Elsevier

Translational Research

Volume 181, March 2017, Pages 27-48
Translational Research

In-Depth Review of Biology and Treatment of Bone Disorders
Osteogenesis imperfecta: new genes reveal novel mechanisms in bone dysplasia

https://doi.org/10.1016/j.trsl.2016.11.005Get rights and content

Osteogenesis imperfecta (OI) is a skeletal dysplasia characterized by fragile bones and short stature and known for its clinical and genetic heterogeneity which is now understood as a collagen-related disorder. During the last decade, research has made remarkable progress in identifying new OI-causing genes and beginning to understand the intertwined molecular and biochemical mechanisms of their gene products. Most cases of OI have dominant inheritance. Each new gene for recessive OI, and a recently identified gene for X-linked OI, has shed new light on its (often previously unsuspected) function in bone biology. Here, we summarize the literature that has contributed to our current understanding of the pathogenesis of OI

Introduction

Osteogenesis imperfecta (OI) is a heterogeneous group of rare genetic disorders of connective tissue. With its primary impact on bone, it is also known as “brittle bone disease.” Typical clinical manifestations are bone fragility, low bone mass, skeletal deformities, and short stature. It is also a systemic disorder that involves extraskeletal structures. This encompasses various clinical manifestations, including dentinogenesis imperfecta, blue-gray sclera, hearing impairment, joint hypermobility, muscle hypotonia, restrictive pulmonary disease, and cardiovascular abnormalities. Overall, the prevalence of OI is estimated at 1 in 15,000–20,000 births. The severity of OI ranges from perinatal lethal and severely deforming types to very mild forms without deformity. Genetic heterogeneity of OI is further complicated by extensive phenotypic variability of each genetic locus and different modes of inheritance.

More than 80% of OI cases are due to dominantly inherited mutations in COL1A1 or COL1A2, which encode the α1(I) and α2(I) chains of type I collagen.1 With 2 exceptions (interferon-induced transmembrane protein 5 [IFITM5] and WNT1), mutations in noncollagen genes are associated with recessive forms of OI.2 One recessive type is X-linked. Generally, homozygous or compound heterozygous loss-of-function mutations result in severely decreased or absent production of protein. Recessive OI genes can be generally categorized based on the cellular pathways in which their molecular functions are executed (Table I). Recessive OI genes that are involved in collagen biosynthesis, post-translational modification, and processing include cartilage-associated protein (CRTAP), LEPRE1, PPIB, FKBP10, SERPINH1, PLOD2, TMEM38B, and BMP1. The mutations in SP7 and WNT1 indicate the significance of osteoblast development and activity. Delineation of mutations in membrane-bound transcription factor peptidase site 2 (MBTPS2) and CREB3L1 underscores the critical importance of regulated intramembrane proteolysis (RIP) in skeletal development. Improved understanding of osteoblast differentiation and bone matrix mineralization will be derived from studies of OI caused by mutations in IFITM5 and SERPINF1. Overall, identification of new OI-causing genes has greatly extended the scope of cellular and biological pathways for OI pathogenesis. Here, we review the recent advances in mechanistic implications of those genes in OI pathogenesis.

Type I procollagen, like other fibrillar collagens, is characterized by N- and C-terminal propeptide sequences flanking a central helical region containing uninterrupted Gly-X-Y repeats (where X and Y represent any amino acid but are frequently proline and hydroxyproline). In type I procollagen, 3 α-chains (two α1 and one α2 chain) are assembled, beginning with chain assortment and alignment in the C-propeptide. The type I procollagen molecule undergoes post-translational modification in the endoplasmic reticulum (ER). Procollagen maturation requires the cleavage of the propeptides by specific N- and C-terminal propeptidases.

Mutations in the two type I collagen genes, COL1A1 and COL1A2, cause autosomal dominant OI. The OI-causing mutations in type I collagen can be divided into quantitative defects, with the synthesis of structurally normal collagen at about half the normal amount, and structural defects, that result in synthesis and secretion of abnormal collagen molecules. The quantitative defects are generally caused by premature termination codons in one COL1A1 allele, which initiate the nonsense-mediated decay of transcripts from that allele. The resulting matrix deficiency causes mild type I OI.3, 4, 5, 6, 7

Collagen structural mutations that alter the chain sequence in the triple helical domain result in a wide phenotypic range from lethal type II OI to moderate type IV OI. Most commonly, these mutations cause a substitution of one of the invariant glycine residues in the triple helical domain of α1(I) or α2(I) chains, because of the crucial role of glycine in helix formation.8, 9 Mutations in COL1A1 that result in the substitution of glycine by serine, cysteine, and arginine account for 78.6% of all α1(I) substitutions. About one-third of all independent glycine substitutions in α1(I) result in perinatal lethal type II OI; especially, substitution by amino acids with charged or branched side chains. About a quarter of all COL1A1 glycines at which substitutions have been identified now have 2 or more independent occurrences of substitutions; these residues are scattered along the length of the α1(I) chain and are often associated with CpG dinucleotides. Interestingly, recurrences at the same α1(I) glycine residue frequently have different clinical outcomes.10

Mutations in COL1A2 that result in α2(I) glycine substitutions by serine, aspartic acid, and valine comprise about three-quarters of known independent events in the helical domain, although substitutions by arginine, valine, and alanine should predominate if mutations occur randomly in the first and second positions of the COL1A2 glycine codons. Substitutions by alanine are markedly under-represented in α2(I), as they were in α1(I), which likely represents an ascertainment bias caused by the generally milder outcome of alanine substitutions. A smaller proportion of α2(I) glycine substitutions resulted in the lethal type II OI phenotype (less than 20%, compared to 35.6% in α1[I]). As for α1(I), substitutions by charged residues generally lead to more severe clinical outcomes. Unlike α1(I), valine substitutions have a lethal outcome of less than one-fifth of occurrences, whereas cysteine has a lethal outcome in a quarter of cases. The distribution of lethal substitutions along the chain also differs in α1(I) and α2(I). In both chains, substitutions in the amino end of the helix have a nonlethal outcome. In the remainder of the α2(I) chain, there is an alternation of lethal and nonlethal regions with the lethal regions coinciding with the proteoglycan-binding regions on the collagen chain. In contrast, α1(I) has 2 lethal regions that coincide with multiple ligand binding regions (MLBRs) on the α1(I) chain in the collagen monomer.11 However, there is no genotype-phenotype correlation that is sufficiently accurate to be used for making a decision as to whether an individual child will be lethal or nonlethal. The general phenotype-genotype correlations should not be the deciding factor in pregnancy terminations.

Abnormal matrix and delayed osteoblast development influence cell signaling between osteoblasts and osteoclasts, increasing bone remodeling and contributing to bone weakness. Ultrastructural examination of OI bone revealed increased numbers of osteocytes, with multiple osteocytes in some lacunae. In Brtl mouse, osteoclast numbers are elevated in femora, uncoupled from osteoblast numbers. Brtl osteoclast precursors from marrow are larger, more numerous, and more intensely tartrate resistant acid phosphatase (TRAP) stained than in wild type. The RANKL/OPG ratio is normal in Brtl bone; so, other soluble factors secreted by mutant osteoblasts may stimulate osteoclast development. In the oim/oim mouse, elevation of the RANKL/OPG ratio and higher expression of TNF-α were detected in sorted immature osteoblasts, supporting cell-cell signaling as a key aspect of elevated bone turnover in OI.12, 13 Cross-links in collagen fibrils are important for preosteoblast maturation. In OI, collagen located at the surface of fibrils had fewer crosslinks than in the fibril interior. Contact with cross-link–deficient matrix by OI bone cell populations could contribute to impaired osteoblast maturation and increased osteoclast recruitment. Also, the effects of collagen heterogeneity in dominant OI could be mediated in part by abnormal crosslinking.14, 15

