Chapter 63. Collagens, Elastic Fibers, and Other Extracellular Matrix Proteins of the Dermis

The extracellular matrix (ECM) is a complex network of different components, responsible for determining and maintaining tissue architecture, and for mediating a number of important biological events. It is composed of a large number of diverse protein families, each constituted by many different individual members. These include the collagens, encoded by 42 different genes, elastin and associated microfibrillar proteins, fibronectin, proteoglycans, and many more molecules. (See Table 63-1.)

Although all of these proteins are genetically, structurally, and biologically diverse, a common denominator is that most of them have a modular structure, and they are composed of one, a few, or several copies of a limited set of individual structural modules, also called domains.1 These can be combined in multiple ways giving rise to proteins as diverse as fibrillin and laminin (Fig. 63-1). Specific functions have been unraveled for some of these domains, for instance, interaction with other ECM proteins, cell adhesion-promoting activity, cytokine trapping, and regulation. Therefore, the ECM has a critical role for many cellular functions, including proliferation, survival, polarity, differentiation, expression of specific genes, and migration.2–4 All different cell types, such as mesenchymal, epithelial, and endothelial cells, and also inflammatory and tumor cells, participate in the production of distinct ECM macromolecules, and are all influenced by interactions with these compounds. It is well established that the ECM determines the biophysical properties of connective tissues. More recently it became clear that, conversely, stiffness and compliance of connective tissues are important factors for the regulation of cellular functions.5–8

The proteins of the collagens family are the main structural components of the connective tissues and the major extracellular proteins of the human body. In human skin, collagens comprises approximately 80% of the dry weight of the dermis. The classical and first recognized physiologic role of collagens in the skin is to provide tensile properties that allow the skin to serve as a protective organ against external trauma. In addition, it is clear that collagens displays important biological properties regulating multiple cellular and tissue activities.

The Collagens Triple Helix

The bulk of collagens in dermis is deposited as large bundles of regularly oriented fibers composed of fibrils that are aligned in a parallel manner, which results in a pattern of cross-striations that can be visualized by electron microscopy (Fig. 63-2). The most prominent cross-striations appear as repeating bands spaced approximately 70 nm apart. The major component of these fibrils is type I collagens, the first identified member of the collagens family and the most abundant collagens in the dermis and most other connective tissues.9–11 The characteristic structural features of collagens were deduced from the study of type I collagens, which is therefore considered as the prototype of the collagens family.

The type I collagens molecule has an approximate molecular mass of 290 kDa and is composed of three polypeptide chains, each approximately 94 kDa. These three polypeptides, known as the α chains, are coiled around each other much like strands of rope, so that the collagens monomer has a triple-helical structure. This conformation gives the molecule a rigid, rod-like shape with approximate dimensions of 1.5 × 300 nm (see Fig. 63-3). The special structure of the collagens triple helix is largely explainable by the unusual amino acid composition of the α chains. Specifically, each α chain of type I collagens has approximately 1,000 amino acids, and glycine (Gly), the smallest amino acid, accounts for approximately one-third of the total number of amino acids, evenly distributed along the polypeptide. Consequently, the polypeptide chains of type I collagens can be described as repeating triplets represented as (Gly-X-Y)333. Although the X and Y positions of the triplets can be occupied by any amino acid, they are often occupied by proline and hydroxyproline, respectively. These two amino acids account for approximately 22% of the total amino acid composition of type I collagens. The relatively high content of these amino acids and the characteristic distribution of glycine in every third position are necessary for the triple-helical conformation of the collagens molecule, and hydroxyproline plays a critical role in stabilizing the triple helix at body temperature. The triple-helical conformation defines the so-called collagenous domain, and it gives type I collagens many of its unique properties.12 In particular, it is essential for assembly into fibrils. Mutations that affect the formation of the stable triple helix prevent collagens from forming fibers, which results in serious defects of connective tissue function and a spectrum of clinical phenotypes.13–15

Genetic Heterogeneity of Collagens

Collagens comprises a family of closely related yet genetically distinct proteins. In the human genome, there are as many as 42 different genes encoding α chains with variable amino acid sequences. These α chains correspond to at least 28 different types of collagens, which have been assigned Roman numerals I to XXVIII (Table 63-2). These are subdivided into several subfamilies, mainly according to the length and number of collagenous and noncollagenous domains. In addition to well-characterized collagens, short triple-helical collagenous segments are present in other proteins, including acetyl cholinesterase, the C1q component of the complement system, gliomedin, pulmonary surfactant proteins, macrophage scavenger receptors, emilins (elastin microfibril interface-located glycoproteins), and ectodysplasin A, a product of the gene mutated in X-linked anhidrotic ectodermal dysplasia.16 However, these proteins are not included in the collagens family because the collagenous domains are not a predominant part of the molecules, and the proteins do not function primarily as structural components of the ECM of connective tissue.

The genetically distinct collagens are distributed into several classes, mainly according to the length and number of triple-helical collagenous segments and noncollagenous domains present in the molecule, and on the basis of the architecture of their assembly in tissues (see Table 63-2). A classification and an overview of the major collagens types contributing to skin physiology and pathology are presented in the following paragraphs. Some collagens types are not discussed in detail because they do not appear to be present in the skin in significant quantities, or they do not seem to contribute to the physiologic properties of dermis, or because information is still lacking.

The first class groups the fibril-forming collagens types I, II, III, V, XI, XXIV, and XXVII. Except for collagens II, XXIV, and XXVII that were not detected in skin, all other fibril-forming collagens types are expressed in the dermis. These collagens types consist of a single 300-nm long collagenous domain, with very short nontriple-helical extensions (Fig. 63-4), and they assemble into relatively large fibrils.11,12,17 Type I collagens, the most widely distributed and the most extensively characterized form of collagens, accounts for approximately 80% of the total collagens of adult human dermis. It is formed by two identical α chains, designated α1(I), and a third chain, called α2(I), that is clearly different in its amino acid composition. Thus, the chain composition of type I collagens is [α1(I)]2α2(I). Type III collagens represents approximately 10% of the total collagens in the adult human dermis.18 It is composed of three identical α1(III) chains, distinguished from the chains of type I collagens by a relatively high content of hydroxyproline and glycine and the presence of one cysteine residue. Collagens types I and III form the relatively broad extracellular fibers that are primarily responsible for the tensile strength of the human dermis. Mutations in the type I and III collagens genes can result in connective tissue abnormalities in the skin and joints, among other tissues, in different forms of Ehlers–Danlos syndrome and fragility of bones in osteogenesis imperfecta.13–15 Type V collagens forms a subfamily of similar interrelated collagens resulting from various assemblies of three different types of α chains, α1(V), α2(V), and α3(V). The predominant form in the skin is [α1(V)]2α2(V), where it represents less than 5% of the total collagens. In dermis, type V collagens is associated with major collagens fibers consisting of type I and III collagens, and type V collagens has been postulated to regulate the fiber diameter during fibrillogenesis. The importance of type V collagens has been demonstrated by the discovery of mutations in type V collagens genes in patients with classical autosomal dominant forms (types I and II) of Ehlers–Danlos syndrome.19 It is of interest that a clinically similar, classical type of Ehlers–Danlos syndrome can be based on the absence of tenascin-X expression, which leads to an autosomal recessive form of the disease.20 Tenascin-X is developmentally associated with collagens fibrillogenesis, and its absence could explain the presence of defective collagens fibers similar to those found in patients with mutations in type V collagens genes.

Another class comprises the network-forming collagens types IV, VIII, and X, which are, however, very different structurally (Fig. 63-4). Type IV collagens is a typical component of basement membranes, where it forms a lattice rather than the fibers characteristic of dermal collagens. Six different α (IV) chains exist, α1(IV) to α6(IV), and participate in the assembly of heterotrimers. In human skin, α1(IV) to α4(IV) chains are present in the basement membrane at the dermal–epidermal junction, with [α1(IV)]2α2(IV) being the predominant heterotrimer.17,21 Type IV collagens chains contain imperfect Gly-X-Y triplets, a feature that confers flexibility to the triple-helical domain of the molecule. A noncollagenous globular domain at the amino-terminal part of the molecule mediates dimer formation, while a short segment at the carboxyl-terminal region of the molecule allows for tetramer assembly, altogether resulting in the so-called “chicken-wire” assembly network for collagens type IV.17 Although skin disorders associated with genetic mutations in type IV collagens have not been described, mutation in the COL4A1 gene has been identified in a family with autosomal dominant porencephaly and infantile hemiparesis,22 and mutations in the COL4A5 gene result in Alport syndrome,23 an X-linked renal disease. Furthermore, autoantibodies recognizing the collagens α3(IV) chain underlie Goodpasture syndrome,23 and autoantibodies targeting the collagens α5(IV) chain are associated with a novel autoimmune disease characterized by subepidermal blisters and renal insufficiency.24

