell scaffoldings and signaling chains are characterized by different domains, as shown. The domain platforms. Cold Spring Harb. Perspect. Biol. 3, a004911 (2011), Cold Spring structures depicted represent only one isoform for each chain type, but Harbour Laboratory Press, and REF. 141, Nature Publishing Group. Figure major differences between the basic composition of the chains (such as part d is modified, with permission, from Schwarzbauer, J. E. & DeSimone, the lack of globular regions in β and γ chains) are also present in the other D. W. Fibronectins, their fibrillogenesis, and in vivo functions. Cold Spring isoforms. Laminins interact with cell surface receptors, such as integrins, Harb. Perspect. Biol. 3, a005041 (2011), Cold Spring Harbour Laboratory primarily through globular domains in the chain. Fibronectin domain Press, and REF. 82, modified with permission from Annual Review of Cell Dev. structure and the domains to which extracellular matrix (ECM) molecules Biol., Volume 26 by Annual Reviews, http://www.annualreviews.org. and cell surface receptors bind are indicated (part d). The fibronectin EGF, epidermal growth factor; LM domain, laminin domain.

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 777

Table 2 | Non-collagenous ECM proteins Protein Location Description Refs Agrin Skeletal muscles Large PG that is crucial in the development of the neuromuscular junction; contains 142–144 laminin G domains, Kazal-type serine protease inhibitor domains and epidermal growth factor domains Integrin-binding Bone Phosphorylated and sulphated glycoprotein that is crucial for the structure 145–147 sialoprotein (also known as of mineralized matrix; binds to calcium and hydroxyapatite and mediates cell bone sialoprotein) attachment through an RGD sequence that recognizes the vitronectin receptor Matrilins Cartilage and Non-collagenous, disulphide-bonded proteins that are involved in the formation of 148–151 connective tissues filamentous networks; contains von Willebrand factor A domains Elastin Arteries, connective One of two components of elastic fibres that provide structural integrity along with 152–155 tissue, lungs, skin reversible extensibility and deformability; forms covalently crosslinked polymers and bladder Fibrinogen Blood Soluble glycoprotein, which, after cleavage by thrombin, is converted to fibrin 153,156, monomer molecules that self-associate to form insoluble, homopolymeric ‘clots’ 157 Fibrillin Connective tissues Predominant component of microfibrils, which forms a sheath around elastin in the 158,159 and cardiovascular formation of elastic fibres tissues Fibulins Basement Calcium-binding glycoprotein that is found in fibrillar matrices and blood 153,160, membranes, blood 161 and arteries Netrins Basement Laminin-related protein that is involved in axon guidance and the development of 162,163 membranes the vasculature, lung, pancreas, muscle and the mammary gland SPARC (also known as Bone Cysteine-rich, acidic, calcium-binding glycoprotein that is required for collagen 164,165 osteonectin) calcification, mineralization initiation and mineral crystal formation Secreted phosphoprotein 1 Bone and kidneys Negatively-charged glycoprotein that is involved in the attachment of cells to 146,166, (also known as osteopontin) mineralized bone matrix 167 Reelin Brain Large glycoprotein that is involved in guiding neuron and glial cell positioning in the 168,169 developing brain Tenascins Connective tissues Glycoproteins that are involved in mediating inflammatory and fibrotic processes 119,153, 170,171 Vitronectin Blood Adhesive glycoprotein that is involved in blood coagulation and wound healing 172,173 ECM, extracellular matrix; PG, proteoglycan.

Additional ECM components, such as laminin or Y‑shaped (three arms) or rod-shaped (single arm) struc- fibronectin, function as bridges between structural ECM ture76 (FIG. 3c). Different chain isoforms combine to gener- molecules to reinforce this network, as well as to con- ate distinct assembled heterotrimers76. These laminins are nect the ECM to cells and to soluble molecules within described by their combined chain numbers; for example, the extracellular space. These connectors tend to be α1β1γ1 is known as laminin 111. This nomenclature was multidomain glycoproteins that contain binding motifs preceded by single number assignments when there were for other ECM proteins, growth factors and cell surface fewer known isoforms. The original designations remain receptors1. In this Review, we focus on laminin and preferred in some settings so, for example, it is possible fibronectin as examples and point to others in TABLE 2 to find references to laminin 5 instead of α3Aβ3γ2. For a for further reference. useful table that cross-references these different systems, see REF. 77. So far, five α-chains (α1–α5), four β-chains Laminins. Laminins are a family of large, mosaic glyco (β1–β4) and three γ-chains (γ1–γ3) are known, and they proteins composed of globular, laminin-type epidermal combine to form 16 different laminin heterotrimers77,78 growth factor (EGF)-like repeats and α‑helical domains Although laminin 111 has been shown to self-assemble Integrins (FIG. 3c). They are primarily located in the basal lamina into aggregates, laminins primarily function as bridges A large family of heterodimeric and some mesenchymal compartments. Interactions between other molecules79,80. transmembrane proteins, which between modular domains within the laminin molecule are present in the plasma membrane as heterodimers of and other proteins enable laminins to mediate inter Fibronectin. Many of the ECM proteins described above α- and β-subunits and function actions between cells via cell surface receptors (such interact with cells through crucial connections with as receptors for cell adhesion as integrins and dystroglycan) and other components of the multidomain protein fibronectin to regulate cell molecules. the ECM (such as nidogens (also known as entactins), adhesion, migration and differentiation81. Fibronectin perlecans and collagens)76 (FIG. 3c). is secreted as a large ECM glycoprotein that assembles Nidogens (Also known as entactins). Laminins consist of α, β and γ chains that combine via via cell-mediated processes into fibrillar structures 82 Glycoproteins and key parts the triple-helical coiled-coil domain in the centre of each around cells . Each fibronectin subunit consists of three of the basement membrane. chain to form structures with a cruciform (cross-shaped), modules of repeating units, each of which has distinct