The Brtl mouse model, a knock-in model for type IV OI, is heterozygous for an α1(I) Gly349Cys substitution. About 30% of mutant pups are lethal, whereas nonlethal Brtl mice reproduce the phenotype, histology, biochemistry, and biomechanics of OI. A recent study also demonstrated abnormal cytoskeleton by immunohistochemistry selectively in cells from lethal Brtl mice. The aberrant cytoskeleton was associated with impaired osteoblast proliferation, collagen deposition, and integrin and transforming growth factor-beta (TGF-β) signaling.16 A knock-in mouse model for the α2(I) chain, which reproduces the mutation found in a large Amish pedigree, has a heterozygous Gly610Cys substitution. Long bones of the so-called Amish mice are less fragile than Brtl femora.17 Another mouse model known as Aga2 has a dominant mutation located in the terminal C-propeptide that was created using an N-ethyl-N-nitrosourea mutagenesis strategy. The Aga2 mouse has perinatal lethal and severe surviving forms. The Aga2 osteoblasts have increased apoptosis due to intracellular retention of abnormal collagen chains.18

Once the procollagen molecules are secreted from the osteoblasts, they undergo an extracellular maturation process in which the N- and C-propeptides are removed by specific proteases. Mutations altering the propeptide cleavage sites cause specific phenotypic variants of OI.19 Processing defects of the C-propeptide have a dominant form caused by substitutions in the cleavage site and a recessive form caused by defects in the processing enzyme; both lead to high bone mass OI.20, 21, 22, 23, 24, 25 Heterozygous substitutions in both residues (Ala-Asp) of the C-propeptide cleavage site in COL1A1 and COL1A2 have been reported, resulting in the incorporation of pC-collagen into the extracellular matrix (ECM) and in collagen fibrils with irregular crosssections.22, 23, 24, 25 Prepubertal children with these defects have mild OI with few fractures, normal stature, sclerae and teeth, and increased DXA Z-scores. Their bone tissue showed a substantial increase in mineralization by bone mineralization density distribution (BMDD) and fourier transfer infrared spectroscopy (FTIR), even above the hypermineralization of classical OI. High bone mass in these patients may be initiated by localization of pC-collagen on the fibril surface, functioning as nucleators of mineralization.22

Recently, homozygous or compound heterozygous mutations in BMP1, encoding the metalloproteases bone morphogenetic protein-1 (BMP1) and its longer isoform mammalian Tolloid (mTLD), were identified in 5 children with a severe autosomal recessive form of OI and in 4 individuals with mild-to-moderate bone fragility. BMP1/mTLD functions as the procollagen carboxy-(C)-proteinase for procollagen types I to III and trimming other ECM substrates. Their phenotype was characterized by numerous fractures, short stature with rhizomelia, deformity of the limbs, and variable kyphoscoliosis. BMP1/mTLD-deficiency in humans not only results in delayed cleavage of the type I procollagen C-propeptide but also hinders processing of the small leucine-rich proteoglycan prodecorin, itself a regulator of collagen fibrillogenesis, highlighting the significance of BMP1/mTLD-deficiency on ECM organization.23, 24, 25

Defects in processing the N-terminal propeptide of type I collagen also comprise dominant and recessive forms. The rare EDS arthrochalasia subtype results from complete or partial skipping of exon 6 in either COL1A1 or COL1A2, which encodes the N-proteinase cleavage site. Incompletely processed procollagen chains, in which the C- but not the N-propeptide has been cleaved, are secreted and incorporated into the growing collagen fibrils. The phenotype of this rare EDS variant is characterized by severe, generalized joint hypermobility and recurring joint dislocations, including bilateral congenital hip dislocation, skin hyperextensibility, atrophic scarring, mild dysmorphic features, short stature, blue sclerae, and osteopenia. In another variant, patients display a phenotype that combines clinical manifestations of both OI and EDS. The reports on such patients with a mixed OI/EDS phenotype demonstrate that they harbor a mutation in the most N-terminal part of the type I collagen helical region in either the α1(I) or α2(I) chain, which unfolds the region containing the N-terminal cleavage site. Since the N-proteinase requires a proper tertiary structure and will not cleave unfolded chains, N-propeptide processing is impaired to a variable extent. Thus, the patients have OI/EDS with the OI component caused by the location of the mutation in the collagen helix and the EDS component caused by impaired N-propeptide cleavage. Incorporation of pN-collagen disturbs normal collagen fibrillogenesis and results in thin fibrils.26

Mutations in the genes encoding CRTAP and prolyl 3-hydroxylase 1 (P3H1 encoded by LEPRE1) were the first identified causes of recessive OI. The physiologic importance of a functional prolyl 3-hydroxylation complex for normal bone formation is highlighted by the fact that deficiency of CRTAP or P3H1 causes a very severe to lethal bone dysplasia in mice and children.27, 28, 29, 30, 31 Null mutations in CRTAP or LEPRE1 cause OI types VII and VIII, respectively, both of which result in overmodification of the full collagen helical region.28, 29, 30

CRTAP and P3H1, together with cyclophilin B (encoded by PPIB), form a complex within the ER that post-translationally modifies the specific proline residues, α1(I)/α1(II) Pro986 and α2(I) Pro707 in unfolded collagen α-chains (Fig 1). In addition to prolyl 3-hydroxylation, the P3H1/CRTAP/CyPB complex also functions as a PPIase, since CyPB is a peptidyl-prolyl cis-trans isomerase for type I collagen, and as a chaperone, preventing type I collagen chains from forming premature aggregates in the ER.32 P3H1 and CRTAP stabilize each other; absence of one results in degradation of the other. Recent data suggest that prolyl 3-hydroxylation of Pro986 is not required for the structural stability of collagen; however, the lack of 3-hydroxylation at Pro986 may affect the nearest neighbor relationships required for the optimum cross-linking arrangement and supramolecular assembly to enable mineral crystal deposition together with retained fibril strength.33, 34, 35

CRTAP (the helper protein of the complex) was first identified in hypertrophic chick chondrocytes in culture.36 In the skeletal system, CRTAP is expressed in long bone epiphyses, articular cartilage, and lower hypertrophic zone cartilage in the chick37 and in the growth plate, bone collar, and calcified cartilage of the chondrosseous junction in the mouse.29 Furthermore, CRTAP is also found to be expressed by osteoblasts and osteoclasts in bone.

Mice homozygous for knock-out alleles of the Crtap gene have a severe osteochondrodysplasia with rhizomelia and osteoporosis. They develop progressive and severe kyphoscoliosis by 6 months of age, exhibiting prenatal and postnatal growth delay with shortening of the proximal limb bones and osteopenia. Absence of CRTAP resulted in loss of α1(I) and α1(II) Pro986 3-hydroxylation. Moreover, bone histomorphometry revealed reduction in osteoid deposition and bone formation rate, and a delay in matrix apposition rate, in spite of normal osteoblast number. Collagen fibrillogenesis is also affected, altering dermal fibril diameters.29

Although type VII OI had been clinically identified in patients with rhizomelia, the phenotype of Crtap−/− mice was among the critical factors that led to the identification of CRTAP mutations as the first cause of recessive OI. Type VII OI is a lethal/severe recessive chondrosseous dysplasia. Fractures and limb deformities are present at birth. Infants with type VII OI frequently die in the first year of life as the result of pulmonary insufficiency or infections.29 The severity of this form of OI varies based on the type of CRTAP mutation.28, 29 Patients with null defects almost invariably die in the perinatal period.28, 38 However, an extended pedigree from Quebec,38 which was actually the index pedigree for type VII OI, has moderately severe OI, with rhizomelia and white sclerae, associated with a hypomorphic CRTAP mutation.

