KIF14 negatively regulates Rap1a-Radil signaling during breast cancer progression

Abstract

The small GTPase Rap1 regulates inside-out integrin activation and thereby influences cell adhesion, migration, and polarity. Several Rap1 effectors have been described to mediate the cellular effects of Rap1 in a context-dependent manner. Radil is emerging as an important Rap effector implicated in cell spreading and migration, but the molecular mechanisms underlying its functions are unclear. We report here that the kinesin KIF14 associates with the PDZ domain of Radil and negatively regulates Rap1-mediated inside-out integrin activation by tethering Radil on microtubules. The depletion of KIF14 led to increased cell spreading, altered focal adhesion dynamics, and inhibition of cell migration and invasion. We also show that Radil is important for breast cancer cell proliferation and for metastasis in mice. Our findings provide evidence that the concurrent up-regulation of Rap1 activity and increased KIF14 levels in several cancers is needed to reach optimal levels of Rap1-Radil signaling, integrin activation, and cell-matrix adhesiveness required for tumor progression.

Figure 1.

KIF14 interacts with the Radil PDZ domain. (A) Recombinant GST-RadilPDZ domain was used as bait in phage display selections using a library of random heptapeptides. 25 unique binding peptides were recovered from the screen and analyzed using a sequence logo generator to identify the preferred binding sequence. The resulting C-terminal binding motif [FI]-x-WV was searched against UniProtKB/Swiss-Prot human database; and the C termini of five proteins were found to match this sequence. (B) KIF14 coimmunoprecipitates with Radil. Lysates from HEK293T cells were immunoprecipitated (IP) using α-Radil or control IgG antibodies, followed by Western blotting (WB) with α-Radil or α-KIF14 antibodies. (C) Radil interacts with KIF14 via its PDZ domain. Lysates from HEK293T cells stably expressing Strep-HA-Radil or Strep-HA-RadilΔPDZ were affinity purified (AP) using streptavidin Sepharose followed by WB using α-HA or α-KIF14 antibodies. (D) The KIF14 PDZ ligand is required to bind to Radil. HEK293T cells stably expressing Strep-HA-Radil were transiently transfected with eGFP-KIF14 wild-type (wt) or eGFP-KIF14-IQAA (WV→AA mutant) constructs. Lysates were subjected to AP with streptavidin beads followed by Western blotting with α-KIF14 and α-HA antibodies. (E) Radil binds to KIF14 directly. Recombinant GST-RadilPDZ was incubated with purified FLAG-tagged KIF14-ctail (wild-type or IQAA mutant) that were immobilized on FLAG-M2 beads. After 1 h of binding, beads were washed and protein binding determined by Western blotting using α-Radil and α-FLAG (rabbit) antibodies.

Figure 2.

KIF14 recruits Radil on microtubules. (A) Shown is localization of mCherry-Radil with eGFP-KIF14 (a–c), mCherry-Radil with eGFP-KIF14-IQAA (d–f), and mCherry-RadilΔPDZ with eGFP-KIF14 (g–i). HEK293T cells were cotransfected with low amounts of Radil expression plasmids together with KIF14 cDNA constructs. (B) Localization of endogenous Radil on microtubules is KIF14 dependent. MDA-MB-231 cells expressing scrambled- Radil or KIF14 shRNA were lysed in BRB80 buffer and cytosolic tubulin was polymerized by addition of 2 mM GTP and 20 µM paclitaxel. Rigor binding of motor proteins was allowed by addition of 2 mM AMP-PNP and the microtubule–motor protein mixture pelleted by centrifugation. The fractionated proteins were detected by Western blotting. P, pellet fraction; S, supernatant fraction.

Figure 3.

KIF14 negatively regulates Radil during MDA-MB-231 cell spreading. (A) The KIF14–Radil complex is present in MDA-MB-231 cells. Immunoprecipitation was performed using α-Radil or control IgG antibodies followed by Western blotting (WB) with α-Radil or α-KIF14 antibodies. (B) Shown is a protein–protein interaction heat map for FLAG-mRadil and FLAG-mRadilΔPDZ. Lysates from cells stably expressing FLAG-mRadil or FLAG-mRadilΔPDZ were subjected to pull-downs with α-FLAG M2 beads and the purified protein complexes were subsequently analyzed by LC-MS/MS. Colors indicate the number of experiments in which the identified proteins were detected as depicted in the legend (right). (C) Rap1–Radil signaling is required for MDA-MB-231 cell spreading. Cells expressing control shRNA, Radil shRNA#4, or overexpressing Rap1GAP cDNA were plated on fibronectin for the indicated times and cell spreading was quantified by measuring the cell area (right). (D) KIF14 negatively regulates cell spreading. Cells expressing KIF14 shRNA# 816 were plated on fibronectin-coated plates and cell spreading was monitored over the time points indicated by quantifying cell surface areas. Images were captured from five random fields for each time point. The curves were fitted by one phase-association model (see Materials and methods). (E) KIF14 negatively regulates Rap1a–Radil signaling. The increased cell spreading observed in cells expressing KIF14 shRNA is rescued by coexpressing Radil shRNA or Rap1GAP. Cells were plated on a fibronectin-coated surface (0.5 µg/ml) and allowed to adhere and spread for 180 min. At the end of the experiment cells were imaged (left) and their spreading quantified (right). Also see Fig. S2. Bars, 50 µm. Error bars, ±SEM.

Figure 4.

