Osteoporosis International

, Volume 25, Issue 2, pp 399–405

A brilliant breakthrough in OI type V

Authors

  • S. Lazarus
    • University of Queensland Diamantina Institute
    • Department of Endocrinology, Royal Brisbane and Women’s Hospital
    • University of Queensland Centre for Clinical Research
  • P. Moffatt
    • Shriners Hospital for Children
    • University of Queensland Diamantina Institute
    • Department of Endocrinology, Royal Brisbane and Women’s Hospital
    • University of Queensland Centre for Clinical Research
    • University of Queensland Diamantina Institute
Review

DOI: 10.1007/s00198-013-2465-8

Cite this article as:
Lazarus, S., Moffatt, P., Duncan, E.L. et al. Osteoporos Int (2014) 25: 399. doi:10.1007/s00198-013-2465-8

Abstract

Interferon-induced transmembrane protein 5 or bone-restricted ifitm-like gene (Bril) was first identified as a bone gene in 2008, although no in vivo role was identified at that time. A role in human bone has now been demonstrated with a number of recent studies identifying a single point mutation in Bril as the causative mutation in osteogenesis imperfecta type V (OI type V). Such a discovery suggests a key role for Bril in skeletal regulation, and the completely novel nature of the gene raises the possibility of a new regulatory pathway in bone. Furthermore, the phenotype of OI type V has unique and quite divergent features compared with other forms of OI involving defects in collagen biology. Currently it appears that the underlying genetic defect in OI type V may be unrelated to collagen regulation, which also raises interesting questions about the classification of this form of OI. This review will discuss current knowledge of OI type V, the function of Bril, and the implications of this recent discovery.

Keywords

BrilHyperplastic callusIFITMInterosseous membrane calcificationNext generation sequencingOsteogenesis imperfecta

Osteogenesis imperfecta

Osteogenesis imperfecta (OI) refers to a heterogeneous group of heritable disorders of bone fragility and is amongst the commonest skeletal dysplasias, with prevalence of 6–7/100,000 [1]. It is usually inherited in an autosomal dominant (AD) fashion, with a high frequency (∼25–30 %) of de novo mutations; autosomal recessive forms are much less common.

Mutations affecting the structure of type 1 collagen were the first identified causes of OI [27], and mutations in COL1A1 and COL1A2 still account for the vast majority of cases of AD OI [810]. Autosomal recessive (AR) forms of OI remain rare worldwide and not surprisingly are more commonly seen in consanguineous families. AR forms of OI are rarely due to mutations in COL1A1 or COL1A2 [11] but more commonly are due to mutations in genes involved in the synthesis and processing of type 1 collagen. These include CRTAP [12], LEPRE1 [13] and PPIB [14] which code for proteins comprising the 3' prolyl hydroxylation complex; chaperone proteins such as SERPINH1 [15], FKBP10 [16] and SERPINF1 [17, 18]; and BMP1 [19], involved in procollagen processing.

In recent years, the molecular basis of a number of other OI-like syndromes, with bone fragility and (usually) other distinct clinical features, have also been established. Of note, these forms of bone fragility do not always arise from mutations in genes currently known to affect type 1 collagen production. For example, mutations in Sp7, coding for the osteoblast-specific transcription factor osterix [20], truncation of SERPINF1 encoding for pigment epithelium-derived factor (PEDF) [17, 18] and mutations in WNT1 [21, 22] have also been demonstrated in OI patients.

Mutations causing OI have recently been reviewed extensively in Forlino et al. [23], and a full summary of current genes associated with OI can be found at https://oi.gene.le.ac.uk.

OI type V and IFITM5/BRIL

Another exception to the dominance of type 1 collagen in the aetiology of OI appears to be the recent discovery that OI type V is caused by mutations in IFITM5 [24, 25], which codes for interferon-induced transmembrane protein 5, also known as bone restricted IFITM-like protein (Bril) [26].

