Mitochondrial ROS cause motor deficits induced by synaptic inactivity: Implications for synapse pruning

Abstract

Developmental synapse pruning refines burgeoning connectomes. The basic mechanisms of mitochondrial reactive oxygen species (ROS) production suggest they select inactive synapses for pruning: whether they do so is unknown. To begin to unravel whether mitochondrial ROS regulate pruning, we made the local consequences of neuromuscular junction (NMJ) pruning detectable as motor deficits by using disparate exogenous and endogenous models to induce synaptic inactivity en masse in developing Xenopus laevis tadpoles. We resolved whether: (1) synaptic inactivity increases mitochondrial ROS; and (2) chemically heterogeneous antioxidants rescue synaptic inactivity induced motor deficits. Regardless of whether it was achieved with muscle (α-bungarotoxin), nerve (α-latrotoxin) targeted neurotoxins or an endogenous pruning cue (SPARC), synaptic inactivity increased mitochondrial ROS in vivo. The manganese porphyrins MnTE-2-PyP⁵⁺ and/or MnTnBuOE-2-PyP⁵⁺ blocked mitochondrial ROS to significantly reduce neurotoxin and endogenous pruning cue induced motor deficits. Selectively inducing mitochondrial ROS—using mitochondria-targeted Paraquat (MitoPQ)—recapitulated synaptic inactivity induced motor deficits; which were significantly reduced by blocking mitochondrial ROS with MnTnBuOE-2-PyP⁵⁺. We unveil mitochondrial ROS as synaptic activity sentinels that regulate the phenotypical consequences of forced synaptic inactivity at the NMJ. Our novel results are relevant to pruning because synaptic inactivity is one of its defining features.

Graphical abstract

Highlights

• •Exogenous and endogenous synaptic inactivity models increase mitochondrial ROS.
• •Manganese porphyrins rescue synaptic inactivity induced motor deficits.
• •Mitochondrial ROS may select inactive synapses for pruning during development.

1. Introduction

Mitochondrial reactive oxygen species (ROS), namely superoxide anion (O₂^.-) and hydrogen peroxide (H₂O₂), are of considerable biological interest because they can signal but can also cause damage [1], [2], [3], [4]. The factors that control the production and removal of mitochondrial ROS are well understood (reviewed in [5], [6], [7]). ATP demand, [NADH] and proton motive force (∆p) control mitochondrial ROS production in forward mode whereas antioxidant enzyme activity controls their removal [5], [6], [7], [8], [9], [10]. Key mitochondrial antioxidant enzymes include manganese superoxide dismutase, peroxiredoxin 3 and glutathione peroxidase 1 [11], [12], [13]. Using existing knowledge of mitochondrial ROS production and removal (the known), it is possible to make experimentally tractable predictions about situations wherein a role for mitochondrial ROS is suspected but unverified (the unknown).

Developmental synapse pruning refines burgeoning connectomes [14], [15]. Neuronal activity regulates pruning: activity protects synapses from and inactivity selects synapses for pruning, respectively [16], [17], [18]. Mitochondrial ROS are proposed to regulate pruning because neuronal activity should divergently regulate their production [19], [20]. Neuronal activity should fulfil essential criteria for comparatively low mitochondrial ROS production (i.e. high respiration, high ATP demand and low ∆p [5]). Conversely, neuronal inactivity should fulfil essential criteria for significant mitochondrial ROS production in forward mode (i.e. low respiration, low ATP synthesis and [NADH] build-up [5]). In support, skeletal muscle activity and inactivity decreases and increases mitochondrial ROS, respectively [21], [22], [23], [24]. Further, surgically abolishing skeletal muscle activity increases mitochondrial ROS ex vivo [25], [26], [27], [28]. We propose that: mitochondrial ROS are endogenous synaptic activity sentinels.

