Chloride Channel ClCN7 Mutations Are Responsible for Severe Recessive, Dominant, and Intermediate Osteopetrosis†
The authors have no conflict of interest.
Among 94 osteopetrotic patients presenting with a severe clinical picture and diagnosed early in life, 12 bore mutations in the ClCN7 gene, but only 7 of them had the expected two recessive mutations. The remaining five patients seem to be heterozygous for a ClCN7 mutation, and significant variations were observed in the clinical manifestations of their disease, even within the same family.
Introduction: Human osteopetroses are a heterogeneous group of diseases that include both infantile severe, autosomal recessive (ARO) and adult autosomal dominant (ADO) forms. Two genes, Atp6a3 (TCIRG1) and ClCN7, have been shown to be associated with human ARO, the latter of which is also thought to be responsible for ADO-II. However, patients with an intermediate phenotype have been described: the genetic basis of these observances is unknown.
Materials and Methods: In this study, we report the clinical and molecular analysis of 94 patients in which a diagnosis of severe osteopetrosis was made within the first 2 years of age. Both TCIRG1 and CLCN7 genes were sequenced in all patients and the molecular findings were correlated to clinical parameters.
Results and Conclusions: In 56 of 94 patients with a classical picture of ARO, TCIRG1-dependent recessive mutations were found. In contrast, ClCN7 mutations were found in 12 cases (13%) of severe osteopetrosis, but only 7 of them had two recessive mutations identified: in 6 of these 7 cases, central nervous system manifestations were noted, and these patients had a poor prognosis. The remaining five cases were heterozygous for a ClCN7 mutation, including two brothers from a large family with a history of ADO-II in which the presence of a second ClCN7 mutation was formally excluded. Despite an early and severe clinical presentation, these five patients all reached adulthood, suggesting that the degree of dominant interference with chloride channel function can vary widely. Our findings suggest that recessive ClCN7-dependent ARO may be associated with CNS involvement and have a very poor prognosis, whereas heterozygous ClCN7 mutations cause a wide range of phenotypes even in the same family, ranging from early severe to nearly asymptomatic forms. These findings have prognostic implications, might complicate prenatal diagnosis of human osteopetroses, and could be relevant to the management of these patients.
Infantile malignant autosomal recessive osteopetrosis (ARO; OMIM 259700) is a severe bone disease with a fatal outcome generally within the first decade of life. It results from a deficit in bone resorption caused by osteoclast malfunction or formation and is associated with a number of severe clinical manifestations, including macrocephaly, cranial nerve dysfunction such as deafness and blindness, hepatosplenomegaly, and severe bone marrow failure beginning in early infancy or in fetal life. Deafness and blindness are generally thought to represent effects of pressure on cranial nerves exerted by the overgrowth of bone, but the possibility of a primary defect in sensory cells has also been raised.(1) Forms of osteopetrosis associated with neurologic deficits have also been described,(2–4) although it is not yet clear whether these forms represent a variant of the same disorder or distinct diseases.
Autosomal dominant osteopetrosis (ADO) has also been described and generally is associated with less severe clinical features. It is often subdivided clinically into two forms, type I (ADO-I) and type II (ADO-II, OMIM 166600).(5–7) An intermediate form has also been described (OMIM 259710): the presence of several affected sibs in some families with intermediate osteopetrosis suggests an autosomal recessive inheritance,(8) while in other reports, the pattern of inheritance has not been established. Although the clinical picture is less severe than that of ARO, some of these patients are diagnosed in infancy and have significant clinical findings; therefore, for these forms, the term “mild” does not seem to be appropriate.
In the last few years, cellular and molecular investigations of hereditary osteopetroses have greatly contributed to the dissection of this heterogeneous disease. Normal bone homeostasis is a dynamic and complex process that relies on a delicate balance between the activity of osteoblasts and osteoclasts. Osteoclasts are multinucleated giant cells that are derived from hematopoietic precursors and are responsible for bone resorption. To achieve normal bone resorption, at least two processes seem to be important for the maintenance of osteoclast function: (1) the establishment of a specialized membrane structure, the so-called ruffled membrane, which is an area of intense membrane infolding thought to be actively involved in bone resorption,(9–11) and (2) the production of a low pH microenvironment at the contact surface between the cell and the bone matrix, the so-called resorption lacuna. Osteoclasts adhere to the bone surface, seal the resorption lacuna, and activate a specific membrane-bound V-ATPase complex, generating the acid environment necessary to dissolve the mineral component of bone, primarily composed of hydroxyapatite crystals. Subsequent release of lysosomal enzymes, including cathepsin K, eventually leads to organic matrix degradation.
