Mechanisms of maturation and ageing of collagen

Abstract

The deleterious age-related changes in collagen that manifest in the stiffening of the joints, the vascular system and the renal and retinal capillaries are primarily due to the intermolecular cross-linking of the collagen molecules within the tissues. The formation of cross-links was elegantly demonstrated by Verzar over 40 years ago but the nature and mechanisms are only now being unravelled. Cross-linking involves two different mechanisms, one a precise enzymically controlled cross-linking during development and maturation and the other an adventitious non-enzymic mechanism following maturation of the tissue. It is this additional non-enzymic cross-linking, known as glycation, involving reaction with glucose and subsequent oxidation products of the complex, that is the major cause of dysfunction of collagenous tissues in old age. The process is accelerated in diabetic subjects due to the higher levels of glucose. The effect of glycation on cell-matrix interactions is now being studied and may be shown to be an equally important aspect of ageing of collagen. An understanding of these mechanisms is now leading to the development of inhibitors of glycation and compounds capable of cleaving the cross-links, thus alleviating the devastating effects of ageing.

Introduction

The overall shape and function, in terms of flexibility and locomotion, of the human skeletal system depend on a basic framework of collagen fibres (Alexander, 1981). The collagen fibres are essentially inextensible and therefore provide mechanical strength and through that strength confer and maintain form whilst allowing flexibility between various organs of the body. Thus, the randomly orientated fibres of the skin permit considerable extension of the tissue until the fibres themselves are loaded. The fibres of tendons are aligned in parallel and therefore loaded instantly, permitting maximum transfer of the energy of muscle contraction to the skeleton. The fibres of bone are organised in concentric layers to maximise for torsional and compressive stresses and the rigidity is conferred on the bone by mineralisation. The surface of cartilage on the bone extremities allows efficient movement of the skeleton through lubricating joints following muscular contraction, and stability is provided by a small proportion of fine collagen fibres in a mainly polysaccharide gel. The cornea provides an example of well ordered fibres in precise layers at a defined angle to each other thereby allowing the transmission of light. In contrast to these fibrous structures the network structure of basement membranes provides a filtration system and an attachment site for cells, for example the kidney glomeruli and arterial basement membranes. This biological diversity of function of collagenous tissues is primarily due to the fact that these fibres are biopolymers of one of several genetically distinct collagens, which are to some extent tissue specific as will be discussed later.

During ageing, changes occur in the collagenous framework. These changes in the physical properties of the fibres are reflected in the well-documented increases in stiffness of skin, tendon, bone and joints in old age. The major changes are an increase in rigidity of the tissue, the fibres ultimately becoming brittle (Torp et al., 1975, Viidik, 1982, Uitto, 1986). Such changes are clearly deleterious to the optimal functioning of the locomotive system, the elastic vascular system and the filtration properties of the basement membranes. Recent studies have emphasised the importance of cell-matrix interactions, particularly during development, and preliminary studies indicate that a reduction in the efficacy of cell-collagen interactions also occurs during ageing, the effect being particularly important in the case of basement membranes (see later).

In order to understand the mechanism of these age-related changes it is essential to understand the role of collagen in determining the mechanical properties of the tissue; the changes in the cross-linking as a major cause of the observed change in mechanical properties; the role of the collagen types comprising the fibrous framework, and changes in the activity and collagen type expression of the cells associated with a particular tissue, occurring with senescence.

Several structural features of collagen bestow particular tissues with mechanical properties appropriate to their widely varying functions. The total collagen content of the tissue is obviously an important determinant of mechanical strength. The ability of the tissue to sustain an applied load is also determined by the orientation of the fibres which can vary markedly between tissues, e.g. unidirectional in tendon, laminated in cornea and random in skin. The collagen fibrils possess little strength in flexion or torsion but exhibit a very high tensile strength. The tensile strength increases considerably with age. In tendon the toe region of the stress-strain curve is shortened as the fibre packing is tighter and at the same time the stiffness of the fibre increases. Little change occurs in the linear modulus during maturation but in old age there is an increase in failure stress accompanied by a decrease in failure strain (Fig. 1), indicating increased cohesion between the microfibrils thus preventing slippage (Torp et al., 1975).

The diameter of the collagen fibres also plays a significant role in determining the mechanical properties of the tissue. For example, the plastic deformation or ‘creep’ of tissues is directly related to the proportion of small diameter fibrils (Parry, 1988). Conversely, the ability to withstand high stress levels is related to the proportion of large diameter fibrils. As the diameter increases the flexibility of the tissue decreases and there is a decrease in the ability to resist crack propagation. The variation in diameter of collagen fibres between tissues is illustrated by tendon (200 nm), skin (approx. 100 nm), cartilage (approx. 50 nm), and cornea (20 nm) and this can be related to their relative mechanical properties. In some tissues there is a bimodal distribution of fibre diameter, the voids between the large fibres being filled by small fibres, thus allowing a high collagen content but maintaining a flexibility of the tissue (Parry et al., 1978). With maturation, fibril diameters increase and may be either unimodular or bimodally distributed, whilst with senescence, diameters may decrease and tend towards bimodality in many tissues (Parry et al., 1978, Jones, 1991).

