|Home | About | Journals | Submit | Contact Us | Français|
Pseudoachondroplasia (PSACH) and multiple epiphyseal dysplasia (MED) are relatively common skeletal dysplasias resulting in short-limbed dwarfism, joint pain, and stiffness. PSACH and the largest proportion of autosomal dominant MED (AD-MED) results from mutations in cartilage oligomeric matrix protein (COMP); however, AD-MED is genetically heterogenous and can also result from mutations in matrilin-3 (MATN3) and type IX collagen (COL9A1, COL9A2, and COL9A3). In contrast, autosomal recessive MED (rMED) appears to result exclusively from mutations in sulphate transporter solute carrier family 26 (SLC26A2). The diagnosis of PSACH and MED can be difficult for the nonexpert due to various complications and similarities with other related diseases and often mutation analysis is requested to either confirm or exclude the diagnosis. Since 2003, the European Skeletal Dysplasia Network (ESDN) has used an on-line review system to efficiently diagnose cases referred to the network prior to mutation analysis. In this study, we present the molecular findings in 130 patients referred to ESDN, which includes the identification of novel and recurrent mutations in over 100 patients. Furthermore, this study provides the first indication of the relative contribution of each gene and confirms that they account for the majority of PSACH and MED. Hum Mutat 33:144–157, 2012. © 2011 Wiley Periodicals, Inc.
Pseudoachondroplasia (PSACH; MIM# 177170) and multiple epiphyseal dysplasia (MED; MIM# 132400) are relatively common skeletal dysplasias that can be inherited as either autosomal dominant (PSACH and AD-MED) or recessive (AR-MED; rMED) conditions [Briggs and Chapman, 2002; Superti-Furga and Unger, 2007].
PSACH usually manifests in the second year of life and is characterized by moderate to severe disproportionate short stature, ligamentous laxity, and degenerative joint disease. MED is a clinically variable disease that manifests in early-to-mid childhood with joint pain and stiffness, mild to moderate short stature, and early onset osteoarthritis [Barrie et al., 1958; Fairbank, 1947; Rimoin et al., 1994]. At least one other disorder overlaps phenotypically with MED; familial hip dysplasia (Beukes type; MIM# 142669) [Cilliers and Beighton, 1990], which is mapped to chromosome 4q35 [Roby et al., 1999] and has been grouped with AD-MED in the 2010 revision of the “International Nosology and Classification of Genetic Skeletal Disorders” [Warman et al., 2011].
PSACH is believed to result exclusively from mutations in the gene encoding cartilage oligomeric matrix protein (COMP; MIM# 600310), as does the largest proportion of AD-MED [Briggs and Chapman, 2002; Briggs et al., 1995; Hecht et al., 1995]. Presumed autosomal recessive forms of PSACH [Dennis and Renton, 1975; Wynne-Davies et al., 1986; Young and Moore, 1985] were proposed to be caused by germline mosaicism and this has been proven by COMP analysis [Ferguson et al., 1997; Hall et al., 1987]. However, a disorder resembling PSACH without a COMP mutation has been described [Spranger et al., 2005], but the genetic basis of this PSACH variant remains undetermined.
AD-MED is a much more heterogeneous disorder, both at the phenotypic and genetic levels. In addition to COMP mutations, it can also result from mutations in the genes encoding matrilin-3 (MATN3, EDM5; MIM# 607078) and type IX collagen (COL9A1, EDM6; MIM# 120210; COL9A2, EDM2; MIM# 600204; COL9A3, EDM3; MIM# 600969, respectively) [Briggs and Chapman, 2002; Chapman et al., 2001; Czarny-Ratajczak et al., 2001; Muragaki et al., 1996; Paassilta et al., 1999; Unger et al., 2008]. Furthermore, several studies have suggested that a variable proportion of AD-MED can result from mutations in other genes [Jakkula et al., 2005; Zankl et al., 2007], but the identities of these genes have not yet been determined. AR-MED (rMED) can result from homozygosity or compound heterozygosity for mutations in the gene encoding SLC26A2 (EDM4; MIM# 226900) [Hastbacka et al., 1999; Rossi and Superti-Furga, 2001; Superti-Furga et al., 1999] and is the mild end of the phenotypic spectrum that includes achondrogenesis 1B and diastrophic dysplasia [Rossi and Superi-Furga, 2001].
