Mitochondrial matrix-localized Src kinase regulates mitochondrial morphology

The architecture of mitochondria adapts to physiological contexts: while mitochondrial fragmentation is usually associated to quality control and cell death, mitochondrial elongation often enhances cell survival during stress. Understanding how these events are regulated is important to elucidate how mitochondrial dynamics control cell fate. Here, we show that the tyrosine kinase Src regulates mitochondrial morphology. Deletion of Src increased mitochondrial size and reduced cellular respiration independently of mitochondrial mass, mitochondrial membrane potential or ATP levels. Re-expression of Src targeted to the mitochondrial matrix, but not of Src targeted to the plasma membrane, rescued mitochondrial morphology in a kinase activity-dependent manner. These findings highlight a novel function for Src in the control of mitochondrial dynamics. Supplementary Information The online version contains supplementary material available at 10.1007/s00018-022-04325-y.


Introduction
Mitochondria are considered as the cellular powerhouse due to their role in oxidative phosphorylation (OXPHOS) and ATP production, but they are also involved in other key physiological processes including Ca 2+ homeostasis, apoptosis and steroidogenesis. Mitochondria form networks into the cytosol which are constantly modified by fission and fusion events, a process known as mitochondrial dynamics [1]. Morphology of individual organelle can shift from tubular to more fragmented or elongated structures according to the metabolic state of the cell, to respond to cellular cues and maintain cellular homeostasis [1,2]. For instance, mitochondria elongate during mild stress [3,4], whereas inhibition of OXPHOS or mitochondrial depolarization induces mitochondrial fragmentation [5,6]. These morphological transition states are mainly regulated by dynamin GTPase proteins with Dynamin-related protein 1 (Drp1), the main actor of mitochondrial division, Mitofusins 1 and 2 (Mfn1/2) and Optic Atrophy 1 (OPA1) controlling outer mitochondrial membrane (OMM) and inner mitochondrial membrane (IMM) fusion, respectively [1,7].
The tyrosine kinase Src was among the first oncogenes to be described [8]. It is involved in various physiological and oncogenic processes, ranging from metabolism, proliferation and differentiation to survival, motility and angiogenesis [8,9]. Src has been found at multiple subcellular compartments including endosomes and plasma membrane [10], the Golgi apparatus [11,12] as well as mitochondria [13,14]. Src targets various mitochondrial proteins involved in different steps of metabolism including OXPHOS, fatty acid oxidation, pyruvate metabolism, ketone body production and production of reactive oxygen species [13,[15][16][17]. We recently observed that metastatic triple-negative breast cancer cells show higher activity of intramitochondrial Src [16]. Interestingly, metastatic properties of these cells depend on mitochondrial dynamics. Indeed, metastatic breast cancer cells display fragmented mitochondria, and inhibition of mitochondrial division via silencing of Drp1 or promotion of mitochondrial fusion by over-expression of Mfn1 blocks metastasis in these cells [18]. These findings suggest that Src could affect mitochondrial dynamics to control oncogenic properties, at least in specific cancer cells. However, the contribution of Src to adjustments of mitochondrial shape is currently not well understood.
The aim of this study was to examine and describe the potential role of Src in the regulation of mitochondrial dynamics. We observed that deletion or silencing of Src increased mitochondrial size, whereas its over-expression had the opposite effect. Deletion of Src also reduced cellular respiration and affected the levels of the mitochondria-shaping proteins Drp1, MiD51 and OPA1, whereas no effect was observed on mitochondrial mass, mitochondrial membrane potential and ATP levels. Re-expression of Src rescued alterations of mitochondrial morphology but not cellular respiration or levels of mitochondria-shaping proteins. Mechanistically, we propose that mitochondrial matrixlocalized Src modulates mitochondrial shape independently of OXPHOS. Indeed, re-expression of Src targeted to the mitochondrial matrix rescued mitochondrial morphology, in a kinase activity-dependent manner, whereas re-expression of Src targeted to plasma membrane had no impact on mitochondrial shape. Overall, these findings highlight a direct contribution of intramitochondrial Src and its kinase activity to the control of mitochondrial dynamics.