778 | DECEMBER 2014 | VOLUME 15 www.nature.com/reviews/molcellbio

structures: type I, type II and type III structures81,82 (FIG. 3d). Fibrillar collagen-rich tendons. Fibrillar ECM that con- These modules contain binding motifs that are impor- sists of substantial amounts of collagens I, II and III, such tant in facilitating the interaction of fibronectin with cell as the ECM found in tendons, ligaments and cartilage, surface receptors, such as integrins, collagen and gelatin, has distinct structural and biomechanical properties and intramolecular units that enable self-assembly of the that facilitate the effective transmission and absorption molecule82,83. Intramolecular disulphide bonds between of cyclical tensile forces while avoiding injury87–89. These each type I and type II module stabilize the folded ter- properties arise from the specialized axial and longitudi- tiary structure of the fibronectin subunit, and fibro nal structural organization of the collagen fibre hierarchy, nectin dimers form via antiparallel disulphide bonds at which provides a scaffold for cell and macromolecular the C terminus82. attachment89,90. The tendon is primarily a uniaxial force- Fibronectin fibril assembly involves interactions transmitting connective tissue consisting of bundles of between its RGD and synergy sequences with correspond- parallel fibrillar ECM and resident tendon fibroblasts ing binding sites within cell surface receptors such as inte- (known as tenoblasts) that connect bone to muscle. Both grins81,82,84. Initial binding between fibronectin dimers and the structure and the function of this fibrillar ECM pro- integrins leads to the activation and clustering of these vide biochemical and mechanical signals that cooperate receptors82. Integrin clustering is thought to promote furin the integrated regulation of tenoblast proliferation, ther fibronectin–fibronectin intermolecular interactions, survival, differentiation and migration, which ultimately and tethering of fibronectin molecules to the cell surface feedforward to maintain physiological ECM synthesis enables cell-mediated contractility to exert force on the and assembly91. fibronectin molecule and to change its conformation81. The ECM that constitutes the tendon consists of This change in conformation exposes cryptic binding collagen (65–80% of its dry mass, primarily collagen I) sites within the fibronectin molecule and also enables and elastin (1–2% of its dry mass) embedded in a well- it to take a form that is more conducive to self-assembly hydrated, proteoglycan-rich matrix synthesized by teno into fibrils. The maturation and the movement of fibrillar blasts90. In the mature tendon, bundles of aligned colla adhesions within the cell is thought to guide the formation gen fibrils are organized into a functionally continuous of fibronectin fibrils82,85. Generation of force along these fibrous ECM, which enables them to guide motion and adhesions further controls the assembly of the fibronectin to transmit forces89,92. Throughout tendon development, network and the maturation of multiple clusters of inte- multiple independent steps of collagen fibrillogenesis grin-based adhesions brings nascent fibronectin bundles and ECM assembly take place and eventually contribute together to form a larger structure82,85. to a functionally mature tissue that has a complex ECM In addition to the key molecular players we have hierarchical structure. described in this Review, the ECM is composed of In the first of three distinct steps in tendon collagen many other classes of macromolecules that we have not fibrillogenesis, initial collagen fibril assembly occurs near included owing to space constraints. For brief descrip- the surface of tendon fibroblasts, with collagen incorpo- tions of their chemical composition and where in the rating into microfibrils, which are the lowest order fila- body they may be found, as well as excellent references mentous structures (approximately 4 nm in diameter), that may enrich understanding of these important having an axial D periodicity16,88–90,93 (FIG. 2a). Through components, see TABLE 2. spatially regulated protein secretion, tenoblasts help to guide this initial assembly step; evidence suggests that Organizing the ECM fibripositors of tenoblasts contiguously present these The ECM is not simply a collection of proteins. Different microfibrils from one cell to another along the tendon tissues have unique and specialized ECM components axis through side‑to‑side and end‑to‑end tenoblast and ECM organization, which enables each ECM to carry alignment88,94,95. Procollagen processing, homotypic out tissue-specific roles, including structural support, and heterotypic collagen interactions and other matrix the transmission of forces and macromolecular filtra- macromolecules influence the nucleation and the growth tion1,86. Three ECM landscapes, spanning diverse struc- of these microfibril intermediates (see above). Impor- tures and functions, are discussed in this section, with tantly, these micrfibrils remain semi-flexible, which a focus on how each supramolecular ECM is fabricated enables the repositioning of these structures within the and organized. With a specialized axial and longitudi- tendon tissue throughout tissue maturation and force nal organization, the first ECM assembly example con- transmission88. sists of the fibrillar collagen-rich matrix that comprises In the second step, longitudinal fibril growth occurs mechanical load-bearing tendons. To provide an example with microfibril intermediates fusing tip‑to‑tip as a result of a non-fibrillar but protein-rich ECM, the basal lamina of lateral associations between adjacent tapered ends, and is discussed as the second example. With a much more thus incorporating into continuous, mechanically func- 87,96,97 Fibrillar adhesions amorphous, GAG-rich structure, the final example dis- tional fibrils (FIG. 2a). Intermediate microfibrils are Cell–matrix connections that cussed is the perineuronal net (PNN), which is found in present in two structural forms — unipolar fibrils that are typified by their location brain ECM. The unique structure–function relationships have C and N termini and N,N‑bipolar fibrils, in which near the centre of the cell and exhibited by these ECMs are highlighted. This section the molecules have their N termini pointed towards their higher length to width 15 ratio compared with focal describes the ECMs of two specific tissues (tendon and one of the tips . End‑to‑end fusion specifically requires 97 adhesions at the periphery brain) as well as the basal lamina, which is found in many the C terminus of unipolar fibrils . SLRPs often coat the of cells. tissues. body of microfibril intermediates, which promotes

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 779

the fusion of the proteoglycan-free ends and inhibits Basal lamina. The basal lamina is a set of cell-adherent, premature side‑to‑side fusions97. organized layers of ECM molecules that form the sup- The third step involves the axial growth of the micro- porting structure for epithelial, endothelial, muscle and fibrils via incorporation of, and interactions with, other fat cells, the central nervous system and the Schwann microfibrils, as well as with microdomains of extracellular cells of peripheral nerves98–100. These ‘sheet-like’ layers macromolecules such as FACIT collagens and proteo appear early in development, segregating tissues, func- glycans (FIG. 2b). These collagen-interacting molecules tioning as macromolecular filters and providing sites regulate and restrict this axial growth and limit fibril for cell adhesion86,101,102. The basal lamina shields tissues diameter, which provides resident tendon fibroblasts from disruptive biochemical and biophysical stresses with a mechanism for regulating this ECM assembly, and mediates communication among cells in the tissue resulting in fibril structures ranging in size from 10 nm and between cells and their external environment101. to 500 nm88,89 (FIG. 2a). Although ‘basal lamina’ and ‘basement membrane’ are Fibres ranging in size from 1 μm to 20 μm form the often used interchangeably in some tissues, such as skel- next level of structure, with multiple fibrils grouped etal muscle, the basement membrane is comprised of together by an endotenon connective tissue containing two distinct structures: the basal lamina and an external, nerves, as well as blood and lymphatic vessels89. Fibres are fibrillar reticular lamina, which is directly linked to the cell bundled into 15–400 µm diameter triangular subfascicles, membrane by the basal lamina103. at which point they are compressed by surrounding struc- The basal lamina predominately assembles at or tures and stabilized via interactions with FACIT collagens near cell surfaces104. The resulting ECM architecture and SLRPs89,90. Subfascicles are bundled into 150–1000 µm consists of interconnected polymers of collagen IV and fascicles, which are further bundled into tertiary bundles VII (FIG. 4a,b), laminins bound to nidogen glycoproteins ranging from 1000 μm to 3000 µm89,90. Finally, multiple and HSPGs99. Laminin deposits on the surface of cells tertiary bundles are enclosed by the epitenon connec- self-assemble into sheets through terminal domain inter tive tissue, which, similar to the endotenon, contains the actions that are organized by cell surface molecules, vascular, lymphatic and nerve sources for the tendon such as integrins and dystroglycan, and by binding to tissue89,90 (FIG. 2a). nidogen through a single globular domain102,105 (FIG. 4c). Throughout the tissue, non-collagenous matrix, which Collagen IV networks formed through interactions consists of glycoproteins, proteoglycans, GAGs and other between its N‑and C‑terminal domains (FIG. 4a) then small molecules, surrounds and stabilizes collagen fibrils provide a three-dimensional scaffold for the association and fibres. The negatively charged groups of the sulphated of laminin–nidogen complexes into less ordered aggre- proteoglycans and GAGs bind water, which improves the gates, which interact with the collagen IV structure mechanical properties of the tissue under both shear and through binding sites on the nidogen molecule102,106. compressive stresses90. The stability and mechanical prop- These collagen IV networks are thought to stabilize the erties of this tissue are further strengthened and stabilized assembling basal lamina through covalent crosslinks101. by collagen crosslinking, which is carried out by enzymes In addition, sulphated proteoglycans, such as agrin, such as lysyl oxidase, as well as by non-enzymatic glyca- perlecan and collagen XVIII, can incorporate into the tion89. Taken together, this hierarchical fibrillar structure lamina, which provides areas of negative charge and results in tendon tissue that is capable of transmitting tethers growth factors107. These layers then recruit other forces from muscle to bone without sustaining injury91. basal lamina proteins, many of which have binding Although collagen fibrillogenesis has been intensively sites for one another, which leads to the formation of a studied for multiple tissues, many outstanding questions mature basal lamina. Collagen VII can further stabilize and areas of active research remain. Further elucidation of this structure, in addition to connecting the basal lamina the biochemical mechanisms that underlie collagen fibril to the reticular lamina to form the basement membrane, assembly will enable the development of rational therapies via interactions with collagen IV (FIG. 4b,c). The inclusion and tissue-engineering strategies for tissue repair and/or of tissue-specific isoforms of laminin and collagen IV for the regeneration of adult tissues with mechanical prop- with different proteoglycan species and accessory pro- erties that are similar to those that are observed in nor- teins provides specific properties to the basal lamina in mal development and function. In addition, through the different locations. The basal lamina seems to be pre- continued investigation and clarification of the structures dominantly attached through laminins to cell surface and functions of the various domains along the collagen receptors and sulphated glycoproteins104. These inter fibril, along with how these domains affect supramolecu- actions are responsible for the activation of intracellular lar assembly, novel insights into the roles of collagen at signalling, such as that involved in establishing cell a tissue level in fibrotic diseases may highlight new tar- polarity. Fascicles Bundles of fibres. gets to limit or to inhibit fibrosis. Improvements in high- The structure and composition of the basal lamina resolution imaging, the establishment of in vivo genetic directly contributes to its functions within tissues: to Reticular lamina knockout (and overexpression) models and the introduc- anchor the epithelium to the dermis, to guide develop- A collagen-rich layer that is tion of novel in vitro self-assembling ECMs add to the ment and differentiation and to function as a mechani- often found below the basal available range of tools that investigators can use to con- cal barrier to cells and macromolecules. The basement lamina that connects the combined basement tinue elucidating the underlying mechanisms respon- membrane is deposited early in development (soon after membrane to the sible for the different levels of supramolecular collagen implantation) and helps to establish polarity in the epi- connective tissue. assembly15. blast108. Expression of ECM components within the basal