The extensive genetic heterogeneity of MED combined with wide-ranging clinical variability, including both intra- and interfamilial variability, and various complications such as osteochondritis dissecans and mild myopathy provide a diagnostic challenge for the nonexpert [Makitie et al., 2004; Newman et al., 2000; Unger, 2002; Unger et al., 2008; Zankl et al., 2007]. In order to better understand the molecular genetics of MED, we screened for COMP, COL9A1, COL9A2, COL9A3, MATN3, and SLC26A2 mutations in over 100 patients referred to the European Skeletal Dysplasia Network (ESDN) via the on-line case manager (www.ESDN.org). In many of these patients, a clinical diagnosis of PSACH or MED was confirmed (or suspected) by the expert panel of the ESDN prior to mutation screening. However, we also included a cohort of patients, which the expert panel felt were not classical examples of these diseases due to a variety of unusual clinical and/or radiographic features (detailed in Supp. Table S1). Indeed, in many of these cases an alternative diagnosis was suggested prior to mutation screening. However, the inclusion of these patients was important for identifying phenotypic outliers of the “classical” PSACH and MED disease spectrum and to also identify specific radiographic and/or clinical features that are generally uncharacteristic of molecularly confirmed MED.
All cases were submitted on-line via the secure case manager site (https://cm.esdn.org/). Every case was then reviewed and discussed by the ESDN panel members. Following discussion, DNA samples from the patient, and when available affected and unaffected family members, were sent for mutation screening in Manchester (for PSACH and AD-MED) or Lausanne (AR-MED). Screening of COMP (exons 8–19), MATN3 (exon 3), and the type IX collagen genes (only the exon sequence and splice donor/acceptor sites of exon 8 of COL9A1 and exon 3 of COL9A2 and COL9A3) was performed as previously described [Jackson et al., 2004; Kennedy et al., 2005a; Zankl et al., 2007]. This screening protocol reflected our current knowledge of all known locations of PSACH and AD-MED mutations in the type III repeat and C-terminal regions of COMP, the A-domain of MATN3, and the COL3 domain of type IX collagen. Screening of SLC26A2 was performed as previously described [Rossi and Superti-Furga, 2001]. All mutations were confirmed in a second PCR reaction. Primer sequences and PCR conditions for exons 1–7 of COMP, exons 3–6 of MATN3, exons 1–3 and 5–6 of MATN1, exons 2, 6+7 of MATN4, and exon 50 of COL2A1 are presented in Supp. Table S2. These exons encode important structural and/or functional domains in COMP (type II EGF-like repeats), MATN3 (EGF-like repeats), matrilin-1 (A-domains), matrilin-4 (A-domains), and type II collagen (triple-helical region).
Proof of pathogenicity was defined by one or more of the following criteria; (1) a previously published mutation with co-segregation in a family and/or absent in controls, (2) a de novo mutation or co-segregation in this study, (3) alteration of an evolutionary conserved known functional residue in either the N-type motif or C-type motif of the type III repeat region of COMP or the A-domain of MATN3, (4) biochemical evidence of a pathogenetic affect.
As part of this 7-year study (2003–2010), we screened DNA from 28 PSACH patients for mutations in COMP, 77 patients (suspected AD-MED and variants) for mutations in COMP, MATN3, and the three type IX collagen genes (COL9A1, COL9A2, and COL9A3), and 22 patients for mutations in SLC26A2 (suspected rMED).
COMP is a modular protein comprising an amino-terminal coiled-coil oligomerization domain, four type II (EGF-like) domains, seven type III (CaM-like) repeats, and a C-terminal globular domain (CTD).
