A human mitochondrial DNA (mtDNA) transition, m.1555A>G, in the 12S rRNA gene causes non-syndromic hearing loss. However, this pathological mutation is the wild-type allele in orangutan mtDNA. Here we rule out different genetic factors as the reason for its fixation in orangutans and show that aminoglycosides negatively affect the oxidative phosphorylation function by decreasing the synthesis of mtDNA-encoded proteins and the amount and activity of respiratory complex IV. These drugs also diminish the growth rate of orangutan cells. The m.1555G nucleotide is also the wild-type allele in other mammal species and they might be at risk of suffering a mitochondrial disorder if treated with aminoglycosides. Therefore, pharmacogenomic approaches should be used to confirm this possibility. These observations are important for human health. Due to the fact that old age and high frequency are criteria widely used in mitochondrial medicine to rule out a genetic change as being a pathological mutation, our results prevent against simplistic genetic approaches that do not consider the potential effect of environmental conditions. Hence, these results suggest that some ancient and highly frequent human population polymorphisms, such as those defining mtDNA haplogroups, in mitochondrial rRNA genes can be deleterious in association with new environmental conditions. Therefore, as the discovery of ribosomal antibiotics has allowed to fight infectious diseases and this breakthrough can be considered an important scientific advance or ‘progress’, our results suggest that ‘progress’ can also have a negative counterpart and render detrimental many of these mtDNA genotypes.
The human mitochondrial DNA (mtDNA) m.1555A>G transition in the 12S ribosomal RNA (rRNA) gene (MT-RNR1) was first reported as causing non-syndromic hearing loss (1) and, since then, it has been frequently associated to this pathological condition (2). Beside this, transmitochondrial cell lines or cytoplasmic hybrids (cybrids), produced by transferring mitochondria from homoplasmic mutant patients to cells without mtDNA (rho0 cells), had decreased mitochondrial protein synthesis and oxygen consumption and increased the cell-doubling time when compared with cybrids without this mutation (3). Because 12S rRNA is required for the expression of the 13 mtDNA-encoded polypeptides of the oxidative phosphorylation system (OXPHOS), which is the major oxygen consumer of the cell, and because mitochondrial energy is important for cell division, these results confirmed the pathogenicity of this mutation. However, the m.1555G allele appears to be the wild-type allele in orangutan (Pongo spp.) mtDNA (4–6) (Fig. 1) and, apparently, they do not show hearing problems (7–9). Thus, some mtDNA rare mutations, such as human m.1555G, are etiologic factors for very severe diseases and highly frequent polymorphisms, such as orangutan m.1555G, have not phenotypical effects.
Phenotypes depend on the interaction of genetic background and environment. The genetic makeup of humanity has changed little during the last 10 000 years. As an example, major European mtDNA haplogroups (phylogenetically related mtDNA genotypes) were already present in Europe at that time (10). However, in recent times, our lifestyle has been transformed to the point that there is now a mismatch between our ancient genetic constitution and certain aspects of our daily lives. These rapid changes have far outpaced any possible genetic adaptation. Therefore, this discordance can act as a promoter of the so-called diseases of civilization (11). Mitochondria are unintended off targets of different therapeutical drugs (12,13) and other natural and commercial xenobiotics, such as pesticides, can affect mitochondrial function. For example, over 60 different families of compounds are known to inhibit respiratory complex I (14). Thus, environmental changes associated to ‘progress’, through new interactions with the mitochondrial genome, might affect individual health status.
Here, we explore the reasons for the fixation of the m.1555A>G pathological mutation in the orangutan mitochondrial genome and, to analyze interactions between new environmental stresses and ancient mtDNA makeups, we use orangutan cells as a model of an ancient mtDNA genetic variant and aminoglycosides as an example of a new environmental factor. We have found that the exposition to these ribosomal antibiotics renders detrimental the orangutan mtDNA genotype. Our results suggest that some ancient and highly frequent human population polymorphisms in mitochondrial rRNA genes, such as those defining mtDNA haplogroups, can be deleterious in association with new environmental conditions.
