m.3700G>A/MT-ND1 (Families 1 and 2)

The mtDNAs from the probands of family 1 (Italy) (Figure S1) and the previously reported family 2 (Germany) [27] harbored the non-synonymous nucleotide change m.3700G>A. The phylogenetic relationship between the two probands’ sequences show that this mutation is not shared by descent, but it is due to two independent mutational events (Figure 1): the first radiates from the root of H, the most common haplogroup in Europe [34], whereas the second can be further classified into the H26a subclade [35]. This nucleotide change, never reported as a population polymorphism, causes the amino acid substitution p.A132T in the ND1 protein, which has been previously proposed as pathogenic for LHON in the German family [27]. The RFLP analysis of the m.3700G>A mutation showed that it was virtually homoplasmic in DNA from peripheral blood (data not shown). A DNA sample from urinary epithelium available only for individual II:1 of family 1 was also homoplasmic mutant. Both cases were apparently sporadic and no further DNA was available for analyses from maternal relatives. Thus, following the initial description of family 2, the pathogenic role of the rare m.3700G>A mutation is confirmed, with the detection of an Italian LHON patient (Family 1), as a separate mutational event.

Family 1 harbored also a second non-synonymous change m.13759G>A/MT-ND5. This nucleotide change is a polymorphic marker of the Asian haplogroups F1a’c’f and M7c1d [35], and it has been also sporadically observed in mtDNAs belonging to other haplogroups.

m.3733G>A-C/MT-ND1 (Families 5, 9 and 14)

The m.3733G>A mutation was detected in two previously described Italian families (Families 1 and 2 in Valentino et al. [8], here renamed families 5 and 14, respectively) on haplogroups H10 and X2b, thus indicating that it resulted from two distinct mutational events (Figure 1). The mutation m.3733G>A causes the amino acid substitution p.E143K and in both families was homoplasmic in the affected individuals, whereas many unaffected maternal relatives were heteroplasmic [8].

The mtDNA from family 5 revealed the presence of only the other non-synonymous change m.9091A>G/MT-ATP6, which has been found in several healthy subjects belonging to haplogroup H10 [35], thus a role in LHON pathogenicity is very unlikely.

The mtDNA sequence from family 14 was instead characterized by numerous non-synonymous changes (see Table 1). As shown in Figure 1, most of these (m.8393C>T, m.13708G>A and m.13966A>G) are ancient polymorphisms of haplogroup X2b. The remaining two substitutions (m.7859G>A and m.11084A>G) are phylogenetically more recent. The m.7859G>A nucleotide change was previously reported in a pediatric case of mitochondrial encephalopathy (patient 2 in Uusimaa et al. [36]), but no clear pathogenicity was assigned because it was homoplasmic. Moreover, it has also been found in different normal individuals associated with at least two different Asian haplogroups (Y2 and M4) [35]. Similarly, the m.11084A>G substitution has also been initially reported as pathogenic for MELAS in an Australian patient [37], but subsequently identified as a polymorphism of the Asian haplogroups A4e1 and M2a1a1a [35]. Thus, most probably these changes are polymorphic variants, but their potential synergistic role is further discussed based on their conservation pattern.

In the course of this study we also found a family from Germany (Family 9 in Figure S1) harboring a heteroplasmic (see Figure 3) variant at the very same position but with a different nucleotide change, m.3733G>C, leading to a different amino acid substitution (p.E143Q). Its mtDNA belongs to sub-haplogroup J1c3e1, a clade characterized by several non-synonymous variants affecting MT-ND1-6 and MT-CYB genes (Table 1, Figure 1). Beside those changes established as markers of J1c3e1, we also note the presence of the change m.14502T>C/MT-ND6. This variant has been previously reported as the only pathogenic change in three unrelated sporadic LHON cases from China, in one Chinese LHON pedigree characterized by complete penetrance in combination with the m.14484T>C primary LHON mutation, and in four additional Chinese unrelated families carrying the m.11778G>A mutation [38]–[40]. The latter two scenarios parallel the one that we observed in the mtDNA sequence of family 9, where the m.14502T>C mutation is in association with the heteroplasmic m.3733G>C mutation. In this context, the m.14502T>C change might worsen the complex I deficiency caused by the 3733G>C transversion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. RFLP evaluation of heteroplasmy.