780 | DECEMBER 2014 | VOLUME 15 www.nature.com/reviews/molcellbio

a Collagen IV b Collagen VII

7S domain NC1 domain

NC1 hexamer

7S domain dimer

Collagen VII NC1– Dimer Collagen IV NC1 c Basal lamina Collagen VII α, β and γ chains interact Collagen IV

Laminin Nidogen Nidogen mediates Assembled basal lamina Cell surface receptors laminin–collagen IV bind globular region binding

Plasma membrane Integrin DG Figure 4 | Basal lamina assembly. Two major collagens participate in the assemblyNature of the Reviews basal lamina:| Molecular collagen Cell BiologyIV and collagen VII. a | Collagen IV is characterized by an amino-terminal 7S domain and a carboxy-terminal NC1 domain — these are globular domains that facilitate the formation of hexamers and dimers, respectively. These then form highly ordered lattice structures, as shown. b | Collagen VII forms long dimers that then bend and connect with the globular regions of collagen IV lattices. c | Components of the basal lamina, and how they interact, are shown on the left-hand side. Within the basal lamina, laminins form a sheet on the cell surface by recruiting soluble laminin. This sheet is reinforced via interactions among chains of different laminin molecules and by binding between laminin globular regions and cell surface receptors (for example, integrins and dystroglycan). Collagen IV binds to laminins via nidogens and forms a scaffold on top of the laminins. This scaffold is further secured by collagen VII, which, in its unbent form (sometimes termed anchoring fibrils), can link the basal lamina to the reticular lamina to form the basement membrane. Assembled basal lamina is shown on the right-hand side. Figure parts a and b are modified from REF. 100, Nature Publishing Group. Figure part c is modified from REF. 100, Nature Publishing Group, with permission, from Yurchenco, P. D. Basement membranes: cell scaffoldings and signaling platforms. Cold Spring Harb. Perspect. Biol. 3, a004911 (2011), Cold Spring Harbor Laboratory Press and from REF. 141, Nature Publishing Group. DG, dystroglycan.

lamina is tightly controlled in order to regulate intra to bind and to retain growth factors and chemokines, cellular signalling via activation of different receptors for which have crucial roles in directing differentiation and these proteins and, in turn, to regulate cell fate determi- guiding tissue repair and regeneration103,113,114. nation at different stages of development109–111. The scaf- Although much has been elucidated about the fold structure of the basal lamina, along with its secure basal lamina, our understanding of its functions and connections to the dermis, means it provides excellent interactions with cells is by no means complete. There support for the epithelium. It is highly crosslinked and remain interesting unanswered questions regarding functions as an efficient barrier to both cellular and the regulatory processes in place for the production of molecular traffic between distinct layers of tissue. How- various components of the basal lamina, how the basal ever, it is also thin enough to permit transit in certain lamina imposes cell polarity and, through this, influ- conditions, such as when immune cells infiltrate dur- ences cell behaviour, the many roles of the proteoglycans ing inflammation112. Many of its components contain in directing cell fate, and more. Research to address binding domains that allow the basement membrane these questions uses a combination of cell biology,

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 781

useful tissue in which to highlight proteoglycan‑rich ECM properties and functions. For a full discussion of neural Hyaluronic acid ECM and its various components, see REFS 56,115,116. Link protein CSPG Early in development, the neural ECM is produced Proteoglycan and secreted by both neurons and glial cells117. The Binding to composition and organization of neural ECM is highly cell-surface CSPG dynamic throughout development and varies in different Tenascin-R 56,59,118 Cell surface parts of the adult brain . It is characterized by amor- CSPG phous protein aggregates and is similar in organization to the proteoglycan component of cartilage. In both developing and adult brains, hyaluronic acid, which is a large, linear polysaccharide, functions as a central organizing molecule around which the other components of the ECM 61,117 Neuron aggregate . Glycoproteins within the tenascin family have an oligomeric structure that facilitates this assembly, Perineuronal nets and proteoglycans with hyaluronic acid‑binding domains surrounding cell on their N termini and glycoprotein-binding domains on their C termini connect these proteins69,70,119. In contrast to the highly regulated structures in collagen-rich matrices or in the basal lamina, neural ECM aggregates form a latticed network that varies in its composition and organization throughout development120–122. Depending on the stage of development, this network can be mostly made up of neurocans, versicans, and tenascin C, as is found in the first few weeks of life, or brevican and aggrecan, as is the case in the adult brain122–124. These different proteoglycan profiles correspond with a ‘loose’ network that is made up of larger proteoglycans in early development, which is replaced and tightened with smaller proteo glycans as the brain develops69,122. These loose networks provide space for the developing brain to grow and to extend the neural network; tight networks in later life restrict growth and reinforce existing connections. An exception to the mainly unorganized nature of the brain ECM are PNNs, so‑called because they form a Figure 5 | Perineuronal nets. Perineuronal netsNature (PNNs) Reviewssurround | cellsMolecular of the Cellbrain Biology and regular structure around the cell bodies of neurons, den- act as inhibitors of both growth and migration. PNNs are lattices of hyaluronic acid, drites and the initial segments of axons122. The formation proteoglycans and tenascin molecules. The basic structure of a PNN is depicted here. of PNNs is initiated by the production of hyaluronan by Hyaluronic acid chains coated with proteoglycans are linked by trimeric tenascin R hyaluronan synthases on the inner surface of the plasma molecules to form a fairly organized structure around neurons. The chondroitin sulphate membrane125. Linear unsulphated hyaluronic acid chains proteoglycans (CSPGs) within the net bind to cell surface CSPGs and can thus control cell behaviour. Figure modified from REF. 115, Nature Publishing Group. deposited on the cell surface bind to extracellular CSPGs; these interactions are reinforced by link proteins (which are poorly characterized glycoproteins known to bind to novel approaches in synthetic chemistry and glycobiol- both hyaluronic acid and proteoglycans) and tenascin R125 ogy, and cutting-edge imaging techniques to visualize (FIG. 5). In this network, the N terminus of CSPG binds large-scale ECM structures. to hyaluronic acid and the C terminus of CSPG binds to tenascin R, forming trimeric crosslinkers that set up a sto- Proteoglycan-rich matrices of the brain. Proteoglycan- ichiometric relationship that determines the organization rich ECM can be found in cartilage and in the central of the PNN125. nervous system and lacks the distinctive superstructural The brain ECM, as with all other ECMs, provides organization that characterizes ECM surrounding other structural support for cells within the organ. It also tissues. Instead, it is characterized by amorphous aggre- actively participates in the function of the brain; for gates of HSPGs and CSPGs, crosslinked by other small example, the formation of PNNs is perfectly timed with proteoglycans. However, that is not to say that these the termination of plasticity in the developing brain, ECMs lack organization altogether. In this section, we termed the ‘critical period’ (REFS 125,126). Conversely, focus on the brain ECM as an example. delay of PNN formation or degradation of the PNN is ECM in the brain contains fibrous components such associated with prolonged, or a return of, plasticity127. as collagens, fibronectins and other molecules discussed Furthermore, the degradation of CSPGs within the PNN above. However, these proteins represent a small percent- is associated with a restoration of neural plasticity124,127. age of the brain ECM, which predominantly consists of The density and ‘tightness’ of the ECM around neurons hyaluronic acid, proteoglycans and GAGs. Thus, it is a can inhibit their growth at the end of development69.

