Mitochondria contact focal adhesions during migration. Previous reports indicate that regulation of calcium flux via the Mitochondrial Calcium Uniporter (MCU) and energy levels via the AMP-activated protein kinase (AMPK) are two pathways through which mitochondria affect cell motility14,15. However, it is unknown what migratory structures mitochondria might be controlling with during migration.
We speculated that mitochondria might be interacting with Focal Adhesions (FA) since these adhesive structures position themselves at the periphery of a cell and are required for many forms of mammalian cell motility20. To test this idea, we first used a confocal microscope to visualize mitochondria and FA in NIH3T3 fibroblasts (Fig. 1A and 1B). We visualized FA and mitochondria by staining cells with antibodies against Talin and Tomm20. Talin is a component of mature FA that links transmembrane integrin receptors to actin fibers in the cytosol21. Tomm20 is an import receptor translocase found on the outer mitochondrial membrane. In fixed cells (Fig. 1A, arrowed), multiple FA in the cell periphery are observed tethered to mitochondria. Surface rendering of the fixed cells (Fig. 1B) also shows FA/mitochondrial tethering. Quantitation of this interaction (Fig. 1C) revealed that approximately one quarter (25.6 + 6%) of all FA are in contact with mitochondria and approximately one fifth (20.4 ± 5%) of the total FA length is in contact with mitochondria. Approximately 8.0 ± 2.1% of all mitochondria and 7.3 ± 2% of the total cellular mitochondrial length is in contact with FA. Critically, FA in contact with mitochondria are larger than those not contacting them (Fig. 1D). FA tethering to mitochondria (n = 1,551) have, on average, an area of 5.77 ± 1.5 µm2, significantly (t-test, p < 5 x 10− 4) larger that the 5.37 ± 3.6 µm2 area of FA not tethered to mitochondria (n = 4,774). Thus, FA represent a new addition to a growing list of cellular structures that are in physical contact with mitochondria.
Interaction between mitochondria and other cellular structures is now understood to be an important aspect of the homeostasis of multiple organelles9. For example, mitochondria make important functional contacts with the endoplasmic reticulum (ER), vacuoles and peroxisome9,13. Mitochondrial contact with the ER regulates mitochondrial division12 and also allows ER-derived lipids to move to mitochondria. The larger size of FA contacting mitochondria suggests that mitochondria could be controlling cell adhesion and migration through them.
Mitochondria move to the leading edge during migration. Because of the importance of FA during cell migration, we next investigated mitochondrial/FA interaction in the context of cell migration. In both breast and ovarian cancer cells, mitochondria infiltrate into the leading edge lamellipodia during migration14,15,22. To determine whether or not this was the case in the non-transformed NIH3T3cells, we used live cell microscopy to image mitochondria in freely migrating NIH3T3 mouse fibroblasts (Fig. 2a, Supplementary Video 1). In migrating cells, mitochondria can be observed moving towards the leading edge. To extend these findings, we tracked individual mitochondria in NIH3T3 cells expressing a fluorescently tagged Cortactin protein (Fig. 2b, Supplementary Video 2). Cortactin localises prominently to the lamellipodial edge in migrating cells and regulates cell migration by binding to the Arp 2/3 complex, stabilizing actin branches and activating multiple signaling pathways necessary for motility23. As shown in Fig. 2B, multiple mitochondria with various starting locations move towards the Cortactin polarized edge (red arrow) during migration (direction indicated by white arrow). In migratory NIH3T3 cells, as in other cell types24,25, mitochondria move along microtubules. To determine if this was the case during cell migration, we imaged mitochondria (Tomm20) microtubules (tubulin) in migrating cells (Fig. 2c, Supplementary Video 3). Here, mitochondria can be seen to move along microtubules towards the leading edge during cell migration.
Mitochondria move to FA during cell migration. To determine whether or not mitochondria at the leading edge were specifically migrating to FA, we imaged migrating cells for both FA (Talin) and mitochondria (Tomm20). In these cells, we observe that as the cell moves forward, multiple mitochondria infiltrate the leading edge and interact with FA there (Fig. 3A, supplementary video 4). During this process, we also observe mitochondrial fission events prior to mitochondria moving to FA (Fig. 3A labelled “f”). To test this idea that mitochondrial interaction with FA might demonstrate leading edge/trailing edge polarization, we quantified the number of FA interacting with mitochondria relative to their position within the motile cell. We dived migrating cells (n = 5) into Leading Edge, Trailing Edge and Centre areas and measured the fraction of FA in each section contacting mitochondria. As shown in Fig. 3B, leading edge FA have almost eight times as many mitochondrial contacts per FA relative to the centre (25.6 ± 5% vs 3.27 ± 4%) and three times as many compared to the trailing edge (8.2 ± 4%). Taken together, the directional movement of mitochondria to FA during migration and the leading edge-dependent polarization of FA/mitochondrial contacts indicates that mitochondrial interaction with FA is a functional part of the cell migration program.
