Introduction

The V-ATPases are large multi-subunit complexes involved in the regulation of the pH of intracellular organelles as well as of specific extracellular compartments of eukaryotic cells [1]. In mammalians, the majority of the V-ATPase subunits have several isoforms which are often tissue-specific [1]. The most extensively studied are the four isoforms of subunit “a” because they have been implicated in different human diseases. In particular, a3, encoded by the TCIRG1 gene [MIM 604592], is expressed in osteoclasts at higher levels when compared to the other isoforms and is essential for bone resorption; as a demonstration it has been found mutated in human autosomal recessive osteopetrosis (ARO; [MIM 259700]; OPTB1) [2]. The same gene through an alternative splicing and the usage of an ATG in exon 7, gives rise to another protein isoform TIRC7 which is a T cell-specific surface protein expressed upon lymphocyte activation [3].

Direct structural information concerning the a3 isoform is lacking and the only data has been elicited from studies of site-directed mutagenesis of the corresponding subunit in yeast [4]. At the same time, the higher level of a3 expression on the osteoclast ruffled border, together with the possibility of being compensated for by the remaining isoforms in other compartments, suggests that the a3 isoform could be selectively inhibited with the aim to reduce bone resorption [5, 6]. The selective blockade of the V-ATPase thus obtained might represent a novel therapeutic strategy for the treatment of bone diseases characterised by excessive bone loss, including osteoporosis. The characterisation of inactivating mutations in this gene in human ARO can be useful in the process of drug design by identifying essential regions or single residues for protein function.

Here, we contribute to the understanding of the structure of the human a3 isoform by reporting the identification of 41 novel mutations in the TCIRG1 gene in a large group of 67 unpublished ARO patients; in all of them, both mutant alleles were identified. This data, together with the molecular and clinical characterisation of the relatives of these patients, tends to rule out the possibility of a dominant (mono-allelic) form of TCIRG1-deficient osteopetrosis.

Within the 41 novel mutations, we describe two large deletions and two different nucleotide changes affecting the donor site of exon 1, all found at the heterozygous state in patients where standard methods had previously allowed to identify only a single mutated allele. Therefore, we can add new insights to the investigation of “missing” mutations in this gene, leading to a complete molecular diagnosis essential for patient management and prenatal diagnosis.

The molecular data herein reported provides additional information for the on-going studies aimed at the development of inhibitors of the V-ATPase a3 isoform for the treatment of osteolytic diseases. In particular, osteoporosis and osteopetrosis can be regarded as the two sides of the same coin; new data on osteoclast function coming from the investigation of osteopetrotic patients could be instrumental in identifying novel strategies to the inhibition of bone resorption, which is one of the most used approaches to the treatment of osteoporotic patients.

Methods

Samples

As a centre of international reference for the molecular diagnosis of osteopetrosis, we have been collecting material from a large number of patients with a clinical diagnosis of severe osteopetrosis made on the basis of the following criteria at presentation: early diagnosis (before 2 years of age), high bone mineral density, bone marrow failure with consequent pancytopenia, hepatosplenomegaly and cranial nerve palsies.

Specimens, including blood and DNA samples, were collected from patients after their parents provided informed consent. This research complies with the standards established by the local ethical committee and the granting agency.

TCIRG1 gene mutation analysis

Mutation analysis of the TCIRG1 gene was performed using direct DNA sequencing of PCR-amplified exons (the mutation nomenclature conforms to www.hgvs.org/mutnomen). The TCIRG1 gene was amplified from genomic DNA (genomic sequence AF033033) in the conditions previously described [7]. For PCR amplification and sequencing of exon 1, which was not routinely investigated in previous mutation analyses, the following primers were used in the same conditions: forward primer 5′-GCCTCGAAGTGAGAGGAGAG-3′ and reverse primer 5′-GGAACGCTGGGCTGTGTTGG-3′.

The search for the known genomic deletions involving exons 11 to 13 was performed as already described [8] using the rapid diagnostic detection strategy involving a forward primer in intron 10 (8066 F; 5′-CTTCGCTCTGTCGCGCAGG-3′), a reverse primer in exon 12 (8721R; 5′-GGCCCATAAGCAGGAGCAG-3′) and an additional reverse primer (9931R; 5′-CAAGTGGTTGGCAGCCAGGC-3′) located in exon 14. The effect was verified at the cDNA level with the primers: forward primer in exon 10 5′-GGCCAGCTTCCAGGGCATCG-3′ and the reverse primer 9931R in exon 14.

Prediction of the protein domains was based on the potential topological domains reported in UniProt database (Q13488; www.uniprot.org) and refined by homology with the yeast subunit a (Vph1p) [9]. The putative structural consequences of mutations were estimated in silico by sequence analysis and multiple sequence alignment information.

Results

Since our last mutational report concerning ARO [10], we have identified 67 new TCIRG1-deficient patients. Among them, 30 bore known mutations, while 37 carried mutations not previously reported in literature, to the best of our knowledge, for a total of 41 novel mutant alleles (Table 1). All of these mutations are in the process to be introduced in a dedicated database (http://bioinf.uta.fi/TCIRG1base).