Src decreases mitochondrial size
Although indirect evidence suggests that Src might impact mitochondrial shape, its exact role in the regulation of mitochondrial dynamics remains unknown. To address this, the morphology of the mitochondrial network was first examined using confocal microscopy in (1) Src ++ mouse embryonic fibroblasts (MEFs) expressing Src but knocked-out for Yes and Fyn (two other members of the Src kinases family) and (2) SYF MEFs knocked-out for Src, Yes and Fyn [19], named hereafter Src +/+ and Src −/− MEFs, respectively. We observed that the deletion of Src (Fig. S1a) decreased the percentage of cells with tubular mitochondria and increased the number of cells with elongated mitochondria (Fig. 1a, b). Quantitative analyses of regions of interest (ROIs) showed that the deletion of Src also decreased the number of mitochondria and increased the area of individual organelles in ROIs (Fig. 1a, c), supporting the mitochondrial elongated phenotype observed in Src −/− MEFs. To examine whether the downregulation of Src increases mitochondrial size independently of Yes and Fyn, Src was silenced using shRNA in naive MEFs (Fig. S1b). Similar to what was observed in Src −/− MEFs, shSrc increased the percentage of MEFs with elongated mitochondria (Fig. S1c,  d), decreased the number of individual mitochondria and increased the area of mitochondria in ROIs (Fig. S1c, e). Src was also silenced using siRNA in the human cell line HeLa (Fig. S1f). Again, we observed that lower expression of Src increased the percentage of HeLa cells with elongated mitochondria (Fig. 1d, e), decreased the number of individual mitochondria and increased the area of individual organelles in ROIs (Fig. 1d, f). The impact of Src deletion on mitochondrial morphology was confirmed using transmission electron microscopy since individual organelles of Src −/− MEFs had higher area and perimeter as compared to Src +/+ mitochondria (Fig. 1g, h). Strikingly, upregulation of Src had opposite effects as compared to its deletion. Indeed, over-expression of Src-HA in Hela cells increased the percentage of cells with fragmented mitochondria (Fig. 1i, j) and increased the number of mitochondria with smaller area in ROIs (Fig. 1i, k). Overall, these findings reveal for the first time that Src reduces mitochondrial size.
To understand how the downregulation of Src leads to mitochondrial elongation, we first examined the levels of a panel of mitochondria-shaping proteins by immunoblotting ( Fig. 2a-c). Results showed that levels of the pro-fission Drp1 and its mitochondrial receptor MiD51 were decreased in Src −/− MEFs, whereas no change was Data are presented as mean ± S.E.M. For the analyses of mitochondrial phenotype, data with different letters are statistically different (p < 0.05), according to two-way ANOVA followed by Tukey's multiple comparison test. For all other analyses, *p < 0.05, **p < 0.01, ***p < 0.001 according to Student's t test ◂ observed in FIS1 or MiD49 levels ( Fig. 2a, b). Levels of Drp1 were also lower in cytosolic and mitochondrial fractions of Src −/− MEFs, but no differences were observed when normalized to the total level of Drp1 in TCL (Fig.  S1h), suggesting no defects in Drp1 recruitment to mitochondrial membranes in Src −/− MEFs. Activity and translocation of Drp1 to mitochondria are partly controlled by phosphorylation at its S616 and S637 residues [20][21][22]. When normalized to Drp1 levels, phosphorylation of both S616-and S637-Drp1 were not different between Src +/+ and Src −/− MEFs (Fig. 2a, b), further corroborating that Src does not affect activity or subcellular distribution of Drp1. Among pro-fusion proteins, levels of the IMM fusion regulator OPA1 were increased in Src −/− MEFs without changing the ratio between long and short OPA1 isoforms (Fig. 2a, c) or OPA1 oligomerization (Fig. S1i). The levels of the OMM fusion protein MFN2 did not change upon deletion of Src (Fig. 2a, c). Surprisingly, levels of the OMM fusion protein MFN1 were decreased in Src −/− MEFs (Fig. 2a, c), suggesting it is not involved in Src-mediated regulation of mitochondrial morphology. These findings indicate that the mitochondrial elongation induced by deletion of Src is associated to lower levels of specific pro-fission proteins (Drp1 and MiD51) and higher levels of the pro-fusion protein OPA1.