We identified type III repeat region COMP mutations in 27 of the 28 patients with PSACH (>96%; Table 1; Fig. 1), which were distributed between seven exons (exons 9, 10, 11, 13, 14, 16, and 18) and comprised missense mutations (67%) or small deletions (30%) and deletions/insertions (3%). We did not identify any PSACH missense mutations in exons 8, 12, 15, 17, or 19 of COMP, which is consistent with our previous findings [Kennedy et al., 2005a] (Fig. 1), although the biological significance of this observation remains unknown. Ten of the mutations (37%) were novel while the common p.Asp473del mutation was identified in six patients (22%). The CTD mutations p.Thr529Ile, pGly719Ser, and p.Thr585Arg, which we and others have previously described [Briggs et al., 1998; Jakkula et al., 2003; Kennedy et al., 2005b], were identified in four patients thus confirming the clustering of the CTD mutations into distinct regions [Kennedy et al., 2005b]. We also screened COMP in three patients with atypical PSACH but did not identify a mutation (Table 2; Fig. 2).
We identified COMP mutations in 37 patients with MED (Table 3), which were distributed between nine exons (exons 8–14, 16, and 18) and comprised missense mutations (>86%), small in-frame deletions (~5%), duplications (~5%), insertions (<3%), and deletion/insertions (<3%). We did not identify any MED missense mutations in COMP exons 15, 17, and 19, which are again consistent with our previous findings [Kennedy et al., 2005a] (Fig. 1). Fifteen (~40%) of the mutations were novel, while the recurrent mutations p.Asp385Asn, p.Asn523Lys, and p.Arg718Pro/Trp were each identified in three patients [Ballo et al., 1997; Kennedy et al., 2005b; Mabuchi et al., 2003]. Interestingly, MED patient ESDN-00594 was found to have two potential COMP missense mutations; p.Gly501Asp in the Type III-repeat region and p.Gln756Arg in the CTD (Table 3).
MATN3 mutations were identified in 13 MED patients and comprised predominantly of missense mutations (~92%) and a novel in-frame deletion, all within exon 2 encoding the single A-domain of MATN3 (Table 3). Nine mutations affected residues forming the internal β-sheet of the A-domain (i.e., βB, βD, βE, and βF), while four mutations affected residues in one of the six external α-helices (i.e., α4, α5, or α6) [Fresquet et al., 2007]. The recurrent mutations p.Thr120Met and p.Arg121Trp [Cotterill et al., 2005; Jackson et al., 2004] were each identified in more than one patient. We also identified an in-frame deletion/insertion (c.513_530del), which is predicted to result in a p.Asp171_Glu177delinsGlu in a single family with MED. MED patient ESDN-00594, who had previously tested heterozygous for p.Gly501Asp and p.Gln756Arg in COMP, was also shown to be heterozygous for p.Val245Met in MATN3 (Tables 3 and and44).
Finally, we identified COL9A2 mutations in five MED patients and a COL9A3 mutation in a single MED patient (Table 3). The COL9A2 mutations that we identified were all in the splice donor site of exon 3 and were therefore consistent with previous findings [Fiedler et al., 2002; Holden et al., 1999; Muragaki et al., 1996]. In our cohort of patients, mutations were identified at positions +3, +5, and +7 of the splice donor consensus sequence (i.e., +1C C G g t g a g t+9) and are therefore consistent with those previously identified, with just one exception. MED patient ESDN-01003 was heterozygous for c.186+4a>c (i.e., +7 in consensus sequence), which has not been previously described. We did not directly assay whether this specific mutation would affect splicing, but the patient tested mutation negative for COMP, MATN3, COL9A1, and COL9A3, and the c.186+4a>c mutation co-segregated with the affected mother and brother of the proband.
Two patients had the relatively common c.186+2t>c change, which has previously been reported in several families of European origin [Muragaki et al., 1996]. We also identified a c.148-2a>g mutation in intron 2 of COL9A3 in ESDN-00986. Although this specific sequence change has not been previously published, a c.148-2a>t mutation has been shown to be pathogenic and to result in the skipping of exon 3 of COL9A3 due to the loss of the consensus “a” at the −2 position of a splice acceptor site [Paassilta et al., 1999].