Human osteopetroses in most, but not all, cases(12) show normal or increased numbers of osteoclasts, and it has been suggested that in humans, the affected gene(s) is more likely to relate to the maintenance of the functional capacity of mature osteoclasts. This suggestion has been supported by the fact that the gene defects found thus far in ARO and ADO-II, as well as the one found in osteopetrosis associated to renal tubular acidosis (OMIM 259730), are all involved in the metabolic pathways leading to acidification of the external microenvironment. The a3 subunit (ATP6a3 or TCIRG1) of the V-ATPase complex, the primary pump involved in proton extrusion by osteoclast cells, has been found mutated in more than 50% of the patients affected by ARO.(13–15) The ClCN7 gene, which provides the chloride conductance required for an efficient proton pump, has recently been implicated in ADO-II,(16) but has also been shown to be mutated in two cases of ARO.(1,16) The clinical implications of these molecular findings have not yet been investigated.
We have collected a large series of patients with classical ARO and patients with less severe clinical courses. In this study, we report the clinical and molecular analysis of this large series, in which the molecular findings were related to the clinical presentation and course, with the aim of investigating whether the particular gene involved and/or the specific mutation identified are of clinical relevance with regard to diagnosis, presentation, or prognosis.
MATERIALS AND METHODS
As part of an international study, we have collected material from 94 unrelated patients with a clinical diagnosis of severe osteopetrosis made based on the following criteria at presentation: early diagnosis (before 2 years of age), radiographic evidence of markedly increased bone density, pancytopenia, hepatosplenomegaly, and cranial nerves defects.
The presence of all the abnormalities was required for entry in the study, although in a few cases (patient 6 for the cranial defect and patients 8 and 12 for blood count), one single criterion was not considered because the specific data were not available.
Samples were obtained from Italy, the United States, Canada, Turkey, Costa Rica, Yugoslavia, Croatia, Switzerland, Sweden, Germany, Israel, the Netherlands, United Kingdom, Belgium, and France. Specimens include DNA, frozen peripheral blood cells, fibroblasts or Epstein-Barr virus (EBV)-transformed lymphoblast cell lines from patients, and DNA samples from their parents; they were collected with informed consent. Clinical, radiological, and laboratory data were collected for genotype/phenotype correlation studies. One additional case of classical ADO-II (case 13) suggested by multiple spontaneous fractures and X-ray high bone density is also reported.
ClCN7 and TCIRG1 gene mutation analysis
The ClCN7 gene was amplified at both genomic (genomic sequences accession no. AL031600 and AL031705) and RNA (cDNA sequence accession no. AF224741) levels. The strategy and primers are shown in Table 1.
Table Table 1. Primers Used for Amplification of Genomic or Transcript C1CN7 Gene
All the reactions were performed in 25 μl of final volume with 0.4 U Taq polymerase, 1.5 mM MgCl2, 150 μM dNTP, 10 pmol of each oligonucleotide primer, and 20 ng of purified DNA. The thermocycling conditions used for amplification consisted of an initial denaturation step at 94°C for 3 minutes, followed by 34 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and 72°C for 30 s.
Sequencing was performed directly on the polymerase chain reaction (PCR) products purified from the gel using the dideoxynucleoside chain termination method (USB).
The organization of the ClCN7 protein and locations of the transmembrane helices were predicted with several bioinformatics tools. Related sequences were searched from databases and used in multiple sequence alignment. In the case of missense mutations (G240R, P249R, R286Q, M332V, L490F, R526W, L614P, G677V, S744F, R767Q) not previously described by Cleiren et al.,(16) 100 chromosomes from normal unrelated donors were also investigated by direct sequence analysis or enzymatic digestion.
