Research ArticleCell Biology

The PPFIA1-PP2A protein complex promotes trafficking of Kif7 to the ciliary tip and Hedgehog signaling

See allHide authors and affiliations

Science Signaling  09 Dec 2014:
Vol. 7, Issue 355, pp. ra117
DOI: 10.1126/scisignal.2005608


The primary cilium is required for Hedgehog (Hh) signaling in vertebrates. Hh leads to ciliary accumulation and activation of the transmembrane protein Smoothened (Smo) and affects the localization of several pathway components, including the Gli family of transcriptional regulators, within different regions of primary cilia. Genetic analysis indicates that the kinesin protein Kif7 both promotes and inhibits mouse Hh signaling. Using mass spectrometry, we identified liprin-α1 (PPFIA1) and the protein phosphatase PP2A as Kif7-interacting proteins, and we showed that they were important for the trafficking of Kif7 and Gli proteins to the tips of cilia and for the transcriptional output of Hh signaling. Our results suggested that PPFIA1 functioned with PP2A to promote the dephosphorylation of Kif7, triggering Kif7 localization to the tips of primary cilia and promoting Gli transcriptional activity.


The Hedgehog (Hh) signaling pathway plays critical roles during embryonic development and in adult animals for the patterning and homeostasis of tissues. Cells respond to Hh ligands using partially conserved signal transduction machineries that regulate target genes affecting the fate, proliferation, and survival of several progenitor and stem cell populations. Consequently, mutational activation of the Hh pathway is linked to several developmental disorders and human cancers such as basal cell carcinoma and medulloblastoma (1).

Primary cilia, which are microtubule-based organelles that protrude from the plasma membrane of interphase cells, are signaling centers that coordinate different developmental signaling pathways (2, 3). When cilia ontogeny or function is defective, human diseases and developmental disorders arise, such as Bardet-Biedl syndrome, Joubert syndrome, and polycystic kidney disease, which are collectively referred to as ciliopathies (3, 4). The realization that the primary cilium fulfills essential roles during Hh signaling has modified our understanding of how responding cells sense, integrate, and react to the presence of this morphogen (5, 6). Reciprocally, because it is essential for pathway activity, cilia dysfunction almost invariably results in defective Hh signaling (7, 8).

In vertebrates, three Gli proteins (Gli1, Gli2, and Gli3) act as transcriptional mediators of the pathway to activate and repress Hh-dependent target genes (9). Genetic and biochemical evidence has established Gli2 as the major activator (GliA) of the pathway (10, 11), whereas partial proteolysis by the ubiquitin-proteasome system cleaves Gli3 into a transcriptional repressor (GliR) (12, 13). Gli1 is an Hh target gene in most cellular contexts where it potentiates pathway activation as part of a positive feedback loop (14). Thus, under resting conditions, formation of GliR keeps the output of the pathway in the “off” state, whereas the presence of Hh inhibits GliR and promotes GliA formation. The fine balance of GliA and GliR abundance therefore dictates the graded output of Hh signaling. In the resting state, the Hh receptor Patched-1 (Ptch1) localizes to cilia where it actively prevents the accumulation of the seven-transmembrane protein Smoothened (Smo) (15). Under these conditions, the inhibitor SuFu (Suppressor of Fused) interacts with Gli proteins (1618), prevents spurious activation of Gli target genes, and promotes the formation of GliR. During pathway activation, the abundance of Ptch1 in the cilia decreases, whereas Smo, Kif7, SuFu, and Gli accumulate into primary cilia and are enriched in various cilia subdomains (7, 19, 20). Smo activation leads to dissociation of Gli proteins from SuFu (21, 22), inhibits GliR formation (23), and promotes GliA activation and its translocation into the nucleus to modulate context- and Hh-specific transcriptional programs (24). Perturbation of normal intraflagellar transport leads to defective localization of several core pathway components in the primary cilia, thus revealing that these proteins dynamically traffic and are stabilized in cilia during pathway activation (25, 26). The precise understanding of how Hh can promote the trafficking and accumulation of several pathway members within cilia and the functional importance of this process for the transcriptional output of the pathway remains incomplete.

Kif7 mutations affecting Hh signaling and cilia functions are found in human ciliopathies (27, 28). Genetic evidence has also established that the kinesin protein Kif7 fulfills important functions during murine Hh signal transduction. For instance, analysis of Kif7 knockout mice has uncovered negative and positive roles for Kif7 during neural tube patterning, limb morphogenesis (2932), chondrogenesis (33), and hair follicle development (34). Kif7 interacts with Gli proteins and accumulates at the tips of primary cilia when the Hh pathway is active. Examination of Kif7–/– mouse embryonic fibroblasts (MEFs) revealed reduced Gli3R (2931) and decreased Hh-promoted accumulation of Gli at the tips of primary cilia (30, 35), suggesting that Kif7 may be important for these processes. The precise molecular mechanisms underlying the positive and negative roles of Kif7, its biochemical and regulatory mechanisms, and how the disease-causing mutations affect Kif7-dependent processes are poorly understood. More work is thus needed to further understand how Smo activation influences Kif7 trafficking in cilia and how this, in turn, affects Gli localization, processing, and function to promote Hh-target gene expression.

Here, we used Kif7–/– and Ptch1–/– MEFs to study the mechanisms of Kif7 during Hh signaling. We report that in MEFs, Kif7 fulfilled a predominantly positive role downstream of Smo activation because it promoted the ciliary tip accumulation of Gli proteins and the induction of Gli target genes. We further described a protein complex, consisting of liprin-α1 (PPFIA1) and the serine/threonine phosphatase PP2A, which interacted with Kif7. Our results indicated that the PPFIA1-PP2A complex promoted the Hh signaling–dependent localization of Kif7 and Gli proteins at the tips of cilia and the activation of Hh target genes. We showed that Kif7 phosphorylation dictated its subcellular localization at the tips of primary cilia and determined whether it inhibited or activated Hh signaling. Our findings are consistent with a model in which Smo activation is coupled to the trafficking of Kif7 and Gli proteins at cilia tips and to their transcriptional output through the dephosphorylation of Kif7 mediated by PP2A.


Kif7 promotes Gli ciliary tip localization and Hh signaling

The correlation of the accumulation of Kif7 and Gli3 proteins at the tips of cilia during pathway activation suggests that Kif7 may be required for the ciliary trafficking of Gli proteins (30, 35). To directly test this, we showed that upon activation of Hh signaling with the Smo agonist SAG, the trafficking of Gli2 and Gli3 to the tips of cilia, but not the accumulation of Smo, was decreased in Kif7–/– MEFs (Fig. 1A). The lentiviral delivery of Kif7-GFP (green fluorescent protein) in Kif7–/– MEFs rescued Kif7 and Gli ciliary localization to the tips of cilia upon SAG treatment. In the absence of stimulation with SAG, overexpression of Kif7-GFP was not sufficient to promote localization of Gli proteins to the tips of cilia or to induce Gli1 mRNA expression (Fig. 1B and fig. S1). In Kif7–/– MEFs, the basal expression of the Hh target gene Gli1 was minimally affected, but its induction by SAG treatment was compromised when compared to MEFs reconstituted with Kif7 (Fig. 1C). This result suggested that the formation of GliA is impaired in the absence of Kif7. We conclude that during Hh pathway activation, Kif7 promotes the localization of Gli proteins at the tips of cilia and that in MEFs, Kif7 chiefly exerts a positive signaling role.