Whether mitochondrial ROS regulate pruning is unknown because it is a novel idea associated with two formidable technical challenges. First, studying pruning is challenging in the brain owing it its intricate connectome. The neuromuscular junction (NMJ)—a peripheral synapse between a motor neuron and skeletal muscle fibre—is an ideal alternative owing to its relative simplicity, size and accessibility [29]. The NMJ is tractable because its development requires pruning to transition from the poly to mono-innervated state [29], [30], [31]. Second, pruning occurs over a protracted time-course (order of weeks) and encompasses in utero and ex utero developmental stages [30]. It is, however, possible to use neurotoxins to broadly recapitulate pruning by forcing synaptic inactivity in species that develop ex utero, such as Xenopus laevis (X. laevis) [32]. Neurotoxins obviate the need for invasive surgical procedures that can provoke adaptive redox responses to injury [33], [34], [35]. A key advantage: neurotoxins make the micro functional outcomes of pruning (partial loss of innervation) detectable as motor deficits at the macro level owing to their mass action [36]. Whether they induce pruning is unclear; but their ability to, at worst, mimic its functional consequences (i.e. motor deficits owing to lost innervation) is clear. To begin to resolve whether mitochondrial ROS regulate pruning, we used multiple forced synaptic inactivity models to determine if: synaptic inactivity increases mitochondrial ROS; and chemically heterogeneous antioxidants rescue synaptic inactivity induced motor deficits.

2. Results

2.1. The manganese porphyrin MnTE-2-PyP⁵⁺ reduces α-BTX—a post-synaptic inactivity model—induced motor deficits

The neurotoxin α-bungarotoxin (α-BTX) induces forced synaptic inactivity by ligating the nicotinic acetylcholine receptor to competitively antagonise cholinergic neurotransmission ([36], [37] Fig. 1A). We used α-BTX to induce forced synaptic inactivity in X. leavis tadpoles. To verify α-BTX induced motor deficits, we assessed evoked swimming frequency: α-BTX decreased evoked swimming frequency by 15.5 ± 1.4% compared to control (Fig. 1B). To determine if α-BTX increases mitochondrial ROS in vivo, we used two intensity based mitochondria-targeted (i.e. triphenylphosphonium cation conjugated) probes: mito-dihydroethidium (Mito-SOX) and mitochondria peroxy yellow 1 (MitoPY1) [38], [39], [40]. Free radicals (e.g. O₂^.-) oxidise Mito-SOX to fluorescent ethidium or 2-hydroxyethidium products whereas H₂O₂ and/or peroxynitrite oxidise MitoPY1 to a fluorescent phenol by a boronate de-protection mechanism [41], [42], [43], [44], [45], [46], [47]. α-BTX increased Mito-SOX and MitoPY1 oxidation by 40.5 ± 4.9% and 30.5 ± 3.5%, respectively, compared to control (Fig. 1C-D).

Fig. 1.

α-BTX induced motor deficits are redox regulated. (A) α-BTX mechanism of action scheme. Left: skeletal muscle activity stems from acetylcholine (ACh) binding to post-synaptic nicotinic acetylcholine receptor (nAChR) to permit Na⁺ entry. Right: α-BTX ligates the nAChR to render it impermeable to Na⁺ leading to forced post-synaptic inactivity induced motor deficits. (B) Evoked swimming frequency (expressed as %δ control) by condition (α-BTX, α-BTX plus: MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- or MitoTempol; n = 10 in each condition). (C) Mito-SOX oxidation (expressed as %δ control) by condition (n = 10 per condition). (D) MitoPY1 oxidation (expressed as %δ control) by condition (n = 10 per condition). Error bars are SEM. Concentrations and incubations: α-BTX (8 µM for 30 min), MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- conditions = 1 µM for 30 min. MitoTempol = 20 µM for 30 min. # denotes significant difference vs control. * denotes significant difference vs α-BTX.