782 | DECEMBER 2014 | VOLUME 15 www.nature.com/reviews/molcellbio

Deposition of proteoglycans provides an adhesive surface aspect of maintaining cell and tissue homeostasis. These for growth factors and then cells during axon regenera- processes follow different temporal and spatial rules in tion after injury or in neurite growth125,128–130. The differ- various tissues, with load-bearing tissues such as ten- ential expression of proteoglycans has been implicated in dons exhibiting highly ordered, fibrillar structures and various neurological conditions, including Alzheimer’s the continually evolving brain showing a less organized, disease and epilepsy65. In particular, it has been suggested GAG-rich ECM28,61,117. Therefore, disruption of the rela- that increased secretion of CSPGs after injury inhibits tive abundance of ECM proteins or their interactions regeneration and repair, which may exacerbate an initially with one another has important consequences for the small injury131,132. The manner in which ECM deposition behaviour and the fate of cells within that tissue5,133,134. and degeneration is related to neural pathogenesis is an In order to study how perturbations to ECM remodel- active area of research. As researchers unravel the com- ling influence pathological conditions such as cancer and plexities that are responsible for the context-dependent cardiovascular disease, it is necessary to understand its effects of ECM alteration in the nervous system, they ground state. Researchers have made great progress in are also making progress towards understanding how to characterizing the many components of the ECM and the mitigate long-term effects of damage to the neural ECM. manners in which these components interact with one another. Although this remains an ongoing area of inves- Conclusions and future perspectives tigation, the next step is to more thoroughly describe the The ECM has important roles in regulating cell and tissue temporal and spatial mechanisms by which ECM pro- development and function1,86. The ECM acts with speci- duction and assembly are regulated in cells and tissues. ficity in different tissues of the body by taking advantage Genetic modifications offer a way to perturb these factors of the wide variety of ECM proteins that exist, as well as in a controlled manner, and advances in in vivo imaging unique structures formed by interactions among these techniques will facilitate observations of the resulting ECM components4. This diversity results in highly dis- effects. Not only would this understanding benefit tissue tinctive ECMs that can be directly linked to the particular engineers seeking to build organs in vitro, but pinpointing roles they have in their respective tissues. As such, the such details may also identify more nuanced approaches production and assembly of the ECM is an important to therapy in cancer, inflammatory diseases and more.

1. Mecham, R. P. Overview of extracellular matrix. 14. Ricard-Blum, S. The collagen family. Cold Spring Harb. surface changes associated with fibril growth. Curr. Protoc. Cell Biol. 10, Unit 10.1 (2012). Perspect. Biol. 3, a004978 (2011). Dev. Biol. 71, 228–242 (1979). 2. Lu, P., Takai, K., Weaver, V. M. & Werb, Z. Extracellular 15. Hulmes, D. J. S. in Collagen Struct. Mech. 28. Birk, D. E. & Trelstad, R. L. Extracellular matrix degradation and remodeling in development (ed. Fratzl, P.) 15–47 (Springer, 2008). compartments in tendon morphogenesis: collagen and disease. Cold Spring Harb. Perspect. Biol. 3, This is a comprehensive review of intracellular and fibril, bundle, and macroaggregate formation. a005058 (2011). extracellular collagen biosynthesis and collagen J. Cell Biol. 103, 231–240 (1986). 3. Rozario, T. & DeSimone, D. W. The extracellular matrix fibril assembly. 29. Birk, D. E., Zycband, E. I., Winkelmann, D. A. & in development and morphogenesis: a dynamic view. 16. Hulmes, D. J. S. Building collagen molecules, fibrils, Trelstad, R. L. Collagen fibrillogenesis in situ: fibril Dev. Biol. 341, 126–140 (2010). and suprafibrillar structures. J. Struct. Biol. 137, 2–10 segments are intermediates in matrix assembly. 4. Naba, A. et al. The matrisome: in silico definition and (2002). Proc. Natl Acad. Sci. USA 86, 4549–4553 (1989). in vivo characterization by proteomics of normal and 17. Myllyharju, J. Intracellular post-translational 30. Canty, E. G. et al. Coalignment of plasma membrane tumor extracellular matrices. Mol. Cell. Proteomics 11, modifications of collagens. Top Curr. Chem. 247, channels and protrusions (fibripositors) specifies the M111.014647 (2012). 115–147 (2005). parallelism of tendon. J. Cell Biol. 165, 553–563 5. Jakeman, L. B., Williams, K. E. & Brautigam, B. In the 18. Bella, J., Eaton, M., Brodsky, B. & Berman, H. M. (2004). presence of danger: the extracellular matrix defensive Crystal and molecular structure of a collagen-like This paper shows how embryonic fibroblasts response to central nervous system injury. Neural peptide at 1.9 A resolution. Science 266, 75–81 produce an ECM that is rich in elongated, parallel, Regen. Res. 9, 377–384 (2014). (1994). collagen fibrils, through the initiation of collagen 6. Fratzl, P. et al. Fibrillar structure and mechanical 19. Persikov, A. V., Ramshaw, J. A. M., Kirkpatrick, A. & fibrillogenesis by targeting collagen properties of collagen. J. Struct. Biol. 122, 119–122 Brodsky, B. Electrostatic interactions involving lysine fibril-containing Golgi-to-plasma membrane (1998). make major contributions to collagen triple-helix carriers (GPCs) to plasma membrane protrusions. 7. Bonnans, C., Chou, J. & Werb, Z. Remodeling the stability. Biochemistry 44, 1414–1422 (2005). 31. Starborg, T. et al. Using transmission electron extracellular matrix in development and diseases. 20. Bourhis, J.‑M. et al. Structural basis of fibrillar microscopy and 3View to determine collagen fibril size Nature Rev. Cell Mol. Biol. http://dx.doi.org/10.1038/ collagen trimerization and related genetic disorders. and three-dimensional organization. Nature Protoc. 8, nrm3904 (2014). Nature Struct. Mol. Biol. 19, 1031–1036 (2012). 1433–1448 (2013). 8. Humphrey, J. D., Dufresne, E. R. & Schwartz, A. M. This paper elucidates the crystal structure of the 32. Kalson, N. S. et al. Nonmuscle myosin II powered Mechanotransduction and extracellular matrix procollagen C propeptide to reveal an exquisite transport of newly formed collagen fibrils at the homeostasis. Nature Rev. Cell Mol. Biol. http://dx.doi. structural mechanism of α-chain selectivity during plasma membrane. Proc. Natl Acad. Sci. USA 110, org/10.1038/nrm3896 (2014). intracellular trimerization. E4743–E4752 (2013). 9. Rowe, R. G. & Weiss, S. J. Breaching the basement 21. McLaughlin, S. H. & Bulleid, N. J. Molecular 33. Porter, S., Clark, I. M., Kevorkian, L. & Edwards, D. R. membrane: who, when and how? Trends Cell Biol. 18, recognition in procollagen chain assembly. Matrix Biol. The ADAMTS metalloproteinases. Biochem. J. 386, 560–574 (2008). 16, 369–377 (1998). 15–27 (2005). 10. Schaefer, L. & Schaefer, R. M. Proteoglycans: from 22. Lees, J. F., Tasab, M. & Bulleid, N. J. Identification of 34. Hopkins, D. R., Keles, S. & Greenspan, D. S. The bone structural compounds to signaling molecules. the molecular recognition sequence which determines morphogenetic protein 1/tolloid-like Cell Tissue Res. 339, 237–246 (2010). the type-specific assembly of procollagen. EMBO J. metalloproteinases. Matrix Biol. 26, 508–523 This paper shows the biological structure and 16, 908–916 (1997). (2007). functions of the primarily extracellular SLRPs and 23. Boudko, S. P., Engel, J. & Bächinger, H. P. The crucial 35. Seidah, N. G. & Prat, A. Precursor convertases in the highlights SLRP-associated genetic diseases and role of trimerization domains in collagen folding. secretory pathway, cytosol and extracellular milieu. signalling events. Int. J. Biochem. Cell Biol. 44, 21–32 (2012). Essays Biochem. 38, 79–94 (2002). 11. Myllyharju, J. & Kivirikko, K. I. Collagens, modifying 24. Ellgaard, L. & Ruddock, L. W. The human protein 36. Pappano, W. N., Steiglitz, B. M., Scott, I. C., enzymes and their mutations in humans, flies and disulphide isomerase family: substrate interactions and Keene, D. R. & Greenspan, D. S. Use of BMP1/TLL1 worms. Trends Genet. 20, 33–43 (2004). functional properties. EMBO Rep. 6, 28–32 (2005). doubly homozygous null mice and proteomics to 12. Ricard-Blum, S. & Ruggiero, F. The collagen 25. Wess, T. J. Collagen fibril form and function. identify and validate in vivo substrates of bone superfamily: from the extracellular matrix to the cell Adv. Protein Chem. 70, 341–374 (2005). morphogenetic protein 1/tolloid-like membrane. Pathol. Biol. (Paris). 53, 430–442 26. Shoulders, M. D. & Raines, R. T. Collagen structure metalloproteinases. Mol. Cell. Biol. 23, 4428–4438 (2005). and stability. Annu. Rev. Biochem. 78, 929–958 (2003). 13. Brodsky, B. & Persikov, A. V. Molecular structure of the (2009). 37. Kadler, K. E., Holmes, D. F., Trotter, J. A. & collagen triple helix. Adv. Protein Chem. 70, 301–339 27. Trelstad, R. L. & Hayashi, K. Tendon collagen Chapman, J. A. Collagen fibril formation. Biochem. J. (2005). fibrillogenesis: intracellular subassemblies and cell 316, 1–11 (1996).