Fig. 1 Loss of Kif7 impairs Gli trafficking and Gli1 mRNA induction.

(A) The SAG-induced accumulation of Gli2 and Gli3 at the tips of cilia is impaired in Kif7–/– MEFs, whereas the SAG-induced ciliary accumulation of Smo is unaffected. Acet Tub, acetylated tubulin. (B) Stable expression of Kif7-GFP in Kif7–/– MEFs (Kif7–/– MEFs+Kif7-GFP) rescues the normal localization of Kif7, Gli2, and Gli3 to the tips of cilia after SAG treatment. (C) The time- and SAG-dependent induction of the Hh target gene Gli1 is impaired in Kif7–/– MEFs but is rescued in Kif7–/– MEFs + Kif7-GFP. Gli1 mRNA expression was calculated relative to that in Kif7–/– MEFs + Kif7-GFP treated with SAG for 24 hours. Quantification of ciliary tip localization and quantitative real-time polymerase chain reaction (qPCR) data are presented as means ± SEM across three biological replicates. NT, nontreated control. *P < 0.05. Scale bars, 1 μm.

PPFIA1 associates with Kif7

To further understand the regulation and functional importance of protein trafficking during Hh signaling, we used mass spectrometry (MS) to sample the composition of Gli3- and Kif7-containing protein complexes purified from mammalian cells (Fig. 2, A and B, and table S1). Gli3 interacted with the Hh signaling regulators SuFu and Kif7. Consistent with the localization of Kif7 in cilia, several centrosome and basal body proteins were identified in Kif7 complexes. We also identified PPFIA1, a member of the liprin family of scaffolding proteins previously recognized for their role in synaptogenesis, in Kif7 purifications (36). In support of this finding, Kif7 was detected in PPFIA1 immunoprecipitates by liquid chromatography (LC)–MS/MS and endogenous Kif7 and PPFIA1 coimmunoprecipitated (Fig. 2C). Stimulation of cells with the Smo agonist SAG increased the interaction between PPFIA1 and Kif7 (Fig. 2C), hinting that formation of this complex is important during Hh signaling. Domain-mapping experiments using different Kif7 and PPFIA1 mutants showed that these proteins interacted through their coiled-coil domains (fig. S2). Consistent with previous reports (3739), several peptides assigned to regulatory and catalytic subunits of PP2A were identified in PPFIA1 immunoprecipitates. Endogenous regulatory PP2A subunit PPP2R1A could only be coimmunoprecipitated together with FLAG-Kif7 when PPFIA1 was overexpressed in cells (Fig. 2D). These results suggested that Kif7 interacts indirectly with the PP2A phosphatase through binding to PPFIA1. We conclude that PPFIA1 and the PP2A phosphatase form a protein complex with Kif7.

Fig. 2 MS analysis of Kif7, Gli3, and PPFIA1 protein complexes.

(A) Heat map representation of high-confidence Kif7, Gli3, and PPFIA1 interacting proteins identified by FLAG immunoprecipitation (IP)–MS. Sum of the spectral counts across two biological replicates for each of the bait is represented as a color code and sorted from highest to lowest spectral counts for the PPFIA1 bait. (B) Cytoscape representation of Kif7, Gli3, and PPFIA1 interactomes. Red edges indicate that the protein-protein interaction has been previously reported. See inset for legend. (C) Coimmunoprecipitation of endogenous Kif7 with PPFIA1 from lysates of MEFs treated or not with SAG. Right: Quantification of three independent experiments. WB, Western blot. IgG, immunoglobulin G. (D) Coimmunoprecipitation of regulatory subunit of PP2A (PPP2R1A) and FLAG-Kif7 with or without overexpression of Strep (streptavidin)–HA (hemagglutinin)–PPFIA1. The interaction of Kif7 with PPP2R1A was only detected by Western blot when PPFIA1 was overexpressed. Western blots are representative of three biological replicates.

PPFIA1 is required for efficient trafficking of Kif7, Gli2, and Gli3 to the tips of cilia and for Hh signaling

PPFIA1 plays important roles in transporting cargoes in neurons (40). In this context, PPFIA1 interacts with and recruits cargo vesicles to the neuron-specific kinesin motor protein Kif1A (41). Drosophila and Caenorhabditis elegans PPFIA1 homologs stimulate kinesins by promoting their activity and/or specific subcellular localization (4244). This led us to hypothesize that PPFIA1 could promote Kif7 trafficking in cilia and consequently impinge on the localization of Gli proteins and on Hh signaling. Consistent with this notion and confirming previous large-scale proteomic results (45, 46), we found that PPFIA1 localized to the basal body and to the cilia axoneme (Fig. 3A and fig. S3B). PPFIA1 basal body/cilia localization did not change in response to Smo modulation, suggesting that its localization is not Hh-responsive (fig. S3C).

Fig. 3 PPFIA1 is required for localization of Kif7 and Gli proteins to the tips of cilia and for Hh signaling.

(A) PPFIA1 localizes to the primary cilium. Indirect immunofluorescence images of MEFs stained for PPFIA1 (green), acetylated tubulin (red), and pericentrin (blue). (B) Immunofluorescence images of MEFs (top and middle) and Ptch1–/– MEFs (bottom) expressing Kif7-GFP (green) stained for acetylated tubulin (red) and pericentrin (blue). (C) shRNA-mediated knockdown of Ppfia1 or Smo impaired the ciliary tip localization of Kif7 during activation of the Hh pathway in MEFs (left) and in Ptch1–/– MEFs (right). (D) Immunofluorescence images of fixed MEFs and Ptch1–/– MEFs stained for Gli2 and Gli3 (green) and acetylated tubulin (red). (E) Knockdown of Kif7 or Ppfia1 with shRNAs impaired the SAG-promoted and constitutive localization of Gli2 and Gli3 at the tips of cilia in MEFs (left) and in Ptch1–/– MEFs (right), respectively. (F) CRISPR-Cas9–mediated knockout (KO) of Kif7 (left) or Ppfia1 (right) in C3H10T1/2 cells impaired Gli1 mRNA induction with SAG. Gli1 mRNA expression was calculated relative to that in SAG-treated wild-type cells. (G) Representative images of whole embryos (top) or magnified views of dotted area (bottom) injected with control, Kif7, or Ppfia1 morpholinos at the one-cell stage. Images were taken at 30 hours after fertilization. In control-injected embryos, somites have a stereotypic chevron shape, whereas they are abnormal in Kif7 or Ppfia1 morphant embryos. Quantification of somite angles (bottom panel). n = 20 embryos per condition; two batches of embryos were analyzed. Quantification of ciliary tip localization, qPCR data, and somite angles are presented as means ± SEM across three biological replicates. *P < 0.05. Scale bars, 1 μm. Western blots are representative of three biological replicates.

