Dynamic Recruitment of CDK5RAP2 to Centrosomes Requires Its Association with Dynein

CDK5RAP2 is a centrosomal protein known to be involved in the regulation of the γ-tubulin ring complex and thus the organization of microtubule arrays. However, the mechanism by which CDK5RAP2 is itself recruited to centrosomes is poorly understood. We report here that CDK5RAP2 displays highly dynamic attachment to centrosomes in a microtubule-dependent manner. CDK5RAP2 associates with the retrograde transporter dynein-dynactin and contains a sequence motif that binds to dynein light chain 8. Significantly, disruption of cellular dynein-dynactin function reduces the centrosomal level of CDK5RAP2. These results reveal a key role of the dynein-dynactin complex in the dynamic recruitment of CDK5RAP2 to centrosomes.


Introduction
In animal cells, centrosomes are the principal microtubuleorganizing centers that control the temporal and spatial distribution of the microtubule network. To initiate the assembly of microtubule filaments and to anchor a radial array of microtubules, centrosomes require the presence of c-tubulin, a highly conserved protein that exists in a macromolecular structure called the c-tubulin ring complex (cTuRC) [1][2][3]. It is therefore of significant interest to identify the molecules that link the cTuRC to centrosomes. CDK5RAP2 is a centrosomal protein whose mutations lead to autosomal recessive primary microcephaly, a disorder caused by defective proliferation and cell-fate determination of neural progenitors during neurogenesis [4][5][6]. We previously showed that CDK5RAP2 associates with the cTuRC and helps attach it to centrosomes [7,8]. The cTuRC-binding domain in CDK5RAP2 was delineated as a short sequence stretch that is highly conserved in c-tubulin complex-targeting proteins of lower organisms, including Drosophila centrosomin and fission yeast Mto1p and Pcp1p [7]. Furthermore, CDK5RAP2's cTuRCbinding domain stimulates the microtubule-nucleating activity of the cTuRC [8], and a disruption of CDK5RAP2 function results in the disorganization of interphase microtubules and the formation of anastral mitotic spindles [7]. Thus, CDK5RAP2 plays an essential role in the organization of microtubules by centrosomes.
The centrosome's pericentriolar material (PCM) dynamically exchanges its molecular contents with the cytoplasm, and several PCM components, such as the matrix proteins pericentrin and PCM-1, are recruited to centrosomes via microtubule-dependent mechanisms [9][10][11]. These mechanisms require the minus end-directed microtubule motor protein dynein in association with dynactin, a large molecular complex necessary for dynein functions [12,13] [14][15][16][17]. CDK5RAP2, like c-tubulin and many other centrosomal proteins, is present in a large cytoplasmic pool, but how it is targeted to centrosomes has remained unclear. Here we report that CDK5RAP2 is dynamically recruited to centrosomes in a microtubule-dependent manner and that CDK5RAP2 associates with dynein and depends on the dynein-dynactin complex for its localization at centrosomes.

Dynamic attachment of CDK5RAP2 to centrosomes requires microtubules
To examine the dynamics of centrosomal CDK5RAP2 by the fluorescence recovery after photobleaching (FRAP), we constructed a stable cell line expressing GFP-CDK5RAP2 at a level similar to that of the endogenous protein (Fig. 1A). GFP-CDK5RAP2 was enriched at centrosomes in these cells (Fig. 1B) and its signal at these sites was then eliminated by photobleaching. The signal, however, reappeared quickly after bleaching, with 50% recovery occurring within ,1 min and reaching ,60% of the original fluorescence intensity with prolonged incubation (Fig. 1C). This is in accord with previous observations of CDK5RAP2's dynamic localization at centrosomes [18,19]. Strikingly, when FRAP was carried out on these cells after treating them with nocodazole to depolymerize microtubules, the GFP-CDK5RAP2 signal failed to recover at centrosomes during the recording period (Fig. 1B, nocotreated). These results demonstrate that intact microtubules are needed for the dynamic recruitment of CDK5RAP2 to centrosomes.
Protein transport along microtubules involves cytoplasmic dynein associated with dynactin, a large complex comprising at least 11 different subunits that include p150 glued and dynamitin [20]. Transport of most dynein cargos is blocked when the dynactin complex is disrupted by over-expressing either the coilcoil domain 1 (CC1) of p150 glued or the dynamitin protein [21][22][23][24]. To probe the involvement of dynein-dynactin in the centrosomal targeting of CDK5RAP2, we over expressed CC1 in cells and then examined the centrosomal content of CDK5RAP2. Centrosomes were labeled with an anti-centrin antibody. In the presence of excess CC1, the centrosomal level of CDK5RAP2 was reduced by ,80% (Fig. 1D), suggesting that dynein-dynactin mediates the transport of CDK5RAP2 towards centrosomes.

