The Extracellular Matrix

Collagens are the major proteins in the ECM

The collagens are a family of proteins that comprise about 30% of total protein mass in the body. As the primary structural components of the ECM in connective tissues, collagens have an important role in tissue architecture and integrity, and in mediating a variety of cell–cell and cell–matrix interactions. To date, more than 25 different types of collagens have been identified. They are composed of related, but distinct, peptide chains and vary greatly in their distribution, organization and function in tissues.

The left-handed triple helical structure of collagen is unique among proteins

The collagens are heterotrimeric proteins composed of three individual peptide chains. The structural hallmark of collagens is their triple-helical structure, formed by folding of the three component peptide chains. X-ray diffraction analysis indicates that three left-handed helical chains are wrapped around one another in a rope-like fashion, to form a superhelix structure (Fig. 29.1). The left-handed helix is more extended than the α-helix of globular proteins, having nearly twice the rise per turn and only three, rather than 3.6, amino acids per turn of the helix. Every third amino acid is glycine, because only this amino acid, with the smallest side chain, fits into the crowded central core. The characteristic, repeating sequence of collagen is Gly-X-Y, where X and Y can be any amino acid but most often X is proline and Y is hydroxyproline. Because of their restricted rotation and bulk, proline and hydroxyproline confer rigidity to the helix. The intra- and interchain helices are stabilized by hydrogen bonds, largely between peptide NH and CO groups. The side chains of the X and Y amino acids point outward from the helix, and thus are on the surface of the protein, where they form lateral interactions with other triple helices or proteins.

Fibrillar collagens provide tensile strength to tendons, ligaments and skin

Fibril-forming collagens include types I, II, III, V, and XI (see Table 29.1). Collagen fibrils can be formed from a mixture of different fibrillar collagens. For instance, dermal collagen fibrils are hybrids of type I and type III collagen, and fibrils in corneal stroma are hybrids of type I and type IV collagen. Type I is the most abundant fibrillar collagen and occurs in a wide variety of tissues; others have a more limited tissue distribution (see Table 29.1). Type I and related fibrillar collagens form well-organized, banded fibrils and provide high-tensile strength to skin, tendons, and ligaments. As indicated above, collagens are heterotrimers composed of three α-helical peptide chains (see Fig. 29.1). The type I collagen heterotrimer is composed of two α1(I) chains and one α2(I) chain. Each of these peptide chains contains about 1000 amino acids and has a triple-helical domain structure along almost the entire length of the molecule. The collagen fibrils are formed by lateral association of triple helices in a ‘quarter-staggered’ alignment in which each molecule is displaced by about one-quarter of its length relative to its nearest neighbor (Fig. 29.2). The quarter-staggered array is responsible for the banded appearance of collagen fibrils in connective tissues. The fibrils are stabilized by both noncovalent forces and interchain crosslinks derived from lysine residues (see below).

Non-fibrillar, lattice-forming collagens are major structural components of basement membranes

Nonfibrillar collagens are a heterogeneous group containing triple-helical segments of variable length, interrupted by one or more intervening nonhelical (noncollagenous) segments. This group includes basement membrane collagens (the type IV family), fibril-associated collagens with interrupted triple helices (FACITs), and collagens with multiple triple-helical domains with interruptions, known as multiplexins. Nonfibrillar collagens associate with the fibrillar collagens, forming microfibrils and network or mesh-like structures.

Basement membranes are relatively thin layers of ECM found on the basal aspect of epithelial cells and surrounding some other cell types including myocytes, Schwann cells and adipocytes. The basement membrane has a number of functions including anchorage of cells to surrounding connective tissue and filtration.