We found that 42% of MEFs stably expressing Kif7-GFP and stimulated with SAG exhibited Kif7 localization at the tips of cilia (Fig. 3, B and C, left panel). We next determined that, of the four liprin-α genes (Ppfia1 to Ppfia4), MEFs express predominantly Ppfia1 with trace amounts of the other three Ppfia genes (fig. S3A). Depletion of Ppfia1 using two independent short hairpin RNAs (shRNAs) (fig. S3E) significantly reduced the percentage of cells with Kif7 localization at the tips of cilia upon stimulation of the pathway with SAG (Fig. 3C). Knockdown of Smo resulted in a similar reduction in ciliary tip localization of Kif7 (Fig. 3C). To rule out off-target effects, we stably expressed human PPFIA1 (hPPFIA1) after infection of cells with the shRNA that targets mouse Ppfia1, which rescued Kif7 trafficking to the tips of cilia (fig. S4, A and C). We also introduced Kif7-GFP into MEFs derived from Ptch1–/– knockout mice (Ptch1–/– MEFs), which exhibit constitutive Hh pathway activity (47), and found that 57% of cells had constitutive Kif7 localization at the tips of cilia (Fig. 3, B and C, right panel). Depletion of Ppfia1 or Smo with shRNAs or inhibition of pathway activity with the Smo antagonist cyclopamine similarly decreased the proportion of Ptch1–/– MEFs with Kif7 at the tips of cilia (Fig. 3C, right panel). These results suggest that PPFIA1 promotes the localization of Kif7 to the tips of cilia during Hh signaling.

Our results demonstrated that Kif7 was required for robust ciliary trafficking of Gli proteins and target gene induction during activation of the Hh pathway (Fig. 1). Because depletion of Ppfia1 inhibited Kif7 trafficking to the tips of cilia during pathway activation, we predicted that this would also affect Gli2 and Gli3 localization and, consequently, Hh target gene activation. Knockdown of Kif7 or Ppfia1 in MEFs and Ptch1–/– MEFs repressed the SAG-induced and the constitutive localization of Gli2 and Gli3 to the tips of cilia in these cells, respectively (Fig. 3, D and E). Depletion of Ppfia1 inhibited the SAG-induced Gli1 mRNA expression in MEFs and its constitutive activation in Ptch1–/– MEFs to similar amounts obtained with Kif7 knockdown (fig. S3D), and this was rescued upon overexpression of hPPFIA1 (fig. S4, B and C), thereby supporting a requirement for PPFIA1 during Hh pathway activation. Furthermore, C3H10T1/2 cell lines with homozygous disruption of Ppfia1 or Kif7 using the CRISPR (clustered regularly interspaced short palindromic repeats)–Cas9 (48) gene editing system showed an attenuated response to SAG that was partially rescued upon expression of complementary DNAs (cDNAs) encoding PPFIA1 or Kif7 (Fig. 3F). Ppfia1–/– cells exhibited a reduction in Gli3R formation similar to that seen in Kif7–/– cells (2931) (fig. S3F).

Last, to further examine the role of PPFIA1 in Hh signal transduction, we asked if morpholino-mediated knockdown of Ppfia1 in zebrafish would phenocopy the somitogenesis defects caused by interference with Kif7 function or with other Hh pathway components (27, 49). In zebrafish, the amount of Hh signaling influences the fates of skeletal muscle cells. High amounts of Hh signaling are required for specification of muscle pioneer cells, slow muscle cells require intermediate amounts of Hh activity, and fast muscle cells form in the absence of Hh activity (49). Improper regulation of Hh signaling therefore leads to defective somitogenesis with individual somite losing their characteristic chevron shape. Injection of morpholinos targeting Kif7 or Ppfia1 into embryos led to similar phenotypes (Fig. 3G), with morphant embryos having significantly wider somite angles (Kif7, 96.6°; Ppfia1, 104.6°) than control morpholino-injected embryos (83.7°) (quantification in bottom panel). Our data suggest that loss of Ppfia1 phenocopies loss of Kif7 in vivo. We conclude that PPFIA1 positively affects Hh signaling by promoting the localization of Kif7 and Gli proteins in the tips of cilia.

The PP2A inhibitor okadaic acid inhibits Kif7 and Gli trafficking in cilia and blocks Hh signaling

Endogenous PPFIA1 coimmunoprecipitated with the catalytic subunit of PP2A (PP2Ac) from MEF lysates (Fig. 4A), thus supporting our MS data. We used okadaic acid, an inhibitor of the PP1 and PP2A family of serine/threonine phosphatases, to test for the potential involvement of PP2A in ciliary trafficking of Kif7, Gli2, and Gli3 as well as for Hh pathway activity. Okadaic acid treatment inhibited the SAG-induced or constitutive accumulation of Kif7 (Fig. 4B) and Gli2 and Gli3 (Fig. 4C) at the tips of cilia in MEFs and Ptch1–/– MEFs, respectively. As a control, we showed that the ciliary accumulation of Smo was not affected by okadaic acid (fig. S5, A and B). Consistent with a requirement of Kif7 and Gli proteins trafficking in cilia for pathway activation, okadaic acid treatment inhibited the induction of Gli1 mRNA in response to SAG treatment (Fig. 4D). Consistent with the role of Kif7 in GliR formation, okadaic acid treatment slightly decreased Gli3R abundance without altering the amount of the full-length forms of Gli2 or Gli3 (fig. S5C). These results suggest the possibility that PP2A activity is required during Hh signaling to promote the ciliary tip trafficking of Kif7 and Gli proteins.

Fig. 4 The protein phosphatase PP2A promotes Hh signaling through control of Kif7 and Gli protein localization at the tips of cilia.

(A) Coimmunoprecipitation of PPFIA1 and the catalytic subunit of PP2A (PP2Ac) from MEF lysate. (B and C) The PP2A inhibitor okadaic acid inhibits the localization of Kif7 (B) or Gli2 and Gli3 (C) to the tips of cilia in SAG-treated MEFs or in Ptch1–/– MEFs. (D) Okadaic acid inhibits the activation of the Hh target gene Gli1 by SAG in MEFs or its constitutive activation in Ptch1–/– MEFs. Gli1 mRNA expression was calculated relative to that in SAG-treated MEFs or untreated Ptch1–/– MEFs. Quantification of cilia tip localization and qPCR data are presented as means ± SEM across three biological replicates. Western blot is representative of three biological replicates. *P < 0.05.