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 783

38. Wenstrup, R. J. et al. Type V collagen controls the 65. Wade, A., McKinney, A. & Phillips, J. J. Matrix 88. Zhang, G. et al. Development of tendon structure and initiation of collagen fibril assembly. J. Biol. Chem. regulators in neural stem cell functions. Biochim. function: regulation of collagen fibrillogenesis. 279, 53331–53337 (2004). Biophys. Acta http://dx.doi.org/10.1016/j. J. Musculoskelet. Neuronal Interact. 5, 5–21 39. Chapman, J. A. The staining pattern of collagen fibrils. bbagen.2014.01.017 (2014). (2005). I. An analysis of electron micrographs. Connect. Tissue 66. Johnson, K. G. et al. Axonal heparan sulfate 89. Wang, J. H.‑C. Mechanobiology of tendon. J. Biomech. Res. 2, 137–150 (1974). proteoglycans regulate the distribution and efficiency 39, 1563–1582 (2006). 40. Olsen, B. R. Electron microscope studies on collagen. I. of the repellent slit during midline axon guidance. 90. Kannus, P. Structure of the tendon connective Native collagen fibrils. Z. Zellforsch. Mikrosk. Anat. 59, Curr. Biol. 14, 499–504 (2004). tissue. Scand. J. Med. Sci. Sports 10, 312–320 184–198 (1963). 67. Häcker, U., Nybakken, K. & Perrimon, N. Heparan (2000). 41. Petruska, J. A. & Hodge, A. J. A subunit model for the sulphate proteoglycans: the sweet side of 91. Tresoldi, I. et al. Tendon’s ultrastructure. Muscles. tropocollagen macromolecule. Proc. Natl Acad. Sci. development. Nature Rev. Mol. Cell. Biol. 6, 530–541 Ligaments Tendons J. 3, 2–6 (2013). USA 51, 871–876 (1964). (2005). 92. Mienaltowski, M. J. & Birk, D. E. Structure, physiology, 42. Hulmes, D. J. & Miller, A. Quasi-hexagonal molecular 68. Maeda, N., Ishii, M., Nishimura, K. & Kamimura, K. and biochemistry of collagens. Adv. Exp. Med. Biol. packing in collagen fibrils. Nature 282, 878–880 Functions of chondroitin sulfate and heparan sulfate in 802, 5–29 (2014). (1979). the developing brain. Neurochem. Res. 36, 93. Craig, A. S., Birtles, M. J., Conway, J. F. & Parry, D. A. 43. Hulmes, D. J., Wess, T. J., Prockop, D. J. & Fratzl, P. 1228–1240 (2011). An estimate of the mean length of collagen fibrils in rat Radial packing, order, and disorder in collagen fibrils. 69. Yamaguchi, Y. Lecticans: organizers of the brain tail-tendon as a function of age. Connect. Tissue Res. Biophys. J. 68, 1661–1670 (1995). extracellular matrix. Cell. Mol. Life Sci. 57, 276–289 19, 51–62 (1989). 44. Orgel, J. P. R. O., Irving, T. C., Miller, A. & Wess, T. J. (2000). 94. Tozer, S. & Duprez, D. Tendon and ligament: Microfibrillar structure of type I collagen in situ. 70. Lundell, A. et al. Structural basis for interactions development, repair and disease. Birth Defects Res. Proc. Natl Acad. Sci. USA 103, 9001–9005 between tenascins and lectican C‑type lectin domains: C. Embryo Today 75, 226–236 (2005). (2006). evidence for a crosslinking role for tenascins. 95. Richardson, S. H. et al. Tendon development requires 45. Schmitt, F. O., Hall, C. E. & Jakus, M. A. Electron Structure 12, 1495–1506 (2004). regulation of cell condensation and cell shape via microscope investigations of the structure of collagen. 71. Dellett, M., Hu, W., Papadaki, V. & Ohnuma, S. Small cadherin-11‑mediated cell-cell junctions. Mol. Cell. J. Cell. Comp. Physiol. 20, 11–33 (1942). leucine rich proteoglycan family regulates multiple Biol. 27, 6218–6228 (2007). 46. Bruckner, P. Suprastructures of extracellular matrices: signalling pathways in neural development and This paper describes how the precise regulation of paradigms of functions controlled by aggregates maintenance. Dev. Growth Differ. 54, 327–340 cell shape via cell–cell junctions and the coaxial rather than molecules. Cell Tissue Res. 339, 7–18 (2012). alignment of plasma membrane channels in (2010). 72. McEwan, P. A., Scott, P. G., Bishop, P. N. & Bella, J. longitudinally stacked cells drives tendon 47. Ameye, L. & Young, M. F. Mice deficient in small Structural correlations in the family of small leucine- formation in the developing limb. leucine-rich proteoglycans: novel in vivo models for rich repeat proteins and proteoglycans. J. Struct. Biol. 96. Sasaki, N. & Odajima, S. Elongation mechanism of osteoporosis, osteoarthritis, Ehlers–Danlos syndrome, 155, 294–305 (2006). collagen fibrils and force-strain relations of tendon at muscular dystrophy, and corneal diseases. 73. Schaefer, L. & Iozzo, R. V. Biological functions of the each level of structural hierarchy. J. Biomech. 29, Glycobiology 12, 107R–116R (2002). small leucine-rich proteoglycans: from genetics to 1131–1136 (1996). 48. Kalamajski, S. & Oldberg, A. The role of small leucine- signal transduction. J. Biol. Chem. 283, 97. Graham, H. K., Holmes, D. F., Watson, R. B. & rich proteoglycans in collagen fibrillogenesis. 21305–21309 (2008). Kadler, K. E. Identification of collagen fibril fusion Matrix Biol. 29, 248–253 (2010). 74. Chen, S. & Birk, D. E. The regulatory roles of small during vertebrate tendon morphogenesis. The process 49. Molnar, J. et al. Structural and functional diversity of leucine-rich proteoglycans in extracellular matrix relies on unipolar fibrils and is regulated by collagen- lysyl oxidase and the LOX-like proteins. Biochim. assembly. FEBS J. 280, 2120–2137 (2013). proteoglycan interaction. J. Mol. Biol. 295, 891–902 Biophys. Acta 1647, 220–224 (2003). 75. Oohira, A., Matsui, F., Tokita, Y., Yamauchi, S. & (2000). 50. Eyre, D. R., Weis, M. A. & Wu, J.‑J. Advances in Aono, S. Molecular interactions of neural chondroitin 98. Paulsson, M. Basement membrane proteins: structure, collagen cross-link analysis. Methods 45, 65–74 sulfate proteoglycans in the brain development. assembly, and cellular interactions. Crit. Rev. Biochem. (2008). Arch. Biochem. Biophys. 374, 24–34 (2000). Mol. Biol. 27, 93–127 (1992). 51. Eyre, D. R. & Glimcher, M. J. Collagen cross-linking. 76. Timpl, R. & Brown, J. C. The laminins. Matrix Biol. 14, 99. Martin Rupert, G. R. T. Laminin and other basement Isolation of cross-linked peptides from collagen of 275–281 (1994). membrane components. Annu. Rev. Cell Biol. 3, chicken bone. Biochem. J. 135, 393–403 (1973). 