From collagen chemistry towards cell therapy – a personal journey

Abstract

The Fell–Muir Award requires the recipient to deliver a lecture and a review manuscript which provides a personal overview of significant scientific developments in the field of matrix biology over the period of the recipient's career. In this context, this review considers the collagen family of structural proteins and the advances in biochemical, molecular biological and genetic techniques which led to the elucidation of the structure, synthesis and function of this important group of extracellular matrix constituents. Particular attention is focussed on early research on the identification and assembly of the soluble precursors of collagen types I and II, and the identification of the precursor of basement membrane collagen type IV. In subsequent studies investigating the maintenance of the chick chondrocyte phenotype in culture, the influence of the extracellular milieu was found to influence markedly both cell morphology and collagen gene expression. These studies led to the discovery of collagen type X whose expression is restricted to hypertrophic chondrocytes at sites of endochondral ossification. Such research provided a prelude to investigations of mammalian endochondral ossification which is known to be aberrant in a variety of human chondrodysplasias and is reactivated in bone fracture repair and in osteoarthritis. The cloning of bovine and then human collagen type X genes facilitated studies in relevant human diseases and contributed to the discovery of mutations in the COL10A1 gene in families with metaphyseal chondrodysplasia type Schmid. Clustering of mutations in the C-terminal domain of the type X collagen molecule has now been widely documented and investigations of the pathogenic mechanisms in animal models are beginning to suggest the prospect of novel treatment strategies.

Introduction

The outstanding contributions made by Dame Honor Fell and Professor Helen Muir in the field of tissue biology laid the foundations on which have been built many fundamental aspects of cell biology and biomedicine, especially in relation to skeletal development and disease. Members of the British Society for Matrix Biology have expressed considerable delight that the Society should choose to recognize the work of these two outstanding scientists by establishing the Fell–Muir award. I felt able to offer personal support for this move for I had not only met and discussed aspects of my own research with each of them but my own career was significantly influenced by their contributions to UK science. It was, however, both a huge surprise and a great honour to become the first recipient of the Fell–Muir award – an honour I am pleased to accept on behalf of the many talented graduate students, post-doctoral associates and research fellows who have worked and collaborated with me over the last 40 years at the University of Manchester. During this period, I have been privileged to devote my research career to a field of research that was initially best encompassed by the term ‘connective tissue biology’ but in more recent years has become a fundamental element of Cell Biology, impacting on critical areas of cell differentiation and development, and the pathology of both acquired and heritable diseases.

When the officers of the BSMB invited me to accept this award it was on the condition that I was prepared to write ‘a review of your perceived highlights of interesting and significant science that you have experienced in the field of Matrix Biology’. When I started to think about the task I became increasingly alarmed about the magnitude and difficulty of the assignment. However, I chose my title in order to stress that this review will be a very subjective essay in a complex multifaceted field to which so many laboratories have contributed, and to which so many scientists around the world have made seminal discoveries that have moved the field from the periphery to centre stage of modern cell biology. In this context, I have chosen to reference these developments wherever possible with review articles that provide a direct link into the relevant literature.

My entry into this field of large and sticky extracellular matrix macromolecules began on my appointment in October 1966 as an Assistant Lecturer in Medical Biochemistry at the University of Manchester. I never cease to be amazed at the gamble taken by the Head of Department, Professor David Jackson, in offering me the post based on my postgraduate training in plant tissue culture in the Botany School, University of Oxford. David Jackson had returned after a 10-year period in the USA, to Manchester Medical School to establish a new Department of Medical Biochemistry. His remit was to develop new courses for medical and dental undergraduates in the rapidly developing discipline of biochemistry; and his research plans focused on the establishment of a new Collagen Research Group. The fact that I had experience in cell culture techniques, albeit with plant cells, was the key to Jackson’s interest in my recruitment – incidentally, to a position for which I had not applied. He perceived that mammalian cell and organ culture approaches offered exciting opportunities to explore key aspects of collagen synthesis and assembly into the huge insoluble fibres that give strength to our connective tissues. As a postgraduate student, I had acquired direct experience of the totipotency of isolated plant cells in culture, and I had become aware of the work of Honor Fell at the Strangeways Research Laboratory in Cambridge on the differentiation of mammalian tissue explants in organ culture (Vaughan 1987). Little did I know that 20 years later her conclusions relating to the reciprocal relationship between cells and their environment would underpin my research on cell-matrix interactions influencing chondrocyte differentiation, collagen gene expression, and long bone development.