In summary, we identified mutations in 27 patients with PSACH and 56 patients with AD-MED. The MED mutations that we identified in our patient cohort were found in the COMP (66%), MATN3 (24%), COL9A2 (8%), and COL9A3 (2%) genes. We did not identify a COL9A1 mutation in any patient sample analyzed. These data confirm recent studies showing that COMP mutations are the predominant cause of MED [Zankl et al., 2007], while type IX collagen gene mutations account for only about 10% of the currently known mutations in AD-MED.
We screened 22 patients for mutations in SLC26A2 that had a diagnosis consistent with AR-MED as determined by the ESDN expert panel. Sixteen (16/22, ~73%) of these patients were either homozygous, or compound heterozygous, for SLC26A2 mutations (Table 3). More specifically, of those 16 patients, 13 (13/15, ~86%) were homozygous for the common p.Arg279Trp AR-MED mutation [Superti-Furga et al., 1999], while one patient was compound heterozygous (p.Arg279Trp and IVS1+2T>C). Three other patients were compound heterozygous, with the common “Finnish” mutation (IVS1+2T>C) [Hastbacka et al., 1999] and p.Cys653Ser both occurring twice; the remaining two mutations being p.Ala715Val and p.Phe256Ser. The latter mutation had not been observed prior to this study; its absence in well over 200 control samples and the proximity to two other known pathogenic mutations (p.Gly255Glu and p. Gly259Val) confirm its putative pathogenicity.
We failed to identify COMP, MATN3, COL9A1, COL9A2, COL9A3, or SLC26A2 mutations in 30 patients that had originally been referred to ESDN with a working diagnosis of PSACH or MED and variants (Supp. Table S1). We therefore decided to extend the screening of some of these patients to include exons 1–7 of COMP, exons 3–6 of MATN3, exons 1–3 and 5–6 of MATN1, and exons 2 and 6+7 of MATN4 (Supp. Table S1). We specifically chose these exons because they encode important structural and/or functional domains in glycoproteins that are structural components of the cartilage growth plate. For example, the EGF-like repeats of COMP and MATN3 and the EGF-like and A-domains of matrilin-1 and -4 are important for protein integrity and interactions in cartilage [Wagener et al., 2005]. Moreover, mutations in the first EGF-like repeat of MATN3 had been reported to cause recessive spondylo-epi-metaphyseal dysplasia (SEMD, MATN3 related) [Borochowitz et al., 2004] and confer susceptibility to hand osteoarthritis [Stefansson et al., 2003].
In one patient with MED and in the one remaining patient with “classical” PSACH, we identified novel mutations in exons 5 and 7 of COMP, respectively (Table 5; Fig. 3). MED patient ESDN-00521 was heterozygous for c.500G>A, which is predicted to result in a p.Gly167Glu substitution in the second EGF-like repeat of COMP. In PSACH patient ESDN-01040, we identified a heterozygous c.700C>T, which resulted in a p.Pro234Ser substitution in the fourth EGF-like repeat of COMP; both of these unclassified variants had not previously been reported and were not present in the dbSNP database version 130 (May 2009). In contrast, no mutations were identified in the additional exons of MATN1, MATN3, and MATN4 that we screened, which is consistent with our previous studies [Jackson et al., 2004].
Finally, we extended our screening to include exon 50 of COL2A1 since a recurrent mutation (p.Gly1170Ser) in this exon has been shown to cause Legg-Calve-Perthes (LCP) disease in four families [Miyamoto et al., 2007; Su et al., 2008], and there is clear clinical overlap between LCP and MED [Herring and Hotchkiss, 1987; Ikegawa et al., 1991]. Patient ESDN-00050 (mild MED) was heterozygous for c.3535G>A, which is predicted to result in a p.Gly1179Arg substitution and patient ESDN-00283 (MED) was heterozygous for c.3527G>T, which is predicted to result in a p.Gly1176Val substitution (Table 5).
In this study, we undertook a comprehensive clinical and molecular approach to define the genetic basis of PSACH and MED in a cohort of 130 patients referred to ESDN. All of these patients had been referred to ESDN between 2003 and 2010 with various working diagnoses of PSACH (27); PSACH-MED (3); MED (66) (including variants described as Fairbank (2), polyepiphyseal dysplasia (1), MED with sacroiliitis (1), MED with neuropathy (1) and Perthes (2)); rMED (9); spondyloepiphyseal dysplasia (SED)/MED (1); SED (4); SEMD (2) (including a variant described as SEMD-JL ); SMD (1); PPRD (2); mild DTD (3), or without any formal diagnosis (4).