Electrophoresis on Metaphor gel of the restriction fragments is shown. Wild type and mutant fragments are indicated with the corresponding size, expressed as base pairs (bps). Non affected negative control individual is indicated as C. DNA samples were extracted from whole blood, except when indicated: whole blood is indicated as B, urinary epithelium is indicated as U, platelet fraction is indicated as P, hairs are indicated as H and skeletal muscle is indicated as M. Molecular weight marker is indicated as MWM.

https://doi.org/10.1371/journal.pone.0042242.g003

m.4171C>A/MT-ND1 (Family 13)

The complete sequencing of mtDNA from the proband of the small French family 13 (Figure S1), with two maternally related affected males revealed the heteroplasmic m.4171C>A mutation (Figure 3), which was previously proposed as pathogenic in two Korean families [15] and further found in a sporadic case and one family from China [16], [17]. Both Korean mtDNAs harbored apparently identical sequences belonging to haplogroup A, thus raising the possibility of a single ancestral mutational event. However, in one family the proband was apparently a sporadic case and his mother was heteroplasmic, supporting the alternative scenario of two independent mutational events. The Chinese cases were unrelated to the Korean, and one family (haplogroup N9a1) displayed an almost complete penetrance [17]. In all these cases further changes affecting MT-ND1-6 and MT-CYB genes were thought to be putatively synergistic with the m.4171C>A mutation. The mtDNA of our proband belongs to haplogroup J2b1 (Figure 1) and this is the first report of the m.4171C>A mutation associated with LHON on a non-Asian background.

The addition of our case, obviously unrelated, confirms the pathogenic role of the m.4171C>A mutation and its association with haplogroup J2b (presenting numerous non-synonymous polymorphisms) and further suggests the need for synergistic variants to increase its pathogenicity. Moreover, this mtDNA also harbored another novel variant, m.7632T>C/MT-COII, whose synergistic role cannot be excluded (Table 1).

m.10663T>C/MT-ND4L (Family 16)

The complete sequencing of mtDNA from the proband of this sporadic French case of African descent (Benin) revealed a haplogroup L2a1 background (Figure 1 and Figure S1) carrying the m.10663T>C mutation. This mutation was previously reported in five pedigrees all belonging to different haplogroup J sub-clades [23], [24], thus indicating multiple independent mutational events. Our case is the first in a non-J mtDNA background and harbored also a combination of non-synonymous polymorphisms m.8701A>G/MT-ATP6, m.9053G>A/MT-ATP6 and m.10398A>G/MT-ND3, for which a synergistic role, similar to the haplogroup J polymorphisms, may be considered.

m.14459G>A/MT-ND6 (Family 8)

The LHON proband from this Italian family carried a virtually homoplasmic m.14459G>A mutation (Table 1 and 2, Figure S1). The pathogenicity of this mutation is well established and notorious for a variable clinical expression, ranging from LHON, dystonia or both, and Leigh syndrome [28], [30], [41], [42]. Here, two maternal relatives of the proband, reported as wheelchair bound and suffering of “paralysis”, were possibly affected with spastic dystonia. The mtDNA of the family here reported belongs to haplogroup J1c3, thus reinforcing the previously discussed synergistic role of this clade.

m.14482C>A/MT-ND6 (Families 10 and 15)