To determine if mitochondrial ROS play a functional role, we used chemically heterogeneous antioxidants (Supplementary Table 1). Specifically, Mn(III) meso-tetrakis(2-pyridyl)porphyrin (MnTE-2-PyP⁵⁺), Mn(III) meso-tetrakis(N-(n-butoxyethyl)pyridinium-2-yl (MnTnBuOE-2-PyP⁵⁺), Mn(III) meso-tetrakis(4-carboxylatophenyl)porphyrin (MnTBAP^3-) and 2,2,6,6-tetramethyl-4-[5-(triphenylphosphonio)pentoxy] piperidin-1-oxy (MitoTempol) [48], [49], [50], [51], [52]. MnTBAP^3- and MitoTempol were inefficacious: failing to rescue α-BTX induced motor deficits or reduce MitoSOX and MitoPY1 oxidation. MnTnBuOE-2-PyP⁵⁺ successfully reduced MitoPY1 oxidation by 20.3 ± 4.2% compared to α-BTX, but failed to significantly reduce α-BTX induced Mito-SOX oxidation or motor deficits. MnTE-2-PyP⁵⁺ successfully blocked mitochondrial ROS reducing Mito-SOX and MitoPY1 oxidation by 38.6 ± 4.7% and 31.2 ± 5.0%, respectively, compared to α-BTX (Fig. 1C-D). MnTE-2-PyP⁵⁺ significantly reduced α-BTX induced motor deficits (Fig. 1B), restoring evoked swimming frequency to control values—no statistical difference between MnTE-2-PyP⁵⁺ and control.

2.2. The manganese porphyrin MnTE-2-PyP⁵⁺ abolishes α-LTX—a pre-synaptic inactivity model—induced motor deficits

The neurotoxin α-latrotoxin (α-LTX) forces pre-synaptic inactivity by binding to nerve terminals to cause mass Ca²⁺ influx, vesicle release and ATP depletion ([53], [54], [55] Fig. 2A). α-LTX induced motor deficits: reducing evoked swimming frequency by 8.3 ± 0.8% compared to control (Fig. 2B). To determine if α-LTX increased mitochondrial ROS, we assessed Mito-SOX and MitoPY1 oxidation. α-LTX increased Mito-SOX and MitoPY1 oxidation by 38.9 ± 6.3% and 41.9 ± 5.2%, respectively, compared to control (Fig. 2C-D). To determine if mitochondrial ROS were functional, we used antioxidants. MnTBAP^3- and MitoTempol were inefficacious: failing to rescue α-LTX induced motor deficits or reduce MitoSOX and MitoPY1 oxidation. MnTnBuOE-2-PyP⁵⁺ successfully reduced Mito-SOX oxidation but failed to significantly reduce α-LTX induced MitoPY1 oxidation or motor deficits. MnTE-2-PyP⁵⁺ decreased Mito-SOX and MitoPY1 oxidation by 25.6 ± 7.2% and 25.9 ± 5.1%, respectively, compared to α-LTX (Fig. 2C-D). MnTE-2-PyP⁵⁺ abolished α-LTX induced motor deficits (Fig. 2A-B).

Fig. 2.

α-LTX induced motor deficits are redox regulated. (A) α-LTX mechanism of action scheme. α-LTX binds to nerve terminals to cause mass neurotransmitter vesicle depletion, leading to pre-synaptic inactivity. (B) Evoked swimming frequency (expressed as %δ control) by condition (α-LTX, α-LTX plus: MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- or MitoTempol; n = 10 in each condition). (C) Mito-SOX oxidation (expressed as %δ control) by condition (n = 10 per condition). (D) MitoPY1 oxidation (expressed as %δ control) by condition (n = 10 per condition). Error bars are SEM. Concentrations and incubations: α-LTX (15 nM for 30 min), MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- conditions = 1 µM for 30 min. MitoTempol = 20 µM for 30 min. # denotes significant difference vs control. * denotes significant difference vs α-LTX.