Two further collagens, type VI collagens and type VII collagens, are unique and each has its own distinct molecular structure and supramolecular assembly. Type VI collagens forms specific microfibrils in the dermis, while type VII collagens forms the anchoring fibrils of the dermal–epidermal junction. The microfibril-forming type VI collagens is relatively abundant in a variety of tissues, including skin. In humans, five distinct α chains—α1(VI), α2(VI), α3(VI), α5(VI), and α6(VI)11—have been identified (Table 63-2). The α1(VI), α2(VI), and α3(VI) chains fold into a triple-helical domain of approximately 100 nm in length, with globular domains at both ends (Fig. 63-4). The supramolecular assembly involves the formation of antiparallel dimers, then tetramers, which in turn align in rows, leading to relatively thin microfibrils, independently of the broad collagens fibers.25,26 The microfibrils of type VI collagens may perform an anchoring function by stabilizing the assembly of the broad collagens fibers as well as basement membranes. Studies have now demonstrated that mutations in each of the three type VI collagens genes can lead to different forms of congenital muscular dystrophy, with multiple joint contractures, laxity of distal joints, and characteristic skin involvement.27

Type VII collagens, the major, if not the exclusive, component of the anchoring fibrils at the dermal–epidermal junction, has an unusually long triple-helical region of approximately 450 nm.28 It has only one type of α chain, α1(VII), and the triple-helical collagenous domain is flanked by a large, nonhelical (noncollagenous) domain, NC-1, at the amino terminus and a shorter carboxyl-terminal nonhelical domain, NC-2 (Fig. 63-4). Type VII collagens molecules become organized into anchoring fibrils, which extend from the lower part of the lamina densa to the papillary dermis, through the formation of antiparallel dimers linked through their carboxyl-terminal ends. The large amino-terminal, noncollagenous domains of type VII collagens interact with type IV collagens and laminin 332 components of the dermal–epidermal basement membrane, and it has been suggested that most anchoring fibrils form U-shaped loops that entrap broad dermal collagens fibers consisting of type I and III collagens. Thus, alterations in the expression, structure, or molecular interactions of type VII with other basement membrane components could result in skin fragility. Such a situation is exemplified by dystrophic epidermolysis bullosa (DEB), a group of mechanobullous diseases characterized by blistering of the skin after minor trauma (see Chapter 62). In fact, distinct mutations in the type VII collagens gene, COL7A1, have been demonstrated in more than 500 families with different forms of DEB, and none of the families with DEB studied so far has revealed mutations in genes other than that encoding the α1(VII) polypeptide of type VII collagens.29–31 Furthermore, type VII collagens serves as the autoantigen in the acquired form of epidermolysis bullosa, an autoimmune disorder with circulating antitype VII collagens antibodies (see Chapter 60).

A further class groups the Fibril-Associated Collagens with Interrupted Triple helices, in short FACITs, including types IX, XII, XIV, XVI, XIX, XX, XXI, XXII, and XXVI.11,32 An association with collagens fibers has been demonstrated for several of these collagens, hence the concept that they serve as important molecular bridges for the organization and stability of extracellular matrices. These collagens types form homotrimers with a relatively short triple-helical domain, and 2–4 noncollagenous domains (Fig. 63-4). In the skin, type XII and type XIV collagens associate with the large collagens fibrils of the dermis.

Type XV collagens and Type XVIII collagens are basement membrane-associated collagens. They have been called multiplexins because they contain multiple noncollagenous domains in their collagenous sequences.33 Interest in these collagens stems from the fact that proteolysis liberates fragments endowed with functions different from those of the original intact molecules.33,34

Type XIII, XVII, XXIII, and XXV collagens are transmembrane proteins.35 Their supramolecular assemblies remain unknown, and, except for types XIII and XVII, their presence in skin has not been established. Type XVII collagens (Fig. 63-4) is particularly important for skin physiology and pathology. It is a transmembrane protein anchored in the membrane of basal keratinocytes, with an intracellular domain and a large extracellular domain, or ectodomain, which is a component of the basement membrane at the dermal–epidermal junction. Immunoelectron microscopy has localized type XVII collagens to hemidesmosomes of basal keratinocytes and to the thin anchoring filaments originating from the hemidesmosomes and spanning the lamina lucida toward the lamina densa of the basement membrane.36 The ectodomain of type XVII collagens consists of 15 collagenous domains with the characteristic repeating Gly-X-Y sequences that form triple helices. These collagenous domains are separated by noncollagenous segments of variable sizes, and consequently type XVII collagens is a protein characterized by alternating collagenous and noncollagenous segments.37 The ectodomain colocalizes and interacts with laminin 332, also a component of anchoring filaments36,38 (see also Chapter 53). Type XVII collagens was initially identified as the 180-kDa bullous pemphigoid antigen (BPAG2), which was recognized by circulating autoantibodies in the sera of patients with bullous pemphigoid or herpes gestationis.39 The importance of the type XVII collagens was also shown by mutations in the corresponding gene (COL17A1) that underlie a nonlethal variant of junctional epidermolysis bullosa, generalized atrophic benign epidermolysis bullosa (see Chapter 62). These patients have protracted, lifelong blistering of the skin, atrophic scarring, alopecia, and nail dystrophy.37,40

From Collagens Genes to Supramolecular Assemblies

The collagens genes, like most eukaryotic genes, are large, multiexon genes interrupted at several points by noncoding DNA sequences called introns.41 For expression, the entire gene is transcribed into a high-molecular-weight precursor messenger RNA (mRNA), which undergoes posttranscriptional modifications, such as capping, polyadenylation, and intron splicing to yield a linear, uninterrupted coding sequence with 5′ and 3′ untranslated flanking regions. The mature mRNA is then transported into the cytoplasm and translated to the corresponding polypeptide. Thus, the eukaryotic gene coding for a protein is much larger than would be predicted from the amino acid sequences of the final protein.

With few exceptions, the collagens genes are widely scattered throughout the human genome (see Table 63-2). Knowledge of the precise chromosomal location of the genes coding for collagens in human skin has allowed identification of polymorphic markers within the genes and in the flanking DNA for use in genetic linkage studies. In addition, sophisticated mutation detection strategies, based on scanning of the genes, have led to identification of a large number of mutations in different collagens genes with characteristic phenotypic consequences.

Under physiologic conditions, fibril-forming collagens molecules spontaneously assemble into insoluble fibers.17 This observation presented a logistic problem, because it was difficult to visualize how a collagens molecule could be synthesized inside the cell and then secreted into the extracellular space without premature assembly of the molecules into insoluble fibers. The solution to this problem was found in the demonstration that fibril-forming collagens, and some other collagens, is initially synthesized as a larger precursor molecule, procollagen, which is soluble under physiologic conditions.

The precursor polypeptides of procollagen, so-called pre-proα chains, are synthesized on the ribosomes of the rough endoplasmic reticulum in fibroblasts and related cells (Fig. 63-5). This initial translation product contains an amino-terminal signal (or leader) sequence rich in hydrophobic amino acids. This sequence serves as a signal for attachment of the ribosomes to the membranes of the rough endoplasmic reticulum and vectorial release of the nascent polypeptides into the cisternae of the rough endoplasmic reticulum. During the transmembrane transport of the polypeptides, the signal sequence is enzymatically removed in a reaction catalyzed by signal peptidase, and the polypeptides, termed proα chains, are released inside the lumen of the rough endoplasmic reticulum. For some collagens types, in particular the fibril-forming collagens, proα chains are larger than collagens α chains because they contain additional peptide sequences at both ends of the molecule.

Posttranslational Modifications of Polypeptide Chains

After the assembly of amino acids into pre-proα chains on the ribosomes, the polypeptides undergo several modifications before the completed collagens molecules are deposited into extracellular fibers (Fig. 63-5). Most of these posttranslational modifications are catalyzed by specific enzymes (Table 63-3), and include (1) synthesis of hydroxyproline by hydroxylation of selected prolyl residues; (2) synthesis of hydroxylysine by hydroxylation of selected lysyl residues; (3) attachment of carbohydrates, galactose, or glucosylgalactose, onto certain hydroxylysyl residues; (4) chain association, disulfide bonding, and triple-helix formation; (5) proteolytic conversion of procollagens to collagens; and (6) fiber formation and cross-linking. Current evidence indicates that modification reactions (1)– (4) are intracellular events, whereas proteolytic conversion, fiber formation, and cross-linking probably take place extracellularly (see Fig. 63-5).

