Daughter centriole elongation is controlled by proteolysis.

The centrosome is the major microtubule-organizing center of most mammalian cells and consists of a pair of centrioles embedded in pericentriolar material. Before mitosis, the two centrioles duplicate and two new daughter centrioles form adjacent to each preexisting maternal centriole. After initiation of daughter centriole synthesis, the procentrioles elongate in a process that is poorly understood. Here, we show that inhibition of cellular proteolysis by Z-L3VS or MG132 induces abnormal elongation of daughter centrioles to approximately 4 times their normal length. This activity of Z-L3VS or MG132 was found to correlate with inhibition of intracellular protease-mediated substrate cleavage. Using a small interfering RNA screen, we identified a total of nine gene products that either attenuated (seven) or promoted (two) abnormal Z-L3VS-induced daughter centriole elongation. Our hits included known regulators of centriole length, including CPAP and CP110, but, interestingly, several proteins involved in microtubule stability and anchoring as well as centrosome cohesion. This suggests that nonproteasomal functions, specifically inhibition of cellular proteases, may play an important and underappreciated role in the regulation of centriole elongation. They also highlight the complexity of daughter centriole length control and provide a framework for future studies to dissect the molecular details of this process.

INTRODUCTION

Centrosomes are the major microtubule-organizing centers during interphase and mitosis in most mammalian cells (Azimzadeh and Bornens, 2007). The centrosome consists of a pair of centrioles surrounded by pericentriolar material. Each centriole is composed of nine triplet microtubules arranged in a cylindrical manner (Strnad and Gonczy, 2008). A centrosome consists of an older, mature centriole that is distinguishable from the younger (daughter) centriole by distal and subdistal appendages, which are important for microtubule nucleation and anchoring (Paintrand ; Piel ; Azimzadeh and Bornens, 2007). The maternal centriole also acts as a basal body to initiate the formation of primary cilia (Azimzadeh and Bornens, 2007).

To orchestrate bipolar mitotic spindle formation, the centrosome duplicates precisely once before mitosis. Centrosome duplication begins during late mitosis–early G1 when the two centrioles disengage (Tsou and Stearns, 2006). Procentrioles start to form in the G1 phase of the cell cycle, adjacent to each of the two preexisting centrioles. Recent results suggest that activation of polo-like kinase 4 (PLK4) on the wall of the maternal centrioles is an early event during procentriole assembly (Habedanck ; Cunha-Ferreira ). After activation of PLK4, hSAS-6, which is a component of the cartwheel structure that forms the basis of centriolar ninefold symmetry (Leidel ; Nakazawa ; Strnad ), is recruited to nascent procentrioles (Kleylein-Sohn ). Centrosomal P4.1-associated protein (CPAP) and other structural proteins subsequently initiate daughter centriole elongation. It has recently been shown that CPAP promotes centriolar microtubule assembly through the binding of tubulin heterodimers, thereby facilitating procentriole elongation (Kleylein-Sohn ; Kohlmaier ; Schmidt ; Tang ). CP110 is then recruited to the growing end of the daughter centriole, possibly forming a cap under which regulated tubulin subunit assembly can occur (Chen ; Kleylein-Sohn ; Schmidt ). The two new centrosomes are held together by a proteinaceous linker, composed of C-Nap1 and other proteins (Mayor ), until late G2 phase when centrosome separation occurs and spindle pole formation is initiated.

Daughter centriole elongation begins during S phase, and centrioles reach ∼80% the length of the maternal centriole in late G2 (Chretien ). The daughter centriole reaches the full length of the maternal centriole, but normally not beyond, during the following cell cycle (Azimzadeh and Bornens, 2007). Besides CPAP and CP110, little is currently known about the mechanisms regulating the length of daughter centrioles.

We have shown previously that cells treated with the proteasome inhibitor Z-L3VS for 48 h show a concurrent assembly of multiple daughter centrioles around single maternal centrioles (centriole multiplication), indicating that normal daughter centriole biogenesis is restrained by proteolysis (Duensing , 2009). Here, we used a panel of proteasome inhibitors and show that cells exposed to Z-L3VS or MG132 contain abnormally elongated daughter centrioles that reach approximately 4 times the length of normal daughter centrioles. We demonstrate that the ability of Z-L3VS and MG132 to potently induce daughter centriole elongation correlates with their ability to inhibit the hydrolysis of casein, a well-established substrate of intracellular proteases. Combining our assay system of Z-L3VS–induced centriole elongation with a small interfering RNA (siRNA) screen targeting 127 known centrosomal proteins (Andersen ), we were able to identify seven centrosomal proteins that attenuated daughter centriole elongation when knocked down (FOP, CAP350, CPAP, hSAS-6, Cep170, ninein, and C-Nap1) and two centrosomal proteins that promoted this process when depleted (Cep97 and CP110). Our results reveal an unexpected complexity of daughter centriole length control and highlight the critical role of proteolysis in this process.