Recently, the consequences of CRTAP deficiency on bone material properties was evaluated using quantitative backscattered electron imaging (qBEI) to assess BMDD in femurs from 12-week-old Crtap−/− mice and transiliac bone biopsies from 4 children with hypomorphic CRTAP mutations. The bone tissue of both Crtap−/− animals and OI type VII patients had significantly increased mean calcium (CaMean) and calcium peak compared to wild-type littermates and control children, respectively. The heterogeneity of mineralization (CaWidth) was reduced in Crtap−/− mice but normal in OI type VII patients. The fraction of highly mineralized bone matrix (CaHigh) was remarkably increased in the patients, associated with the persistence of primary bone. The BMDD data suggested that CRTAP deficiency results in a shift toward higher mineral content of the bone matrix similar to classical OI with collagen gene mutations.39

Crtap−/− mice also have abnormal connective tissue in the lungs, kidneys, and skin, consistent with systemic dysregulation of collagen homeostasis. Both Crtap−/− mice lung and renal glomeruli showed increased cellular proliferation. Histologically, the lungs showed increased alveolar spacing, whereas the kidneys showed evidence of segmental glomerulosclerosis, with abnormal collagen deposition, possibly because some CRTAP is secreted.40, 41 A recent report suggested excess TGF-β signaling is present in Crtap−/− bone, based on higher expression of TGF-β target genes and Smad activation. Anti–TGF-β treatment using the neutralizing antibody 1D11 improved the bone and lung abnormalities in Crtap−/− mice. However, the role of TGF-β as an effector in tissue pathology, rather than an elevated bystander, remains to be definitively demonstrated.

The LEPRE1 gene, which encodes prolyl 3-hydroxylase 1, the enzymatic component of the rough endoplasmic reticulum-resident complex, was first isolated from a rat yolk sac tumor cell line as a novel matrix chondroitin sulfate proteoglycan.42 P3H1 expression is localized to tissues rich in fibrillar collagens in mouse. Type VIII OI, which is caused by homozygosity for null LEPRE1 alleles, is clinically similar to type VII OI. Type VIII OI patients present with severe to lethal OI, white sclera, rhizomelia, and undertubulation of long bones. Some individuals with this type of mutation have been reported to attain their third decade.43, 44, 45, 46, 47 Those who survive into childhood have extremely low BMD, severe growth deficiency, and bulbous metaphyses. A LEPRE1 founder mutation, originating in West Africans and also carried by a low percentage of African Americans, occurs in almost half of the type VIII OI cases reported.48

Lack of P3H1, due to mutations in LEPRE1, results in severe recessive OI, which is very often lethal in the perinatal period and resembles OI types IIB and III.49, 50 The P3H1-deficient mice have a milder phenotype than type VIII OI patients. The general phenotype of the P3H1 deficient mouse model51 is characterized by abnormalities in skin, tendon, and bone due to collagen fibril disturbances, as well as hearing loss.52 P3H1-deficient mice were viable and fertile but had shorter, less radiodense, and weaker bones than wild-type littermates. Collagen folding and secretion by skin fibroblasts was delayed, leading to overmodification of collagen molecules and disturbances in the assembly of higher order collagen fibrils.

P3H1 deficiency in mice results in low trabecular bone volume and mineral apposition rate but osteoid maturation time and osteoblast and osteoclast surfaces are normal. Quantitative backscattered electron imaging yielded hypermineralization of bone matrix, as in types VII and VIII OI patients and classical dominant OI. It thus appears that P3H1 deficiency leads to decreased deposition of ECM by osteoblasts and increased incorporation of mineral into the matrix.53 Because P3H1 and CRTAP form a mutually stabilizing complex in the ER, it would be expected that the histology of type VII and VIII OI patients and the knockout murine models for P3H1 and CRTAP would be similar to each other but might differ from OI with a collagen structural abnormality. In fact, bone from patients with nonlethal type VIII OI showed features similar to type VII OI bone. Distinctive features of type VIII OI bone histology included extremely thin and sparse trabeculae and focal accumulation of unmineralized osteoid. The proportion of bone undergoing primary mineralization is increased in type VIII OI, in comparison to type VII and I OI. Reduction of trabecular thickness in femora of P3h1−/− mice was not as severe as in patients, consistent with a less-severe phenotype in P3h1−/− rodents than children. The type VIII OI patient bone had a porous and trabecularized cortex, suggesting active cortical bone remodeling and supported by elevated serum TRACP-5b and increased eroded surface, although osteoclast number and surface were not elevated.54 Notably, neither the CRTAP nor the P3h1−/− mice have high bone turnover; instead, they have reduced bone formation and normal to low osteoblast and osteoclast indices, indicating an intrinsic osteoblast defect.55

To begin to address the question of whether 3-hydroxylation itself was critical to recessive OI, or whether the lack of the functions of the ER-resident complex were rather at fault, knock-in mice carrying a single amino acid substitution in the catalytic site of P3h1 (Lepre1 [His662Ala]) were generated. The mutation substantially abolished P3h1 3-hydroxylation activity, and significant overmodification of the collagen helix was not detected by gel electrophoresis. However, the inactive P3H1 retained the ability to form a complex with CRTAP, as well as the collagen chaperone function. The mice had normal size and growth plate histology but their trabecular bone mass was reduced. This data suggests differential tissue consequences to selective inactivation of P3H1 hydroxylase activity vs complete ablation of the complex.53 This question will be further clarified by generation of a knock-in mouse with a substitution at α1(I) Pro986 that cannot be modified.

Cyclophilin B, encoded by PPIB, is the third member of the 3-hydroxylation complex. Mutations in this gene cause type IX OI, which differs phenotypically from P3H1 or CRTAP deficiency, and falls into 2 groups. Some children had lethal OI, whereas others had moderately severe bone dysplasia.56, 57 The major phenotypic difference is that patients with type IX OI do not have rhizomelia; however, they share the white sclerae and normal dentition of types VII and VIII.56 Type IX OI also differs biochemically: in the absence of P3H1 or CRTAP, α1(I) Pro986 3-hydroxylation is nearly absent and collagen folding is delayed, resulting in collagen helical overmodification, whereas lack of CyPB is associated with α1(I)Pro986 3-hydroxylation levels ranging from 1/3 to normal, as well as normal helical modification.58, 59

CyPB is ubiquitously expressed and its stability is independent of CRTAP/P3H1. It is an ER-localized member of the immunophilin family of proteins with peptidyl-prolyl cis-trans isomerase (PPIase) activity.60, 61 CyPB plays a key role as a member of several foldase and chaperone complexes, with partners including BiP, GRP94, PDI, and calreticulin, to facilitate folding of multiple substrates within the ER.54, 62 Cis-trans isomerization of peptidyl-prolyl bonds is especially important for type I collagen folding because prolines constitute approximately one-fifth of its sequence. Studies published over 20 years ago demonstrated that exposure of cells to the cyclophilin inhibitor cyclosporin A (CsA) slows the rate of collagen folding and results in overmodification of lysyl residues.63 Thus, CyPB was thought to be the major, and possibly unique, collagen peptidyl-prolyl cis-trans isomerase.64 However, normal folding of collagen in the absence of CyPB suggests that there is redundancy for this critical function.

Ppib−/− mice recapitulate type IX OI. They are small, with reduced femoral areal bone mineral density, bone volume (BV/TV) and mechanical properties, and increased femoral brittleness. Like type IX OI patients, they do not have rhizomelia. Biochemical studies reveal novel cell- and tissue-specific dysregulation of collagen helical lysyl hydroxylation and glycosylation in the absence of CyPB, independent of impaired collagen folding. Furthermore, reduced hydroxylation of specific collagen helical lysine residues led to a shift in the pattern of intermolecular crosslinks in bone tissue, and reduced collagen deposition into matrix. These studies establish novel functions for CyPB in regulating collagen biosynthesis and post-translational modification.56, 57, 59, 64, 65 In tendon collagen, CypB deficiency resulted in lower lysine hydroxylation in the helical cross-linking sites and increased hydroxylation in the telopeptide cross-linking sites. This resulted in generation of hydroxylysine aldehyde–derived crosslinks, that were absent from wild type and Ppib+/− mice. CypB was shown to interact with all lysyl hydroxylase isoforms (isoforms 1–3) and a lysyl hydroxylase-2 chaperone, FK506-binding protein. Tendon collagen in Ppib−/− mice showed severe organizational abnormalities.66