Although it was my tissue culture experience that provided my entry ticket to the field of collagen chemistry it was not until I spent 2 years leave of absence from Manchester in Dr Darwin Prockop’s laboratory in Philadelphia (1970–1972) that I undertook biosynthetic studies with isolated cells. Nevertheless I had come to realize that new methods and advances in technology were the key to progress in biological research. As a postgraduate in Oxford I frequently attended seminars in Sir Hans Krebs’ Department of Biochemistry, and on more than one occasion Sir Hans stated that he tended to judge the competence of his students on the number of techniques they had mastered. These thoughts have always guided my approach to the training of postgraduates in my laboratory, for methodology and technology coupled with rigorous planning and careful thought provide the basis for tackling the research problems we seek to address. Indeed this theme of the influence of technological developments on the advance of research underpins this review of the remarkable progress which has characterized the field of matrix biology over the last 30–40 years.

Introduction to collagen chemistry

A dictionary definition of collagen describes it as ‘a fibrous protein of connective tissue and bones that yield gelatine on boiling’– derived from the Greek kolla for glue. Thus the word collagen, meaning glue former, came into use during the nineteenth century but the economic importance of collagen dates back to prehistory when the manufacture of leather and glue was first undertaken. Indeed when I entered this field, the existing knowledge of the biochemistry of collagen was derived largely from studies conducted by chemists in the leather industry and the glue and gelatine industry. At the same time modern views of the diseases of connective tissues began to emerge when the techniques of x-ray diffraction and of electron microscopy allowed collagen fibres to be dissected in biophysical detail and when the cryostat encouraged immunofluorescence microscopy. At this time pathologists still had a tendency to use the terms ‘connective tissue disease’ and ‘collagen disease’ synonymously, although it soon became clear that more specific definitions were required to describe the numerous distinct diseases, hereditary and acquired, that exhibited an involvement of connective tissue matrix components. (See Schubert & Hamerman 1968, for an excellent overview of the field at that time).

The recognition that collagen was probably the most abundant protein in the human body, and the major constituent of most connective tissues, made it a focus of considerable interest. Its critical function to give strength to, and to maintain the structural integrity of various tissues and organs, was recognized to be a property of the insoluble collagen fibres. At the same time there was evidence suggesting collagen fibres were essentially metabolically inert and did not turn-over, and thereby developed early concepts that collagen fibres represented rather uninteresting structural elements of the matrix, and the real challenges in biology lay with the cells therein.

The matrix itself was a difficult complex to investigate comprising as it did, fibrous elements (collagen fibres and elastic fibres) embedded within what was often referred to as the ‘ground substance’. This term is a mistranslation of that used by the early German histologists, ‘Grundsubstanz’, meaning fundamental substance. This substance is essentially the extracellular, non-fibrillar, amorphous matrix which in the 1960s was said to comprise acid mucoproteins, mucopolysaccharides, and other ill defined components, plasma proteins, ions, water, etc. Purifying collagen fibres from the other components was itself a major problem for collagen chemists. However, the advent of the technique of SDS-polyacrylamide gel electrophoresis permitted sensitive analysis of the protein components of tissue extracts, and carbohydrate analyses, including gas liquid chromatography procedures, greatly aided the monitoring of purification procedures (Grant & Jackson 1968; Grant et al. 1971); and coincidentally provided improved characterisation of the components of the ground substance. Indeed the term mucoprotein started to disappear and the concept of non-collagenous structural glycoproteins as components of connective tissues was introduced (Robert et al. 1970).

