Roscovitine, a selective CDK inhibitor, was purchased from Biomol (Plymouth Meeting, PA). The proteasome inhibitor MG132, the MEK1/2 inhibitor U0126, and nocodazole were from Peptide Institute (Osaka, Japan), Wako (Osaka, Japan), and Sigma-Aldrich (St. Louis, MO), respectively. Anti-Kif26b rabbit polyclonal antibody was previously described [9]. Antibodies against phosphorylated Kif26b (anti-phospho-Thr1859 and anti-phospho-Ser1962 Kif26b antibodies) were generated by immunization of rabbits with phosphorylated peptide (phospho-Thr1859; CYSKIpTPPRKP (1855–1864) and phospho-Ser1962; CLDTPpSPVRKT (1958–1967)) conjugated to KLH and the resulting sera then affinity-purified. The following commercially available antibodies were also used: rabbit polyclonal anti-c-Myc antibody, rabbit polyclonal anti-Nedd4 antibody and mouse monoclonal anti-ubiquitin antibody (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit polyclonal anti-phospho CDK/MAPK substrate (phospho-Ser) antibody, mouse monoclonal anti-phospho CDK/MAPK substrate (phospho-Thr) antibody, and rabbit polyclonal anti-CDK5 antibody (Cell Signaling, Danvers, MA), rabbit polyclonal anti-FLAG antibody and mouse monoclonal anti-β-tubulin antibody (Sigma-Aldrich), and Alexa Fluor 488 goat anti-rabbit IgG and Alexa Fluor 546 goat anti-mouse IgG (Invitrogen, Carlsbad, CA).

The full-length Kif26b construct and the Kif26b-ΔC and Kif26b-C deletion constructs were previously described [9]. The other Kif26b deletion constructs were generated by digestion by appropriate restriction enzymes or by PCR. The amino acid residue numbers of the deletion constructs are as follows: Motor (1–798) and ΔMotor (799–2112). The full-length GAKIN/KIF13B construct was previously described [12]. The cDNA for full-length mouse Nedd4 was kindly gifted by Dr. Shinji Masui at the Center for iPS Cell Research and Application, Kyoto University. The Nedd4 deletion constructs were generated by PCR with appropriate primers. The amino acid residue numbers of the deletion constructs are as follows: C2 (1–249), ΔC2 (250–886), WW (250–549), and HECT (550–886). Point mutations in Kif26b (T1859A, S1962A, and T1859A/S1962A) and Nedd4 CS (the catalytically inactive C854S mutation) were introduced with a site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA). All DNA fragments were sequenced for verification.

HEK293, COS-7, and HeLa (American Type Culture Collection) cells were maintained in DMEM supplemented with 10% fetal bovine serum. Cells were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. For organ culture, kidneys dissected from embryonic day (E) 14.5 embryos were cultured as previously described [13]. MG132 was used at a final concentration of 20 µM.

Duplex siRNAs for human Nedd4 were purchased from Invitrogen. The siRNA nucleotide sequences were as follows: Nedd4-siRNA#1, sense (5′-UAUUGGUGACAACUAUUUCUGAUCC-3′) and antisense (5′-GGAUCAGAAAUAGUUGUCACCAAUA-3′), Nedd4-siRNA#2, sense (5′-UUCAAUUGCCAUCUGAAGUUUAUCC-3′) and antisense (5′-GGAUAAACUUCAGAUGGCAAUUGAA-3′), and negative control, sense (5′-AAAUGGCGUCGCGUUCCUUUAGAGG-3′) and antisense (5′-CCUCUAAAGGAACGCGACGCCAUUU-3′).