OI type V was defined by Glorieux et al. [27] and is distinguished from other forms of bone fragility by the presence of interosseous membrane calcification (or, rather, new bone formation) [28] and hyperplastic callus formation. Although a single mutation appears to be responsible for OI type V, in common with other forms of AD OI, the clinical phenotype can vary [24, 2830]. Calcification of the forearm interosseous membrane is the most consistent clinical feature, with almost all cases in published case series demonstrating this feature and its associated limitation of forearm supination/pronation [24, 2729, 31, 32]. In many patients, radial head dislocation occurs as a consequence, causing a characteristic elbow deformity (Fig. 1) [24, 27, 29, 32]. Hyperplastic callus is perhaps the most conspicuous feature of OI type V, where exuberant callus formation mimicking the appearance of osteosarcoma occurs during fracture healing, although this is not seen in all patients (estimates of 8–65 %) [24, 2729, 31, 32]. In patients who have undergone bone biopsy, histology has consistently shown mesh-like bone lamellation [27, 28].
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-013-2465-8/MediaObjects/198_2013_2465_Fig1_HTML.gif
Fig. 1

Radiographs from an OI type V patient showing the characteristic bilateral calcification of the forearm interosseous membrane (a) and radial head dislocation (b). Photograph of the elbow deformity (hand discolouration is unrelated) (c). Images courtesy of A/Prof Craig Munns, The Children's Hospital at Westmead, Australia

There is marked variability in bone fragility. Some patients manifest profound bone fragility with multiple fractures, particularly in childhood [24]. In our own cohort, we have one adult carrying the mutation who to date has never had a clinical fracture (unpublished data). Rauch et al. reported that 90 % of patients in their cohort had vertebral compressions, although whether these represented fracture or the consequence of abnormal vertebral growth could not be determined [28]. Fracture variability may have resulted in an ascertainment bias in the clinical diagnosis of OI type V, as cases with low fracture rates may have been missed.

Affected individuals also have variable height and BMD, although the use of bisphosphonate therapy may alter these phenotypic features [28]. Other less consistently reported radiological features include the presence of a hyperdense metaphyseal band and a widened irregular appearance to the growth plate in younger children [27, 33]. Blue sclerae have been reported in two cases [29]; dentinogenesis imperfecta has not been reported.

Through whole exome sequencing of individuals with OI type V and their unaffected relatives, two groups independently identified a heterozygous mutation in the 5' UTR of IFITM5, c.-14C > T, that co-segregated completely in affected individuals [24, 30]. The mutation was predicted to generate a premature in-frame start codon, and transfection of wild type and mutant IFITM5 into HEK293 cells confirmed a mutant protein that was five amino acids longer than the wild type [30]. We recently confirmed the presence of the longer mRNA transcript in the bone from an affected patient with OI type V (S. Lazarus, unpublished data). Subsequent studies reported the same mutation in all other patients diagnosed with OI type V that have been sequenced to date [28, 29] including our own Australasian cohort (S Lazarus, unpublished data).

Not surprisingly, this has generated great interest in what was previously a little-known protein. IFITM5 was originally annotated as a member of the IFITM gene family through in silico analysis based on homology and chromosomal localisation only with no experimental data. The IFITM family contains five members clustered on chromosome 11: IFITM1, IFITM2, IFITM3, IFITM5 and IFITM10. Until recently, no functional roles had been delineated for any of the IFITM family members. A number of publications have now demonstrated that IFITMs 1, 2 and 3 play key roles in defence against viral infection [34, 35]. However, none of these studies have demonstrated a role for IFITM5 specifically in viral response, and, unlike other family members, this gene does not show a response to interferons [36, 37].