The development of ion-exchange chromatography had led to an ability to determine the amino acid composition of proteins and the Edman degradation technique had made sequence analysis a possibility. Accordingly the unusual amino acid composition of collagen with its high glycine content combined with substantial quantities of proline and hydroxyproline served to emphasize the unique nature of this protein. Such studies combined with small angle x-ray diffraction and ultra centrifugation studies had identified the triple-helical rod-like structure of the collagen fibre subunits. The nature of the subunits became a little clearer with the isolation in neutral salt at 4 °C of a soluble collagen preparation from skin which by warming to 37 °C could be reconstituted into fibres, indistinguishable in the electron microscope from those seen in intact tissues. This soluble subunit was termed tropocollagen – first formed collagen – and considered to be the precursor of collagen fibrillogenesis (Gross et al. 1954), although how a fibroblast at 37 °C in vivo could control this process and avoid the intracellular formation of the insoluble aggregates remained to be explained. Characterization of the tropocollagen molecule by a variety of techniques – CM-cellulose chromatography, ultracentrifugation, light scattering, and subsequently by gel electrophoresis – made it possible to conclude that tropocollagen comprised 3 polypeptide (α chains) wound around each other in a unique type of helical structure to form a thin rigid rod with a molecular mass of around 300,000 Da. Each tropocollagen molecule was shown to consist of two chains of one type, designated α1, and one chain of another type, designated α2. It was deduced that each α chain must contain about 1000 amino acids and the unusual nature of the helix was attributable to the unusual amino acid sequences of the α chains. Throughout most of each polypeptide chain every third amino acid was shown to be glycine and hence collagen came to be considered a polypeptide of tripeptide units with the formula (Gly-X-Y)n, where X is often proline and Y is often 4-hydroxyproline (Lowther 1963). Other studies made it possible for McKusick (1966) to tabulate the characteristics of collagen as shown in Table 1. Of particular note is the susceptibility of collagen to bacterial collagenases, thereby providing a further characteristic used in detecting and defining collagenous proteins in extracts.

Table 1.

Characteristics of collagen – a partial tabulation

Category based on method of study	Characteristic features
Histologic properties	Tinctorial characteristics:
	Acid fuchsin – bright red staining
	Periodic acid and Schiff’s reagent – faint red staining
	Silver agents – staining poor or absent
	Dilute acids and alkalis – swelling
Electron microscopic features	640 Å periodicity
Chemical features	14% hydroxyproline by weight; hydroxylysine also a unique amino acid; low content of tyrosine, methionine, and histidine; absence of cystine and tryptophane; 1% hexosamine (associated carbo-hydrate)
Shrinkage characteristics	Temperature: 60–65 °C. (shrinkage to about one third original length) certain electrolyte solutions
Behaviour toward enzymes	Attacked by pepsin
	Attacked by ‘collagenases’ of Clostridium histolyticum and Cl. Welchii
	Resistant to trypsin, chymotrypsin, papain, hyaluronidase
Isotope tracer studies of metabolic turnover rate	Relative metabolic inertia
X-ray crystallography	Characteristic pattern(s)
Immunology	Very low antigenicity of unaltered collagen, viz., use for suture material

Reproduced from McKusick (1966).

Collagen synthesis and new collagen types

The presence of 4-hydroxyproline in collagen provided a marker of collagenous molecules and the demonstration that this hydroxylated imino acid was the product of a post-translational modification of peptide-bound prolyl residues was a further distinctive feature of collagen synthesis. The discovery of the role of ferrous iron (Fe⁺⁺), molecular oxygen and vitamin C as cofactors required by collagen prolyl-4-hydroxylase added further interest in the potential significance of this hydroxylation process. It had been noted by Gustavson (1955) in comparative studies on collagens from fish found at different latitudes, and therefore living at different temperatures, that higher hydroxyproline content was associated with fish at higher temperatures. He, therefore, speculated that hydroxyproline may be required to stabilize the triple helical structure at physiological temperature – a conclusion that was eventually proven by classical studies by Berg and Prockop (1973) in analysing the thermal stability of synthetically generated hydroxylated and unhydroxylated collagen molecules.

The value of hydroxyproline as both a measure of collagen content of tissues and as a urinary product attributable to collagen degradation in the body became well established (Kivirikko 1970). However, understanding how fibroblasts, osteoblasts, chondrocytes, etc., could synthesize a collagen molecule which was capable of being assembled into an insoluble fibre many times longer than a cell and exhibiting a tensile strength as high as 100 kg/cm² (equivalent to high tensile steel) was a major challenge (Grant & Prockop 1972). Furthermore, electron micrographs of collagen fibres in a variety of tissues had revealed important differences in tissue architecture including marked variation in the sizes of the collagen microfibrils. The average diameter of the microfibrillar unit was shown to be 150–250 Å in tissues such as cartilage, about 300 Å in cornea, about 600 Å in skin, and 300–1300 Å in tendon. In contrast, the lens capsule, which on the basis of hydroxyproline content was considered to be approx 70% collagenous, appeared to have a rather amorphous structure of exceedingly fine microfibrils lacking the characteristic collagen staining patterns. Understanding these differences and the mechanisms by which tissues and their collagen fibres are assembled continue to be a subject of intense research several decades after the first evidence for procollagen, the soluble precursor of collagen fibres, was reported in several laboratories around 1970.