Src does not impact on mitochondrial fusion and fission rates
To examine whether the mitochondrial elongation induced by Src downregulation was associated with deregulation of mitochondrial fusion or fission, we first quantified real-time mitochondrial fusion and fission events in Src +/+ and Src −/− MEFs expressing a GFP targeted to the mitochondrial matrix (mtGFP) by live cell confocal microscopy, as previously described [23]. Our findings suggest no difference in fusion and fission rates upon deletion of Src ( Fig. 2d; Movies 1 and 2). Similar results were obtained when these events were quantified in HeLa cells treated with siSrc ( Fig. S1g; Movies 3 and 4). Mitochondrial fusion was further evaluated using Src +/+ and Src −/− MEFs expressing mitochondria-targeted photoactivable GFP (mt-PAGFP) [24]. Quantification of GFP fluorescence in stimulated ROIs of cells co-expressing mt-PAGFP and the mitochondrial marker mtDsRed indicated similar mitochondrial fusion rates between Src +/+ and Src −/− MEFs after 10 min (Fig. 2e, f). We also treated Src +/+ and Src −/− MEFs expressing mtGFP with the uncoupler carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) to induce Drp1-dependent mitochondrial fission [25]. We observed that Src +/+ and Src −/− MEFs expressing mtGFP have similar fragmentation levels upon treatment with the uncoupler FCCP, suggesting no defects in the fission machinery per se, at least under this specific stress (Fig. 2g, h). These findings indicate that while we observed a dysregulation of some of the main regulators of mitochondrial morphology, including Drp1 and OPA1, the downregulation of Src increases mitochondrial size without affecting the overall mitochondrial fusion or fission rates.
Next, we observed that the changes of mitochondrial morphology induced by downregulation of Src were not associated with alterations of mitochondrial mass, as shown by immunoblotting several mitochondrial proteins (Fig. 3a, b), assessing citrate synthase activity (Fig. 3c), and by staining Src +/+ and Src −/− MEFs with Mitotracker™ green which labels mitochondria independently of the mitochondrial membrane potential (Fig. 3d).
However, deletion of Src significantly decreased cellular respiration in MEFs (Fig. 3e). Similarly, silencing of Src in HeLa cells also decreased oxygen consumption (Fig. S1j), suggesting that the changes of mitochondrial size induced by Src downregulation could be due to defects in OXPHOS. However, staining Src +/+ and Src −/− MEFs with tetramethylrhodamine (TMRM) showed that deletion of Src had no impact on mitochondrial membrane potential (Fig. 3f). Finally, cytosolic ATP levels, as measured by the ATP probe GoAteam2 [26], were not different between Src +/+ and Src −/− MEFs (Fig. 3g). ATP levels were also comparable after treatment with the ATP synthase inhibitor oligomycin, used as a proxy to measure cytosolic ATP generated by the mitochondrial ATP synthase (Fig. 3g, h). Thus, these findings suggest that the Src-dependent decrease in oxygen consumption does not lead to altered mitochondrial membrane potential nor ATP production. Overall, we propose that the Src-dependent changes in mitochondrial size are not due to a global mitochondrial dysfunction.