Mutation analysis of the TCIRG1 gene was performed as previously described.(15)
The mutations and patient information were collected and put into a database (http://bioinf.uta.fi/TCIRG1base/).
We analyzed the TCIRG1 and ClCN7 genes in 94 patients clinically diagnosed with severe osteopetrosis; in 56 (60%), defects were identified in the TCIRG1 gene. Of these, 53 had mutations on both alleles, whereas 3 had mutations in only one allele; these 3 patients have previously been reported.(15) In 12 families (13%), mutations were documented in the ClCN7 gene. In 26 cases (28%), no defects were identified in either the TCIRG1 or ClCN7 gene. In 6 of these 26 patients, whose parents were consanguineous, a whole genome search with approximately 300 polymorphic markers was performed to identify regions of homozygosity. This allowed us to formally exclude the TCIRG1 and CLCN7 genes as playing a role in the pathogenesis of the disease in these patients, because no homozygosity was found in the regions harboring these two genes, except in one case who was homozygous in the TCIRG1 region (A Frattini and G Casari, unpublished findings, 2002). These findings confirm that at least a third gene is involved in the pathogenesis of human ARO.
The clinical history of all the ClCN7 mutated cases is described below and summarized in Table 2.
Table Table 2. Clinical Features of C1CN7-Dependent Osteopetrotic Patients
Cases 1–7 are from unrelated families of various ancestry, in which parents were unaffected. They presented with a severe clinical picture compatible with autosomal recessive osteopetrosis. In 6 of the 7 families, central nervous system abnormalities were observed radiographically or clinically, including cerebral atrophy (cases 1 and 3), macrocephaly with neurological impairment (case 4) or hydrocephalus (case 6), autism (case 5), and spinocerebellar ataxia (case 2). In the seventh family, two affected siblings were evaluated (case 7a, male; case 7b, female). The brother suffered his first fracture at 1 year of age, and at 24 years of age, he has vision impairment, mandibular osteomyelitis, anemia, thrombocytopenia, and splenomegaly; he has had a total of approximately 80 fractures. A bone biopsy showed multiple activated osteoclasts. In contrast to her brother, the affected sister, at 31 years of age, has anemia, but has not developed transfusion dependence; she has optic atrophy and visual impairment and has had approximately 100 fractures.
In cases 8–12, the early and severe presentation of the probands suggested an ARO diagnosis, but in one case, this occurred within a family with classical ADO-II (family 8). In this family, two children (cases 8a and 8b) were diagnosed as infants. Several members of the family, including their mother and maternal grandmother, were affected with a dominant form of the disease with multiple bone fractures only. In addition to bone fractures, both siblings developed severe optic nerve atrophy secondary to optic canal narrowing, requiring neurosurgical intervention. The sister, who is 4 years older than the brother, was surgically treated at 2 years of age and did not fare as well as the brother, who underwent surgery at a younger age (11 months) because of his sister's medical course. The sister, now 17 years of age, is legally blind. The brother, at the age of 13 years, is legally blind in the left eye and has 20/20 visual acuity of the right eye. No other neurological defects have been reported. The severity of their clinical presentation is somewhat unusual for an ADO-II diagnosis.
Case 9 was diagnosed as an infant with an abnormal complete blood count and signs of optic nerve compression and has had numerous fractures. The parents have no history of an increased likelihood of fractures. He is now 10 years old, blind, and growth retarded, but his blood counts are normal.
Case 10 was diagnosed with severe osteopetrosis in early infancy with blindness, severe anemia, and hepatosplenomegaly. A frequent transfusion requirement and the development of hemochromatosis necessitated splenectomy at 14 years of age. He developed a partial palsy of the VII and the VIII cranial nerve and severe obstruction of the carotid and vertebral arteries with extreme collateral formation. He had dental impaction and multiple episodes of osteomyelitis of the mandible, requiring numerous stomatologic interventions. He had several pathological fractures. At last visit, he was 26 years old and of normal intelligence. Chronic headaches are a major problem. A bone biopsy revealed an increased number of multinucleated osteoclasts.
