Genetic Diseases of Connective Tissues: Cellular and Extracellular Effects of ECM Mutations

Home , Alport syndrome, Kniest dysplasia, Leprecan, Marshall syndrome, Pseudoachondroplasia, Stickler syndrome

REVIEWS

Genetic diseases of connective tissues: cellular and extracellular effects of ECM mutations

John F. Bateman*, Raymond P. Boot-Handford‡ and Shireen R. Lamandé* Abstract | Tissue-specific extracellular matrices (ECMs) are crucial for normal development and tissue function, and mutations in ECM genes result in a wide range of serious inherited connective tissue disorders. Mutations cause ECM dysfunction by combinations of two mechanisms. First, secretion of the mutated ECM components can be reduced by mutations affecting synthesis or by structural mutations causing cellular retention and/or degradation. Second, secretion of mutant protein can disturb crucial ECM interactions, structure and stability. Moreover, recent experiments suggest that endoplasmic reticulum (ER) stress, caused by mutant misfolded ECM proteins, contributes to the molecular pathology. Targeting ER stress might offer a new therapeutic strategy.

Osteogenesis imperfecta The major connective tissues of the body, such as might pave the way to new therapeutic opportunities. (OI). A genetic bone disorder skin, tendon, ligaments, cartilage and bone, provide Comprehensive information about gene mutations caused by abnormalities of the structural and informational framework that is has been covered in earlier reviews and is available collagen I structure or synthesis necessary for development. The extensive extracellu- in online databases, so rather than repeating infor- that results in poorly formed and fragile bones. Multiple lar matrix (ECM) of connective tissues is a complex mation we provide a summary of the mutation types distinct clinical manifestations interacting network of proteins, glycoproteins and and diseases that result from ECM gene mutations. range in severity from mild to proteoglycans providing the dynamic and essential Our main aim is to bring together a discussion of the congenital lethal. three-dimensional environment that supports the molecular pathways of pathophysiology that have maintenance, growth and differentiation of cells. In been elucidated by studies using cells from patients, addition to providing a highly organized framework, genetically manipulated cell lines or mutant mouse the ECM mediates signals to and from cells that are disease models. In particular, we highlight the realiza- involved in important biological processes such as cell tion that ECM mutations exert important pathogenic differentiation and migration during development, and effects inside the cell, as well as in the ECM outside repair processes. the cell. *Murdoch Childrens The diversity and functional importance of the ECM Research Institute and is illustrated by the occurrence of numerous genetic and ECM integrates cells into functional assemblies Department of Paediatrics, acquired connective tissue disorders. Mutations in indi- The ECM is diverse, with precisely regulated combi- University of Melbourne, vidual ECM genes cause conditions such as osteogenesis nations of molecular components providing tissue- Royal Childrens Hospital, imperfecta chondrodysplasias Ehlers– Parkville, Victoria 3052, (OI), numerous , specific properties ranging from the rock-hard nature Australia. Danlos syndrome and Marfan syndrome, and although of bone to the elasticity of ligament and skin, and from ‡Wellcome Trust Centre for these conditions are individually rare, collectively they the longevity of adult articular cartilage to the transient Cell-Matrix Research, Faculty are a considerable health burden. Furthermore, studies nature of growth plate cartilage. ECM components can of Life Sciences, University of Manchester, Oxford Road, on these conditions have been of considerable impor- be broadly placed into groups: structural components; Manchester M13 9PT, tance in understanding fundamental aspects of ECM matricellular proteins that have little structural role but United Kingdom. assembly and function. modulate cell–ECM interactions and growth factor Correspondence to J.B. Here we will review recent advances in our under- and protease activity; cell surface receptors and ancil- e-mail: standing of the molecular genetics and pathophysiol- lary proteins that interact with the ECM and perform [email protected] doi:10.1038/nrg2520 ogy of diseases caused by mutations in ECM genes, signalling and structural roles; and proteins involved Published online including insights into the important pathogenic in ECM homeostasis and remodelling, such as pro- 10 February 2009 role of endoplasmic reticulum (ER) stress, which teinases and their inhibitors. Although the different

Chondrodysplasia connective tissue matrices vary in composition and Molecular pathology of ECM gene mutations A disturbance in the detailed architecture, there are basic similarities in the Many mutations have been characterized in the struc- development of cartilage, types of molecular components and how these compo- tural components of the ECM and in the enzymes primarily affecting the long nents form interacting structural networks. The major involved in their post-translational processing and bones. More than 200 forms are recognized, presenting a ECM components include collagens, proteoglycans and folding. The molecular basis of how these mutations clinical range from mild to a large number of non-collagenous glycoproteins cause the myriad of connective tissue disorders depends severe arrested growth and and proteins1–8 (BOX 1). on the function of the gene product, its tissue distribu- dwarfism, to congenital lethal. Many ECM structural components assemble into tion and the nature of the mutation. Despite this phe- multimers in the cell. This intracellular assembly is an notypic diversity, unifying features in the molecular Ehlers–Danlos syndrome A group of inherited disorders important prerequisite for the multivalent extracellular mechanisms that lead to tissue pathology are emerging of collagen synthesis and fibril interactions that occur between the ECM components from recent studies. This Review will focus on human formation that result in a range which generate the architecturally precise matrix that is disease mutations (TABLE 1) and related mouse models of pathologies, including joint crucial for function. It can also be an important deter- (supplementary information s1 (table)). A compre- laxity and hypermobility, and skin and blood vessel fragility. minant of how ECM gene mutations cause disease, as hensive description of ECM gene-targeted knockouts we discuss below. The details of the biosynthetic path- is provided by Aszodi et al.14. Marfan syndrome way and multilevel assembly are best understood for An inherited disorder the collagen family, and for this reason much of our Loss-of-function mutations presenting with long bone discussion will focus on collagen as the prototypical The most common genetic causes of reduced synthe- overgrowth, and defects of the heart valves and aorta. It is ECM protein to illustrate common themes in ECM sis of a gene product are mutations that result in the caused by mutations in the assembly and disease mechanisms. The central fea- introduction of premature termination codons (PTCs). microfibrillar protein fibrillin 1. tures of intracellular synthesis and assembly of several The presence of a PTC triggers an mRnA surveillance collagen types and cartilage oligomeric matrix protein process and nonsense mediated decay (nMD), whereby Articular cartilage The permanent cartilage that (COMP, also known as thrombospondin 5, TsP5) are aberrant mRnAs are distinguished from normal mRnAs 15–18 forms the smooth articulating presented in FIG. 1. and are rapidly degraded (FIG. 2). There are many PTC surface of joints. It is a dense The interactions of ECM with cells integrates them mutations in ECM structural genes, and for several of connective tissue with an into functional assemblies and provides two-way con- these it has been directly shown that the PTC mutations extracellular matrix rich duits for signalling and mechanotransduction9,10. ECM result in nMD and haploinsufficiency. These include PTC in collagen II and the 19,20 proteoglycan aggrecan. components interact with cells, and thus connect to the mutations in collagen I in OI , collagen II in Stickler cytoskeleton through a range of cell surface receptors, syndrome21,22, collagen vI in Bethlem myopathy23 and predominantly integrins11. In addition to these direct collagen X in metaphyseal chondrodysplasia, Schmid type24. roles in signalling, many ECM components bind Recessive PTC mutations lead to the absence of collagen growth factors and thus act as a reservoir controlling vI in ullrich’s congenital muscular dystrophy25,26 and their bioavailablity. These important regulatory roles of collagen vII in dystrophic epidermolysis bullosa27,28. It is ECM function are beyond the scope of this article and likely that the majority of other PTC mutations in ECM are covered in several recent reviews6,12,13. genes will also be shown to cause nMD of the mRnA transcribed from the mutant allele. Although nMD can be 100% efficient29, more com- Box 1 | Extracellular matrix organization monly it reduces the abundance of PTC-containing transcripts to approximately 5–25% of the normal allele A diverse range of extracellular matrix (ECM) structural components have been transcript, so there can be small amounts of mutant described, including 28 distinct collagen subtypes, hyalectins, proteoglycans and a large number of non-collagenous proteins. Many of these matrix proteins are truncated protein produced that have the potential to modular, and are assembled from a limited set of protein domains, or modules, that exert a dominant negative or gain-of function effect. The are utilized in ECM and non-ECM proteins to provide specific functional or extent of nMD is likely to be an important contribu- structural characteristics. The major components of the ECM are members of the tor to the clinical outcome30, and this can be difficult to collagen protein family, which provide the backbone scaffolding that is essential for predict from the mutation alone. The efficiency of nMD the interaction of ECM components to provide tissue structure and integrity depends on the position of the mutation in the gene rela- (reviewed in REFS 5,7,8). The characteristic feature of collagens is that they contain tive to sequence elements that designate nMD compe- a triple helical collagen domain in which glycine occurs at every third position of tency, such as the exon–exon boundaries defined during the protein sequence. The triple helical domain can be the sole protein module in the splicing or, in the case of COL10A1 (collagen, type X, mature protein, as is the case for the fibril-forming collagens type I, II and III. These a1), the position of the PTC relative to the 3′ uTR31. fibrillar collagens have uninterrupted triple helical domains that assemble into the highly organized tensile fibrils of many tissues such as skin (collagen I and III), bone nMD competency can also show gene and tissue specifi- 29 (collagen I) and cartilage (collagen II). However, other members of the collagen city . In general, dominant heterozygous PTC mutations family have interruptions in their triple helix, and in many cases the collagen module lead to approximately half the amount of normal protein is only a small component of the mature protein (for example, collagen VI). The and have a milder clinical phenotype than structural distinct domains of collagens drive formation of different molecular structures. gain-of-function mutations in the same gene. Functional collagen fibrils are commonly heterotypic co-assemblies, such as loss-of-function mutations in the genes involved in collagens I, III and V in skin, and collagens II, IX and XI in cartilage3,107. These ECM protein processing, folding and post-translational composites provide important structural characteristics that are further modified modification can also result in connective tissue disease. 1,2 by interactions with small leucine-rich proteoglycans and other non-collagenous For example, mutations in ADAMTs2, the enzyme that components to produce the architecturally precise ECM that is crucial for the removes the collagen I n-propeptide before fibril forma- biomechanical function of the tissue. tion (FIG. 1), causes recessive Ehlers–Danlos syndrome32,