Cells were harvested with lysis buffer (50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 10% glycerol, 5 mM EDTA, 0.5% Triton X-100, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), and phosphatase inhibitor cocktail (Thermo, Waltham, MA). For analysis of ubiquitinated proteins, SDS was added to a concentration of 0.1%. Lysates were centrifuged at 17,500 × g at 4°C, and the supernatants were incubated with the appropriate antibodies and protein G-Sepharose beads (Pierce, Rockford, IL) for 2 h at 4°C. Immunoprecipitates were washed four times with lysis buffer, separated by SDS-PAGE, and transferred to a polyvinylidene difluoride membrane. Immunoblotting analyses were performed as previously described [14] with a minor modification. In brief, membranes were blocked with 10% fat-free dry milk in phosphate-buffered saline or 5% BSA in Tris-buffered saline with 0.05% Tween 20 and incubated with primary antibodies and then with alkaline phosphatase-conjugated anti-mouse IgG or anti-rabbit IgG (Promega, Madison, WI).

E14.5 mouse embryos were fixed in 10% formalin and processed for paraffin-embedded sectioning. Immunohistochemistry and in situ hybridization and were performed using an automated Discovery System (Roche, Basel Switzerland) according to the manufacturer’s protocols. A cDNA fragment of Nedd4 (1648–2664 bp) spanning the open reading frame was used as a riboprobe. All experimental procedures and protocols for animal studies were approved by the Committee on Animal Research of Kumamoto University (B23-179).

Immunofluorescence analysis was performed as previously described [15]. For tubulin staining, cells grown on cover slips were fixed with methanol for 3 min on ice and rehydrated by three times 5-min washes with phosphate-buffered saline (PBS). Cells were blocked for 30 min with 2% BSA in PBS, incubated for 1 h with the primary antibody diluted in blocking buffer. After three times 5-min washes with PBS, cells were incubated for 30 min with appropriate secondary antibodies diluted in blocking buffer. Coverslips were then washed with PBS and mounted, and immunofluorescence was observed using a confocal scanning laser microscope (FLUOVIEW FV1000; Olympus, Tokyo, Japan). Images were processed with Adobe Photoshop software.

FLAG-Kif26b or FLAG-GAKIN/KIF13B was expressed in COS-7 cells. Cells were lysed with PHEM buffer (50 mM Pipes, 50 mM HEPES (pH 7.2), 10 mM EGTA, 5 mM MgCl₂, and 1% Triton X-100) containing complete protease inhibitor (Roche) and centrifuged at 100,000 × g for 30 min at 4°C. The lysates were supplemented with 0.2 mg/ml of a taxol-stabilized microtubule solution (purified from porcine brain as previously described [16]) incubated at room temperature for 30 min in the presence of 20 µM taxol and 2 mM AMP-PNP, and centrifuged at 100,000 × g for 15 min at 20°C. The resulting supernatant and pellet were used as S1 and P1, respectively. The P1 fraction was resuspended in PHEM containing 20 µM taxol and 10 mM ATP, incubated at room temperature for 30 min, and centrifuged at 100,000 × g for 15 min at 20°C. The resulting supernatant and pellet from the second centrifugation were used as S2 and P2, respectively.

The in vitro kinase assay was performed as previously described with a minor modification [15]. In brief, a recombinant C-terminal tail fragment of Kif26b (Kif26b-C) expressed in Escherichia coli (BL21) with an N-terminal glutathione S-transferase (GST) tag was purified with Glutathione Sepharose 4B (GE Healthcare, Buckinghamshire, United Kingdom). CDK1/cyclin B, CDK2/cyclin A, CDK5/p25 and CDK6/cyclin D3 were purchased from Millipore. Kinase activity was assayed in 50 mM Tris-HCl (pH 7.5), 5 mM MgCl₂, 2 mM EGTA, 0.5 mM PMSF, and 0.5 mM DTT for 30 min at 30°C in the presence of 0.1 mM ATP.

In vitro ubiquitination assays for Kif26b proteins were performed as previously described [17] with the following minor modifications. The FLAG-Kif26b used as a substrate was ectopically expressed in HEK293 cells and purified by anti-FLAG immunoprecipitation. The precipitates were incubated for 1 h at 37°C with 2 µg of purified GST-Nedd4 protein, 12 µg ubiquitin (Sigma-Aldrich), 100 ng E1 (His-UBE1, Boston Biochem, Cambridge, MA), and 1 µg E2 (His-Ubc7, Boston Biochem). The reactions were washed three times with lysis buffer and then analyzed by SDS-PAGE followed by immunoblotting.

