The respiratory complex I (electrogenic NADH:quinone oxidoreductase) has been considered to act exclusively as a H+ pump. This was questioned when the search for the NADH-driven respiratory Na+ pump in Klebsiella pneumoniae initiated by Peter Dimroth led to the discovery of a Na+-translocating complex in this enterobacterium. The 3D structures of complex I from different organisms support the idea that the mechanism of cation transport by complex I involves conformational changes of the membrane-bound NuoL, NuoM and NuoN subunits. In vitro methods to follow Na+ transport were compared with in vivo approaches to test whether complex I, or its individual NuoL, NuoM or NuoN subunits, extrude Na+ from the cytoplasm to the periplasm of bacterial host cells. The truncated NuoL subunit of the Escherichia coli complex I which comprises amino acids 1–369 exhibits Na+ transport activity in vitro. This observation, together with an analysis of putative cation channels in NuoL, suggests that there exists in NuoL at least one continuous pathway for cations lined by amino acid residues from transmembrane segments 3, 4, 5, 7 and 8. Finally, we discuss recent studies on Na+ transport by mitochondrial complex I with respect to its putative role in the cycling of Na+ ions across the inner mitochondrial membrane.
There are three types of respiratory NADH:quinone oxidoreductases: the proton- or Na+ ion-translocating complex I from mitochondria (termed NDH-1 in bacteria) [1,2], the non-electrogenic NADH:quinone oxidoreductases (NDH-2) found in eukaryotes  and prokaryotes , and the Na+ ion-translocating NADH:quinone oxidoreductases (Na+-NQR) exclusively found in bacteria [5,6]. These enzymes catalyse the oxidation of NADH and reduction of Q (ubiquinone) to QH2 (quinol), which acts as a reducing substrate of enzyme complexes further along the respiratory chain. NADH:quinone oxidoreductases of the NDH-2 type do not participate in the generation of a transmembrane voltage, i.e. they are non-electrogenic. Complex I (or NDH-1) and Na+-NQR in contrast tap into the exergonic energy released by the redox reaction to transport H+ or Na+ across the inner membrane of mitochondria or bacteria. The accumulation of positively charged ions on the outside of the membrane leads to an electrical and chemical imbalance, summarized as ΔμH+ [PMF (protonmotive force)] or ΔμNa+ [SMF (sodium motive force)]. Prokaryotes may contain all three types of NADH:quinone oxidoreductases, but the expression of each is usually tightly regulated . The chromosome of Escherichia coli, for example, features genes for both NDH-1 and NDH-2. Bacteria of the genus Vibrio contain NDH-2 and Na+-NQR and a few bacteria such as Klebsiella pneumoniae harbour all three types of enzymes .
Looking for Na+-NQR, but finding complex I
It was Peter Dimroth who began to search for an Na+-NQR in K. pneumoniae . This enterobacterium operates a unique Na+ cycle during the anaerobic fermentation of citrate which is taken up in symport with Na+ by a citrate carrier . The chemical Na+ gradient which drives accumulation of citrate in K. pneumoniae is generated by its Na+-translocating oxaloacetate decarboxylase, a membrane-bound Na+ pump which converts the oxaloacetate (formed from citrate by citrate lyase) into pyruvate and CO2 . No NAD(P)H is formed in this pathway, yet K. pneumoniae needs reducing equivalents for the biosynthesis of cell material starting from citrate, so Peter Dimroth speculated about the existence of a Na+-NQR in K. pneumoniae. In the reverse mode, this enzyme could reduce NAD+ to NADH driven by the SMF provided by the oxaloacetate decarboxylase . It was later shown that NAD(P)H in citrate-fermenting K. pneumoniae is provided by the NAD(P)+-reducing hydrogenase which does not require the SMF . Still, these cells exhibited Na+-stimulated NADH oxidation activity concomitant with the uptake of Na+ into inverted membrane vesicles which did not collapse in the presence of protonophores . These features are the hallmarks of a respiratory primary Na+ pump like the Na+-translocating NQR first