Radil and KIF14 differentially regulate integrin activation and focal adhesion dynamics. (A) Shown are representative TIRF images of active β1 integrin (9EG7) staining in MDA-MB-231 cells expressing scrambled shRNA, Radil shRNA #4, KIF14 shRNA #816, or Radil shRNA #4 + KIF14 shRNA #816. Cells were plated on fibronectin-coated coverslips (0.5 µg/ml) for 120 min, fixed, and stained with 9EG7 antibody. (B) FACS analysis showing increase in activated β1 integrin (9EG7) upon depletion of KIF14. Total integrin expression levels were determined by Western blot using a β1 integrin antibody that recognizes the cytoplasmic tail. β-Tubulin was used as loading control. (C) Focal adhesion sites were imaged using TIRF microscopy and vinculin staining in cells treated as above. Threshold images were used to quantify FA areas and numbers. (D) Graphical representation of FA area versus number in individual cells taken from three independent experiments. (E) Average FA area for all cells quantified in D. (F) Confocal images of F-actin in cells treated as indicated and following spreading for 120 min on fibronectin coated coverslips (n = 3). A line was drawn across cells and the intensity profile plotted (below) to show the relative distribution of actin stress fibers in cells. Also see Fig. S3. Bars, 20 µm. Error bars, ±SEM.

Figure 5.

KIF14 knockdown alters focal adhesion dynamics. (A) Venus-Paxillin dynamics was assessed by time lapse TIRF microscopy in MDA-MB-231 cells expressing scrambled or two different KIF14 shRNAs. Cells were plated on fibronectin (0.2 µg/ml)-coated glass slides. Each cell was observed over 54 min and a picture was taken every 3 min. Merge shows overlay of images taken at time 0 and 54. Red = 0 min; green = 54 min. (B) Quantification of Venus-Paxillin dynamics shown in A. Bars, 5 µM. Error bars, ±SEM.

Figure 6.

Rap1a activation releases Radil and KIF14 from microtubules. (A) Confocal images of HEK293T cells expressing mCherry-Radil and eGFP-KIF14 with or without HA-Rap1aQ63E or FLAG-Rap1GAP. Expression of constitutively active Rap1a leads to dislocation of Radil and KIF14 from microtubules and recruitment of Radil to the plasma membrane. Bars, 10 µm. (B) Confocal images of MDA-MB-231 cells expressing FLAG-Dishevelled2 (FLAG-Dsh2) or FLAG-Radil transduced with scrambled shRNA or KIF14 shRNA #816. A line was drawn across each cell and the profile of protein distribution depicted below each image (The data shown are from a single representative experiment out of three repeats.). Bars, 20 µm.

Figure 7.

Radil and KIF14 are required for MDA-MB-231 cell migration and invasion. (A) MDA-MB-231 cells were transduced with the indicated shRNAs. 72 h after transduction 5 × 10⁴ cells were seeded on the upper chamber of Transwells and 20% fetal bovine serum (FBS) in DMEM was applied to the lower chamber. Cells were allowed to migrate for 10–12 h and migrated cells counted from pictures of 4–5 random fields. (B) Cell invasion was assessed as above except Transwells coated with 1 µg/ml of matrigel were used. Cells were allowed to invade through the matrigel for 20 h. (C) MDA-MB-231 cells stably expressing the shRNA-resistant murine full-length mRadil or mRadilΔPDZ were transduced with scrambled, or two different Radil shRNAs. The migratory potential of these cells was assessed as described above. Expression levels of FLAG-mRadil and FLAG-mRadilΔPDZ are shown on the right. Cortactin used as loading control. (D) Transwell cell migration and (E) invasion assays with cells expressing KIF14 shRNAs. Cells were processed as above. (F) MDA-MB-231 cells were transduced with scrambled shRNA or KIF14 shRNA #816 in the presence of FLAG-GFP, eGFP-KIF14, or eGFP-KIF14-IQAA. The different cells were then subjected to Transwell assays. Expression levels of eGFP-KIF14 and eGFP-KIF14-IQAA are shown on the right. KIF14 shRNA #816 targets the 3′ UTR and was thus used for rescue experiments. Bars, 50 µm. Error bars, mean ± SEM.

Figure 8.

Radil is required for breast cancer cell metastasis. (A) MDA-MB-231 cells stably expressing Renilla luciferase were transduced with scrambled shRNA or Radil shRNA #4 and their migratory potential was tested using Transwell assays before injection in mice. (B) MDA-MB-231 cells were injected via tail vein in NOD-SCID mice. Metastasis and homing of cells to the lungs was monitored over time using in vivo bioluminescence imaging. Shown are images from day 35. (C) Mice were sacrificed 45 d after inoculations. Lungs from mice injected with saline or with scrambled shRNA or Radil shRNA–expressing cells were collected, fixed, and stained with Bouin’s solution. Shown are representative images from each condition (top). Quantification of the visible tumor nodules is provided in the table (below). Scrambled shRNA, n = 12; Radil shRNA #4, n = 11; statistics, Student’s t test. (D) Total number of nodules counted on the lungs of each animal. Error bars, ±SEM. (E) In vitro proliferation of MDA-MB-231 cells treated with scrambled or Radil shRNA #4. (F) Quantification of tumor end volume (in cm³) on day 31. Scrambled shRNA, n = 5; Radil shRNA #4, n = 5. Red line denotes median value. Bars, interquartile range. (G) Quantification of tumor end weight (in grams) on day 31.

Figure 9.

Proposed model describing the roles of Radil and KIF14 during cancer cell migration. (A) In highly motile cancer cells, KIF14 is up-regulated and sequesters Radil on microtubules. This enables optimal Radil–Rap1 signaling, inside-out integrin signaling, and cell–matrix adhesive properties required for efficient cell migration. (B) Depletion of Radil inhibits integrin activation, reduces cell–matrix adhesion, and causes loss of traction and cell motility. (C) Depletion of KIF14 leads to the release of Radil from microtubules, thereby increasing the pool of Radil available to associate with activated Rap1 at the plasma membrane. This leads to hyperactivated integrins, increases focal adhesion formation, and decreases motility.