Fibroblast growth factor receptor influences primary cilium length through an interaction with intestinal cell kinase

Abstract

Vertebrate primary cilium is a Hedgehog signaling center but the extent of its involvement in other signaling systems is less well understood. This report delineates a mechanism by which fibroblast growth factor (FGF) controls primary cilia. Employing proteomic approaches to characterize proteins associated with the FGF-receptor, FGFR3, we identified the serine/threonine kinase intestinal cell kinase (ICK) as an FGFR interactor. ICK is involved in ciliogenesis and participates in control of ciliary length. FGF signaling partially abolished ICK's kinase activity, through FGFR-mediated ICK phosphorylation at conserved residue Tyr15, which interfered with optimal ATP binding. Activation of the FGF signaling pathway affected both primary cilia length and function in a manner consistent with cilia effects caused by inhibition of ICK activity. Moreover, knockdown and knockout of ICK rescued the FGF-mediated effect on cilia. We provide conclusive evidence that FGF signaling controls cilia via interaction with ICK.

Keywords: FGFR; ICK; cilia length; fibroblast growth factor; intestinal cell kinase.

Fig. 1.

FGFRs interact with ICK, MAK, and CCRK. (A) IP of FLAG-tagged ICK with V5-tagged wild-type (WT) FGFR3 or activating FGFR3 mutant K650M in 293T cells, or (B and C) FLAG-tagged MAK or CCRK with V5-tagged wild-type FGFR3 or FGFR3-K650M in 293T cells. Actin serves as a loading control. (D) IP of ICK with FGFR1, FGFR2, and FGFR4 demonstrating the ICK association with FGFR1 and FGFR4 but not FGFR2. (E) Wild-type NIH 3T3 cells were transfected with FLAG-tagged ICK together with V5-tagged FGFR3; Ick^Flag NIH 3T3 cells were transfected only with V5-tagged FGFR3. The antibodies against protein tags were used in the PLA (red); FGFR3 antibody was used to counterstain the transfected cells (green). As a negative control, cells were transfected with FGFR3 and an empty vector (WT), or by GFP (WT and Ick^Flag). Numbers of PLA dots per cell were calculated and plotted (Student’s t test, ***P < 0.001). (Scale bars, 10 µm.) Two clones of Ick^Flag NIH 3T3 cells, B11, and E5, were analyzed. (F) IP of endogenous FLAG-tagged ICK with endogenous FGFR1 in Ick^Flag NIH 3T3 cells; actin serves as a loading control. (G–I) Endogenous ICK forms a complex with endogenous FGFR1 in NIH 3T3 cells. (G) Scheme of the procedure, comprising ultracentrifugation, BN-PAGE, SDS/PAGE, and Western blot. (H) Cofractionalization of FGFR1 and ICK-FLAG in Ick^Flag(B11) NIH 3T3 cells (*). WT NIH 3T3 cells were used as a control. (I) Native complexes (fractions #6 and/or #7) were separated using BN-PAGE, followed by second dimension SDS/PAGE. Orange box shows separation of the ∼669-kDa complex containing FGFR1 and ICK-FLAG (arrow).

Fig. 2.

FGFR3 interacts with ICK via C terminus and Y724. (A) Truncated or mutated FGFR3 variants prepared for this study. Positions of point-mutated residues are indicated in red. (B) IP of FGFR3 with FLAG-tagged ICK in 293T cells. Actin serves as loading control. FGFR3 variants lacking C terminus or carrying the Y724F mutation did not co-IP with ICK. Arrows indicate FGFR3.

Fig. 3.

The ⁷⁵¹VLTVTSTDEY⁷⁶⁰ motif in FGFR3 is required for the interaction with ICK. (A) Averaged fluorescence intensities from three replicates of the peptide microarray involving peptides from FGFR3 C-terminal region. (B) Ribbon and surface representations of the crystal structure of the TK domain of FGFR3 (PDB ID code 4K33). Residue Y724 and C-terminal region implicated in ICK binding by co-IP experiments and peptide microarray analysis are highlighted in blue and orange/green, respectively. Orange, α-helix; green, the residue T⁷⁵⁵ (unstructured). Note that the absence of structural information for residues ⁷⁵⁶STDEY⁷⁶⁰ is suggestive of structural disorder. (C) The putative ICK interacting motif on FGFR3 (⁷⁵¹VLTVTSTDEY⁷⁶⁰) was replaced by the analogous sequence from FGFR2 (⁷⁶⁰ILTLTTNEEY⁷⁶⁹), creating the FGFR3-R2-C-t chimera. The co-IP of FGFR3-R2-C-t with ICK compared with wild-type FGFR3 (Student’s t test; ***P < 0.001).

