*3.4. Topology of the Phenylalanine Residue Targeted by the m.8909T*>*C Mutation*

Complete high-resolution structures of mitochondrial ATP synthase were described recently [36,47–49,52–54]. Near the middle of the membrane are located two universally conserved residues, an acidic one in subunit *c* (*c*E<sup>59</sup> in yeast) and a positively charged arginine residue in subunit *a* (*a*R<sup>176</sup> in yeast) that are functionally essential [36,55–57] (Figure 4B). Two hydrophilic clefts, one on the *p*-side and *n*-side of the membrane facilitate proton movement from one side of the membrane to the other. Inside the *p*-side channel, *c*E<sup>59</sup> takes a proton from the intermembrane space (IMS) and releases it into the *n*-side cleft after an almost complete rotation of the *c*-ring. The phenylalanine residue 145 (128 in *H.s.*) that is changed into serine by the m.8909T>C mutation is located at the bottom of the *n*-side cleft within a cluster of hydrophobic residues (*aW126, aF141, aF142, aL144, aF145, aY166*) beneath the helical domain of subunit *a* (*a*H5) that runs along the subunit *c*-ring (Figure 4C). Replacement of *aF<sup>145</sup>* by a polar serine residue may weaken these hydrophobic interactions, and this is possibly responsible for the partial defects in ATP synthase assembly/stability observed in the *aF145S* mutant (as described above).

α β β **Figure 4.** Evolutionary conservation of the phenylalanine residue targeted by the m.8909T>C mutation, and its topology in yeast ATP synthase. (**A**) Amino-acid alignments of subunits *a* from *Homo sapiens* (*H.s.*), *Bos taurus* (*B.t.*), *Xenopus laevis* (*X.l.*), *Arabidopsis thalian*a (*A.t.*), *Schizosaccharomyces pombe* (*S.p.*), *Podospora anser*ina (*P.a.*), *Yarrowia lipolytica* (*Y.l.*) and *Saccharomyces cerevisiae* (*S.c.*). At the top and bottom, the residues are numbered according to the *H.s.* protein and mature *S.c.* protein (i.e., without the first 10 N-terminal residues that are cleaved during assembly [42]), respectively. Strictly conserved residues are in white characters on a red background while similar residues are in red on a white background with blue frames. α-helices (*a*H4-5 and *a*Hβ) in the *S.c.* protein marked above the amino-acid alignments are according to [36]. The essential arginine (*a*R<sup>159</sup> in humans, *a*R<sup>176</sup> in yeast) is on a blue background. The yeast *aF<sup>145</sup>* residue corresponding to *a*F<sup>128</sup> in *H.s.* targeted by the m.8909T>C mutation is on a green background. The grey arrows mark the residues belonging to the hydrophobic cluster surrounded by *a*H4, *a*Hβ and *a*H5. (**B**) View of the entire *c*-ring and subunits *a, b, 8, I, f and k* from the IMS and the pathway along which protons are transported from the intermembrane space to the mitochondrial matrix. The side chains of the residues that are essential (*a*R<sup>176</sup> and *c*E59) and important (*a*E162, *a*D244, *a*E223, H185) to this transfer are drawn as stick with their carbon atoms in white. The *p*-side and *n*-side clefts are shown as grey surfaces. (**C**) Enlargement of the region where is located the *aF145S* mutation. The mutated *aF<sup>145</sup>* residue in green belongs to a hydrophobic cluster of six residues (*aW126, aF141, aF142, aL144, aF145, aY166*), which may stabilize or ease the folding of the bottom of the *n*-side cleft and the proper positioning of *a*H5 along the *c*-ring.

### **4. Discussion**

With the advent of next generation sequencing methods, numerous novel variants of the mitochondrial DNA (mtDNA) are continuously identified in patients suffering from mitochondrial disorders. They are frequently found in only a limited number of cases, sometimes in a single individual (as with the case here reported), which makes it difficult to conclude about their pathogenicity. In a study aiming to probe the possible implication of mtDNA alterations in renal disease in the region of Nanjing, we examined more than 5000 patients with a biopsy-proven kidney dysfunction. Patients retained for entire mtDNA sequencing additionally showed at least two symptoms commonly observed in mitochondrial diseases like diabetes mellitus, deafness, neuromuscular and cardiac manifestations. Another criterion was the presence of abnormal mitochondrial ultrastructure and decreased cytochrome *c* oxidase (COX) and NADH (Complex I) dehydrogenase activities in kidney biopsies. In the patient with all these features here reported, we identified a novel variant in the mitochondrial *ATP6* gene, m8909T>C. It was homoplasmic in blood, urine sediments, and kidney (Figure 2B). It was detected in combination with the well-known pathogenic m.3243A>G mutation in a leucine tRNA gene (*MT-TL1)* [38], at a lesser abundancy (50–90%). There is no doubt that defects in mitochondrial translation induced by the m.3243A>G mutation largely contributed to the disease process and the severe decreases in CI and CIV activities, with only a minimal loss of SDH activity as was observed in other patients carrying this mutation [58].