Kif7 dephosphorylation promotes its localization at the tip of primary cilium and Hh signaling

The above results identify Kif7 as a putative target of PP2A and suggest that posttranslational control of Kif7 phosphorylation may be important for its localization at the tips of cilia and for Hh signaling. Stimulation of C3H10T1/2 cells with SAG resulted in a decrease in Kif7 phosphorylation, as determined by faster migration of the Kif7 immunoreactive band on Phos-tag–containing gels (Fig. 5A, left). In contrast, treatment of Ptch1–/– MEFs with the Smo antagonist cyclopamine led to an upward shift of Kif7, suggesting that Kif7 is phosphorylated when the pathway is inhibited (Fig. 5A, right). MS of FLAG-Kif7 immunopurified from human embryonic kidney (HEK) 293T cells identified three phosphorylation sites that corresponded to mouse Ser897, Ser969, and Ser1337. We then designed a targeted high-resolution multiple reaction monitoring (MRMHR) MS method to quantify the abundance of each of these phosphopeptides in FLAG-Kif7 immunoprecipitates. The abundance of the C-terminal 1330–1339 phosphopeptide that contains phosphorylated Ser1332 (which corresponds to phosphorylated Ser1337 in mouse Kif7) was the only phosphopeptide that showed varying abundance according to okadaic acid treatment, suggesting that phosphorylation of this residue might be dependent on PP2A (Fig. 5B, fig. S6A, and table S2). The abundance of the nonphosphorylated version of these peptides and of three other nonphosphorylated control peptides used as internal standards did not substantially vary (Fig. 5B, fig. S6A, and table S2). We then expressed Kif7 wild type and Kif7-S1337A in MEFs and determined that the electrophoretic mobility of the Kif7-S1337A mutant was faster than Kif7 wild type when using Phos-tag–containing gels, suggesting that this site was indeed targeted by phosphorylation (Fig. 5C).

Fig. 5 Kif7 phosphorylation dictates its ciliary tip localization and Hh signaling.

(A) Western blots of lysates from C3H10T1/2 cells treated or not with SAG (left) or from Ptch1–/– MEFs treated or not with cyclopamine (right), using Phos-tag gels and anti-Kif7 antibodies. (B) The abundance of three Kif7 phosphorylation sites was quantified by MRMHR in cells treated or not with okadaic acid. Both the nonphosphorylated (gray) and phosphorylated peptides (blue) were quantified. Numbering refers to the amino acid positions in human Kif7. The dashed box indicates the only peptide (containing Ser1332 in human and Ser1337 in mouse Kif7) whose abundance was increased by okadaic acid treatment. “*” denotes phosphoserine. The amounts indicated are from a representative MRM experiment; for the results of the second experiment, see table S2. (C) Western blots of lysates from MEFs overexpressing Kif7-WT (wild type) or Kif7-S1337A mutant using Phos-tag gel and anti-Kif7 antibodies. (D) Kif7-S1337A mutant exhibits constitutive localization at the tips of cilia in MEFs (first and second panels). Expression of Kif7-S1337A also leads to ligand-independent localization of Gli2 and Gli3 to the tips of cilia (third panel). Kif7-S1337A expression is sufficient to induce Gli1 mRNA expression, which was calculated relative to that in cells expressing WT Kif7 (fourth panel). (E) The Kif7-S1337D mutant displays impaired ciliary tip localization in response to SAG treatment compared to Kif7 WT (left). Expression of this mutant inhibits the ciliary tip localization of Gli proteins (middle) and the SAG-mediated activation of Gli1 mRNA expression, which was calculated relative to SAG-treated cells expressing WT Kif7 (right). (F) Okadaic acid treatment does not inhibit the induction of Gli1 mRNA expression in cells expressing the Kif7-S1337A mutant (left) or its constitutive ciliary tip localization (right), suggesting that PP2A could be modifying this residue. Gli1 mRNA expression was calculated relative to untreated cells. Quantification of ciliary tip localization and qPCR data are presented as means ± SEM across three biological replicates. Western blot figure is representative of three biological replicates. *P < 0.05. Scale bar, 1 μm.

To investigate the functional importance of phosphorylation of Ser1337 for Kif7 function, we generated MEFs stably expressing the Kif7-S1337A-GFP or Kif7-S1337D-GFP point mutants that mimic the unphosphorylated and phosphorylated states of this residue, respectively (fig. S6J). A large proportion of cells (~44%) exhibited constitutive, ligand-independent localization of Kif7-S1337A to the tips of cilia (Fig. 5D). Conversely, Kif7-S1337D did not localize at the tips of cilia in a ligand-independent fashion, but its localization to cilia tips was impaired in SAG-treated MEFs when compared to wild-type Kif7 (Fig. 5E). Because we showed that Gli proteins were cargoes of Kif7 in cilia (Fig. 1), we reasoned that their ciliary tip localization should also be affected in cells expressing the Kif7 mutants. Expression of Kif7-S1337A was indeed sufficient to induce the accumulation of Gli2 and Gli3 at the tips of cilia (Fig. 5D, middle graph), whereas expression of Kif7-S1337D reduced the localization of Gli2 and Gli3 at the tips of cilia in SAG-stimulated MEFs (Fig. 5E, second panel).

Because the Kif7-S1337A mutant is constitutively at the tips of cilia, a localization that correlates with the state of activation of the Hh pathway, we next tested whether its expression is sufficient to activate Hh signaling. In MEFs, stable expression of Kif7-S1337A, but not of wild-type Kif7, led to a robust 30-fold induction of Gli1 mRNA abundance (Fig. 5D, third panel). Other Hh target genes, such as Ptch1 and Hhip, were also induced upon Kif7-S1337A expression (fig. S6D). Moreover, the constitutive ciliary localization of Kif7-S1337A and its ability to activate Hh signaling were Smo-independent because the Smo antagonist cyclopamine had no effect (fig. S6, B and C). However, the Gli inhibitor GANT-61 (Gli antagonist 61) (50) partially inhibited the Kif7-S1337A–dependent Gli1 mRNA induction (fig. S6C). Reciprocally, expression of Kif7-S1337D decreased SAG-induced localization of Gli proteins at the tips of cilia and Gli1 mRNA induction (Fig. 5E, bottom panel). A form of Kif7 with similar mutations at another phosphosite Ser969 did not show altered ciliary tip localization, and expression of this mutant did not affect the induction of Gli1 (fig. S6, E and F). Furthermore, expression of Kif7-S1337A in either Kif7–/– or Ppfia1–/– cells was also sufficient to increase Gli1 mRNA (fig. S6, G and H).

Our MS analysis thus identified Ser1337 as an okadaic acid–sensitive Kif7 phosphorylation site. Because Hh signaling is inhibited by okadaic acid and Kif7-S1337D expression and activated by Kif7-S1337A expression, we surmised that PP2A inhibition could lead to phosphorylation of Ser1337 and that PP2A-mediated dephosphorylation of Kif7 at this residue could be required for Kif7 localization to the tips of cilia and Hh target gene activation. Our results showed that Kif7-S1337A–mediated activation of Gli1 transcription or its ligand-independent localization at the tips of cilia was unaffected by okadaic acid treatment (Fig. 5F). Thus, Kif7-Ser1337 dephosphorylation appears to be a key step in Kif7 trafficking to the tips of cilia, and our results suggest that PP2A-mediated dephosphorylation of this residue is required for Gli-mediated transcriptional output.