Both families, one Italian and one German of Turkish ancestry, were previously reported to carry the m.14482C>A mutation [11], [27], but the complete sequence analysis was not available. The Italian sequence belongs to haplogroup J1c and harbors all the non-synonymous variants characterizing this haplogroup, whereas the German family is a member of haplogroup I5a, which is also characterized by three non-synonymous variants m.5074T>C/MT-ND2, m.10398A>G/MT-ND3 and m.13780A>G/MT-ND5. A different nucleotide change affecting the same position m.14482C>G/MT-ND6 has been previously reported [9] in an unrelated family of Turkish ancestry, which noticeably belonged to the haplogroup I. It is conceivable that haplogroup I potentially plays a role similar to J. Furthermore, a third patient carrying the m.14482C>A mutation has been reported by screening a large cohort of patients with optic atrophy from France [43]. In summary, mutations at position 14482, although affecting a poorly conserved amino acid (Table 2) that is also changed by the well-established LHON primary mutation m.14484T>C, must be considered pathogenic.

m.14495A>G/MT-ND6 (Family 3)

The proband from the Italian family 3 carried the heteroplasmic m.14495A>G mutation (see Table 2, Figure 3 and Figure S1), which represents the third unrelated occurrence of this mutation in association with LHON [7]. All the cases, including our (sub-haplogroup H66), belong to haplogroup H and the mutational events were possibly recent, as suggested by their heteroplasmic status.

m.14568C>T/MT-ND6 (Families 4, 6, 7, 11 and 12)

This previously reported transition [12]–[14] was found in five different mtDNAs, whose phylogenetic analysis indicates independent mutational events (Figure 1).

In the mtDNA from family 6 (haplogroup U6a1a1, Figure S1) the m.14568C>T mutation was found heteroplasmic in combination with five non-synonymous nucleotide changes (see Table 1 and Olivieri et al. [44]). They are all known polymorphic variants, with the m.4172T>A mutation defining this specific subclade. This polymorphism induces the p.L289Q change by affecting the same codon of the previously discussed m.4171C>A pathogenic mutation. In addition, the m.14568C>T mutation was heteroplasmic in multiple tissues from the proband, being approximately 90% in skeletal muscle and urinary sediment, but 60% in blood leukocyte and platelet fractions. Similar results were obtained by analyzing multiple tissues from the proband’s brother, whereas all the remaining maternally related individuals available for analysis harbored a virtually homoplasmic wild type mtDNA except for the proband’s mother and maternal aunt who had traces of mutant mtDNA (Figure 3). The RFLP analysis of the m.4172T>A variant revealed that it was homoplasmic in all maternal relatives and all tissues investigated (data not shown).

The same pathogenic mutation m.14568C>T was also found in two unrelated French families, which belonged to haplogroups K1a (family 7) and J1c2e (family 11), respectively. In both cases the mutation was homoplasmic in the probands. Besides the haplogroup J-related non-synonymous changes of mtDNA from family 11, the mtDNA from family 7 harbored a collection of non-synonymous variants (Table 1).

Finally, we also re-examined by complete mtDNA sequencing the two originally reported German cases carrying the m.14568C>T mutation (families 4 and 12) [12], [14]. The proband from family 4 belonged to haplogroup H1c1 with no further non-synonymous changes besides the pathogenic mutation, whereas the proband from family 12 clustered within haplogroup J1c2, thus harbored all non-synonymous changes characterizing this haplogroup [35]. Overall, two of the five families carrying the m.14568C>T mutation were associated with haplogroup J1c, and one further family had the interesting association with a variant affecting the same amino acid changed by the LHON pathogenic m.4171C>A mutation, again emphasizing the importance of the mtDNA background sequence variation.

