Aurora Kinase A Is Involved in Controlling the Localization of Aquaporin-2 in Renal Principal Cells

Abstract

The cAMP-dependent aquaporin-2 (AQP2) redistribution from intracellular vesicles into the plasma membrane of renal collecting duct principal cells induces water reabsorption and fine-tunes body water homeostasis. However, the mechanisms controlling the localization of AQP2 are not understood in detail. Using immortalized mouse medullary collecting duct (MCD4) and primary rat inner medullary collecting duct (IMCD) cells as model systems, we here discovered a key regulatory role of Aurora kinase A (AURKA) in the control of AQP2. The AURKA-selective inhibitor Aurora-A inhibitor I and novel derivatives as well as a structurally different inhibitor, Alisertib, prevented the cAMP-induced redistribution of AQP2. Aurora-A inhibitor I led to a depolymerization of actin stress fibers, which serve as tracks for the translocation of AQP2-bearing vesicles to the plasma membrane. The phosphorylation of cofilin-1 (CFL1) inactivates the actin-depolymerizing function of CFL1. Aurora-A inhibitor I decreased the CFL1 phosphorylation, accounting for the removal of the actin stress fibers and the inhibition of the redistribution of AQP2. Surprisingly, Alisertib caused an increase in actin stress fibers and did not affect CFL1 phosphorylation, indicating that AURKA exerts its control over AQP2 through different mechanisms. An involvement of AURKA and CFL1 in the control of the localization of AQP2 was hitherto unknown.

Keywords: AQP2; AURKA; AVP; actin cytoskeleton; cofilin-1.

Figure A1

Supplement to Figure 3. Effects of Aurora-A inhibitor I derivatives on the localization of AQP2 in MCD4 cells. MCD4 cells were left untreated or treated with DMSO (0.3%, 1 h) or compound (30 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min) where indicated. AQP2 was detected by immunofluorescence microscopy as described in the legend of Figure 1. Shown are representative images of four independent experiments; scale bar 30 µm.

Figure 1

Aurora-A inhibitor I inhibits the cyclic adenosine monophosphate (cAMP)-induced aquaporin-2 (AQP2) redistribution to the plasma membrane of (A) immortalized mouse medullary collecting duct (MCD4) and (B) primary rat inner medullary collecting duct (IMCD) cells. The cells were left untreated or treated with DMSO (0.3%, 1 h) or Aurora-A inhibitor I (1 µM, 10 µM or 30 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min) or arginine-vasopressin (AVP; 100 nM, 30 min) where indicated. AQP2 was detected by immunofluorescence microscopy using specific primary antibody (AQP2, E-2) and Cy3-coupled anti-mouse secondary antibody (red). Nuclei were stained with DAPI (cyan). Shown are representative images of three independent experiments; scale bar 30 µm. Magnified views (zoom) are from the indicated boxes. The intensities of plasma membrane and perinuclear immunofluorescence signals arising from AQP2 were determined. The ratios of plasma membrane to perinuclear fluorescence signal intensities were calculated. Ratios >1 indicate a predominant localization at the plasma membrane. Shown are means ± SEM of three independent experiments with a total of 75 cells per condition. Statistically significant differences compared to DMSO-treated control cells are indicated, **** p < 0.0001, *** p < 0.001.

Figure 2

Inhibition of Aurora kinase A (AURKA) but not Aurora kinase B (AURKB) inhibits the redistribution of AQP2. (A) MCD4 or (B) primary IMCD cells were left untreated or treated with DMSO (0.3% or 0.2%, 1 h), Alisertib (AURKA inhibitor; 30 µM or 20 µM, 1 h), or Barasertib-HQPA (AURKB inhibitor; 30 µM or 20 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min) or arginine-vasopressin (AVP; 100 nM, 30 min) where indicated. AQP2 was detected by immunofluorescence microscopy as described in the legend of Figure 1. Shown are representative images of four independent experiments; scale bar 30 µm. AQP2 was quantified using a segmentation-based approach [42]. Ratios > 1 indicate a predominant localization at the plasma membrane. Shown are means ± SEM of four independent experiments with ≥25 cells per image per condition. Statistically significant differences are indicated, **** p < 0.0001, ** p < 0.01, * p < 0.05.