Case 11a presented with anemia and thrombocytopenia, partial compression of the optic nerve canal, and radiological signs of osteopetrosis. A diagnosis of autosomal recessive OP was suspected, and she underwent bone marrow transplantation (BMT). However, retrospective examination of his parents (performed after the birth of the proband) showed that the father (case 11b) had a radiological picture of bone involvement and a history of two fractures, both related to trauma. Case 11b is now 4 years old, in very good health, and is able to write, read, and speak two languages.
Case 12 presented as an infant with optic nerve compression and increased bone density. A clinical diagnosis of ARO was made. Her parents and one sister were not affected. She did not undergo BMT, and she is now 30 years old and blind. She experienced many fractures.
Molecular analysis was also performed on a case with classical ADO-II (case 13), a female who was diagnosed at 8 years of age by X-ray after a metatarsal fracture sustained during a sporting event. She has had one additional fracture since then but no other symptoms. There is no positive family history.
ClCN7-dependent recessive osteopetrosis
In our analysis, we found seven families in which ClCN7 mutations were identified in both alleles. These data are summarized in Table 3. Two patients (cases 1 and 2), were born from consanguineous parents and were homozygous for the described mutation, whereas the other five were compound heterozygotes (patients 3–7). Patient 1 bore a single T nucleotide insertion, causing a frameshift and a premature stop codon (D145X), whereas patient 2 had a single nucleotide change, causing a missense mutation (R767Q) in the COOH-terminal portion of the protein. For both patients, the same mutation was found at the heterozygous level in their parents.
Table Table 3. Molecular Findings in C1CN7-Dependent Osteopetrotic Patients
Patient 3 had two single nucleotide changes, causing two missense mutations (G240R and R526W), the first being present in the mother and the second in the father. Patient 4 had a single nucleotide change causing a missense mutation (L614P) in one allele and a genomic deletion involving exon 17 on the second allele (parent DNA was not available). Patient 5 had a single nucleotide change, causing a nonsense mutation (E374X) in the first allele and a nucleotide change in intron 10, creating a new splice site causing an in frame insertion of 54 nucleotides in ClCN7 transcript, starting after codon G306. No parent DNA was available. Patient 6 had two missense mutations, M332V and R767W. The molecular findings in these six patients are in complete agreement with the clinical diagnosis of severe ARO and with the fact that all their parents were asymptomatic.
In family 7, two siblings were affected (7a and 7b). As shown in Table 3, both were compound heterozygotes for two different single nucleotide changes, causing two missense mutations, a maternally derived P249R and a paternally derived S744F. This latter mutation was also shared by a healthy brother. Therefore, the family history suggests that both the mutations are necessary for the clinical manifestation of osteopetrosis, although the possibility of incomplete penetrance cannot be formally excluded.
ClCN7-dependent osteopetrosis with a single mutated allele
Six patients from five families (cases 8–12) were found to carry a single ClCN7 mutation, all represented by a single nucleotide change causing missense mutations in cases 8–11 and a stop mutation in case 12 as shown in Table 3.
Patients 8a and 8b (brother and sister) were both heterozygous for a missense mutation at codon 677 (G677V). Many members of the maternal ancestry were affected by dominant osteopetrosis on clinical grounds, because they reached adulthood and had children. To explain the severity of the presentation in the third generation, it could be hypothesized that the father, in whose family there is no history of osteopetrosis, could have provided a recessive mutation; in this case, both the severely affected siblings from the third generation could have inherited the same paternal chromosome. We tested this possibility by analyzing several single nucleotide polymorphisms (SNPs) in the ClCN7 gene in the affected siblings (A Frattini, unpublished data, 2003); the analysis demonstrated that they inherited two different paternal chromosomes, ruling out an additional ClCN7 mutation as the cause of the worsening of the clinical picture in family 8.
Patients 9 and 10 was heterozygous for a G215R and L490F mutation, respectively.
In case 11a, a putative diagnosis of ARO was suggested, and BMT was performed. However, after the diagnosis was made in this child, bone abnormalities consistent with ADO were found in her father, while the mother was normal. The affected child showed a G215R mutation on only one allele, and interestingly, her father and her asymptomatic paternal grandmother carried the same amino acid change. Three additional changes, a V418M and two silent changes (T19233C and C12974T), were found in the proband, her mother, and her maternal grandmother, but these changes were shared by many normal controls. In family 12, a diagnosis of ARO was established at 6 months, although no BMT was performed. However, an extensive search identified only one maternally derived mutated allele, bearing a nonsense mutation (W180X). No mutation was found in the father and her sister.