Several hypotheses have been advanced to explain the fixation of human pathological mutations in closely related species, such as a recent founder effect, the inefficiency of natural selection due to post-reproductive phenotypes or the presence of compensatory mutations (15).
Orangutan m.1555A>G transition is an ancient mutation
To rule out m.1555G as an error in the orangutan reference sequence, a private polymorphism or a recent founder effect, we collected the MT-RNR1 reference sequences for Bornean (Pongo pygmaeus) and Sumatran orangutans (Pongo abelii) (4,5) as well as 15 partial sequences from both species (16). The 17 sequences have guanine in this position (Fig. 2A). These results rule out this mutation as a sequencing error or a recent mutation because the two species diverged ∼400 000 years ago (17).
Negative selection is acting on nucleotide position m.1555
The inefficiency of natural selection that is due to a post-reproductive phenotype can also be ruled out as the reason for the fixation because 53 out of 55 deaf individuals from the Israeli-Arab pedigree, where this mutation was first found, became deaf in infancy or early childhood (1,18). Perhaps mitochondrial deafness due to this pathological mutation is a male-specific phenotype and, because mtDNA is inherited exclusively by maternal lineage (19), natural selection does not affect its population frequency (20). However, 26 of those 53 deaf individuals from the Israeli-Arab pedigree were female (18). In any case, these observations do not confirm the action of natural selection on this mtDNA genetic variant. It has recently been found that orangutans use less energy relative to body mass than nearly any eutherian mammal, including sedentary humans. This slow life history of orangutans results from decreased energy throughput (21). It is possible that this physiological adaptation was accompanied by a mutational relaxation in mtDNA-encoded OXPHOS genes. To test this possibility, we checked in the orangutan mtDNA 40 mitochondrial tRNA (mt-tRNA) and 2 mitochondrial rRNA (mt-rRNA) human pathologic mutations (1,22,23). Only two (i.e. m.1555A>G/MT-RNR1 and m.5703G>A/MT-TN) were the wild-type allele in orangutans. However, the tRNA mutation was probably compensated by another m.5687C>T/MT-TN transition (Supplementary Material, Fig. S1). Additionally, the m.1555A position is conserved in 306 [conservation index (CI) = 96.8%] of 316 mammal species examined but the mean CI of 20 positions separated by 50 nucleotides along the MT-RNR1 gene (positions m.648, m.698, m.748 and so on) is 70.2% (Supplementary Material, Fig. S2). Moreover, using the ConSurf tool (http://consurf.tau.ac.il) which calculates CI while taking into account the topology of the phylogenetic tree, we obtained a CI = 8, being 9 completely conserved. These results suggest that negative selection is indeed acting on the orangutan mtDNA-encoded OXPHOS genes and in particular on nucleotide position m.1555.
Compensatory mutations do not appear to be the cause for the fixation of m.1555G in orangutan mtDNA
Compensatory mutations have been hypothesized as a probable explanation for many of the fixations of mtDNA-encoded protein (24) and tRNA (25) disease mutations. Compensatory mutations are also frequently found in mtDNA-encoded rRNAs (26). However, despite the existence of compensatory mutations at helices 39 and 40 of the decoding site from the orangutan rRNA, the pathologic m.1555A>G mutation has not been compensated by an m.1494C>T transition (Fig. 1).
Mitochondrial ribosomal protein S12 (MRPS12) is a critical component of the decoding center (27) and compensatory interactions between ribosomal proteins and rRNA variants have already been reported (28). We analyzed human and orangutan MRPS12 sequences and observed only two differences in the mature proteins. Human R18 and R60 amino acids correspond to Q18 and Q60 amino acids in orangutan (Supplementary Material, Fig. S3A). However, these positions do not interact with the rRNA mutant nucleotide in a reconstruction of bacterial ribosomes (29) (Fig. 3A). Therefore, intra- and inter-gene structural compensatory mutations do not appear to be the cause of the fixation of the m.1555G pathological mutation.