Fig. 4.

FGF signaling alters ICK’s subcellular distribution. (A) 293T cells were transfected with FLAG-tagged ICK together with V5-tagged wild-type FGFR3 or its active mutant K650M. The antibodies against protein tags were used in the PLA (red); FGFR3 antibody was used to counterstain the transfected cells (green). As a negative control, cells were transfected with FGFR3 and an empty vector. Numbers of PLA dots per cell were calculated and plotted (Student’s t test, ***P < 0.001). (Scale bar, 10 µm.) (B) Increased cytosolic localization of transfected ICK in a 293T cell cotransfected with FGFR3-K650E, determined by ICK and FGFR3 immunocytochemistry (Scale bar, 20 µm.). (C and D) Altered ICK subcellular distribution in 293T cells expressing FLAG-tagged ICK, treated with FGF2; ICK was visualized by FLAG immunocytochemistry. (C) Typical localization patterns of ICK (DIC, differential interference contrast). (Scale bar, 20 µm.) (D) Percentages of cells in each category of ICK localization (Student’s t test, **P < 0.01). (E) 293T cells were transfected with FLAG-tagged wild-type FGFR3, active FGFR3-K650E, or K650M, or empty vector, and immunoblotted for phosphorylated (p) FGFR3. ICK and FGFR3 were visualized by immunocytochemistry, and ICK subcellular localization was determined.

Fig. 5.

FGFRs phosphorylate and partially inactivate ICK. (A–C) Tyrosine phosphorylation of ICK, MAK, or CCRK in cell-free kinase assay with recombinant FGFR3 and recombinant ICK, MAK or CCRK, detected by Western blot with pY antibody. (D) FGFR3-mediated tyrosine phosphorylation of ICK in cell-free kinase assay analyzed by MS. Footnotes: (1) Tyrosine position relative to human ICK sequence (NP_055735.1); (2) number of experiments varies due to limited sequence coverage in some MS samples; (3) conservation in Homo sapiens ICK/MAK, Mus musculus Ick/Mak, Gallus gallus ICK/MAK, Xenopus laevis ick/mak, Danio rerio mak, Drosophila melanogaster DmeI_CG42366; (4) conservation in H. sapiens, M. musculus, G. gallus, X. laevis, D. rerio but not in D. melanogaster; (5) representative peptide sequence found in MS, pY underlined; (6) ICK activating dual phosphorylation motif T¹⁵⁷-D-Y¹⁵⁹; (7) conservation only in ICK, not in MAK. (Middle) Sequence alignment of N-terminal region of ICK, conserved residues in gray; Y15 in red. (Bottom Left) Three-dimensional homology model of ICK kinase domain showing ATP binding site, activation loop, and substrate binding site in yellow, green, and purple, respectively, Y15 in red. (Bottom Right) Electrostatic representation of ATP binding site with bound ATP. Electronegative and electropositive sites are in blue and red, respectively. (E) The kinase activity of ICK immunoprecipitated from FGF2-treated 293T cells, measured using MBP as a substrate, in the presence of [³²P]-ATP. ICK activity is presented as a relative [³²P]MBP/MBP ratio (“Obtained”), and compared with the expected values based on ICK amounts entering the kinase assay quantified using Coomassie stained gel (ICK Coomassie) or ICK immunoblot (ICK IP) (Student’s t test, ***P < 0.001). The percentages express the extent of inhibition of the ICK kinase activity in FGF2-treated cells. (F) ICK kinase activity in 293T cells treated with FGF2, determined as a degree of ICK-mediated Thr908 phosphorylation of expressed Raptor. The ICK activity is presented as a relative pRaptor/Raptor ratio (“Obtained”) and compared with the expected values based on ICK levels. The percentages express the extent of inhibition of the ICK kinase activity in FGF2-treated cells.

Fig. 6.