Kif7-PPFIA1 interaction is increased in Kif7 ciliopathy mutant

Mutations in Kif7 have been identified in various human ciliopathies including acrocallosal, Bardet-Biedl, Meckel, and hydrolethalus syndromes (27, 51). Some of these mutations occur in the coil-coiled domain of Kif7 that we mapped as the PPFIA1 binding site. We derived stable cell lines expressing four randomly selected coiled-coil mutant Kif7 proteins and performed MS on Kif7 immunoprecipitates to determine whether the mutations affected the association of Kif7-interacting proteins (Fig. 6A). The immunoprecipitates for Kif7-L759P (mouse Kif7-L764P), a mutation found in Bardet-Biedl syndrome, contained an increased number of peptides corresponding to PPFIA1 and PP2A subunits, suggesting increased association of this mutant Kif7 for the PPFIA1-PP2A complex (Fig. 6B). This result was confirmed by increased precipitation of endogenous PPFIA1 with Kif7-L759P when compared to wild-type Kif7 (Fig. 6C). We hypothesized that enhanced binding of Kif7-L759P to the PPFIA1-PP2A complex would promote its dephosphorylation and localization at the ciliary tip. MEFs (32%) expressing Kif7-L759P-GFP exhibited ligand-independent localization of Kif7-L759P at the tips of cilia (Fig. 6D). Expression of Kif7-L759P was sufficient to induce Gli1 expression sevenfold compared to wild type, although it was expressed at lower amounts. Disease-causing Kif7 mutations result in decreased Gli3R production (27), and we found that cells expressing the L759P mutant also exhibited decreased Gli3R abundance (fig. S6I). Together, these results suggest that some of the Kif7 human mutations may directly or indirectly affect ciliary tip localization and activity by altering its phosphorylation state.

Fig. 6 Disease-causing Kif7 mutations lead to increased interaction of Kif7 with the PPFIA1-PP2A complex.

(A) Four disease-causing Kif7 mutants found in human ciliopathies. (B) The protein-protein interaction profiles of the Kif7 mutants were compared to Kif7-WT using FLAG IP-MS. More peptides attributed to PPFIA1 and subunits of PP2A were identified in Kif7-L759P immunoprecipitates than in Kif7-WT immunoprecipitates. Mean sum of total peptide ± SEM across three independent experiments are listed in the table. (C) Lysates of cells stably expressing FLAG Kif7-WT or FLAG Kif7-L759P were immunoprecipitated with FLAG antibodies and probed with anti-PPFIA1 antibodies using Western blotting. (D) The Kif7-L759P mutant stably expressed in MEFs localizes to the tips of primary cilia in a ligand-independent manner. (E) Kif7-L759P expression leads to intermediate amounts of ligand-independent Gli1 activation compared to the phosphorylation-deficient Kif7-S1337A mutant. Gli1 mRNA expression was calculated relative to that in cells expressing wild-type Kif7. Quantification of ciliary tip localization and qPCR data are presented as means ± SEM across three independent experiments. Western blot figures are representative of three biological replicates. *P < 0.05. (F) Working model. When the Hh pathway is inactive, Kif7 is phosphorylated and ciliary trafficking of Gli proteins is low. At steady state, Kif7 and Gli proteins are enriched at the base of cilia, but a basal flux of Kif7-Gli trafficking in cilia occurs to allow processing of Gli3 into Gli3R (left). During pathway activation, Smo accumulates in cilia and promotes the recruitment of the PPF1A-PP2A complex to Kif7. This leads to increased dephosphorylation of Kif7 on Ser1337 and increased localization of Kif7 and Gli proteins at the cilia tips, leading to SuFu dissociation from Gli2 and Gli3 proteins, resulting in Gli2 activation and reduction of Gli3R (right). IFT, intraflagellar transport.


How the signaling information carried by Hh ligands is transduced from active Smo to regulation of Gli proteins is poorly understood. Trafficking of Gli proteins from the base to the tip of the primary cilium, a process thought to involve Kif7, leads to dissociation of Gli proteins from the negative regulator SuFu to allow for their nuclear translocation and activation of target genes. Kif7 is important for the localization of Gli proteins at the tips of cilia during pathway activation (30, 35), although it has been unclear whether Kif7 is directly required for Hh-ligand activation of Gli signaling. Here, we found that in Kif7–/– MEFs or Kif7–/–C3H10T1/2 cells, the ability of SAG to promote the accumulation of Gli proteins at the tips of cilia and to induce expression of the Hh target gene Gli1 is compromised. In the absence of ligand, Kif7 promotes GliR formation and redundantly functions with SuFu to repress Gli activity. The dual negative and positive roles of Kif7 thus reflect its function in trafficking Gli proteins in cilia under resting conditions and during Hh-dependent activation to underlie GliR and GliA formation, respectively (Fig. 6F). We note that Kif7 deficiency in Kif7–/– MEFs or Kif7–/–C3H10T1/2 cells does not result in increased ligand-independent pathway activation, suggesting that Kif7 principally fulfills a positive role in these contexts and that SuFu amounts are likely sufficient to inhibit Gli. The use of these cell systems enabled us to specifically study the mechanisms by which Kif7 positively promotes Hh signaling. Our results differ from similar experiments that suggest that the induction of Gli1 expression by Hh is relatively normal in MEFs derived from Kif7–/– mice (30). The reason for this discrepancy is unclear but could be due to heterogeneity in Hh responsiveness of different MEF cultures (wild-type MEFs compared to Kif7–/– MEFs) or differences in Shh (Sonic Hedgehog) compared to SAG treatment. Nevertheless, our results that Gli protein trafficking to tips of cilia and robust SAG-induced Gli1 activation were restored when Kif7 was reintroduced into Kif7–/– MEFs, combined with the blunting of the SAG-mediated response when we knocked out Kif7 in C3H10T1/2 cells, indicate that Kif7 is required for efficient Hh signal transduction in these contexts. However, because the ligand-promoted response is not completely eliminated, possible redundant mechanisms supporting Gli activation may exist. Kif7 has been implicated in the organization of the cilium tip (35). Here, loss of Kif7 results in redistribution of Gli2 from the cilium tip to the axoneme. Although we observed a strong reduction of Gli2 and Gli3 localization at the tips of cilia in Kif7 mutant cells, we did not observe a localization of these proteins to the axoneme. A difference in antibody sensitivity is a possible explanation for these differences.

Using MS, we identified PPFIA1 and PP2A as Kif7-interacting proteins. Using loss-of-function experiments, we showed that PPFIA1 is required for Kif7 function. Indeed, in addition to inhibiting localization of Kif7 at the cilium tip during pathway activation, knockdown of Ppfia1 phenocopied several defects seen upon loss of Kif7 function, including reduced accumulation of Gli proteins at the tips of cilia, lower induction of the Gli1 target gene, and defective somitogenesis in zebrafish embryos.

