Mutations underlying human inherited disease vary greatly in size, from gross rearrangements to single base-pair substitutions. They can also originate via a number of different molecular mechanisms. Irrespective of the type of lesion, the nature, location and frequency of human gene mutations are shaped to a considerable extent, and often in remarkably predictable ways, by the local DNA sequence environment (for a review, see
Almost 45% of the genomic sequence of the human TAZ gene is represented by interspersed repeated sequences (SINES and LINES) and 76% of these (35% of the TAZ gene sequence) are Alu sequences. The proportion of the whole human genome that is Alu sequence is estimated to be around 10.6%
. A high proportion of Alu repeats within a gene sequence can promote gross gene rearrangements
, as reported for example for the PAFAH1B1 (LIS1) gene
. We report here an intragenic deletion in the TAZ gene, which appears to have been mediated by recombination between two AluY repeats. The deletion junctions in Pt1 were located within the two highly homologous and similarly oriented AluY repeats (Figure
3A). Non-allelic homologous recombination (NAHR) between Alu elements is a likely mechanism of mutation in this patient
. An Alu-mediated TAZ gene deletion that ablates the entire gene has also been reported
. These data suggest that the TAZ gene is located within a genomic environment that is particularly rich in repeats that can promote genetic rearrangements.
HGMD Professional v.2012.2 (
 update 29th June 2012) reports a total of 104 TAZ gene mutations of which 35.6% are missense, 12.5% are nonsense, 18.3% affect splicing, 19.2% are microdeletions, 7.7% are microinsertions and 6.7% are large gene rearrangements. Of the eight gross gene rearrangements reported, three were published
[31, 47, 48] whereas the remaining five can be retrieved from the “Human Tafazzin Gene Mutation and Variation Database”
 to which they were submitted directly, without supporting details. One mutation in the latter group is reported to involve the deletion of exons 6–11. In our patient sample, we identified deletions of exons 6–11 in two patients (Pt1 and Pt2) who harbored two different genetic rearrangements. Defining the exact breakpoints of such rearrangements is a prerequisite to identify heterozygous carriers of TAZ gene rearrangements, and for prenatal diagnosis.
Despite the high density of Alu repeats in the TAZ gene, the deletion observed in Pt2 was not mediated by NAHR between Alu elements. In this patient, the genetic rearrangement was found to be more complex than in Pt1 (one 38 bp deletion and one >4 kb deletion) and might have been prompted by and Serial Replication Slippage (SRS)
. Consistently with this assertion, the sequences flanking the breakpoints exhibit microhomologies as depicted in Figure
3B. Sequential slippage events at the 3’ end of the nascent strand during DNA replication could have been primed by these microhomologies
. We postulate that the SRS model, instead of the more recently proposed FoSTeS model
, is the more likely mechanism of the rearrangement in Pt2 since the SRS model assumes that the replication slippage occurs at closely adjacent sites and gives rise to relatively small rearrangements (up to 10 kb), whereas the FoSTeS model is better suited to explain template switching that occurs over long distances, generating much larger DNA rearrangements (120 to 550 kb)
. Whatever the underlying mechanism, characterization of gross rearrangements in the TAZ gene is essential for identifying heterozygous carriers among female relatives of the probands and, as it is the case for Pt2, to confirm the carrier status of close female relatives.
In addition to these gross gene rearrangements, we identified two frameshift variants in the TAZ genes of our patients: a small 14 bp deletion (Pt5) and a small duplication (Pt6). Ball et al.
 suggested that a combination of slipped mispairing mediated by direct repeats and secondary structure formation promoted by symmetric elements could account for most microdeletions and microinsertions. We analyzed the sequence environment as suggested
, considering the 10 bp of DNA sequence flanking the c.678_691del14 deletion detected in Pt5 (ACAGTCCGCCctacttcccccgctTTGGACAGGT). This region is G/C rich and the deletion is flanked by CCGC repeats, consistently with a slipped mispairing model. At the protein level, the c.678_691del14 deletion is predicted to lead to a p.Tyr227Trpfs*79 frameshift.