discovered by Tokuda and Unemoto  in Vibrio alginolyticus. Quite unexpectedly, the Na+-stimulated NADH:quinone oxidoreductase enriched from solubilized membranes of citrate-grown K. pneumoniae cells contained subunits from complex I, but not from the Na+-NQR . NADH-driven Na+ transport activity assigned to complex I was also observed in the related enterobacterium Escherichia coli . Reconstitution of the enzyme in proteoliposomes supported our initial hypothesis that complex I from K. pneumoniae was capable of Na+ transport instead of, or in addition to, proton translocation . Our view was challenged by Bertsova and Bogachev  who, upon studying aerobically grown cells from K. pneumoniae, detected an enzyme belonging to the NQR family. They observed H+ transport by complex I from K. pneumoniae and concluded that “the NQR-type enzyme, not the NDH-1-type enzyme, catalyses sodium-dependent NADH oxidation in K. pneumoniae” . That study raised severe doubts over our earlier finding of a Na+-transporting complex I in K. pneumoniae grown anaerobically on citrate . Bogachev and co-workers later showed that the “Na+-NQR of K. pneumoniae is not induced during anaerobic growth on citrate as the sole carbon and energy source” , but the authors did not discuss the implications of their findings with respect to the nature of the Na+-transporting NADH dehydrogenase clearly present in those cells . If Na+-NQR is not present in citrate-fermenting K. pneumoniae cells, what is the identity of the enzyme catalysing NADH-driven Na+ extrusion? In our view, it is the Na+-translocating complex I. The results reported by Bogachev and colleagues  are in perfect accordance with our earlier finding that complex I represents the respiratory Na+ pump in K. pneumoniae grown anaerobically on citrate .
Reports describing Na+ transport by complex I from different organisms have become more frequent and are summarized in the following sections. Furthermore, alternative routes for the coupling cations (Na+ and/or H+) through the membrane-bound NuoL subunit of the E. coli complex I  are presented. We then discuss the possible implications of a Na+-transporting complex I for the cycling of Na+ ions across the inner mitochondrial membrane. Complex I and the Na+-NQR are not related to each other with respect to their amino acid sequences or cofactor composition. We recently reported the first 3D structure of the Na+-NQR , but the resolution of 16 Å (1 Å=0.1 nm) is yet too low to speculate about the mechanism(s) of redox-driven Na+ translocation. For the description of the subunits, cofactors and catalytic properties of Na+-NQR, we refer the reader to [5,6,21].
Architecture of complex I from mitochondria and bacteria
Complex I is L-shaped and consists of a hydrophilic peripheral domain which reaches into the mitochondrial matrix (or bacterial cytosol) and a membrane domain comprising numerous membrane-spanning subunits. Bovine complex I is made of 45 different polypeptides , and complex I from the yeast Yarrowia lipolytica is composed of 40 different polypeptides . The mitochondrial complex I contains 14 core subunits conserved from prokaryotes, which can be divided into seven hydrophilic and seven hydrophobic proteins. The latter are mitochondrially encoded and correspond to the membrane-embedded NuoA (ND3), NuoH (ND1), NuoJ (ND6), NuoK (ND4L), NuoL (ND5), NuoM (ND4) and NuoN (ND2) of the bacterial enzyme (corresponding subunits from mitochondrial complex I in parentheses). The hydrophilic core subunits can be attributed to the peripheral arm of complex I which consists of the dehydrogenase fragment (NuoE, NuoF and NuoG) and the connecting fragment (NuoB, NuoC, NuoD and NuoI). Sazanov and Hinchliffe  reported X-ray structures of the peripheral arm of complex I from Thermus thermophilus, of the holo-complex I from T. thermophilus [25,26], and of the membrane arm of the E. coli complex I . Crystallographic analysis of mitochondrial complex I, double the size of the bacterial complex I, revealed a similar architecture with respect to the conserved core regions .