The protein phosphatase PP2A has previously been linked to Hh signaling in Drosophila, where it appears to have multiple roles in dictating signaling output. For instance, mts, which encodes the PP2A catalytic subunit in flies, has been identified as a gene required for Hh signaling in a small interfering RNA screen (52). Subsequent studies have shown that mts interacts with and dephosphorylates Smo (53) and Ci (Cubitus interruptus) (54), leading to inhibition and activation of Hh signaling, respectively. PP2A also appears to be required for mammalian Hh signaling because it is required for Hh-dependent activation of COUP-TFII (chicken ovalbumin upstream promoter–transcription factor 2) (55). Here, treatment of MEFs with the PP1 and PP2A inhibitor okadaic acid inhibited the trafficking of Kif7 and Gli proteins to the tips of primary cilia and the activation of Hh-dependent target genes without affecting Hh-dependent localization of Smo in cilia (fig. S5). These results suggest that an okadaic acid–sensitive phosphatase acts downstream of Smo but upstream of Gli to inhibit vertebrate Hh signaling. Combined with the identification of the Kif7-PPFIA1-PP2A interaction, this led us to test Kif7 as a possible substrate of PP2A. PPFIA1 interacts with the neuron-specific motor protein Kif1A (41, 44) and stimulates the trafficking of synaptic vesicles (42, 43). Our discovery that phosphorylation dictates localization of Kif7 raises the possibility that this may represent a more general mechanism that controls kinesin function in diverse contexts. Similar to Kif7, the C terminus of Kif1A contains a serine residue flanked by several arginine amino acids, suggesting that a similar regulation may occur. However, unlike conventional kinesins, which directionally move along microtubule tracks, the motor domain of Kif7 inhibits microtubule growth by influencing microtubule plus-end dynamics (35). Whether Kif7 phosphorylation influences its tubulin dynamics activity remains to be evaluated.

Although several aspects are conserved between Drosophila and vertebrate Hh signaling, differences exist. The evolutionary adaptation of the primary cilium as a hub directing Gli protein processing and Hh signaling in vertebrate is likely a reason underlying these differences. In flies, the Kif7 homolog Cos2 also mediates both positive and negative functions during Hh signaling (56, 57). Upon pathway activation, Cos2 is rapidly phosphorylated by Fused, a phosphorylation event that triggers the redistribution of Cos2 from the cytoplasm to the plasma membrane, as well as the magnitude of pathway activation through Ci nuclear translocation and transcriptional activity (58). Analogous to Drosophila Hh signaling, we report here that Kif7 phosphorylation also directs its subcellular localization and the transcriptional output of the pathway. Unlike flies, however, vertebrate Kif7 is phosphorylated under basal conditions and is dephosphorylated during Hh signaling; thus, phosphorylation is a posttranslational modification that controls Kif7 accumulation at the tips of primary cilia. Because Fused is dispensable for mammalian Hh signaling, the kinase that is responsible for Kif7 phosphorylation remains to be determined.

Genetic studies have identified several Kif7 mutations in patients with hydrolethalus, Joubert, Bardet-Biedl, and acrocallosal syndromes. Defects in Gli target gene expression and impaired Gli3 processing confirm that Hh signaling is defective in these disorders. Kif7 mutations also induce Golgi fragmentation, centrosome duplications, and ciliogenesis defects, suggesting that Kif7 may also have additional Hh-independent roles contributing to disease (28). How these disease-causing mutations perturb Kif7 functions is unknown. Results from our study suggest that in the case of at least one mutation, Kif7-L759P, found in Bardet-Biedl syndrome, the association with the PPFIA1-PP2A complex is promoted, and this leads to increased localization of Kif7 at the cilium tip and ligand-independent activation of Gli1 transcription. In light of these findings, whether other disease-causing mutations affect Hh pathway activity by influencing protein-protein interactions, Kif7 phosphorylation, and/or ciliary localization needs to be examined.


Antibodies and reagents

Antibodies were purchased from the following vendors: mouse anti-HA.11 clone 16B12 (Covance); mouse anti-FLAG, mouse anti–β-tubulin clone TUB 2.1, and mouse anti–acetylated tubulin clone 6-11B-1 (Sigma-Aldrich); rabbit anti-Gli3 (sc-20688, for immunofluorescence), rabbit anti-GFP (sc-8334), goat anti-PPFIA1 (sc-54039), and goat anti-Vangl1 used as IgG control (sc-46557, all from Santa Cruz Biotechnology); goat anti-Gli3 (EMD Millipore, AF3690, for Western blot); rabbit anti-pericentrin (Abcam, ab4448); rabbit anti-Smo (LSBio, LS-A2668); goat anti-Gli2 (R&D, AF3635, for Western blot); rabbit anti-Kif7 (31); rabbit anti-Gli2 (for immunofluorescence) (33); rat anti-PPP2CA (Abcam, ab77830); and rabbit anti-PPP2R1A (Cell Signaling, no. 2041). Secondary antibodies conjugated to Alexa Fluor 488, 594, and 647 were purchased from Life Technologies. Secondary antibodies conjugated to horseradish peroxidase were purchased from Jackson ImmunoResearch Laboratories Inc. Okadaic acid sodium salt was purchased from BioShop Canada Inc. (209266-80-8) and was used at 50 nM for 3 hours for all experiments, except for MRM experiments where cells were treated at 150 nM for 2.5 hours. GANT-61 (Tocris) was used at 5 μM for 6 hours, cyclopamine was used at 1 μM for 24 hours, and SAG (Santa Cruz Biotechnology, sc-212905) was used at 200 nM for 24 hours for all experiments except for that in Fig. 2C in which SAG was added for 1.5 hours. Phos-tag Acrylamide Aqueous Solution (Wako, 304-93526) was used according to the manufacturer’s instructions.

Tissue culture, transfections, and lentiviral transductions

HEK293T cells, MEFs, and C3H10T1/2 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich) and penicillin/streptomycin. HEK293T cells were transfected using polyethylenimine (Polysciences Inc., cat no. 23966). All lentiviral particles were produced in HEK293T cells by cotransfection of vesicular stomatitis virus glycoprotein (3 μg), psPAX2 (8 μg), and lentiviral plasmids (8 μg) in 30 to 40% confluent monolayer cell culture grown in 10-cm plates. Medium containing virus particles was collected after 24 and 48 hours. MEFs were subsequently transduced in the presence of polybrene (10 μg/ml) (Sigma-Aldrich). Twenty-four hours after infection, the viral medium was replaced with fresh DMEM with 10% FBS, and where appropriate, cells were selected with puromycin (2 μg/ml) 48 hours after viral infection. For experiments with MEFs or C3H10T1/2 cells, cells were grown to confluency and then serum-starved with 0.5% FBS in DMEM to induce ciliogenesis for 24 to 48 hours. Primary Kif7–/– MEFs were immortalized through overexpression of SV40 large T antigen.

Short hairpin RNAs

MISSION shRNA clones in the lentiviral plasmid pLKO.1-puro were acquired from Sigma-Aldrich. Kif7 shRNA clones TRCN0000090438 and TRCN0000090439 were labeled as Kif7 shRNA no. 1 and no. 2, respectively. PPFIA1 shRNA clones TRCN0000251544 and TRCN0000251546 were labeled as PPFIA1 no. 1 and no. 2. Smo shRNA clone TRCN0000026312 labeled as Smo was also from Sigma-Aldrich. Sequences of shRNAs used are listed in table S3.


pEGPF-mKif7, pEGFP-mKif7delN (647–1348), and pEGFP-mKif7delC (1–724) have been previously described (31). Kif7-GFP was cloned into the pLenti-puro plasmid. Human Kif7 delN (674–1349) was PCR-amplified from a brain cDNA library and was cloned into the pIRES-puro-FLAG plasmid. The Lenti-mKif7-GFP and Lenti-FLAG-mKif7 mutants were generated by QuikChange PCR mutagenesis. Ciliopathy mutations are named according to human Kif7 mutations; for the corresponding mouse Kif7 mutations, see table S3. pCDNA5.1-FLAG-hPPFIA1 and FLAG-PPP2R1A were obtained from the Gingras Laboratory. The respective cDNAs were PCR-amplified from Mammalian Gene Collection clones and Gateway cloned into pDEST 5′ Triple FLAG and pcDNA5/FRT/TO. hPPFIA1 was also subcloned into the pIRES-puro-Strep-HA plasmid. For truncation mutations of Kif7 and PPFIA1, see table S3. All PCR-amplified regions were verified by sequencing. pBABE-neo largeT cDNA was obtained from Addgene and used to immortalize the Kif7–/– MEFs. Primers used for cloning are listed in table S3. Detailed descriptions of the different plasmids and sequences are provided upon request.