A functional compensation has been recently proposed to explain the population survival of a homoplasmic mt-tRNA pathologic mutation (30). The m.1555A>G mutation causes 12S rRNA hypermethylation at mtDNA positions m.1583A and m.1584A, induces a faulty retrograde mitochondrial biogenesis signal and predisposes cells to stress-induced cell death (31). Therefore, a less active orangutan mitochondrial rRNA methyltransferase (MTFB1) might functionally compensate for the rRNA mutation. After comparing human and orangutan MTFB1, we only found one difference in a protein motif required for these methylations (32) (Supplementary Material, Fig. S3B). However, this amino acid is not interacting with the rRNA helix that includes the methylated adenines (Fig. 3B).
Other genes less related to the 12S rRNA could be responsible for the fixation. Thus, the A10S mutation in mitochondrial 5-methylaminomethyl-2-thiouridylate-methyltransferase (TRMU) leads to a marked failure in mt-tRNA metabolism and contribute to the impairment of mitochondrial protein synthesis, then aggravating the mitochondrial dysfunction associated to the m.1555G mutation (33). However, orangutan and human harbor the same amino acid at this position (Supplementary Material, Fig. S3C). Hence, a functional compensatory mutation does not appear to be the reason for the fixation of the pathological m.1555G mutation.
Although they diverged between 12 and 16 million years ago, human and orangutan genomes are 97.4% similar and these genes related to 12S rRNA function do not show substantial differences among these species. Bornean and Sumatran Orangutans diverged 400 000 years ago and they are similar in a 99.7% (17). Their genomes have remained very stable during this time. Therefore, although we cannot completely rule out potential genetic compensations, our results suggest that the absence of the phenotypical effect of the m.1555G pathological mutation probably does not depend on the interaction with other genetic factors.
Aminoglycosides negatively affect OXPHOS function in orangutan cells
Since none of the previously advanced hypotheses seems to explain the fixation of this pathological mutation in orangutan, it is possible that the m.1555G position does not have complete penetrance and requires a co-factor to become pathological (3). In fact, antibiotic aminoglycosides trigger hearing loss in individuals harboring this mutation (1), and their effects have been confirmed in cybrid cell lines. Thus, mitochondrial translation in a HeLa cybrid clone repopulated with the homoplasmic mutation showed a high susceptibility to streptomycin (34) and, despite there were no differences in mitochondrial protein synthesis, cytochrome oxidase activity, oxygen consumption and ATP synthesis between homoplasmic mutant and isogenic control osteosarcoma 143B cybrids, mutant cybrids grew less in the presence of aminoglycosides (35).
The m.1555 equivalent position in the bacterial rRNA (16S rRNA) forms a Watson–Crick base pair with nucleotide 1494, which is important for aminoglycoside binding. Human wild-type mitochondrial 12S rRNA lacks this base pair, but it is rebuilt by the m.1555G mutation (Fig. 1). Thus, human mutant and orangutan wild-type mt-rRNAs are more similar to the bacterial rRNA and favor aminoglycoside binding, which can produce toxic-side effects.
A clue about the effect of aminoglycosides on orangutan cells came from studies with inter-specifies cybrids (xenocybrids). Thus, some xenocybrid clones were obtained by fusing orangutan rho0 cells and human cytoplasts (enucleated cells) (36). However, no clone was obtained by using human rho0 cells and orangutan cytoplasts (37). Maybe, beside the defective interactions between nuclear chromosomes- and mtDNA-encoded products, aminoglycosides in the cell culture behave as another stress condition for orangutan mtDNAs.
To study the effect of aminoglycosides on orangutan cells, we analyzed four cell lines: (i) orangutan (Pongo pygmaeus) primary fibroblasts from skin explants; (ii) human osteosarcoma 143B cybrids harboring the m.1555G mutation as a positive control; (iii) bonobo (Pan paniscus) primary fibroblasts from skin explants; and (iv) human osteosarcoma 143B cybrids harboring the m.1555A nucleotide as negative controls.
We first checked the species origin and the genotypes of these cell lines by sequencing. The m.1542C nucleotide has not been found in >3000 human (38) and 37 bonobo (4,16,39,40) sequences, but it is present in orangutan mtDNA. The m.1556T has not been found in >3000 human (38) and 17 orangutan (4,5,16) sequences, but it is found in bonobo mtDNA. Then, the T/T, T/C and C/C genotypes at m.1542/m.1556 positions confirm that these cell lines are from bonobo, human and orangutan, respectively (Fig. 2B). The m.1551A nucleotide characterizes orangutan primary fibroblasts as belonging to Pongo pygmaeus (Fig. 2A). Moreover, the m.1555G nucleotide in this orangutan sequence, along with the other 17 sequences from Figure 2A, reconfirm that this is not a sequencing error or a recent polymorphism.