77. Aumailley, M. et al. A simplified laminin nomenclature. 57–85 (1987). 52. Lucero, H. A. & Kagan, H. M. Lysyl oxidase: an Matrix Biol. 24, 326–332 (2005). 100. Kalluri, R. Basement membranes: structure, assembly oxidative enzyme and effector of cell function. 78. Miner, J. H. & Yurchenco, P. D. Laminin functions in and role in tumour angiogenesis. Nature Rev. Cancer Cell. Mol. Life Sci. 63, 2304–2316 (2006). tissue morphogenesis. Annu. Rev. Cell Dev. Biol. 20, 3, 422–433 (2003). 53. Rucker, R. B. & Murray, J. Cross-linking amino acids in 255–284 (2004). This is an excellent review highlighting the collagen and elastin. Am. J. Clin. Nutr. 31, This paper discusses how different laminin proteins constituents, structure and assembly of the 1221–1236 (1978). affect invertebrate and vertebrate tissue basement membrane, with emphasis on 54. Orgel, J. P., Wess, T. J. & Miller, A. The in situ morphogenesis through the induction and the angiogenesis and vascular basement membranes. conformation and axial location of the intermolecular maintenance of cell polarity, the establishment of 101. Yurchenco, P. D. Basement membranes: cell cross-linked non-helical telopeptides of type I collagen. barriers between tissue compartments and the scaffoldings and signaling platforms. Cold Spring Harb. Structure 8, 137–142 (2000). protection of adherent cells from anoikis. Perspect. Biol. 3, a004911 (2011). 55. O’Leary, L. E. R., Fallas, J. A., Bakota, E. L., 79. Mercurio, A. M. & Shaw, L. M. Laminin binding 102. Yurchenco, P. D. & Schittny, J. C. Molecular architecture Kang, M. K. & Hartgerink, J. D. Multi-hierarchical self- proteins. Bioessays 13, 469–473 (1991). of basement membranes. FASEB J. 4, 1577–1590 assembly of a collagen mimetic peptide from triple 80. Yurchenco, P. D., Smirnov, S. & Mathus, T. Analysis of (1990). helix to nanofibre and hydrogel. Nature Chem. 3, basement membrane self-assembly and cellular 103. Sanes, J. R. The basement membrane/basal lamina of 821–828 (2011). interactions with native and recombinant skeletal muscle. J. Biol. Chem. 278, 12601–12604 56. Bandtlow, C. E. & Zimmermann, D. R. Proteoglycans glycoproteins. Methods Cell Biol. 69, 111–144 (2003). in the developing brain: new conceptual insights (2002). 104. Yurchenco, P. D. & Patton, B. L. Developmental and for old proteins. Physiol. Rev. 80, 1267–1290 81. Schwarzbauer, J. E. & DeSimone, D. W. Fibronectins, pathogenic mechanisms of basement membrane (2000). their fibrillogenesis, and in vivo functions. Cold Spring assembly. Curr. Pharm. Des. 15, 1277–1294 57. Knudson, C. B. & Knudson, W. Cartilage Harb. Perspect. Biol. 3, a005041 (2011). (2009). proteoglycans. Semin. Cell Dev. Biol. 12, 69–78 82. Singh, P., Carraher, C. & Schwarzbauer, J. E. Assembly 105. Takagi, J., Yang, Y., Liu, J.‑H., Wang, J.‑H. & (2001). of fibronectin extracellular matrix. Annu. Rev. Cell Dev. Springer, T. A. Complex between nidogen and laminin 58. Cui, H., Freeman, C., Jacobson, G. A. & Small, D. H. Biol. 26, 397–419 (2010). fragments reveals a paradigmatic β-propeller interface. Proteoglycans in the central nervous system: role in This is an excellent review highlighting the major Nature 424, 969–974 (2003). development, neural repair, and Alzheimer’s disease. steps, molecular interactions and cellular 106. Pöschl, E. et al. Collagen IV is essential for basement IUBMB Life 65, 108–120 (2013). mechanisms involved in assembling fibronectin into membrane stability but dispensable for initiation of its 59. Schwartz, N. B. & Domowicz, M. Proteoglycans in a fibrillar matrix. assembly during early development. Development brain development. Glycoconj. J. 21, 329–341 83. Hynes, R. O. in Guidebook to the Extracellular Matrix, 131, 1619–1628 (2004). (2004). Anchor and Adhesion Proteins. 2nd edn (eds Kreis, T. 107. Iozzo, R. V., Zoeller, J. J. & Nyström, A. Basement 60. Zhang, L. (ed.) Glycosaminoglycans in Development, & Vale, R.) 422–425 (Oxford University Press, 1999). membrane proteoglycans: modulators Par Excellence Health and Disease (Academic, 2010). 84. García, A. J., Schwarzbauer, J. E. & Boettiger, D. of cancer growth and angiogenesis. Mol. Cells 27, 61. Deepa, S. S. et al. Composition of perineuronal net Distinct activation states of α5β1 integrin show 503–513 (2009). extracellular matrix in rat brain: a different differential binding to RGD and synergy domains of 108. Costello, I., Biondi, C. A., Taylor, J. M., Bikoff, E. K. & disaccharide composition for the net-associated fibronectin. Biochemistry 41, 9063–9069 (2002). Robertson, E. J. Smad4‑dependent pathways control proteoglycans. J. Biol. Chem. 281, 17789–17800 85. Ilic´, D. et al. FAK promotes organization of fibronectin basement membrane deposition and endodermal cell (2006). matrix and fibrillar adhesions. J. Cell Sci. 117, migration at early stages of mouse development. 62. Simon Davis, D. A. & Parish, C. R. Heparan sulfate: a 177–187 (2004). BMC Dev. Biol. 9, 54 (2009). ubiquitous glycosaminoglycan with multiple roles in 86. Hay, E. D. Extracellular matrix alters epithelial 109. Wiradjaja, F., DiTommaso, T. & Smyth, I. Basement immunity. Front. Immunol. 4, 470 (2013). differentiation. Curr. Opin. Cell Biol. 5, 1029–1035 membranes in development and disease. Birth Defects 63. Bishop, J. R., Schuksz, M. & Esko, J. D. Heparan (1993). Res. C. Embryo Today 90, 8–31 (2010). sulphate proteoglycans fine-tune mammalian 87. Birk, D. E., Zycband, E. I., Woodruff, S., 110. Paralkar, V. M., Vukicevic, S. & Reddi, A. H. physiology. Nature 446, 1030–1037 (2007). Winkelmann, D. A. & Trelstad, R. L. Collagen Transforming growth factor-β type 1 binds to 64. Esko, J. D., Kimata, K. & Lindahl, U. in Essentials of fibrillogenesis in situ: fibril segments become long collagen IV of basement membrane matrix: Glycobiology. 2nd edn Ch. 16 (Cold Spring Harbor, fibrils as the developing tendon matures. Dev. Dyn. Implications for development. Dev. Biol. 143, 2009). 208, 291–298 (1997). 303–308 (1991).