The insolubility of the triple-helical tropocollagen molecule under physiological conditions at 37 °C had suggested that tropocollagen might not be the true biosynthetic precursor of extracellular collagen. Not only was solubility an issue but how the cell assembled the triple helix comprising two α1 chains and one α2 chain, and achieved the post-translational synthesis of appropriate hydroxyproline and hydroxylysine residues, intrigued collagen researchers. Insight into these processes came from several sources but the development in Prockop’s laboratory at the University of Pennsylvania, of matrix-free cells prepared from 17-day embryonic chick tendons (Dehm & Prockop 1971) led to major advances in our understanding of collagen synthesis. Such embryonic tissues devote a huge proportion of their protein synthetic capacity to collagen production. By removing the tendon extracellular matrix by digestion with bacterial collagenase, tendon fibroblasts are released into suspension (approx 3 × 10⁶cells per embryo) and for 4–6 h in culture they will continue to synthesize and secrete the soluble collagen precursor, procollagen (Dehm et al. 1972). This molecule was eventually shown to possess N- and C- terminal extensions to the collagen helix, which confer solubility on the procollagen molecule and play a significant role in helix formation and fibre assembly (Harwood et al. 1977; Prockop et al. 1979).

I was fortunate to join the Prockop laboratory in September 1970, on leave from my position in Manchester, and as a joint post-doc with Dr Nicholas Kefalides we resolved to investigate the synthesis of the collagenous proteins associated with the basement membrane structure which forms the lens capsule. This project was of considerable interest in part because of the importance of basement membranes in human pathology but especially because of a major controversy surrounding the nature of basement membrane collagen – indeed whether such an entity existed (Kefalides 1970; Spiro 1970).

The lens capsule is a product of lens epithelial cells whose main role is the production of crystallins of the lens. Accordingly the incorporation of [¹⁴C]proline into proteins by matrix-free epithelial cells from the lenses of 19-day chick embryos was predominantly into non-collagenous proteins, but the assay for peptide-bound hydroxy[¹⁴C]proline facilitated the identification of collagenous proteins (Grant et al. 1972a). When the procedure was combined with SDS-agarose chromatography it became possible to identify a discrete entity which was the first biosynthetic evidence for type IV collagen – a molecule comprising collagenous polypeptides somewhat larger than the tendon procollagen (pro-α) chains (Grant et al. 1972b). It also became apparent that the synthesis and secretion of lens capsule collagen took around 60 min, in contrast to the 20 min required for tendon procollagen. This delay in secretion could be attributed in large part to the delay in intracellular helix formation, a process which seemed to be directly related to the formation of disulphide bonds between the collagen polypeptides (Grant et al. 1973).

These studies prompted analogous experiments with embryonic chick matrix-free tendon and cartilage cells which confirmed a close relationship between intracellular helix formation and interchain disulphide bond formation through cystine residues at the C-terminal extension peptides (for review see Grant & Jackson 1976). At this time it had been shown that cartilage collagen existed as a genetically distinct type II collagen comprised of three identical α1(II) chains. The synthesis and secretion of the chick type II procollagen was found to take approx 40 min – a value intermediate between the values found for the type I tendon precursor and the type IV lens capsule precursor. Subsequently evidence accumulated to suggest that procollagen secretion required the molecule to be in a native helical configuration to achieve release from the cell. In addition the relationship between hydroxyproline content and inter-chain disulphide bonding in helix formation became apparent on the subsequent discovery that collagen prolyl-4-hydroxylase (i) required a non-helical substrate, and (ii) exists as a tetramer (α₂β₂) where the β-subunits in isolation possess protein disulphide isomerase activity (Kivirikko et al. 1989).

There remained many questions about the intracellular translocation of procollagen and its route of secretion from the cell. However, the ability to isolate large numbers of matrix-free cells from embryonic chick tendons and sternal cartilage allowed us to develop subcellular-fractionation procedures which permitted the investigation of the intracellular location of key post-translational events and the secretory route of procollagen polypeptides from membrane bound ribosomes through the smooth endoplasmic reticulum to the extracellular space (Grant et al., 1975). However, the mechanism by which the cell monitored helix formation and achieved pro-α chain selection, alignment and protein folding into the native helical structure was unknown – and was to occupy the interest of researchers for many more years (Engel & Prockop 1991; Lees et al. 1997; Lamande & Bateman 1999). Indeed, collagen synthesis proved to be an ideal model system of intracellular protein folding and, such studies became highly relevant to heritable diseases e.g. osteogenesis imperfecta, achondroplasia and Stickler syndrome etc., where mutations of fibrillar collagen genes gave rise to proteins with imperfections in helix assembly (Prockop & Kivirikko 1984; Prockop 1992). The field of intracellular protein folding subsequently assumed an importance with respect to a wide range of human diseases with recognition that accumulation of misfolded or unfolded mutant proteins in the endoplasmic reticulum (ER) induces ER stress and may seriously affect the viability of cells.