MATERIALS AND METHODS

Antibodies

Rabbit anti-Cep97 and CP110 were a kind gift from Brian Dynlacht (NYU Cancer Institute, New York, NY; Spektor ). Mouse anti-Cep170 and mouse anti-ninein were a kind gift from Erich A. Nigg (Max Plank Institute of Biochemistry, Martinsried, Germany; Guarguaglini ). Rabbit anti-CPAP was a kind gift from Tang K. Tang (Institute of Biomedical Sciences, Taipei, Taiwan; Tang ). Rabbit anti-CAP350 was obtained commercially from Novus Biologicals (Littleton, CO). Rabbit anti-FGFR1OP was obtained commercially from Proteintech Group (Chicago, IL). Mouse anti-C-Nap1 was obtained commercially from BD Biosciences Transduction Laboratories (San Jose, CA), and mouse anti-hSAS6 was obtained commercially from Santa Cruz Biotechnology (Santa Cruz, CA).

Cell Culture and Inhibitor Treatment

The human U-2 OS osteosarcoma cell line was obtained from American Type Culture Collection (Manassas, VA) and maintained in DMEM (Lonza Walkersville, Walkersville, MD) supplemented with 10% fetal bovine serum (Mediatech, Herndon, VA), 50 U/ml penicillin, and 50 mg/ml streptomycin (Lonza Walkerville). U-2 OS cells were engineered to stably express a centrin-1-green fluorescent protein (GFP)-encoding construct (kindly provided by Michel Bornens, Institut Curie, Paris, France; Piel ). Proteasome inhibitors were used at the following concentrations that were associated with at least 50% viable cells after 72 h: Z-L3VS (used at a 1 μM concentration; BIOMOL Research Laboratories, Plymouth Meeting, PA), MG132 (used at a 1 μM concentration), MG262 (used at a 0.001 μM concentration), lactacystin (used at a 1 μM concentration; all Boston Biochem, Cambridge, MA), and epoxomicin (used at a 0.01 μM concentration; Calbiochem, San Diego, CA) were dissolved in dimethyl sulfoxide (DMSO). In all experiments, solvent controls were included using 0.1% DMSO.

siRNA

Proteins to be tested in the siRNA screen were depleted using RNA duplexes (QIAGEN, Valencia, CA) targeting known centriolar associated proteins as described previously (Andersen ; Supplemental Table 1). For the centriole elongation screen, U-2 OS/centrin-GFP cells were grown in 12-well tissue culture plates, on coverslips, with 0.5 ml of DMEM free of antibiotics. Cells were transfected with 3 μl of 20 μM annealed RNA duplexes using Oligofectamine (Invitrogen, Carlsbad, CA) transfection reagent. Twenty-four hour posttransfection cells were treated with 1 μM of the proteasome inhibitor Z-L3VS and were analyzed 72 h postinhibitor addition. Primary and secondary siRNA screens were performed using target sequences shown in Supplemental Table 1. RNA duplexes that yielded a >20% increase or decrease of the proportion of cells with elongated centrioles normalized to Z-L3VS-treated, control siRNA-transfected cells in both rounds of screens were considered hits.

For each immunofluorescence or immunoblotting experiment, U-2 OS/centrin-GFP cells were grown in 60-mm tissue culture dishes with 2 ml of DMEM free of antibiotics. Cells were transfected with 12 μl of 20 μM annealed RNA duplexes using Oligofectamine transfection reagent and treated as described above. Knockdown efficiency was monitored by Western blot analysis or immunofluorescence microscopy (Supplemental Figure 1).