Synthesis of Hydroxyproline, Hydroxylysine, and Attachment of Carbohydrates

A characteristic feature of collagens is the presence of hydroxyproline and hydroxylysine residues.9 Free hydroxyproline and hydroxylysine are not incorporated into nascent polypeptide chains, but result from the hydroxylation of prolyl and lysyl residues, respectively (Fig. 63-6). The hydroxylation reactions are catalyzed by enzymes belonging to the prolyl and lysyl hydroxylase families.9,41–45 These enzymes require molecular oxygen, ferrous iron, α-ketoglutarate, and a reducing agent, such as ascorbate as cosubstrates or cofactors for the reactions (Fig. 63-6 and Table 63-3). Hydroxylation begins while the proα chains are growing on the ribosomes, and it is completed soon after the release of full-length polypeptide chains from the ribosomes (Fig. 63-7). The hydroxyproline in collagens is found in two isomeric forms, trans-3-hydroxy-l-proline, and the most abundant trans-4-hydroxy-l-proline. A critical amount of trans-4-hydroxy-l-proline is a prerequisite for the folding of α chains into the triple helix, the conformation required for secretion of procollagens molecules out of the cells. Because triple-helix formation takes place in the cisternae of the rough endoplasmic reticulum, and because prolyl hydroxylases do not hydroxylate prolyl residues if the collagens substrate is in a triple-helical conformation, hydroxyproline formation must be completed before the procollagens molecules leave this cellular compartment. Thus, in the absence of hydroxyproline, the critical triple-helical structure of collagens would not form under physiologic conditions, and no functional collagens fibers would appear in the extracellular space. Also, because prolyl hydroxylase requires a reducing agent, such as ascorbate, for its activity, ascorbic acid deficiency leads to a decreased formation of collagens fibers. This explains some of the clinical manifestations in scurvy, such as poor wound healing and decreased tensile strength of the connective tissues.46 An analogous situation may exist in tissues with hypoxia, because molecular oxygen is a specific requirement for the formation of hydroxyl groups in hydroxyproline. Studies in animal models demonstrate that wound healing is relatively poor under hypobaric conditions, and in such situations the low oxygen levels may limit the synthesis of hydroxyproline.47 This observation may also explain the decreased healing tendency of wounds and ulcers in peripheral tissues that are anoxic due to relatively poor blood supply. It may also relate to the recent appreciation that prolyl hydroxylases are genuine oxygen sensors. In addition to the prolyl-4-hydroxylase playing a critical role in the hydroxylation of prolyl residues on nascent collagens polypeptide chains, there are additional prolyl-4-hydroxylase isoforms that are responsible for the hydroxylation of two proline residues that earmarks the α subunit of the hypoxia-inducible transcription factor (the master regulator of hypoxia-inducible genes) for proteasomal degradation.44,48,49

Lysyl hydroxylation is of critical importance because hydroxylysyl residues serve either as an attachment site for galactosyl and glucosylgalactosyl residues during the intracellular synthesis of procollagen, or to the formation of cross-links that stabilize the collagens matrix in the extracellular space.9,45,50 The intracellular modification of lysyl residues is a sequential event, with first hydroxylation of lysyl residues, followed by the attachment of galactosyl residue to the hydroxyl group of hydroxylysine with an O-glycosidic bond and then a glucose moiety is attached to some of the galactosyl residues (see Figs. 63-5 and 63-7). Therefore, the synthesis of hydroxylysine is a prerequisite for the glycosylation of collagens (see Table 63-3). Three lysyl hydroxylase isoforms exist, and one of them, the lysyl-3-hydroxylase (LH3), displays three activities: (1) lysyl hydroxylase, (2) galactosyltransferase, and (3) glucosyltransfrase activities.45 The glycosylation reactions use uridine diphosphate sugars as a source of the carbohydrate and require Mn2+ as a cofactor. In addition to the glycosylation of hydroxylysyl residues in the triple-helical portion of the molecule, the nonhelical extensions contain complex carbohydrates, consisting mainly of mannose. Deficiency of lysyl hydroxylase has been identified in patients with the scoliotic form (type VI) of Ehlers–Danlos syndrome, characterized by hyperextensible skin, loose-jointedness, severe kyphoscoliosis, and ocular fragility. A connective tissue disorder caused by mutations of the lysyl hydroxylase 3 gene has been recently identified in humans. The disorder is associated with abnormalities in several organs, including skin, and the phenotype has features that overlap with a number of known collagens disorders.51

Chain Association, Disulfide Bonding, and Triple-Helix Formation

A critical step in the intracellular biosynthesis of procollagens is the association of three proα chains and subsequent folding of the collagens portion of the polypeptides into a triple helix (see Fig. 63-7). The noncollagenous peptide extensions on the individual proα chains assume globular conformations soon after their translation, and this conformation contains the specific information that directs the correct association of the three proα chains. Such a mechanism might explain the association of proα1 and proα2 chains in a proper 2:1 ratio during the synthesis of type I procollagen. It would also explain the rapid and efficient association of the proα chains and folding of the molecule into the triple helix. The association of the extensions at the carboxyl-terminal ends of the polypeptide chains appears to facilitate folding of the molecules into the triple helix, perhaps by providing a nucleation site from which the formation of the triple helix is propagated throughout the collagenous portion of the molecule.

Conversion of Procollagenss to Collagens

After secretion into the extracellular space, procollagens molecules are converted to collagens by limited proteolysis, which removes the extension peptides on the molecule.52 The conversion of type I procollagens to collagens is catalyzed by two specific enzymes, (1) procollagens N-proteinase and (2) procollagens C-proteinase, that separately remove the amino-terminal and carboxyl-terminal extensions, respectively (see Table 63-3). Furthermore, the N-proteinase catalyzing the conversion of type III procollagens to collagens is a separate, specific enzyme.

The N-proteinase capable of cleaving type I procollagens belongs to the ADAMTS (a disintegrin and metalloprotease with thrombospondin motifs) family of extracellular proteases, and is called ADAMTS-2.53 The activity of this enzyme is dependent on the native conformation of the amino-terminal propeptides in procollagen, because it does not catalyze the removal of the extension peptides from isolated proα chains. Furthermore, partially purified N-proteinase is inhibited by metal chelators, which suggests a requirement for divalent cations.

C-proteinase is required for the removal of the carboxyl-terminal extension from type I, II, III, and V procollagen, which allows the fully processed molecules to form functional fibers.52 Cloning of type I procollagens C-proteinase revealed that it is identical to bone morphogenic protein-1, a metalloprotease implicated in pattern formation during development of diverse organisms and also capable of inducing ectopic bone formation.54 The activity of C-proteinase/bone morphogenic protein-1 is stimulated by procollagens C-proteinase enhancers, glycoproteins that bind to the C-terminal propeptide of type I procollagen.55 Thus, the conversion of procollagens to collagens is a complex, carefully controlled process, and lack of removal of either the amino- or the carboxyl-terminal extensions results in impaired tensile strength of collagens fibers in the skin. Specifically, deficiency in the removal of the amino-terminal propeptide of type I collagens in vivo causes dermatosparaxis, a disease of fragile skin, originally recognized in various animal species, and more recently recognized in humans.56 Specifically, the human counterpart is the dermatosparaxis type of Ehlers–Danlos syndrome (type VIIc), caused by deficiency in N-proteinase activity. It should be noted that a disease with similar phenotype, the arthrochalasia type of Ehlers–Danlos syndrome (types VIIa and VIIb), can be caused by mutations in the type I collagens genes (COL1A1 and COL1A2, respectively) at the cleavage site for the N-proteinase.57

Collagens Assembly and Cross-Linking

Collagens assembly is best defined for fibrillar collagens. After removal of the extension peptides in the extracellular space, the collagens molecules spontaneously align to form fibers. However, these fibers do not attain the necessary tensile strength until the molecules are linked together by specific covalent bonds known as cross-links.58 The most common forms of cross-links in collagens are derived from lysine or hydroxylysine. The first step in the cross-linking of collagens is the enzymatic conversion of some of the lysyl and hydroxylysyl residues to the corresponding aldehyde derivatives by removal of the ϵ-amino groups (see Figs. 63-5 and 63-7). The aldehydes then form cross-links by two kinds of reactions. One reaction involves condensation of an aldehyde with an ϵ-amino group still present in another unmodified lysine or hydroxylysine to form a Schiff base-type of covalent cross-link. The second type of reaction is an aldol condensation between two aldehydes. In addition to these cross-links, collagens contains several more complex cross-links that also involve lysyl or hydroxylysyl residues. The lysine- and hydroxylysine-derived cross-links can be either intramolecular, occurring between two adjacent α chains in the same collagens molecule, or intermolecular, stabilizing the alignment of neighboring collagens molecules along microfibril structures.

The first step in collagens cross-linking, the oxidative deamination of certain lysyl and hydroxylysyl residues, is catalyzed by lysyl oxidase. This enzyme requires copper as a cofactor, and its activity is readily inhibited by nitriles, such as β-aminopropionitrile, which produce lathyrism in animals. Because the cross-links of collagens provide the tensile strength required in certain tissues, a defect in the formation of these covalent bonds can lead to a disturbance in connective tissue function. An example is occipital horn syndrome (previously known as Ehlers–Danlos syndrome type IX), which results from reduced lysyl oxidase activity (see Chapter 137). The primary defect resides in perturbed copper metabolism caused by mutations in a copper transport enzyme protein, an adenosine triphosphatase encoded by the gene MNK-1 that is also involved in Menkes syndrome.59 As a result, serum copper levels are reduced, which leads to reduced lysyl oxidase activity.