In our previous report, we searched for proteins that interact with Kif26b in the developing kidney and found several, including nonmuscle myosin heavy chain type IIB, that were pulled down by a purified GST-tagged C-terminal tail fragment of Kif26b [9]. We analyzed the precipitated proteins by mass spectrometry and identified an approximately 120 kDa protein as Nedd4, a HECT E3 ubiquitin ligase (Figure S1). To examine the interaction between Kif26b and Nedd4, we performed overexpression and immunoprecipitation analyses. These showed obvious complex formation between Kif26b and wild type (WT) Nedd4 (Figure 1A). Interestingly, the catalytically inactivated mutant of Nedd4 (Nedd4 CS) showed a stronger affinity against Kif26b, suggesting Kif26b as a potential substrate for a HECT domain ubiquitin ligase.

We next investigated which region of Kif26b was involved in the interaction with Nedd4. We expressed Myc-tagged Nedd4 and FLAG-tagged full-length or partial Kif26b constructs (Figure 1B) in HEK293 cells and immunoprecipitated the FLAG-tagged proteins from the cell lysates with anti-FLAG beads. Among the Kif26b deletion mutants, ΔMotor and C interacted with Myc-Nedd4, suggesting that the C-terminal tail region of Kif26b is required for the interaction with Nedd4 (Figure 1C). We next constructed Nedd4 deletion mutants (Figure 1D) and performed pull-down assay to test their abilities to associate with Kif26b. As shown in Figure 1E, the WW domains of Nedd4 mediated its interaction with Kif26b. As the WW domains of Nedd4 have been implicated in substrate recognition [18], these results suggested that Nedd4 might ubiquitinate Kif26b.

As shown above, the WW domains of Nedd4 interacted with the C-terminal tail region of Kif26b. The WW domains have been grouped into four classes according to their ligand preferences [19]. The WW domains of Nedd4 belong to Class-I WW domains, which bind to proteins containing PPxY (PY) motifs or phosphoserine/phosphothreonine residues [20]–[22]. As Kif26b does not contain PY motifs, we examined whether phosphorylation of Kif26b affected its interaction with Nedd4.

We first addressed whether the C-terminal tail region of Kif26b could be phosphorylated. We searched the C-terminal tail region of Kif26b for protein kinase consensus sequences and found that this region harbored two potential CDK phosphorylation sites (Figure 2A). Analyses of combinational peptide libraries have determined the consensus sequence for CDK phosphorylation to be a Ser (S)- or Thr (T)-Pro (P) motif followed by a basic amino acid (Arg (R), Lys (K), or His (H)) ([S/T]Px[R/K/H]) [23].

To examine whether CDKs could phosphorylate Kif26b, we performed an in vitro kinase assay using purified active CDK proteins. We found that CDK2 and CDK5 prominently phosphorylated GST-Kif26b C (Figure 2B). CDK1 showed potent but weaker kinase activity against Kif26b, but the activity of CDK6 was low. These results indicate that Kif26b is phosphorylated by CDKs, mainly CDK2 and CDK5. We next confirmed phosphorylation of Kif26b in cultured cells. FLAG-Kif26b was expressed in HEK293 cells, immunoprecipitated with anti-FLAG beads, and treated with or without alkaline phosphatase (AP). Immunoblotting with antibodies specific to the phosphorylated CDK consensus sequence was positive for Ser- or Thr-phosphorylated FLAG-Kif26b (Figure 2C), and the signal was abolished by treatment with either the CDK inhibitor Roscovitine or AP. Because the antibodies against phosphorylated CDK substrates are also known to recognize the motifs phosphorylated by mitogen-activated protein kinases (MAPKs), we investigated whether MAPKs could phosphorylate Kif26b by transfecting HEK293 cells with FLAG-Kif26b and treating with U0126, an inhibitor of the MAPKs MEK1/2. The FLAG-tagged proteins were immunoprecipitated from the lysates with anti-FLAG beads. Immunoblotting analysis revealed that U0126 treatment did not affect the binding of anti-phospho-CDK substrate antibodies to FLAG-Kif26b (Figure S2A), suggesting that Kif26b is phosphorylated specifically by CDKs.