Aspects of the regulation of collagen synthesis

By the mid-1970s at least five different structural genes for collagens had been described. The demonstration of some tissue specificity in the expression of these genes suggested that in order to achieve an understanding of normal growth and development in vertebrates an understanding of the control of collagen biosynthesis would be required. There was, however, little known about the regulation of collagen synthesis at the level of either translation or transcription. Accordingly, we undertook a series of studies on the isolation and characterization of procollagen mRNA and its ability to translate into authentic pro-α chains in cell-free protein synthesizing systems. We were the first to report the successful synthesis of pro-α chains in vitro using a wheat-germ cell-free system (Harwood et al. 1975). However in subsequent work the mRNA-dependent rabbit reticulocyte-lysate system proved most efficient (Cheah et al. 1979) and provided convincing evidence that in common with a number of secreted proteins, procollagen is synthesized as pre-pro-α chains consistent with the ‘signal hypothesis’ of Blobel and Dobberstein (1975).

Reports from several laboratories, including our own, had indicated that the size of procollagen mRNAs was consistent with their being monocistronic. The question, therefore, remained of how a fibroblast synthesizing type I collagen with the heteromeric composition [α1(I)]₂α2(I) coordinated the expression of the relevant mRNAs and achieved the chain alignment and triple helical assembly of the constituent chains in a 2:1 ratio. Insight into chain alignment and trimerization eventually emerged in studies which identified molecular recognition sequences in the C-terminal domains which determine the type-specific assembly of procollagen (Lees et al. 1997; Kielty & Grant 2002). Understanding the coordination of the transcription of collagen genes and the translation of respective mRNAs remained a challenge especially as there was evidence to suggest that cells in culture were capable of synthesizing more than one genetic type of collagen, e.g. types I and III by skin fibroblasts. In addition there was a developing body of literature suggesting that by growing cells on different substrata it was possible to manipulate cell phenotype in culture with a concomitant change in collagen gene expression. Indeed, the differentiation of a number of cell types including epidermal basal cells, hepatocytes, mammary epithelial cells and fibroblasts were all shown to be affected by culture on a collagen substratum.

In this context our studies started to focus on chondrocytes explanted to plastic tissue-culture dishes which initially synthesize cartilage-specific macromolecules, but lose their differentiated phenotype during subsequent culture in vitro. The common observation is that chondrocytes tend to assume a fibroblastoid morphology with a switch from the predominant synthesis of type II to type I collagen. In fact a number of culture parameters had been shown to affect chondrocyte behaviour in vitro, including cell density and position within a colony; and cell-produced matrix macromolecules including exogenous collagen, proteoglycan and fibronectin had all been observed to influence the quantity of cartilage-specific components synthesized by chondrocytes in vitro.

Against this background, a research programme was initiated which was expected to focus on cell–matrix interactions, the control of collagen gene expression, and the maintenance of the chondrocyte phenotype. There was, of course, an aspiration that such studies would have a fundamental relevance to human disease, especially osteoarthritis, where a loss of cartilage and failure to achieve appropriate repair are key features.

Chondrocyte gene expression and endochondral ossification

In seeking to stabilize collagen gene expression in cultured chondrocytes, we agreed that adopting a culture system more representative of their environment in vivo should be considered. Initial studies undertook a comparison of the growth characteristics, morphology, and collagen biosynthesis of embryonic chick sternal chondrocytes cultured within 3-dimensional type I collagen matrices and on plastic tissue-culture dishes (Gibson et al. 1982). Subsequent studies investigated chondrocyte cultures in type II collagen gels (Bates et al. 1987), and in composite collagen-proteoglycan gels ± fibronectin (Thomas & Grant 1988).