The kinase activity of mitochondrial matrix-localized Src is sufficient to impact on mitochondrial size
Src was previously shown to localize to different cellular compartments including mitochondria in several cellular and animal models, using multiple techniques such as transmission electron microscopy [13,15,27,28]. Subcellular fractionation followed by immunoblotting showed that Src is present in mitochondria-enriched fractions obtained from MEFs, HEK293 and HeLa (Fig. 4a), as previously observed [13,15,16,29]. Trypsin sensitivity assay then confirmed that Src is present inside mitochondria since it was protected from degradation when mitochondrial-enriched fractions were treated with trypsin, in contrast to the OMM protein TOM20 (Fig. 4b). To further determine the localization of Src at different mitochondrial subcompartments, mitochondria-enriched fractions were treated with increasing concentrations of digitonin, which dose-dependently release the intermembrane space protein Smac and the IMM matrixfacing protein ATP5a (Fig. 4c). We observed that the endogenous Src appears to be located in different mitochondrial compartments including the IMM/mitochondrial matrix compartment (Fig. 4b), as previously observed [13,15]. Immunoblotting of the ER lumen protein ERp57 excluded bias linked to the well-known contamination of mitochondria-enriched fractions with ER since ERp57 was released differently than Src (Fig. 4c).
To test whether the impact of Src on mitochondrial morphology depends on its subcellular localization, we generated different tagged Src constructs: (i) untargeted Src, (ii) Src targeted to the mitochondrial matrix (mtSrc), and (iii) Src targeted to the plasma membrane (pmSrc). The constructs Src-V5, mtSrc-V5 and pmSrc-V5 were then used to rescue levels of Src in Src −/− MEFs in specific subcellular compartments ( Fig. 4d and Fig. S2a). Expression of Src-V5 or mtSrc-V5 both reversed the changes of mitochondrial shape induced by the deletion of Src, whereas no change was observed upon expression of pmSrc-V5 (Fig. 4d, e). Similarly, over-expression of Src-V5 and mtSrc-V5 in HeLa cells decreased mitochondrial size (Fig. S2b-d). Over-expression of Src targeted to the plasma membrane had no effect on the number and the area of individual mitochondria in ROIs although pmSrc-V5 changed the number of cells between the different mitochondrial phenotypes. (Fig. S2b-d). Globally, our results suggest that Src localized in the mitochondrial matrix is sufficient to affect mitochondrial shape.
Since re-expression of Src in Src −/− MEFs rescued mitochondrial morphology, we tested whether it would also reverse the alterations of Drp1, MiD51 and OPA1 levels. Surprisingly, expression of Src-HA and mtSrc-HA in Src −/− MEFs did not reverse the decrease of Drp1 and MiD51 levels or the increase of OPA1 levels induced by the deletion of Src (Fig. 5a, b). Considering that downregulation of Src decreases cellular respiration ( Fig. 3e and Fig. S1j), we also examined whether expression of Src-HA and mtSrc-HA would affect oxygen consumption in Src −/− MEFs. Similar to Drp1, MiD51 and OPA1 levels, expression of Src-HA and mtSrc-HA in Src −/− MEFs did not rescue cellular respiration (Fig. 5c). Altogether, these results suggest that Src affects mitochondrial size and cellular respiration via different mechanisms.
Expression of Src-HA and mtSrc-HA both reversed the mitochondrial elongation phenotype induced by siSrc in HeLa cells (Fig. 6a, b). To examine if these morphological changes depend on the kinase activity of Src, we performed similar rescue experiments in HeLa cells treated with siSrc using constitutively active (CA) and kinase-dead (KD) mutants of mtSrc [30], named mtSrc(CA)-FLAG and mtSrc(KD)-FLAG, respectively. Interestingly, mtSrc(CA)-FLAG, but not the inactive mutant mtSrc(KD)-FLAG, reversed the mitochondrial elongation induced by siSrc in HeLa cells (Fig. 6c, d), suggesting that mitochondrial matrix-localized Src affects mitochondrial shape in a kinase activity-dependent manner. The same constructs were then over-expressed in HeLa cells: mtSrc(CA)-FLAG reduced mitochondrial size, whereas the inactive mutant mtSrc-(KD)-FLAG had no effect (Fig. S3a, b). This gain-of-function experiment confirmed that the kinase activity of Src inside the mitochondrial matrix is sufficient to induce the Src-dependent alterations of mitochondrial shape.