Generation of C3H10T1/2 knockout cell lines with CRISPR-Cas9 system

Single-guide RNAs (sgRNAs) targeting mKif7 and mPpfia1 were selected and cloned in px330 according to instructions from (48). Kif7 or PPFIA1 homology arms surrounding the predicted sgRNA cut location were amplified by PCR from mouse genomic DNA and cloned into a donor plasmid designed to introduce a puromycin resistance expression cassette through homologous recombination (HR). C3H10T1/2 cells were transfected with px330, and the HR donor plasmid was transfected with Lipofectamine 2000 (Life Technologies). Cells were selected with puromycin 48 hours after transfection. Single cells were isolated by serial dilution and screened by PCR for homozygous disruption of the targeted alleles. Sequences for the sgRNA and the primers used for cloning are listed in table S3.

Reverse transcription PCR and qPCR

Total RNA was extracted from cells, using TRIzol (Life Technologies) according to the manufacturer’s instructions. For cDNA synthesis, 2 μg of total RNA was reverse-transcribed using random primers and SuperScript II reverse transcription kit (Life Technologies). Real-time PCR was performed in a 7900HT Fast Real-Time PCR system using 2.5 μl of the synthesized cDNA product plus the SYBR Green and primer mix in a final volume of 20 μl (Life Technologies). SYBR gene expression assays (Applied Biosystems) were used to measure the mRNA expression of genes in triplicate. Mean relative gene expression was determined using the ΔΔCt method normalized to cyclophilin mRNA (59). Primers used are listed in table S3. qPCR figures present averages and SEM of three independent experiments.

Affinity purification, immunoprecipitation, and Western blot analysis

Cells were lysed and protein complexes were affinity-purified using streptavidin Sepharose beads (GE Healthcare) or anti–FLAG M2 agarose beads (Sigma-Aldrich) or immunoprecipitated with the indicated antibodies. In brief, cells were solubilized in lysis buffer containing 0.1% NP-40, 20 mM tris-HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 10 mM NaF, 0.25 mM Na3VO3, 100 mM β-glycerophosphate, and protease inhibitor cocktail (Sigma) for 1 hour at 4°C. Lysates were spun down at 20,000g for 10 min at 4°C. Supernatants were collected and affinity-purified with antibodies or with appropriate affinity resins for 24 hours, followed by extensive washing of the beads with lysis buffer. Copurified proteins were eluted from the beads at 95°C for 10 min using 4× Laemmli buffer containing β-mercaptoethanol (Sigma-Aldrich), resolved by SDS–polyacrylamide gel electrophoresis, and transferred onto nitrocellulose or polyvinylidene difluoride membranes (Pall) for Western blot analysis.

Zebrafish morpholino experiments

Zebrafish microinjection and handling were carried out using standard techniques. Translation blocking morpholinos were obtained for both Kif7 (5′-GCCGACTCCTTTTGGAGACATAGCT-3′) and PPFIA1 (5′-GGTGGGCATCACCTCGCACATCATC-3′) from Gene Tools. The Kif7 morpholino has been previously characterized (27). Eight nanograms of either Kif7 or Ppfia1 morpholino was injected per embryo. Embryos were imaged at 30 hours after fertilization, and for each embryo, a single somite in the middle of the yolk extension was chosen for quantification. Somite angles were measured using ImageJ software. Two batches of embryos were analyzed.

Immunofluorescence microscopy and image acquisition

Cells grown on coverslips were fixed with cold 100% methanol for 5 min. Coverslips were covered with antibodies in 1% normal donkey serum in phosphate-buffered saline overnight at 4°C. Slides were mounted on coverslips with Vectashield mounting medium (Vector Laboratories). Laser scanning confocal images were acquired using a Plan-Apochromat 63×/1.4 numerical aperture oil immersion objective on a confocal microscope (LSM 700, Carl Zeiss) operated with ZEN software. EGFP (enhanced GFP)/Alexa Fluor 488, Alexa Fluor 594, and Alexa Fluor 647 fluorophores were excited individually with 488-, 543-, and 638-nm lasers, respectively, with appropriate filter sets. Uncompressed images were processed with Zeiss ZEN software black edition. To quantify cilium tip localization, images were taken as Z-stacks and rendered as three-dimensional projections. Cilia with Kif7 or Gli at the tips were counted (30 cells in 10 to 15 independent images were used for each condition). Cilia tip localization was calculated as the percentage of cilia exhibiting localization of Kif7 or Gli proteins at the tips. Average and SEM were calculated from three independent experiments. Percentage of protein of interest at cilium tip was analyzed with Fisher’s exact test.

FLAG affinity purification protocol for MS

Lysates of HEK293T cells expressing FLAG-hKif7, FLAG-hPPFIA1, and FLAG-hGli3 were processed as previously described using the M2-FLAG magnetic bead affinity purification protocol and on-bead digest (60). Twenty-five percent of the volume of the digested sample was analyzed on a TripleTOF 5600 (AB SCIEX), using a Nanoflex cHiPLC system at 200 nl/min (Eksigent ChromXP C18 3 μm × 75 μm × 15 cm column chip). Buffer A was 0.1% formic acid in water; buffer B was 0.1% formic acid in acetonitrile (ACN). The high-performance liquid chromatography (HPLC) delivered an ACN gradient over 120 min (2 to 35% buffer B over 85 min, 40 to 60% buffer B over 5 min, 60 to 90% buffer B over 5 min, hold buffer B at 90% for 8 min, and return to 2% buffer B for 105 min). The parameters for acquisition were 1 MS scan (250 ms; mass range, 400 to 1250), followed by up to 50 MS/MS scans (50 ms each). Candidate ions with charge states +2 to +5 and above a minimum threshold of 200 counts/s were isolated using a window of 0.7 atomic mass unit (amu). Previous candidate ions were dynamically excluded for 20 s with a 50-mD window.