Proposed mechanisms of proton transport by complex I
The groundbreaking structures of complex I from bacteria and mitochondria lay the foundation for deciphering the mechanism of redox-driven cation translocation by complex I. A mechanism including conformational change-driven cation transport by three cation channels through the membrane-bound NuoL, NuoM and NuoN subunits is discussed in [27–31], but experimental evidence is still lacking. From the structural similarity of the transmembrane segments 4–8 in NuoL (or NuoM, or NuoN) with transmembrane segments of the NuoH subunit, Sazanov and co-workers proposed that NuoH forms part of a fourth cation channel through complex I . Q binds at the interface between the peripheral arm and the membrane part  to accept electrons donated by the FeS centres and, initially, by FMN and NADH, located on peripheral subunits. A unique amphipathic helix (helix HL) reaches from NuoL situated at the tip of the membrane arm to the membrane domain beneath the peripheral arm where the Q-binding site is situated . Helix HL on the cytoplasmic side of the membrane arm, together with the β-hairpin–helix element formed by neighbouring NuoL/NuoM/NuoN subunits on the periplasmic side, might function as transducers of conformational changes triggered by redox reactions in the peripheral arm of complex I . The large membrane-bound NuoL, NuoM and NuoN subunits share a common ancestor and exhibit sequence similarity to subunits of Na+/H+ antiporter complexes belonging to the Mrp family . This similarity is also evident from the arrangement of transmembrane helices in NuoL/NuoM/NuoN  and in the single-subunit Na+/H+ antiporter NhaA . Subunits NuoL/NuoM/NuoN comprise several conserved charged or polar amino acid residues within their membrane region which are situated in close proximity to broken or partially unwound transmembrane helices  (Figure 1). This is a typical feature shared with NhaA, which binds Na+ in a binding site formed by two adjacent interrupted transmembrane helices. In NhaA, this critical region within the membrane-embedded part of the protein undergoes a conformational change during exchange of Na+ against H+ [34,35]. The evolutionary relationship of subunits NuoL/NuoM/NuoN  with secondary transporters belonging to the Mrp family was used as an argument for their sole participation in proton transport [28,30]. Yet, these cation/H+ antiporters have the capacity to transport Na+ (or K+) in addition to protons [31,36,37], so passage of Na+ (or K+) through NuoL/NuoM/NuoN is equally feasible.
Putative cation channels through the membrane-bound subunit NuoL from E. coli complex I
The directionality of Na+ translocation by complex I from bacteria
How has Na+ transport by bacterial complex I been measured? With the purified complex I from K. pneumoniae reconstituted in proteoliposomes with its NADH-oxidizing (cytoplasmic) side facing the external medium, we followed NADH oxidation and quinol formation simultaneously using a diode array spectrophotometer . In an assay performed in parallel with the same batch of proteoliposomes, Na+ entrapped in liposomes was determined during the first 60 s after the start of the reaction by addition of NADH to the external medium. To quantify the Na+ in the liposomes, the external Na+ was removed by passage of aliquots over a cation-exchange material. There is a delay before the first data point, which is due to the time required for loading of an aliquot onto the cation exchanger, and its elution with H2O. The Na+ content in the eluate is determined by atomic absorption spectroscopy, or by γ-counting of 22Na+ added as tracer. Using this approach, the K. pneumoniae complex I exhibited specific activities of 2.4 μmol·min−1·mg−1 for quinol formation and 4.7 μmol·min−1·mg−1 for Na+ translocation. The Na+ transport proceeded from the ‘cytoplasmic’ to the ‘periplasmic’ side of complex I . Using proteoliposomes containing both complex I and the Na+-dependent F1Fo ATP synthase (with the NADH-oxidizing and ATP-hydrolysing sides facing the external medium), complex I catalysed Na+ uptake during NADH:quinone oxidoreduction coupled to ATP synthesis or QH2:NAD+ oxidoreduction by complex I driven by the SMF established by the ATPase, depending on the experimental conditions . In the latter experiment, reverse electron transfer by complex I was accelerated by the transport of Na+ from its ‘periplasmic’ towards its ‘cytoplasmic’ side. Complex I from K. pneumoniae thus has the capacity to translocate Na+ in either direction, depending on the substrate availability and the applied membrane potential.