Over-expression of mitochondrial matrix-localized Src prevents mitochondrial elongation during starvation
During starvation, mitochondria elongate to protect themselves from autophagic degradation and prevent cell death [4,31]. We thus hypothesized that the Src-dependent modulation of mitochondrial dynamics could be involved in this process. To address this, we first treated Src +/+ and Src −/− MEFs with Hank's balanced salt solution (HBSS) during 5 h, a treatment known to induce starvation-induced mitochondrial elongation [4]. We observed more Src +/+ MEFs with elongated mitochondria upon starvation, whereas no change was observed in Src −/− MEFs (Fig. 7a, b), confirming that the mitochondrial network of Src −/− MEFs was already in an elongated state before starvation. The mitochondrial morphology was also examined in HeLa cells over-expressing different Src constructs and treated with HBSS during 5 h. As expected, the HBSS treatment induced mitochondrial elongation in control HeLa cells (Fig. 7c, d). Mitochondrial elongation induced by starvation was, however, inhibited in cells over-expressing mtSrc-FLAG but not in cells expressing the inactive mtSrc(KD)-FLAG (Fig. 7c,  d), suggesting that the activity of intramitochondrial Src is decreased during starvation to allow mitochondrial elongation. However, we observed higher phosphorylation of the activating residue of Src (i.e., pY416-Src), indicating that Src activity increased upon HBSS treatment in both total cell lysates and mitochondria-enriched fractions (Fig. 7e, f). This finding indicates that the regulation of mitochondrial shape by endogenous mitochondrial matrix-localized Src is likely not involved in starvation-induced mitochondrial elongation.

Discussion
The present work examined the impact of the tyrosine kinase Src on mitochondrial morphology. We show that lower expression of Src increases mitochondrial size, whereas its over-expression shortens mitochondria. We suggest that the kinase activity of Src inside the mitochondrial matrix adjusts mitochondrial shape.
Mitochondrial activity and morphology are tightly connected and treatment with inhibitors of OXPHOS leads to mitochondrial fragmentation [5,6]. For instance, the uncoupler FCCP or the mitochondrial ATP synthase inhibitor oligomycin induce mitochondrial fragmentation via the cleavage of OPA1 by the protease OMA1, leading to inhibition of mitochondrial fusion and unopposed mitochondrial fragmentation [32]. Previous works showed that pharmacological inhibition or genetic downregulation of Src decreases OXPHOS, mitochondrial membrane potential and ATP production [13,15,33,34]. Here, the downregulation of Src did not decrease mitochondrial membrane potential or ATP levels, at least in the cytosol, suggesting that the alterations of oxygen consumption induced by downregulation of Src observed here and in previous work [13,15,[33][34][35] did not translate into decreased mitochondrial membrane potential or ATP levels in MEFs and HeLa cells. Our results thus suggest that the Src-dependent regulation of mitochondrial morphology is independent of OXPHOS defects. In contrast, mitochondria can elongate upon various stress stimuli. Starvation triggers cAMP-PKA signaling and inhibition of Drp1-mediated mitochondrial fragmentation leading to unopposed mitochondrial fusion [4,31]. UV irradiation, actinomycin D and cycloheximide also induce mitochondrial elongation. In such conditions, mitochondrial fusion increases due to stabilization of the pro-fusion long OPA1 isoforms by SLP-2 [3]. This stress-induced mitochondrial hyperfusion is independent of mitochondrial depolarization or changes in ATP levels [3,4], similar to our observations about Src-dependent mitochondrial elongation. However, our findings suggest that the regulation of mitochondrial size by endogenous Src is not involved in starvation-induced mitochondrial elongation. Overall, the functional impact of the Src-mediated modulation of mitochondrial shape is still unclear, and future works will be necessary to address this.