For the vast majority of patients referred to ESDN with a provisional diagnosis of PSACH, the panel agreed with the diagnosis and a COMP mutation was subsequently found (Table 1). This observation implies that PSACH is relatively straightforward to diagnose given suitable radiographs and clinical summary, however, it should be noted that at least 21 of the 27 PSACH referrals came from geneticists within clinical genetics departments, including eight referrals from members of the ESDN panel, suggesting that the cases came from clinicians with experience in skeletal dysplasias. The three exceptions were ESDN-00074, ESDN-00618, and ESDN-00695, but atypical clinical and radiographic features excluded PSACH and suggested an alternative diagnosis prior to sequencing the COMP gene (Table 2; Fig. 2).
In the other 56 patients in whom we identified a COMP, MATN3 or type IX collagen gene mutation, the majority had been referred to ESDN with a diagnosis of MED (Table 3), which would indicate that the “classical” forms of MED (i.e., those patients in whom we identified a mutation) are also relatively easy to diagnose. This was particularly the case for those patients in whom we identified MATN3, COL9A2, or COL9A3 mutations [i.e., 18/19 (94%) patients with these mutations had a correct diagnosis upon referral to ESDN; see Table 3]. Once again these referrals came almost exclusively from geneticists within clinical genetics and/or pediatrics departments and included nine referrals from members of the ESDN panel.
Finally, in those patients in which we identified a DTDST mutation, 69% (11/16 patients) had originally been referred with a diagnosis of MED or rMED; with PPRD (2) and mild DTD/SED/SEMD (3) being proposed as a diagnosis in five cases. This observation would therefore suggest that rMED is slightly harder to diagnose than classical AD-MED. This may be explained partly by the fact that unlike AD-MED where family history is often positive, most cases of rMED lack a family history and therefore physicians may be less inclined to think of a possible genetic cause of the disorder. Likewise, all the referrals came from geneticists within clinical genetics departments and also included three referrals from members of the ESDN panel.
In spite of the diagnostic difficulties, we can confirm that autosomal recessive MED (rMED) accounts for approximately one-fourth of total MED cases, as has been suggested by earlier studies (15, 16). This relatively high incidence is driven by the frequency of the p.R279W mutation, which is by far the most common DTDST mutation in the European population [Ballhausen et al., 2003; Barbosa et al., 2010; Rossi and Superti-Furga, 2001]. Interestingly, it has been reported that DTDST mutations are not a common cause of MED in some Asian populations [Itoh et al., 2006]. On the whole, this study suggests that ESDN receives a significant number of referrals from geneticists and/or pediatricians with an interest in, and knowledge of, skeletal dysplasias, which is reflected in the relatively high level of correct diagnosis on referral.
The range of COMP mutations that we identified in the PSACH patients was similar to those previously published [Kennedy et al., 2005a] and included missense mutations that resulted in the substitution of conserved glycine, aspartic acid, asparagine, and cysteine residues, which are important for the folding, structural integrity, and calcium binding of the type III repeats [Tan et al., 2009]. We also identified the common p.Asp473del mutation in six PSACH patients and more complex deletions in three other patients, thus in-frame deletions were identified in approximately 33% of PSACH patients (9/27), which was slightly less than the 43% that we have previously reported [Kennedy et al., 2005a]. The range of COMP mutations that we identified in the MED patients was more diverse than those found in PSACH and in addition to the substitution of conserved glycine, aspartic acid, asparagine, and cysteine residues, we also identified missense mutations that resulted in the substitution of nonconserved proline (p.Pro276Arg) and serine (p.Ser298Leu) residues and a conserved alanine (p.Ala311Asp) residue all within the linker or the T31 repeat of the Type III region (Fig. 4). We also identified a broad range of in-frame deletion, duplication, and deletion/insertion mutations including the previously reported p.Asp473dup [Delot et al., 1999]. It is an interesting observation that p.Asp473del consistently causes PSACH, while p.Asp473dup always causes MED. Presumably, the insertion of a aspartic acid reside into the C-type motif of T36 is less deleterious to protein folding and structure than its deletion.