Ca2+ release from the ER and sarcoplasmic reticulum (SR) regulates important cellular functions.67 Two trimeric intracellular cation (TRIC) channels, TRIC-A and TRIC-B, are localized to ER, SR, and nuclear membranes.68, 69 Gene knockout studies indicate that TRIC channels support Ca2+ release from intracellular stores. Double knockout mice lacking both TRIC subtypes die during early embryonic stages, and their cardiomyocytes exhibit impaired RyR-mediated Ca2+ release.70 TRIC channels, therefore, may contribute to regulating the flux required for physiologic Ca2+ release in various cell types. Tric-b−/− mice die immediately after birth due to respiratory failure, and the alveolar epithelial cells from these mice exhibit insufficient production and secretion of surfactant lipids likely due to diminished IP3R-mediated Ca2+ release.71

Recently, a novel form of recessive OI (Type XIV), caused by a founder mutation, was reported in Bedouin families from Israel and Saudi Arabia. The probands from these families presented with a moderately severe bone phenotype, with bowing and fragility of long bones leading to multiple fractures in infancy. Blue sclerae but normal teeth, facies, and hearing were reported. Two groups independently identified the same mutation in TMEM38B, a deletion of exon 4 and surrounding intronic sequence.72, 73 A second mutant allele was also identified in an Albanian child born to consanguineous parents who harbored a genomic deletion including exons 1 and 2 of TMEM38B74 and had a similar phenotype. TMEM38B encodes the ER membrane protein TRIC-B, which has been proposed to function as a monovalent cation channel that regulates the kinetics of recovery for IP3R-mediated Ca2+ release from ER, the major site of the intracellular Ca2+ storage.70

The molecular mechanism through which absence of TRIC-B causes an OI phenotype was addressed in a recent report. Absence of TRIC-B from fibroblasts and osteoblasts of patients with type XIV OI was demonstrated to decrease Ca2+ flux from ER to cytoplasm and to increase the PERK4 pathway of ER stress. Multiple steps in type I collagen synthesis and modification were dysregulated in TRIC-B–deficient cells. Interestingly, collagen helical lysine hydroxylation is reduced 20%–30% in TRIC-B–deficient cells, despite increased LH1; this can be compared with the 85% reduction in collagen helical lysine hydroxylation seen in EDS type VI. These alterations in collagen modification and decreased secretion of collagen from patient osteoblasts indicate that the mechanism of type XIV OI falls within the collagen-related paradigm of recessive OI.75 Complementary data on bone histology from the lethal Tric-b−/− pups show reduced collagen in matrix and impaired bone mineralization. Murine knockout osteoblasts had impaired collagen release without a decrease in collagen-encoding transcripts, whereas their osteoclasts were similar to wild type.76

Heat shock protein 47 (HSP47; encoded by SERPINH1) is an ER-localized chaperone that preferentially recognizes the folded state of the type I procollagen trimer and helps to maintain it.77, 78 Recessive mutations in Hsp47 that lead to an OI phenotype were reported in humans and dog. Type X OI, represented by a single affected child, is caused by a recessive heterozygous mutation in SERPINH1. A single patient with a homozygous missense mutation (Leu78Pro) rendering Hsp47 unstable has severe OI, characterized by macrocephaly, an open anterior fontanel, high and prominent forehead, midface hypoplasia with shallow orbits, blue sclerae, and dentinogenesis imperfecta. He had short and bowed extremities, with generalized joint laxity and scoliosis. He died at the age of 3.5 years due to sudden onset of respiratory distress.79, 80 His mutation resulted in degradation of virtually all the abnormal HSP47 via the proteasomal degradation. The unstable HSP47 had a minimal effect on procollagen post-translational modification. Although overall type I procollagen production was effectively normal, secreted type I collagen was protease sensitive at multiple sites because of its misfolded configuration. A dachshund line with an HSP47 (p.Leu326Pro) had a severe recessive form of OI, characterized by marked osteopenia, thin bones with in homogeneous and shallow trabeculation in the foreleg, joint hyperlaxity, and undermineralization of the teeth.81

Serpinh1−/− mice are embryonic lethal at 10 dpc, suggesting a pivotal role during development81 and pleiotropic effects on collagen-containing tissues.80, 82 Type I collagen secreted from Serpinh1−/−fibroblasts is more susceptible to protease digestion, due to misfolding of its triple helix. Also, mature, processed type I procollagen was severely reduced in tissue, since type I procollagen accumulated as aggregates in the ER.83 Cleavage of the proα1(I) N-propeptide was also deficient in Serpinh1−/− fibroblasts, consistent with localized structural alterations of the substrate procollagen sequences.

FKBP10 encodes FKBP65, an ER-localized PPIase and chaperone molecule.84 FKBP65 is the largest member of the immunophilin subfamily that binds FK506 and has 4 PPIase domains (Fig 1). FKBP65 has multiple ligands, including type I collagen and elastin.85, 86 Deficiency of FKBP65 causes recessive type XI OI and was first reported in 2010 in cases of progressive deforming OI.87

The phenotype of Bruck syndrome (BS) type I, with congenital joint contractures as well as osteoporosis, fragile bones, and short stature, overlapped with the phenotype of FKBP10 mutations causing OI.88, 89, 90, 91 Bruck I was quickly identified to be allelic with OI XI. Further, individuals with the same mutation in FKBP10, even siblings, may have OI with or without contractures, showing the contractures to be a variable manifestation of the mutation.87, 89, 90, 92 Interestingly, homozygosity for a small in-frame deletion in exon 5 of FKBP10 is the cause of Kuskokwim syndrome, a congenital contracture syndrome with minimal skeletal findings. Thus, the phenotypic spectrum of FKBP10 mutations encompasses OI alone, OI with contractures (BS), and a congenital contracture syndrome.

The type I collagen produced by cells with FKBP65 deficiency has normal helical post-translational modification and normal α1(I) Pro986 3-hydroxylation, suggesting FKBP65 does not make a significant contribution to collagen folding.87, 93 A homozygous frameshift mutation in FKBP10 in a Palestinian child with moderately severe OI without contractures provided the first demonstration that type I collagen crosslinking and deposition of collagen into matrix are strongly FKBP65 dependent. Similarly, Fkbp10−/− mouse embryos are postnatally lethal and display reduced collagen crosslinking in calvarial bone.94 In humans and mice, there is near absence of collagen telopeptide hydroxylation in proband-secreted collagen. It was recently shown that FKBP65 binds to LH2 (lysyl hydroxylase 2) but not to LH1 or LH3. Binding of FKBP65 to LH2 is not hampered by a loss of LH2 activity or partial loss of FKBP65 PPIase activity. However, partial loss of PPIase activity hampers the dimerization of LH2, which is the active LH conformation, resulting in a striking decrease in collagen telopeptide lysyl hydroxylation.95

PLOD2 encodes LH2, the key enzyme responsible for hydroxylation of lysine residues outside the major triple helix of type I collagen, crucial to formation of mature intermolecular crosslinks in bone and cartilage.87, 96, 97 The hydroxylation of type I collagen telopeptide lysine residues is a key step in collagen biosynthesis, as telopeptide hydroxylysines are the precursors of a series of biochemical reactions, known as the hydroxyallysine route, which finalize with the formation of intermolecular lysylpyridinoline and hydroxylysylpyridinoline crosslinks within collagen fibrils (Fig 1). Collagen crosslinks provide stability and tensile properties to collagen fibrils.94, 98, 99

PLOD2 and FKBP10 are causative genes in BS, a condition resembling OI but that is also typically associated with congenital joint contractures, multiple fractures, and short stature. Mutations in 6 consanguineous BS families showed changes in either PLOD2 or FKBP10. Screening revealed compound heterozygous mutations in PLOD2 in 2 brothers, 1 affected with mild AR-OI and the other with mild BS. Thus, PLOD2, in addition to causing BS, is also associated with AR-OI phenotypes of variable severity. In addition, BS patients were also demonstrated to have reduced levels of hydroxylysine-derived crosslinks in type I collagen from bone.100