Collagen I,II,III Collagen VI Collagen X,VIII COMP

Polypeptide synthesis

Post-translational modifications Proline and lysine hydroxylation Hydroxylysine glycosylation N-linked N-linked glycosylation glycosylation

Multimer assembly Triple helix formation Pentamer Proline cis–trans isomerization formation Stabilization by HSP47

Higher order assembly Endoplasmic reticulum

N-linked oligosaccharide Golgi network modification

Intracellular

Extracellular

Procollagen processing C- and N-terminal propeptide cleavage Facilitation of Fibrillogenesis collagen fibrillogenesis Crosslinking ECM suprastructure formation

Figure 1 | supramolecular assembly pathways. Major commonalities and differences in the synthesis and assembly Growth plate cartilage pathway of three different functional classes of collagen types — fibril-forming collagens I, II and III, microfibillar A transient cartilage type Nature Reviews | Genetics that drives bone growth and collagen VI and network-forming collagens X and VIII — and cartilage oligomeric matrix protein (COMP). Collagen is located at one or both ends synthesis, chain assembly and formation of the triple helical domain is similar for most collagen types (reviewed in of long bones between the REFS 8,40,41). The collagen precursor chains are co-translationally translocated into the endoplasmic reticulum (ER) epiphysis and the diaphysis. lumen, where specific post-translational modifications occur. Three collagen α-chains associate specifically via their The chondrocytes of the 42 C-terminal domains to form heterotrimers or homotrimers.. The helical collagens are trafficked via the Golgi network to growth plate undergo the plasma membrane, and secreted into the extracellular space41. With collagen VI the individual collagen helices are specific maturation steps not secreted as monomers but assemble intracellularly into antiparallel overlapping dimers (two triple-helical collagen leading to hypertrophy and VI molecules), which then align to form tetramers (four triple-helical collagen VI molecules). The fibril-forming collagens replacement with bone during are secreted as precursor forms, called procollagens, with N- and C-terminal non-collagenous domains. These domains endochondral bone formation before puberty. are removed by the action of specific proteases, and the collagens are assembled into dense fibrils with a characteristic D-periodicity. The fibril is stabilized by covalent lysine- and hydroxylysine-derived crosslinks41. With collagen X and VIII Haploinsufficiency there is no evidence that the N- and C- terminal non-collagenous domains are processed, and they are thought to play a A condition in a diploid part in the formation of a tetrahedron of four homotrimers. It has been proposed that these tetrahedrons could then organism in which a single form hexagonal lattices by secondary interactions involving terminal and helical sequences24. Collagen VI is secreted as functional copy of a gene tetrameric structures of four collagen VI molecules that aggregate end-to-end to form long thin periodically beaded results in a phenotype, microfibrils. COMP monomers associate via N-terminal recognition sequences into homopentamers, which, after such as a disease. secretion, can interact with and facilitate collagen I and II fibril formation. COMP pentamers interact with numerous other extracellular matrix (ECM) components, including collagen IX, matrilins and aggrecan. Stickler syndrome A mild inherited chondrodysplasia with early degenerative joint and vertebral and mutations in arylsulphatase E33 and diastrophic abnormalities in collagen I helix formation, resulting in changes and often retinal dysplasia sulphate transporter (DTDST, also known as OI37,38. LEPRE1 encodes prolyl-3-hydroxylase 1 (P3H1, detachment and blindness. SLC26A2)34 affect sulphation of glycosaminoglycans and also known as leprecan) and forms a molecular com- Bethlem myopathy cause chondrodysplasias. Mutations of lysyl hydroxylase 2 plex in the ER with CRTAP and cyclophilin B (CYPB; A genetic disease associated (PLOD2), a collagen post-translational processing also known as peptidylprolyl isomerase B, PPIB)39, and with muscle weakness. This enzyme, cause Bruck syndrome35, and PLOD3 mutations this complex acts as a molecular chaperone for efficient congenital form of muscular have been identified in a patient with complex features helix formation (FIG. 1). It is likely that this impairment dystrophy caused by collagen 36 VI mutations is less severe than that overlap several collagen disorders . Homozygosity of collagen helix formation has gain-of-function con- the allelic disorder, Ullrich or compound heterozygosity for mutations in carti- sequences similar to those discussed below for protein congenital muscular dystrophy. lage associated protein (CRTAP) and LEPRE1 cause structural mutations.

Protein folding and assembly mutations deleterious effects on the ECM are thought to result from Extensive studies of dominant structural mutations in reduced protein levels owing to intracellular degradation ECM components have led to the currently accepted of the mutant polypeptide and/or secretion of mutant model that the tissue pathology results from the matrix protein that disrupts the organization of the ECM effects exerted by the mutant protein on the ECM. The (FIG. 2). These effects will be discussed first, followed by

Table 1 | Examples of mutations in ECM structural proteins causing human disease ecM component Gene(s) Principal tissue(s) affected Principal disease(s) inheritance Aggrecan ACAN Cartilage Spondyloepiphyseal dysplasia, Kimberley type AD COMP COMP Cartilage, ligaments Multiple epiphyseal dysplasia, pseudoachondroplasia AD Collagen I COL1A1, COL1A2 Bone Osteogenesis imperfecta AD COL1A1, COL1A2 Skin, joints Ehlers–Danlos syndrome, type VII AD COL1A2 Skin, joints, heart Ehlers–Danlos syndrome, cardiac valvular form AR Collagen II COL2A1 Cartilage, eyes Spondyloepiphyseal dysplasia, spondy- AD loepimetaphyseal dysplasia, achondrogenesis, hypochondrogenesis, Kniest dysplasia, Stickler syndrome Collagen III COL3A1 Blood vessels Ehlers–Danlos syndrome, type IV AD Collagen IV COL4A1 Kidney, skin, basement membranes Familial porencephaly, hereditary angiopathy AD COL4A3, COL4A4 Kidney, skin, basement membranes Alport syndrome, benign familial haematuria AR, AD COL4A5, COL4A6 Kidney, skin, basement membranes Alport syndrome, leiomyomatosis X Collagen V COL5A1, COL5A2 Skin, joints Ehlers–Danlos syndrome, type I, II AD Collagen VI COL6A1, COL6A2, Muscle Bethlem myopathy, Ullrich congenital muscular AD, AR COL6A3 dystrophy Collagen VII COL7A1 Skin, dermal–epidermal junction Dystrophic epidermolysis bullosa AD, AR Collagen VIII COL8A2 Cornea Fuchs corneal dystrophy AD Collagen IX COL9A1, COL9A2, Cartilage Multiple epiphyseal dysplasia AD COL9A3 COL9A1 Cartilage Autosomal recessive Stickler syndrome AR Collagen X COL10A1 Cartilage, growth plate Metaphyseal chondrodysplasia, Schmid type AD Collagen XI COL11A1, COL11A2 Cartilage, eyes Stickler syndrome, Marshall syndrome AD COL11A2 Cartilage, ears Otospondylomegaepiphyseal dysplasia AD, AR COL11A2 Ears Deafness AD, AR Decorin DCN Cornea Congenital stromal corneal dystrophy AD Elastin ELN Arteries, skin Supravalvular aortic stenosis, cutis laxa AD Fibrillin 1 FBN1 Skeleton, eyes, cardiovascular Marfan syndrome, ectopia lentis, Shprintzen– AD Goldberg syndrome, Weill–Marchesani syndrome Fibrillin 2 FBN2 Skeleton Contractural arachnodactyly AD Fibronectin FN1 Kidney Glomerulopathy AD Fibulin 4 FBLN4 Skin Cutis laxa AR Fibulin 5 FBLN5 Eyes Age-related macular degeneration AD FBLN5 Skin Cutis laxa AD, AR Laminin LAMA2 Muscle Congenital muscular dystrophy AR LAMA3, LAMB3, Skin, dermal–epidermal junction Epidermolysis bullosa, junctional AR LAMC2 LAMB2 Kidney, eyes Pierson syndrome AR Matrilin 3 MATN3 Cartilage Multiple epiphyseal dysplasia AD Perlecan HSPG2 Cartilage, basement membranes Schwartz–Jampel syndrome, dysegmental dysplasia AR Silverman–Handmaker type Tenascin XB TNXB Skin Ehlers–Danlos-like syndrome AR TNXB Skin Ehlers–Danlos syndrome, type III AD AD, autosomal dominant; AR, autosomal recessive; COMP, cartilage oligomeric matrix protein (also known as thrombospondin 5); ECM, extracellular matrix; X, X-linked.