FGF regulates the length of primary cilia via ICK. (A) Primary cilia length extension in NIH 3T3 cells treated with FGF2. Cilia were visualized by ARL13B, acetylated tubulin (AcTu) and pericentrin immunostaining, measured in 3D and plotted. Black dots, individual cilia; red bars, medians. [Scale bars: 5 µm (cells) and 1 µm (cilia).] (B) Rescue of FGF2-mediated cilia extension with two independent Ick shRNAs (Ick #1 and Ick #2). Ick transcript levels were monitored by qPCR at 24 h (beginning of serum starvation) and 36 h (FGF2 treatment) after transfection, and normalized to Gapdh expression. The columns show Ick expression levels relative to the scrambled control (red dashed line). Cilia length was measured 48 h after transfection and graphed. (C) Rescue of FGF2-mediated cilia extension in Ick^Flag(B11) NIH 3T3 cells stably transfected with DOX-inducible shRNA-expressing construct targeting ICK expression. ICK protein levels were monitored by immunoblot after 4 d with DOX (beginning of FGF2 treatment), normalized to actin, and plotted. Cells expressing scrambled (Scr.) shRNA upon DOX were used as a control. Student’s t test, **P < 0.01, ***P < 0.001; n.s., not significant.

Fig. 7.

FGF increases cilia length via ICK. (A) CRISPR/Cas9 targeting of Ick in triploid NIH 3T3 cells. Ick^CRISPR #1–3, three clones with all three alleles targeted; the positions of deletions are indicated (N.m.d., nonsense mediated decay). (B) Ick expression was analyzed by qPCR, and normalized to Gapdh. (C) ICK protein levels were analyzed by Western blot. (D) 293T cells were transfected with FLAG-tagged wild-type ICK, kinase-dead ICK-E80K or ICK variants harboring in-frame deletions as shown in A, and Western blot for ICK activating phosphorylation (p) at Y159 was used to determine the kinase activity. (E) Cilia in Ick^CRISPR cells were visualized by ARL13B, acetylated tubulin (AcTu), polyglutamylated tubulin (PolygluTu) or γ-tubulin (γTu) immunostaining. Both long and extremely short cilia signals are shown for Ick^CRISPR cells. (Scale bar, 1 µm.) Missing AcTu staining is indicated (arrow). (F) Ick^CRISPR cells were treated with FGF2 for 12 h and the cilia length was measured; black dots, individual cilia; red bars, medians. (G) Ick^CRISPR #3 cells were transfected with wt ICK or ICK-E80K, treated with FGF2, and the cilia length was measured. GFP transfection was used as a control. FGF2 extended primary cilia length in wild-type ICK add-back cells but not in ICK-E80K cells. (H) Human control and ICK-E80K fibroblasts were serum starved and immunostained to visualize cilia. (Scale bar, 1 µm.) Cells were treated with FGF2 and the cilia length was measured. (I) Micromasses produced from limb buds of either Ick^+/+ (control) or Ick^−/− E12 mouse littermates were treated with FGF2 for 24 h and stained with hematoxylin or ARL13B/γ-tubulin. [Scale bars: 1 mm (Left) and 5 µm (Right).] The cilia lengths were measured and graphed. Student’s t test, **P < 0.01, ***P < 0.001; n.s., nonsignificant.

Fig. 8.

Model describing regulation of primary cilia length by the FGF–ICK pathway. (A) Under basal conditions, endogenous FGF signaling restricts ICK activity to the level required for optimal length of the cilia. (B) Experimental up-regulation of ICK activity, by increased expression of active ICK, shortens the primary cilia (12, 15, 16). Similar effect is achieved in cells with abolished FGF signaling, which are unable to inhibit ICK, leading to up-regulation of ICK kinase activity and ciliary shortening (3, 4). (C) Down-regulation of ICK activity, either by transient increase in activity of FGF signaling, or by small chemicals, expression of kinase-inactive ICK mutants (*), Ick knockout or knockdown, extends primary cilia (4, 12, 13, 15, 16, 33). (D) Under certain conditions, inactivation of ICK results in profound dysregulation of primary cilia (cilia disaster), manifested by low cilia stability and extreme length variability (10, 33). A similar phenotype is achieved by strong activation of FGF signaling via disease-associated FGFR3 mutations (*) (4, 5), suggesting that the strength of the FGF stimulus regulates optimal ciliary length through ICK.