Control of Collagens and Extracellular Matrix Production

A major question in ECM biology concerns the mechanisms that control the deposition of collagens and other ECM proteins in tissues. Such control must exist in vivo, because the consequences of uncontrolled ECM accumulation are demonstrated by diseases such as progressive systemic sclerosis (see Chapter 157) and various other fibrotic conditions.60,61

The accumulation of ECM in tissues can be controlled at several different levels of biosynthesis and degradation. Several observations suggest that an important control mechanism acts at the level of mRNA formation through regulation of the transcriptional activity of gene expression.62,63 The transcriptional regulation of gene expression involves both cis-acting elements and trans-acting factors. The cis-acting elements are nucleotide sequences in the promoter region of the gene that serve as binding sites for trans-acting cellular proteins, which can upregulate or downregulate the transcriptional promoter activity. Some of the trans-acting factors are nuclear receptors, such as the retinoic acid receptors (RAR and RXR)49 that form a complex with the ligand (a retinoid) and then bind to the retinoic acid-responsive elements (RARE) in the target gene. Retinoids, such as all-trans-retinoic acid, modulate type I collagens gene expression both in vitro and in vivo.64 In particular, quiescent nonproliferating cells can be stimulated by retinoic acid to activate collagens gene expression. This may have relevance to the elevated collagens synthesis observed in photodamaged dermis after topical application of all-trans-retinoic acid (see Chapter 109).

Diverse growth factors, cytokines, and chemokines are released into the extracellular space from different types of cells, such as fibroblasts, keratinocytes, and endothelial and inflammatory cells. By both paracrine and autocrine mechanisms these factors elicit intracellular signaling pathways acting on the transcriptional and/or translational levels. One of the most powerful regulators of collagens and ECM production is transforming growth factor β (TGF-β), a member of a family of growth factors involved in many physiological and pathological conditions.65 Interestingly, TGF-β can induce the expression of other growth factors, in particular the expression of connective tissue growth factor in fibroblasts. Also, cytokines such as platelet-derived growth factor, interleukin-1 and -4, and chemokines, such as monocyte chemotactic protein-1 and -3, are thought to induce collagens production.66

Degradation and Turnover

In the extracellular space, most collagens form polymers described above, which in turn interact with other proteins of the ECM through multiple protein–protein interactions. It gives rise to large multimolecular assemblies and to the organization of the ECM proteins into specific networks.17 In addition, many growth factors and cytokines specifically interact with ECM proteins, and they are therefore stored in the networks. There is a continuous remodeling of the multimolecular assemblies, at a rate that can be significantly enhanced during development, wound healing or in certain disease processes. This remodeling results in the release of a large variety of biologically active peptides, including growth factors.34

The remodeling of the ECM is a cooperative multistep process involving a variety of molecular pathways and mechanisms. The initial step depends on the presence of proteinases capable of initiating the degradation of the ECM proteins. These enzymes comprise the matrix metalloproteinase (MMP) family,67–71 which includes collagenases, gelatinases, stromelysins, matrilysins, metalloelastase, enamelysin, and the membrane-type MMPs (Fig. 63-8 and Table 63-4). Interstitial collagenase (MMP-1) was the first enzyme of the MMP family to be discovered and was defined by its ability to break down triple-helical collagens that is resistant to most proteases.72 Human skin collagenase was initially isolated in active form from the culture medium of skin explants and subsequently as a proenzyme from monolayer fibroblast cultures.73 Other cell types present in skin, including keratinocytes, endothelial cells, monocytes, and macrophages, express an identical enzyme. Like the other MMPs, MMP-1 contains intrinsic zinc in the active site and requires calcium for its activity and thermostabilization.

MMP-1 degrades some of the collagens present in skin, such as types I, III, and VII, but not others like types IV and V.74 At physiologic pH and temperature, the enzyme catalyzes a single cleavage three-quarter of the distance from the amino terminus in each of the three α chains comprising the triple-helical, native collagens monomer. Specifically, MMP-1 cleaves the α1(I) chain at a Gly-Ile (glycine-isoleucine) bond (residues 775 and 776) and the α2(I) chain at a Gly-Leu (glycine-leucine) bond in the homologous region. Ten other Gly-Ile or Gly-Leu bonds within the triple-helical domain of interstitial collagens are not cleaved, which suggests that the local conformation of the collagenase cleavage site is a major factor in determining substrate specificity. Indeed, MMP-1 catalyzes multiple cleavages in the denatured chains of all collagens types at Gly-Ile and Gly-Leu bonds. Thus, it appears that the triple-helical conformation of native collagens can drastically alter MMP-1 ability to catalyze cleavages that would be permissible from primary sequence consideration alone.

Human neutrophil interstitial collagenase (MMP-8) attacks collagens at the same site, as does human MMP-1, to produce the characteristic three-quarter/one-quarter collagens fragments. Human MMP-8 is highly homologous to MMP-1, but has a higher degree of glycosylation. It is not immediately secreted but is stored in neutrophil granules and released on stimulation. Although both fibroblast (MMP-1) and neutrophil (MMP-8) collagenases have similar affinities for type I and III collagens, fibroblast collagenase degrades soluble type III collagens with a higher turnover rate, whereas neutrophil collagenase degrades soluble type I collagens more rapidly. For both enzymes, the differences in specificity against monomeric collagens are largely abolished when the substrates are reconstituted into the insoluble fibrillar forms found in tissues.75

Important mechanistic information on collagens degradation has recently emerged, and specifically, it has been shown75 that activated collagenase (MMP-1) moves processively on collagens fibrils in which the mechanism of movement is a biased diffusion with the biased component of the motion depending on the proteolysis of collagens and not on ATP hydrolysis. Most importantly, the closely related MMPs, MMP-2 (gelatinase) MMP-9, and the extracellular portion of the membrane-associated metalloproteinase MT1–MMP, can also diffuse on collagens fibrils. These studies also show that a collagenolytic MT1–MMP complex exists at the cell surface. The MT1–MMP complex anchored at the cell surface has profound implications concerning the generation of the forces required for cell locomotion, not only in the normal physiologic processes of tissue remodeling but also in pathologic processes, such as tumor invasion.

Collagenase-3 (MMP-13), first cloned from breast carcinoma, is homologous to rodent collagenase.69 The purified recombinant enzyme cleaves triple-helical collagens, giving the characteristic three-quarter/one-quarter fragments. However, unlike interstitial and neutrophil collagenases, collagenase-3 acts five to ten times more rapidly on soluble type II collagens, a cartilage-specific collagens, than on types I and III. At physiologic temperature, the collagens cleavage products (i.e., the three-quarter/one-quarter fragments), which are initially triple helical, become soluble, are thermally unstable, and denature spontaneously, forming gelatin (denatured collagens). These denatured gelatin polypeptides are then susceptible to attack by other proteinases. Thus, the interstitial collagenases stand at a key point in connective tissue metabolism: they initiate the proteolytic events that result in the degradation of collagens and overall turnover of the extracellular matrices.

Structure and Development of the Elastic Fibers

Elastic fibers form a network responsible for connective tissue elasticity and resilience specific of various organs.76,77 The relative concentration of elastic fibers is highest in the aorta and arterial blood vessels and in the lungs, while in the skin they are only a minor component. Specifically, in sun-protected adult human skin, the elastin content is 1%–2% of the total dry weight of dermis.78 In the papillary dermis, elastic fibers are present either as bundles of microfibrils (oxytalan fibers) or with small amounts of cross-linked elastin (elaunin fibers). In the reticular dermis, the elastic fibers consist primarily of elastin, and they are oriented horizontally in a network with vertical extensions to the papillary dermis in the form of oxytalan fibers. Examination of connective tissues by transmission electron microscopy has demonstrated that mature elastic fibers consist of two distinct components (Fig. 63-9). An amorphous, electron-lucent core consists primarily of elastin, surrounded by distinct electron-dense microfibrillar structures that have a regular diameter of 10–12 nm, and which is composed of a variety of microfibrillar proteins such as fibrillins and fibulins.

Elastin and the microfibrillar proteins exist in close association in various connective tissues; however, the relative proportions of these components vary during embryonic development. Elastic fibers that form during the first trimester of fetal development consist of bundles of microfibrils, which are thought to form a scaffold into which the elastin molecules will align while fetal age is increasing. Analysis of mature, fully developed elastic fibers suggests that elastin represents well over 90% of the total content of such fibers (see Fig. 63-9). However, it should be noted that skin as well as other tissues contain microfibrils devoid of elastin.