A recurrent mutation at the 5'-end of IFITM5 causes type V OI, the only dominantly inherited type of OI that is not due to collagen structural abnormalities. Type V OI (OMIM 610967) presents with moderate-to-severe skeletal deformity. Despite the general similarity of its clinical severity to type IV OI in fracture incidence, long-bone deformities, vertebral compression fractures, and scoliosis, many type V OI patients have distinctive clinical manifestations.101 The common clinical and radiographic features of type V OI include calcification of the interosseous membranes of the forearm and lower leg, hyperplastic callus formation at fracture sites or after surgical interventions, radial head dislocation, and a radiodense metaphyseal line. Their bone tissue has a mesh-like lamellation pattern under polarized light.101, 102 Some type V OI patients also have facial dysmorphism, including a short, upturned nose, a small mouth, and a prominent chin. Sclerae may be white or greyish-blue; dentinogenesis imperfecta is absent.103 Bisphosphonate therapy significantly exacerbated callus hyperplasia in type V OI, which was reversed by cessation of treatment.104

Type V OI has significant inter-individual phenotypic variability, even within the same family, which is remarkable given the fact that the IFITM5 causative mutation is identical among patients with type V OI.105, 106 The molecular cause of type V OI is a heterozygous mutation in the 5′-UTR of IFITM5 (c.−14C>T)107, 108, 109 that generates an alternative in-frame start codon and adds 5 residues (MALEP) to the N-terminus of bone restricted ifitm-like protein (BRIL). The IFITM5 gene is located on chromosome 11p15.5. It contains 2 exons and encodes a 134-amino acid transmembrane (TM) protein. BRIL is highly expressed in osteoblasts and developing bone. Wild-type and the longer mutant variant of BRIL proteins coexist in type V OI bones.110 Osteoblasts from a type V OI patient yielded increased mineralization in culture compared with control.111 Both early and late markers of osteoblast differentiation were increased in cultured osteoblasts from a type V OI patient. This gain-of-function mechanism at the osteoblast level appears to underlie the overactive tissue mineralization seen in patients.111

A transgenic mouse model expressing the type V OI mutant IFITM5 under the control of an osteoblast-specific col1a1 2.3-kb promoter had a neomorphic effect in bone.112 This mouse exhibited perinatal lethality and abnormal skeletal development, including abnormal rib cage formation, long bone deformities, and fractures.112 Overexpression of wild-type IFITM5 did not display a noticeable bone phenotype. Osteoblasts from the transgenic type V OI mouse model displayed delayed mineralization in vitro.112 A mouse model for type V OI that was recently generated by CRISPR-Cas9 gene engineering displayed severely malformed skeleton.113 In contrast, Bril knock-out/LacZ knock-in mouse did not present any developmental and reproductive problems and did not show any appreciable mineralization defects in their skeleton (unpublished data).114

A de novo heterozygous mutation different from that causing type V OI was identified in the coding region of IFITM5 in a young adult with extreme short stature and prenatal onset of severe OI.115 The same IFITM5 mutation (c.119C>T; p.Ser40Leu) was also identified in 2 young children with prenatal onset of severe OI, confirming the relationship of genotype and phenotype.116, 117 The index patient had no hallmarks of OI type V. Instead, she had “atypical” type VI OI clinically with bone histology that showed osteoid accumulation, loss of normal lamellar orientation, and a fish-scale pattern under polarized light, along with elevated childhood serum alkaline phosphatase.115 Cultured osteoblasts and fibroblasts from the patient displayed a decreased expression of SERPINF1 and minimal secretion of pigment-epithelium derived factor (PEDF), although SERPINF1 sequences and the circulating PEDF level were normal.115 Notably, SERPINF1 expression and PEDF secretion were increased in type V OI.111

The 2 IFITM5 mutations have opposite effects on SERPINF1 expression, PEDF secretion, and osteoblast marker expression. SERPINF1 expression and PEDF secretion are increased in type V OI osteoblasts, whereas they are decreased in BRIL p.Ser40Leu osteoblasts. Since BRIL and PEDF do not appear to interact directly, this implicates a pathway connecting BRIL and PEDF which impacts bone mineralization.

IFITM5 is named as an IFITM gene mainly because of its location adjacent to the IFITM family on mouse and human chromosomes 7 and 11, respectively.118 IFITM proteins are involved in immune response signaling, germ cell maturation, and development. Despite the family name, only Ifitm1 and Ifitm3 have been shown to be regulated by interferons.119 IFITM5 protein is also known as BRIL because of its strong expression in bone tissues,118 although it has low expression in skin as well. In vitro studies provided evidence that BRIL plays a role in osteoblast differentiation and mineralization. Ifitm5 expression increases in osteoblasts until early mineralization, and declines as mineralization progresses.120 Other studies demonstrated that Ifitm5 expression increases around the time of bone nodule formation and declines when late markers of osteoblast development are expressed, such as Bglap (encoding osteocalcin).121 Perplexingly, Ifitm5−/− mice do not recapitulate the in vitro findings, having no apparent abnormality in bone development.121

BRIL had been described to occur in N-out/C-out orientation with 2 TM domains until experimental evidence was presented to show that the correct orientation has N-in/C-out topology.122 Furthermore, the Phe54 residue of human BRIL is likely to be involved in the oligomerization of BRIL molecules, as seen in other IFITM proteins. S-palmitoylation of mouse BRIL promotes its interaction with FKBP11 (FK506-binding protein 11) to form an FKBP11-CD81-FPRP/CD9 complex at the osteoblast cell membrane, which induces the expression of immunologically relevant genes.123, 124 Human BRIL is also palmitoylated on 3 cysteine residues (Cys50, Cys51, and Cys84). When overexpressed from mutant constructs, BRIL with the N-terminal MALEP addition found in type V OI is palmitoylated normally and localizes to the plasma membrane, whereas BRIL p.Ser40Leu, causing atypical type VI OI, fails to be palmitoylated on Cys50 and Cys51 and is sequestered in the Golgi. The trafficking of mutant BRIL protein in heterozygous human cells has not been reported. The possibility of intracellular signaling via BRIL remains to be investigated, as does the detailed mechanism by which IFITM5 mutations cause their distinctive OI symptoms.

Type VI OI (OMIM 613982) is an autosomal recessive form caused by homozygous or compound heterozygous null mutations in SERPINF1, located on chromosome 17p13.3 (OMIM 172860).125, 126, 127, 128, 129 SERPINF1 encodes PEDF, a 418-amino acid secreted glycoprotein well-known for its neurotrophic and antiangiogenic properties.130 Type VI OI patients have no circulating PEDF.131 Heterozygous SERPINF1-null mutant carriers have a reduced level of circulating PEDF but are asymptomatic.132 Type VI OI patients appear normal at birth or initially mild, and then develop severe and progressively deforming bone dysplasia.133 Blue sclerae, dentiogenesis imperfecta, and hearing impairment do not occur in type VI OI. Histologically, type VI OI bone is distinguished by a fish-scale pattern of bone lamellation under polarized light.133 In addition, broad seams of unmineralized osteoid and a prolonged mineralization lag time are indicative of a primary mineralization defect and a critical role for PEDF in mineralization during bone formation. The skeletal phenotype of type VI OI has been recapitulated in Serpinf1−/− mice, which displayed defective mineralization and an increased mineral-to-matrix ratio that leads to brittle bone.134 Given that the absence of PEDF impairs proper transition of osteoblast to osteocyte, PEDF is required at the early stage of mineralization.135 PEDF also influences mineralization by suppressing expression of mineralization inhibitors such as SOST.136

In addition to the SERPINF1 frameshift or nonsense mutations that lead to premature termination, 3 in-frame insertion or deletion mutations have also been reported in type VI OI cases.131, 137, 138 A 9-nucleotide duplication, adding 3 amino acids (p.91_93dup), subjects PEDF to intracellular degradation without causing ER stress; overexpression of this PEDF mutant caused decreased deposition of type I collagen and mineralization in vitro.139 A recurrent 3-nucleotide deletion causing the elimination of Phe277 (p.F277del; ΔF277), a highly conserved residue within the collagen-binding domain, affects PEDF interaction with type I collagen.138, 139 Deletion of exon 5 (ΔE5), which encodes the Lys146-Lys213 region, eliminates residues necessary for heparin binding.131, 140, 141 Both ΔF277 and ΔE5 mutations caused defective PEDF secretion and ER stress.139