Our findings suggest that Src regulates mitochondrial shape downstream of mitochondria-shaping proteins. Indeed, although levels of Drp1, MiD51 and OPA1 were different between Src +/+ and Src −/− MEFs, re-expression of Src in Src −/− MEFs rescued the changes of mitochondrial morphology but not of the levels of these mitochondriashaping proteins. Thus, the different levels of Drp1, MiD51 and OPA1 between Src +/+ and Src −/− MEFs could have been indirect consequence of the deletion of the kinase itself, to compensate for the chronic deficiency of Src. Similarly, deletion of Src did not affect processing and oligomerization of OPA1. Thus, Src could control mitochondrial shape downstream of the canonical fission/fusion machineries. Drp1-independent mitochondrial division can occur during hypoxia or deferiprone-induced mitophagy in mammalian cells [36]. Elevated cytosolic Ca 2+ can also shorten mitochondria independently of Drp1 [37,38]. Moreover, some types of mitochondria-derived vesicles are released from mitochondria in a Drp1-independent manner [39]. To our knowledge, there is, however, no evidence for any mechanism leading to mitochondrial elongation independently of the pro-fusion proteins OPA1 and Mfn1/2. Although this was not the focus of this work, it would be critical to identify the mechanisms linking Src and mitochondrial dynamics in future studies.
Src is present in several subcellular compartments, including mitochondria [10,13,15]. Our experiments confirm previous observations showing that a pool of Src is, at least partly, localized inside the organelle [13,15,28]. Targeting Src to the mitochondrial matrix was sufficient to reduce mitochondrial size. These findings suggest that Src could target proteins inside the organelle to control its shape. We recently identified several potential targets of intramitochondrial Src which could be involved in the morphological changes observed in the present work [17]. For example, it has been shown that Src could interact with SLP-2 [17] which is required for stress-induced mitochondrial hyperfusion [3]. Similarly, we showed potential interaction of intramitochondrial Src with prohibitin 2 and the mitochondrial contact site and cristae organizing system Mic60 [17] which are also key players in the regulation of mitochondrial architecture [40,41]. It is also possible that Src is present in other compartments of the organelle from where it could impact on mitochondrial architecture and functions. Similarly, over-expression of Src targeted to the plasma membrane impacted the number of cells among the different mitochondrial phenotypes, suggesting that the different subcellular pools of Src could impact on mitochondrial morphology via specific pathways. Further investigation could thus allow to identify new pathways important for the regulation of mitochondrial shape and examine whether Src modulates mitochondrial morphology and functions In conclusion, the present work highlights a novel function for Src as a regulator of mitochondrial morphology. This may provide insights into the contribution of Src signaling in several cellular processes and diseases such as the involvement of Src in different pathologies, including cancer. The role of Src in oncogenesis and the metabolic rewiring observed in cancer cells is well described [8]. Considering that metastasis of some cancer cells depends on appropriate regulation of mitochondrial dynamics, the intramitochondrial Src signaling and its impact on mitochondrial morphology could represent a novel therapeutic avenue.

Constructs and site-directed mutagenesis
The constructs encoding hemagglutinin (HA)-or V5-tagged mouse Src (Src-HA and Src-V5) and HA-or V5-tagged mouse Src targeted to the mitochondria (mtSrc-HA and mtSrc-V5) were generated by amplification of the sequence of Src from pCMV5-Src (#13663, Addgene, MA, USA) and fused to HA or V5 with or without the mitochondrial leading sequence (MLS) of cytochrome c oxidase VIIIa, as described [16]. Constructs encoding FLAG-tagged human wild-type Src, constitutively active (CA) Src and kinase-dead (KD) Src fused to the MLS of the very long chain acyl-CoA dehydrogenase (named hereafter mtSrc-FLAG, mtSrc(CA)-FLAG and mtSrc(KD)-Flag, respectively) were obtained from Addgene (#44652, #44654 and # 44653, respectively). Src-V5 was targeted to the plasma membrane (pmSrc-V5) using the first 26 amino acids of Lck (derived from Lck-GFP obtained from Addgene #61099) as previously shown [42,43].