Patients 13, as expected in consideration of her ADO diagnosis, was heterozygous for a single R286Q mutation.
Polymorphic repeat does not account for variation in the osteopetrotic phenotype
Because the clinical picture in the last generation in several families seems to be more severe than in parents, we hypothesized a possible role of the polymorphic repeat present in intron 8.(16) In our normal population, we found from four to seven repeats, but no increase in the number of the repeats was seen from parents to siblings. Therefore, it is unlikely that an elongation of these repeats is responsible for the severity of the disease.
Despite the fact that, in murine studies, inactivation of many genes involved in both differentiation and function of osteoclasts give rise to severe osteopetrosis,(17) the genes found mutated in human osteopetroses affect a single aspect of mature osteoclast, specifically the ability to generate acidification in the extracellular environment.(1,13,18) While mutations in carbonic anhydrase are associated to a renal defect, recessive abnormalities in TCIRG1 and ClCN7 genes give rise to relatively homogeneous pictures of severe ARO. However, the data presented here, together with finding that ClCN7, but not TCIRG1, is also responsible for ADO, suggest that the spectrum of the disease observed with defects within these two genes is different.
TCIRG1-dependent mutations reported thus far seem to be limited to severe ARO. All TCIRG1-ARO patients analyzed by us and described in the literature are diagnosed early in infancy and present with classical ARO clinical picture. Most TCIRG1 mutations are genetically null, consisting of nonsense, deletion, and splicing abnormalities.(15) For this reason we have hypothesized that hypomorphic TCIRG1 mutations could also cause less severe (intermediate) osteopetrosis cases, but we were unable to find such mutations in patients with intermediate or dominant osteopetrosis.
In contrast, the spectrum of disease caused by ClCN7 mutations is wider. Cleiren et al. have clearly demonstrated that adult-onset ADO is caused by heterozygous missense mutations in the ClCN7 gene.(16) They also reported a single patient with a homozygous missense mutation and a typical picture of ARO. A clinically similar patient who was a compound heterozygote for a nonsense and a missense mutation was also reported by Kornak et al.(1) Our data extend these reports by documenting that ClCN7-dependent ARO represents a substantial fraction of ARO, accounting for about 13% of patients presenting with a severe clinical picture.
Interestingly, in six of seven of the families in which ARO was documented to be caused by two ClCN7 mutations, central nervous system abnormalities were noted. In two of these instances, the patients were survivors of BMT, and complications of the transplant process could account for these changes in whole or in part. In addition, it is difficult to distinguish primary neurologic changes related directly to the ClCN7 gene defects and secondary effects caused by skull abnormalities affecting the flow of cerebral spinal fluid or sequelae of chronic airway obstruction, which has been observed in individuals with ARO.(19) Nevertheless, our data suggests that ARO caused by ClCN7 defects should be evaluated routinely for neurologic findings, both clinically and radiologically. It is possible that this subset could overlap with a previously recognized osteopetrosis form associated with infantile neuroaxonal dystrophy (OMIM 600329). The implications of this issue for transplantation are clear; whereas improvement in the bony changes of osteopetrosis was documented after transplant in at least one patient (patient 5), the central nervous system abnormalities directly related to ClCN7 mutations cannot be expected to be corrected by BMT. Therefore, information regarding this issue could potentially be an important variable in determining the prognosis for individuals with ClCN7 defects. Only prospective studies closely correlating the genotype of osteopetrotic patients and their clinical course will provide the necessary data to address this question.