Table 1 Novel mutations in the TCIRG1 gene
Full size table

These pathological variants include 11 missense, 6 nonsense, 13 small deletions/insertions, 9 splice site mutations and 2 genomic deletions and are scattered throughout the entire gene.

All the missense mutations affect highly conserved amino acid residues among different species including yeast; the only exception being p.Leu399Arg, which however falls in a conserved region. Of note, 6 out of 11 missense mutations are changes to arginine and lie in putative transmembrane domains (TMD) of the a3 subunit of the proton pump, as predicted on the basis of the membrane topology of its homologue in yeast, Vph1p [9]. It has been suggested that introducing an arginine into a TMD can be pathogenetic; its large size and high polar character would disrupt the native structure through steric and/or electrostatic means, regardless of the wild-type residue it is replacing [11]. This would suggest the functional relevance of these amino acid changes, supporting the hypothesis of their pathogenicity.

The nonsense mutations fall in a region of the sequence that is shared by TCIRG1 and TIRC7, which could suggest that patients bearing these truncating variants might display a more severe immunological phenotype. However, the clinical data available do not support this hypothesis: while we and others have observed reduced levels of immunoglobulins in some TCIRG1-deficient patients ([17] and our unpublished data), the presence of disruptive mutations affecting both TCIRG1 and TIRC7 was not necessarily associated with hypogammaglobulinemia. On the other hand, no major defects in the T cell compartment have been reported. To verify the possibility of a genotype/phenotype correlation, it would be necessary to further investigate the immune function in larger cohorts of ARO patients with mutations in the TCIRG1 gene, with specific regard to T-lymphocyte activation. However, the rarity of the disease and the severity of the clinical course often leading to death early in life, pose ethical and technical concerns to these kinds of studies.

Thirteen mutations involve the deletion or insertion of one or a few nucleotides; for all but one, the putative effect is a frame shift and the formation of a stop codon at a variable distance; for the remaining one, the in-frame deletion of 10 amino acids and the insertion of a new one is predicted to occur (p.Trp216_Gly225delinsCys).

Among the nine splice site mutations, eight directly affect the donor or acceptor sites, while one involves a more distant position (c.807+5G>C). We have previously shown the pathological role of different mutations far from the classical splice site in TCIRG1 gene [10]; therefore, it is likely that this novel intronic nucleotide change might also be responsible for the disease. In addition, in the very same position a different nucleotide change has been recently reported in Chuvashian osteopetrotic patients and extensive functional analysis has clearly demonstrated its pathological role [12].

Two out of nine splice site mutations are located in the first intron and are different nucleotide changes in the same position (c.−5+1G>C and c.−5+1G>T). Interestingly, in intron 1, we also found a deletion of 116 nucleotides starting from the donor splice site (c.−5+1_−5+116del). Even though exon 1 of the TCIRG1 gene is not transcribed, the putative effect of these mutations is an aberrant splicing or an interference with transcription initiation, which we were unable to verify at the cDNA level. However, it is noteworthy that we could not detect these variants in more than 100 chromosomes, thus supporting the hypothesis of a pathologic role.

Finally, using the same approach that we have previously described [8], we identified a novel large genomic deletion of more than 1.1 kb, encompassing exons 12 and 13 (g.8670_9803del) (Fig. 1a). At variance with the one already published, this deletion preserves the reading frame but removes 83 amino acids (p.Gln438_Trp520del) in the transmembrane region. In addition, its breakpoints do not seem to involve repetitive elements, which could suggest a mechanism different from an Alu-mediated homologous recombination (Fig. 1b).

Fig. 1
figure 1

Characterization of the 1.1 kb deletion (g.8670_9803del) in TCIRG1. a Rapid screening of mutation g.8670_9803del; in lane A, the patient shows the presence of the wild-type band of 655 bp (amplicon 8066 F-8721R) together with an additional one of 732 bp corresponding to the allele carrying the 1.1 kb deletion (g.8670_9803del; amplicon 8066 F-9931R); in lane B, the previously published Pt2 [8] shows the presence of the wild-type band and an additional 585 bp generated from the allele carrying the 1.3 kb deletion (g.8280_9560del), previously published (amplicon 8066 F-9931R). PCR amplification of the same region in a normal donor (ND) produced only the wild-type band, whereas in these conditions, the expected amplicon of about 1.9 kb spanning the region of the deletion is not generated with primers 8066 F and 9931R. No bands are evident in the lane of the negative control (C−). MW, 1 kb DNA ladder (Invitrogen). b Schematic representation of the TCIRG1 genomic region spanning exons 11–14. The novel 1.1 kb deletion (g.8670_9803del) is depicted by the dashed area, while the 1.3 kb deletion (g.8280_9560del) previously published is depicted by the shaded area. Primers 8066 F, 8721R and 9931R were used for rapid diagnostic detection of the deletion. Orientation and type of the repetitive sequence elements comprised in this region are indicated by arrowheads, whose size is proportional to the length of the corresponding repeat