Metaphyseal information that implicates the cellular consequences formation of the collagen triple helix. The structural chondrodysplasia, of disturbances to protein folding as a major gain-of- basis of this vital self-organizing step has been recently Schmid type function component of the molecular pathology of these reviewed42. Folding of the C-terminal trimerization A form of chondrodysplasia ECM structural mutations. domains probably involves interactions with ER-resident that is caused by mutations in collagen X, a component of Collagen mutations have been well studied and serve molecular chaperones such as immunoglobulin-heavy- growth plate cartilage. Growth to illustrate the effect of structural mutations, such as chain-binding protein (BiP; also known as heat shock plates are structurally altered missense mutations and in-frame deletions, on protein 70 kDa protein 5, HsPA5)43,44. and the chondrocytes are assembly at many different levels during synthesis and not surprisingly, mutations of the C-terminal propep- disorganized, causing mild secretion. There are several excellent reviews containing tide of many collagens interfere with folding of the clinical abnormities of bone growth, such as bowed legs detailed information on collagen synthesis and assem- domains, and prevent or severely impede trimer assem- 8,40,41 and hip problems. bly . Briefly, collagen trimers are assembled in the bly. Heterozygous dominant mutations in the collagen I ER via interactions of trimerization domains that direct proα1(I) C-propeptide in patients with OI are the cause of selection of the appropriate partner chains to form one abnormal procollagen trimerization, resulting in delayed of the 28 heterotrimeric or homotrimeric forms (FIG. 1). triple helix folding and reduced secretion44. This compro- For most collagens the trimerization domains are at the mised assembly leads to intracellular degradation of the C terminus of the protein. This initial association provides mutant misfolded proα1(I) chains via the ER-associated the correct chain registration required for subsequent proteasomal pathway45, and results in a major collagen I deficiency in bone. In the OI mouse model, Oim, a C propeptide mutation in the collagen I proα2(I) chain also prevents association of the mutant proα2(I) with Premature stop codon or null allele Loss of function: the proα1(I) chain. In Oim/Oim homozygotes this Haploinsufficiency reduced secretion of results in production of collagen I that contains only wild-type gene product proα1(I) trimers rather than the functional collagen I heterotrimers46. A similar mutation in the C-propeptide Normal folding of collagen II also prevents assembly in the Dmm/Dmm Structural mutation mouse, which is dwarfed and has major cartilage defects47. Mutations in COL10A1 that cause metaphyseal chon- Protein folding drodysplasia, schmid type, cluster in the C-terminal defect trimerization domain24 — further highlighting the • Upregulation of UPR- importance of the initial chain association step. The responsive genes Folded 48 • Reduced protein UPR abnormal crystal structure of the trimerization domain predicts translation protein that mutations that disrupt the hydrophobic core of the tion domain are likely to prohibit correct folding, resulting in Reduced secretion of exclusion of affected collagen X chains from trimers. The Adapta Misfolded proteins degraded mutated gene product structure also suggested that some types of trimerization Endoplasmic (ERAD and autophagy) • Altered interactions domain mutation could permit trimer formation but reticulum • Reduced stability • Altered functionality perturb subsequent collagen X supramolecular network assembly (FIG. 1) or interactions in the cartilage matrix. Altered gene expression • Reduced secretion of ECM proteins However, studies on mutant collagen X in transfected • Disruption of thology cells and in transgenic mice have shown that both classes Apoptosis

Pa development, growth and tissue maintenance of missense mutations cause misfolding and severely compromise trimer assembly49–52. Cellular consequence Extracellular consequence The next stage of collagen assembly, formation of the Figure 2 | the extracellular matrix (ecM) disease paradigm. The existing model for triple helix, is also crucial to collagen function (reviewed Nature Reviews | Genetics ECM mutation pathophysiology proposes that the extracellular consequences in REFS 8,40,41). In most collagen types, helix formation account for the molecular pathology. Reduced synthesis owing to regulatory along the repetitive Gly–X–Y collagen domain sequence mutations or decay of mRNA-containing premature termination mutations results in progresses from the C terminus, where the chains are deficiency of the protein in the ECM, thereby compromising function (green box). held in register by trimerized C-terminal domains, However, structural misfolding mutations have a dominant negative effect, leading to towards the n terminus (FIG. 1). Helix formation and sta- partial or complete cellular retention and/or degradation of mutant proteins, and bilization involves cis–trans isomerization of prolyl pep- normal proteins being assembled into mutant-containing multimers. This results in a tide bonds by peptidyl-prolyl cis–trans isomerase and severe protein deficiency and, if the mutant abnormally folded protein is secreted, collaboration of the ER-resident foldase protein disul- a further deleterious effect on ECM stability or function (yellow boxes). The new paradigm for understanding ECM mutations also considers cellular consequences phide isomerase (PDI), prolyl-4-hydroxylase (P4H), the 39 that might result from endoplasmic reticulum (ER) stress, such as the unfolded protein CRTAP–CYPB–P3H1 complex and 47 kDa heat shock response (UPR), which is induced by retention of misfolded proteins in the ER (blue protein (HsP47, also known as serpin 1)53. Prolines in boxes). The UPR is initially an adaptive response but, if unresolved, can lead to the Y position of collagen Gly–X–Y triplet sequences are changes in gene expression that result in disruption of cellular gene expression hydroxylated by P4H. This step is crucial, as hydroxy- patterns, and eventually apoptosis and pathology. The relative contribution of the proline provides the hydrogen bonding force necessary extracellular and cellular consequences to the molecular pathology is likely to show to stabilize the collagen helix. In addition to prolines, considerable mutation and gene specificity. ERAD, ER-associated degradation. some lysine residues are also hydroxylated by lysyl