Our initial studies led to the discovery of a new low molecular weight collagen, initially called G collagen but eventually assigned as type X (ten) collagen; and also detected were two other collagenous polypeptides (designated H and J chains) which provided the first biosynthetic description of chains of type IX collagen, another cartilage-specific collagen. On the basis of their electrophoretic mobilities the G, H and J collagenous polypeptides were calculated to have M_r approx 59,000, 69,000 and 84,000 respectively – significantly smaller that the pro-α chains of the fibrillar collagen types I and II (Gibson et al. 1983).

Our interest focused on type X collagen which on pepsin treatment gave rise to a triple-helical product comprising three chains of Mr 45,000 – i.e. about half the size of the helix of collagens I and II. Further studies with explants of chick embryo cartilages demonstrated that type IX collagen was a major biosynthetic product of sternal cartilage but surprisingly there was no evidence for synthesis of type X collagen. In contrast, embryonic chick tibiotarsal and femoral cartilages were found to synthesize type X collagen in organ culture explants but expression was restricted to the zone of hypertrophic chondrocytes (Kielty et al. 1984). Large amounts of type X collagen were recoverable from hypertrophic chondrocytes in culture and the native molecule was shown by rotary shadowing to occur as a short rod-like molecule 148 mm in length with a terminal globular expression which was lost on pepsin treatment (Kielty et al. 1985). Further aggregation studies in vitro revealed that under neutral conditions at 34 °C, individual type X collagen molecules associate rapidly into multimeric clusters via their carboxyl-terminal globular domains generating a regular hexagonal lattice by lateral association of the juxtaposed triple-helical domains from adjacent multimeric clusters (Kwan et al. 1991). These structural arrays are analogous to the hexagonal lattices formed by type VIII collagen, another member of the short chain collagen family, found in corneal Descemet’s membrane (Sawada et al. 1990; Shuttleworth 1998).

Unlike type VIII collagen which has a widespread distribution in vertebrate tissues, type X collagen was shown to have a very restricted localization in the growth plate of developing long bones. Purification of chick type X collagen from our hypertophic chondrocyte cultures allowed us to prepare a rabbit anti-chick type X collagen antiserum and monoclonal antibodies against chick type X collagen. Immunohistochemistry of developing chick tibiotarsus, and the secondary ossification centre of the epiphysis revealed that type X reactivity occurs only in the zone of hypertrophic chondrocytes, and as such must play a key role in the process of endochondral ossification. It remains unclear whether type X collagen plays a role in the initiation of matrix mineralization within the growth plate. Certainly, modulation of collagen X gene expression in hypocalcaemic and normocalcaemic rickets (Kwan et al. 1989), and by agents that promote mineralization (e.g. calcium β-glycerophosphate) or inhibit it (e.g. levamisole, an inhibitor of alkaline phosphatase) implies a functional relationship between type X collagen deposition and mineralization of cartilage (Thomas et al. 1990). It remains conceivable that type X collagen provides an appropriate structural support that is permissive for calcification and allows for the remodelling phases associated with vascular invasion and long bone development.

Collagen X genes, human diseases, and animal models

Experience gained with the chick chondrocyte cultures maintained in 3D-collagen gels allowed us to undertake analogous experiments with fetal bovine epiphyseal growth plate chondrocytes which were also induced to make significant quantities of bovine type X collagen (Thomas et al. 1990; Marriott et al. 1991). Such studies provided a prelude to investigations of mammalian endochondral ossification which is known to be aberrant in a variety of human chondrodysplasias and is reactivated in bone fracture repair and in osteoarthritis. We were fortunate that the rabbit anti-chick type X collagen cross-reacted with bovine type X collagen in spite of the fact that we found bovine type X collagen α-chains to possess interchain disulphide bonds within the helical segment, whereas chick collagen X had no such interchain linkages (Ayad et al. 1987). These antibodies were to be critical in immunoscreening a cDNA expression library prepared from polyA⁺RNA isolated by oligo(dT)-cellulose affinity chromatography from extracts of cultured fetal bovine chondrocytes exhibiting maximal synthesis of type X collagen (Thomas et al. 1991a).

The complete primary structure of the bovine α1(X) collagen chain was determined by nucleotide sequencing of overlapping cDNA clones. We were able to highlight the most conserved regions of the molecule in comparison with the chick α1(X) chain and also confirmed that both the mammalian and chick collagen X genes are highly condensed, in contrast to all other vertebrate collagen genes which possess multiple exons (Chu & Prockop 1993). Then using a restriction fragment (containing 360 bp of terminal non-collagenous domain) derived from the bovine α1(X), cDNA screening of a human genomic library generated human genomic clones encoding the human collagen α1(X) chain within just two exons. In addition, it was possible by a combination of somatic cell hybrid screening and hybridization in situ to assign the human collagen X gene (COL10A1) to the distal end of the long arm of chromosome 6 at the locus 6q21-6q23 (Thomas et al. 1991b).