Pereira and co-workers used 23Na NMR spectroscopy to analyse the change in external Na+ concentration induced by complex I embedded in native membrane vesicles with its ‘cytoplasmic’ side facing the external medium . This technique allows determination of the external Na+ concentration in situ, so, in contrast with our method , there is no need to separate membrane vesicles from the surrounding reaction medium prior to analysis. Upon addition of NADH to vesicles from Rhodothermus marinus [40,41] or E. coli , there was a marked increase in the external Na+ concentration which was caused by complex I, as demonstrated by the use of specific inhibitors. This observed net flux of Na+ from the ‘periplasmic’ to the ‘cytoplasmic’ side upon NADH oxidation by complex I from R. marinus and E. coli seems to be contradictory to the observed NADH-driven Na+ transport from the ‘cytoplasmic’ to the ‘periplasmic’ side of complex I from K. pneumoniae [17,38]. Considering that Na+/H+ antiporters may transport Na+ in either direction, and taking into account that complex I from R. marinus and E. coli generated a pH gradient upon NADH:quinone oxidoreduction, Pereira and co-workers proposed that ‘‘complex I has proton pump and Na+/H+ antiporter activities’’ . Support for the assumption of Na+/H+ antiporter module(s) in complex I comes from the relation of the NuoL/NuoM/NuoN subunits with subunits of the Mrp-type antiporter . Do subunits NuoL/NuoM/NuoN actually exchange Na+ for H+ in the complex I-bound state? To answer this question, we will need tools to continuously monitor the reactions performed by complex I (and its individual antiporter-like subunits). The published methods to detect NADH-driven Na+ transport by complex I rely on discontinuous assays, with a time lag of 10 s before analysis of internal Na+ , or 5 min before analysis of external Na+ . In the ideal case, we should be able to monitor H+ and Na+ movements, and the formation of transmembrane voltage, in a simultaneous manner. Another important question is whether complex I, under the given experimental condition, operates in the forward or in the reverse mode of electron transfer. Furthermore, the state of complex I used in the transport assays should be characterized. For example, Na+/H+ antiport activity has only been detected with the ‘deactive’ form of bovine complex I .
Na+ transport by complex I or its isolated NuoL, NuoM or NuoN subunits was also investigated in vivo using bacterial hosts which are sensitive towards Na+, since they lack Na+/H+ antiporters. The Na+-sensitive E. coli EP432 devoid of the antiporters NhaA and NhaB  showed increased resistance to Na+ stress when cells were grown under conditions which promote the synthesis of complex I . Recently, the Hägerhäll group reported that growth of a Na+-sensitive Bacillus subtilis strain lacking MrpA was restored when subunit NuoL from E. coli complex I was expressed in trans . The authors proposed that, in the complex I-bound state, NuoL comprises a Na+ channel, whereas NuoM and NuoN mediate H+ transport . These in vivo studies indicate that the accumulation of Na+ to toxic concentrations in the cytoplasm of Na+/H+ antiporter-deficient bacteria is diminished by the action of complex I, or its individual antiporter-related subunits. The results support the notion that complex I transports Na+ from the cytoplasm to the periplasm during bacterial growth.
In summary, the investigations performed with bacterial complex I from K. pneumoniae, R. marinus and E. coli suggest that NADH oxidation is linked to Na+ transport, but they do not yet allow us to unequivocally decide whether complex I, or its individual antiporter-like subunits, operates an Na+ pump or an Na+/H+ antiporter respectively.
Na+ transport by mitochondrial complex I
Using submitochondrial particles from the yeast Yarrowia lipolytica, and its solubilized complex I reconstituted in proteoliposomes, we showed that mitochondrial complex I from this lower eukaryote has the capacity to catalyse NADH-driven Na+ translocation from the negatively charged (matrix, or N-) to the positively charged (cytoplasm, or P-) side of the membrane. NADH-driven Na+ transport was sensitive towards rotenone, a specific inhibitor of complex I, but was resistant to protonophores. This led us to propose that the Y. lipolytica complex I, similar to the K. pneumoniae enzyme, acts as a primary redox-driven Na+ pump . Recently, Roberts and Hirst  showed that the deactive form of mitochondrial complex I of Bos taurus catalyses Na+/H+ antiport. The authors did not detect Na+/H+ antiport activity with purified complex I from Yarrowia lipolytica. In the forward mode, during NADH oxidation, transport of Na+ and H+ by bovine complex I exhibited the same directionality as observed with the Y. lipolytica complex, e.g. from the N- to the P-side of the membrane . The mechanism of Na+ transport by mitochondrial complex I is far from being understood, but an important outcome of these studies is that the ability to transport Na+ is not restricted to bacterial complex I. Further support for Na+ transport by mitochondrial complex I comes from our observation that subunit ND5 (NuoL homologue) from human complex I inserted in membranes of the endoplasmic reticulum from Saccharomyces cerevisiae increases the salt-sensitivity of the host cells .