Interspecies Conservation Analysis of Pathogenic Mutations and Non-synonymous Variants

The mutation m.3700G>A/MT-ND1 causes the substitution of the non-polar p.A132 with a polar T and affects a highly conserved position (Table 2). Similarly, the mutations m.3733G>A/MT-ND1 and m.3733G>C/MT-ND1 induce the amino acid substitutions p.E143K or p.E143Q, respectively. The position affected is extremely conserved (Table 2) and both mutations cause a change in the chemical properties from the acidic glutamic acid, which is negatively charged, to either the positively charged, basic lysine or the uncharged, polar glutamine. Remarkably, these three mutations (m.3700G>A, m.3733G>A and m.3733G>C) affect the same highly conserved region of the ND1 protein (Figure 2), suggesting an important role of this conserved domain in complex I function or structure. The mutation m.4171C>A/MT-ND1 also affects the highly conserved position p.L289 (Table 2), which is invariant in mammals, and leads to the substitution with methionine. Interestingly, the polymorphism m.4172T>A/MT-ND1, found in family 6 mtDNA and considered a polymorphic marker of haplogroup U6a2, affects the same p.L289, causing the replacement of the non-polar leucine with the negatively charged glutamic acid, suggesting a probable synergistic role of this polymorphism in LHON pathogenesis co-occurring with the primary mutation m.14568C>T/MT-ND6.

The m.10663T>C/MT-ND4L mutation induces the conservative amino acid substitution valine to alanine at position 65; the residue is moderately conserved along eukaryotes but becomes highly conserved in vertebrates and mammals (Table 2).

Four of the five mutations in the MT-ND6 gene are concentrated in a hyper-conserved region of the protein and mostly affect highly conserved positions. The m.14459G>A/MT-ND6 mutation causes a conservative p.A72V replacement. The p.A72 is highly conserved along vertebrates and becomes invariant in mammals, underlying a major functional role of this amino acid in the ND6 subunit. Similar features characterize the amino acid substitution generated by the mutation m.14495A>G/MT-ND6. This mutation causes the replacement of the highly conserved non-polar p.L60, which is invariant in vertebrates and mammals (Table 2), with a polar serine. Within this highly conserved amino acid stretch of ND6, a remarkable exception is the residue p.M64, affected by the mutations m.14482C>A-G. In fact, this mutation induces a conservative change (p.M64I) and hits a position with a low degree of conservation (see Table 2 and Figure 2) which is, however, contiguous to invariant stretches of amino acid positions characterizing this highly conserved domain of the MT-ND6 gene in mammals (Figure 2). Finally, the m.14568C>T/MT-ND6 mutation affects the amino acid residue p.G36 that is moderately conserved along eukaryotes and vertebrates, but becomes invariant in mammals, again suggesting the acquisition of a particular role of this position during evolution. Two different predictive tools (PolyPhen2 and SIFT) were also used to evaluate the pathogenic potential of the nine novel rare mtDNA variants. All mutations were classified as “possibly/probably damaging” or “not tolerated” by at least one software, and for 67% of these (six out of nine, Table 2), both algorithms highlighted the possible pathogenic effect.

Considering the 28 non-synonymous polymorphic variants identified in the 16 completely sequenced mtDNAs (see Figure 1), 14 were over the threshold of 70% conservation in mammals (Table S2). As already noted, the m.4172T>A/MT-ND1 polymorphism, a marker of haplogroup U6a2, stands out by affecting the same invariant p.L289 (100% conservation) as the LHON primary mutation m.4171C>A/MT-ND1. A further group of polymorphisms with very high degree of conservation in mammals (90%–100%) included five variants: m.7632T>C is a private mutation of a J2b1 sample; m.13145G>A and m.15257G>A are markers of J subclades; the remaining two (m.9055G>A and m.14927A>G) define two different U branches (U8b and U6a1, respectively). Other seven polymorphisms had a conservation pattern between 70% and 90%. Finally, we also applied the two in-silico predictors (PolyPhen2 and SIFT) to further assess the functional impact of these polymorphic changes on mtDNA encoded proteins. Nine amino acid substitutions were independently predicted as damaging or not tolerated by these tools (Table S2). As expected, the polymorphism m.4172T>A/MT-ND1 is the only one that is predicted to have a functional impact by both algorithms, because it hits the same p.Leu298 amino acid of the LHON primary mutation m.4171C>A/MT-ND1, which is invariant in mammals.