Closer analysis of the gene revealed that although chromosomal localisation suggests a common heritage, IFITM5 sequence homology with other IFITM-family members is weak, suggesting it is not closely related functionally to other IFITMs (Fig. 2) [26, 38]. Amino acid sequence identity between the IFITM family members is 75–90 %, whereas IFITM5 only shares 30 % identity with the other IFITM proteins with most of this homology lying in the transmembrane domains. Importantly, there is no homology in the N- and C-termini extracellular regions which are likely to define the function of the protein through interactions with other cells/proteins (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-013-2465-8/MediaObjects/198_2013_2465_Fig2_HTML.gif
Fig. 2

a Protein sequence alignment showing homology of human IFITM family members. Strong homology is seen between IFITM 1–3 but very poor homology is seen between IFITM5 and IFITM1-3 particularly in the N- and C-terminal extracellular regions. Extracellular domains (yellow shading), transmembrane region (red shading) and intracellular domain (green shading). Black shading highlights residues conserved across all IFITM proteins, grey shading shows those conserved across only IFITM1-3. b Bril membrane orientation showing the N- and C-termini extracellular domains. The factors that interact with the extracellular domain are currently unknown as are the pathways mediating the intracellular signalling

A functional role for IFITM5 in bone was first indicated from a screen undertaken to identify novel bone genes and bone regulatory pathways using a virus-based signal-trap system for membrane-bound and secreted proteins [39]. Screening the rat UMR106 osteosarcoma cell line, we identified IFITM5 as being expressed in bone. We then undertook extensive localisation and functional studies (see below) [26] and consequently renamed IFITM5 Bone restricted ifitm-like gene (Bril) to better reflect its skeletal role. Bril has also been detected in bone in a number of other studies [24, 4043]. Through extensive tissue screening at both the RNA and protein level, we demonstrated that Bril was only expressed in osseous tissues and was present in both flat and long bones throughout life, although expression decreased in aged bones, perhaps in part due to reduced numbers of osteoblasts. Bril staining in adult bone showed Bril localisation in active osteoblasts in the growth plate and on periosteal surfaces. Bril was not detected in chondrocytes or osteocytes. This in vivo localisation suggesting Bril to be an osteoblast-specific gene was further supported by the demonstration of Bril expression in a number of osteoblast cell lines and primary osteoblasts, with no expression in non-osseous cell lines. Co-localisation of Bril and bone sialoprotein (a mineralisation-associated protein) in the mineralising nodules of primary rat osteoblasts further suggested a role in the matrix development/mineralisation process [26].

To ascertain a function for Bril in bone cells, we conducted both overexpression and knock-down studies of Bril in osteoblast cell lines. Overexpression of Bril in UMR106 and primary osteoblasts resulted in a dose-responsive increase in mineralisation. Conversely, knockdown in MC3T3 cells with Bril-specific shRNA showed markedly decreased mineralisation [26].

Although Bril clearly demonstrated specific expression in the skeleton and functional effects in vitro, a role for Bril was yet to be demonstrated in vivo. A number of mouse models have been generated with deletion of Bril. Hanagata's group generated a Bril knockout mouse which has a minor reduction in bone length during prenatal development but no adult skeletal abnormalities [44]. IfitmDel mice, in which the whole Ifitm locus is deleted [45], display defects in viral host defence [34]; however, our analysis of these mice did not show any skeletal defects (G Thomas, unpublished data). We have also generated a Bril-knockout mouse which also has no skeletal phenotype (P Moffatt, unpublished data). Database searching in the Database of Genomic Variants and DECIPHER identified two individuals with heterozygous genomic deletions resulting in complete deletion of Bril [30]. Neither of these individuals exhibited severe skeletal defects. Thus, an in vivo role for Bril in bone was unclear.

However, the two recent publications reporting the identification of an IFITM5/Bril mutation as a cause of OI type V have strongly demonstrated that Bril does in fact play a major role in human skeletal physiology [24, 30]. The in silico analyses presented in these papers suggested the addition of 5 amino acids would not affect protein localisation or glycosylation [30]. With the animal models suggesting that deficiency in Bril has little effect on the skeleton, it seems more likely that the mutations underlying OI type V result in a gain-of-function in Bril. In vitro, overexpression of Bril in osteoblasts results in increased mineralisation [26]; this may contribute to hyperplastic callous formation. Further experimental validation is required either through generation of transgenic mice overexpressing Bril or knock-in mice expressing the human mutation.

