Glutamylation of centrosomes ensures their function by recruiting microtubule nucleation factors

Abstract

Centrosomes are tubulin-based organelles that undergo glutamylation, a post-translational modification that conjugates glutamic acid residues to tubulins. Although centrosomal glutamylation has been known for several decades, how this modification regulates centrosome structure and function remains unclear. To address this long-standing issue, we developed a method to spatiotemporally reduce centrosomal glutamylation by recruiting an engineered deglutamylase to centrosomes. We found that centrosome structure remains largely unaffected by centrosomal hypoglutamylation. Intriguingly, glutamylation physically recruits, via electrostatic forces, the NEDD1/CEP192/γ-tubulin complex to centrosomes, ensuring microtubule nucleation and proper trafficking of centriolar satellites. The consequent defect in centriolar satellite trafficking leads to reduced levels of the ciliogenesis factor Talpid3, suppressing ciliogenesis. Centrosome glutamylation also promotes proper mitotic spindle formation and mitosis. In summary, our study provides a new approach to spatiotemporally manipulate glutamylation at centrosomes, and offers novel insights into how centrosomes are organized and regulated by glutamylation.

Keywords: Centriolar satellites; Centrosomes; Glutamylation; Microtubules; Primary Cilia.

Figure 1. Rapid recruitment of a deglutamylase to centrosomes for reducing centrosomal glutamylation.

(A) Schematic diagram of our newly developed approach, which recruits an engineered deglutamylase from the cytosol to centrosomes via a chemically inducible dimerization system. The cytosolic FKBP-tagged deglutamylase is recruited to the FRB-tagged centrosomal targeting protein (CEP170C) upon rapamycin treatment (Rapa). The study evaluates the physiological roles of centrosomal glutamylation in ciliogenesis, as well as the structural integrity and functions of centrosomes. (B) Rapid recruitment of FKBP-tagged proteins from the cytosol to centrosomes. COS7 cells co-transfected with Cerulean3 (Ce3)-FRB-CEP170C and Neon-FKBP were treated with rapamycin (Rapa; 100 nM) for 30 s. Dashed lines indicate the cell boundary. Scale bar, 10 μm. (C) Video frames of the centrosome region highlighted in the dashed squares in (B). (D) Normalized fluorescence intensity of Neon-FKBP at centrosomes upon treatment with DMSO control (0.1%; red) or rapamycin (Rapa, 100 nM; green). Data represent the mean ± SEM. n = 21 and 24 cells in the DMSO and rapamycin groups, respectively, from four independent experiments. The arrow indicates the time of treatment. (E) COS7 cells were co-transfected with Ce3-FRB-CEP170C and either CCP5CD-Neon-FKBP or the enzymatically inactive mutant, CCP5CDDM-Neon-FKBP. After 24 h of transfection, cells were treated with 0.1% DMSO or rapamycin (Rapa; 100 nM) for 30 min, followed by immunostaining with GT335 antibody (red). The right panels show magnified images of the areas outlined by the dashed squares. Scale bar, 10 μm. (F) Normalized density of glutamylated tubulin at centrosomes after rapamycin-induced translocation of Neon-FKBP (green), CCP5CD-Neon-FKBP (red) or CCP5CDDM-Neon-FKBP (blue) for the indicated times. Data represent the mean ± SEM. (n = 811, 725, and 674 cells for the Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively, from three to five independent experiments). The arrow indicates the time of rapamycin treatment. Student’s t tests were performed, and the corresponding P values are indicated. Source data are available online for this figure.

Figure 2. Centrosomal hypoglutamylation has minimal impact on centrosome structure.