Figure 3

Aurora-A inhibitor I derivatives inhibit the forskolin-induced redistribution of AQP2 to the plasma membrane of MCD4 cells. Cells were left untreated or treated with DMSO (0.3%, 1 h) or compound (30 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min). AQP2 was detected by immunofluorescence microscopy as described in the legend of Figure 1. Shown are representative images of four independent experiments; scale bar 30 µm. AQP2 was quantified using a segmentation-based approach [42]. Ratios > 1 indicate a predominant localization at the plasma membrane. Shown are means ± SEM of four independent experiments with ≥25 cells per image per condition. Statistically significant differences between resting and forskolin-stimulated cells are indicated, *** p < 0.001, ** p < 0.01, * p < 0.05. Additional images can be found in Appendix A (Figure A1).

Figure 4

Aurora-A inhibitor I derivatives inhibit the AVP-induced redistribution of AQP2 to the plasma membrane of primary IMCD cells. Cells were left untreated or treated with DMSO (0.2%, 1 h) or compound (20 µM, 1 h) alone or in combination with arginine-vasopressin (AVP; 100 nM, 30 min). AQP2 was detected by immunofluorescence microscopy and quantified as described in the legend of Figure 1. Shown are representative images of three independent experiments; scale bar 30 µm. Ratios > 1 indicate a predominant localization at the plasma membrane. Shown are means ± SEM of three independent experiments with a total of 75 cells per condition. Statistically significant differences compared to DMSO-treated control cells are indicated, **** p < 0.0001, ** p < 0.01.

Figure 5

Effects of Aurora-A inhibitor I and its derivatives on the viability of MCD4 cells. Cell proliferation was assessed using the CellTiter 96^® Aq_ueous One Solution Cell Proliferation Assay (Promega; Walldorf, Germany). MCD4 cells were left untreated (Utr) or treated with DMSO (0.3%, 1 h) or compound (30 µM, 10 µM or 1 µM, 1 h). As negative controls, i.e., to induce maximal toxicity, cells were incubated with ethanol (EtOH). Shown are means ± SEM of three or six independent experiments. Statistically significant differences compared to untreated cells are indicated, **** p < 0.0001, ** p < 0.01, * p < 0.05.

Figure 6

Aurora-A inhibitor I and its derivatives are mostly selective for AURKA. (A) Thermal shift assay to determine binding of Aurora-A inhibitor I and its derivatives to different kinases. Staurosporine served as a positive control; n = 3. Unit ΔT_m (°C). (B) The affinity of the test compounds to intracellular AURKA/B/C was determined by NanoBRET measurements. Staurosporine and NVP-Tae684 served as positive controls; n = 2; shown are means ± SD. Affinities are listed in molar concentration (M) if not stated otherwise. Affinities are color-coded with the lowest values on dark red background; lowest binding is indicated by no background.

Figure 7

AURKA inhibitors do not affect global protein kinase A (PKA) or protein phosphatase activity in MCD4 cell lysates. (A) PKA activity in MCD4 cell lysates was determined using the PepTag^® Non-Radioactive Protein Kinase Assay (Promega; Walldorf, Germany). Cells were left untreated or treated with DMSO (0.3%, 1 h) or compound (30 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min). cAMP (1 µM) was added to induce maximal PKA activity. PKA catalytic (C) subunit (2 µg/mL) was used in positive controls and omitted in negative controls. Relative PKA activity was determined by calculating the ratio of phosphorylated to non-phosphorylated peptides, which were semi-quantitatively analyzed by densitometry. Shown are representative agarose gels and means ± SEM of three independent experiments. Statistically significant differences are indicated, *** p < 0.001, ** p < 0.01, * p < 0.05. (B) Phosphatase activity in MCD4 cell lysates was determined by para-Nitrophenylphosphate (pNPP)-based activity assays. Cells were left untreated or treated with DMSO (0.3%, 1 h) or compound (30 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min). Cell lysates were incubated with a pNPP substrate solution and absorbance measured at 405 nm. Shown are means ± SEM of two to five independent experiments.

Figure 8

AURKA inhibitors affect the organization of actin stress fibers in MCD4 cells. MCD4 cells were left untreated (utr) or treated with DMSO (0.3%, 1 h) or compound (30 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min). AQP2 was detected by immunofluorescence microscopy using specific primary antibody (AQP2, E-2) and Cy3-coupled anti-mouse secondary antibody (red). Filamentous (F-)actin was stained using an iFluor 488-conjugated phalloidin (grey). Nuclei were stained with DAPI (cyan). Shown are representative images of three independent experiments; scale bar 30 µm. Magnified views (zoom) of actin stress fibers are from the indicated boxes.