Type IV collagen assembles into a flexible mesh-like network. This collagen contains a long triple-helical domain interrupted by short noncollagenous sequences. These interruptions in the helical domain block continued association of two triple helices, oblige them to find another partner, and thus contribute to formation of a lattice-type structure. In the kidney, the thickened basement membrane (100–200 nm thick) on the basal aspect of the glomuerular capillary endothelial cells plays an essential role as a macromolecular filter (see Chapter 23). The meshwork of ECM proteins in the basement membrane restricts the passage of large molecules from the blood into the urine. In addition, the inclusion of negatively charged proteoglycans (described later in this chapter) in the glomerular basement membrane restricts the passage of charged molecules. Anomalies in type IV collagen in the glomerular basement membrane result in several glomerular diseases including Goodpasture's syndrome and Alport syndrome. Goodpasture's syndrome is a rare autoimmune disease caused by the production of antibodies that specifically bind to type IV collagen of basement membranes. This inflammatory condition leads to progressive worsening of basement membrane function in the kidney and sometimes in the lung. Alport syndrome results from mutations in the type IV collagen chains which cause defective collagen scaffold assembly within the basement membrane. The symptoms of both of these syndromes progress from blood in the urine (hematuria) to urine containing excessive protein (proteinuria) and eventually to kidney failure.

Collagen synthesis begins in the rough endoplasmic reticulum (RER)

After synthesis in the RER, the nascent collagen polypeptide undergoes extensive modification, first in the RER, then in the Golgi apparatus, and finally in the extracellular space, where it is modified to a mature extracellular collagen fibril (Fig. 29.3). A nascent polypeptide chain, preprocollagen, is synthesized initially with a hydrophobic signal sequence that facilitates binding of ribosomes to the endoplasmic reticulum (ER) and directs the growing polypeptide chain into the lumen of the ER. Post-translational modification of the protein begins with removal of the signal peptide in the ER, yielding procollagen. Three different hydroxylases then add hydroxyl groups to proline and lysine residues, forming 3- and 4-hydroxyprolines and δ-hydroxylysine. These hydroxylases require ascorbate (vitamin C) as a cofactor (Fig. 29.3, step 1). Vitamin C deficiency leads to scurvy as a result of alterations in collagen synthesis and crosslinking (see Chapter 11).

O-linked glycosylation occurs by the addition of galactosyl residues to hydroxylysine by galactosyl transferase; a disaccharide is formed by addition of glucose to galactosyl hydroxylysine by a glucosyl transferase (Fig. 29.3, step 2). These enzymes have strict substrate specificity for hydroxylysine or galactosyl hydroxylysine, and they glycosylate only those peptide sequences that are in noncollagenous domains. N-linked glycosylation also occurs on specific asparagine residues in nonfibrillar domains. The nonfibrillar collagens, with a greater extent of nonhelical domains, are more highly glycosylated than fibrillar collagens. Thus, the extent of glycosylation may influence fibril structure, interrupting fibril formation and promoting interchain interactions required for a meshwork structure. Intra- and interchain disulfide bonds are formed in the C-terminal domains by a protein disulfide isomerase, facilitating the association and folding of peptide chains into a triple helix (Fig. 28.3, steps 3–5). At this stage, the procollagen is still soluble and contains additional, nonhelical extensions at its N- and C-terminals.

Procollagen is finally modified to collagen in the Golgi apparatus

After assembly into the triple helix, the procollagen is transported from the RER to the Golgi apparatus, where it is packaged into cylindrical aggregates in secretory vesicles, then exported to the extracellular space by exocytosis. The nonhelical extensions of the procollagen are now removed in the extracellular space, by specific N- and C-terminal procollagen proteinases (Fig. 29.3, step 6). The ‘tropocollagen’ molecules then self-assemble into insoluble collagen fibrils, which are further stabilized by the formation of aldehyde-derived intermolecular crosslinks. Lysyl oxidase (not to be confused with lysyl hydroxylase involved in formation of hydroxylysine) oxidatively deaminates the amino group from the side chains of some lysine and hydroxylysine residues, producing reactive aldehyde derivatives, known as allysine and hydroxyallysine. The aldehyde groups now form aldol condensation products with neighboring aldehyde groups, generating crosslinks both within and between triple-helical molecules. They may also react with the amino groups of unoxidized lysine and hydroxylysine residues to form Schiff base (imine) crosslinks (Fig. 29.4). The initial products may rearrange, or be dehydrated, or reduced to form stable crosslinks, such as lysinonorleucine. Studies with β-aminopropionitrile, which inhibits the enzyme lysyl oxidase, have illustrated that collagen crosslink formation is a major determinant of tissue mechanical properties and strength.