784 | DECEMBER 2014 | VOLUME 15 www.nature.com/reviews/molcellbio

111. Streuli, C. H. Control of mammary epithelial 132. Bartus, K., James, N. D., Bosch, K. D. & Bradbury, E. J. 154. Pepe, A., Bochicchio, B. & Tamburro, A. M. differentiation: basement membrane induces tissue- Chondroitin sulphate proteoglycans: key modulators of Supramolecular organization of elastin and elastin- specific gene expression in the absence of cell–cell spinal cord and brain plasticity. Exp. Neurol. 235, related nanostructured biopolymers. Nanomed. 2, interaction and morphological polarity. J. Cell Biol. 5–17 (2012). 203–218 (2007). 115, 1383–1395 (1991). 133. Li, A.‑H., Liu, P. P., Villarreal, F. J. & Garcia, R. A. 155. Wagenseil, J. E. & Mecham, R. P. New insights into 112. Camelo, A., Dunmore, R., Sleeman, M. A. & Dynamic changes in myocardial matrix and relevance elastic fiber assembly. Birth Defects Res. C. Embryo Clarke, D. L. The epithelium in idiopathic pulmonary to disease: translational perspectives. Circ. Res. 114, Today 81, 229–240 (2007). fibrosis: breaking the barrier. Front. Pharmacol. 4, 916–927 (2014). 156. Weisel, J. W. Fibrinogen and fibrin. Adv. Protein Chem. 173 (2014). 134. Page-McCaw, A., Ewald, A. J. & Werb, Z. Matrix 70, 247–299 (2005). 113. Torricelli, A. A. M., Singh, V., Santhiago, M. R. & metalloproteinases and the regulation of tissue 157. Mosesson, M. W., Siebenlist, K. R. & Meh, D. A. Wilson, S. E. The corneal epithelial basement remodelling. Nature Rev. Mol. Cell Biol. 8, 221–233 The structure and biological features of fibrinogen and membrane: structure, function, and disease. Invest. (2007). fibrin. Ann. NY Acad. Sci. 936, 11–30 (2001). Ophthalmol. Vis. Sci. 54, 6390–6400 (2013). 135. Fan, D., Takawale, A., Lee, J. & Kassiri, Z. Cardiac 158. Ramirez, F., Sakai, L. Y., Dietz, H. C. & Rifkin, D. B. 114. Brownell, A. G. & Slavkin, H. C. Role of basal lamina in fibroblasts, fibrosis and extracellular matrix remodeling Fibrillin microfibrils: multipurpose extracellular tissue interactions. Ren. Physiol. 3, 193–204 in heart disease. Fibrogen. Tissue Repair 5, 15 (2012). networks in organismal physiology. Physiol. Genom. (1980). 136. Nickla, H., Klug, W. S. & Cummings, M. R. Concepts of 19, 151–154 (2004). 115. Lau, L. W., Cua, R., Keough, M. B., Haylock-Jacobs, S. Genetics. (Prentice Hall, 1997). 159. Kielty, C. M. et al. Fibrillin: from microfibril assembly & Yong, V. W. Pathophysiology of the brain 137. Jäälinoja, J., Ylöstalo, J., Beckett, W., Hulmes, D. J. S. to biomechanical function. Phil. Trans. R. Soc. 357, extracellular matrix: a new target for remyelination. & Ala-Kokko, L. Trimerization of collagen IX α-chains 207–217 (2002). Nature Rev. Neurosci. 14, 722–729 (2013). does not require the presence of the COL1 and NC1 160. Timpl, R., Sasaki, T., Kostka, G. & Chu, M.‑L. Fibulins: This paper shows the organization of the brain domains. Biochem. J. 409, 545–554 (2008). a versatile family of extracellular matrix proteins. ECM, emphasizing the pathophysiological roles of 138. Liu, Y., Ramanath, H. S. & Wang, D.‑A. Tendon tissue Nature Rev. Mol. Cell. Biol. 4, 479–489 (2003). CSPGs after neural injury. engineering using scaffold enhancing strategies. 161. De Vega, S., Iwamoto, T. & Yamada, Y. Fibulins: 116. Plantman, S. Proregenerative properties of ECM Trends Biotechnol. 26, 201–209 (2008). multiple roles in matrix structures and tissue functions. molecules. Biomed. Res. Int. 2013, 981695 139. Edwards, I. J. Proteoglycans in prostate cancer. Cell. Mol. Life Sci. 66, 1890–1902 (2009). (2013). Nature Rev. Urol. 9, 196–206 (2012). 162. Cirulli, V. & Yebra, M. Netrins: beyond the brain. 117. Ruoslahti, E. Brain extracellular matrix. Glycobiology 140. Rudd, T. R. et al. A key biological signalling system Nature Rev. Mol. Cell Biol. 8, 296–306 (2007). 6, 489–492 (1996). promises new routes to intervention in medically 163. Lai Wing Sun, K., Correia, J. P. & Kennedy, T. E. 118. Barros, C. S., Franco, S. J. & Müller, U. Extracellular important processes. Diamond Science [online], Netrins: versatile extracellular cues with diverse matrix: functions in the nervous system. Cold Spring http://www.diamond.ac.uk/Science/Research/ functions. Development 138, 2153–2169 (2011). Harb. Perspect. Biol. 3, a005108 (2011). Highlights/study2.html (2014). 164. Bradshaw, A. D. Diverse biological functions of the 119. Chiquet-Ehrismann, R. & Tucker, R. P. Tenascins and 141. Marinkovich, M. P. Tumour microenvironment: laminin SPARC family of proteins. Int. J. Biochem. Cell Biol. 44, the importance of adhesion modulation. Cold Spring 332 in squamous-cell carcinoma. Nature Rev. Cancer 480–488 (2012). Harb. Perspect. Biol. 3, a004960 (2011). 7, 370–380 (2007). 165. Brekken, R. A. & Sage, E. H. SPARC, a matricellular 120. Dityatev, A. Remodeling of extracellular matrix and 142. Bezakova, G. & Ruegg, M. A. New insights into the protein: at the crossroads of cell-matrix epileptogenesis. Epilepsia 51 (Suppl. 3), 61–65 roles of agrin. Nature Rev. Mol. Cell. Biol. 4, communication. Matrix Biol. 19, 816–827 (2001). (2010). 295–308 (2003). 