**3. Discussion**

All mammalian cells produce ATP by glycolysis and OXPHOS. The balance between these processes is di fferent in each cell type [**?** ], and its disturbance can cause disease. Here, we show a clearly compromised OXPHOS function in LS patient fibroblasts and neurons; however only basal respiration was altered in patient iPSCs. This indicates that the defect in iPSC lines is less evident than in fibroblasts or neurons, probably due to the main dependence of iPSCs on glycolysis [**???** ]. This conclusion is supported by the low levels of oxygen consumed by iPSCs: basal respiration in control iPSCs was approximately half and maximum respiration one-third of that of control fibroblasts.

Theoretically, a defect in complex I of the respiratory chain could be bypassed by an increase in the electron flow through complex II that could restore ΔΨmit and ATP levels. In this regard, succinate fuels complex II but it cannot cross biological membranes and does not reach mitochondria. Recently, several cell-permeable succinate prodrugs have been described [**?** ]. As LS patient neurons manifested a decreased complex I respiration, we attempted to rescue that phenotype using one of the succinate prodrugs, NV241, described by Ehinger et al. [**?** ]. Administration of the succinate prodrug increased both routine (basal) respiration and the complex II contribution to maximal respiration to similar levels in patient and control iPSC-derived neurons. However, patient neurons responded with increased respiration, possibly uncoupling, to the drug vehicle (DMSO), which rendered the drug response data inconclusive for the patient iPSC-derived neurons. For the control neurons, the stimulating e ffects of complex II on respiration were significant. In the initial publication describing the cell-permeable succinate prodrugs, respiration and spare respiratory capacity were increased by prodrug administration in LS patient fibroblasts with a recessive *NDUFS2* mutation.

The question remains as to why iPSCs with their low dependence on OXPHOS manifest a combined RC defect with decreased activities of complexes I and III, while fibroblasts, which rely on mitochondrial OXPHOS more than iPSCs [**?** ], do not show a diminished activity of complexes III and IV similar to muscle. Due to technical problems it was not possible to measure complex I deficiency. Therefore, we cannot discard the possibility of an isolated complex I deficiency in fibroblasts. Sequencing of mitochondrial complex I related genes, however, ruled out the possibility that control fibroblasts had mutations that were not previously detected. One possible explanation for normal specific activity of complexes III and IV in fibroblasts is that the alteration of the ETC function in patient fibroblasts could lead to a decreased ΔΨmit, which in turn could induce an increase in the selective elimination of abnormal mitochondria through mitophagy and mask the underlying defect when enzymatic activities are measured. We do observe, as the main finding in patient fibroblasts, diminished mitochondrial mass; considering that mitophagy is the core mechanism of mitochondrial quality and quantity control [**?** ], it is reasonable to think that mitophagy could account for the decrease in mitochondrial mass. This compensatory e ffect has already been described in other fibroblasts with OXPHOS mutations [**????** ], and its protective role during pathogenesis is recognized [**?** ]. However, in one of these studies, no increase of mitophagy in fibroblasts harboring the m.13513G>A mutation was observed [**?** ]. We therefore cannot rule out the possibility that the observed reduction in protein levels and activity in CS could be the consequence of a decreased mitochondrial biogenesis or an increase in the random elimination of mitochondria by bulk macroautophagy. This alternative explanation, in which no selection against abnormal mitochondria occurs, would not, however, explain the normal activity found in complexes III and IV in the fibroblasts.

In contrast to what happen in fibroblasts, iPSCs, similar to cancer cells, are considered to mainly rely on glycolysis [**???** ]. In theory, the low OXPHOS function in iPSCs would lead both in controls and patients to a decrease of ΔΨmit and the consequent activation of mitophagy [**? ?** ]. It has been demonstrated that iPSCs display higher ΔΨmit than di fferentiated cells [**? ?** ], even using hydrolase activity of ATP synthase to maintain ΔΨmit [**?** ]. The maintenance of normal ΔΨmit would prevent mitophagy and the compensatory e ffect on patient iPSCs; as a consequence, no abnormal complexes would be removed, and a compromised activity of complexes I and III would be detected even when the defect is not evident at physiological conditions.

For disease modeling, it is important to recapitulate the principal pathological features of the disease. LS is characterized by the presence of bilateral symmetric necrotic areas in the basal ganglia or the brain stem which correspond with regions of demyelination, neuronal death and astrocytic gliosis [**? ?** ]. Diseased neurons manifested a compromised respiratory capacity and evident neuronal death after being replated on mouse astrocytes. These results indicate that this in vitro model, at least in part, recapitulates the in vivo phenotype of LS. Moreover, increased lactate levels in blood or cerebrospinal fluid are common criteria in the diagnosis of MD. Here, both cell types analyzed (fibroblasts and iPSCs) manifest elevated production of lactate, independently of the main pathway used by the cell type to generate energy.