hydroxylases and some of these are further modified by For secretion from the cell, the collagen triple helix addition of galactose or galactosyl-glucose. Formation of must be correctly folded. Collagen molecules contain- the triple helix prevents further post-translational modi- ing mutant chains are secreted poorly and are largely fication. After helix formation, HsP47 stabilizes the helix retained within the ER. Collagen chains containing and prevents aggregation of the collagen in the ER, and mutations that affect initial chain association, such as is important for efficient secretion, processing and fibril those in the proα1(I) C-propeptide, are removed by ret- formation53 (FIG. 1). rotranslocation of monomeric unfolded mutant collagen Mutations that interfere with the triple helix are by chains into the cytosol followed by proteasomal degra- far the most prevalent group of collagen mutations. The dation (ER-associated degradation, ERAD)44,45. However, most common of these are glycine substitutions, which there is no evidence that ERAD degrades molecules interrupt the obligatory Gly–X–Y repeat sequence, containing a triple-helical glycine substitution56. Indeed, causing misfolding and a structurally abnormal helix. although collagens containing helix mutations have Glycine mutations in collagen I cause OI; in collagen II unstable poorly formed triple helices, they do associate they cause a range of chondrodysplasias, including at their C termini and form trimers, and this is likely to spondyloepiphyseal dysplasia, Kniest dysplasia, achon- preclude them from retrotranslocation and ERAD. drogenesis, hypochondrogenesis and stickler syn- There are no published data on the mechanisms of drome; and in collagen III they cause Ehlers–Danlos degradation of collagen trimers that contain helix muta- syndrome type Iv. Although collagen I glycine mutations. However, some clues exist from studies using cells tions will be discussed as the archetype, glycine from Hsp47-null mice. In HsP47-deficient cells collagen mutations in the triple helical domain of most collagen triple helix formation and stability is impaired, and the types will have similar destabilizing effects, but the clini- improperly folded triple helices form insoluble aggre- cal consequences will depend on the structural role of gates in the ER53. Autophagy, a degradation mechanism the helix in the particular collagen and the role of the commonly deployed to degrade protein aggregates57, is collagen type in ECM architecture. therefore a likely mechanism for degradation of colla- Glycine mutations generally cause major disruptions gen trimers containing helix mutations. Mutant chain- to helix folding, delaying helix propagation at the site of containing collagen that exits the cell can also have the mutation. This pause in helix formation exposes the important extracellular affects. If incorporated into col- Dystrophic epidermolysis unfolded portions of the chains that lie n-terminal to lagen fibrils it might have a destabilizing effect and be bullosa the mutation to additional post-translational modifi- selectively degraded58, or it might compromise the inter- A severe genetic disorder cation, resulting in increased hydroxylation and glyco- actions of collagen with other ECM ligands54, disturbing resulting in extremely fragile sylation. In OI, 682 of the 832 independent mutations ECM architecture and stability. skin and recurrent blister formation caused by reported (82%) are glycine substitutions in either the Collagen vI provides an example of mutations that 54 mutations in collagen VII. collagen I proα1(I) or proα2(I) chains . substitutions can also affect the higher levels of supramolecular assem- in the most n-terminal 20% of the helix are non-lethal, bly. Glycine substitutions and exon-skipping mutations Dominant negative whereas more C-terminal substitutions are of vari- towards the n terminus of the helix can have a severe A form of mutation that interferes with the function of able phenotype but are, in general, more severe. This dominant negative effect by interfering with intracellular its wild-type allele product. is broadly consistent with predictions from the long- formation of the larger multimers, which is necessary for held view that mutations that disturb helix formation secretion and microfibril formation59–61 (FIG. 1). Another Bruck syndrome closer to the site of propagation (the C terminus) are striking example of how structural mutations can domi- A recessive form of more disruptive. However, for substitutions in the most nantly affect ECM protein assembly is provided by COMP osteogenesis imperfecta, with joint contractures C-terminal 80% of the helix there is no apparent cor- in two skeletal dysplasias, pseudoachondroplasia and 62,63 caused by mutations in the relation between disease severity and the position of multiple epiphyseal dysplasia . structural mutations in collagen-modifying enzyme the mutation54, or disease severity and regions of dif- COMP affect protein folding and assembly, in many cases lysyl hydroxylase 2. ferent local helix stability55, suggesting a more complex causing retention of the misfolded protein within the ER. (FIG. 1) ER-associated degradation relationship between mutations and phenotype. Because COMP assembles into a pentamer , almost (ERAD). An intracellular quality By contrast, the nature of the substitution is important all the multimers will contain at least one structurally control pathway that directs in disease severity. In COL1A1, substitution by amino abnormal mutant chain, resulting in intracellular reten- retrotranslocation of normal acids with charged or branched side chains (Asp, Arg tion of COMP. Accumulation of intracellular COMP and misfolded proteins from or val) causes more disruption to the helix, and usually can also result in co-retention of interacting partners, the endoplasmic reticulum 54 (REFS 64,65) to the cytoplasm for results in lethal OI phenotypes . A significant conse- including collagen IX and matrilin 3 . proteasomal degradation. quence of impaired collagen folding is reduced secretion of trimers containing mutant collagen chains. As Misfolded ECM proteins cause ER stress Autophagy just one mutant chain in a trimer will impair helix In addition to dominant effects of mutations exerted In autophagy the cell is degraded largely from within, formation, heterozygous mutations have a dominant through reduced rates of synthesis and secretion, or dis- with little or no help from negative effect. In heterotrimers, such as collagen I turbed interactions in the ECM as discussed above, mis- phagocytes. Bulk cytoplasm [α1(I)]2α2(I), three-quarters of the collagen trimers folded mutant ECM proteins such as COMP, collagens and organelles are sequestered contain one or more abnormal chains if one COL1A1 and matrilin 3 have recently been shown to induce signif- within double-membrane- allele is mutated. In the case of homotrimers, such as icant ER stress and trigger the unfolded protein response bound vesicles. These ultimately fuse with the collagen II [α1(II)]3, seven-eights of the trimers will (uPR). ER stress and the uPR have been extensively 66–69 lysosome and their contents have abnormal helix folding if one COL2A1 allele is reviewed , and the main features are summarized in are degraded. mutated. BOX 2. The uPR pathway evolved to allow cells to adjust

Box 2 | The unfolded protein response

One of the major functions of the endoplasmic Endoplasmic reticulum (ER) is the correct folding and maturation reticulum Folded protein of proteins and glycoproteins that are destined for Misfolded secretion. This protein folding factory imposes protein stringent quality control, such that only correctly modified functional proteins leave the ER108. Incorrectly folded proteins are bound by immunoglobulin-heavy- BiP chain-binding protein (BiP; also known as heat shock 70 kDa protein 5, HSPA5). This removes BiP from the ER luminal domains of the three major transmembrane IRE1 stress sensors, IRE1, PERK and ATF6, which activates ATF6p90 PERK the sensors and initiates the unfolded protein response (UPR). IRE1 activation is transmitted to the cytosolic domain Selective of IRE1, which contains a serine/threonine kinase and proteolysis site-specific RNase activities. The RNase cleaves X-box binding protein 1 (XBP1) mRNA; this splicing leads to eIF2α ATF6p50 phosphorylation the production of an active transcription factor, XBP1s. This transcription factor acts by binding to the Upregulation UPR-responsive elements in the promoters of a subset of XBP1 mRNA Selective translation of genes to stimulate the synthesis of chaperones as a mechanism to cope with the unfolded protein load. XBP1 mRNA A second arm of the UPR is mediated by the splicing Translational ATF4 attenuation transcription factor ATF6, which contains a basic leucine zipper domain. The release of BiP from the luminal domain allows uncleaved ATF6 (ATF6p90) to transit to the Golgi network, where it is cleaved to generate an Chaperones Mutant degradation active cytosolic fragment (ATF6p50). ATF6p50 migrates Apoptosis to the nucleus and binds to the promoters of genes • ERAD • Autophagy containing an ER stress-responsive element. Release of BiP from the luminal domain of PERK induces dimerization of PERK, which can then Protective Deleterious phosphorylate the translational initiation factor eIF2α. Duration of UPR This prevents the formation of the translational initiation complex, and thus downregulates general translation and reduces the protein foldingNa load.ture Re Althoughviews | Genetics phosphorylated eIF2α inhibits translation of most mRNAs, it also promotes translation of a subset of stress response genes, including activating transcription factor 4 (ATF4). ATF4 subsequently upregulates the transcription of numerous genes that are involved with amino acid metabolism and transport, oxidation–reduction reactions, and ER stress-induced apoptosis genes such as CHOP68. In addition to translational attenuation, misfolded proteins can be removed by ER-associated degradation (ERAD) and/or autophagy57,109,110, which are complementary processes to reduce the unfolded protein load and promote cell survival. In the ERAD process, the misfolded or unfolded proteins are retrotranslocated from the ER to the cytoplasm, where they are ubiquitylated and degraded by proteasomes. Autophagy is a collection of pathways that result in sections of the cytoplasm, including organelles, becoming sequestered into membrane-bound compartments that then fuse with lysosomes, where their contents are degraded by acid hydrolases. The initial activation of the UPR is cytoprotective, offering the cells an opportunity to return to protein folding homeostasis. But with increased duration of unfolded protein load the balance in activities of the three pathways changes, and prolonged ER stress results in deleterious effects such as apoptosis.

the folding capacity of the ER to differing protein folding components, it also induces ER stress. Evidence of ER loads. The uPR is deployed as a cytoprotective strategy stress comes from analyses of patient cells and cells trans- to restore protein folding homeostasis when misfolded fected with mutant COMP, which show co-localization proteins are present in the ER, but it can also contribute of molecular chaperones, such as calnexin, HsP47 to the pathophysiology of many heritable ECM disor- (REF. 74), PDI74,75, calreticulin75,76 and BiP75,76, with the ders (FIG. 2). Although this is a recent concept for ECM mutant COMP in the ER. Furthermore, phosphorylation disorders, it is unsurprising in light of the many studies of the eukaryotic translation initiation factor eIF2α, an that implicate elevated ER stress and its consequences as ER stress marker, was increased in COs cells expressing significant contributors to the pathology of an increasing mutant COMP77. In a knockin mouse model contain- number of human disorders70–73. ing a mild pseudoachondroplasia Comp mutation, no Proteasome In pseudoachondroplasia and multiple epiphy- intracellular accumulation was noted, although a uPR A large cytoplasmic protein seal dysplasia, COMP structural mutations that cause ensued, characterized by upregulation of the chaper- complex that degrades proteins to which ubiquitin has misfolding result in a characteristic distension of the ones BiP and calreticulin and activation of eIF2α and 62,63 78 been added by a process that ER and retention of mutant COMP. Although this activating transcription factor 6 (ATF6) . These data requires ATP. disrupts the normal secretion of crucial cartilage ECM suggest that although the extent of the trafficking defect