(A–E) NIH3T3 cells co-transfected with Ce3-FRB-CEP170C and the indicated constructs were treated with either 0.1% DMSO or rapamycin (Rapa, 100 nM) for 30 min and then immunostained with antibodies against Centrin2 (A), Pericentrin (B), CEP83 (C), CEP164 (D), and ODF2 (E). The right panels show magnified images of the areas outlined by the dashed squares. Scale bar, 10 µm. (F–J) Normalized density of the indicated centrosome proteins after 0.1% DMSO (green) or rapamycin (100 nM; red) treatment in cells transfected with the indicated constructs (A–E). Data represent the mean ± SD. (n = 388 (F), 343 (G), 311 (H), 210 (I), and 224 cells (J) from three to five independent experiments). (K). 3D-SIM images of centrosomes in COS7 cells transfected with CCP5CD-mCh-FKBP-P2A-FRB-CEP170C upon 0.1% DMSO or rapamycin treatment. Cells were immunostained with antibodies against mCherry (mCh; red) and acetylated tubulin (act-tub; a marker of centrioles; green). Scale bar, 1 µm. (L, M) Normalized intensity of acetylated tubulin (L) and centriole length (M) in cells from (K) after 30 min of 0.1% DMSO (red) or 100 nM rapamycin (blue) treatment. Data represent as the mean ± SEM. n (from left to right) = 80, 93, 53, and 60 cells in (L); 18, 26, 10, and 12 cells in (M); from three independent experiments. Student’s t tests were performed, and P values are indicated. Source data are available online for this figure.

Figure 3. Centrosomal glutamylation is essential for microtubule nucleation.

(A–D) COS7 cells were co-transfected with Ce3-FRB-CEP170C and either CCP5CD-Neon-FKBP or CCP5CDDM-Neon-FKBP. At 80–90% confluency, transfected cells were treated with 100 nM rapamycin (Rapa) or 0.1% DMSO for 30 min and then placed on ice for 40 min to depolymerize microtubules. Cells were subsequently washed with DMEM for 1 min, allowed to recover for 3 min at room temperature (RT), and then fixed for immunostaining with anti-α-tubulin antibody. The right panels show magnified images of the areas outlined by the dashed squares. Hollow arrowheads indicate regrown microtubules, as shown by α-tubulin-labeled filaments extending from the centrosomal region. Scale bar, 10 μm. (B) Quantification of centrosome-derived microtubule intensity with or without 3 min recovery in cells transfected with the indicated constructs. Data represent the mean ± SEM (n = 270, 499, and 439 cells for the Neon-FKBP (green), CCP5CD-Neon-FKBP (red), and CCP5CDDM-Neon-FKBP groups (blue), respectively, from four to five independent experiments). (C) COS7 cells were co-transfected with Ce3-FRB-CEP170C, CCP5CD-mCh-FKBP, and EB1-YFP. One day post-transfection, EB1-YFP-positive comets were monitored by live-cell imaging in the same cell before and after rapamycin-induced centrosomal hypoglutamylation. Centrosomal microtubule tracks were drawn based on EB1-YFP time-lapse imaging with 2 min duration. Insets show higher-magnification images of the centrosomal regions. Scale bar, 10 μm. (D) Quantification of EB1-YFP comet frequency emitted from centrosomes in cells co-transfected with Ce3-FRB-CEP170C and the indicated constructs before and after treatment with rapamycin (100 nM) for 30 min. n = 11, 13 and 10 cells in the Neon-FKBP, CCP5CD-Neon-FKBP and CCP5CDDM-Neon-FKBP groups, respectively, from four to five independent experiments. Student’s t tests were performed, and P values are indicated. Source data are available online for this figure.

Figure 4. Glutamylation physically recruits NEDD1 and γ-tubulin via electrostatic forces for microtubule nucleation.