In addition, our data suggest that ClCN7 mutations are also responsible for intermediate forms. The relationship of these forms with both ARO and ADO is not straightforward, because both dominant and recessive inheritance has been suggested.(20,21) Our findings suggest that at least some of the intermediate forms represent a continuum with both ARO and ADO. ADO is sometimes referred to as benign form, with the diagnosis often being made incidentally by X-rays after the occurrence of fractures in adult age. However, some patients are diagnosed at earlier ages because of the presence of blindness, fractures, and/or anemia; these more severe phenotypes can coexist with benign ADO cases within the same family.(22) The data reported in our studies contribute to the clarification of these questions. In the original families described as intermediate forms, a recessive transmission was inferred because of the presence of several siblings in the same family in the presence of asymptomatic parents: the existence of several siblings was thought to rule out the occurrence of de novo dominant mutations (see OMIM 259710). However, our data suggest that dominant ClCN7 mutations can be responsible for a range of presentations wider than previously appreciated and that the same mutation can cause a broad range of phenotypes even in the same family, because we found mildly affected or even asymptomatic patients together with others showing severe symptoms, including anemia and pancytopenia, which are usually associated with ARO. A dominant pattern of transmission with incomplete penetrance and/or variable expressivity is likely to be the best explanation. The alternative explanation, mentioned by Cleiren et al.,(16) could be that an additional recessive mutation from the second parent contributed to the different phenotypes. Although the possibility exists that additional mutations were not detected by our analysis, it is noteworthy that severe cases with a single ClCN7 mutation do not show any primary neurological involvement and live longer, with a consequent better prognosis than classical recessive ClCN7-dependent patients (see patients 8–12). In addition, polymorphism analysis performed in family 8 allowed us to show that the two affected brothers inherited a different chromosome from their unaffected father, ruling out a contribution of the second allele to their phenotype. Finally, phenotype variability within ADO families, including several with asymptomatic obligate carriers,(23) is a common event, and given the low frequency of recessive alleles, it would be difficult to postulate that undetected mutations are involved in all these cases.
Further complexity is added by the finding that the expressivity of a specific ClCN7 mutation is variable, as exemplified by the phenotype variability found between families. An R767W mutation has been reported in an ADO white patient by Cleiren et al.,(16) while we found the same mutation in a recessive white proband whose parents where phenotypically normal (case 6). In addition, a different homozygous mutation in the same codon (R767Q) was found in a recessive Pakistani patient whose parents were again asymptomatic (case 2). The same G215R mutation was the only one found in families 9 and 11, as well as in families described by Cleiren et al.(16) However, the clinical picture varied from being relatively severe in probands 9 and 11a to the father and the grandmother of proband 11a who were diagnosed only after her birth. In family 8, two siblings presented with a relatively severe picture, clearly different from those found in several members of the same family, and finally, in family 12, the same mutation was found in the asymptomatic mother and her severely affected daughter.
The factors involved in this phenotypic variability remain unknown. It is likely that this variability is caused by other unknown loci or to complex interactions with genes regulating chloride channel function or involved in osteoclast metabolism. One such candidate is the TCIRG1 gene itself, which functions in a metabolic pathway interacting with that of ClCN7. However, a normal TCIRG1 coding sequence was found in all ClCN7-dependent patients. Alternatively, this variability could be caused by different levels of ClCN7 expression associated with promoter polymorphisms, a possibility that is worthy of further investigation.
In conclusion, our findings contribute to further dissection of the heterogeneity of ARO, documenting that ClCN7-dependent ARO constitutes a substantial portion of severe cases, including a number in which central nervous system findings were observed. In addition, in contrast to what has been observed with the TCIRG1 gene, whose mutations have been reported only in classical ARO, dominant ClCN7 mutations are responsible for a continuum of symptoms and can cause both classical ADO and intermediate/severe cases of osteopetrosis. Although further studies may establish a relation between the specific protein mutants and the pattern of inheritance, it is likely that genetic analysis alone will be insufficient to fully predict the clinical phenotype, because of incomplete penetrance and variable expressivity. This makes prenatal diagnosis more cumbersome and difficult to interpret, at least until the contributing factors are unambiguously identified.
We thank Prof M Vihinen for database construction and management. The technical assistance of Enrica Mira Catò is acknowledged. This study was partially supported by Grants Genomica Funzionale (PV), grant from Programma Nazionale Cellule Staminali (to PV), Grant RBNE019J9W from MIUR-FIRB (AV and PV), grants from Cofin 2003 (AA), grant from Research to Prevent Blindness, Inc. (TY), and grants from the Mabel E. Leslie Research Endowment Funds (TY). CS is a recipient of a fellowship from FIRC. This is manuscript no. 67 of the Genoma 2000/ITBA Project funded by CARIPLO.