Protease Assay

For analyzing protease inhibition, U-2 OS cells were grown in 60-mm tissue culture dishes, on coverslips, and treated for 72 h with the following proteasome inhibitors at the indicated concentrations: ZL3VS (1 μM), MG132 (1 μM), MG262 (0.001 μM), lactacystin (1 μM), epoxomicin (0.01 μM), or 0.1% DMSO as a control. After this treatment, the cells were overlaid with ∼500 ng of BODIPY TR-X-casein diluted in digestion buffer (10 mM Tris-HCl, pH 7.8, with 0.1 mM sodium azide) (Invitrogen), or 50 μl of the digestion buffer alone as a control and permeabilized with 500 mg of 425–600 μM acid-washed glass beads (Sigma, St. Louis, MO). The cells were then incubated for 24 h at 37°C in media containing the appropriate proteasome inhibitor or DMSO control, fixed in 4% paraformaldehyde/phosphate-buffered saline (PBS), and mounted with 4,6-diamidino-2-phenylindole (DAPI). Cells were analyzed using an AX70 epifluorescence microscope (Olympus, Tokyo, Japan) equipped with a SpotRT digital camera.

Immunoblotting

For immunoblot analyses, cell lysates were prepared by scraping cells into lysis buffer (1% NP-40, 50 mM Tris-HCl, pH 8.0, 100 mM sodium fluoride, 30 mM sodium pyrophosphate, 2 mM sodium molybdate, 5 mM EDTA, and 2 mM sodium orthovanadate in distilled H2O) containing protease inhibitors (10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 μM phenylmethylsulfonyl fluoride, and 2 μM vanadate). Lysates were incubated for 1 h with rotating at 4°C and then cleared by centrifugation for 30 min at 13,000 rpm at 4°C. Protein concentrations were determined by the Bradford assay (Bio-Rad Laboratories, Hercules, CA). Thirty micrograms of protein was loaded on a 4–12% Bis-Tris or 3–8% Tris-acetate gel (Invitrogen) and blotted onto a nitrocellulose membrane.

Immunofluorescence Microscopy

For immunofluorescence microscopic analyses, cells grown on coverslips were fixed in 4% paraformaldehyde/PBS for 15 min at room temperature, washed in PBS, and permeabilized with 1% Triton-X-100 in PBS for 20 min. After blocking in 10% normal donkey serum (Jackson ImmunoResearch Laboratories, West Grove, PA), cells were incubated with primary antibody overnight. The next morning, cells were warmed at 37°C for 2 h, washed in PBS, and incubated with rhodamine red-conjugated secondary antibody for 2 h and mounted with DAPI. Cells were analyzed using an AX70 epifluorescence microscope (Olympus) equipped with a SpotRT digital camera.

Electron Microscopy (EM)

For transmission electron microscopy, the samples were fixed for at least 1 h in a 2% glutaraldehyde solution buffered with PBS. After washing in three changes of PBS, the samples were placed in a 1% osmium tetroxide solution buffered with PBS for 1 h, followed by a series of rinses with ethanol solutions of increasing concentration (50, 70, 95, and 100%). The samples were then placed in a 1:1 mixture of Epon Araldite resin and propylene oxide and held overnight in a desiccator. The following day, the Epon Araldite and propylene oxide mixture was removed and replaced with 100% Epon Araldite resin. The samples were infiltrated with the resin for an additional 8 h, placed in embedding molds, and polymerized for 48 h at 60°C. Serial thin sections were cut using a Reichert-Jung Ultracut E ultramicrotome and a DDK Diamond knife. Thin sections were picked up on copper grids and stained with 1% uranyl acetate and Reynold's lead citrate. The sections were viewed on an H-7100 TEM transmission electron microscope (Hitachi High Technologies America, Pleasanton, CA). Digital images were obtained using an AMT Advantage 10 CCD Camera System (Advanced Microscopy Techniques, Danvers, MA) and Image software (National Institutes of Health, Bethesda, MD).

Statistical Methods

Student's two-tailed t test for independent samples was used where applicable.

RESULTS

Inhibition of the Proteasome Induces Abnormal Elongation of Daughter Centrioles

To examine the role of proteolysis in centriole biogenesis, we treated U-2 OS cells stably expressing GFP-tagged centrin (U-2 OS/centrin-GFP), with the proteasome inhibitor Z-L3VS (Bogyo ). Treatment of cells for 72 h with Z-L3VS was found to induce a significant elongation of centrioles (Figure 1A). Elongated centrioles were typically arranged in a “flower”-like pattern around a large centrin-GFP dot and described before in Z-L3VS–treated cells (Duensing ). Occasionally elongated centrioles were found to have a bifurcated end (Figure 1A, bottom left, arrow). This observation is similar to previous reports when abnormal centriole elongation was induced through prolonged overexpression of the protein CPAP (Kohlmaier ; Schmidt ; Tang ). Z-L3VS–induced abnormally elongated centrioles were stable and persisted in a significant proportion of cells for up to 72 h after removal of the drug (data not shown).