Of note, recently, a de novo coding mutation in IFITM5/Bril (S40L) was reported in a child aged 6 years who had blue sclerae, macrocephaly, scoliosis and long bone bowing but no history of ligamentous calcification [25]. Some of these features have also been reported in patients carrying the c.-14C > T mutation [29]. Thus, mutations in IFITM5/Bril may cause a spectrum of bone disease broader than the clinical phenotype of OI type V. This would not be unique for skeletal disease—for example, FGFR3 mutations may result in several different clinical diseases including hypochondroplasia, achondroplasia and thanatophoric dysplasia (OMIM 134934).

One key piece of information still lacking is the mode of action of Bril, as is the case for all of the IFITM proteins. Interaction with endosome proteins has been proposed [46], but the motifs mediating these interactions are not present in Bril. A recent conference report suggested that a different mutation in Bril might contribute to OI through an interaction with PEDF [25]. Defects in PEDF are thought to cause OI type VI, and PEDF knockout mice display a phenotype resembling OI type VI [47] with enhanced mineralisation. A direct interaction between PEDF and Bril has yet to be demonstrated; however, Bril has also been shown to be regulated by members of the bone-associated hedgehog and Sp transcription factor families [48]. There is a pressing need to identify which molecules interact with Bril, presumably through its extracellular domains, and which intracellular signalling pathways are activated through the intracellular domain. Such studies may start to explain why mutations in 5' UTR region cause the OI type V phenotype. It may be that Bril does in fact interact with collagen or collagen-associated proteins, thereby aligning OI type V more closely with the aetiology of other forms of OI. It is also possible that other apparently non-collagen-related genes may be found to cause bone fragility syndromes. Certainly, the recent advances of massive parallel sequencing are resulting in an explosion of genetic discoveries in monogenic disorders including many skeletal dysplasias [49, 50]. Such advances will further add to our understanding of conditions such as OI.

Implications

  1. (a)

    A novel pathway in bone

    Bril is completely novel in bone biology, both for its lack of homology with other molecules and the lack of conserved signalling/interacting motifs that might point to a mode of action. Thus, it seems possible that elucidation of the mechanism of Bril function will result in the discovery of a new pathway in bone regulation or highlight new levels of regulation of known skeletal biology. In the last 20 years, only two novel bone regulatory pathways have been described, the OPG-RANKL-RANK [51] and Wnt [52] pathways. Both pathways have resulted in novel therapies for bone disease—anti-RANKL antibodies (denosumab) [51] and anti-SOST antibodies [53]. Similarly, Bril may also point to novel therapeutic approaches for bone disease.

     
  2. (b)

    Where does OI type V fit into the classification system for OI?

    The recent discovery of the cause of OI type V forces a revisit to its classification too. This was not unexpected; indeed, Van Dijk et al. suggested as “it is unknown whether the genetic causes of OI type V…affect collagen type I biosynthesis…we propose to consider these types as new syndromes instead of particular types of OI” [1]. Furthermore, the ubiquitous feature of OI type V appears not necessarily to be bone fragility, but the presence of intramembranous ossification resulting in elbow deformity and propensity to hyperplastic callous—and in many cases, the manifestation of elbow malformation precedes the first clinical fracture. It would seem logical, therefore, both due to its clinical features and its unique genetic aetiology, to move OI type V from inclusion in the overarching five Sillence categories to a recently proposed related group “syndromes of bone fragility or osteoporosis plus additional features” [54], in which the underlying genetic defect would contribute significantly to the classification. This is not without precedent—classification of osteopetrosis changed with increased understanding of the underlying aetiologies [54, 55]. Given that the commonest clinical feature of OI type V appears to be periosteal new bone formation associated with bone fragility, this could provide the starting point for a new nomenclature to capture this most recent breakthrough in bone.

     

Conflicts of interest

None

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2013