might be mutation specific, protein misfolding resulting mutations might exist and require further study. Recent in uPR activation seems to be a common feature with studies on cells transfected with a form of collagen II that COMP structural mutations. These studies also demon- contains an arginine to cysteine mutation towards the C strate reduced chondrocyte proliferation and increased terminus of the helix86 demonstrated a uPR character- and spatially dysregulated apoptosis78. Apoptosis, the ized by upregulation and binding of BiP to the mutant downstream consequence of unresolved ER stress, has unfolded protein and the expression of apoptosis mark- also been observed in other in vivo79,80 and in vitro77,81 ers. More n-terminal arginine to cysteine mutations, or studies on COMP mutations. a helical glycine to glutamic acid substitution, did not Mutations in matrilin 3 can also cause the pseudoa- bind BiP or elicit an apoptotic response86. By contrast, chondroplasia or multiple epiphyseal dysplasia pheno- BiP and CHOP were upregulated in a mouse expressing type63, and recent studies implicate matrilin misfolding a collagen II helical glycine to cysteine mutation52. and uPR activation in a disease model produced by a Although these studies support an important role for knockin mutation of matrilin 3 (REF. 82). In this model, the uPR in cell dysfunction, the most convincing data chondrocyte proliferation was reduced, levels of the so far comes from studies on collagen X mutations in chaperones BiP and GRP94 were increased, and apop- metaphyseal chondrodysplasia, schmid type. Mutations tosis was dysregulated in the growth plates of affected in the collagen X C-terminal trimerization domain com- mice. These data strongly support the hypothesis that a promise trimer assembly. In vitro studies show that the component of the pathophysiology of COMP and mat- mutant collagen X protein misfolds, forming aberrant rilin 3 misfolding mutations is activation of the uPR. disulphide-bonded dimers50, causing upregulation of Although a direct link between the uPR and chondro- ER chaperones including BiP, splicing of X-box bind- cyte proliferation and apoptosis remains to be proven, it ing protein 1 (XBP1) mRnA, and ERAD, resulting in seems likely that the combined effects of the uPR and little or no collagen X secretion50. Activation of the uPR diminished secretion of functional COMP–collagen has been confirmed in more detail in transgenic mouse IX–matrilin 3 assemblies into the cartilage ECM account models51,52, in which mutations in the Col10a1 trimeri- for the phenotypes in these disorders. zation domain are expressed in growth plate cartilage. For fibrillar collagens I and II, evidence is also mount- Expression of the mutant protein led to upregulation of ing that misfolding mutations initiate a uPR with delete- BiP and CHOP, and to XBP1 splicing, indicative of the rious cellular consequences. In OI, COL1A1 C-terminal uPR. Importantly, although CHOP was upregulated, trimerization domain mutants bind to BiP and upregu- apoptosis did not occur and the ER-stressed chondro- late expression of both BiP and GRP94 (REFS 43,44). cytes survived. However, there were significant changes These mutant chains are targeted for degradation via to hypertrophic chondrocytes, such as cell cycle re-entry the proteasomal ERAD system45. A mouse model of OI and expression of genes from the prehypertrophic stages (Aga2) provides further support for the contribution of of cartilage development. This was proposed as an adap- the uPR to the clinical phenotype83. In this mouse model tive response to the uPR; by downregulating collagen X a collagen I C-propeptide mutation causes ER retention expression and reverting to a less mature phenotype of collagen and increases caspase-induced apoptosis and the cells survive, although at a significant cost to their levels of HsP47 and CHOP in osteoblasts, both in vitro normal function. and in vivo. These data provide support for the proposal Because collagen X secretion is reduced as a result of that changes in cell behaviour as a consequence of uPR intracellular degradation of the misfolded mutant protein, activation are an important component of the pathology it was considered possible that the major changes in gene in conditions caused by problems with collagen trimer expression could result from the effect of the reduced association in the ER. collagen ECM, rather than from a direct downstream This raises the important question of whether the regulatory consequence of the misfolded collagen X more common collagen helix mutations also trigger ER causing a uPR. However, this effect can be excluded stress and whether this contributes to the clinical pheno- because heterozygous and homozygous Col10a1-null type of OI (for collagen I mutations), chondrodysplasias mice do not have any of these changes in cellular dif- (for collagen II mutations) and other collagenopathies. ferentiation87. Recent studies with a knockin Col10a1 Chondrocyte Helix mutations allow initial chain association, but then trimerization domain mutation (R.P.B.-H., unpublished Cartilage cells that produce impair subsequent folding of the triple helix. Early stud- data) also show characteristic growth plate expansion, the structural components of cartilage. ies suggested that the glycine mutations in the colla- upregulation of BiP, GRP94 and the protein disulphide gen I helix do not bind BiP, but instead bind to another isomerase ERp72, and other hallmarks of the uPR, Osteoblast ER-resident foldase, protein disulphide isomerase, along with profound downstream alterations to gene A mesenchymal cell with the which has isomerase and chaperone activities84. studies expression patterns. capacity to differentiate into bone tissue. in an OI mouse model with an engineered Col1a1 helix glycine mutation provided a preliminary indication that Implications for therapy Mesenchymal stem cells CHOP, a key pro-apoptotic regulator85, is increased in Therapeutic strategies for ECM disorders caused by Multipotent mesenchymally bone when this mutation is present. As CHOP is upreg- structural mutations, such as in forms of OI, have focused derived stem cells that can ulated by the uPR, these results suggest that helix gly- on approaches to increase expression of the normal ECM differentiate into a variety of cell types, including cine mutations can also trigger a form of uPR. However, product, in this case collagen I, or to suppress mutant 88 osteoblasts, chondrocytes, because BiP was not upregulated in this model, protein expression . Cell therapy approaches using myocytes and adipocytes. alternative mechanisms for sensing helix misfolding mesenchymal stem cells88,89 with the ability to differentiate

into bone cells have been explored in vitro and in animal mutations in Marfan syndrome101. These mutations models of OI. Allogeneic mesenchymal stem cells admin- reduce the levels of extracellular fibrillin-rich microfi- istered in small clinical trials showed poor engraftment, brils, which normally act as a transforming growth factor-β although bone growth improvements were reported in (TGFβ) reservoir, resulting in disturbances to the normal some patients. Gene therapy approaches to inducing regulation of TGFβ signalling. In these instances, treat- increased target gene expression have been explored ment of mouse models and patients with TGFβ antago- in vitro and in mouse models, but for the treatment of nists to attenuate TGFβ signalling is providing important patients these suffer from the current limitations therapeutic benefits102,103, even though other ECM of these techniques88. As many of the structural muta- structural deficiencies and possibly unfolded protein tions exert a gain-of-function effect either by interfer- effects are not corrected. ence with the interactions, assembly and integrity of A novel therapeutic strategy for an ECM disease is the ECM, an important target in the development suggested by studies on the collagen vI knockout mouse of therapeutic approaches has involved mutant gene muscular dystrophy model. Myofibres from these mice expression knockdown. A number of approaches have have ultrastructural ER and mitochondrial defects and been explored, including antisense oligonucleotides, increased apoptosis, which are thought to be a result of ribozymes and small interfering RnAs88. Although these mitochondrial depolarization and Ca2+ deregulation104. offer potential and are being explored for efficacy, they Treatment with cyclosporine A, an inhibitor of the mito- are hampered by the need to develop mutation-specific chondrial transition pore, rescued the ultrastructural knockdown tools to achieve near complete ablation of defects and decreased apoptosis104. An open pilot trial mutant gene expression, as production of even small with cyclosporine A in five patients with collagen vI amounts of structurally abnormal protein can exert mutations also showed reduced apoptosis in muscle strong dominant negative effects. biopsies105. Although these early results are encouraging, The emerging importance of protein folding abnor- improvement in muscle function has not been demon- malities and the concomitant ER stress in the pathology strated and the molecular basis for the cyclosporine A of a range of connective tissue disorders offers the pos- effect is controversial and requires further study106. sibility of more ‘generic’ new treatment strategies. If the misfolded protein load in the ER can be reduced to levels Perspectives and future directions that can be managed by the cell, then the serious delete- The long-standing view has been that mutations cause rious outcomes of an unresolved uPR, such as apoptosis, ECM dysfunction by combinations of two mechanisms, could be ameliorated. This could be achieved by correct- both of which ultimately have an impact extracellularly ing the protein folding defect, stimulating rapid degra- on the quality and integrity of the matrix that surrounds dation or blocking translation of the mutant protein90,91. cells. The first mechanism involves a quantitative reduc- One promising approach is the use of pharmacological tion in ECM components by mutations affecting synthe- agents, such as small chemical chaperones, which can sis, or by structural mutations causing cellular retention stabilize proteins in their native conformation and res- and/or degradation. second, secretion of mutant protein cue mutant protein folding and/or trafficking defects91–94. can disturb the ECM qualitatively, compromising crucial Overexpression of the endogenous chaperone BiP can interactions, structure and stability. reduce ER stress95, and recent studies identified a small However, in this Review we have presented recent chemical that induces BiP and protects against ER stress evidence suggesting that there is another significant in neurons96. player in the molecular pathology of these disorders: Another therapeutic approach that offers possibilities ER stress. This ER stress results from the intracellular Allogeneic is stimulation of the bulk destruction of ER containing the effect of misfolded ECM proteins in the ER eliciting the In allogeneic transplants, cells, mutant protein by autophagy using rapamycin, an mTOR uPR. The relative contribution of each of the intracel- organs or tissues from any inhibitor, or other drugs that enhance autophagy97–99. lular and extracellular components to pathophysiol- human other than self or a (FIG. 2) monozygotic twin are used In another approach, a selective inhibitor of dephos- ogy will depend on the mutation and will be 100 for therapeutic purposes. phorylation of eIF2α protected cells from ER stress . context dependant. In most cases it would seem likely Manipulation of the downstream consequences of the that both gain-of-function uPR consequences and mTOR uPR, such as apoptosis, also offers therapeutic potential alterations to the ECM, either by reduced secretion, The mammalian target of in the treatment of heritable ECM disorders. However, altered interactions or composition, will contribute to rapamycin is a serine/threonine protein kinase that regulates it is important to temper our enthusiasm with the cau- the disease mechanism. However, new experiments in cell growth, proliferation, tionary realization that some of these approaches might which ER stress is triggered in hypertrophic chondro- survival, protein synthesis result in increased secretion of mutant dysfunctional cytes in vivo by expressing an exogenous misfolding and transcription. protein with the potential to exert deleterious effects protein under the control of the collagen X promoter