2.3. MnTE-2-PyP⁵⁺ and MnTnBuOE-2-PyP⁵⁺ reduce SPARC—an endogenous pruning cue—induced motor deficits

Despite multiple advantages (e.g. known mechanisms) [36], [37], [53], [54], neurotoxins lack physiological relevance beyond envenomation. Consequently, we used Secreted Protein Acidic and Rich in Cysteine (SPARC)—an endogenous pruning cue—to provide a physiologically relevant model ([56], [57] Fig. 3A). SPARC reduced evoked swimming frequency by 12.0 ± 1.7% compared to control (Fig. 3B). To determine if SPARC increased mitochondrial ROS, we assessed Mito-SOX and MitoPY1 oxidation. SPARC increased Mito-SOX and MitoPY1 oxidation by 42.1 ± 5.5% and 49.5% ± 5.8%, respectively, compared to control (Fig. 3B-C). To determine if mitochondrial ROS were functional, we used antioxidants. MnTBAP^3- and MitoTempol were inefficacious: failing to rescue SPARC induced motor deficits or reduce MitoSOX and MitoPY1 oxidation. MnTE-2-PyP⁵⁺ and MnTnBuOE-2-PyP⁵⁺ blocked the SPARC induced increase in mitochondrial ROS. MnTE-2-PyP⁵⁺ and MnTnBuOE-2-PyP⁵⁺ reduced Mito-SOX (33.7 ± 4.2% and 28.77 ± 4.6%, respectively) and MitoPY1 (39.2 ± 4.8% and 38.17 ± 2.9%, respectively) oxidation compared to SPARC (Fig. 3C-D). MnTE-2-PyP⁵⁺ and MnTnBuOE-2-PyP⁵⁺ significantly reduced SPARC induced motor deficits (Fig. 3B), restoring evoked swimming frequency to control values—no statistical difference between MnTE-2-PyP⁵⁺ and MnTnBuOE-2-PyP⁵⁺ vs control.

Fig. 3.

SPARC induced motor deficits are redox regulated. (A) SPARC mechanism of action scheme. SPARC binds to weak synaptic inputs to selectively induce pruning—via a yet to be fully elucidated mechanism. Withdrawing inputs (retraction bulbs) are inactive; hence, global SPARC treatment can silence multiple NMJs to induce motor deficits. (B) Evoked swimming frequency (expressed as %δ control) by condition (SPARC, SPARC plus: MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- or MitoTempol; n = 10 in each condition). (C) Mito-SOX oxidation (expressed as %δ control) by condition (n = 10 per condition). (D) MitoPY1 oxidation (expressed as %δ control) by condition (n = 10 per condition). Error bars are SEM. Concentrations and incubations: SPARC (50 nM for 30 min), MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- conditions = 1 µM for 30 min. MitoTempol = 20 µM for 30 min. # denotes significant difference vs control. * denotes significant difference vs SPARC.

2.4. MitoPQ recapitulates exogenous and endogenous synaptic inactivity model induced motor deficits

If exogenous and endogenous forced synaptic inactivity models induce mitochondrial ROS, it should be possible to chemically recapitulate motor deficits by increasing their levels. A related advantage: one can leverage known redox chemistry to gain new insight regarding the redox control of forced synaptic inactivity. We used mitochondria-targeted paraquat (Mito-PQ) to selectively induce matrix O₂^.- [58] (Fig. 4A). Unlike respiratory chain inhibitors (e.g. rotenone [59]), MitoPQ leaves ∆p and ATP synthesis unperturbed [58]. MitoPQ successfully increased mitochondrial ROS evidenced by increased Mito-SOX (44.7 ± 5.3%) and MitoPY1 (34.2 ± 4.3%) oxidation compared to control (Fig. 4B-C). MitoPQ decreased evoked swimming frequency by 19.9 ± 1.4% compared to control (Fig. 4D). We exclude the possibility that MitoPQ simply kills cells to induce motor deficits because primary muscle-nerve co-cultures remain viable after MitoPQ treatment (Supplementary Fig. 1). To explore the functional role of mitochondrial ROS, we used antioxidants. MnTnBuOE-2-PyP⁵⁺ reduced Mito-SOX (43.1 ± 4.2%) and MitoPY1 (51.3 ± 4.4%) oxidation compared to MitoPQ (Fig. 4B-C). MnTnBuOE-2-PyP⁵⁺ reduced MitoPQ induced motor deficits (Fig. 4D). Each antioxidant reduced MitoPQ induced motor deficits—perhaps due to MitoPQ indiscriminately targeting multiple cells consistent with greater (~ 5%) motor deficits compared with α-BTX, α-LTX and SPARC.

Fig. 4.