CDK5RAP2 associates with dynein
Previously we showed that CDK5RAP2 associates with microtubules using a microtubule sedimentation assay [7]. When the same sedimentation assay was performed here in the presence of the ATP analog AMP-PNP, which facilitates the binding of molecular motors to microtubules [9], the co-sedimentation of CDK5RAP2 with microtubules was significantly enhanced ( Fig. 2A). This indicated the involvement of motor proteins in mediating the association of CDK5RAP2 with microtubules. Thus, to examine the potential interaction of CDK5RAP2 with dynein, DIC, a core component of dynein, was immunoprecipitated. We found that CDK5RAP2 was present in anti-DIC but not control IgG immunoprecipitates (Fig. 2B). Furthermore, DIC co-immunoprecipitated with a FLAG-tagged CDK5RAP2 construct spanning residues 706-1893 but not with construct 1-706, and any truncation of the 706-1893 construct eliminated its association with DIC (Fig. 2C). Therefore, we conclude that a large carboxy-terminal region of CDK5RAP2 mediates its interaction with dynein.
Next, to ask whether CDK5RAP2 and dynein colocalize in cells, immunofluorescence microscopy and time-lapse microscopy were performed. Endogenous CDK5RAP2 was detected in cytoplasmic particles which were also labeled by the anti-DIC antibody (Fig. 3A). The expression of DIC was demonstrated by immunoblotting (Fig. 2B). Time-lapse microscopy carried out on GFP-CDK5RAP2 stably expressed cells showed that the cytoplasmic particles containing GFP-CDK5RAP2 moved along microtubules towards centrosomes ( Fig. 3B and Movie S1). The particles traveled at ,0.5 mm/s, a speed similar to that of motor dynein movement [25,26]. Collectively, these results point to a dynamic transport of CDK5RAP2 with dynein to centrosomes.

CDK5RAP2 interacts with dynein light chain 8
To ascertain whether CDK5RAP2 directly interacts with dynein, we biochemically isolated proteins that bind to CDK5RAP2. In these screens the CDK5RAP2 fragment 706-1241 specifically captured DLC8, which was identified by mass spectrometry (Fig. 4A). To verify this binding we co-transfected cells with CDK5RAP2 fragments and DLC8 for immunoprecip- itation. In these assays DLC8 associated with CDK5RAP2 706-1241 and the 706-925 sequence stretch within it, but not with other regions of CDK5RAP2 (Fig. 4B). However, since coimmunoprecipitation of two proteins does not necessarily mean direct interaction between them, we expressed CDK5RAP2 706-925 and DLC8 in bacteria and purified the recombinant proteins for ''pull-down'' assays: the 706-925 fragment was incubated with GST-DLC8 or GST and then GST was captured using GSHbeads. CDK5RAP2 706-925 was readily detected in pull-downs of GST-DLC8 but not of GST alone (Fig. 4C), indicating the direct binding of 706-925 to DLC8.
DLC8 recognizes and interacts with the motif K/R-X-T-Q-T (where X is any amino acid) [14], and sequence scanning of CDK5RAP2 706-925 revealed such a motif (Fig. 4D). The sequence in this region of human CDK5RAP2 is conserved in mouse and chicken orthologs (Fig. 4D) but is not found in centrosomin from Drosophila. Within the DLC8-binding motif, the dipeptide Gln-Thr was previously shown to be involved in mediating the interaction with DLC8 [14]. Thus, we created the double mutation Q874A/T875A in CDK5RAP2 and expressed the mutant and wild-type CDK5RAP2 in cells to test their binding to DLC8 by immunoprecipitation. The double mutation eliminated the DLC8-binding activity of CDK5RAP2 (Fig. 4E), indicating that this short sequence found in 706-925 is critical for CDK5RAP2's interaction with DLC8.