Transforming growth on the ECM. It is therefore vital that we gain a more are indicating that the initiation of an uPR can, by factor-β comprehensive molecular understanding of the contri- itself, lead to growth plate cartilage pathology simi- (TGFβ). A secreted protein that bution of the extracellular and intracellular components lar to that seen with collagen X misfolding mutations controls cellular proliferation, to inherited ECM disease pathology. (R.P.B.-H., unpublished data). These findings make differentiation and other In this Review we have concentrated on how struc- it clear that in disorders involving ECM protein mis- functions in most cells. It has a role in immunity, cancer, tural mutations and protein misfolding cause ECM folding, the relative contribution of the cellular effects heart disease and Marfan disorders, but it is important to recognize other disease of the uPR and its downstream consequences, such syndrome. mechanisms, such as those exemplified by fibrillin 1 as apoptosis and altered gene expression, and the

extracellular dominant negative disturbance of the ECM to what extent are the mutant proteins degraded, and by on pathophysiology must be thoroughly assessed. which pathways? Developing this level of understanding There are numerous important questions that will of the role of the uPR in ECM protein misfolding dis- need to be addressed before we can fully understand the orders will require additional mouse genetic models in molecular pathology. which uPR-triggering pathways are which the uPR and ECM effects can be assessed in the used and which downstream signalling and gene expres- in vivo developmental context, along with in vitro studies sion pathways are activated? How are outcomes affected on transfected cells in which protein expression levels and by mutant protein levels and duration of expression? And timing can be experimentally manipulated.

1. Iozzo, R. V. The biology of the small leucine-rich 23. Lamandé, S. R. et al. Reduced collagen VI causes 39. Vranka, J. A., Sakai, L. Y. & Bachinger, H. P. proteoglycans. Functional network of interactive Bethlem myopathy: a heterozygous COL6A1 Prolyl 3-hydroxylase 1, enzyme characterization and proteins. J. Biol. Chem. 274, 18843–18846 (1999). nonsense mutation results in mRNA decay and identification of a novel family of enzymes. J. Biol. 2. Svensson, L., Oldberg, A. & Heinegard, D. Collagen functional haploinsufficiency. Hum. Mol. Genet. 7, Chem. 279, 23615–23621 (2004). binding proteins. Osteoarthr. Cart. 9 (Suppl. A), 981–989 (1998). 40. Canty, E.G. & Kadler, K. E. Procollagen trafficking, S23–S28 (2001). 24. Bateman, J. F., Wilson, R., Freddi, S., Lamande, S. R. processing and fibrillogenesis. J. Cell Sci. 118, 3. Eyre, D. R. Collagens and cartilage matrix homeostasis. & Savarirayan, R. Mutations of COL10A1 in Schmid 1341–1353 (2005). Clin. Orthop. Relat. Res. S118–S122 (2004). metaphyseal chondrodysplasia. Hum. Mutat. 25, 41. Kielty, C. M., Hopkinson, I. & Grant, M. E. in Connective 4. Alford, A. I. & Hankenson, K. D. Matricellular proteins: 525–534 (2005). Tissue and its Heritable Disorders. Molecular, Genetic, extracellular modulators of bone development, 25. Lucarini, L. et al. A homozygous COL6A2 intron and Medical Aspects. (eds Royce, P. M. & Steinmann, remodeling, and regeneration. Bone 38, 749–757 mutation causes in-frame triple-helical deletion and B.) 103–147 (Wiley-Liss, Inc., New York, 1993). (2006). nonsense-mediated mRNA decay in a patient with 42. Khoshnoodi, J., Cartailler, J. P., Alvares, K., Veis, A. & 5. Heino, J. The collagen family members as cell Ullrich congenital muscular dystrophy. Hum. Genet. Hudson, B. G. Molecular recognition in the assembly adhesion proteins. Bioessays 29, 1001–1010 (2007). 117, 460–466 (2005). of collagens: terminal noncollagenous domains are key 6. Lamoureux, F., Baud’huin, M., Duplomb, L., Heymann, 26. Peat, R. A., Baker, N. L., Jones, K. J., North, K. N. & recognition modules in the formation of triple helical D. & Redini, F. Proteoglycans: key partners in bone Lamandé, S. R. Variable penetrance of COL6A1 null protomers. J. Biol. Chem. 281, 38117–38121 (2006). cell biology. Bioessays 29, 758–771 (2007). mutations: implications for prenatal diagnosis and 43. Chessler, S. D. & Byers, P. H. BiP binds type I 7. Kadler, K. E., Baldock, C., Bella, J. & Boot-Handford, genetic counselling in Ullrich congenital muscular procollagen pro alpha chains with mutations in the R. P. Collagens at a glance. J. Cell Sci. 120, dystrophy families. Neuromuscul. Disord. 17, carboxyl-terminal propeptide synthesized by cells from 1955–1958 (2007). 547–557 (2007). patients with osteogenesis imperfecta. J. Biol. Chem. 8. Myllyharju, J. & Kivirikko, K. I. Collagens, modifying 27. Christiano, A. M., Amano, S., Eichenfield, L. F., 268, 18226–18233 (1993). enzymes and their mutations in humans, flies and Burgeson, R. E. & Uitto, J. Premature termination A seminal paper that identified binding of the ER worms. Trends Genet. 20, 33–43 (2004). codon mutations in the type VII collagen gene in chaperone BiP to mutant procollagen, providing Succinct review of the collagen protein family, recessive dystrophic epidermolysis bullosa result the first molecular evidence for what subsequently including supramolecular structure, biosynthesis and in nonsense-mediated mRNA decay and absence became known as an ER stress response to mutant mutations in collagens and their modifying enzymes. of functional protein. J. Invest. Dermatol. 109, collagen I in OI. 9. Nelson, C. M. & Bissell, M. J. Of extracellular matrix, 390–394 (1997). 44. Lamandé, S. R. et al. Endoplasmic reticulum-mediated scaffolds, and signaling: tissue architecture regulates 28. Hovnanian, A. et al. Characterization of 18 new quality control of type I collagen production by cells development, homeostasis, and cancer. Annu. Rev. Cell mutations in COL7A1 in recessive dystrophic from osteogenesis imperfecta patients with mutations Dev. Biol. 22, 287–309 (2006). epidermolysis bullosa provides evidence for in the pro alpha 1 (I) chain carboxyl-terminal 10. Gieni, R. S. & Hendzel, M. J. Mechanotransduction distinct molecular mechanisms underlying defective propeptide which impair subunit assembly. J. Biol. from the ECM to the genome: are the pieces now in anchoring fibril formation. Am. J. Hum. Genet. 61, Chem. 270, 8642–8649 (1995). place? J. Cell Biochem. 104, 1964–1987 (2007). 599–610 (1997). This paper described the role of ERAD in the 11. Danen, E. H. & Sonnenberg, A. Integrins in regulation 29. Bateman, J. F., Freddi, S., Nattrass, G. & Savarirayan, R. degradation of mutant misfolded type I procollagen, of tissue development and function. J. Pathol. 201, Tissue-specific RNA surveillance? Nonsense-mediated a further early indication that ER stress might be 632–641 (2003). mRNA decay causes collagen X haploinsufficiency in involved with these collagen I mutations 12. Whitelock, J. M., Melrose, J. & Iozzo, R. V. Diverse cell Schmid metaphyseal chondrodysplasia cartilage. 45. Fitzgerald, J., Lamandé, S. R. & Bateman, J. F. signaling events modulated by perlecan. Biochemistry Hum. Mol. Genet. 12, 217–225 (2003). Proteasomal degradation of unassembled mutant 47,11174–11183 (2008). 30. Khajavi, M., Inoue, K. & Lupski, J. R. type I collagen pro-alpha1(I) chains. J. Biol. Chem. 13. ten Dijke, P. & Arthur, H. M. Extracellular control of Nonsense-mediated mRNA decay modulates 274, 27392–27398 (1999). TGFβ signalling in vascular development and disease. clinical outcome of genetic disease. Eur. J. Hum. 46. Chipman, S. D. et al. Defective pro alpha 2(I) collagen Nature Rev. Mol. Cell Biol. 8, 857–869 (2007). Genet. 14, 1074–1081 (2006). synthesis in a recessive mutation in mice: a model of 14. Aszodi, A., Legate, K. R., Nakchbandi, I. & Fassler, R. 31. Tan, J. T. et al. Competency for nonsense-mediated human osteogenesis imperfecta. Proc. Natl Acad. Sci. What mouse mutants teach us about extracellular reduction in collagen X mRNA is specified by the 3′ USA 90, 1701–1705 (1993). matrix function. Annu. Rev. Cell Dev. Biol. 22, UTR and corresponds to the position of mutations in 47. Fernandes, R. J., Seegmiller, R. E., Nelson, W. R. & 591–621 (2006). Schmid metaphyseal chondrodysplasia. Am. J. Hum. Eyre, D. R. Protein consequences of the Col2a1 Comprehensive review on the use of mutant mouse Genet. 82, 786–793 (2008). C-propeptide mutation in the chondrodysplastic Dmm models to explore the structure and function of the 32. Colige, A. et al. Novel types of mutation responsible for mouse. Matrix Biol. 22, 449–453 (2003). ECM. the dermatosparactic type of Ehlers–Danlos syndrome 48. Bogin, O. et al. Insight into Schmid metaphyseal 15. Frischmeyer, P. A. & Dietz, H. C. Nonsense-mediated (type VIIC) and common polymorphisms in the ADAMTS2 chondrodysplasia from the crystal structure of the mRNA decay in health and disease. Hum. Mol. Genet. gene. J. Invest. Dermatol. 123, 656–663 (2004). collagen X NC1 domain trimer. Structure 10, 8, 1893–1900 (1999). 33. Brunetti-Pierri, N. et al. X-linked recessive 165–173 (2002). 16. Schell, T., Kulozik, A. E. & Hentze, M. W. Integration of chondrodysplasia punctata: spectrum of arylsulfatase 49. Wilson, R., Freddi, S. & Bateman, J. F. Collagen X splicing, transport and translation to achieve mRNA E gene mutations and expanded clinical variability. chains harboring Schmid metaphyseal quality control by the nonsense-mediated decay Am. J. Med. Genet. A 117A, 164–168 (2003). chondrodysplasia NC1 domain mutations are pathway. Genome Biol. 3, REVIEWS1006 (2002). 34. Dawson, P. A. & Markovich, D. Pathogenetics of the selectively retained and degraded in stably transfected 17. Weischenfeldt, J., Lykke-Andersen, J. & Porse, B. human SLC26 transporters. Curr. Med. Chem. 12, cells. J. Biol. Chem. 277, 12516–12524 (2002). Messenger RNA surveillance: neutralizing natural 385–396 (2005). 50. Wilson, R., Freddi, S., Chan, D., Cheah, K. S. & nonsense. Curr. Biol. 15, R559–R562 (2005). 35. Ha-Vinh, R. et al. Phenotypic and molecular Bateman, J. F. Misfolding of collagen X chains 18. Isken, O. & Maquat, L. E. Quality control of eukaryotic characterization of Bruck syndrome (osteogenesis harboring Schmid metaphyseal chondrodysplasia mRNA: safeguarding cells from abnormal mRNA imperfecta with contractures of the large joints) mutations results in aberrant disulfide bond function. Genes Dev. 21, 1833–1856 (2007). caused by a recessive mutation in PLOD2. Am. J. Med. formation, intracellular retention, and activation of the Reviews the molecular mechanisms of NMD. Genet. A 131, 115–120 (2004). unfolded protein response. J. Biol. Chem. 280, 19. Willing, M. C. et al. Osteogenesis imperfecta type I: 36. Salo, A. M. et al. A connective tissue disorder caused 15544–15552 (2005). molecular heterogeneity for COL1A1 null alleles of type I by mutations of the lysyl hydroxylase 3 gene. 51. Ho, M. S. et al. COL10A1 nonsense and frame-shift collagen. Am. J. Hum. Genet. 55, 638–647 (1994). Am. J. Hum. Genet. 83, 495–503 (2008). mutations have a gain-of-function effect on the growth 20. Dalgleish, R. The human type I collagen mutation 37. Morello, R. et al. CRTAP is required for prolyl plate in human and mouse metaphyseal database. Nucleic Acids Res. 25, 181–187 (1997). 3- hydroxylation and mutations cause recessive chondrodysplasia type Schmid. Hum. Mol. Genet. 16, 21. Richards, A. J. et al. High efficiency of mutation osteogenesis imperfecta. Cell 127, 291–304 (2006). 1201–1215 (2007). detection in type 1 Stickler syndrome using a two- With reference 38 this study provides new insights 52. Tsang, K. Y. et al. Surviving endoplasmic reticulum stage approach: vitreoretinal assessment coupled with into how mutations affecting the collagen folding stress is coupled to altered chondrocyte differentiation exon sequencing for screening COL2A1. Hum. Mutat. machinery, in addition to collagen I mutations, can and function. PLoS Biol. 5, e44 (2007). 27, 696–704 (2006). cause collagen defects and OI. An important paper providing the first description 22. Snead, M. P. & Yates, J. R. Clinical and molecular 38. Baldridge, D. et al. CRTAP and LEPRE1 mutations in of alterations to gene expression and cellular genetics of Stickler syndrome. J. Med. Genet. 36, recessive osteogenesis imperfecta. Hum. Mutat. 29, differentiation as a consequence of mutant 353–359 (1999). 1435–1442 (2008). collagen-induced ER-stress in vivo.