Figure 9

AURKA inhibitors affect the organization of actin stress fibers in primary IMCD cells. Primary IMCD cells were left untreated (utr) or treated with DMSO (0.2%, 1 h) or compound (20 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min) or arginine-vasopressin (AVP; 100 nM, 30 min). AQP2 was detected by immunofluorescence microscopy using specific primary antibody (AQP2, E-2) and Cy3-coupled anti-mouse secondary antibody (red). F-actin was stained using an iFluor 488-conjugated phalloidin (grey). Nuclei were stained with DAPI (cyan). Shown are representative images of three independent experiments; scale bar 30 µm. Magnified views (zoom) of actin stress fibers are from the indicated boxes.

Figure 10

AURKA inhibitors affect the phosphorylation of cofilin-1 (CFL1) in MCD4 cells. Cells were left untreated or treated with DMSO (0.3%, 1 h) or compound (30 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min). Cell lysates were prepared and CFL1, phosphorylated (p-)CFL1, and cadherin as loading control were detected by Western blotting using specific antibodies. The signals were semi-quantitatively analyzed by densitometry and normalized to the loading control. Shown are representative blots and means ± SEM of three to five independent experiments. Statistically significant differences compared to DMSO-treated control cells are indicated, *** p < 0.001, ** p < 0.01, * p < 0.05.

Figure 11

AURKA inhibitors affect the phosphorylation of CFL1 in primary IMCD cells. Cells were left untreated or treated with DMSO (0.2%, 1 h) or compound (20 µM, 1 h) alone or in combination with forskolin (Fsk; 30 µM, 30 min) or arginine-vasopressin (AVP; 100 nM, 30 min). Cell lysates were prepared and CFL1, phosphorylated (p-)CFL1 and Hsp60 as loading control were detected by Western blotting using specific antibodies. The signals were semi-quantitatively analyzed by densitometry and normalized to the loading control. Shown are representative blots and means ± SEM of three independent experiments.

Figure 12

AURKA and AURKB inhibitors do not affect the phosphorylation of LIM kinase-1 (LIMK-1) in primary IMCD cells. Cells were left untreated or treated with forskolin (Fsk; 30 µM, 30 min), DMSO (0.2%, 1 h), or compound (20 µM, 1 h) alone or in combination with arginine-vasopressin (AVP; 100 nM, 30 min). Cell lysates were prepared and LIMK-1, phosphorylated (p-)LIMK-1 (threonine (Thr) 508), and Hsp60 as loading control were detected by Western blotting using specific antibodies. The signals were semi-quantitatively analyzed by densitometry and normalized to the loading controls. Shown are representative blots and means ± SEM of three independent experiments. Statistically significant differences are indicated, ** p < 0.01.

Figure 13

Aurora-A inhibitor I and Alisertib inhibit the redistribution of aquaporin-2 (AQP2) in renal principal cells by different mechanisms. (A) Under physiological conditions, the binding of arginine-vasopressin (AVP) to the vasopressin V2 receptor (V2R) activates adenylyl cyclase (AC) and increases intracellular cyclic adenosine monophosphate (cAMP). The redistribution of AQP2 to the plasma membrane involves actin remodeling. The actin regulatory protein cofilin-1 (CFL1) is phosphorylated (P) at serine (Ser) 3 and thereby inactivated by LIM kinase 1 (LIMK-1) and Aurora kinase A (AURKA). Inactive CFL1 promotes the formation and maintenance of filamentous (F-)actin stress fibers. The depolymerization of F-actin into globular (G-)actin monomers is induced after reactivation of CFL1 by dephosphorylation by slingshot phosphatase 1 (SSH-1) and chronophin (CIN). (B) Aurora-A inhibitor I inhibits AURKA and decreases the CFL1 phosphorylation in stimulated cells. It thereby promotes the F-actin depolymerization. (C) Alisertib inhibits AURKA and increases the amount of F-actin stress fibers without affecting the CFL1 phosphorylation, indicating that another signaling pathway is involved.