Figure 1.

Inhibition of the proteasome stimulates aberrant centriole elongation. (A) Fluorescence microscopic analysis of U-2 OS cells stably expressing centrin-GFP after either treatment with 0.1% DMSO (top left) or 1 μM Z-L3VS (remaining panels) for 72 h. Arrow indicates a bifurcated distal end of an elongated centriole. Arrowhead points to an example of an elongated centriole with segmentation of the centrin-GFP signal. (B and C) Quantification of centriole length in U-2 OS/centrin GFP cells with abnormally elongated centrioles after treatment with 0.1% DMSO or 1 μM Z-L3VS for 72 h. Centriole length measurements were performed using ImageJ and are expressed in micrometers as described in Results. (D) Quantification of the percentage of U-2 OS/centrin-GFP cells containing elongated centrioles after treatment with Z-L3VS (1 μM), MG132 (1 μM), lactacystin (1 μM), MG262 (0.001 μM), and epoxomicin (0.01 μM) for 72 h. Cells treated with 0.1% DMSO for 72 h were used as controls. All these inhibitor concentrations were associated with at least 50% cell viability. Each bar represents mean and SE of at least two independent experiments with a minimum of 100 cells counted per experiment. Asterisks indicate statistically significant differences (p ≤ 0.0001; Student's t test for independent samples). (E) Immunofluorescence microscopic analysis of red fluorescent BODIPY TR-X dye-labeled casein cleavage in U-2 OS/centrin-GFP cells treated with Z-L3VS (1 μM), MG132 (1 μM), or lactacystin (1 μM). Cells treated with 0.1% DMSO for 96 h were used as controls. Protease-catalyzed cleavage of the red fluorescent dye-labeled BODIPY TR-X-casein relieves the inherent quenching of the dye and results in brightly fluorescent BODIPY TR-X dye-labeled peptides. Bar, 10 μm.

To assess the increase in centriole length induced by Z-L3VS, we measured the length of daughter centrioles in control and Z-L3VS–treated cells by using the ImageJ software (http://rsbweb.nih.gov/ij/) and compared these measurements to a 10-μm reference standard to obtain the length of daughter centrioles expressed in micrometers (Figure 1, B and C). The average length of centrioles in DMSO-treated control cells (0.6 μm) is very similar to the length obtained for centrioles in a report recently published by Tang et al., which validates our method of centriole length measurement (Figure 1B; Tang ). In cells treated with Z-L3VS, centrioles were, on average, 2.2 μm long, or 4.4-fold longer than DMSO-treated control cells. The maximum length of individual centrioles (n = 30) in Z-L3VS–treated cells was considerably longer and varied significantly compared with the shorter and more constant length of individual centrioles in DMSO-treated control cells (n = 30; Figure 1, B and C). Together, these results underscore the intrinsic ability of centrioles to elongate to several times their normal length.

We next tested a panel of different proteasome inhibitors for their ability to induce abnormal centriole elongation using a drug concentration that resulted in a cell viability of at least 50%. Quantification of cells with elongated centrioles revealed a significant 95.8-fold increase in populations treated with 1 μM Z-L3VS for 72 h (31.6%; p ≤ 0.0001) compared with DMSO-treated controls (0.33%; Figure 1D). A significant increase of cells with elongated daughter centrioles, although to a lesser extent, was also detected in cells treated with the proteasome inhibitor MG132 (17.3%; p ≤ 0.0001) but not in cells treated with MG262, lactacystin, or epoximicin (Figure 1D).

To confirm the role of proteolysis in centriole length control, we repeated these experiments with the proteasome inhibitors that did not initially induce centriole elongation at higher drug concentrations, albeit at the expense of cell viability. Cells treated with a 10-fold increase in concentration of MG262, a fivefold increase in concentration of epoxomicin, and a 10-fold increase in concentration of lactacystin exhibited elongated centrioles in 2.0, 3.7, and 2.3% of cells, respectively (data not shown).