Subcellular fractionation
Mitochondria-enriched fractions were obtained as previously described [44]. Briefly, cells were harvested and resuspended in mitochondrial isolation buffer (250 mM sucrose, 1 mM EDTA, 5 mM HEPES, pH 7.4) supplemented with 1% protease inhibitor cocktail and 2 mM sodium orthovanadate. Cells were lysed with 15 strokes using a 25-gauge syringe on ice and centrifuged at 1500g for 5 min (4 °C). The resulting supernatant (TCL) was centrifuged at 12,500g for 10 min (4 °C). The obtained supernatant was considered as the cytosol-enriched fraction, whereas the pellet was resuspended in the mitochondrial buffer and a cycle of centrifugations at 1500g and 12,500g was repeated. The final pellet was considered as the mitochondria-enriched fraction.
For assessing distribution of Src among mitochondrial subcompartments, mitochondria-enriched fractions were resuspended in 1 M HEPES-KOH buffer (pH 7.4) in the presence of 0.125, 0.25 or 0.5% of digitonin (Sigma-Aldrich, MO, USA) and incubated at room temperature (RT) with continuous shaking at 1000 rpm. After 15 min, samples were centrifuged at 12,500g at 4 °C for 10 min. The resulting pellets and supernatants were processed for SDS-PAGE. The presence of proteins from different mitochondrial subcompartments in pellets and supernatants was analyzed using immunoblotting.

Fluorescence microscopy
For the evaluation of mitochondrial membrane potential and mitochondrial mass, Src +/+ and Src −/− MEFs were treated with 200 nM tetramethylrhodamine methyl ester (TMRM, Invitrogen, CA, USA) and with 100 nM Mitotracker green™ (Invitrogen, CA, USA), respectively, during 30 min at 37 °C. Cells were then rinsed and directly imaged with the EVOS Fl Auto 2 (Invitrogen, CA, USA) at 37 °C in 5% CO 2 and 95% humidity using a 40 × objective (LPLAN 40 ×, 0.65NA, EVOS), and appropriate excitation and emission filters. For each independent experiment, fluorescence intensity was quantified in 30 cells using Fiji software (NIH, MD, USA).
For analysis of the mitochondrial morphology, cells were first presented as fragmented, tubular and elongated. The mitochondrial network was classified as fragmented when mitochondria are short and spherical; elongated when > 50% of mitochondria are longer than 5 μm and highly interconnected; tubular when the mitochondrial network appeared as an intermediate between fragmented and elongated. At For analysis of the mitochondrial number and individual mitochondrial area, ROIs of 225 μm 2 were randomly selected within peripheral regions of cells from max projection images and followed by manual thresholding, as previously described [45]. Mitochondrial number and size in ROI were obtained using the Analyze particles plugin in Fiji with a minimum area of 0.2 μm 2 . At least 10 cells were analyzed for each independent experiment (n = 3, minimum of 30 cells in total per condition).
The number of mitochondrial fusion and fission events in live cells was analyzed as previously described [23]. Briefly, cells expressing mtGFP were imaged every 2.5 s for 450 s, and then analyzed using Fiji in at least 5 cells in 3-5 independent experiments (total of 15 cells for MEFs and a total of 25 cells for HeLa). Fusion and fission events were calculated in ROI of 225 μm 2 randomly selected within peripheral regions of cells from max projection images. A single fission event was noted when one mitochondrion divided in two and remained separated for at least the next two timeframes. A fusion event was noted when two adjacent organelles connected and remained fused for at least the next three timeframes.