Patch clamp recordings revealed normal electrophysiological function of individual patient neurons, indicating that mtDNA mutation m.13513G>A does not impair Na<sup>+</sup> or K<sup>+</sup> currents or the ability of neurons to fire APs. This is important, because it indicates that electrically functional neurons can be derived from LS patient NSCs. However, those neurons manifested a marked dysregulation of calcium homeostasis.

Ca2+ is an intracellular signal responsible for controlling numerous cellular processes [**?** ]. The Ca2+ signaling network requires first the ON mechanisms, by which there is a 10-fold rise of cytoplasmic Ca2+, followed by the OFF mechanisms, which ensure that Ca2+ is recovered to basal levels [**?** ]. The increase of cytoplasmic calcium is the consequence of the entry of external Ca2+ through di fferent types of channels (VOCs, ROCs, SOCs) and the release of Ca2+ from internal stores [**?** ]. At the same time, the OFF mechanisms include the extrusion of calcium to the outside, its return to internal stores (endoplasmic reticulum and mitochondrion) and its association with cytosolic bu ffers (parvalbumin, calbindin-D28k, calretinin) [**? ?** ]. In the presynaptic membrane of neurons, the rise of cytoplasmic Ca2+ triggers the release of neurotransmitters. As important as the rise is the bu ffering of that calcium to ensure the possibility of a new activation. In fact, sustained high concentrations of cytoplasmic Ca2+ are associated with neuronal death [**?** ] through apoptosis if there is ATP available or necrosis if there is ATP depletion [**? ?** ]. It has been postulated that the addition of KCl induces a slow depolarization of membrane potential (ΔΨ), consequent activation of Ca2+ voltage-operated channels (VOCs) and Ca2+ influx into the cell [**? ?** ]. Cells that respond to this KCl are called evoked cells [**?** ]. The analysis of the response of evoked cells harboring the mtDNA mutation m.13513G>A manifested a slower bu ffering capacity and an increased refractory period. This is not surprising, because mitochondria play a significant role in shaping global Ca2+ signals [**?** ] through direct or indirect participation [**?** ]. The decreased bu ffering capacity that we observe can lead to higher amounts of calcium in the cytoplasm and consequent neuronal necrosis observed in LS patients [**?** ]. The concrete mechanism by which this mtDNA mutation causes inappropriate bu ffering of calcium remains unknown. Recently, neural progenitor cells (NPCs) harboring the homoplasmic m.9185T>C mutation in the *MT-ATP6* causative of episodic paralysis with spinal neuropathy (NPC\_ATP6) has been reported [**?** ]. In contrast to our results, NPC\_ATP6 mutated cells manifested a decreased calcium release (ON mechanisms) rather than the decrease in calcium bu ffering observed here (OFF mechanism). These di fferences could be the consequence of the ΔΨmit, in which mutations that a ffect the ATP synthase increase ΔΨmit and a ffect calcium release, while mutations in the ETC provoke a diminishment in ΔΨmit and a ffect calcium bu ffering. In support of our findings, iPSC-derived neurons from other neurodegenerative diseases such as Parkinson's disease (PD) or frontotemporal lobar degeneration tauopathy (FTLD-Tau) have also shown an impaired calcium homeostasis as the underlying pathogenic mechanism [**? ?** ]. It is also very interesting that the necrosis observed after myocardial ischemia or stroke [**? ?** ] occurs in response to a lack of oxygen to the mitochondria, suggesting that MD, similarly to PD and TFLD-Tau and other conditions such as myocardial infarction or ischemic stroke, could occur through the same calcium overload process, which leads to necrosis. Although other studies have concluded that mutations in the *MT-ND5* gene in fibroblasts or cybrids can be associated to calcium handling defects [**? ?** ], the phenotype of the neurons, the cell type in which the disease occurs, remained unexplored. Here, we demonstrate defective OXPHOS and a clear dysregulation in calcium homeostasis in neurons harboring the m.13513G>A mutation, with profound consequences for the pathology.

## **4. Materials and Methods**

#### *4.1. Cell Culture, iPSC Generation and Characterization*

This study was approved by the Institutional Ethical Committee of the Autonoma University of Madrid according to Spanish and European Union legislation and complies with the principles of the 1964 Helsinki declaration. Normal human dermal fibroblasts (NHDFs) were purchased from Promocell (C12300) and maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% FBS, 50 U/mL penicillin and 50 μg/mL streptomycin. Leigh fibroblasts from a described patient harboring the heteroplasmic mutation in the mtDNA m.13513G>A (p.D393N) were kindly provided by Dr. Francina Munell from the Hospital Universitario Vall d'Hebron (Barcelona, Spain), with prior informed consent for study participation. LS fibroblasts were maintained in the same growth medium as NHDF but supplemented with 50 μg/mL uridine. A previously described iPSC line, LND554SV.4 [**?** ], and a control iPSC line reported here (N44SV.1) have been derived, maintained and characterized following the protocols described in [**?** ].