Our focus then turned to investigation of type X collagen gene expression in human disease. Initially we undertook in situ hybridization analysis of osteoarthritic femoral-head cartilage and demonstrated a dramatic induction of collagen X gene expression by chondrocytes in areas where there appeared to be a re-initiation of endochondral bone formation, specifically at sites of subchondral bone sclerosis and at the interface between osteophyte cartilage and bone. No collagen X mRNA or protein was detectable in normal age and sex matched femoral heads. These data provided the first definitive evidence that osteoarthritic chondrocytes are stimulated to produce type X collagen and that new bone formation in OA joints is at least in part initiated by local changes in chondrocyte metabolism (Hoyland et al. 1991).

Recognizing that studies of heritable diseases have frequently given insight into the relationship between structure and function of a given gene product, we sought to investigate whether mutations within the COL10A1 gene were responsible for any of the clinically defined chondrodysplasias. This heterogeneous group of heritable disorders is characterized by aberrant growth and shape of the developing skeleton with most forms exhibiting morphological changes in the cartilage growth plates; and many of these conditions exhibit secondary osteoarthritis. We used the PCR and the single-stranded conformational polymorphism techniques to analyse the coding and upstream promoter regions of the COL10A1 gene in a number of individuals with forms of chondrodysplasia. No mutations causing achondroplasia, pseudo achondroplasia, or thanatophoric dysplasia were identified within the COL10A1 gene (Sweetman et al. 1992); but we found amino acid substitutions of conserved residues in the C-terminal domain of the α1(X) chain in two unrelated families with metaphyseal chondrodysplasia type Schmid (Wallis et al. 1994).

The most common form of metaphyseal dysplasia is the Schmid variety, which is inherited as an autosomal dominant trait, characterized by mild to moderate short stature, genu varum, and a waddling gait in infancy, with subsequent lumbar lordosis and shortness of stature. Abnormalities become apparent in early childhood with onset of weight bearing and increased mobility. Radiographically the metaphyses are expanded and irregular, with associated coxa vara. These changes are maximal at the hip and knee joints, whereas the skeleton is otherwise virtually normal (Beighton 1988).

The mutations that we found, and the many others now reported in the literature (Bateman et al. 2005) concentrate in the C-terminal domain of the type X collagen molecule. This domain we had shown to be important in the macromolecular assembly of the collagen X hexagonal lattice (Kwan et al. 1991) and on the basis of amino acid sequence analyses we had concluded that a conserved cluster of aromatic residues in the C-terminal domain in collagens X, VIII, and other fibrillar collagens, were probably crucial to the trimerization of collagen molecules (Brass et al. 1992). This clustering of almost all of the Schmid-causing mutations (be they missense or non-sense) within the region of the gene encoding the carboxyl-NC1 domain remains something of an enigma, but clearly argues for a failure of intracellular type X collagen assembly and secretion. In some instances it is apparent that the mutant type X collagen mRNA is destroyed by nonsense mediated mRNA decay and no mutant protein is made. In seeking to explain the pathogenic mechanisms associated with such type X collagen mutations the case has been made for a haplosufficiency of the protein (Bateman et al. 2003). However, the development of mouse models of the Schmid chondrodysplasia are now enabling pathogenic mechanisms to be examined in vivo, and it is becoming increasingly clear that missense mutations in COL10A1 trigger an endoplasmic reticulum (ER) stress in hypertrophic chondrocytes that may well be an integral part of the disease mechanism (Tsang et al. 2007).