Biology of the Elastic Fibers

Characterization of elastin was hindered for a long time by its extreme insolubility in mature animal tissues. However, it was discovered that this insolubility is attributable to the presence of complex covalent cross-links, known as desmosines (see Section “Fibrillogenesis and Cross-Linking of Elastin”), whose formation can be prevented by maintaining the animals on a copper-deficient diet or by feeding them lathyrogens, such as β-aminopropionitrile, which inhibit elastin and collagens cross-linking. Once the formation of the cross-links is prevented, a large fraction of newly synthesized elastin can be extracted from the tissues.79

Cloning of cDNAs corresponding to the human elastin mRNA, approximately 3.5 kilobases (kb) in size, has allowed elucidation of the entire primary sequence of tropoelastin80–84 (Fig. 63-10). The basic molecular unit of elastin is a linear polypeptide, known as tropoelastin, that consists of approximately 800 amino acids with a molecular mass of around 70 kDa. Similar to collagens type I, glycine accounts for about one-third of the amino acid residues in elastin; however, it is not evenly distributed as it is in a typical collagens sequence. Instead, the glycine residues are grouped in valine- and proline-rich regions, which are interspersed with alanine-rich sequences. Elastin also contains some hydroxyproline, but the relative content of this amino acid is considerably lower than that in collagens, and the values for hydroxyproline are variable. Elastin does not contain hydroxylysine or carbohydrate moieties covalently linked to the polypeptide chain.

A characteristic feature of elastic fibers is the presence of cross-links that covalently bind elastin polypeptide chains into a fiber network. The deduced amino acid sequence depicts alternating segments of cross-link domains, characterized by the presence of lysyl residues separated by two or three alanine residues, and hydrophobic domains. The two major cross-link compounds, desmosine and its isomer, isodesmosine, are structures that appear to be unique to elastin.79 The content of desmosines in various elastin preparations has been shown to be fairly constant, approximately 1.5 residues per 1,000 amino acids. Consequently, assay of desmosine and isodesmosine can provide a quantitative measure of the elastin content in tissues.79

Constitutive and Alternative Splicing of Elastin

The human elastin gene consists of 34 separate exons spanning a total of 45 kb of genomic DNA (see Fig. 63-10). The information stored in the DNA sequence is transcribed in the nucleus of the cells to a large mRNA precursor, which undergoes several posttranscriptional modifications, including splicing and polyadenylation. The processed molecules are then transported to the cytoplasm, where the functional mRNAs serve as templates for the synthesis of elastin polypeptides during translation. Several lines of evidence suggest that the rate of elastin biosynthesis is largely regulated by the abundance of the functional mRNA, and consequently, assay of elastin mRNA levels allows measurements of the rate of elastin biosynthesis in tissues and cells.

An early observation during the isolation of human elastin cDNAs was that several overlapping clones were identical, with the exception of short sequences that were absent in some clones but present in others.81,83,84 Comparison of these sequences in the cDNAs with the corresponding genomic DNA segments indicated that the differences were due to alternative splicing of certain exons during posttranscriptional processing of the elastin pre-mRNA. In fact, at least six exons in the human gene have been reported to be subject to alternative splicing, and several of the variant mRNAs are translated into protein. This mechanism can provide significant variation in the amino acid composition of individual elastin polypeptides and presumably in the function of the elastic fibers in different tissues. However, the developmental significance and tissue specificity of alternative splicing have not been elucidated in detail.

Regulation of Elastin Gene Expression

The elastin promoter contains a remarkable constellation of potential binding sites for transcriptional regulatory factors indicative of complex transcriptional regulation. These include multiple Sp1 and AP2 binding sites, putative glucocorticoid-responsive elements, and phorbol ester tumor promoter, 12-O-tetradecanoylphorbol and cyclic adenosine monophosphate-responsive elements. The absence of a TATA box in the promoter region suggests that there may be multiple sites of transcriptional initiation, and various molecular tests support this notion.82 Functional analyses of the human elastin promoter segment have been carried out by constructing a panel of promoter-reporter gene (such as chloramphenicol acetyl transferase) constructs.85 Use of these constructs in transient transfection of a variety of cultured cells, including human skin fibroblasts, indicated that the core promoter necessary for basal expression of the gene is contained within the region −128 to −1 (in reference to translation initiation site −1, +1), and the upstream sequences contain several upregulatory and downregulatory elements. Development of transgenic mice expressing the human elastin promoter has revealed that 5.2 kb of DNA flanking the human elastin gene contains the elements necessary for tissue-specific expression.86 In addition to the 5′ upstream sequences, the first introns of both the bovine and human elastin genes contain regions of extremely strong sequence homology. Because it has been demonstrated that the first intron of three different collagens genes contains segments that act as enhancer elements of the promoter activity,87 the strong conservation within intron 1 of the elastin gene suggests the possible presence of an enhancer in this gene as well.

Cytokine and Hormonal Regulation of Elastin Synthesis

The presence of multiple cis-acting elements in the elastin promoter region suggests that elastin gene expression is subject to modulation at the transcriptional level by trans-acting factors. For example, previous studies have indicated that tumor necrosis factor-α decreases elastin mRNA abundance primarily by suppressing promoter activity.88 In other studies, TGF-β has been shown to upregulate the abundance of human elastin mRNA by up to approximately 30-fold,89 but evidence from transient transfection assays suggests that this upregulation is, at least in part, posttranscriptional. In fact, assay of elastin mRNA half-life suggests that TGF-β stabilizes the elastin mRNA, which leads to elevated steady-state levels. These observations collectively suggest that mediators released from inflammatory cells can modulate elastin gene expression, and such modulation may play a role in diseases characterized by altered accumulation of elastic fibers in tissues.

Vitamin D3 also modulates elastin gene expression. Specifically, incubation of fibroblasts with vitamin D3 results in an 80%–90% decrease in total accumulation of tropoelastin, accompanied by a parallel decrease in steady-state levels of the corresponding mRNA.90 At the same time, insulin-like growth factor-1 enhances elastin gene expression at the transcriptional level.91 Although there is preliminary evidence for modulation of elastin gene expression by selective growth factors and hormone-like substances, the precise details are largely unknown.

Elastin Biosynthesis

Elastin biosynthesis involves several specific steps necessary for the assembly of elastic fibers. Several studies show that smooth muscle cells in culture synthesize relatively large quantities of elastin, which suggests that they may be the major source of elastin in tissues rich in elastic fibers, such as vascular connective tissue.92,93 The amount of elastin synthesized by cultured human skin fibroblasts is relatively small; nonetheless, fibroblasts may be the primary source of elastin in the dermis. After the completion of translation, newly synthesized polypeptides are translocated into the cisternae of the rough endoplasmic reticulum, and then transferred from the rough endoplasmic reticulum into the extracellular space. Several observations suggest that the secretion of elastin polypeptides is a process that involves the microtubules of the cells and that the molecules may be transferred out of the cells packaged in Golgi vacuoles or related vesicles.

Fibrillogenesis and Cross-Linking of Elastin

The formation of desmosines occurs in the extracellular space, and the first step is the oxidative deamination of lysyl residues to corresponding aldehydes, known as allysines. This conversion is catalyzed by copper-requiring enzymes, lysyl oxidases.94,95 The desmosines are formed by the fusion of three allysines and a fourth unmodified lysyl residue in two adjacent tropoelastin chains. Thus, the desmosines link the individual elastin polypeptides into an insoluble network.

Lysyl oxidases, a group of closely related enzymes, require copper and molecular oxygen as cofactors. It has been shown that the activity of lysyl oxidases is reduced with copper deficiency and that it is associated with a disorder of connective tissue.96,97 In particular, the individual tropoelastin polypeptides remain soluble and the elastin-rich tissues are fragile.

Degradation and Remodeling of Elastin

Although the metabolic turnover of elastin is very slow compared with that of proteins in general, a portion of the body's elastin is continuously degraded and may partly be replaced by newly synthesized fibers.98 In addition, degradation of elastin is markedly increased in a variety of pathologic conditions.99 Thus, the tissues containing elastin must contain proteolytic enzymes that are capable of degrading elastic fibers.

Evidence for elastase, a specific elastolytic enzyme, was first obtained from study of the pancreas, and since then elastolytic enzymes have been detected in several other tissues, as well as in a variety of cell types, including polymorphonuclear leukocytes, monocytes/macrophages, and platelets.100 However, elastases from different sources (pancreas vs. polymorphonuclear leukocytes) have different cleavage specificities, as shown by peptide mapping of degradation products.101

The classic elastases are serine proteases that degrade insoluble elastic fibers at neutral or slightly alkaline pH. The activity of these enzymes is inhibited by serum factors, such as α1-antitrypsin and α2-macroglobulin. In addition to the classic elastases, which are serine proteases, others were shown to be metalloenzymes requiring calcium for their activity. One such enzyme is secreted by macrophages isolated from human alveolar macrophage exudates.102

Microfibrillar Proteins

At the electron microscopic level, the microfibrils of the dermis appear in cross section as an outer electron-dense shell surrounding an inner lucid core, and in longitudinal section as a beaded chain, which suggests that they may be composed of more than one protein. Moreover, there is no immunoreactive elastin associated with microfibrils located close to the dermal–epidermal junction, but as microfibrils traverse the dermis, they are associated with an increasing amount of amorphous elastin. Because of their insolubility and apparent complexity, chemical characterization of the microfibrils has progressed slowly until recently. These structural proteins can now be divided into several groups based on their molecular characteristics. Table 63-5 lists these proteins and provides some distinguishing features, including their human chromosomal localizations. These groups include several gene families that share common structural motifs.