The m.8909T>C variant leads to replacement of a well conserved phenylalanine residue with serine in ATP synthase subunit *a* (*aF128S*) (Figure 4A), which prompted us to investigate the consequences of an equivalent of this mutation (*aF145S*) in yeast. Several lines of evidence demonstrate a partial impairment of ATP synthase function in the mutant compared to wild type yeast: (i) respiratory growth showed a higher sensitivity to suboptimal concentrations of oligomycin, a chemical that inhibits ATP synthase; (ii) as was observed in many yeast ATP synthase defective mutants, the rate of mitochondrial oxygen consumption was diminished; (iii) ATP was produced less rapidly; and (iv) the presence in BN gels of partial ATP synthase assemblies (free F<sup>1</sup> and *c*-ring particles) attested for a compromised ability of the mutated subunit *a* to be stably incorporated into ATP synthase (Figure 3). According to recently published atomic structures of yeast ATP synthase [36,54], these effects are presumably the consequence of a disorganization of a cluster of hydrophobic residues supposedly important to help subunit *a* to adopt a stable functional conformation around the *c*-ring. Similar ATP synthase defects were previously observed in yeast models of *MT-ATP6* mutations with a well-established pathogenicity, like m.9176T>C [13] and m.8993T>C [11]. It is thus a reasonable hypothesis that the m.8909T>C variant could by itself have detrimental consequences on human health.

In line with this hypothesis, the patient here described showed a very severe clinical presentation in comparison to 35 previously reported patients suffering from kidney problems and presumed to carry only the m.3243A>G mutation (in most of them the mtDNA was not entirely sequenced). The median onset age of disease was 26 years (Table S1). More than 50% of these patients were first diagnosed with a renal disease due to persistent proteinuria. Glomerular lesions were observed in 21 patients, FSGS in 15 patients, and 3 cases showed TIN (Table S2) [59–81]. At the time of diagnosis, two relatively aged patients (41 and 47 years old) developed ESRD. A total of 21 patients were recorded with a median follow-up period of 72 months (range 24–240 months). Kidney function remained stable in two patients after 24 and 72 months, respectively, and one patient died of heart failure 60 months after diagnosis. Eighteen patients evolved toward renal failure or ESRD (Table S1) within 10-years of follow-up after diagnosis (Figure S1). Our patient, carrying both m.8909T>C and m.3243A>G, developed a kidney disease at the age of 14 and showed an extremely rapid progression to ESRD and death after only 8 months after diagnosis of a nephropathic syndrome.

As the m.8909T>C mutation was homoplasmic in the analyzed cells and tissues, it is likely that it is maternally inherited. Unfortunately, the patient's mother did not consent to be sequenced and followed medically. She was apparently healthy at the time her daughter was admitted at the hospital. We had no more contact with her since the death of her child. From her apparent good health, at least

at the age she brought her daughter at the hospital, we may conclude that the m.8909T>C mutation has no dramatic consequences on mitochondrial function, which is consistent with the relatively mild effects of an equivalent of this mutation on yeast ATP synthase. However, when combined to a more severe allele like m.3243G>A, it may have the potential to accelerate a disease process. Previous studies already concluded that well-known pathogenic *ATP6* variants can act in synergy with other genetic determinants in patients with a very severe clinical presentation. For instance, while alone it usually provokes mild clinical phenotypes [82], the m.9176T>C mutation was identified in a case of fulminant and fatal Leigh syndrome [23]. As there was no major difference in the amount of mutated mtDNA in relative's probands, it was concluded that additional mitochondrial or nuclear genetic determinants were responsible for the different phenotypic expression. A second study similarly reported a rapid clinical evolution leading to sudden infant death syndrome (SIDS) in families segregating the m.8993T>G mutation [83] or the m.10044A>G mutation of the mt-tRNAGly gene [84], which suggested that mtDNA abnormalities should be considered as contributing to SIDS. Because of a lack of data from the mother of the patient here described, it would be premature to claim that m.8909T>C is a pathogenic mutation. However, considering the very severe clinical presentation of this patient, and the detrimental consequences of m.8909T>C on yeast ATP synthase, it is a reasonable proposal that this mutation has the potential to impact human health, in particular when combined to other genetic abnormalities that compromise mitochondrial function.