MitoPQ induced motor deficits are redox regulated. (A) MitoPQ mechanism of action scheme. Within the mitochondrial matrix, MitoPQ accepts an electron from complex I to yield a MitoPQ radical. The MitoPQ radical reacts with molecular oxygen (O₂) to generate O₂^.-, leading to an increase in matrix [O₂^.- and H₂O₂] (the latter as a consequence of MnSOD mediated O₂^.- dismutation). (B) Mito-SOX oxidation (expressed as %δ control) by condition (MitoPQ, MitoPQ plus: MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- or MitoTempol n = 10 per condition). (C) MitoPY1 oxidation (expressed as %δ control) by condition (n = 10 per condition). (D) Evoked swimming frequency (expressed as %δ control) by condition (n = 10 in each condition). Error bars are SEM. Concentrations and incubations: MitoPQ (5 µM for 60 min), MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- conditions = 1 µM for 30 min. MitoTempol = 20 µM for 30 min. #denotes significant difference vs control. *denotes significant difference vs MitoPQ.

3. Discussion

If mitochondrial ROS are synaptic activity sentinels, their production should be unmasked by silencing synaptic activity. We robustly tested whether forced synaptic inactivity increases mitochondrial ROS using mechanistically diverse exogenous and endogenous models. Their ability to induce motor deficits—the canonical outcome of synaptic inactivity at the NMJ—was verified using a well-accepted behavioural assay, before we expolited the optical accessibility of X. laevis tadpoles to assess mitochondrial ROS in vivo using established techniques [38], [39], [40], [41], [42], [43], [44], [45]. Forced synaptic inactivity rapidly (within 30 min) increases mitochondrial ROS in vivo, regardless of how it is achieved. The outcome is the same irrespective of whether one deploys muscle (α-BTX) or nerve (α-LTX) targeted neurotoxins or endogenous pruning cues (SPARC) to silence the NMJ. We acknowledge that global increases in mitochondrial ROS are described: specific roles of nerve or muscle mitochondria are important open questions. It will also be necessary to unravel how synaptic inactivity increases mitochondrial ROS with reference to sites and modes of their production [5], [6], [7]. We are, however, the first to describe rapid increases in mitochondrial ROS in vivo across diverse synaptic inactivity models. That mechanistically diverse forced synaptic inactivity models increase mitochondrial ROS, supports the idea that mitochondrial ROS are endogenous synaptic activity sentinels.

Forcing synaptic inactivity en masse makes the phenotypical consequences of pruning (partial loss of innervation) at the micro level detectable as motor deficits at the macro level, enabling us to resolve whether mitochondrial ROS were functional or mere epiphenomena. Our forced synaptic inactivity models make functional studies possible because their effects are reversible [36], [55]. If mitochondrial ROS are functional, antioxidants should rescue motor deficits, provided they block mitochondrial ROS [60], [61], [62], [63], [64]. The latter clause is key because failing to block mitochondrial ROS seemed to negate the ability of MnTBAP^3- and MitoTempol to reduce motor deficits induced by neurotoxins or SPARC. MnTE-2-PyP⁵⁺ and/or MnTnBuOE-2-PyP⁵⁺ block mitochondrial ROS to significantly reduce forced synaptic inactivity induced motor deficits. Consistent with their ability to accumulate in mitochondria and react with O₂^.-/H₂O₂ [48], [49], [52], [65]. However, given their pleiotropic redox biochemistry [48], how MnTE-2-PyP⁵⁺ and MnTnBuOE-2-PyP⁵⁺ act is an open question. MnTE-2-PyP⁵⁺ and/or MnTnBuOE-2-PyP⁵⁺ were effective regardless of how forced synaptic inactivity was achieved: significantly reducing motor deficits induced by disparate exogenous and endogenous models. Their efficacy suggests NMJ blockade fails to fully explain the motor deficits observed; perhaps, denervation contributes, consistent with synapse loss (pruning) as SPARCs modus operandi [56], [57]. Our antioxidant studies strongly suggest that mitochondrial ROS cause forced synaptic inactivity induced motor deficits.