53. Ishida, Y. et al. Type I collagen in Hsp47-null cells is 74. Vranka, J. et al. Selective intracellular retention of 95. Reddy, R. K. et al. Endoplasmic reticulum aggregated in endoplasmic reticulum and deficient in extracellular matrix proteins and chaperones chaperone protein GRP78 protects cells from N-propeptide processing and fibrillogenesis. Mol. Biol. associated with pseudoachondroplasia. Matrix Biol. apoptosis induced by topoisomerase inhibitors: Cell 17, 2346–2355 (2006). 20, 439–450 (2001). role of ATP binding site in suppression of caspase-7 54. Marini, J. C. et al. Consortium for osteogenesis 75. Hecht, J. T. et al. Calreticulin, PDI, Grp94 and BiP activation. J. Biol. Chem. 278, 20915–20924 imperfecta mutations in the helical domain of type I chaperone proteins are associated with retained (2003). collagen: regions rich in lethal mutations align with COMP in pseudoachondroplasia chondrocytes. 96. Kudo, T. et al. A molecular chaperone inducer collagen binding sites for integrins and proteoglycans. Matrix Biol. 20, 251–262 (2001). protects neurons from ER stress. Cell Death Differ. Hum. Mutat. 28, 209–221 (2007). 76. Dinser, R. et al. Pseudoachondroplasia is caused 15, 364–375 (2008). A comprehensive review of collagen I mutations in through both intra- and extracellular pathogenic 97. Williams, A. et al. Novel targets for Huntington’s OI and discussion of the possible molecular effects pathways. J. Clin. Invest. 110, 505–513 (2002). disease in an mTOR-independent autophagy pathway. of the mutations. 77. Hashimoto, Y., Tomiyama, T., Yamano, Y. & Nature Chem. Biol. 4, 295–305 (2008). 55. Makareeva, E. et al. Structural heterogeneity Mori, H. Mutation (D472Y) in the type 3 repeat Describes a novel mTOR-independent pathway of type I collagen triple helix and its role in domain of cartilage oligomeric matrix protein that regulates autophagy, and drugs that can osteogenesis imperfecta. J. Biol. Chem. 283, affects its early vesicle trafficking in endoplasmic act upon this pathway to stimulate autophagy 4787–4798 (2008). reticulum and induces apoptosis. Am. J. Pathol. 163, as a putative therapeutic strategy in some protein 56. Forlino, A., Kuznetsova, N. V., Marini, J. C. & Leikin, S. 101–110 (2003). misfolding diseases. Selective retention and degradation of molecules with 78. Pirog-Garcia, K. A. et al. Reduced cell proliferation 98. Sarkar, S. et al. Small molecules enhance autophagy a single mutant alpha1(I) chain in the Brtl IV mouse and increased apoptosis are significant pathological and reduce toxicity in Huntington’s disease models. model of OI. Matrix Biol. 26, 604–614 (2007). mechanisms in a murine model of mild Nature Chem. Biol. 3, 331–338 (2007). 57. Ding, W. X. & Yin, X. M. Sorting, recognition and pseudoachondroplasia resulting from a mutation in 99. Rubinsztein, D. C., Gestwicki, J. E., Murphy, L. O. & activation of the misfolded protein degradation the C-terminal domain of COMP. Hum. Mol. Genet. Klionsky, D. J. Potential therapeutic applications pathways through macroautophagy and the 16, 2072–2088 (2007). of autophagy. Nature Rev. Drug Discov. 6, 304–312 proteasome. Autophagy 4, 141–150 (2008). Provides direct evidence for triggering of the UPR in (2007). 58. Bateman, J. F. & Golub, S. B. Deposition and selective a mouse knockin model of a human COMP mutation. 100. Boyce, M. et al. A selective inhibitor of eIF2α degradation of structurally-abnormal type I collagen 79. Hecht, J. T. et al. Chondrocyte cell death and dephosphorylation protects cells from ER stress. in a collagen matrix produced by osteogenesis intracellular distribution of COMP and type IX Science 307, 935–939 (2005). imperfecta fibroblasts in vitro. Matrix Biol. 14, collagen in the pseudoachondroplasia growth plate. 101. Robinson, P. N. et al. The molecular genetics of 251–262 (1994). J. Orthop. Res. 22, 759–767 (2004). Marfan syndrome and related disorders. J. Med. 59. Baker, N. L. et al. Dominant collagen VI mutations are 80. Schmitz, M. et al. Transgenic mice expressing D469Δ Genet. 43, 769–787 (2006). a common cause of Ullrich congenital muscular mutated cartilage oligomeric matrix protein (COMP) 102. Brooke, B. S. et al. Angiotensin II blockade and dystrophy. Hum. Mol. Genet. 14, 279–293 (2005). show growth plate abnormalities and sternal aortic-root dilation in Marfan’s syndrome. 60. Lamandé, S. R., Shields, K. A., Kornberg, A. J., malformations. Matrix Biol. 27, 67–85 (2008). N. Engl. J. Med. 358, 2787–2795 (2008). Shield, L. K. & Bateman, J. F. Bethlem myopathy 81. Weirich, C. et al. Expression of PSACH-associated 103. Habashi, J. P. et al. Losartan, an AT1 antagonist, and engineered collagen VI triple helical deletions mutant COMP in tendon fibroblasts leads to increased prevents aortic aneurysm in a mouse model of Marfan prevent intracellular multimer assembly and protein apoptotic cell death irrespective of the secretory syndrome. Science 312, 117–121 (2006). secretion. J. Biol. Chem. 274, 21817–21822 characteristics of mutant COMP. Matrix Biol. 26, 104. Irwin, W. A. et al. Mitochondrial dysfunction (1999). 314–323 (2007). and apoptosis in myopathic mice with collagen VI 61. Pace, R. A. et al. Collagen VI glycine mutations: 82. Leighton, M. P. et al. Decreased chondrocyte deficiency. Nature Genet. 35, 367–371 (2003). Perturbed assembly and a spectrum of clinical proliferation and dysregulated apoptosis in the 105. Merlini, L. et al. Cyclosporin A corrects mitochondrial severity. Ann. Neurol. 64, 294–303 (2008). cartilage growth plate are key features of a murine dysfunction and muscle apoptosis in patients with 62. Posey, K. L., Yang, Y., Veerisetty, A. C., Sharan, S. K. & model of epiphyseal dysplasia caused by a matn3 collagen VI myopathies. Proc. Natl Acad. Sci. USA Hecht, J. T. Model systems for studying skeletal mutation. Hum. Mol. Genet. 16, 1728–1741 (2007). 105, 5225–5229 (2008). dysplasias caused by TSP-5/COMP mutations. 