Interestingly, we identified the recurrent p.Asn523Lys (c.1569C>G) mutation in three MED patients from our panel (ESDN-00123, 00382, and 00751), all of whom were from the Netherlands. Furthermore, this same mutation was previously identified in a large South African kindred of Dutch descent [Ballo et al., 1997], suggesting that this is an ancestral mutation. We also identified the recurrent p.Asp385Asn (c.1153G>A) mutation in two British and one Dutch family with MED (ESDN-00049, 00509, and 00597), in addition to p.Asp585Asn and p.Asp385del mutations, which points to a key role for Asp385 in the structure of COMP.
Finally, when just considering the 34 different COMP missense mutations that we identified in the type III region, sequence alignments reveal that 85% (29/34) of them affect residues in the C-type motif of the linker and T31-7 repeats (Fig. 4). This suggests that conserved residues in this motif are more important for coordinating calcium binding and/or the folding of COMP; the five mutations that we identified in the N-type motif all cause MED.
Mutations in exons 14–18 of COMP, which encodes the C-terminal domain were identified in approximately 13% (8/64) of PSACH and MED patients and were once again clustered at specific residues as previously noted [Kennedy et al., 2005b]. These observations reinforce the hypothesis that Thr529, Thr585, Arg718, and Gly719 are important for the structure and/or function of the C-terminal domain of COMP. Indeed, we have now identified every possible amino acid substitution of Thr585 [i.e., pThr585Met, p.Thr585Lys (unpublished data), and p.Thr585Arg], which all result from either a transversion or transition at nucleotide c.1754, suggesting that this nucleotide is particularly susceptible to mutation, but also confirming that methionine at residue 585 is vital for the correct folding and/or functioning of COMP.
MED mutations in MATN3 were again found in exon 2, which encodes the single A-domain of MATN3 and affected residues in either the β-stands (70%) or α-helices (30%). The identification of an in-frame deletion/insertion (c.513_530del), which is predicted to result in a p.Asp171_Glu177delinsGlu in the α4 helix, is the first mutation of this kind to be identified in MATN3 and thus extends the type of mutations in MATN3 that can cause MED. The clustering of mutations in exon 2 demonstrates the importance of the A-domain and most studies have demonstrated that these mutations disrupt the folding of this domain [Cotterill et al., 2005], which elicits and unfolded protein response [Nundlall et al., 2010].
Interestingly, in all six patients in whom we identified a COL9 mutation, the ESDN panel had predicted or suggested a type IX collagen defect prior to mutation screening. The ability of the panel to accurately predict a COL9 mutation was a result of the previously documented differences in the clinical and radiographic presentation of MED caused by a type IX collagen mutation compared to COMP and MATN3 mutations [Lachman et al., 2005; Unger et al., 2008, 2001]. COL9-MED is generally the mildest form of MED and is characterized by joint pain and stiffness presenting in the first decade of life, while radiographic abnormalities are primarily restricted to the knees with relative sparing of the hips. Interestingly, three of the six patients (ESDN-0638, 0926, and 0997) were from the Netherlands and two of these shared the same c.186+2C>T mutation in COL9A2, which was originally identified in a large Dutch kindred in 1986 [Muragaki et al., 1996; van Mourik et al., 1998] and more recently in a second large family of Dutch origin [Jackson et al., 2010; Versteylen et al., 1988]. These data might suggest that the c.186+2t>c mutation in COL9A2 is another ancestral mutation in the Dutch population; however, haplotype analysis should be performed to test this hypothesis further. In contrast, three British families with COL9A2-MED all had a different mutation; c.186+4a>c (ESDN-01003 in this study), c.186+5g>c [Holden et al., 1999], and c.186+6t>g [Barrie et al., 1958; Briggs et al., 1994; Spayde et al., 2000]. Finally, although it has been proposed that COL9-MED mutations are more prevalent in Japan [Itoh et al., 2006], it is interesting to note that of the 14 COL9A2 mutations published to date, 12 have actually been identified in families from Northern Europe (UK , Netherlands , Germany , Sweden , and unspecified ). Our study has now demonstrated that mutations in COL9A2 and COL9A3 account for approximately 10% of mutations in molecularly confirmed MED, which is slightly less than the 16% reported by Itoh and colleagues [Itoh et al., 2006]. In contrast, COMP mutations accounted for 66% of mutations (37% in [Itoh et al., 2006]) and MATN3 for 24% of mutations (47% in [Itoh et al., 2006]). These differences in the relative proportions of COMP, MATN3, and COL9 mutations may be due to ascertainment bias or ethnic differences.