Cation transport by the transporter-related NuoL subunit from complex I
With the 3D structure of the membrane arm of E. coli complex I now available , it is possible to search for putative cation channels through NuoL which are retained in NuoLN. Subunit NuoL (PDB code 3RKO) was analysed for cavities with a 1.4 Å radius probe using the program CAVER . Tunnels that could support ion translocation (Figure 1) and important amino acid residues lining the putative channels (Figure 2) were visualized with PyMOL (http://www.pymol.org). The CAVER algorithm finds paths to the solvent from a starting point located within the interior of a protein . Depending on the starting co-ordinates, different segments of a tunnel (i.e. backbone or branches) are found. The depicted channels thus represent only plausible constellations out of several theoretically possible pathways.
Comparison of the 3D structure of subunit NuoL with a 3D model of the truncated NuoL subunit encompassing transmembrane helices 1–11
Efremov and Sazanov  proposed that NuoL comprises two opposing half-channels connected by a series of polar or charged residues in the middle of the membrane. They argue that the first putative channel is closed to the periplasm and the second channel is closed to the cytoplasm by hydrophobic residues and thus a single proton translocating pathway is most likely. Indeed, residues such as Phe123, Leu130, Trp143, Ala261 and Leu265 in channel 1 (coloured black in Figures 1 and 2) and Leu308, Phe341, Leu345 and Phe346 in channel 2 (coloured white in Figures 1 and 2) create hydrophobic ‘choking points’ that seem to block off further progress of the respective pathways. Interestingly, these residues are strongly conserved in complex I and subunits of the Mrp-type antiporters, suggesting a functional role. Our analysis of the structure of NuoL revealed that the areas behind the choking points contain cavities linking to the surface that could accommodate protons or Na+ ions. These cavities are lined with hydrophilic residues, most of which are conserved (Figure 2). One can thus propose that, instead of two half-channels (Figure 1, left-hand panel), two continuous channels exist, the passage through which is controlled by the conserved hydrophobic residues, acting as gates opened and closed by conformational change (Figure 1, middle panel). The first such channel (shown in black) is accessible from the cytoplasmic side as described in . The polar side chains of His100, Tyr106, Met107, Thr120, Tyr157 and Thr257 create a passage to the main cavity lined by Met127, Glu144, Ser150, Thr174, Asp178, Lys229 and Met258. This passage is interrupted by the strongly hydrophobic and bulky Phe123 side chain. Phe123 is highly conserved and mutations in its human homologue Phe124 are linked to MELAS (mitochondrial myopathy, encephalopathy, lactic acidosis and stroke), PD (Parkinson's disease) and Leigh disease [50–52]. A second choking point formed by Trp143, Leu265, Leu130 and Ala261 constricts the periplasmic access of channel 1, which is supported by the sulfur of Met139 and the side chains of Tyr264, Arg268, Ser85, Asp134, Thr269 and Asp82. Mutations in Asp134 and Asp82 were shown to decrease proton pumping activity . Channel 2 (shown in white) comprises the periplasmic half-channel as described in , lined by Asp400, Glu494, Ser497 and the main cavity residues Ser311, Ser314, Gln315, Tyr318, His334, Thr337, Ser398, Lys399, Thr425, Tyr428 and Thr429. It then passes the central residues Lys342, His338, Thr312 and His254, but instead of using the carbonyl oxygen of the Ala255 main chain as a bridge to channel 1, direct access to the cytoplasm by Met243, Ser250, Lys305, Thr247, Asp303, Gln360, Ser349 and Glu359 is also plausible. Mutation of Asp303 has been shown to decrease proton pumping activity  and the human Met243 homologue was mutated in patients suffering from MELAS syndrome . In analogy to channel 1, the direct connection between periplasm and cytosol is restricted by a hydrophobic choking point formed by Leu308, Phe341, Leu345 and Phe346. An animated representation of the channels shown in Figure 2 is provided at http://www.biochemsoctrans.org/bst/041/bst0411280add.htm. The putative cavities reaching from the surface of the protein to the central plane of NuoL may also be interpreted in terms of two crossing channels (Figure 1, right-hand panel). This model assumes two half channels (Efremov and Sazanov ) and includes a second cation pathway which utilizes the putative cavities opposing the two half channels.