Thr-1859 and Ser-1962 in the C-terminal tail region of Kif26b met the criteria for possible CDK substrates (Figure 2A). In addition to these sites, there are another seven CDK consensus sequences in the remaining regions of Kif26b. Accordingly, we constructed Kif26b mutants in which these Ser/Thr residues were replaced with Ala and tested whether antibodies against phosphorylated CDK substrates could recognize these mutant proteins. Immunoprecipitation and immunoblotting analyses revealed that mutation of Thr-1859 or Ser-1962 (constructs Kif26b T1859A or Kif26b S1962A, respectively), but not other residues, of Kif26b significantly decreased its reactivity with antibodies against phosphorylated CDK substrates (Figure 2D). Moreover, an in vitro kinase assay demonstrated that purified CDK5 efficiently phosphorylated GST-Kif26b-C, but not GST, and that mutation of Thr-1859, Ser-1962, or both Thr-1859 and Ser-1962 of Kif26b to Ala impaired its phosphorylation by CDK5 (Figure S2B). These results suggest that CDKs directly phosphorylate Kif26b at both Thr-1859 and Ser-1962. To further confirm phosphorylation of Kif26b, antibodies specific for phosphorylated Thr-1859 or Ser-1962 were raised in rabbits immunized with synthetic peptides. The affinity-purified antibodies (anti-phospho-Thr1859 Kif26b and anti-phospho-Ser1962 Kif26b antibodies) recognized WT but not Ser/Thr-mutant FLAG-Kif26b immunoprecipitated with anti-FLAG beads (Figure S2C).

Because the C-terminal tail region of Kif26b was required for the interaction between Kif26b and Nedd4 (Figure 1B and C), we next investigated whether phosphorylation of this region could affect the interaction of Kif26b with Nedd4. HEK293 cells were co-transfected with FLAG-Kif26b and Myc-Nedd4, treated with Roscovitine, and lysed, and FLAG-tagged proteins were immunoprecipitated from the lysates with anti-FLAG beads and subjected to immunoblotting. Roscovitine treatment clearly interfered with the interaction between Kif26b and Nedd4 (Figure 3A). We also tested the ability of Ala-substitution mutants of Kif26b to associate with Nedd4. Overexpression and immunoprecipitation analyses showed that Kif26b with both the T1859A and S1962A substitutions, but not with either alone, had significantly reduced affinity for Nedd4 (Figure 3B). These results suggest that phosphorylation of Kif26b at either Thr-1859 or Ser-1962 enhances the interaction between Kif26b and Nedd4.

As Nedd4 is known to catalyze protein polyubiquitination, we examined whether Nedd4 mediated polyubiquitination of Kif26b. We expressed FLAG-Kif26b, Myc-Nedd4, and Myc-Ubiquitin (Ub) in HEK293 cells. At 48 h post-transfection, cells were treated for 8 h with the proteasome inhibitor MG132 and Kif26b was immunoprecipitated with anti-FLAG beads. We observed a smeared signal in the anti-FLAG bead precipitates of the lysates of cells expressing FLAG-Kif26b and Myc-Ub (Figure 4A). This smear signal was enhanced by further overexpression of Myc-Nedd4 WT but not Myc-Nedd4 CS. The smeared signal presumably represents polyubiquitinated Kif26b, as it persisted in the precipitates from the lysates treated with 2% SDS and boiled for 10 min to disrupt any noncovalent interactions between Kif26b and other proteins (Figure S3A). This method has previously been reported to remove associated proteins from the immunoprecipitated protein of interest [24]. Furthermore, we also performed in vitro ubiquitination assays. FLAG-Kif26b was expressed in HEK293 cells and purified by immunoprecipitation with anti-FLAG beads as an ubiquitination substrate. Then, GST, GST-Nedd4 WT, or GST-Nedd4 CS purified from E. coli and ubiquitin, together with other E1 and E2 proteins, were added to initiate the ubiquitination reaction. The positive smeared signal was observed and was dependent on the catalytic activity of Nedd4 (Figure 4B). We also found that polyubiquitination of Kif26b involved Lys48 but not Lys63 of ubiquitin (Figure S3B). Lys48-linked polyubiquitin chains have been reported to target proteins for proteasomal degradation [25]. Therefore, as shown in Figure 1A with Nedd4 WT and CS, Kif26b appears to be not only a substrate for Nedd4 but also to be targeted by it for proteasomal degradation.