Inhibition of mitochondrial activity reduces Focal Adhesion size. To examine a functional role for mitochondrial interaction in FA structure, we treated NIH3T3 cells with oligomycin. Oligomycin inhibits mitochondrial ATP generation by preventing ATP synthase activity26 and has previously shown to retard migration of ovarian cancer cells14. We treated cells with oligomycin and used an image analysis program to quantitate FA length four hours later. The mean number of focal adhesions per cell is not affected by oligomycin: control NIH3T3 cells have an average of 29.8 ± 16 FA/cell (n = 177), not statistically different than the 29.2 ± 14 and 29.8 ± 21 FA/cell in 1µM and 2µM oligomycin treated cells (Fig. 4A, n = 190 and n = 160 respectively). However, oligomycin treated cells have shorter FA compared to controls. Cells treated with 1 µM or 2 µM oligomycin each have an average FA length of 2.33 ± 1.6 µm and 2.22 ± 1.8 µm respectively, significantly smaller than the average FA length of 2.79 ± 1.9 µm in untreated cells (Fig. 4A, t-test, p < 4 x10− 5 and p < 5 x 10− 4 ). Cumulative Distribution of FA length in cells shows that oligomycin significantly decreases FA length compared to controls (Fig. 4A, (Smirnov Kolmogorov test, p < 6 x 10− 4 and p < 1 x 10− 7). In particular, 1 µM and 2 µM oligomycin treated cells have respectively 51 and 55% of FA shorter than 1.8 µm compared to 47% of control cells.
To further explore the relationship between mitochondrial contact and FA, we artificially tethered mitochondria to FA. We generated constructs containing the mitochondrial targeting sequence of the human Bak gene (cBak), a glycine spacer and an mEmerald tagged full length human Talin or GFP control (Fig. 4B). Similar constructs have been used to study interaction of vinculin and talin outside of endogenous FA27. We then transfected the constructs into NIH3T3 cells and stained the transfectants for Mitochondria (Tomm20) and FA (phosphor-FAK, Fig. 4C). Both the GFP and Talin fusion proteins show predominant mitochondrial localization as well as some co-localization with FA. Cells transfected with the Talin construct show an increased number of FA/mitochondrial contacts compared to those transfected with the GFP control (Fig. 4D), indicating that the Talin fusion is functioning as expected. In cells transfected with cBak-Talin (n = 60), 26.9 ± 18% of FA/per cell are in contact with mitochondria, significantly higher (t-test, p < 0.0001) that the 18.2 ± 11% in cells transfected with cBak-GFP (n = 62 cells). Importantly, FA in cells transfected with cBak-Talin (Fig. 4E; n = 3196 FA) have a mean area of 8.91 ± 5.7 µm2, significantly larger (t-test, p < 0.0001) that the mean FA area in cells transfected with cBak-GFP, 8.43 ± 5.5 µm2 (n = 3899 FA). Cumulative distribution analysis similarly indicates that cBak-Talin transfected cells have significantly larger FA than GFP controls (Smirnov Kolmogorov test, p < 3 x 10− 4). In particular, cBak-Talin cells have 52% of their FA larger than ~ 7 µm2 in area, more than the 48% in GFP transfectants. Thus, tethering of mitochondria to FA increases their size
The regulation of FA size by mitochondria suggests that full FA assembly requires mitochondrial action. A mature FA, generally 2 µm wide x 3–10 µm long, is created from smaller adhesive structures described as focal complexes or nascent adhesions19. Our observation that oligomycin decrease FA length is consistent with the idea that localized ATP generation by mitochondria is a key part of FA maturation. High local concentrations of ATP could support actin polymerization or the signaling processes necessary for full FA assembly. FA size has been shown to regulate cell speed, with shorter focal adhesions were associated with slower cell speeds through a biphasic relationship20. Thus, a requirement for mitochondria in FA maturation provides a plausible explanation for how mitochondrial metabolism might regulate motility.