When oxidative metabolism is impaired, cells cultured in galactose medium without glucose cannot synthesize the bulk of their ATP requirements by glycolysis and the cell growth is inhibited (41). For this reason, we grew these cells without or with paromomycin (2 mg/ml) in the galactose medium. As shown in Supplementary Material, Figure S4, after 72 h, human mutant and wild-type orangutan, but not wild-type human or bonobo, cells with the antibiotic showed a slower growth rate. Another aminoglycoside, geneticin (G418, 1 mg/ml), had a stronger effect on orangutan cells. Thus, orangutan cells died after 48 h with this antibiotic but the growth of bonobo and human cells was not affected (data not shown).
In previous experiments, it was shown that aminoglycosides decreased cell growth in human cybrids harboring the m.1555A>G mutation. However, the OXPHOS function was not affected (35). It would be possible that this growth impairment was not an OXPHOS-mediated effect. Indeed, the abundance of positively charged amino groups predestines aminoglycosides to bind negatively charged molecules, such as non-ribosomal RNAs or phospholipids (42,43). To check whether this growth deficit in orangutan cells was specifically due to an OXPHOS defect and since the catalytic subunits of the OXPHOS complex IV (CIV) are encoded in the mtDNA, we determined the in gel CIV activity. This activity was lower in orangutan cells treated with aminoglycosides (Fig. 4A). In gel CIV activity is not a quantitative assay. Therefore, we measured CIV-specific activity by spectrophotometry. This activity was significantly lower in human mutant and wild-type orangutan cells treated with paromomycin, but exhibited no difference in wild-type human and bonobo cells (Fig. 4B).
In the presence of aminoglycosides, the conformation of rRNA becomes altered and erroneous amino acids can be incorporated to the proteins (44). Therefore, to check whether lower CIV activities were due to anomalous CIV subunits or a lower CIV amount, we determined CIV levels. We observed that CIV quantities were also significantly lower in human mutant and wild-type orangutan cells with the antibiotic (Fig. 5A). Moreover, there was a significant correlation (r = 0.774, P = 0.021) between CIV activity and quantity (Fig. 5B). Hence, lower CIV activities in human mutant and wild-type orangutan cells are mainly due to lesser CIV quantities. The aminoglycoside effect was dependent on the doses. Thus, 4 mg/ml of paromomycin provoked a larger decrease both in CIV activity (47%) and quantity (17%) (data not shown). Streptomycin (100 ng/ml) also produced a decrease in CIV activity (34%) and quantity (19%) in orangutan cells (data not shown). This meant that the inhibitory effect was not specific for paromomycin.
The first step in the CIV assembly pathway is the insertion of the p.MT-CO1 polypeptide into the inner mitochondrial membrane (45). Therefore, to check whether the lower CIV amount was due to lower p.MT-CO1 levels, we measured the quantity of this polypeptide by western blot. The results shown in Figure 5C indicated that paromomycin (2 mg/ml) lowered the amount of this subunit in orangutan cells. Similar to the effect on CIV activity and quantity, an increase in paromomycin concentration (4 mg/ml) produced a larger decrease in the amount of this subunit in wild-type orangutan and human mutant cells (Supplementary Material, Fig. S5). Because p.MT-CO1 is only one of the 13 mtDNA-encoded polypeptides, to confirm that this aminoglycoside could alter the levels of other mtDNA-encoded polypeptides, we performed a mitochondrial protein synthesis assay and observed that all of them were decreased in human mutant and wild-type orangutan cells (Fig. 5D). Therefore, aminoglycosides negatively affect OXPHOS function in orangutan cells.