Type VI OI patients respond poorly to bisphosphate therapy131 but showed improved bone mass with Denosumab, an anti-RANKL antibody.142, 143 This is consistent with PEDF induction of osteoprotegerin (OPG), a decoy receptor of RANKL, which inhibits RANKL-mediated osteoclast differentiation.144 Administration of exogenous PEDF to Serpinf1−/− mice gave rise to 2 different outcomes. When circulating PEDF was restored using an adenoviral vector, the low bone mass phenotype of Serpinf1−/− mice was not normalized.145 Interestingly, another group reported increased bone mass from intraperitoneal injection of PEDF-containing microspheres to Serpinf1−/− mice.146

PEDF is a potent anti-angiogenic factor with a collagen-binding domain that is critical for its anti-angiogenic function.147 Serpinf1−/− mice have a hypervascularization phenotype in bone, as indicated by the increase in CD31-positive endothelial cells in bone,134 consistent with the function of PEDF as a potent inhibitor of vascularization. However, it is unclear whether the hypervascularization inside bone is associated with the delayed mineralization of Serpinf1−/− mice and type VI OI (Fig 2).

PEDF binds directly to LRP6 and acts as an endogenous antagonist, blocking canonical Wnt signaling in some cell types.148, 149 However, several studies demonstrated that PEDF functions as a pro-osteogenic factor and cooperates with Wnt3a to stimulate osteogenic differentiation of mesenchymal stem cells by activating ERK1/2 and Akt.150, 151 A pro-osteogenic function of PEDF was also indicated by increased adipose tissue in Serpinf1−/− mice compared with wild-type controls.151 Intriguingly, however, a recent study identified a conserved motif in PEDF that is found in DKK1, a Wnt antagonist, suggesting that the role of PEDF may vary at different stages of osteoblast differentiation.146 Although PEDF-mediated signaling via PEDF-R (Pnpla2, a.k.a. adipose triglyceride lipase) was shown to protect osteoblasts from glucocorticoid-induced apoptosis,152 it remains to be elucidated which of those known PEDF-interacting cell surface TM proteins130, 153 plays a major role in PEDF-mediated signal transduction for bone formation and mineralization.

A homozygous mutation in SP7, located on chromosome 12q13.13 (OMIM 606633), was identified in a single child who died at 3 years of age. This bone dysplasia, originally designated type XII OI, was referred to as type XIII OI in recent reviews. An Egyptian child is homozygous for a single base pair deletion (c.1052delA) in SP7, causing a frameshift leading to a premature termination codon that would remove the third zinc finger motif of Osterix, if the shortened protein were stable.154 Clinical features of the affected individual include recurrent fractures, bone deformities, generalized osteoporosis, delayed eruption of teeth, normal teeth and hearing, and white sclera.154

The SP7 gene consists of 2 exons encoding Osterix, a 431-amino acid protein that belongs to the Sp subfamily of Sp/XKLF transcription factors.155 Osterix was originally identified as a BMP2-inducible gene.156 SP7 expression is highly restricted to immature osteochondro-progenitor cells and mature osteoblasts. Osterix regulates osteoblast differentiation and mesenchymal stem cell-mediated endochondral ossification156, 157; so, a skeletal effect would be expected from defects in SP7, although no functional data from the reported case is available to distinguish between severe osteoporosis and OI-characteristic bone tissue or biochemistry. A recent study with KLF10 (Kruppel-Like Factor 10) knockout mice, which have osteopenia, demonstrated that KLF10 directly binds to the Sp7 promoter and activates Sp7 expression on TGF-β and BMP2 stimulation in osteoblasts.158 The critical role of Osterix in bone development was also highlighted by the fact that Sp7−/− mouse embryos completely lack bone formation.156 Osterix can interact with Runx2 to coordinately induce expression of the COL1A1 gene.159 Osterix also modulates Wnt signaling. Osterix interaction with Runx2 or HIF1α inhibits Wnt signaling by inducing expression of the antagonists of the Wnt pathway, Sclerostin, and DKK1, respectively.160, 161, 162, 163 Moreover, Osterix may be a critical link between ER stress and osteoblast differentiation, since the IRE1α-XBP1 branch of the unfolded protein response (UPR) mediates osteoblast differentiation by promoting Sp7 transcription164 (Fig 3).

Homozygous nonsense mutations in WNT1 cause recessive OI type XV (OMIM 615220),165, 166, 167, 168 which presents with severe and progressive bone fragility, long bone deformity, osteopenia, short stature, and kyphoscoliosis. Heterozygous missense mutations in WNT1 were also found in individuals with early onset osteoporosis (OMIM 615221).166, 168, 169 WNT1 (wingless-type MMTV integration site family, member 1) is a secreted glycoprotein and a member of the WNT protein family, characterized by a set of 23 conserved cysteine residues that form intrachain disulfide bonds to facilitate folding and maintain WNT tertiary protein.170 Palmitoylation of WNT1 at a single serine residue (Ser224) is required for intracellular trafficking and full activity of the secreted protein.171, 172

Wnt engagement to frizzled (Fzd) and its coreceptor LRP5/6 activates canonical Wnt signaling, which in turn stabilizes β-catenin to induce the expression of Wnt target genes. As exemplified by osteoporosis-pseudoglioma syndrome, sclerosteosis, and Van Buchem disease, mutations in the genes that participate in the canonical Wnt pathway can lead to drastically different bone phenotypes, attesting to the importance of Wnt signaling in bone mass regulation.173 Activation of the Wnt pathway leads to increased bone mass and strength, whereas inhibition results in decreased bone mass.

Homozygous nonsense and frameshift mutations in WNT1 result in either nonsense-mediated mRNA decay or a truncated Wnt1 protein.166, 167, 168 Moreover, some missense mutations in WNT1 that cause recessive OI disrupt the interaction of Wnt with receptors. The WNT1 p.Gly177Cys substitution occurs at a residue highly conserved among human WNTs and disturbs the tight packing of the Wnt1 helical core, hindering its interaction with LRP5/6.166 WNT1 p.Cys143Phe and WNT1 p.Val355Phe decrease the ability of Wnt1 to stimulate canonical Wnt signaling by >90%.165 The WNT1 p.Arg235Trp substitution identified in early-onset osteoporosis disrupts the structure of Wnt1's thumb region and compromises its interaction with Fzd proteins.166

Lineage tracing experiments in transgenic mice showed Wnt1 expression in a subset of osteocytes. However, it is quite remarkable that Wnt1 expression in bone is relatively low, despite the significant impact of Wnt1 loss-of-function mutations on bone. Wnt1 is highly expressed during neuronal development and Wnt1−/− mice are embryonic lethal due to severe central nervous system (CNS) involvement.174 The Swaying mouse (Wnt1sw/sw) carries a spontaneous single nucleotide deletion in the third exon of Wnt1 that results in a premature termination; they manifest ataxia and hypertonia.175 Similarly, a boy with a homozygous nonsense mutation in WNT1 (c.990C>A, p.Cys330*) presented with severe CNS involvement in addition to an OI phenotype.176 A distinct spatiotemporal mechanism appears to play a role in the Wnt1-mediated regulation of bone homeostasis. Wnt1 was recently found to be involved in coupling bone formation and resorption.177 A study with osteoclast-specific conditional knockout mice of Tgfbr2 (Tgfr2Ocl-cKO) revealed that TGF-β signaling in osteoclasts induces expression of Wnt1 that in turn regulates osteoblast differentiation.