(A–C) NIH3T3 cells co-transfected with Ce3-FRB-CEP170C and either CCP5CD-Neon-FKBP or CCP5CDDM-Neon-FKBP were treated with DMSO (0.1%) or rapamycin (Rapa, 100 nM) for 1 h. Following treatment, cells were fixed and immunostained for γ-tubulin (A), CEP192 (B), and NEDD1 (C), respectively. The right panels show magnified images of the areas outlined by dashed squares. Scale bar, 10 μm. (D–F) Quantification of the normalized fluorescence intensity of γ-tubulin (D), CEP192 (E), and NEDD1 (F) in cells from (A), (B), and (C), respectively. Data represent the mean ± SEM. n (Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP) = 87, 142, and 113 cells in (D); 117, 174, and 159 cells in (E); 153, 229, and 192 cells in (F), respectively, from three to four independent experiments. (G, H) NIH3T3 cells were co-transfected with Ce3-FRB-CEP170C and CCP5CD-Neon-FKBP or with Ce3-FRB-CEP170C-NEDD1 and CCP5CDDM-Neon-FKBP. In (G), transfected cells were treated with 100 nM rapamycin (Rapa) for 1 h and then immunostained for γ-tubulin. In (H), transfected cells were incubated on ice with nocodazole (3.3 μM) for 40 min to depolymerize microtubules. Following depolymerization, cells were washed with DMEM for 1 min, allowed to recover for 3 min at RT, and then immunostained with anti-α-tubulin antibody. Hollow arrowheads indicate regrown microtubules, as shown by α-tubulin-labeled filaments extending from the centrosomal region in (H). Scale bar, 2 μm. (I, J) Quantification of the normalized intensity of γ-tubulin in (I) and area of centrosome-derived microtubules in (J) are shown. Data represent the mean ± SEM. n = 212 and 222 cells in CCP5CD and CCP5CDDM groups, respectively, for (I); n = 158 and 149 cells in CCP5CD and CCP5CDDM groups, respectively, for (J); three independent experiments. (K) Schematic of the co-sedimentation experiment protocol. Purified glutamylated or unmodified microtubules were generated and incubated with NIH3T3 cell lysates. The lysate proteins co-sedimented with microtubules (MTs) were collected and subjected to immunoblotting. (L) Immunoblot analysis of cell lysates collected using the protocol in (K), probing for NEDD1, glutamylated tubulin, and α-tubulin. (M) Quantification of NEDD1 levels co-sedimented with unmodified or glutamylated microtubules. Data represent the mean ± SEM from three independent experiments. (N) 3D-SIM images of tubulin glutamylation (Glu-tub) and NEDD1 in COS7 cells. The distribution profiles of tubulin glutamylation and NEDD1 are shown from both top view and side view of centrosomes. Scale bar, 1 µm. (O) Charge potential analysis of the centrosome-binding domain of wild-type (WT) NEDD1 and the negatively charged NEDD1 mutant (Neg). (P) COS7 cells transfected with WT NEDD1 or Neg NEDD1 were immunostained for pericentrin (red). The right panels show magnified images of the areas outlined by dashed squares. Scale bar, 10 µm. (Q, R) Quantification of the percentage of cells with centrosomal or cytosolic localization of NEDD1 (Q) and the centrosome-to-cytosol intensity ratio of NEDD1 (R). Data represent the mean ± SEM in (Q) and mean ± SD (red) in (R). n = 200 and 204 cells in the WT and Neg NEDD1 groups, respectively; Three independent experiments. Student’s t tests were performed, and P values are indicated. Noted: Single focal plane of Ce3-FRB-CEP170C images may not always display a two-centriole pattern. Source data are available online for this figure.

Figure 5. Centrosomal deglutamylation perturbs the dynamics and distribution of centriolar satellites via NEDD1.

(A) NIH3T3 cells co-transfected with the indicated constructs were treated with DMSO (0.1%) or rapamycin (100 nM) for 30 min. Following treatment, cells were fixed and immunostained for PCM1. Representative images show condensed and dispersed PCM1 patterns in CCP5CD-Neon-FKBP-transfected cells. The right panels show magnified images of the areas outlined by dashed squares. Scale bar, 5 µm. (B) Percentage of cells exhibiting the indicated PCM1 pattern from (A). Data represent the mean ± SEM. n = 1035, 385, and 260 cells in Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively, from three independent experiments. (C) Normalized intensity of PCM1 in (A). Data represent the mean ± SD. n = 99, 139, 137 cells in Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively, from three independent experiments. (D) Quantification of the normalized area of PCM1-positive centriolar satellites around centrosomes in (A). Data represent the mean ± SEM. n = 216, 391, and 310 cells in the Neon-FKBP, CCP5CD-Neon-FKBP and CCP5CDDM-Neon-FKBP groups, respectively, from three to five independent experiments. (E) Video frames of NIH3T3 cells transfected with the indicated constructs upon rapamycin treatment (100 nM). Scale bar, 5 µm. (F) NIH3T3 cells co-transfected with the indicated constructs were treated with rapamycin (100 nM) for 1 h and immunostained for PCM1. Scale bar, 5 µm. (G) Percentage of cells exhibiting the indicated PCM1 pattern in (F). Data represent the mean ± SEM. n = 45, 56, 39, and 52 cells from left to right. Three independent experiments. (H) Quantification of the PCM1-positive centriolar satellite area around centrosomes in (E). Data represent the mean ± SEM. n = 42, 53, 36, and 50 cells from left to right. Three independent experiments. Student’s t tests were performed, and P values are indicated. Source data are available online for this figure.