We next sought to determine why Z-L3VS and MG132 were much more potent inducers of abnormal daughter centriole elongation compared with the other potent proteasome inhibitors, such as lactacystin. One explanation for this difference could be that both Z-L3VS and MG132 are known to have nonproteasomal proteolytic inhibitory activities against intracellular proteases, whereas the remaining proteasome inhibitors in our screen have less of these activities.

To address this hypothesis, we devised a new technique to analyze the ability of our panel of proteasome inhibitors to inhibit intracellular protease activity. To this end, we used a well established protease substrate, casein, which is known to be multiply cleaved by many abundant cellular proteases, including chymotrypsin, trypsin, and elastase (Twining, 1984). Protease-catalyzed cleavage of the red fluorescent dye-labeled BODIPY TR-X-casein relieves the inherent quenching of the dye and results in brightly fluorescent BODIPY TR-X dye-labeled peptides. We found that in DMSO-treated control cells, the BODIPY TR-X-dye-labeled casein was cleaved, resulting in the production of smaller dye-labeled peptide products represented by small dots (Figure 1E). Treatment of cells with the proteasome inhibitors MG262, epoxomicin or lactacystin, which were unable to induce centriole length, also resulted in cleavage of the dye-labeled casein into small peptide fragments, albeit to a lesser extent than DMSO-treated cells (Figure 1E and Supplemental Figure 1). In contrast, when cells were treated with either Z-L3VS or MG132, the proteasome inhibitors that were able to induce abnormal centriole elongation, there was little to no cleavage of the dye-labeled casein, suggesting that Z-L3VS and MG132-treamtent prevented hydrolysis of the casein substrate (Figure 1E and Supplemental Figure 1). These results suggest a correlation between inhibition of intracellular proteases and induction of abnormal centriole elongation.

To determine that the elongated centrin-positive structures observed in Z-L3VS–treated U-2 OS/centrin-GFP cells were in fact daughter centrioles, we examined Z-L3VS–treated cells by ultrathin serial section EM. As shown in Figure 2A, the elongated structures observed were bona fide daughter centrioles as indicated by the lack of detectable appendages.

Figure 2.

Inhibition of the proteasome stimulates elongation of daughter but not maternal centrioles. (A) Consecutive serial section electron microscopic analysis of an abnormally elongated centriole after Z-L3VS treatment of U-2 OS/centrin-GFP cells. Arrows point to an elongated daughter centriole. Bar, 500 nm. (B) Consecutive serial section electron microscopic analysis of an elongated daughter centriole after 72-h treatment of U-2 OS/centrin-GFP cells with Z-L3VS. Note the less electron dense area and the continuous microtubules that span this region (arrows). Bar, 100 nm. (C and D) Immunofluorescence microscopic analysis for Cep170 (C) and ninein (D) in U-2 OS/centrin-GFP cells after a 72 h treatment with the proteasome inhibitors Z-L3VS (1 μM) or 0.1% DMSO as control. Nuclei stained with DAPI. Bar, 10 μm.

While analyzing elongated daughter centrioles, we observed regions of the elongated daughter that contained less intense centrin staining (Figure 1A, arrowhead). This finding raised the question whether elongated daughters consist of centrin-containing segments or continuously elongated centriolar structures. To answer this question, we specifically analyzed these regions of abnormally elongated daughter centrioles. As shown in Figure 2B, we found microtubules spanning this electron dense region by ultrathin serial section EM, underscoring that elongated daughter centrioles consist of elongated centriolar microtubules and not discrete centrin-containing segments.

To further substantiate that these elongated structures represent elongated daughter centrioles, we performed immunofluorescence staining of Cep170 and ninein, two markers of mature maternal centrioles (Mogensen ; Guarguaglini ). Elongated centrioles were grouped around a central, maternal centriole and did not colocalize with either Cep170 or ninein, indicating that elongated centrin positive structures were indeed daughter centrioles (Figure 2, C and D).

Together, our results suggest that Z-L3VS and MG132 treatment leads to an abnormal elongation of daughter centriolar microtubules and that normal daughter centriole length is not a result of structural constraints but is controlled by proteolysis.