The Fibrillins

The largest, and perhaps the most important microfibrillar proteins for biology and pathology of the dermis, are the fibrillins, 350-kDa glycoproteins that form an integral part of the microfibril structure.103,104 Electron microscopic images of human monomeric fibrillin synthesized by cultured fibroblasts show an extended flexible molecule, approximately 148-nm long and 2.2-nm wide. Multiple fibrillin molecules align in a parallel head-to-tail fashion to form microfibrils. Molecular cloning studies have so far identified three distinct homologous human genes, (1) FBN1, (2) FBN2, and (3) FBN3, encoding fibrillin-1, -2, and –3, respectively.104 Analysis of the amino acid sequences deduced from cloned cDNAs showed that these proteins contain multiple repeats of a motif initially observed in epidermal growth factor (EGF) precursor molecule. These motifs have either six or eight conserved cysteines, and many of them contain a consensus sequence for calcium binding.103,104 Such EGF-like motifs are quite frequent in ECM proteins, and occur as several copies in, for instance, laminins, fibulins, nidogens, agrin, perlecan, tenascins, and latent TGF-β-binding proteins (LTBPs). Besides important structural functions for tissue architecture, it is now clear that fibrillins regulate signaling events and control directly and indirectly cellular activities. Therefore, they play critical roles during development and disease processes.105–107

The Latent Transforming Growth Factor-β-Binding Proteins

It is of interest that LTBPs contain EGF-like motifs similar to those found in fibrillins (Fig. 63-11 provides a comparison of the structures of fibrillins and LTBPs). TGF-β is always secreted as a latent complex with Latency-Associated Peptide (LAP) and this complex is bound to an LTBP (Fig. 63-12). To date, four distinct genes (LTBP1 to LTBP4) have been identified, coding for LTBPs ranging in size from 125 to 310 kDa.108,109 Although LTBPs may facilitate the secretion of TGF-β or association of the inactive complex to the cell surface where activation takes place, they are also found as free proteins associated with components of the ECM. Development of an LTBP3 knockout mouse revealed craniofacial malformations, perturbed ossification, and development of osteosclerosis and osteoarthritis in the deficient animals. These observations were interpreted to mean that LTBP3 is important for the control of TGF-β action and, specifically, that LTBPs may modulate TGF-β bioavailability. Immunohistologic studies have localized LTBP1 to microfibrils in elastic fibers, which strongly suggests that one or more of the LTBPs may be a component of these fibrils. Furthermore, levels of LTBP1 are altered in a number of pathologic conditions, including solar elastosis.110

The Fibulins

The fibulins are a family of ECM glycoproteins that contain tandem EGF-like repeats similar to fibrillins and LTBPs, and a common carboxyl-terminal globular domain (see Fig. 63-11). Molecular cloning has identified seven fibulin genes (see Table 63-5). Fibulins are widely distributed in connective tissues, including the skin, where they are located within the elastic fibers.111,112 Fibulin-4- and fibulin-5-deficient mice, developed by targeted ablation of the corresponding genes, demonstrate marked reduction in elastogenesis. Fibulin-4-deficient mice have a severe phenotype that leads to perinatal death, primarily due to pulmonary and cardiovascular abnormalities.113 Particularly interesting is the phenotype of fibulin-5 knockout mice, which show progressive laxity of skin in addition to vascular abnormalities and emphysematous lung changes.114,115 The skin changes are reminiscent of those noted in patients with congenital cutis laxa. In fact, mutations in the FBLN4 and FBLN5 genes have been documented in some cases of cutis laxa.116–118

A number of other proteins unrelated to the fibrillins, LTBPs, and fibulins are constituents of the microfibrils. They include the lysyl oxidases,94,95 interface proteins called emilins,119 and several proteins120 belonging to the families of Microfibril-Associated Proteins (MAP) or microfibril-associated glycoproteins (see Table 63-5). Among the latter, microfibril-associated protein 1 is remarkable in that it is extremely acidic, with glutamic acid comprising 23% and aspartic acid 6% of the residues. The extremely acidic nature of the protein suggests that it may have an important function in the assembly of the very basic tropoelastin molecules. Finally, lysyl oxidases are critical for the cross-linking and stabilization of the elastic fiber structures.

Historically, terms such as ground substance or mucopolysaccharides were used to describe proteoglycans because of their histologic appearance and their thick and mucinous nature when isolated. Because of these physical properties, proteoglycans were difficult to study. Our understanding of proteoglycans has markedly advanced as a result of increased information about their core proteins and the subsequent use of molecular tools to study the expression and function of both the protein core and glucosaminoglucan (GAG) components. Now, proteoglycans are recognized as a structurally unique and highly diverse group of macromolecules. The most distinguishing structural characteristic of proteoglycans is that they comprise both a core protein and covalently linked linear carbohydrate chains known as GAGs. Proteoglycans constitute an important portion of external cellular membranes and ECM of the skin. Their ability to bind proteins and alter protein–protein interactions or enzymatic activities has identified them as important determinants of cellular responsiveness in development, homeostasis, and disease. Because GAGs are highly polyanionic and bear a high charge density, they are a critical component of the skin, and together with the core proteins, they impart a unique set of functions that are critical to a large number of biologic processes.

Structure and Synthesis of Proteoglycans

The prototypical proteoglycan consists of a single core protein linked to one or more GAGs (Fig. 63-13). Each core protein has the capacity to accept a variety of GAG chains. Consequently, the nomenclature for proteoglycans is complicated in that individual molecules must be defined based on the core protein and the associated GAGs.

The Glycosaminoglycan Chains

The GAG chains are defined based on the assembly of essential sugar residues.121,122 These sugars are organized as disaccharide pairs that usually consist of an acidic sugar that is either an iduronic acid or glucuronic acid alternating with a hexosamine that is either glucosamine or galactosamine. When the chains are assembled as disaccharides, the choice of sugars and the linkage between them is used to assign a name to the GAG chain. Thus, several terms are used to define GAGs (see Fig. 63-13), including hyaluronic acid, which contains a glucuronic acid alternating with N-acetylglucosamine-linked β1–3 and β1–4; heparan sulfate, containing iduronic or glucuronic acids alternating with N-acetylglucosamine-linked β1–4; chondroitin sulfate, containing glucuronic acid alternating with N-acetylgalactosamine-linked β1–3 and β1–4; and keratan sulfate, containing galactose alternating with N-acetylglucosamine-linked β1–4 and β1–3. A fifth general term used to describe GAGs, and of particular importance to cutaneous biology, is dermatan sulfate. This form of GAG (also known as chondroitin sulfate B) is similar to other chondroitin sulfates except that it contains a high proportion of iduronic acids in place of glucuronic acid and has more variable sulfation. Dermatan sulfate shares features of both chondroitin sulfate (N-acetylgalactosamine) and heparan sulfate (iduronate) (see Fig. 63-13).

The linear chains of linked disaccharide units in a GAG are highly variable in size, ranging from as few as ten disaccharides to several thousand. Thus, the mass of naturally occurring GAGs typically ranges between 5 × 103 Da and 5 × 107 Da for hyaluronic acid. Further variability is introduced into the GAG chain by epimerization reactions and sulfation reactions. The control of these reactions depends on the nature of the GAG, the core protein to which it may be attached, the cell type, and the cell environment. For example, the simplest GAG, hyaluronic acid, is never sulfated, whereas other GAGs can be sulfated to varying degree. The highly sulfated sugars tend to occur in specific regions of the GAG chain and are interspersed with areas of low sulfation. Such discrete domains of low or high sulfation are believed to determine the interactions between proteoglycans and GAGs and their many binding partners. The size, disaccharide composition, and sulfation are of critical importance in understanding GAGs. These parameters influence function and are the mechanism by which instructions are defined within the molecule. Thus, the linear GAG should be considered as a molecule containing information, just as information is encoded in a protein.121,122

Synthesis of GAGs (with the exception of hyaluronan) occurs in the Golgi apparatus, and the sequence information is determined by the activity and location of multiple specific enzymes along this pathway. Today, many of the enzymes that control heparan sulfate synthesis and those enzymes responsible for postsynthetic modifications have been identified and characterized. The sulfated GAGs are all synthesized on core proteins. The tetrasaccharide xylose–galactose–galactose–glucuronic acid is first assembled on the core protein by beginning with a xylosyltransferase that forms a linkage between xylose and a serine residue in the core protein. Subsequent elongation reactions that occur in the Golgi apparatus define the nature of the GAG chain. It is appreciated that part of the information necessary to direct synthesis of the GAG is encoded within the core protein sequence itself. The final product, core protein with attached GAG, defines the proteoglycan (see Fig. 63-13). Hyaluronan, the only GAG produced without attachment to a core protein, is synthesized by an enzyme complex at the plasma membrane and then extruded into the extracellular space.