The c.284dupG duplication in Pt6 (GTTGATGCGTTGgGTGAGGAGGA) might be explained by the modified slipped mispairing model
, whereby misalignment occurs between a specific contiguous sequence on one strand and a second partially homologous, yet interrupted, sequence on the other strand. One repeat copy only becomes capable of base-pairing once an additional base has been inserted or deleted. In the case of Pt6, the first repeat is GTTGA whilst the second repeat is GTTGGA; slipped mispairing would therefore account for the duplication of a G in this first repeat. At the protein level, the c.284dupG mutation is predicted to lead to a frameshift p.Thr96Aspfs*37 with premature termination of translation.
We also detected two single nucleotide substitutions in the TAZ genes: c.641A>G (Pt3) and c.367C>T (Pt4). c.641A>G leads to the substitution p.His214Arg, which, although not predicted to change the polarity of the involved residue, occurs in a residue which is evolutionarily conserved, suggesting an essential role in protein function. We are unable to infer about the possible impact of the missense change directly on the three-dimensional structure of human tafazzin, as the crystal structure is not available, however molecular characterization of mutant TAZ alleles has been performed in a Saccharomyces cerevisiae model of Barth syndrome
. While the corresponding residue in the S. cerevisiae Taz1p ortholog is not conserved (it is an Alanine), the p.His214Arg mutation occurs in a putative unusual membrane anchor region in yeast Taz1p
. Interestingly, mutations in this membrane anchor result in two distinct biochemical fates: low fidelity sorting with increased rates of aggregation and aberrant macromolecular assembly
Pt4 carried the p.Arg123Term nonsense mutation previously reported by van Werkhoven et al.
. According to the Human Tafazzin Gene Mutation and Variation Database
, multiple examples of the p.Arg123Term mutation have been reported, suggesting recurrent mutation at this CpG dinucleotide mutation hotspot.
We found lactic acidosis to be present in 5/6 patients at the time they were brought to clinical attention, suggesting that, in addition to 3-methylglutaconic aciduria and neutropenia, lactic acidosis is an important biochemical marker for suspecting BS. In addition, a deep metabolic decompensation during the first hours of life, including respiratory distress and cyanosis with oxygen desaturation, was associated with the most severe outcome (Pt1, P2 and Pt5). Deep metabolic decompensation associated to these other signs may thus correlate with the severity of the disease more than it does the time of onset of cardiomyopathy that, for instance, was already detected in utero in Pt3 who was still alive at last follow up
[54, 55]. The alteration of cardiolipin metabolism in BS leads to an abnormal lipid structure of the inner mitochondrial membrane. This may result in mitochondrial leakage of metabolites, such as the lactate as we observed in our BS patients. Thus, we could hypothesize that an acute metabolic decompensation during the first hours of life, associated with respiratory distress and oxygen desaturation, or with other signs such as mild hyperammonaemia, hypoglycaemia and elevated transaminases
 could be correlated to a deeper impairment of the mitochondrial function at the onset of BS and thus to an earlier and deeper cardioskeletal tissue damage. However, it cannot be excluded that the severity of the metabolic decompensation can be related also to other still undefined factors, as it has been reported that the pathophysiology of Barth syndrome can unlikely be explained solely on the basis of cardiolipin abnormalities and additional factors could play a role in modifying the disease expression
. Anyhow, these assertions remains to be proven.
Several cases of clinical manifestations of BS being detected in utero, as in the case of Pt3, have been reported
[57, 58]. Therefore, cardiolipin analysis or mutational analysis of the TAZ gene should be considered in cases of heart failure or sepsis during the neonatal period or of in utero cardiomyopathy. Such analysis should also be considered when investigating the causes of natural abortions (e.g. Pt1 and Pt6) as also previously reported
Early diagnosis of BS is a prerequisite for early and effective therapy and follow-up, and for preventing crisis due to sepsis, as suggested by the clinical histories of Pt3 and Pt6. Moreover, prompt molecular characterization of BS patients is vital for prenatal diagnosis and for preventing the recurrence of BS in the same family (e.g. Pt5). Late diagnosis of BS can lead to familial recurrence of the disease as observed in the families of Pt1, Pt2 and Pt4.