Abnormal Elongation of Daughter Centrioles Involves Several Centrosomal Proteins

To determine the underlying mechanisms of abnormal elongation of daughter centrioles, we performed an siRNA screen focusing on 127 proteins known to be associated with centrosomes (Andersen ; Supplemental Table 1). Twenty-four hours after siRNA treatment of U-2 OS/centrin-GFP cells, elongation of daughter centrioles was induced by treatment of cells with the proteasome inhibitor Z-L3VS for 72 h. Protein depletion was assessed by immunoblot analysis or immunofluorescence microscopy (Supplemental Figure 2). Of 127 proteins analyzed, depletion of nine centrosomal proteins reproducibly changed the percentage of cells containing elongated centrioles when normalized to Z-L3VS–treated control siRNA-transfected cells, suggesting that the target proteins may be involved in regulating centriole elongation (Figure 3A). Depletion of seven proteins, FOP, CAP350, CPAP, hSAS6, Cep170, ninein, and C-Nap1 led to a decrease in the number of cells that contained elongated daughter centrioles, whereas depletion of two proteins, Cep97 and CP110, led to an increase in the number of cells that contained elongated daughter centrioles (Figure 3A).

Figure 3.

An siRNA screen to identify centrosomal proteins involved in Z-L3VS–induced abnormal daughter centriole elongation. (A) U-2 OS-centrin/GFP cells were transfected for 24 h with control or siRNA duplexes targeting 127 known centrosomal proteins (Andersen ) followed by a 72-h exposure to the proteasome inhibitor Z-L3VS (1 μM) or 0.1% DMSO as control. Cells were analyzed by fluorescence microscopy to determine the proportion of cells that contained abnormally long daughter centrioles. The bar graph shows negative (black bars) and positive (gray bars) regulators of Z-L3VS–induced daughter centriole elongation. Each bar represents the proportion of cells containing elongated centrioles when normalized to Z-L3VS–treated cells transfected with control siRNA duplexes (set to 100%). Mean and SE of at least two independent experiments with a minimum of 100 cells counted per experiment is shown. (B and C) Immunofluorescence microscopic analysis for FOP (B) and CAP350 (C) in U-2 OS/centrin-GFP cells after a 72-h treatment with the proteasome inhibitors Z-L3VS, MG132, or lactacystin (all 1 μM). Treatment with 0.1% DMSO was used as control. Bar, 10 μm. (D) Coimmunofluorescence microscopic analysis for FOP and CAP350 in U-2 OS/centrin-GFP cells after a 72-h treatment with the proteasome inhibitors Z-L3VS (1 μM) or 0.1% DMSO. Nuclei stained with DAPI.

Depletion of two proteins implicated in microtubule anchoring and stability, FOP and CAP350 (Yan ; Hoppeler-Lebel ; Le Clech, 2008), significantly reduced the number of cells containing abnormally elongated daughter centrioles to 50.7 and 54.7%, respectively, of control siRNA transfected Z-L3VS–treated cells. Furthermore, depletion of Cep97 and CP110, two proteins that were previously shown to suppress the formation of cilia (Spektor ), led to an increase in the number of cells that contained long centrioles to 162.7 and 153.3%, respectively, of control siRNA-transfected Z-L3VS–treated cells. This is in line with previous results suggesting that CP110 plays a negative regulatory role in centriole elongation (Schmidt ; Tang ).

Depletion of hSAS-6, a structural protein that is required for daughter centriole synthesis (Strnad ), decreased the percentage of Z-L3VS–treated cells that contained long daughter centrioles to 76% of control siRNA Z-L3VS–treated cells (Figure 3A). In addition, depletion of CPAP, a protein that has been shown previously to play a role in centriole elongation (Kohlmaier ; Schmidt ; Tang ), also decreased the percentage of Z-L3VS–treated cells with long daughter centrioles to 73.2% of control siRNA Z-L3VS–treated cells (Figure 3A). Remarkably, depletion of proteins implicated either in maintaining centrosome cohesion or associated with maternal centriole appendages, C-Nap1 (Mayor ), ninein (Mogensen ), and Cep170 (Guarguaglini ), also caused a reduction in the number of cells containing long centrioles to 67.1, 60.3, and 72.6%, respectively, compared with control siRNA-transfected Z-L3VS–treated cells (Figure 3A).

Together, these results suggest that Z-L3VS–induced abnormal centriole elongation involves known positive and negative microtubule regulatory proteins as well as proteins involved in centrosome cohesion and microtubule anchoring.