Weak hydrophobic interactions between valine residues permit the flexibility and extensibility of elastin

The flexibility required for function of blood vessels, lungs, ligaments and skin is imparted by a network of elastic fibers in the ECM of these tissues. The predominant protein of elastic fibers is elastin. Unlike the multigene collagen family, there is only one gene for elastin, coding for a polypeptide about 750 amino acids long. In common with collagens, it is rich in glycine and proline residues but elastin is more hydrophobic: one in seven of its amino acids is a valine. Unlike collagens, elastin contains little hydroxyproline and no hydroxylysine or carbohydrate chains, and does not have a regular secondary structure. Its primary structure consists of alternating hydrophilic and hydrophobic lysine and valine-rich domains. The lysines are involved in intermolecular crosslinking, while the weak interactions between valine residues in the hydrophobic domains impart elasticity to the molecule.

The soluble monomeric form of elastin initially synthesized on the RER is called tropoelastin. Except for some hydroxylation of proline, tropoelastin does not undergo post-translational modification. During the assembly process in the extracellular space, lysyl oxidase generates allysine in specific sequences: -Lys-Ala-Ala-Lys- and -Lys-Ala-Ala-Ala-Lys-. As with collagen, the reactive aldehyde of allysine condenses with other allysines or with unmodified lysines. Allysine and dehydrolysinonorleucine on different tropoelastin chains also condense to form pyridinium crosslinks – heterocyclic structures known as desmosine or isodesmosine (Fig. 29.5). Because of the way in which elastin monomers are cross-linked in polymers, elastin can stretch in two dimensions.

Fibronectin and laminin have multiple binding sites for ECM proteins and proteoglycans

Fibronectin is a glycoprotein present as a structural component of the ECM and also in plasma as a soluble protein. Fibronectin is a dimer of two identical subunits, each of 230 kDa, joined by a pair of disulfide bonds at their C-terminals. Each subunit is organized into domains, known as type I, II, and III domains, and each of these has several homologous repeating units or modules in its primary structure: there are 12 type I repeats, two type II repeats, and 15–17 type III repeats. Each module is independently folded, forming a ‘string of beads’ type of structure. At least 20 different tissue-specific isoforms of fibronectin have been identified, all produced by alternative splicing of a single precursor messenger ribonucleic acid (mRNA). The alternative splicing is regulated not only in a tissue-specific manner but also during embryogenesis, wound healing, and oncogenesis. Plasma fibronectin, secreted mainly by liver cells, lacks two of the type III repeats that are found in cell- and matrix-associated forms of fibronectin. Because of its multidomain structure and its ability to interact with cells and with other ECM components, alterations in fibronectin expression affect cell adhesion and migration, embryonic morphogenesis, and cytoskeletal and ECM organization.

Functional domains in fibronectin have been identified by their binding affinity for other ECM components, including collagen, heparin, fibrin, and the cell surface. The type I modules interact with fibrin, heparin and collagen, type II modules have collagen-binding domains, and type III modules are involved in binding to heparin and the cell surface. The specific interactions have been further mapped to short stretches of amino acids. A short peptide containing Arg-Gly-Asp (RGD), present in the tenth type III repeat of fibronectin, binds to the integrin family of proteins present on cell surfaces; this sequence is not unique to fibronectin but is also found in other proteins in the ECM. Another sequence, Pro-X-Ser-Arg-Asn (PXSRN), present in the ninth type III repeat, is implicated in integrin-mediated cell attachment. The loss of fibronectin from the surface of many tumor cells may contribute to their release into the circulation and penetration through the ECM, one of the first steps in tumor metastasis.

Glycosaminoglycans are the polysaccharide components of proteoglycans

The general structures of the glycosaminoglycans (GAGs) are shown in Table 29.2. The disaccharide repeat is different for each type of GAG, but is usually composed of a hexosamine and a uronic acid residue, except in the case of keratan sulfate, in which the uronic acid is replaced by galactose. The amino sugar in GAGs is either glucosamine (GlcNH₂) or galactosamine (GalNH₂), both of which are present mostly in their N-acetylated forms (GlcNAc and GalNAc), although in some of the GAGs (heparin, heparan sulfate) the amino group is sulfated rather than acetylated. The uronic acid is usually D-glucuronic acid (GlcUA) but in some cases (dermatan sulfate, heparin) it may be L-iduronic acid (IdUA). With the exception of hyaluronic acid and keratan sulfate, all the GAGs are attached to protein by a core trisaccharide, Gal-Gal-Xyl; the xylose is linked to a serine or threonine residue of a core protein. Keratan sulfate is also attached to protein, but in that case the linkage is either through an N-linked oligosaccharide (keratan sulfate I) or an O-linked oligosaccharide (keratan sulfate II). Hyaluronic acid, which has the longest polysaccharide chains, is the only GAG that does not appear to be attached to a core protein.