The intracellular accumulation of unfolded collagen polypeptides resulting in cells with dilated ER is well documented in many of the heritable collagen diseases (Kornak & Mundlos 2003) and is a feature of mice homozygous for a Schmid mutation (Boot-Handford, R.P., personal communication). Analogous observations of chondrocyte morphology also characterize other chondrodysplasias where mutations in COMP and matrilin-3 give rise to skeletal abnormalities (Briggs, M.D., personal communication). In this context there is a developing literature on ER stress in a spectrum of diseases associated with protein misfolding or protein aggregation (e.g. cystic fibrosis, α-1 antitrypsin deficiency) and the potential for therapies that affect protein folding is now recognized. In fact a recent report from Özcan et al. (2006) describes remarkable therapeutic success in a mouse model of type II diabetes treated with low molecular weight chaperones that enhance protein folding in the ER. The feasibility of cell therapy based on enhancing ER functional capacity and the trafficking of mutant proteins is an exciting new strategy that merits further investigation in respect of heritable skeletal disorders. Furthermore, the availability of animal models of human disease now permit the interrogation of pathogenesis with a prospect of the emergence of novel treatment strategies.

Further reflections, the Wellcome Trust, and the BSMB

In writing this very personal review of research in my laboratory over the last 40 years, I have attempted to present a coherent story which in large part reflects many aspects of the developments and advances that have illuminated the field of matrix biology. Initially, the reductionist approach of the biochemist provided a characterization of the macromolecular components of the extracellular matrix which facilitated major advances in cell and developmental biology underpinned by new technologies emerging from the disciplines of biophysics, immunology, molecular biology and genetics. Unravelling the complexity of the extracellular matrix has been a real tour de force and investigations of the dynamic relationship between cells and their matrix has brought the fields of matrix biology, cytokines and growth factors into close apposition with cell signalling and gene expression. At the same time the relevance of the extracellular matrix to a vast range of diseases in man has been recognized; and exciting advances in regenerative medicine are now emanating from the interdisciplinary collaborations forged between matrix biologists, materials scientists, engineers and clinicians (Hardingham et al. 2007).

The contributions that have arisen from my laboratory in Manchester have only been possible through my good fortune in the recruitment of talented young researchers from around the world. Their contributions are referenced in the text but in reviewing the core of my research with its focus on collagenous components of the matrix, I have had to ignore the superb work of many postgraduate students and postdocs who have contributed to important aspects of elastic fibre formation (Kielty 2006), basement membrane structure and function (Heathcote & Grant 1981), angiogenesis and calcification (Canfield et al. 2000). It is also appropriate that I mention the contribution of undergraduates in the University of Manchester – science, medical and dental students – for as an academic with an enthusiasm for teaching I have always found that not only does my research inform my teaching but my discussions with undergraduates has always informed my research.

None of my research would have been possible without the generous funding I received over many years from Research Councils, Medical Charities, and above all the Arthritis Research Campaign and the Wellcome Trust. The importance of the Wellcome Trust in supporting biomedical research in the UK is widely acknowledged, and the contribution to the Manchester group has been specifically documented (Balter 1996). At a time in the late 1980s when the Trust and other funding agencies decided to invest fellowships in outstanding young researchers, we were fortunate to recruit a number of such individuals who have subsequently established themselves as international leaders in the matrix biology field (see Kadler et al. 1996; Bishop 2000; Humphries 2000; Briggs & Chapman 2002; Boot-Handford & Tuckwell 2003; Kielty 2006; Streuli 2006). Developing a critical mass of high quality researchers with the vision and ability to develop successful programmes addressing fundamental questions in biology led to a successful bid to the Wellcome Trust and the establishment in 1994 of the Wellcome Trust Centre for Cell-Matrix Research. Here I pay tribute to the advice and guidance I received from Professor Helen Muir, former Wellcome Trustee, who had for many years been supportive of our research on cartilage biology from her perspective as Director of the Kennedy Institute of Rheumatology in London. Helen was a very distinguished scientist – the only UK matrix biologist to be made a Fellow of the Royal Society – and her contributions to the field continue to underpin important aspects of research undertaken in our Wellcome Trust Centre (Hardingham 2005).

Finally, I would wish to thank again the British Society for Matrix Biology for the honour they bestow on me today. For more than 35 years I have enjoyed a close association with the Society, initially in its previous existence as the UK Collagen Club, and the British Connective Tissue Society. I was privileged to chair the Society (1987–1992) and the organizing committee of the Federation of European Connective Tissue Societies (FECTS) meeting in Manchester in 1986. Six days of rain did little to enhance the reputation of Manchester as a vibrant European city but it was a highly successful meeting that attracted over 600 delegates and made a significant profit that put the Society on a sound financial footing. The Society has gone from strength to strength over the last two decades and it has been a very important vehicle in the development of my own career and that of so many UK matrix biologists. I therefore express my sincere thanks for the support I and my colleagues have received from the Society and I wish it continued success in the future.