The panel's success in predicting a COL9 mutation was not repeated with MED resulting from MATN3 mutations and in most cases the panel could not decide between COMP or MATN3 as the causative gene prior to screening (although the COL9 genes were never considered as candidates). These observations would suggest that there are phenotypic features (both clinical and radiographic) shared between COMP-MED and MATN3-MED, which may result from common disease mechanisms. Indeed, recent studies of knock-in mouse models of mild PSACH and MED caused by Comp and Matn3 mutations, respectively, suggest that specific characteristics of growth plate pathophysiology, such as reduced chondrocyte proliferation and increased and/or spatially dysregulated apoptosis are common disease mechanisms [Leighton et al., 2007; Pirog-Garcia et al., 2007].
Interestingly, in two unrelated patients with AD-MED, we identified more than one potential mutation that co-segregated with the phenotype following family studies (Table 4). MED patient ESDN-00359 had three changes identified in COMP; p.Gly404Gly (a nonpathogenic neutral polymorphism), p.Gly501Asp, and p.Asp502Tyr, which are both conserved amino acid residues in the N-type motif of the T37 repeat and help to co-ordinate Ca2+ binding. All three changes co-segregate as a single haplotype since they were also present in the suspected affected mother, but not the unaffected father, of the proband. In this case, it would seem most likely that a single mutational event affecting contiguous nucleotides in the codons of both Gly501 (GGC) and Asp 502 (GAC) would account for these two mutations (p.Gly501Asp and p.Asp502Tyr). Both mutations might conceivably cause MED on their own by affecting highly conserved amino acids important for protein folding and calcium binding, respectively.
ESDN-00594 had four changes; three in COMP and a single change in MATN3, none of which have previously been reported (Table 4). The COMP changes included p.Gly501Asp (also identified in ESDN-00359 and ESDN-00422); p.Gln756Arg, which is at the end of the C-terminal domain of COMP (and interestingly arginine is present at that same position in bovine and rat COMP); and finally c.2274+1G>C which is in the 3′ untranslated region and might affect mRNA stability leading to reduced protein levels, but since COMP-null mice are normal, this change is unlikely to have a phenotypic effect. The MATN3 mutation is at a conserved residue (p.Val245) in the αF strand of the MATN3 A-domain and functional studies show p.Val245Met affects to some extent the trafficking and secretion of MATN3 A-domain [unpublished data]. Importantly, the p.Gly501Asp mutation in ESDN-00422 was recently confirmed as de novo in this family [Stephen Robertson, personal communication], confirming that it is the cause of MED in ESDN-00594. Nevertheless, the intracellular retention of a significant proportion of p.Val245Met suggests the intriguing possibility that it might be a genetic modifier of phenotypic severity.
By extending our standard screening protocol, we identified mutations in COMP and COL2A1 in the remaining PSACH patient and in three MED patients. In PSACH patient ESDN-01040, we identified a heterozygous p.Pro234Ser substitution in the fourth EGF-like repeat of COMP and MED patient ESDN-00521 was heterozygous for p.Gly167Glu in the second EGF-like repeat of COMP. Both of these residues are conserved in murine COMP and the substitution of glycine and proline residues in the EGF-like repeats of fibrilin-1 has been shown to cause Marfan Syndrome [Arbustini et al., 2005; Collod-Beroud et al., 1999]. More recently, we have identified a third COMP EGF-like mutation in a patient with PSACH, p.Gly258Arg, which is in the fourth repeat and again conserved across species (unpublished data), but the precise affect of these COMP mutations remains undetermined and will require extensive studies in vitro.