Multiple non-mitochondrial ROS sources exist (e.g. NADPH oxidase [66], [67], [68]); why, therefore, do we suspect a role for mitochondrial ROS? We include a role for non-mitochondrial ROS sources: ROS-induced ROS release means they likely contribute [69]. However, most do so at a biological cost: sustaining NADPH oxidase activity requires continued NADPH and GTP to supply electrons and activate RAC1, respectively [67], [70], [71]. We favour mitochondrial ROS as the proximal event because their production reports on ATP use without consuming it, placing mitochondrial ROS as energetically efficient synaptic activity sentinels [19], [72], [73], [74], [75]. We used MitoPQ to explore the role of mitochondrial ROS [58]. MitoPQ recapitulated synaptic inactivity induced motor deficits by increasing mitochondrial ROS. MnTnBuOE-2-PyP⁵⁺ blocked mitochondrial ROS to significantly reduce MitoPQ induced motor deficits; suggesting mitochondrial ROS production is the key proximal event. A related corollary: MitoSNO rescues ischemia-reperfusion injury induced denervation by blocking mitochondrial ROS [76].

4. Conclusion

We unveil mitochondrial ROS as synaptic activity sentinels that regulate the phenotypical consequences (i.e. motor deficits) of forced synaptic inactivity at the NMJ. Our novel results are relevant to pruning because synaptic inactivity is one of its defining features.

5. Methods

5.1. Materials and reagents

Materials and reagents details are provided (Supplementary Table 2).

5.2. Xenopus laevis

X. laevis were reared in de-chlorinated water at room temperature. Nieuwkoop–Faber developmental stage 37–38 X. laevis tadpoles were used because the neural control of their swimming behaviour is well-understood [77]. Experiments received institutional ethical approval.

5.3. Treatments

In accordance with previous research [78], [79], stage 37–38 X. laevis tadpoles were anesthetised in 0.1% MS-222 (3-aminobenzoic acid ester) in normal frog ringer (NFR, mM: 116 NaCl, 2 KCl, 1.8 CaCl₂, 5 Na⁺-HEPES, pH 7.35) until their movement ceased (usually within a minute). MS-222 is commonly used to anesthetise X. leavis [80]. After mounting tadpoles on a Sylgaard stage immersed in NFR, a fine dorsal fin cut was made proximal to the myotomal region using a dissection knife to improve α-BTX and α-LTX permeability consistent with previous research [78], [79], [81], [82]. After being allowed to recover for 30 min in NFR, tadpoles were incubated with either control (i.e. just NFR) or NFR plus: SPARC (50 nM for 30 min), α-LTX (15 nM for 30 min), α-BTX (8 µM for 30 min) or MitoPQ (5 µM for 60 min) in a light-protected Eppendorf before being washed in NFR for 3 min. For antioxidant experiments, tadpoles were pre-treated for 30 min with either MnTE-2-PyP⁵⁺ (1 µM), MnTnBuOE-2-PyP⁵⁺ (1 µM), MnTBAP^3- (1 µM) and MitoTempol (20 µM) before being washed in NFR for 3 min and treated as described above. Doses and times were rationally selected based on pilot experiments and/or previous research [36], [48], [49], [50], [51], [55], [57], [58].

5.4. Evoked swimming frequency

In accordance with previous research [78], [83], evoked swimming was induced by manually touching the skin of X. laevis tadpoles with a pipette, which robustly initiates movement. Swimming was filmed at 400 frames per second with a Nikon D7100 camera and calculated as: swimming frequency = frames per second/ frames per cycle. A cycle defined one complete tail oscillation and the frames required to complete one cycle were manually counted to derive evoked swimming frequency. A single swimming episode was recorded at 2 min intervals [78], five episodes were recorded and their average calculated and defined as n = 1. Evoked swimming frequency experiments were performed during the day under room lighting because dimming lights influences swimming behaviour [83].