Homozygous missense mutations in the SPARC (secreted protein, acidic and rich in cysteine) gene were identified in 2 unrelated girls with severe bone fragility.178 The 2 affected individuals had a normal skeleton at birth and then developed severe bone fragility and presented with kyphoscoliosis, muscle hypotonia, and joint hyperlaxity. The SPARC gene is located on chromosome 5q33.1 and contains 10 exons. It encodes osteonectin, a 303-amino acid secreted, monomeric matricellular glycoprotein that binds collagen and other ECM proteins. SPARC has an acidic region at the N-terminus followed by a cysteine-rich follistatin-like domain, which binds activin, inhibin, heparin, and proteoglycans.179 The C-terminal domain of SPARC protein is an extracellular calcium-binding (EC) domain with an EF hand motif that encompasses the collagen-binding domain. SPARC recognizes the Gly-Val-Met-Gly-Phe-HyP (GVMGFO) motifs in fibrillar collagens (“O” denotes 4-hydroxyproline) of the middle and trailing collagen chains. On collagen binding, SPARC undergoes a conformational change in its EC domain, creating a deep pocket that accommodates the phenylalanine residue of the trailing collagen chain (“Phe pocket”).180 The SPARC variants that gave rise to type XVII OI phenotype (OMIM 616507) were homozygous missense changes in exon 7 (c.497G>A; p.Arg166His) and in exon 9 (c.787G>A; p.Glu263Lys).178 The SPARC residues Arg166 and Glu263 are located in EC domain and form an intramolecular salt bridge that is critical for maintaining the Phe pocket.181

SPARC was proposed to cooperate with HSP47 (encoded by SERPINH1, the type X OI gene), ensuring that only properly folded procollagen molecules exit the ER.182 Supporting SPARC activity as an intracellular type I collagen chaperone, cells with the SPARC p.Arg166His mutation displayed a defect in collagen secretion.178 Reduction in collagen fiber size and quantity and increased activity of transglutaminase in Sparc−/− mice indicates that SPARC regulates crosslinking of ECM proteins by transglutaminase enzymes.183 In bone, SPARC is expressed by osteoblasts. Sparc−/− mice develop progressive osteoporosis, due to a defect in bone formation.184 The severe early-onset scoliosis phenotype of the individual with SPARC p.Arg166His was consistent with the disk degeneration developed by Sparc−/− mice.

Recently, we identified the first X-linked recessive form of OI. Novel missense mutations in MBTPS2 were identified in 2 independent OI pedigrees with an apparent X-linked inheritance.185 Affected males have moderate to severe OI with pre- and post-natal fractures of long bone and ribs, low bone mass (i.e. DXA Z-score = −4.7), bone deformity, scoliosis, and thoracic deformity. They have white-to-bluish sclerae and normal teeth. The clinical presentation of MBTPS2 mutations is easily distinguished from that of mutations in plastin 3, which causes an X-linked osteoporosis without prenatal fractures and without the extraskeletal findings of OI. Defects in plastin 3, which is involved in the formation of F-actin bundles, cause a low turnover osteoporosis without hypermineralization of bone matrix.186, 187

The MBTPS2 gene is located on chromosome Xp22.11-p22.13 and encodes a 519-amino acid membrane-bound zinc metalloprotease, site-2 protease (S2P), consisting of 6 TM domains, with both the N- and C-termini facing the cytosol.188 S2P is a critical component of RIP. RIP is an evolutionarily conserved mechanism of signal transduction that facilitates communication between cellular compartments.189 In the well-studied class 2 RIP, regulatory proteins are transported from the ER membrane to the Golgi membrane in times of cell stress or sterol metabolite depletion. In the Golgi membrane, these regulatory proteins are sequentially cleaved by S1P, a serine protease encoded by MBTPS1 (membrane-bound transcription factor peptidase, site 1), and then by S2P, releasing the N-terminal portion of the regulatory protein as an active factor.190 This fragment translocates to the nucleus, where it activates expression of genes for cellular differentiation, UPR, and lipid metabolism.189 Among the ER membrane–bound transcription factors that require 2-step cleavage by RIP for activation are OASIS (old astrocyte specifically induced substance)/CREB3L1 (cyclic amp responsive element binding protein 3-like 1), ATF6 (activating transcription factor 6), and SREBPs (sterol regulatory element binding proteins).

Mutations in MBTPS2 were previously identified in cholesterol-based conditions—ichthyosis follicularis with alopecia and photophobia (IFAP) syndrome (OMIM 308205), a rare X-linked oculocutaneous genodermatosis, and KFSD (keratosis follicularis spinulosa decalvans) (OMIM 308800).191, 192, 193 The IFAP and KFSD mutations are scattered throughout S2P but the most severe phenotypes are associated with mutations near the active site and are more detrimental to its enzymatic function.194 However, individuals with X-linked recessive OI do not have any symptoms associated with IFAP/KFSD, nor do IFAP/KFSD patients have symptoms of OI. The novel MBTPS2 missense mutations causing X-linked recessive OI, MBTPS2 (c.1376A>G; p.Asn459Ser) and MBTPS2 (c.1515G>C; p.Leu505Phe), substitute residues in or near the Asn-Pro-Asp-Gly (NPDG) motif of S2P that is crucial for Zn ion coordination.195 The mutant S2P proteins causing X-linked OI have significantly impaired cleavage of RIP substrates, OASIS, ATF6, and SREBPs. X-linked OI is also collagen-related, the detailed mechanism of which remains to be elucidated. Collagen extracted from the bone tissue with X-linked OI has about half the normal level of hydroxylation of the α1(I) and α2(I) Lys87 residue critical to collagen crosslinking in matrix. This alteration of collagen crosslinking undermines bone strength. Furthermore, osteoblasts with mutant S2P have defective differentiation and reduced collagen secretion. These novel MBTPS2 mutations underscore the importance of a pathway known to be critical for lipid management in normal bone development.

There is additional evidence that components of RIP impact skeletal development and cholesterol metabolism.196 For S1P, the zebrafish mutant gonzo exhibited phenotypes similar to human chondrodysplasias.197 Similarly, cartilage-specific S1P knockout mice displayed severe chondrodysplasias and complete lack of endochondral ossification.198 Although S1P deficiency did not impair osteoblastogenesis itself, as indicated by normal expression of Runx2, loss of S1P in chondroprogenitor cells uncoupled endochondral bone formation from cortical bone formation.198 The recent finding of X-linked OI with MBTPS2 mutations in patients establishes the significance of the RIP pathway for skeletal development.

A homozygous genomic deletion of CREB3L1 was identified in a Turkish family in which 2 siblings had severe OI. The first sibling developed multiple fractures of the extremities, and the second sibling, terminated at 19 weeks gestation, showed thin ribs and fractures at bowed humerus and femora.199 CREB3L1 on chromosome 11p11.2 encodes OASIS, a 519-amino acid transcription factor that is one of the substrates for RIP. OASIS belongs to the CREB/ATF family and contains an N-terminal transcription activation domain and a central basic leucine zipper (bZIP) domain followed by a putative TM domain.200 Before RIP activation, OASIS is inserted into ER membranes with N-terminal transcription activation domain and bZIP domain oriented toward the cytosol. After RIP processing in the Golgi membrane, the released OASIS bZIP domain translocates into the nucleus, where it activates the transcription of UPR genes by binding to box-B elements.201, 202 OASIS is a relatively unstable protein and is regulated by ubiquitination by HMG-CoA Reductase Degradation homolog 1 (HRD1), an ER resident E3 ligase that is involved in degradation of unfolded proteins during ER stress.203 Interaction of OASIS with HRD1 is disturbed by ER stress.

OASIS−/− mice are osteopenic and display growth retardation. The osteopenic phenotype is rescued by crossing the knockout mice with OASIS transgenic mice, which overexpress OASIS under the control of a 2.3-kb osteoblast-specific col1a1 promoter.204 But growth retardation is not normalized by this approach, suggesting that OASIS regulates skeletal development by osteoblast-dependent and -independent mechanisms.204 In both dermal fibroblasts of patients with the CREB3L1 homozygous deletion and in fibroblasts of OASIS−/− mice, type I collagen production was normal.199, 205 However, expression of col1a1 and col1a2 was reduced and a decreased amount of type I collagen was detected in osteoblasts and bones of OASIS knockout mice, suggesting a bone-specific impact of OASIS deficiency.205 When induced by BMP2, OASIS activates col1a1 expression in osteoblasts.205 OASIS also interacts with HIF1α through its hypoxia response element and modulates signaling to regulate vascularization during bone development.206 These findings suggest that OASIS joins other OI-causative proteins, such as PEDF, that connect bone formation and bone vascularization (Fig 2, Fig 4).