Hyaluronic acid, the only nonsulfated glycosaminoglycan, has a unique role in proteoglycan assembly

Hyaluronic acid is composed of repeating units of GlcUA and GlcNAc. This polysaccharide chain is the longest of the GAGs, with molecular weight of 10⁵–10⁷ Da (250–25,000 repeating disaccharide units), and is the only nonsulfated GAG.

Heparin is a small, highly charged GAG with strong anticoagulant activity

Heparin and heparan sulfate consist primarily of repeating disaccharide units of GlcNH₂ with IdUA or GlcUA, respectively. The linkage between the amino sugar and the uronic acid is uniformly 1–4, rather than the alternating 1–4/1–3 linkages seen in other GAGs. Most of the GlcNH₂ units of heparin are N-sulfated, whereas many of the IdUA residues are sulfated at the C-2 hydroxyl group, and the GlcNH₂ residues at the C-6 hydroxyl group. Heparin and heparan sulfate are the most highly charged of the GAGs. Although the structures of these two polymers are closely related, their distribution in the body and their functions are quite different: heparin is a small microheterogeneous molecule (~3000–30,000 Da), found intracellularly as a proteoglycan. It is released into the extracellular space as a free polysaccharide (GAG) and has strong anticoagulant activity (Chapter 7). In contrast, heparan sulfate is bound in the ECM or on the surface of cells, and has only weak anticoagulant activity.

The structure of glycosaminoglycans is determined by the cell's complement of glycosyl and sulfotransferases

Proteoglycans are synthesized by a series of glycosyl transferases, epimerases and sulfotransferases, beginning with the synthesis of the core trisaccharide (Xyl→Gal→Gal) while the protein is still in the RER. Synthesis of the repeating oligosaccharide and other modifications take place in the Golgi apparatus. As with the synthesis of glycoproteins and glycolipids, separate enzymes are involved in individual steps. For example, there are separate galactosyl transferases for each of the galactose units in the core, a separate GlcUA transferase for the core and repeating disaccharides, and separate sulfotransferases for the C-4 and C-6 positions of the GalNAc residues of chondroitin sulfates. Phosphoadenosine phosphosulfate (PAPS) is the sulfate donor for the sulfotransferases. These pathways are illustrated in Figure 29.6, for chondroitin-6-sulfate.

Defects of proteoglycan degradation lead to mucopolysaccharidoses

The degradation of proteoglycans occurs in lysosomes. The protein portion is degraded by lysosomal proteases and the GAG chains are degraded by the sequential action of a number of different lysosomal acid hydrolases. The stepwise degradation of GAGs involves exoglycosidases and sulfatases, beginning from the external end of the glycan chain. This may involve the removal of sulfate by a sulfatase, then removal of the terminal sugar by a specific glycosidase, and so on. Figure 29.7 shows the steps in the degradation of heparan sulfate. As with degradation of glycosphingolipids, if one of the enzymes involved in the stepwise pathway is missing, the entire degradation process is halted at that point and the undegraded molecules accumulate in the lysosome. The lysosomal storage diseases resulting from accumulation of GAGs are known as mucopolysaccharidoses (Table 29.3), because of the original designation of GAGs as mucopolysaccharides. There are more than a dozen such mucopolysaccharidoses, resulting from defects in degradation of GAGs. In general, these diseases can be diagnosed by the identification of specific GAG chains in the urine, followed by assay of the specific hydrolases in leukocytes or fibroblasts.