Mitochondrial fission was also analyzed using live cell imaging of Src +/+ and Src −/− MEFs expressing mtGFP and treated with 10 µM carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) at 37 °C in 5% CO 2 and 95% humidity. Live cells were examined at different time points using the EVOS FL Auto2 imaging system and a 40 × objective (LPLAN 40 ×, 0.65NA, EVOS), with appropriate excitation and emission filters. At least 40 cells were analyzed for each independent experiment (n = 4).
ATP levels were evaluated using the ATP FRET sensor GoAteam2, as described [26]. Src +/+ and Src −/− MEFs expressing GoAteam2 were analyzed 24 h post-transfection, with excitation at 488 nm and emission at 500-535 nm for GFP and 545-625 nm for OFP. To determine ATP levels depending on the mitochondrial ATP synthase, MEFs were treated with oligomycin (1 µg/ml). The OFP/GFP ratio was measured every 5 min in a randomly selected ROI of 25 μm 2 within peripheral regions of cells using Fiji in a minimum of 5 cells in 5 independent experiments (total of 25 cells per condition).
Several cells were analyzed in each independent experiment (as indicated). To minimize bias and make sure the analyzed cells represented the overall population of cells, coverslips were divided in four quarters for each independent experiment, from which an approximate equal number of cells were randomly selected, When appropriate, ROIs were selected within peripheral regions to ensure an appropriate distribution of all organelles, as described previously [45], in contrast to the perinuclear region, which is known to "compact" organelles due to steric hindrance leading to changes in their morphologies.

Electron microscopy
Src +/+ and Src −/− MEFs were fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer pH 7.4 at 4 °C. Samples were then incubated with 1% osmium tetroxide and 1% potassium ferricyanide in 0.1 M sodium cacodylate buffer for 1 h at 4 °C. After three water washes, samples were dehydrated in a graded ethanol series and embedded in epoxy resin. Ultrathin sections (60-70 nm) were obtained with an Ultrotome V (LKB) ultramicrotome, counterstained with uranyl acetate and lead citrate and viewed with a Tecnai G 2 (FEI) transmission electron microscope operating at 100 kV. Images were captured with a Veleta (Olympus Soft Imaging System) digital camera at the Imaging facility of the department of biology at the University of Padova (Italy).
The area of individual mitochondria (mitochondrial area, n = 131-132) and the perimeter of individual mitochondria (mitochondrial perimeter, n = 131-133) were then analyzed using Fiji.

Cellular respiration
Oxygen consumption was measured using the highresolution respirometry system Oxygraph-2k Oroboros (Innsbruck, Austria). Cell respiration was measured with 1 × 10 6 cells mL −1 according to volume-specific flux at 37 °C in 2 mL chambers at a stirring rate of 750 rpm. Three different states of endogenous respiration with intact cells were measured: (i) basal respiration representing the endogenous physiological coupled state, (ii) respiration with oligomycin (2 μg mL −1 ) representing the non-coupled resting respiration, and (iii) maximal uncoupled respiration induced by FCCP (0.5 μM steps with 2.5 μM final concentration) providing a measure of the maximal capacity of ETS under conditions of physiological substrate supply in the intact cells.

Statistical analyses
Data are presented as mean ± SEM. Statistical analyses were performed using GraphPad Prism 9. Data were analyzed using Student t test, one-way or two-way ANOVA followed by Tukey post hoc test, as appropriate. For two-way ANOVA and post hoc tests, statistical differences (p < 0.05) are presented with letters: datapoints with different letters are statistically different (p < 0.05). For instance, a datapoint with the letter a is statistically different (p < 0.05) from datapoints with the letters b or bc, whereas it is not statistically different (p > 0.05) from datapoints with the letters a or ab.