An additional structural feature of the collagen fibre is its possession of a crimp structure, a periodic light and dark banding structure of 100 μm with an angle of 5–25° seen under polarised light, which results from the planar zigzag wave along the fibre path. This crimp is believed to act as a shock absorber system, and is represented within the toe-region of the stress–strain curve, that is, within physiological levels of stress (Fig. 1a). With increasing age the angle of the planar waveform increases thus decreasing the shock absorbing effect (Gathercole and Keller, 1975).

The effect of physical training is of considerable interest to the understanding of the control mechanisms. Collagenous tissues appear to have a general systemic response to exercise, skin, tendon, and bone readily showing increases in weight and strength (Viidik, 1986). It has been suggested that load resistance training is more effective than endurance training in the development of connective tissue (Stone, 1988). Under normal conditions the flexor tendons are stronger than the extensor tendons, but following training the extensor tendons can approach the strength of flexor tendons (Woo et al., 1982). The mechanisms involved are as yet unknown but could be a direct effect on the collagen synthesising cells as evidenced by elevated levels of hydroxylases, by initiation of a growth promoter, and by a change in the blood hormone levels, or a change in the number and activity of the hormone receptors. On the other hand extremes have a different effect, excessive tension on fibres leads to degradation due to the stimulation of MMPs (Bailey et al., 1994).

The collagens exist as thick striated fibres, as non-fibrous networks in basement membranes, as non-striated filamentous structures, or fibril associated molecules. These morphologically different collagen structures are aggregates of one of more of over a dozen different collagen molecules. These molecules are, in fact, a family of closely related but genetically distinct proteins, possessing a basic structure of three polypeptide chains each with a Gly-X-Y repeat forming tightly bound triple helices which subsequently aggregate to form various types of supporting structures (for reviews see Kielty et al., 1993, Comper, 1996). Before further discussion of the mechanisms of ageing of collagen, it would be advantageous at this stage to review the nature of these different collagens (Fig. 2). At the present time there are 19 genetically distinct collagens, but the function of many of the minor collagens has yet to be elucidated.

The majority of the collagens can be classified according to the nature of their aggregated forms.

These collagens are long (300 nm) rod-like molecules which self-assemble in a parallel, quarter-staggered end over-lap arrangement (Fig. 2a) to form fibres possessing a characteristic band pattern, with a periodicity of 67 nm identifiable in the electron microscope. The collagen types in this group are Types I, II, III, and the minor collagens V and XI.

The type IV molecules are very long (400 nm) and flexible due to irregularities in the Gly-X-Y sequence. They form a ‘chicken-wire’ network which acts as the basic framework of basement membranes of vertebrates and invertebrates (Fig. 2b). Four molecules are associated in an antiparallel fashion through the amino termini to form a 110 nm overlap known as the 7S region, whilst the carboxy termini, which are comprised of a large non-triple helical peptide (NC1), interact with the NC1 region of an adjacent molecule to build up the ‘network’.

Types VIII and X collagen also form networks, and are often classed as ‘short-chain’ collagens. Both have been reported to form hexagonal lattices in Descemet’s membrane (Benya and Padilla, 1986) and growth plate cartilage respectively, although their supramolecular structures in other tissues have not yet been elucidated.

Type VI is observed as a loosely packed filamentous structure with an axial repeat of 100 nm and is formed by end-to-end alignment of tetramers (Bruns et al., 1986), Fig. 2c. These fibres occur in many tissues and it has been suggested that it may separate and align larger type I fibres (Bonaldo et al., 1990). Lateral aggregation of the tetramers has also been observed in both normal and diseased tissues.

Several collagens do not form homotypic fibres or networks but are associated with other fibre forming collagens, for example, type IX collagen decorates the surface of the type II collagen fibre (Vaughan et al., 1988) (Fig. 2d). Type IX, which consists of three collagenous domains separated by non-triple helical regions, aligns on the type II collagen fibre in an anti-parallel manner (Wu et al., 1992). Types XII and XIV are generally associated with the surface of type I collagen fibres (Sugrue et al., 1989, Van der Rest and Dublet, 1996), although both have also been localised in foetal cartilage (Watt et al., 1992). One of the three domains of type IX extends into the extracellular space and we have shown it to be more thermally stable (Miles et al., 1998).

Not all collagens fall easily into these groups. Type VII for example forms microfibres which underlie some basement membranes acting as short anchoring fibrils between the membrane and the underlying matrix. These short fibrils are formed by lateral aggregation of anti-parallel overlapped (60 nm) dimers. Many of the remaining collagen types are known only by their DNA sequence, and as such their macromolecular structures are unknown (Comper, 1996).