83. Lisse, T. S. et al. ER stress-mediated apoptosis in a 106. Hicks, D. et al. Cyclosporine A treatment for Ullrich Cell. Mol. Life Sci. 65, 687–699 (2008). new mouse model of osteogenesis imperfecta. congenital muscular dustrophy: a cellular study of 63. Briggs, M. D. & Chapman, K. L. Pseudoachondroplasia PLoS Genet. 4, e7 (2008). mitochondrial dysfunction and its rescue. Brain 16 and multiple epiphyseal dysplasia: mutation review, Describes the UPR and the downstream Nov 2008 (doi:10.1093/brain/awn289). molecular interactions, and genotype to phenotype consequences of apoptosis in an in vivo 107. Hansen, U. & Bruckner, P. Macromolecular correlations. Hum. Mutat. 19, 465–478 (2002). N-ethyl-N-nitrosourea (ENU) mutant mouse with specificity of collagen fibrillogenesis: fibrils of 64. Holden, P. et al. Cartilage oligomeric matrix protein a collagen I OI mutation. collagens I and XI contain a heterotypic alloyed interacts with type IX collagen, and disruptions to 84. Chessler, S. D. & Byers, P. H. Defective folding and core and a collagen I sheath. J. Biol. Chem. 278, these interactions identify a pathogenetic mechanism stable association with protein disulfide isomerase/prolyl 37352–37359 (2003). in a bone dysplasia family. J. Biol. Chem. 276, hydroxylase of type I procollagen with a deletion in the 108. van Anken, E. & Braakman, I. Versatility of 6046–6055 (2001). pro alpha 2(I) chain that preserves the Gly-X-Y repeat the endoplasmic reticulum protein folding Describes the dominant negative effect of COMP pattern. J. Biol. Chem. 267, 7751–7757 (1992). factory. Crit. Rev. Biochem. Mol. Biol. 40, 191–228 mutations that cause intracellular retention of its 85. Forlino, A. et al. Differential expression of both (2005). interacting partners. extracellular and intracellular proteins is involved in 109. Meusser, B., Hirsch, C., Jarosch, E. & Sommer, T. 65. Merritt, T. M., Bick, R., Poindexter, B. J., Alcorn, J. L. the lethal or nonlethal phenotypic variation of BrtlIV, ERAD: the long road to destruction. Nature Cell Biol. & Hecht, J. T. Unique matrix structure in the rough a murine model for osteogenesis imperfecta. 7, 766–772 (2005). endoplasmic reticulum cisternae of Proteomics 7, 1877–1891 (2007). 110. Yorimitsu, T. & Klionsky, D. J. Endoplasmic reticulum pseudoachondroplasia chondrocytes. Am. J. Pathol. 86. Hintze, V. et al. Cells expressing partially unfolded stress: a new pathway to induce autophagy. 170, 293–300 (2007). R789C/p.R989C type II procollagen mutant Autophagy 3, 160–162 (2007). 66. Malhotra, J. D. & Kaufman, R. J. The endoplasmic associated with spondyloepiphyseal dysplasia undergo reticulum and the unfolded protein response. apoptosis. Hum. Mutat. 29, 841–851 (2008). Acknowledgements Semin. Cell Dev. Biol. 18, 716–731 (2007). 87. Kwan, K. M. et al. Abnormal compartmentalization This work was supported by grants from National Health and 67. Bernales, S., Papa, F. R. & Walter, P. Intracellular of cartilage matrix components in mice lacking Medical Research Council of Australia. J.F.B. and S.R.L. are signaling by the unfolded protein response. Annu. Rev. collagen X: implications for function. J. Cell Biol. 136, National Health and Medical Research Council Research Cell Dev. Biol. 22, 487–508 (2006). 459–471 (1997). Fellows. We thank C. Little, P. Farlie, A. Fosang and R. Wilson 68. Rutkowski, D. T. & Kaufman, R. J. That which does not 88. Millington-Ward, S., McMahon, H. P. & Farrar, G. J. for critical reading of the manuscript. kill me makes me stronger: adapting to chronic ER Emerging therapeutic approaches for osteogenesis stress. Trends Biochem. Sci. 32, 469–476 (2007). imperfecta. Trends Mol. Med. 11, 299–305 (2005). 69. Ron, D. & Walter, P. Signal integration in the 89. Dazzi, F. & Horwood, N. J. Potential of mesenchymal DATABASES endoplasmic reticulum unfolded protein response. stem cell therapy. Curr. Opin. Oncol. 19, 650–655 UniProtKB: http://www.uniprot.org Nature Rev. Mol. Cell Biol. 8, 519–529 (2007). (2007). BiP | collagen X | COMP | GRP94 | matrilin 3 Comprehensive review of the molecular signalling 90. Rochet, J. C. Novel therapeutic strategies for the induced by protein misfolding in the ER. treatment of protein-misfolding diseases. Expert Rev. FURTHER INFORMATION 70. Zhang, K. & Kaufman, R. J. The unfolded protein Mol. Med. 9, 1–34 (2007). Bateman laboratory homepage: response: a stress signaling pathway critical for health 91. Cohen, F. E. & Kelly, J. W. Therapeutic approaches http://www.mcri.edu.au/SkeletalBiology and disease. Neurology 66, S102–S109 (2006). to protein-misfolding diseases. Nature 426, Online Medelian Inheritance in Man (OMIM): 71. Zhao, L. & Ackerman, S. L. Endoplasmic reticulum 905–909 (2003). http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM stress in health and disease. Curr. Opin. Cell Biol. 18, 92. Ulloa-Aguirre, A., Janovick, J. A., Brothers, S. P. & The Human Gene Mutation Database at the Institute of 444–452 (2006). Conn, P. M. Pharmacologic rescue of conformationally- Medical Genetics in Cardiff: 72. Marciniak, S. J. & Ron, D. Endoplasmic reticulum defective proteins: implications for the treatment of http://www.hgmd.cf.ac.uk/ac/index.php stress signaling in disease. Physiol. Rev. 86, human disease. Traffic 5, 821–837 (2004). Database of osteogenesis imperfecta and type III collagen 1133–1149 (2006). 93. Perlmutter, D. H. Chemical chaperones: a mutations: http://www.le.ac.uk/genetics/collagen 73. Lin, J. H., Walter, P. & Yen, T. S. Endoplasmic reticulum pharmacological strategy for disorders of protein folding Leiden Muscular Dystrophy pages: http://www.dmd.nl stress in disease pathogenesis. Annu. Rev. Pathol. 3, and trafficking. Pediatr. Res. 52, 832–836 (2002). 399–425 (2008). 94. Arakawa, T., Ejima, D., Kita, Y. & Tsumoto, K. SUPPLEMENTARY INFORMATION Review of how ER stress and the UPR play an Small molecule pharmacological chaperones: from See online article: S1 (table) important part in the pathogenesis of many thermodynamic stabilization to pharmaceutical drugs. All links Are Active in the online Pdf human diseases. Biochim. Biophys. Acta 1764, 1677–1687 (2006).