The Variety of Proteoglycan Core Proteins

For purposes of organization, it is useful to group proteoglycans according to their site of expression by the cell. Specific proteoglycan core proteins have been identified within the cell, attached to the cell surface, bound within the ECM, and released in soluble form (Table 63-6).

Intracellular Proteoglycan: Serglycin

Serglycin is the best-known intracellular proteoglycan and is found within the secretory granules of hematopoietic cells, including mast cells, leukocytes, and eosinophils. The core protein is processed further to contain either heparan sulfate or chondroitin sulfate GAG. The heparan sulfate form is found in serosal mast cells and is a major source from which heparin is pharmacologically derived. The serglycin peptide core is composed primarily of tandem serine–glycine repeats and has an estimated mass of 16–18 kDa before GAG addition and 60–750 kDa after the addition of multiple heterogenous GAG chains.123 Thus, serglycin is a unique proteoglycan because of its distinctive core protein sequence, and intracellular localization. In the skin, serglycin is found whenever mast cells or eosinophils enter the dermal stroma. On release, serglycin becomes a major source for the delivery of highly sulfated heparan sulfate GAG.

Cell Surface Proteoglycans: Syndecans and Glypicans

Several cell surface proteoglycans are very interesting because they act at the interface between the plasma membrane and the extracellular environment. They are attached to the cell surface either by a phospholipid anchor (i.e., glypican family) or as membrane-spanning core proteins (i.e., syndecan family).

The glypicans share the glycophosphatidyl inositol anchorage mechanism and a unique cysteine motif that is likely to impart a compact tertiary structure to the proteoglycans. Consequently, glypicans can be visualized as a compact protein presenting GAG chains in close proximity to the outer surface of the plasma membrane. Six isoforms (glypican-1 to 6) have been described,124–126 and expression of some glypicans may be altered in inflamed skin and chronic wounds.127

The syndecan core proteins consist of a short C-terminal cytoplasmic region, a transmembrane domain, and an extracellular region containing attachment sites for GAG chains. Unlike glypicans, syndecans span the plasma membrane and extend beyond the surface of the cell. In this way, syndecans are in a unique position to connect extracellular GAG to structures within the cytoplasm.124,126 The intracellular domain endows syndecans to participate in several signal transduction pathways, including the protein kinase C pathway.127–129 Syndecan transcripts and proteins are expressed in distinct patterns during development and in mature tissues.130 Syndecan-1 is particularly abundant on keratinocytes and can vary the nature of attached GAG chains as keratinocytes differentiate.131 During wound repair, syndecan-1 and syndecan-4 are highly induced in the dermis and granulation tissue.132 Deletion of syndecan-4 from mice decreases the rate of wound repair.133 During malignant transformation, syndecan-1 expression decreases in the epidermis,134 and expression can alter malignant behavior in select cell types.135 Thus, syndecans have received particular attention as proteoglycans that modify cell function. Because syndecans and other proteoglycans can contain more than one GAG species, either simultaneously or under different cellular circumstances, GAG expression provides another mechanism of regulation in addition to protein expression.

Large Aggregating Extracellular Matrix Proteoglycans: Aggrecan and Versican

Aggrecan, a large proteoglycan found in cartilage, has a core protein containing a region with over 100 serine–glycine dipeptides that serve as attachment sites for up to 130 GAG chains. As many as 100 aggrecan molecules can bind to a single hyaluronic acid molecule. Thus, the overall proteoglycan aggregate (from which the name aggrecan is derived) can have a mass near 200,000 kDa.135

In the skin, fibroblasts produce large aggregating proteoglycans similar to aggrecan, the best known being versican.135,136 The core protein of versican contains attachment sites for 12–15 GAG chains, which are primarily chondroitin sulfate or dermatan sulfate. Versican, like aggrecan, binds hyaluronic acid, which enables it to form large aggregates. In skin, versican has been identified in the dermis (fibroblasts) and epidermis (keratinocytes) and demonstrates selective upregulation in response to TGF-β.137 Thus, the size, distribution, and abundance of versican in the skin suggest that it is an important molecule in the regulation of the behavior of the skin.

Small Extracellular Matrix Proteoglycans: Decorin

Several genes encoding smaller ECM proteoglycans have been identified. One family is characterized by a leucine-rich repeat motif. The prototype of this family is decorin, an approximately 36-kDa secreted proteoglycan.125,135,136 Decorin is a ubiquitous component of connective tissues and it is found in abundance in skin. The decorin core protein has a single dermatan sulfate chain covalently bound to a serine residue at amino acid position 4 and, like many other proteoglycans, also has N-linked oligosaccharides. Thus, after GAG addition, decorin can have a mass of 80 kDa. Decorin received its name from observations that this molecule closely associates with collagens fibrils and can be seen to “decorate” the fibrils in vivo. This interaction is attributed to the ability of the decorin core protein to directly bind to collagens type I.138 The single GAG chain of decorin also binds to tenascin-X, another ECM protein that colocalizes with collagens fibrils in connective tissues.139 These binding interactions contribute to collagens fibril formation and influence function. Interestingly, similar phenotypes showing abnormal collagens morphology are seen in a patient deficient in tenascin-X and in the murine decorin knockout model.140

Function of Proteoglycans

To understand how proteoglycans function in the skin it is essential to appreciate the basics of their organization: (1) proteoglycans are part protein and part GAG; (2) GAGs are heterogeneous but encode specific information; (3) proteoglycan core proteins are expressed in different cellular compartments; and (4) proteoglycan expression and GAG composition vary according to cellular context. These characteristics are important, because many proteoglycan functions depend on their ability to bind other molecules in the environment, and because both the core protein and GAG chains may be mediating or influencing molecular interactions. Similar GAGs can be attached to diverse core proteins. This ability to interchange GAGs among cores adds complexity to the study of proteoglycan function, and in many cases, proteoglycan properties are defined by the GAG chains. It implies that certain genes encoding different core proteins to which similar GAGs are attached may have a similar function. The list of molecules to which the major skin proteoglycans heparan sulfate and dermatan sulfate bind is quite extensive (Table 63-7). Therefore, several functions for proteoglycans have been proposed that include the particular biologic system in which the ligand is involved.

Some of the most compelling experimental evidence for the functions of heparan and dermatan sulfate proteoglycans identifies the ability to bind several growth factors, cytokines, and components of the ECM.141 For instance, heparan sulfate is required for the function of growth factors such as several members of the fibroblast growth factor family,142,143 hepatocyte growth factor/scatter factor,144 and vascular endothelial growth factor.145 In this model, the heparan sulfate proteoglycan is found either at the cell surface or in a soluble form. The molecules then assemble to form a ternary complex at the cell surface between the growth factor, its specific high-affinity signaling receptor, and the proteoglycan.146 Only after this ternary complex is formed does the cell receive the growth factor signal to begin to proliferate, differentiate, or migrate. Heparan sulfate-dependent growth factors, such as members of the fibroblast growth factor family, are induced in healing wounds and influence the wound repair process through the stimulation of keratinocyte proliferation, fibroblast growth, and angiogenesis. Because dermatan sulfate is the predominant GAG in wound fluid and because it can bind a variety of growth factors, cytokines, and ECM proteins, further study of this GAG and the proteoglycans it comprises promises to be an exciting area for future investigation.

In addition to binding to growth factors, heparan sulfate proteoglycans play an important role in adhesion to the ECM. In the current model, cell surface proteoglycans, in conjunction with other matrix-binding molecules such as the integrins, help the cell adhere to the ECM.124,126 Furthermore, the formation of these focal adhesions requires heparan sulfate and the subsequent activation of protein kinase C by a domain in the syndecan-4 core protein cytoplasmic tail. In this setting, both the extracellular GAG and intracellular core protein of the proteoglycan have demonstrable signaling function.128,129

Other proteoglycans also interact with ECM molecules. Chondroitin sulfate and dermatan sulfate bind fibronectin and laminin. As mentioned earlier, decorin, to which dermatan sulfate is typically attached, is known to associate primarily with type I collagens via its core protein.138 In mice, selective disruption of the decorin gene leads to abnormal collagens morphology and increased skin fragility.140 Thus, through targeted disruption of the gene in the mouse, the function of decorin as a stabilizer of collagens fibrils has been directly confirmed.

The function of the large ECM proteoglycans in skin is thought to relate primarily to the physical properties inherent in their large mass and charge density. Hyaluronan has been studied extensively from this perspective because of its extreme hydrophilicity and viscosity in dilute solution. As discussed earlier, hyaluronic acid is pure GAG and is synthesized extracellularly without a core protein. The genes for human hyaluronan synthase have been cloned and identified.147 The expression of hyaluronan is developmentally regulated in skin and changes during wound repair. It has been proposed that the physicochemical properties of hyaluronan serve to expand the matrix and thus aid cell movements. Other physical properties attributed to GAG and large proteoglycan complexes, such as those formed with versican or basement membrane proteoglycans, include their ability to act as anionic filters and elastic cushions and their function in salt and water balance. In fetal skin, the high relative content of hyaluronan has been associated with the ability of fetal skin to heal without scar.