Figure 6. Centrosomal glutamylation is required for ciliogenesis and cilia maintenance.

(A) Experimental protocol to induce acute centrosomal hypoglutamylation followed by serum starvation–induced ciliogenesis. Cells were serum-starved for 4 h or 24 h, after which nascent and mature primary cilia were stained for immunofluorescence analysis (IF). (B) NIH3T3 cells co-transfected with Ce3-FRB-CEP170C and either CCP5CD-Neon-FKBP or CCP5CDDM-Neon-FKBP were serum-starved (SS) for the indicated times according to the protocol in (A). Cilia were labeled by anti-Arl13B antibody. The right panels show magnified images of the areas outlined by dashed squares. Scale bar, 10 µm. (C) Quantification of cilium length in cells from (B). Data (black) represent the mean ± SEM. n = 225, 474, and 392 cells in Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively, obtained from 8 to 10 independent experiments. (D) Percentage of ciliated cells in (B). Data represent the mean ± SEM. n = 534, 399, and 446 cells in Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively, obtained from three independent experiments. (E) Experimental protocol to induce the formation of mature cilia via 24 h of serum starvation, followed by rapamycin (Rapa)-induced basal-body hypoglutamylation for the indicated times. (F) NIH3T3 cells co-transfected with Ce3-FRB-CEP170C and either CCP5CD-Neon-FKBP or CCP5CDDM-Neon-FKBP were treated according to the protocol in (E). Cilia were labeled by anti-Arl13B antibody. The right panels show magnified images of the areas outlined by dashed squares. Scale bar, 10 µm. (G) Quantification of cilium length of cells from (F). Data represent the mean ± SEM. n = 520, 946 and 276 cells in Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively, obtained from six independent experiments. (H) Percentage of ciliated cells in (F). Data represent the mean ± SEM. n = 580, 1093, and 244 cells in Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively, obtained from three to seven independent experiments. Student’s t tests were performed, and P values are indicated. Source data are available online for this figure.

Figure 7. Glutamylation regulates ciliogenesis via NEDD1 and Talpid3.

(A) NIH3T3 cells co-transfected with the indicated constructs were serum-starved for 24 h and then treated with rapamycin (100 nM) for 60 min. Following treatment, cells were fixed and immunostained for Arl13B. Scale bar, 2 µm. (B) Quantification of cilium length of NIH3T3 cells from (A). Data (black) represent the mean ± SD. n = 32, 45, 33, and 42 cells from left to right, respectively. Three independent experiments. (C) NIH3T3 cells co-transfected with the indicated constructs were serum-starved for 24 h and then treated with rapamycin (100 nM) for the indicated times. Following treatment, cells were fixed and immunostained for Talpid3. Scale bar, 5 µm. (D) Normalized Talpid3 intensity in NIH3T3 cells from (C) after rapamycin treatment for the indicated times. Data represent the mean ± SEM. n = 171, 327 and 311 cells in the Neon-FKBP, CCP5CD-Neon-FKBP, and CCP5CDDM-Neon-FKBP groups, respectively. Three independent experiments. (E). NIH3T3 cells co-transfected with the indicated constructs were serum-starved for 24 h and then treated with rapamycin (100 nM) for 1 h. Following treatment, cells were fixed and immunostained for Arl13B. Scale bar, 1 µm. (F) Quantification of cilium length in NIH3T3 cells from (E). Data (black) represent the mean ± SD. n = 134, 83, 155, and 87 cells from left to right. Seven independent experiments. Student’s t tests were performed, and P values are indicated. Source data are available online for this figure.