Discussion
Microtubules help proteins travel to and from centrosomes. Several centrosomal components, such as the matrix proteins pericentrin and PCM-1, are recruited to centrosomes by microtubule-dependent mechanisms, which require the minus end-directed microtubule motor dynein in conjunction with dynactin [9,21,26,27]. Present on centrosomes throughout the cell cycle, CDK5RAP2 plays an important role in the microtubule-organizing function of centrosomes. We have shown here that CDK5RAP2 associates with dynein and that dynein-dependent transport is necessary for the dynamic attachment of CDK5RAP2 to centrosomes. Therefore, this report provides direct evidence for dynein-mediated recruitment of CDK5RAP2 to centrosomes.
CDK5RAP2 localizes dynamically on centrosomes with a turnover rate similar to its homologue centrosomin (Cnn) in Drosophila but significantly greater than that of c-tubulin [19,28], which suggests that CDK5RAP2 and c-tubulin are transported separately to centrosomes. The assembly of c-tubulin on centrosomes appears to involve both microtubule-dependent and microtubule-independent pathways: on the one hand, the ctubulin level at centrosomes has been found to depend on intact microtubules and a functional dynein-dynactin complex [21,26], but on the other hand studies have pointed to c-tubulin recruitment to centrosomes in the absence of microtubules and presumably by passive diffusion [27,[29][30][31][32][33]. It is possible that these two pathways function independently in the recruitment of ctubulin, with perhaps the cellular conditions determining which of the two predominates. Our FRAP experiments showed that the recovery of CDK5RAP2 signal at centrosomes relies on the presence of an intact microtubule network (Fig. 1B), which suggests that the microtubule-dependent mechanism plays a principal role in efficiently trafficking CDK5RAP2 to centrosomes; our data, however, do not exclude alternative pathways for the centrosomal recruitment of CDK5RAP2.
In this study we have demonstrated CDK5RAP2's association with the cytoplasmic dynein motor protein and identified a DLC8binding motif in CDK5RAP2. DLC8, one of the three dimeric light chains of the dynein complex, is a highly conserved protein with diverse functions [34,35]. It acts as an adaptor between dynein and its cargo proteins [36], but may also have motorindependent functions in activities such as protein folding and dimerization, protein-protein interactions, and nuclear import of proteins [15,35,37,38]. We have found that a large region of CDK5RAP2 encompassing the DLC8-binding domain is required for association with dynein. However, the DLC8-binding domain is insufficient for the interaction of CDK5RAP2 with dynein (Fig. 2C), suggesting that the mechanism of CDK5RAP2's association with dynein is more complex. Dynein associates with dynactin, which regulates dynein functions and also links dynein to various cargos [20,[39][40][41]. We found that the centrosomal content of CDK5RAP2 was significantly reduced following the disruption of the dynein-dynactin complex by over-expression of CC1 of p150 glued (Fig. 1C); introduction of excess CC1 produces effects similar to that seen with the over-expression of the dynactin subunit dynamitin [24,42]. Collectively, our results suggest that CDK5RAP2 associates with the cytoplasmic dynein-dynactin complex for retrograde transport.
In conclusion, CDK5RAP2 plays an important role in the structural organization of the PCM and in the microtubule nucleation that occurs therein [8,[43][44][45]. Our present study showing microtubule-dependent targeting of CDK5RAP2 to centrosomes lends further support to the notion that an intact microtubule network is required for centrosome assembly and functions. The centrosomal levels of CDK5RAP2 may be maintained by its well-controlled recruitment, release, and degradation, and thus the coupling of CDK5RAP2 to dyneindynactin may directly regulate centrosomal activities.