Section snippets

Acknowledgments

Conflicts of Interest: All authors have read the journal's policy on disclosure of potential conflicts of interest and have none to declare.

Joan C. Marini is supported by intramural funding from the NICHD, NIH.

The authors thank Jeremy Swan and Nichole Jonas of The Unit on Computer Support Services, Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) for preparing the figures.

References (206)

  • L.M. Ward et al.

    Osteogenesis imperfecta type VII: an autosomal recessive form of brittle bone disease

    Bone

    (2002)
  • N. Fratzl-Zelman et al.

    CRTAP deficiency leads to abnormally high bone matrix mineralization in a murine model and in children with osteogenesis imperfecta type VII

    Bone

    (2010)
  • D.J. Wassenhove-McCarthy et al.

    Molecular characterization of a novel basement membrane-associated proteoglycan, leprecan

    J Biol Chem

    (1999)
  • W.A. Cabral et al.

    A founder mutation in LEPRE1 carried by 1.5% of West Africans and 0.4% of African Americans causes lethal recessive osteogenesis imperfecta

    Genet Med

    (2012)
  • J.A. Vranka et al.

    Prolyl 3-hydroxylase 1 null mice display abnormalities in fibrillar collagen-rich tissues such as tendons, skin, and bones

    J Biol Chem

    (2010)
  • E. Pokidysheva et al.

    Prolyl 3-hydroxylase-1 null mice exhibit hearing impairment and abnormal morphology of the middle ear bone joints

    Matrix Biol J Int Soc Matrix Biol

    (2013)
  • N. Fratzl-Zelman et al.

    Bone matrix hypermineralization in prolyl-3 hydroxylase 1 deficient mice

    Bone

    (2016)
  • G. Jansen et al.

    An interaction map of endoplasmic reticulum chaperones and foldases

    Mol Cell Proteomics MCP

    (2012)
  • F.S. van Dijk et al.

    PPIB mutations cause severe osteogenesis imperfecta

    Am J Hum Genet

    (2009)
  • Y. Ishikawa et al.

    Mutation in cyclophilin B that causes hyperelastosis cutis in American Quarter Horse does not affect peptidylprolyl cis-trans isomerase activity but shows altered cyclophilin B-protein interactions and affects collagen folding

    J Biol Chem

    (2012)
  • B. Steinmann et al.

    Cyclosporin A slows collagen triple-helix formation in vivo: indirect evidence for a physiologic role of peptidyl-prolyl cis-trans-isomerase

    J Biol Chem

    (1991)
  • M. Terajima et al.

    Cyclophilin-B modulates collagen cross-linking by differentially affecting lysine hydroxylation in the helical and telopeptidyl domains of tendon type I collagen

    J Biol Chem

    (2016)
  • M.J. Berridge

    Inositol trisphosphate and calcium signalling mechanisms

    Biochim Biophys Acta

    (2009)
  • E. Rubinato et al.

    A novel deletion mutation involving TMEM38B in a patient with autosomal recessive osteogenesis imperfecta

    Gene

    (2014)
  • J.R. Macdonald et al.

    HSP47 binds cooperatively to triple helical type I collagen but has little effect on the thermal stability or rate of refolding

    J Biol Chem

    (2001)
  • H.E. Christiansen et al.

    Homozygosity for a missense mutation in SERPINH1, which encodes the collagen chaperone protein HSP47, results in severe recessive osteogenesis imperfecta

    Am J Hum Genet

    (2010)
  • Y. Ishikawa et al.

    The rough endoplasmic reticulum-resident FK506-binding protein FKBP65 is a molecular chaperone that interacts with collagens

    J Biol Chem

    (2008)
  • Y. Alanay et al.

    Mutations in the gene encoding the RER protein FKBP65 cause autosomal-recessive osteogenesis imperfecta

    Am J Hum Genet

    (2010)
  • E.D. Setijowati et al.

    A novel homozygous 5 bp deletion in FKBP10 causes clinically Bruck syndrome in an Indonesian patient

    Eur J Med Genet

    (2012)
  • R. Shaheen et al.

    FKBP10 and Bruck syndrome: phenotypic heterogeneity or call for reclassification?

    Am J Hum Genet

    (2010)
  • G. Venturi et al.

    A novel splicing mutation in FKBP10 causing osteogenesis imperfecta with a possible mineralization defect

    Bone

    (2012)
  • D.R. Eyre et al.

    Quantitation of hydroxypyridinium crosslinks in collagen by high-performance liquid chromatography

    Anal Biochem

    (1984)
  • A. Forlino et al.

    New perspectives on osteogenesis imperfecta

    Nat Rev Endocrinol

    (2011)
  • J.C. Marini et al.

    Osteogenesis imperfecta due to mutations in non-collagenous genes: lessons in the biology of bone formation

    Curr Opin Pediatr

    (2014)
  • D.A. Redford-Badwal et al.

    Nuclear retention of COL1A1 messenger RNA identifies null alleles causing mild osteogenesis imperfecta

    J Clin Invest

    (1996)
  • M.C. Willing et al.

    Premature chain termination is a unifying mechanism for COL1A1 null alleles in osteogenesis imperfecta type I cell strains

    Am J Hum Genet

    (1996)
  • D.J. Prockop

    Mutations in collagen genes as a cause of rare and perhaps common diseases of connective tissue

    Acta Paediatr Scand Suppl

    (1991)
  • D.J. Prockop et al.

    Heritable diseases of collagen

    N Engl J Med

    (1984)
  • D.F. Schorderet et al.

    Analysis of CpG suppression in methylated and nonmethylated species

    Proc Natl Acad Sci U S A

    (1992)
  • J.C. Marini et al.

    Consortium for osteogenesis imperfecta mutations in the helical domain of type I collagen: regions rich in lethal mutations align with collagen binding sites for integrins and proteoglycans

    Hum Mutat

    (2007)
  • T.E. Uveges et al.

    Cellular mechanism of decreased bone in Brtl mouse model of OI: imbalance of decreased osteoblast function and increased osteoclasts and their precursors

    J Bone Miner Res

    (2008)
  • H. Fernandes et al.

    The role of collagen crosslinking in differentiation of human mesenchymal stem cells and MC3T3-E1 cells

    Tissue Eng A

    (2009)
  • R.A. Bank et al.

    Pyridinium cross-links in bone of patients with osteogenesis imperfecta: evidence of a normal intrafibrillar collagen packing

    J Bone Miner Res

    (2000)
  • L. Bianchi et al.

    Altered cytoskeletal organization characterized lethal but not surviving Brtl+/− mice: insight on phenotypic variability in osteogenesis imperfecta

    Hum Mol Genet

    (2015)
  • E. Daley et al.

    Variable bone fragility associated with an Amish COL1A2 variant and a knock-in mouse model

    J Bone Miner Res

    (2010)
  • T.S. Lisse et al.

    ER stress-mediated apoptosis in a new mouse model of osteogenesis imperfecta

    PLoS Genet

    (2008)
  • F. Malfait et al.

    Helical mutations in type I collagen that affect the processing of the amino-propeptide result in an osteogenesis imperfecta/Ehlers-Danlos Syndrome overlap syndrome

    Orphanet J Rare Dis

    (2013)
  • S. Symoens et al.

    Type I procollagen C-propeptide defects: study of genotype-phenotype correlation and predictive role of crystal structure

    Hum Mutat

    (2014)
  • K. Lindahl et al.

    COL1 C-propeptide cleavage site mutations cause high bone mass osteogenesis imperfecta

    Hum Mutat

    (2011)
  • V. Martinez-Glez et al.

    Identification of a mutation causing deficient BMP1/mTLD proteolytic activity in autosomal recessive osteogenesis imperfecta

    Hum Mutat

    (2012)
  • Cited by (0)

    View full text