Figure 8. Centrosomal hypoglutamlyation perturbs mitotic spindle formation and prolongs mitosis.

(A) Schematic of the protocol for RO-3306-induced cell synchronization at the G2/M transition, followed by rapamycin (Rapa)-induced centrosomal hypoglutamylation after RO-3306 washout (W/O). Cell morphology was analyzed by immunofluorescence (IF) 30 min after rapamycin or DMSO treatment. (B) HeLa cells co-transfected with the indicated constructs were synchronized at the G2/M transition and released into mitosis for 30 min in the presence of DMSO (0.1%) or rapamycin (Rapa, 100 nM). Mitotic spindles and chromosomes were labeled with anti-α-tubulin antibody (green) and DAPI (blue), respectively. Dashed lines indicate cell boundaries. Scale bar, 10 μm. (C) Quantification of the percentage of cells displaying a normal mitotic spindle pattern as shown in (B). Data are presented as mean ± SEM. n = 90 and 78 cells in CCP5CD and CCP5CDDM groups, respectively; three independent experiments. (D) Representative video frames of HeLa cells co-transfected with the indicated constructs, captured 30 min after treatment with DMSO (0.1%) or rapamycin (100 nM). Red dashed lines indicate the boundaries of mitotic cells. Scale bar, 10 μm. (E) Quantification of mitosis duration (from metaphase to telophase) in HeLa cells co-transfected with mCh-FRB-CEP170C and the indicated constructs, treated with DMSO (0.1%) or rapamycin (100 nM). Data are presented as mean ± SD (black). n = 90 and 78 cells in the CCP5CD and CCP5CDDM groups, respectively. Three independent experiments. Student’s t tests were performed, and P values are indicated. Source data are available online for this figure.

Figure EV1. Specific glutamylation reduction occurs at two centrioles in three different cell types.

(A–C) COS7 cells co-transfected with Ce3-FRB-CEP170C and the indicated constructs were treated with either 0.1% DMSO (green circles) or 100 nM rapamycin (Rapa, red circles) for 1 h in (A, B) or for 24 h in (C). Treated cells were immunostained with GT335 antibody to assess the level of glutamylated tubulin in cytosol (A); the relative difference in glutamylated tubulin intensity between two centrioles (B); and the normalized intensity of centrosomal glutamylation in (C). Data are presented as mean ± SEM. n (from left to right) = 23, 18, 28, 22, 39, and 23 cells in (A); 21, 25, 78, 18, 46, and 13 cells in (B); 47, 27, 71, 46, 85, and 22 cells in (C), collected from 3 independent experiments. Statistical significance was determined using a one-way ANOVA test. “NS” indicates no significant difference among groups in (A, B). Student’s t tests were performed, and P values are indicated in (C). (D) COS7, NIH3T3, and U2OS cells were transfected with Ce3-FRB-CEP170C and CCP5CD-Neon-FKBP. Transfected cells were treated with 0.1% DMSO (green) or rapamycin (100 nM) for 30 min. The normalized level of tubulin glutamylation at centrosomes under the indicated conditions is shown. Data (black) represent the mean ± SD. n = 105, 107, 97, 113, 133, and 112 cells from left to right. Three independent experiments. Student’s t tests were performed, and P values are indicated.

Figure EV2. CEP192 is sufficient to recruit NEDD1 and γ-tubulin to hypoglutamylated centrosomes.

(A, B) COS7 cells co-transfected with the indicated constructs were treated with DMSO (0.1%) or Rapa (100 nM rapamycin) for 1 h. Following treatment, cells were immunostained with antibodies against NEDD1 (A) and γ-tubulin (B), respectively. Scale bar, 10 µm. (C, D) Quantification of the normalized intensity of NEDD1 (C) in cells from (A) and γ-tubulin (D) in cells from (B). Data represent as mean ± SEM. n (from left to right) = 62, 51, 84, 101, 67, 55, 53, and 54 cells in (C); 143, 152, 134, 135, 37, 36, 85, and 69 cells in (D); 3–6 independent experiments. Students’ t tests were performed, and P values are indicated.