Cell culture and transfection
HEK293T and MDA-MB-231 cells were cultured on poly-Dlysine coated dishes or on 16-mm coverslips with DMEM or RPMI1640 medium, respectively, supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. Plasmid transfection was performed using Lipofectamine TM LTX and Plus Reagent or Lipofectamine 2000 (Invitrogen). Stable MDA-MB-231 cell lines expressing GFP-CDK5RAP2 were selected using 400 mg/ml G418 after transfection; individual G418-resistant cells were selected, cultured and tested for expression of GFP-CDK5RAP2 by immunoblotting.

Microscopic imaging
To perform immunostaining, cells grown on coverslips were fixed with methanol at 220uC or with 4% paraformaldehyde at room temperature, permeabilized with 0.5% Triton X-100 in PBS and then stained with primary antibodies for 2 h at room temperature. After extensive washing with PBS, secondary antibodies coupled with AlexaFluor488 or AlexaFluro594 conjugates (Invitrogen) were used for staining for another 1 h. Nuclear DNA was stained with 1 mM Hoechst 33528 (Sigma-Aldrich). Fluorescent images were acquired with an inverted microscope (Eclipse TE2000, Nikon) or a confocal microscope (LSM510 META, Carl Zeiss Microimaging). To perform FRAP, cells were cultured on 35-mm glass-bottom dishes. Selected regions of cells were bleached at 95% laser power for 20 iterations on the confocal microscope. Two photos were captured before photobleaching and time-lapse photos at 2-4 s intervals were obtained after photobleaching to monitor fluorescence recovery.

Immunoprecipitation
At 24 h after transfection, cell extracts (1,2610 6 cells for binding assays and 1610 7 cells for protein identification by mass spectrometry) were prepared in lysis buffer (25 mM Tris-HCl, pH 7.4, 0.5% NP40, 100 mM NaCl, 5 mM MgCl 2 , 5 mM NaF, 1 mM dithiothreitol, Roche Complete Protease Inhibitor Cocktail) and clarified by centrifugation (16,0006g, 15 min). Immunoprecipitation was performed for at least 2 h at 4uC with antibodies coupled to Protein A/G Agarose or with ANTI-FLAG H M2 Affinity Gel (Sigma-Aldrich). Immunoprecipitates were resolved by SDS-PAGE for immunoblotting or for total protein staining in gels with Sypro Ruby (Bio-Rad Laboratories). Proteins present in the bands excised from gels were identified by tandem mass spectrometry (LTQ Velos linear ion trap LC-MS system, Thermo Fisher Scientific).

Microtubule sedimentation assay
Tubulins were purified from porcine brain by two cycles of temperature-dependent assembly/disassembly followed by phosphocellulose chromatography [46]. Cell lysates of CDK5RAP2 were prepared in PEM buffer (80 mM PIPES, pH 6.8, 1 mM MgCl 2 and 1 mM EGTA) supplemented with 50 mM NaCl, 1% Triton X-100 and the Roche Complete Protease Inhibitor Cocktail; lysates were clarified by spinning them in a refrigerated microcentrifuge at full speed for 15 min. Microtubules were preassembled from purified tubulins (50 mg) in the PEM buffer containing 50 mM taxol (Sigma) and 1 mM GTP and then incubated with the cell lysates in 150 ml PEM buffer containing taxol and GTP for 30 min at room temperature. AMP-PNP was used at 0.5 mM in assays as indicated. Samples were centrifuged through a 20% sucrose cushion at 20,0006g for 30 min at 4uC and the resulting pellets were boiled in SDS-PAGE sample buffer and analyzed by immunoblotting.

Statistical analysis
Statistical analysis was performed with two-tailed Student's t-test.

Supporting Information
Movie S1 GFP-CDK5RAP2 was expressed in MDA-MB-231 cells and observed under a confocal microscope. Live cell images were captured at 12 s-intervals. The arrowheads show the centrosomal-targeting of moving particles; one of these particles first traveled along the microtubules and finally fused with a centrosome. (ZIP)