Many components of the ECM regulate cellular behavior directly and indirectly. An indirect mechanism is provided by the modulation and control of the activity of cytokines, growth factors, and chemokines by proteins in the ECM. A direct and most important mechanism relies on the control of cellular activities through adhesive interactions between cell surface receptors and ECM proteins (Fig. 63-14). Several classes of cell surface receptors have been identified, such as the syndecans,124,126 DDRs (Discoidin Domain Receptors),148 and integrins.10,149–152 The latter are noncovalently linked heterodimers of one α and one β subunits. Integrins have a short cytoplasmic domain (except for the 1,000 residue-long intracellular domain of the integrin β4 subunit expressed by keratinocytes), a single stretch transmembrane domain, and a large extracellular domain interacting with specific proteins of the ECM or with counter-receptor present on the surface of circulating cells, including microbial proteins. In human, 18 α and 8 β subunits have been characterized, assembling into 24 different combinations. The integrin cytoplasmic domain lacks enzymatic activity; however, by associating with multiple proteins into signaling platforms, they activate diverse molecular pathways.151,152 Adhesive interactions between integrins and ECM proteins control many cellular functions, such as migration, survival, differentiation, contraction, transmission of forces, and expression of specific genes.149,150 Expression of a distinct set of integrins on the cell surface determines specificity, range, and versatility of the interactions with ECM proteins. For instance, fibroblasts express, among others, integrins with the property to interact with native fibril-forming collagens, fibronectin, and vitronectin.10 In addition, the repertoire of integrins expressed by a given cell type may adapt to the differentiation stage during development and to specific physiological and pathological features of the microenvironment, such as wound healing, aging, inflammation, and fibrosis.153 Integrins are also expressed by inflammatory cells and control their migration and targeting to specific sites of inflammation.154 Finally, it has been suggested that integrins cross talk or act in synergy with growth factors, cytokines, and cognate receptors.155 Together, the many facets of integrins endow these proteins with crucial roles in various physiological and pathological processes such as fibrosis, tissue repair, and carcinoma progression.

The many components of the ECM in the dermal compartment of the skin have an elaborate biosynthesis, maturation, and turnover, often a complex structure, and they are assembled into exquisitely complex networks. Therefore, it is not surprising that many diseases exist due to disturbances in the genetic expression, structure, homeostasis, and assemblies of ECM proteins.

Based on our understanding of the molecular biology and the metabolism of the ECM components, it is now possible to identify discrete points at which defects could occur. These can be located at different levels (Box 63-1). Many mutations in the genes coding for ECM proteins or in the enzymes involved in their posttranslational modifications or their supramolecular assemblies into defined networks have been identified as outlined in preceding paragraphs.

ECM deposition and assembly into insoluble and complex polymers are tightly controlled mechanisms, which undergo substantial variations during development, tissue remodeling, repair, ageing, wound healing, or fibrosis. Several ECM proteins are also targets for circulating antibodies, leading to autoimmune diseases. Most interesting for dermatology are the autoimmune blistering diseases that are discussed in detail in other chapters. There is also accumulating evidence that many ECM proteins manifest regulatory functions, which can be disturbed and lead to disease processes.

The term collagens disease implies that a clinical condition involves an abnormality in the structure, synthesis, or degradation of collagens. This term is frequently used to characterize a clinically heterogeneous group of inflammatory diseases, including lupus erythematosus, scleroderma, and dermatomyositis. In the classic nomenclature, the use of the term collagens disease for these conditions was based on morphologic changes, known as fibrinoid degeneration, that have been interpreted to represent changes in collagens fibers. However, there is currently no evidence for a primary defect in collagens in these diseases. On the basis of available biochemical evidence, scleroderma is the only clinical condition among the classic collagens diseases that involves dysregulation of collagens metabolism (see also Chapter 157).

Collagens is an unusual protein in many respects, and its structure and metabolism involve a number of special features that are critical for deposition of normal collagens fibers. In a primary collagens disease, such a defect could be an inherited abnormality in the structure of the collagens or procollagens or in the enzymes participating in the biosynthesis and degradation of collagens. Indeed, there are many diseases in which a basic biochemical defect in collagens has been described, and several heritable connective tissue diseases with cutaneous involvement are now known to result from specific molecular defects in collagens genes (see Chapter 137). Many of these diseases involve insertions, deletions, or single-base substitutions in the genes that alter the primary structure of the protein.9,13,14 In several acquired diseases, the regulation of collagens gene expression is disturbed, which leads to altered deposition of collagens in tissues. The diversity of collagens pathology is exemplified by Ehlers–Danlos syndrome and fibrotic skin diseases (see Chapters 137 and 157). Ehlers–Danlos syndrome comprises a group of phenotypically similar conditions that frequently result from abnormalities in the structure of collagens or in enzymes modifying collagens molecules.19,20,56,57 In contrast, the fibrotic skin diseases exemplify conditions with altered regulation of collagens gene expression that leads to excessive accumulation of collagens in tissues.61

Several heritable and acquired connective tissue diseases have been shown to be associated with aberrations in the elastic fibers, including pseudoxanthoma elasticum, cutis laxa, and the Buschke–Ollendorff syndrome (Table 63-8; see Chapter 137). In addition, clinically distinct entities with phenotypes closely resembling pseudoxanthoma elasticum (e.g., elastoderma) or cutis laxa (e.g., wrinkly skin syndrome) can be recognized.

There has been significant progress recently in identifying and understanding the molecular mechanisms associated with mutations in the fibrillin genes.156 Heterozygous mutations that affect the structure or lead to a reduced synthesis are the cause of the Marfan syndrome. This disease is clinically characterized by abnormalities and disturbances in the ocular, skeletal, and cardiovascular systems. The patients display a thin skin, with extensive striae distensae; some, however, also have contractures. A careful analysis of mouse models with similar molecular defects revealed that the mutations in the fibrillin gene are associated with increased TGF-β signaling. Interestingly, the phenotype presenting with development of a mitral valve prolapse in this mice could be reverted or delayed by addition of TGF-β neutralizing antibodies. These data indicate that fibrillin-1 is likely to be involved in the tight regulation of TGF-β activity. In a clinical study, application of angiotensin II inhibitors to patients with Marfan syndrome consistently and significantly slowed the progression of aortic root dilatation.157

Multiple cutaneous disorders are associated with abnormal proteoglycan synthesis or deposition. A deficiency of decorin core protein has been described as the cause of a variant form of Ehlers–Danlos syndrome. Even infectious diseases are influenced by proteoglycan expression. A good example is provided by the decorin-deficient mice, which are rather resistant to infection with Borrelia burgdorferi. Presumably, the resistance is caused by a partial dependence of the organism on decorin for binding in the dermis.158 During aging, the composition of GAG in the skin undergoes marked changes and it has been proposed that the decrease in hyaluronic acid explains diminished skin turgor.159,160 Retinoids also modify this response.161 In keloids, hyaluronic acid synthesis is elevated.162 Thyroid hormone also has a marked effect on the synthesis of proteoglycans and GAGs. The manifestation of this response is seen in thyroid dermopathy (Chapter 151), in which the mucinous material deposited is predominantly hyaluronic acid and chondroitin sulfate. Other pathologic skin conditions, including pseudoxanthoma elasticum, scleroderma, psoriasis, a variant form of Ehlers–Danlos syndrome, lichen myxedematosus, and ultraviolet B-irradiated skin, have also been reported to be characterized by abnormal amounts of proteoglycans. These associations do not imply that abnormal proteoglycan metabolism is responsible for these disorders, but they suggest that much of the pathophysiology of skin disease can be influenced by the function of cutaneous proteoglycans.

Mutations in the genes coding for key ECM proteins often lead to functional consequences, which are associated with severe clinical symptoms. Although structural alterations can certainly explain a number of symptoms, it is clear that not all pleiotropic clinical alterations are directly and solely due to the abnormalities in protein structure. Evidence is now accumulating that induction of endoplasmic reticulum stress represents an additional mechanism explaining some of the deleterious effects resulting from mutations in the genes needed for the elaboration of ECM proteins.22,163–165 For instance, the accumulation of misfolded or unfolded mutant polypeptides in the endoplasmic reticulum induces the so-called “unfolded protein response.”166–168 It results in detrimental processes of diverse severity, ranging from increased protein targeting to the proteasome for destruction, autophagy, general reduction in protein synthesis including that of the abnormal protein, to complete cellular dysfunction with apoptosis of the cells (Fig. 63-15). Such a deleterious outcome has been demonstrated for mutations in several ECM genes, often resulting in severe clinical subsets of osteogenesis imperfecta or some inheritable cartilage diseases. However, it remains to be seen whether these mechanisms could play a much broader role and apply to other inherited or acquired disorders of the ECM.