Phosphorylation of Kif26b regulated its interaction with Nedd4 (Figure 3). Therefore, we examined whether phosphorylation also affected polyubiquitination of Kif26b. Overexpression and immunoprecipitation analyses demonstrated that treatment with Roscovitine decreased the amount of polyubiquitinated Kif26b (Figure 4C). Furthermore, the Kif26b T1859A/S1962A mutant showed reduced polyubiquitination when co-expressed with Nedd4 (Figure 4D). These results suggest that Kif26b phosphorylation is important for its polyubiquitination by Nedd4.

In order to verify that endogenous Nedd4 also affected polyubiquitination of Kif26b, we treated HeLa/FLAG-Kif26b cells with siRNAs against Nedd4 and performed immunoprecipitation and immunoblotting analyses. Treatment with two siRNAs against Nedd4 reduced the amount of polyubiquitinated FLAG-Kif26b compared with a control siRNA treatment (Figure 4E), implying that endogenous Nedd4 mediate degradation of Kif26b via the ubiquitin-proteasome pathway.

We next addressed how upstream signaling induced the phosphorylation and polyubiquitination of Kif26b. Kif26b is expressed in the metanephric mesenchyme but is absent in the renal epithelia (described below). This suggests the possibility that epithelialization of the metanephric mesenchyme leads to degradation of Kif26b. Dramatic rearrangements to the microtubule-based cytoskeleton have been reported to occur during the formation of polarized epithelial cells [26], [27]. Therefore, we hypothesized that posttranslational modification of Kif26b could depend on the status of the microtubules.

We first addressed whether Kif26b associated with microtubules. Kif26b, like Kif26a, is classified as a member of the Kinesin-11 family [6]. KIF26A has been reported to associate with microtubules but to lack microtubule-based motor activity [7]. As shown in Figure 5A, amino acid comparison with other KIFs demonstrates that KIF26B and Kif26b harbor neither the “IFAYGQT” nor “DLAGSE” motif, that is highly conserved in KIF motor domains and also differ in several amino acid sequences that function in ATP hydrolysis [6]. To determine whether Kif26b associated with microtubules and possessed ATPase-based motor activity, we performed an ATP-dependent microtubule-dissociation assay using FLAG-Kif26b expressed in COS7 cells. This experiment demonstrated that Kif26b constitutively associated with microtubules, while kinesin-like microtubule-based motor protein GAKIN/KIF13B dissociated from microtubules in response to ATP (Figure 5B). Immunostaining of HeLa/FLAG-Kif26b cells with anti-FLAG antibody showed a mesh-like structure that was obviously co-localized with microtubules (Figure 5C). In addition, nocodazole treatment disrupted the mesh-like distribution of Kif26b, suggesting that Kif26b localization was microtubule-based. Taken together, these results suggest that Kif26b associates with microtubules but lacks microtubule-based motor activity based on ATP hydrolysis.