5.5. Mitochondrial ROS imaging

Mito-SOX and MitoPY1 were used at a working concentration of 5 µM [84]. After a 30 min light-protected incubation to avoid artificial photo-oxidation, tadpoles were washed in NFR supplemented with 0.1% MS-222 (to prevent movement while imaging) for 3 min and placed on a glass cavity slide with methylcellulose for whole-mount redox imaging in a dark room [84]. Care was taken to avoid bubbles when dispensing methylcellulose. If necessary, X. laevis tadpoles were orientated using nylon line [84], before being imaged (Mito-SOX excitation/emission: 510/580 nm; MitoPY1 excitation/emission: 488/517 nm) with a Lecia DMR epifluorescence microscope to capture images of the myotomal region (containing skeletal muscle and motor neurons) with a Nikon D7100 camera. For reference, Supplementary Fig. 4 shows the myotomal region. Regions of interest (ROIs) were analysed on Image J (http://imagej.nhi.gov/ij). Two images of the myotomal region were captured per tadpole and at least three ROIs were analysed per image. After background subtraction and smoothing [33], mean florescence intensity was calculated (n = 1 defines mean fluorescence per tadpole).

6. Primary muscle-nerve co-culture

To derive primary muscle-nerve co-cultures, Nieuwkoop–Faber Stage 22 X. leavis embryos were dissected in sterile NFR [85], [86]. After removing the jelly coat and vitelline membrane with sterile forceps, the neural tube and associated myotomes were manually dissected in sterile Ca²⁺/Mg²⁺-free media (CMF mM: 125 NaCl, 2 KCl, 1.2 EDTA, 5 Na⁺-HEPES, pH 7.35). Cells were left to dissociate in CMF for 2 h before being plated in poly-d-lysine coated (500 µg/ml) 35 mM glass culture dishes in media (50% L-15% and 50% NFR vol/vol) and cultured for 24 h at room temperature under ambient gas conditions before use.

6.1. Cell viability

Primary muscle-nerve co-cultures were treated with control (5 µM DMSO in media) or MitoPQ (5 µM in media) for 60 min before being incubated with Trypan blue solution (0.4% Trypan Blue stock diluted 1:1 with culture medium) for 2 min. After removing Trypan Blue solution, co-cultures were immediately fixed with 5% glutaraldehyde for 10 min at room temperature before being imaged using Leica DM IL LED. Images were captured using Leica MC120 HD Microscope Camera. Cell viability was calculated as: percentage viable cells = [1.00 – (Number of blue cells ÷ Number of total cells)] × 100.

6.2. Statistical analysis

To determine the appropriate statistical test, normal distribution was assessed using a Shapiro-Wilk test. Normally distributed data were analysed using a one-way general linear model with post-hoc Tukey tests. Abnormally distributed data were analysed using a non-parametric equivalent with Dunn's multiple comparison tests with alpha set to ≤ 0.05. Statistical analysis were performed on GraphPad PRISM 7 (https://www.graphpad.com/).

Appendix B. Supplementary material

Supplementary Table 1

Supplementary material

Supplementary Table 2

Supplementary material

Supplementary Fig. 1.

No statistical difference in cell viability between MitoPQ vs control treated primary muscle-nerve co-cultures.

Supplementary Fig. 2.

Representative florescent images of Mito-SOX oxidation in vivo by synaptic inactivity model (α-BTX, α-LTX, SPARC and MitoPQ) and treatment condition (control, MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- and Mito-Tempol) taken from the intermyotomal region of stage 37–38 X. laevis tadpoles using whole-mount redox imaging. Scale bar is 0.1 mm.

Supplementary Fig. 3.

Representative florescent images of MitoPY1 oxidation in vivo by synaptic inactivity model (α-BTX, α-LTX, SPARC and MitoPQ) and treatment condition (control, MnTE-2-PyP⁵⁺, MnTnBuOE-2-PyP⁵⁺, MnTBAP^3- and Mito-Tempol) taken from the intermyotomal region of stage 37–38 X. laevis tadpoles using whole-mount redox imaging. Scale bar is 0.1 mm.

Supplementary Fig. 4.

A stage 37–38 X. laevis tadpole with the location of a representative intermyotomal region sampled (red box) with florescent images of Mito-SOX and MitoPY1 oxidation shown below.