166. Giachelli, C. M. & Steitz, S. Osteopontin: a versatile 121. Lajtha, A. & Banik, N. Handbook of Neurochemistry 143. Meier, T. & Wallace, B. G. Formation of the regulator of inflammation and biomineralization. and Molecular Neurobiology: Neural Protein neuromuscular junction: molecules and mechanisms. Matrix Biol. 19, 615–622 (2000). Metabolism and Function. Vol. 3 (Springer, 2007). Bioessays 20, 819–829 (1998). 167. Hunter, G. K. Role of osteopontin in modulation of 122. Howell, M. D. & Gottschall, P. E. Lectican 144. McMahan, U. J. et al. Agrin isoforms and their role in hydroxyapatite formation. Calcif. Tissue Int. 93, proteoglycans, their cleaving metalloproteinases, and synaptogenesis. Curr. Opin. Cell Biol. 4, 869–874 348–354 (2013). plasticity in the central nervous system extracellular (1992). 168. Fatemi, S. H. Reelin glycoprotein: structure, biology microenvironment. Neuroscience 217, 6–18 145. Fisher, L. W., Torchia, D. A., Fohr, B., Young, M. F. & and roles in health and disease. Mol. Psychiatry 10, (2012). Fedarko, N. S. Flexible structures of SIBLING proteins, 251–257 (2005). 123. Carulli, D., Rhodes, K. E. & Fawcett, J. W. Upregulation bone sialoprotein, and osteopontin. Biochem. Biophys. 169. Magdaleno, S. M. & Curran, T. Brain development: of aggrecan, link protein 1, and hyaluronan synthases Res. Commun. 280, 460–465 (2001). integrins and the Reelin pathway. Curr. Biol. 11, during formation of perineuronal nets in the rat 146. Bellahcène, A., Castronovo, V., Ogbureke, K. U. E., R1032–R1035 (2001). cerebellum. J. Comp. Neurol. 501, 83–94 (2007). Fisher, L. W. & Fedarko, N. S. Small integrin-binding 170. Vincent, T. L., Woolfson, D. N. & Adams, J. C. 124. Frischknecht, R. & Gundelfinger, E. D. The brain’s ligand N‑linked glycoproteins (SIBLINGs): Prediction and analysis of higher-order coiled-coils: extracellular matrix and its role in synaptic plasticity. multifunctional proteins in cancer. Nature Rev. Cancer insights from proteins of the extracellular matrix, Adv. Exp. Med. Biol. 970, 153–171 (2012). 8, 212–226 (2008). tenascins and thrombospondins. Int. J. Biochem. Cell 125. Kwok, J. C. F., Dick, G., Wang, D. & Fawcett, J. W. 147. Ogata, Y. Bone sialoprotein and its transcriptional Biol. 45, 2392–2401 (2013). Extracellular matrix and perineuronal nets in CNS regulatory mechanism. J. Periodontal Res. 43, 171. Tucker, R. P. & Chiquet-Ehrismann, R. The regulation of repair. Dev. Neurobiol. 71, 1073–1089 (2011). 127–135 (2008). tenascin expression by tissue microenvironments. 126. Wang, D. & Fawcett, J. The perineuronal net and the 148. Deák, F., Wagener, R., Kiss, I. & Paulsson, M. The Biochim. Biophys. Acta 1793, 888–892 (2009). control of CNS plasticity. Cell Tissue Res. 349, matrilins: a novel family of oligomeric extracellular 172. Leavesley, D. I. et al. Vitronectin — master controller 147–160 (2012). matrix proteins. Matrix Biol. 18, 55–64 (1999). or micromanager? IUBMB Life 65, 807–818 (2013). 127. Romberg, C. et al. Depletion of perineuronal nets 149. Klatt, A. R., Becker, A.‑K. A., Neacsu, C. D., 173. Preissner, K. T. & Reuning, U. Vitronectin in vascular enhances recognition memory and long-term Paulsson, M. & Wagener, R. The matrilins: modulators context: facets of a multitalented matricellular protein. depression in the perirhinal cortex. J. Neurosci. 33, of extracellular matrix assembly. Int. J. Biochem. Cell Semin. Thromb. Hemost. 37, 408–424 (2011). 7057–7065 (2013). Biol. 43, 320–330 (2011). 128. Grimpe, B. & Silver, J. The extracellular matrix in axon 150. Wagener, R. et al. The matrilins — adaptor proteins in Acknowledgements regeneration. Prog. Brain Res. 137, 333–349 the extracellular matrix. FEBS Lett. 579, 3323–3329 The authors apologize to all colleagues whose work cannot be (2002). (2005). cited owing to space limitations. This work was supported by 129. Fitch, M. T. & Silver, J. Glial cell extracellular matrix: 151. Chen, Q., Johnson, D. M., Haudenschild, D. R. & DOD Breast Cancer Research Program (BCRP) grant boundaries for axon growth in development and Goetinck, P. F. Cartilage matrix protein: expression W81XWH-07‑1‑0538 (to J.K.M.), US National Science Foun- regeneration. Cell Tissue Res. 290, 379–384 patterns in chicken, mouse, and human. Ann. NY Acad. dation (NSF) Graduate Research Fellowship (to G.O.), DOD (1997). Sci. 785, 238–240 (1996). BCRP grants W81XWH-05‑1‑0330 and W81XWH-13‑1‑0216 130. Ishii, M. & Maeda, N. Oversulfated chondroitin sulfate 152. Mithieux, S. M. & Weiss, A. S. Elastin. Adv. Protein (to V.M.W.), US National Institutes of Health National Cancer plays critical roles in the neuronal migration in the Chem. 70, 437–461 (2005). Institute (NCI) grants R01 CA138818, U54 CA143836, R01 cerebral cortex. J. Biol. Chem. 283, 32610–32620 153. Halper, J. & Kjaer, M. Basic components of connective CA085492 and U01 ES019458 (to V.M.W.), and Susan G. (2008). tissues and extracellular matrix: elastin, fibrillin, Komen grant KG110560PP (to V.M.W.). 131. Sharma, K., Selzer, M. E. & Li, S. Scar-mediated fibulins, fibrinogen, fibronectin, laminin, tenascins and inhibition and CSPG receptors in the CNS. Exp. Neurol. thrombospondins. Adv. Exp. Med. Biol. 802, 31–47 Competing interests statement 237, 370–378 (2012). (2014). The authors declare no competing interests.

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 15 | DECEMBER 2014 | 785