We next examined whether the phosphorylation of Kif26b was dependent on the microtubule status. When we treated HeLa/FLAG-Kif26b cells with nocodazole, we observed a significant shift of the FLAG-Kif26b signal toward the higher molecular weight area on SDS-PAGE in a time-course dependent manner (Figure 6A), and pretreatment with Roscovitine inhibited the mobility shift of Kif26b in response to nocodazole. We also found the mobility shift of FLAG-Kif26b induced by nocodazole treatment of HEK (human embryonic kidney) 293 cells stably expressing FLAG-Kif26b (HEK293/FLAG-Kif26b cells). Interestingly, the slower-migrating form of Kif26b disappeared upon AP treatment (Figure 6B), and, anti-phospho-Thr1859 Kif26b or anti-phospho-Ser1962 Kif26b antibody strongly recognized the slower-migrating form of Kif26b (Figure 6C). These results suggest that the mobility shift of Kif26 is attributable to phosphorylation by CDKs, which could be triggered by dissociation of Kif26b from microtubules.

We next addressed whether nocodazole treatment affected polyubiquitination of Kif26b. HeLa or HEK293/FLAG-Kif26b cells were pretreated for 6 h with DMSO or nocodazole and then treated for another 8 h with MG132. Immunoprecipitation and immunoblotting analyses showed that nocodazole treatment increased the polyubiquitination of Kif26b (Figure 6D). Furthermore, we observed a smeared and mobility-shifted FLAG-Kif26b band in response to MG132 treatment. This MG132-induced FLAG-Kif26b mobility shift was further enhanced by treatment with nocodazole. Interestingly, the FLAG-Kif26b mobility shift was not observed when we treated HeLa or HEK293/FLAG-Kif26b cells with the lysosomotropic drug chloroquine, which increases endosomal and lysosomal pH and impairs lysosomal function (Figure 6E). These results suggest that Kif26b dissociation from microtubules is followed by its phosphorylation by CDKs and polyubiquitination by Nedd4, resulting in its degradation via the ubiquitin-proteasomal pathway.

As overexpression analyses had allowed us to determine how Kif26b is regulat`ed by Nedd4, we next sought to verify the interaction between endogenous Kif26b and Nedd4 in the developing kidney. Immunoprecipitation from E14.5 kidney lysates showed that Nedd4 could be identified in anti-Kif26b but not control rabbit IgG precipitates (Figure 7A). The interaction was disturbed by treatment with roscovitine, suggesting that the in vivo complex formation between Kif26b and Nedd4 in the developing kidney is regulated by CDKs. Immunohistochemistry analyses of E14.5 kidneys showed that Kif26b was localized in the peripheral regions of the developing kidney, in which the metanephric mesenchyme containing nephron progenitors exist [28], but not in the epithelial components such as renal vesicles and comma-shaped bodies (Figure 7B). Nedd4 was also detected also in the metanephric mesenchyme as well as in the epithelialized structures (Figure 7B, arrow heads). In situ hybridization for Nedd4 showed an expression pattern similar to that shown by immunostaining (Figure S4). These data suggest the importance of Kif26b-Nedd4 interaction in the undifferentiated mesenchymal population.

Finally, we examined whether phosphorylation and polyubiquitination of endogenous Kif26b took place in the developing kidney. Kif26b was immunoprecipitated from E14.5 kidney lysates by anti-Kif26b antibody. Immunoblotting with anti-phosphorylated Kif26b antibodies showed obvious signals in the anti-Kif26 precipitates, which disappeared upon AP treatment or roscovitine treatment (Figure 7C). Furthermore, we also confirmed the interaction between Kif26b and CDK5 in E14.5 kidney by immunoprecipitation (Figure S5A). These results suggest CDKs can phosphorylate Kif26b in the developing kidney. We next examined Kif26b polyubiquitination in E14.5 embryonic kidneys cultured in vitro with or without MG132. We were able to observe the smeared signal by immunoblotting anti-Kif26b precipitates with anti-ubiquitin antibody (Figure 7D). Collectively, these results suggest that phosphorylation of Kif26b by CDKs in the metanephric mesenchyme enhances its polyubiquitination by Nedd4 and targets it for proteasomal degradation.