A recent founder effect, the inefficiency of natural selection or the presence of compensatory mutations do not appear to explain the fixation of the human m.1555A>G pathological mutation in the orangutan mtDNA. Moreover, results obtained from human cybrids, with the same nuclear background than those here used, showed that the mutation per se was not sufficient to cause an OXPHOS dysfunction in human cybrids (35). However, the exposition to aminoglycosides provokes a decrease in the OXPHOS function of orangutan cells. Because other primates less phylogenetically close to humans, such as Erythrocebus patas, are susceptible to aminoglycosides (46), pharmacokinetic factors can probably be ruled out as responsible for the fixation of the m.1555G mutation in orangutan. Therefore, it is possible that incomplete penetrance of this mutation along with the absence in the orangutan habitat of xenobiotics that interact with this mt-rRNA position are the reason for the fixation of this human pathological mutation in orangutan. Interestingly, gorillas, chimps and humans have shared the same environment because they are African hominids but orangutan's environment is different. They live in South East Asia.
More importantly, m.1555G nucleotide is also the wild-type allele in other 8 out of 316 examined mammal species: a rodent (Thryonomys swinderianus), a ruminant (Budorcas taxicolor) and six bears (Arctodus simus, Melursus ursinus, Ursus arctos, Ursus maritimus, Ursus thibetanus and Ursus thibetanus ussuricus). Because aminoglycosides are commonly used in veterinary medicine (47), these mammals might be at risk of suffering a mitochondrial disorder if treated with aminoglycosides, although this possibility should be confirmed.
On the other hand, these results can be important for human health. Ancient and highly prevalent mtDNA mutations are usually ruled out as pathological mutations. However, as seen in orangutans, this kind of mutations could have phenotypic effects, depending on environmental conditions. Therefore, it is possible that similar mutations, such as some human haplogroup-defining mt-rRNA single nucleotide polymorphisms, are deleterious in association with particular xenobiotics. In this sense, ribosomal antibiotics started to be used in the 1940s to fight bacteria. Because the protein synthesis apparatuses of mitochondria and bacteria are very alike, these compounds might provoke side effects on individuals with particular mtDNA genotypes (48,49). For example, platelet CIV activities and quantities from patients that survived 6 months to sepsis were significantly higher than those of non-survival patients (50) and mtDNA haplogroups were found to modify the chance of sepsis survival at 180 days (51). As antibiotics are frequently used in sepsis patients, it would be possible that the interaction between mt-rRNA genetic variants and particular ribosomal antibiotics behave as a risk factor for sepsis survival.
MATERIALS AND METHODS
Sequence alignments were performed using the multiple sequence alignment program ClustalW2 (http://www.ebi.ac.uk/Tools/msa/clustalw2/). CIs were estimated by alignment of all 316 mammal MT-RNR1 sequences pulled from completely sequenced mtDNAs published in Genbank (http://www.ncbi.nlm.nih.gov/sites/entrez) through the end of 2010. We also used the ConSurf tool (http://consurf.tau.ac.il) to calculate the CI while taking into account the topology of the phylogenetic tree. Molecular models were obtained with the RasMol2.6 program (PDB 2WDG and 1QYR).
Growth conditions, molecular and statistics analysis
Cells were cultured in the galactose medium. Cell culture, molecular-genetics studies, biochemical analysis and statistics were performed according to protocols previously described (52–54). To test the phenotypic effect of aminoglycosides, cells were grown in the presence or absence of different concentrations of paromomycin, geneticin or streptomycin.
This project was supported by grants from Instituto de Salud Carlos III-FIS (PI08-0264, PI10-00662) and Departamento de Ciencia, Tecnología y Universidad del Gobierno de Aragón y Fondo Social Europeo (Grupos Consolidados B33, PIPAMER10-010 and Fundación ARAID-Programa de Apoyo a la I+D+I para jóvenes investigadores 2010). The CIBERER is an initiative of the ISCIII.
We would like to thank Professor Werner Schempp in the Institute of Human Genetics at University of Freiburg (Freiburg, Germany) for his kind gift of orangutan and bonobo primary fibroblasts and Dr Covadonga Gómez-Díaz in the Servicio de Otorrinolaringología at Hospital Miguel Servet (Zaragoza, Spain) for her interpretation of results on orangutan hearing.
Conflict of Interest statement. None declared.