Both of the COL2A1 mutations that we identified (p.Gly1176Val and p.Gly1179Arg) were in suspected MED patients (ESDN-00283 and ESDN-00050) in whom there was limited clinical information and radiographic images in which to make an unambiguous diagnosis (Table 5). However, in both cases while there were some features consistent with MED, it was also noted that there were features not normally associated with MED such as short trunk and severely fragmented hip epiphyses with adjacent metaphyseal anomalies. This observation would suggest that there is some clinical and radiographic overlap between MED and mild SED congenital (SEDc), which is borne out by the fact that similar mutations, p.Gly1173Arg [Sobetzko et al., 2000] and p.Gly1176Ser [Williams et al., 1995], have previously been shown to cause SEDc type. Furthermore, the identification of a recurrent p.Gly1170Ser mutation in patients with Legg-Calve-Perthes disease (LCPD) and/or primary avascular necrosis of femoral head (ANFH) [Miyamoto et al., 2007; Su et al., 2008] also suggests that there are similar disease mechanisms that might cause phenotypes within a LCPD/ANFH-MED-SEDc disease spectrum.
Finally, in those patients in whom we could not identify a mutation in the core exons of our screening protocol, the predominant diagnosis on referral had been MED or rMED (Supp. Table S1: 22/30 [80%]), with the remainder being PSACH (1), DTD (2), SED (3), or unknown (2). This observation would suggest that the mutation negative cases of MED are due to either a mutation in an as yet unknown gene(s), or the diagnosis of MED was incorrect in these patients. Interestingly, on re-review of these cases it was clear that the ESDN panel had not agreed upon a consensus diagnosis for most of these cases. Indeed, there were only two mutation negative patients in which a diagnosis of suspected mild MED was agreed upon by the panel prior to screening (Supp. Table S1; ESDN-00039 and ESDN-00160). In the majority of cases in which we did not identify a mutation in this study (Supp. Table S1: 24/26 [96%]), an alternative diagnosis was suggested prior to mutation screening and for many cases this was either Meyer's disease/hip dysplasia (Beukes)/bilateral LCPD (8/26), a type II collagenopathy (5/26), or SEMD (2). This would suggest that there are forms of familial hip dysplasia, variably described in the literature as Meyer's disease (dysplasia epiphysealis capitis femoris), familial hip dysplasia (Beukes), and bilateral LCPD, that are genetically distinct from the classical forms of MED and do not result from mutations in COMP, MATN3, or type IX collagen. The genetic cause of these diseases remains underdetermined, but the careful use exome sequencing may help identify potential candidate genes. Finally, it is interesting to note that like the mutation positive cases, the majority of mutation negative patients had also been referred by geneticists within clinical genetics departments including six from members of the ESDN panel. This would indicate that in these patients there are clear difficulties in making a correct diagnosis rather than a lack of relevant expertise.
In summary, we have shown that in the context of PSACH and the MED disease spectrum, the classical form of PSACH is relatively straightforward to diagnose provided there is sufficient clinical and radiographic information. In cases of PSACH, a COMP mutation should be identified, however, we have additional evidence to confirm that a PSACH-like phenotype is distinct from classical PSACH and does not result from a COMP mutation [Spranger et al., 2005]. In contrast, the radiographic signs of MED are more subtle and variable, and while the ESDN panel was relatively successful in predicting genotype from the phenotype, MED remains more difficult to diagnose correctly. Our study confirms that accurate review by an expert panel may help in prioritizing the genes to be sequenced and thus reduce both time and cost. Those cases that remain “mutation negative” should be carefully re-reviewed and alternative diagnoses possibly considered. Finally, our comprehensive study also throws doubt on previous studies that have suggested that mutations in the known genes are not the major cause of MED [Jakkula et al., 2005], and we conclude that mutations in COMP, MATN3, and type IX collagen genes account for the vast majority of classical AD-MED.
We would like to thank numerous colleagues who have referred cases to the European Skeletal Dysplasia Network.
Additional Supporting information may be found in the online version of this article