To further characterize the possible functional role of the non-conserved polymorphic variants, we extended the conservation analysis on the protein region surrounding these polymorphisms. Eight out of 15 had an invariant amino acid within four contiguous amino acidic positions (+4/−4 aa, Table S3) and four of these hit a protein region with a local conservation higher or equal to the global conservation of the protein subunit, suggesting that the protein regions containing these variants may represent functional domains. Remarkably, two ancient variants defining haplogroups J (m.13708G>A) and X (m.13708G>A and m.8393C>T) were selected by both analyses, supporting their possible functional role. In particular, the m.13708G>A/MT-ND5 variant has been repeatedly associated with LHON as a secondary mutation.

Ethics Statement

All experimental procedures and written informed consent, obtained from all donors, were reviewed and approved by the following Ethics Committee: Comitato etico indipendente dell’Azienda Ospedaliero-Universitaria di Bologna, Policlinico S.Orsola-Malpighi, Prot. n. 789/2007.

Mitochondrial DNA Complete Sequencing and Phylogenetic Analyses

DNAs were extracted from venous blood. Sequencing of entire mtDNAs and phylogeny construction was performed as previously described [34]. All the mutations are relative to the revised Cambridge Reference Sequence (rCRS) [57]. The complete mtDNA sequences of the 16 case reports have been deposited in GenBank (Table 1).

Evaluation of Heteroplasmy for Rare LHON Mutations

Heteroplasmy/homoplasmy status of the rare LHON mutations was evaluated by RFLP analysis of PCR amplified mtDNA fragments encompassing the position of the mutant nucleotide. The list of the primers and restriction enzymes used for each mutation are listed in Table S4. Fragments were separated by electrophoresis on Metaphor (BioSpa) gel at 3% or 4% and detected by ethidium bromide staining. Quantification of mutational load was carried out by densitometric analysis using AlphaView software package (Cell Biosciences).

Alignment of Protein Sequences and Interspecies Conservation Analysis

The interspecies conservation of the amino acidic positions changed by pathogenic mutations and non-synonymous polymorphic variants, detected in our mtDNA complete sequences, was calculated and, as a general rule, all the polymorphisms above a conservation threshold (70%) were considered “possibly synergistic”.

All the available complete mitochondrial protein sequences were downloaded from the non-redundant SwissProt database (http://www.expasy.org/sprot) and three sequence sets were created, corresponding to eukaryotes, vertebrates and mammals. These sequences were aligned as previously reported [58], the most representative amino acidic residues were identified and their prevalence was calculated at each position within the alignment, setting the value of 70% as conservation threshold.

To assess the possible functional effect of primary rare mutations and non-synonymous polymorphic variants four different prediction tools available on line were applied: PolyPhen2, SIFT, SNP&GO and PROVEAN 1.0 [59]–[62]. The accuracy of these tools was tested on the confirmed pathogenic mtDNA mutations in MT-ND1-6 genes listed in MITOMAP [26]. Eventually, we adopted PolyPhen2 and SIFT that resulted the most accurate algorithms predicting as damaging or not tolerated 15/16 of the tested mutations.

The non-synonymous polymorphic variants were clustered in ancient or recent changes, depending on their position along the phylogenetic tree, and the percentage of conserved and non-conserved amino acid positions was calculated within each cluster. Further analyses of the non-conserved amino acid changes induced by non-synonymous polymorphic variants included the estimate of the local conservation and number of conserved amino acids around the affected position (−10/+10 amino acids), which was compared to the global conservation and the total number of conserved amino acids of the entire protein. Those polymorphisms inducing amino acidic changes at positions with both parameters higher than the global conservation and number of conserved amino acid were considered “possibly synergistic”. Finally, the closest invariant (100% conservation) position for each amino acid change was identified and the polymorphic changes hitting amino acidic positions nearby invariant positions (−4/+4 amino acids) were considered “possibly synergistic”.