The shortened NuoL subunit from E. coli complex I that was truncated after Arg369 (NuoLN) exhibited Na+ (and K+) transport activities in vitro [55,56]. This protein lacks transmembrane helices 12–15, the amphipathic helix HL, and transmembrane helix 16 (Figure 1). Transmembrane helix 12 is broken and yields the two hydrophobic segments TM12a and TM12b . In contrast with the full-length NuoL subunit from E. coli complex I , NuoLN did not confer Na+ resistance on an antiporter-deficient host, but actually increased its sensitivity towards Na+ , suggesting that NuoLN promoted the flux of Na+ into the host cell driven by the electrochemical Na+ gradient. We conclude that there exists at least one continuous channel in NuoLN, probably lined by the transmembrane segments 3, 4, 5, 7 and 8 (shown in black in Figure 2). Transmembrane segment 7 represents a discontinuous helix which is interrupted at the critical Lys229 proposed to participate in H+ transport . Our model assumes (at least) one continuous pathway for cations through NuoLN encompassing the transmembrane segments 1–11 (Figure 2, right-hand panel). In contrast, Efremov and Sazanov  proposed that cation transport involves two half channels built by the transmembrane segments 3–15 of NuoL (Figure 1, left-hand panel).
Hypothesis: Na+ extrusion by complex I completes the Na+ cycle in mitochondria
It has been generally assumed that respiratory complexes in mitochondria exclusively use protons to energize the inner mitochondrial membrane. We now have to consider the possibility that energy conversion by mitochondria does not exclusively rely on the PMF, but may benefit from the conversion of an electrochemical Na+ gradient by complex I. The search for a missing component in the sodium cycle of K. pneumoniae led to the discovery of a bacterial complex I which transports Na+ rather than H+. Is there a sodium cycle in mitochondria, and does complex I play a part in it?
Secondary Na+ transport across the inner mitochondrial membrane may be caused by NHE, the Na+/H+ exchanger with high selectivity for Na+, or KHE, the H+/K+ exchanger which will transport K+ or Na+ with similar affinity . In our scheme depicting possible routes for Na+ into and out of mitochondria (Figure 3), ‘NHE’ stands for the Na+/H+ activity of mitochondria first described by Mitchell and Moyle , but does not refer to a distinct protein, since the molecular identity and composition of NHE and KHE are still under investigation [57,59]. The gradient [Na+]cytosol/[Na+]matrix ranges between 2 and 8 , and it promotes the extrusion of Ca2+ from the matrix to the cytosol by the NCE (Na+/Ca2+ exchanger). Ca2+ enters the mitochondrion via the Ca2+ uniporter, so the mitochondrial Na+ gradient plays an important role in Ca2+ homoeostasis of the cell . Let us consider a situation where ΔpH is diminished, caused by ischaemia or metabolic inhibition. This would result in Na+ flux into mitochondria by NHE (Figure 3) which is not prevented by the transmembrane voltage (positive outside), since transport by NHE is electroneutral . Under these conditions, an efflux of Na+ from the matrix to the cytosol catalysed by complex I would effectively restore the Na+ gradient across the mitochondrial membrane (Figure 3).
Proton- and sodium-ion-transporting systems in mitochondria
Other scenarios are equally feasible. Roberts and Hirst  recently suggested that complex I, in its deactive state, might represent the long-sought NHE of the inner mitochondrial membrane. They argued that a rise in matrix [Na+], for example caused by an opening of the mitochondrial transition pore, would be compensated for by exchange of matrix Na+ with protons from the cytoplasmic side by complex I in its deactive state .
There are many open questions concerning the transport mechanism (primary against secondary), the directionality (from the N- to the P-side, or vice versa) and cation selectivity (H+ against Na+) by complex I, but it seems highly probable that subunits NuoL, NuoM and NuoN directly participate in the cation translocation event(s). Studying the properties of these transporter modules in their isolated state will contribute to our understanding of the coupling mechanism of the complex I, as exemplified by the structure–function analysis of the membrane-bound Fo part of the Na+-translocating ATPase .
Bioenergetics in Mitochondria, Bacteria and Chloroplasts: Third Joint German/UK Bioenergetics Conference, a Biochemical Society Focused Meeting held at Schloss Rauischholzhausen, Ebsdorfergrund, Germany, 10–13 April 2013. Organized and Edited by Fraser MacMillan (University of East Anglia, Norwich, U.K.) and Thomas Meier (Max Planck Institute of Biophysics, Frankfurt am Main, Germany).
We thank Cecilia Hägerhäll, Ricardo O. Louro and Manuela M. Pereira for critically reading the paper before submission.
This work was supported by the Swiss Foundation for Research on Muscle Disease, and the Rehovot-Hohenheim partnership programme (to J.S.).