Constitutive regulation of mitochondrial morphology by Aurora A kinase depends on a predicted cryptic targeting sequence at the N-terminus

Abstract

Aurora A kinase (AURKA) is a major regulator of mitosis and an important driver of cancer progression. The roles of AURKA outside of mitosis, and how these might contribute to cancer progression, are not well understood. Here, we show that a fraction of cytoplasmic AURKA is associated with mitochondria, co-fractionating in cell extracts and interacting with mitochondrial proteins by reciprocal co-immunoprecipitation. We have also found that the dynamics of the mitochondrial network are sensitive to AURKA inhibition, depletion or overexpression. This can account for the different mitochondrial morphologies observed in RPE-1 and U2OS cell lines, which show very different levels of expression of AURKA. We identify the mitochondrial fraction of AURKA as influencing mitochondrial morphology, because an N-terminally truncated version of the kinase that does not localize to mitochondria does not affect the mitochondrial network. We identify a cryptic mitochondrial targeting sequence in the AURKA N-terminus and discuss how alternative conformations of the protein may influence its cytoplasmic fate.

Keywords: AURKA; mitochondria; mitochondrial dynamics.

Figure 1.

AURKA is a constitutive regulator of the mitochondrial network. (a) RPE-1 cells were treated with control (GL2-i) or AURKA siRNA (AURKA-i) for 48 h and mitochondria imaged in live cells using MitoTracker^TM. Areas of cytoplasm marked by white squares are shown enlarged twofold in panels to the side of each image. Mitochondria were analysed for tubular fragment length as described in Material and methods, with raw measurements shown as probability density plots. p < 0.001 for the maximum deviation D = 0.40 (K-S test). (b,c) RPE-1-mRuby-PCNA cells were stained with Mito-ID^® green and imaged after treatment with 100 nM MLN8237 (MLN) or vehicle control (DMSO) for 3 h. (b) Tubular mitochondrial lengths are plotted as probability density curves (left-hand panel) for cells assigned to G1, S or G2 phase according to localization of mRuby-PCNA (middle panel), with cell cycle distribution summarized in the plot shown in the right-hand panel. (c) Probability density plots showing increased mitochondria length in all phases of the cell cycle after MLN treatment. G1: p < 0.001, D = 0.39; S: p < 0.001, D = 0.23; G2: p < 0.001, D = 0.13 (K-S test). (d) Drosophila D.mel-2 cells were treated with MLN for 3 h and processed as in (a). p < 0.001, D = 0.53 (K-S test). For panels (b–d): raw measurements are pooled from three statistically reproducible experimental repeats. For panels (a–d): scale bars, 10 µm in main panels, 1 µm in magnification panels.

Figure 2.

AURKA levels influence mitochondrial morphology in RPE-1 and U2OS cells. (a) RPE-1 and U2OS cells were treated with MitoTracker^TM and imaged under identical conditions. Mitochondrial lengths were measured and plotted as in figure 1. p < 0.001 for the maximum deviation D = 0.37 (K-S test) from two repeats. (b) Equal numbers of U2OS and RPE-1 cells were harvested for cell extracts and calculated quantities loaded onto gels to be examined by quantitative immunoblot. Bar charts show AURKA levels quantified and normalized against number of cells, or against different loading markers. Statistical confidence is indicated as *p < 0.01; **p < 0.001 (Student's t-test), n = 3 repeats. (c,d) U2OS cells were treated with control (GL2-i) or AURKA siRNA (AURK-i) for 48 h and then either processed for quantitative immunoblotting of AURKA levels (c) or stained with MitoTracker^TM (d). Example images show insets indicated by white boxes magnified twofold. Mitochondrial tubular length measurements are plotted as probability density curves. p < 0.001, D = 0.73 (K-S test). (e,f) RPE-1-AURKA-Venus cells were induced for AURKA-Venus expression (AURKA OE) or not (control) with addition of tet for 18 h. Cells were either processed for quantitative immunoblotting of AURKA levels (e) or stained with MitoTracker^TM for mitochondrial tubular length measurements (f), p < 0.01, D = 0.08 (K-S test). tet, tetracycline; endog, endogenous; OE, overexpression. For panels (a,d,f): scale bars, 10 µm in main panels, 1 µm in magnification panels.

Figure 3.

A mitochondrial fraction of AURKA. (a) U2OS and U2OS-AURKA-Venus cells were fractionated by serial centrifugation and probed for the presence of exogenous (exog) and endogenous (endog) AURKA, and of specific markers for cytosolic (GAPDH, MEK1) or mitochondrial (ATP5A1) fractions, by immunoblot. (b) Co-localization of endogenous AURKA (green) and mitochondrial marker TOMM20 (red) examined by IF in RPE-1 cells. (c) Live cell imaging of endogenously tagged RPE-1 mVenus-AURKA knock-in (KI) cells stained with MitoTracker^TM to examine co-localization. (d) RPE-1-AURKA-Venus cells were treated with MitoTracker^TM 24 h after induction of AURKA-Venus expression, then methanol-fixed, stained using GFP antibody and imaged on an OMX system. The image shown is a maximum-intensity projection of a 12 × 0.125 µm stack, with 10-fold magnification of insets. For panels (b,c): scale bars, 10 µm in main panels, 1 µm in magnification panels. For panel (d): scale bars, 10 µm in main panels, 100 nm in magnification panels.

Figure 4.

A cryptic mitochondrial targeting sequence resides in the AURKA N-terminus. (a,b) U2OS cells transiently transfected with AURKA-Venus (AURKA), N-terminally truncated AURKA (AURKAΔ31) or Venus alone were fractionated and probed by immunoblot (a) for the presence of Venus (anti-GFP) or with various markers for cytosolic (MEK1, PI31) and mitochondrial (ATP5A1, TOMM20) fractions. SUG1 was used to control for whole-cell lysate (WCL). Venus levels in the mitochondrial fraction (Mito) were quantitated and normalized against the level in WCL (b) in three separate experiments. **p < 0.01; ***p < 0.001 (Student's t-test). (c) N-terminally truncated AURKA (Δ31) does not cause fragmentation of the mitochondrial network. RPE-1 cells transfected with Venus-tagged AURKA, full-length and -Δ31, or Venus control, were treated with MitoTracker^TM and imaged under identical conditions. Mitochondrial tubular lengths are plotted as probability density curves. The two-sample K-S tests of AURKA and AURKAΔ31 populations give p < 0.001 for maximum deviation D = 0.34, from two repeats. (d) Likely amphipathic helix in the AURKA N-terminus revealed by Kyte–Doolittle plot (left-hand panel) and Helical wheel projection (right-hand panel) [30]. Yellow, hydrophobic; blue and purple, hydrophilic; grey, neutral. (e) MitoProt prediction of mitochondrial targeting probability [31] for in silico sequential N-terminal truncations of AURKA. (f,g) RPE-1 cells were transfected with wild-type or N-terminally truncated versions of AURKA-Venus (Δ6, Δ8) and imaged 24 h later after staining with MitoTracker^TM for measurements of mitochondrial tubular length (f). By two-sample K-S tests, WT versus Δ6: p < 0.001, D = 0.13; WT versus Δ8: p < 0.001, D = 0.25. Cells were then fixed and processed for IF with GFP and TOMM20 antibodies (g, see also electronic supplementary material, figure S5). For panel (g): scale bars, 10 µm in main panels, 1 µm in magnification panels.