Full size image

Discussion

More than 10 years ago, mutations in the TCIRG1 gene were found to be responsible for human autosomal recessive osteopetrosis [2]. Since then, our group and others have published additional pathological changes in the same gene [7, 8, 10, 12–19]. In the present work, we report the identification of 41 novel mutations in 67 unpublished ARO patients. We confirm on a larger series that the TCIRG1-deficient form represents 53% of our cohort of ARO patients. Our findings underline the wide range of molecular defects present in TCIRG1-deficient ARO; in fact, more than 90 different mutations have been described in this gene to date, with only a few cases of recurrent changes found in well-defined populations, namely the Costa Rican, the Belgian and the Chuvashian [7, 10, 12].

In spite of the large number of different mutations in this gene, clinical data shows a homogeneous phenotype shared by TCIRG1-deficient patients; they present with a prominent bone defect, secondary haematological and neurological deficits and no major immunological impairment [20], even though in a small subset of patients, a reduction in immunoglobulin levels has been reported [17]. In addition, irrespectively of the specific molecular defect, this subset of ARO patients responds well to therapy through haematopoietic stem cell transplantation (HSCT) [17, 20].

All the different mutations identified in the TCIRG1 are spread along the gene; they are made up of 19% of missense, 19% of nonsense, 29% of small deletions/insertions, 30% of splice site mutations and 3% of large genomic deletions. Initially, a few patients were reported in whom only a single mutated allele was detected [7, 13] raising the possibility of the existence of a dominant form of the disease due to a single mutated allele in this gene, as occurs with mutations at the ClCN7 locus [21]. We have previously reported that a more complete analysis of the TCIRG1 gene can lead to the identification of mutations that standard methods fail to recognise, completing the molecular diagnosis in some of these patients and ruling out the possibility of a dominant form of osteopetrosis due to TCIRG1 defects [8]. In agreement with this assumption, in all the 67 ARO patients herein reported, we could identify both mutant alleles. Of note, in the present work, we describe two additional large genomic deletions and two splice site mutations in the 5′ UTR of the gene in four patients previously classified as mono-allelic (this untranslated exon was not investigated in our previous standard sequence analysis). While we could not precisely define the biological effect of the mutations in the 5′ UTR, it is likely that they affect the transcription of the gene. The possibility of completing the molecular diagnosis of these patients would give a stronger rationale to the decision to direct them towards HSCT.

The identification of two novel large genomic deletions underlines the importance to look for this type of mutation. Their recognition is particularly relevant and has a diagnostic and prognostic value. Indeed, the failure in detecting large allelic deletions can lead to the misinterpretation of the single variant sequence identified which can be erroneously considered as a homozygous mutation when masked by a coincident exonic deletion. Therefore, the identification of mutations in both alleles is essential for genetic counselling of family members, and, in particular, for a correct prenatal diagnosis.

All the missense mutations reported in literature except one, p.Ala141Pro [10], are located in the region of the putative TMDs, which is highly conserved throughout species up to yeast; this observation makes the conclusions on the biological effects of human mutations deduced from studies in Saccharomyces cerevisiae more relevant. In contrast, according to the SNP database, several coding polymorphisms are commonly found in the NH2 terminal region, which would further support the critical role of the TMD region for the pump activity. In addition, the uniformity of the phenotype in patients bearing defects in this gene would support that the missense mutations identified so far in TCIRG1 are functionally null and do not maintain any residual activity. This information is useful because it suggests which specific amino acid residues of the a3 subunit could be targeted in order to selectively inhibit the activity of the V-ATPase proton pump in the osteoclasts. Indeed, in this cell type, loss of function mutations in the a3 subunit can be compensated for only in lysosomes and not on the plasma membrane, thus highlighting the possibility to obtain an important block on bone resorption while preserving the lysosomal function [5, 6]. Recent studies have underlined the important role that osteoclasts play, under physiological and pathological circumstances, in mediating bone formation by supporting a continued anabolic signalling to osteoblasts in the bone remodelling compartment. In agreement with these findings, the prolonged treatment with several antiresorptive drugs, causing osteoclast cell death, has been associated with a secondary reduction of bone formation which ultimately limits the long-term efficacy of the treatment [22]. On the contrary, the usage of a3-selective inhibitors could result in an uncoupling of bone formation and bone resorption, thereby allowing a continuous increase in BMD improving the bone quality and the potential effectiveness of the treatment. Indeed, in vitro studies on osteoclasts treated with a3-selective inhibitors have already shown that bone resorption is prevented and cells survive longer, thereby mimicking the situation observed in ARO patients bearing mutations in the gene encoding a3 [22].

In conclusion, we provide an important contribution to the molecular dissection of TCIRG1-deficient ARO; overall, these findings pave the way for the development of a new class of drugs for the treatment of common diseases characterised by excessive bone resorption, first of all osteoporosis.