Bottlebrushes, silly putty and reinforced concrete

Proteoglycans are found in association with most tissues and cells. One of their major roles is to provide structural support to tissues, especially cartilage and connective tissue. In cartilage, large aggregates, composed of chondroitin sulfate and keratan sulfate chains linked to their core proteins, are noncovalently associated with hyaluronic acid via link proteins, forming a jelly-like matrix in which the collagen fibers are embedded. This macromolecule, a ‘bottlebrush’ structure known as aggrecan (Fig. 29.8), provides both rigidity and stability to connective tissue. Because of their negative charge, the GAGs bind large amounts of monovalent and divalent cations: a cartilage proteoglycan molecule of 2 × 10⁶ Da would have an aggregate negative charge of about 10,000. The maintenance of electrical neutrality consequently requires a high concentration of counter-ions. These ions draw water into the ECM, causing swelling and stiffening of the matrix, the result of tension between osmotic forces and binding interactions between proteoglycans and collagen. The structure and hydration of the ECM allow for a degree of rigidity, combined with flexibility and compressibility, enabling the tissue to withstand torsion and shock. The hyaluronic acid–proteoglycan–collagen aggregates in vertebral and articular disks have some of the viscoelastic properties of ‘silly putty’, bounce plus resilience, cushioning the impact between bones. These disks compress during the course of the day, expand elastically during the night, and deform gradually with age.

The overall structure of cartilage can be likened to that of the vertical reinforced concrete slabs poured during the construction of large buildings, in which steel rods (collagen fibers) are embedded in an amorphous layer of cement (the proteoglycan aggregates). Collagen stabilizes the network of proteoglycans in cartilage in much the same way that the reinforcing rods in the concrete provide structural strength for the cement walls. The structure of earthquake-resistant buildings, like the ECM, provides a balance between integrity and flexibility.

Although the amounts involved are low compared with those in skin and cartilage, organs such as the liver, brain, or kidney also contain a variety of proteoglycans:

Some proteoglycans or GAGs, especially heparin and heparan sulfate, have important physiologic roles in binding proteins or other macromolecules. Heparin serves as an intracellular binding site for proteinases located in secretory granules of mast cells. Several proteoglycans are involved in binding of proteins and enzymes to the vascular wall. They may also function in the vascular wall to inhibit clot formation by activation of antithrombin III (Chapter 7).

Integrins are plasma membrane proteins that bind to and transmit mechanical signals between the ECM and intracellular proteins

Interactions between cells and the ECM regulate a wide variety of cellular processes including proliferation, migration, differentiation and even survival. Several cell surface receptors have been indentified that mediate these interactions including integrins, discoidin domain receptors, dystroglycan and others. Of these, the integrins appear to be the most ubiquitous form of ECM receptors. Integrins are widely expressed throughout the animal kingdom from sponges to humans. They are heterodimers of α and β chains that have been loosely grouped into subfamilies based upon the component β chain. To date, 18 α and eight β chains have been indentified in mammals. Through various combinations of α and β chains, over 20 different functional integrin heterodimers have been described. The specific combination of α and β chains dictates the specific ECM ligand for a particular integrin heterodimer. However, multiple integrin heterodimers can bind to some ECM components. For instance, α₄β₁, α₅β₁ and α_vβ₃ all interact with fibronectin. Adding to this complexity, several integrin heterodimers bind to multiple ECM components. For instance, α_vβ₃, which was originally described as a vitronectin receptor, can interact with not only vitronectin but also with fibronectin, fibrinogen and osteopontin as well.

In a functional integrin, both the α and β chains span the cell membrane (Fig. 29.9). Typically, each chain has a large extracellular domain, a single transmembrane domain and a short cytoplasmic tail. An exception to this is the β₄ protein, which has an exceptionally long cytoplasmic tail of over 1000 amino acids. The extracellular region of the integrin heterodimer interacts with ECM components in a divalent cation-dependent manner. The integrins are in an optimal position to transmit physical or mechanical signals from the ECM to the interior of the cell. These physical signals can be further distributed through the cell via the actin-containing cytoskeleton and ultimately modulate gene expression in the nucleus. This transmission of physical signals via the ECM–integrin–cytoskeletal axis has been extensively investigated and termed ‘tensegrity’. Physical signals from the ECM can also be transduced into biochemical events in the cytoplasm of the cell via integrins. Unlike some other types of receptors, integrins do not themselves possess enzymatic activity. However, integrins associate with a number of cytoplasmic protein kinases including focal adhesion kinase (FAK) and Src. Activation of integrins initiates enzymatic cascades via these associated kinases that ultimately leads to changes in cell behavior and gene expression.