The biological diversity of function of collagenous tissue is primarily due to the variety of aggregated forms derived from the genetically distinct collagens listed above, which are to some extent tissue specific. Bone and tendon are predominantly fibrous type I collagen, the vascular system contains both types I and III, whilst cartilage contains predominantly type II collagen. The thin basement membranes are primarily type IV collagen. Most collagenous tissues also contain other, ‘minor’ collagens, and in some cases the individual fibres themselves may contain small proportions of another collagen type.

At the present time there is little information on the role of these additional collagens in modifying the physical properties of the fibre. For example, the presence of type III fibres is believed to confer the greater elasticity to embryonic and vascular tissues primarily because of their small diameter, but its effect when co-polymerised with type I is unknown. On the other hand type XI in cartilage and type V in many other tissues, may be involved in the nucleation of fibrous type II and type I respectively, whilst type IX, which occurs on the surface of type II, may play a role in determining the diameter of the fibril (Wotton et al., 1988). Types X and VIII collagens, which are structurally very similar have apparently widely differing functions. Type X is restricted to the calcifying edge of the growth plate and may form both hexagonal structures and type II associated fibrous mats (Reginato and Jimenez, 1991), whilst type VIII is expressed in a wide range of tissues in addition to Descemet’s membrane (Benya and Padilla, 1986), in which its macromolecular structure is not yet clear.

Changes in the collagenous matrix are readily observed during development (Reichenberger and Olsen, 1996) and are often reported as age changes, mainly because the differences in old age are harder to identify. However, it is now becoming clear that there are significant quantitative and qualitative changes in the collagenous tissues in old age. The composition of tissues may vary and involve alteration in the proportion of different collagens in a particular tissue. Few studies however have been carried out on compositional changes during ageing, although some have been studied during development. For example, embryonic dermis contains about 50% type III collagen, but this reduces to about 15% during post-natal growth (Epstein, 1971). However, one study has indicated that in the last few decades of life there is an increase in the proportion of type III collagen present in the dermis (Lovell et al., 1987). This effect could be due to a loss or reduced synthesis of type I collagen or an increase in type III collagen due to a change in the phenotypic expression of the fibroblasts in old tissue. Either way it almost certainly affects the functional properties of the skin. Similarly, the osteoblasts of bone or the chondroblasts of cartilage could alter their phenotypic expression in old age due to a changing environment. However, we have recently shown that in the case of human iliac bone no detectable changes in collagen type could be detected (Bailey et al., 1998). On the other hand we have recently observed changes in the collagen type in the subchondral bone of elderly osteoarthritic subjects (unpublished data). The effect of such changes in the collagen type on the functional properties of the fibre, primarily mechanical, has yet to be established.

The rate of collagen metabolism also varies substantially with maturation and senescence, although the overall turnover rate is comparatively slow. Collagen synthesis decreases steadily with maturation, and with subsequent ageing drops 10-fold in the majority of tissues. Collagen degradation, on the other hand, has been reported in rats to increase with maturation, and as a result the majority of newly synthesised collagen in old rats appears to be destined for degradation (Mays et al., 1991).

The functional properties of the tissue also depend on the primary structure of the collagen molecules themselves, and despite the relatively low rate of collagen turnover, changes can be significant in terms of decades of life as the matrix is slowly being renewed. In old age it is therefore possible that the composition of the fibres can be altered. Indeed, the extent of hydroxylation and glycosylation has been reported to decrease with maturation (Barnes et al., 1974, Royce and Barnes, 1985). Based on studies of the effect of increased lysyl hydroxylation, for example, in osteogenesis imperfecta type II, the fibre diameter is reduced with a consequent decrease in mechanical strength (Byers, 1993). These modifications reflect changes in the phenotypic expression of the fibroblasts, osteoblasts and chondroblasts of the various tissues with increasing age. At the same time modifications of the amino acid side-chains, whether pre- or post-translational could affect the cell/matrix interactions, which would further exacerbate the decline in functional properties of these collagenous tissues. However, little is known of the effects of these changes and they certainly warrant further research.

Isomerization and racemisation of aspartyl residues in proteins (Clarke, 1987) have been shown to increase with age in long lived proteins such as dentine, tooth enamel, and lens protein (Helfman and Bada, 1975, Masters et al., 1977). Recently, type I collagen has been shown to undergo β-isomerization of Asp-Gly (Fig. 3) within the C-telopeptide (Fledelius et al., 1997). This modification was demonstrated in bone tissue by direct bone analysis and indirectly in urine and the extent of isomerization was shown to increase with age. The identification of the β-Asp in urine as a monitor of turnover would be limited since if turnover increased, for example in Paget’s disease, the amount of bone collagen would be underestimated as there would be an excess of newly synthesised collagen that had not undergone isomerization. To be an effective method for ageing or turnover, a baseline level of the rate of racemisation with age under the conditions pertaining in vivo needs to be established.