MS data acquisition and analysis

TripleTOF 5600 .wiff files were converted to .mgf format using ProteinPilot Software before being saved into ProHits (61). Files were analyzed with the iProphet pipeline (62) implemented within ProHits as follows. The database consisted of the human and adenovirus complements of the RefSeq protein database (version 57) supplemented with “common contaminants” from the Max Planck Institute ( and the Global Proteome Machine ( The search database consisted of forward and reverse sequences (labeled “gi|9999” and, more recently, “DECOY”); in total, 72,226 entries were searched. The search engines used were Mascot (2.3.02; Matrix Science) and Comet (2012.01 rev.3), with trypsin specificity (two missed cleavages were allowed), deamidation (NQ), and oxidation as variable modifications. Charges +1, +2, and +3 were allowed, and the precursor mass tolerance was set at 50 ppm, whereas the fragment bin tolerance was set at 0.6 amu. The resulting Comet and Mascot search results were individually processed by PeptideProphet (63) and combined into a final iProphet output using the Trans-Proteomic Pipeline (TPP) (Linux version, v0.0 Development trunk rev 0, Build 201303061711). TPP options were as follows: general options are -p0.05 -x20 -d“gi|9999,” iProphet options are –ipPRIME, and PeptideProphet options are –OpdP. The general TPP options were -p0.05 -x20 -PPM -d“DECOY.” All proteins with a minimal iProphet probability of 0.05 were parsed to the relational module of ProHits. Note that for analysis with SAINT (Significance Analysis of INTeractome), only proteins with iProphet protein probability >0.95 are considered. This corresponds to an estimated false discovery rate of ~0.5%. To identify significant interaction partners from the affinity purification data, the data were subjected to SAINT analysis (64) implemented in ProHits using SAINTexpress proteins with average SAINT score [average probability (AvgP)] greater than 0.98 were considered to be statistically significant. Artifact proteins (such as trypsin and keratin) were manually curated from the final interaction partner list (table S1).

Affinity purification for MRMHR MS

Affinity purification was performed as described above using the on-bead digest protocol, with minor modifications. Briefly, cell pellets were lysed by resuspension (1:4, w/v) in lysis buffer [50 mM Hepes-KOH (pH 8.0), 100 mM KCl, 2 mM EDTA, 0.1% NP-40, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 15 nM Ca, 15 nM okadaic acid, and 1× protease inhibitor cocktail (Sigma, P8340)] followed by two freeze/thaw cycles. Clarified lysate was immunoprecipitated with 25 μl of magnetic anti–FLAG M2 beads (Sigma) for 2 hours at 4°C. One wash with 1 ml of lysis buffer was then performed, followed by an additional wash with 1 ml of 20 mM tris-HCl (pH 8.0), 2 mM CaCl2. For MS analysis, 7.5 μl of 20 mM tris-HCl (pH 8.0) containing 750 ng of trypsin (Sigma; resuspended at 100 ng/μl in tris buffer) was added, and the mixture was incubated at 37°C with agitation for 15 hours. The next morning, the tubes were quickly centrifuged, the beads were magnetized, and the partially digested sample was transferred to a fresh tube before addition of an extra 2.5 μl of 20 mM tris-HCl (pH 8.0) containing 250 ng of trypsin. The samples were incubated for 3 hours at 37°C before addition of 1 μl of 50% formic acid. The samples were stored at –80°C until analysis.

Data acquisition for MRMHR MS

Samples were analyzed on TripleTOF 5600 in two phases: (i) data-dependent acquisition (DDA) and (ii) MRMHR using the same gradient conditions and the same amounts of sample. For DDA, one quarter of the volume of the digested sample was analyzed using a packed tip emitter at 200 nl/min. Buffer A was 0.1% formic acid in water; buffer B was 0.1% formic acid in ACN. The HPLC delivered an ACN gradient over 120 min (2 to 35% buffer B over 85 min, 40 to 60% buffer B over 5 min, 60 to 90% buffer B over 5 min, hold buffer B at 90% at 8 min, and return to 2% buffer B at 105 min). The DDA parameters for acquisition on the TripleTOF 5600 were 1 MS scan (250 ms; mass range, 400 to 1250) followed by up to 20 MS/MS scans (50 ms each). Candidate ions with charge states +2 to +5 and above a minimum threshold of 200 counts/s were isolated using a window of 0.7 amu. Previous candidate ions were dynamically excluded for 20 s with a 50-mD window. Data from the DDA analysis were searched with Mascot (2.3.02, Matrix Science) with trypsin specificity (three missed cleavages were allowed), deamidation (NQ), oxidation, and phosphorylation (STY) as variable modifications. Charges +2, +3, and +4 were allowed, and the parent mass tolerance was set at 50 ppm, whereas the fragment bin tolerance was set at 0.15 dalton, and Kif7 peptides containing sites of phosphorylation, along with several control peptides (table S1), were included in a subsequent targeted MRMHR analysis. MS/MS resulting from the MRMHR acquisition was used to quantify intensities between okadaic acid–treated and control samples.

Statistical analysis

The quantification of cilium tip localization was analyzed by Fisher’s exact test. qPCR and somite angle data were analyzed by Student’s t test (GraphPad). All statistical analyses were considered significant at P < 0.05.


Fig. S1. Rescue of Kif7 expression in Kif7–/– MEFs.

Fig. S2. PPFIA1 and Kif7 interact through their coiled-coil domains.

Fig. S3. PPFIA1 is required for robust Gli1 induction during Hh pathway activation.

Fig. S4. PPFIA1 shRNA rescue experiments.

Fig. S5. Knockdown of Kif7 or PPFIA1 or use of a PP1A/PP2A inhibitor does not interfere with the ciliary localization of Smo during pathway activation.

Fig. S6. Phosphorylation of Kif7 affects its ciliary localization and Hh signaling.

Table S1. High-confidence data set reported in this study.

Table S2. Kif7 peptides quantified by MRMHR.

Table S3. List of primers.


Acknowledgments: We are thankful to the members of C.-c.H.’s laboratory for important discussions. We thank C. Cummins’ laboratory for providing mouse liver and brain cDNAs. We are grateful to Z.-Y. Lin and B. Larsen for help with MS data analysis. S.A. is a Canada Research Chair in Functional Architecture of Signal Transduction, and A.-C.G. is the Canada Research Chair in Functional Proteomics and the Lea Reichmann Chair in Cancer Proteomics. Funding: This work was supported from grants funded by the Canadian Institutes for Health Research to S.A. (grant MOP-84273) and A.-C.G. (grant MOP-84314) and from the Canadian Cancer Society Research Institute to C.-c.H. (grants 700320 and 700774). Y.C.L. was supported by the Alexander Graham Bell Canada Graduate Scholarship from Natural Sciences and Engineering Research Council of Canada. Author contributions: Y.C.L. and S.A. designed and performed most of the experiments. A.L.C., L.D.B.M.-C., and A.-C.G. designed and performed quantitative MS experiments. A.R.D. and I.C.S. designed and performed zebrafish experiments. X.Z., V.P., and C.-c.H. generated plasmids, cell lines, and antibodies and provided important discussions. Y.C.L. and S.A. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The raw files have been deposited in the Mass spectrometry Interactive Virtual Environment (MassIVE) repository housed at the Center for Computational Mass Spectrometry at University California, San Diego ( The main FLAG AP (affinity purification)–MS data set consisting of 17 raw files has been assigned the MassIVE ID MSV000078479 and is available for file transmission protocol download at Password is Kif7-Liu.
View Abstract

Navigate This Article