Z-L3VS Alters the Expression of Several Centriolar Proteins Involved in Length Control

Next, we asked whether Z-L3VS– or MG132-induced abnormal daughter centriole elongation was associated with alteration of the localization, abundance, or both of the nine proteins we identified in our siRNA screen. First, we analyzed the localization of the two major microtubule-stabilizing proteins identified in our screen, FOP and CAP350. Immunofluorescence microscopy of FOP in DMSO-treated control cells showed FOP colocalizing with centrioles (Figure 3B, top), as noted previously (Yan ; Le Clech, 2008). When cells were treated with Z-L3VS or MG132, staining for FOP was found to colocalize along elongated daughter centrioles (Figure 3B, middle). However, when cells were treated with lactacystin, a proteasome inhibitor that is less potent to induce abnormal elongation of daughter centrioles, FOP localization was similar to that of DMSO-treated control cells (Figure 3B, bottom).

Immunofluorescence microscopic analysis of CAP350 in DMSO-treated control cells showed CAP350 localization to both mother and daughter centrioles as reported previously (Yan ; Le Clech, 2008; Figure 3C, top). When CAP350 was analyzed after Z-L3VS or MG132 treatment, we detected CAP350 colocalizing along elongated daughter centrioles, similar to the localization of FOP after treatment with these proteasome inhibitors (Figure 3C, middle). When cells were treated with lactacystin, CAP350 localization was comparable with that of DMSO-treated control cells (Figure 3C, bottom). We confirmed the similar localization of FOP and CAP350 along elongated daughters by performing coimmunofluorescence microscopic analysis after Z-L3VS-treatment (Figure 3D).

To assess the accumulation of FOP and CAP350 protein, respectively, an immunoblot analysis of whole cell extracts from cells treated for 72 h with our panel of proteasome inhibitors was performed. Z-L3VS or MG132 treatment, which both induce abnormal elongation of daughter centrioles, resulted in the accumulation of both FOP and CAP350 to significantly higher levels than in DMSO-treated controls (Figure 4A). The proteasome inhibitors that were unable to induce abnormal daughter centriole elongation at concentrations associated with at least 50% cell viability, did not demonstrate an accumulation of either FOP or CAP350 compared with DMSO-treated control cells (Figure 4A).

Figure 4.

Stabilization of FOP and CAP350 protein by both Z-L3VS and MG132 treatment. (A) Immunoblot analysis of FOP and CAP350 in U-2 OS/centrin-GFP cells after treatment with either 0.1% DMSO or proteasome inhibitors as indicated for 72 h. Immunoblot for actin is shown to demonstrate protein loading. Note accumulation of both CAP350 and FOP in Z-L3VS–treated cells. (B) Quantification of the immunofluorescence intensity of FOP and CAP350 in U-2 OS/centrin GFP cells with abnormally elongated centrioles after treatment with 0.1% DMSO or 1 μM Z-L3VS for 72 h. Fluorescence intensity measurements were performed using ImageJ.

To corroborate that accumulation of FOP and CAP350 was in fact occurring at the centrosome, we performed semiquantitative analysis of centrosomal protein levels by immunofluorescence analysis using ImageJ software (Figure 4B). We found a 4.9-fold increase in FOP and a 5.3-fold increase in CAP350 protein levels in Z-L3VS–treated cells versus DMSO-treated control cells. Furthermore, we analyzed individual centriole length upon transfection of siRNA duplexes targeting either FOP or CAP350, followed by Z-L3VS treatment, as we had previously done (Figure 1, B and C). We found that the maximum length of individual daughter centrioles was overall reduced compared with controls although the length of individual abnormally elongated daughter centrioles still varied widely (data not shown). Together, these results suggest that protein stabilization of both FOP and CAP350 through proteasome inhibition contributes to abnormal elongation of daughter centrioles.

We next examined the localization of CPAP, a known microtubule-interacting protein that has recently been implicated in centriole length control, after treatment with proteasome inhibitors (Figure 5). In DMSO-treated control cells, we saw CPAP localizing to both mother and daughter centrioles (Figure 5, top), as seen previously (Kohlmaier ; Schmidt ; Tang ). When we treated cells with Z-L3VS, CPAP colocalized with the centrin signal of elongated daughter centrioles (Figure 5, middle). However, when cells were treated with the proteasome inhibitor lactacystin at a concentration that did not lead to centriole elongation, CPAP localization was analogous to DMSO-treated control cells (Figure 5, bottom). This suggests that treatment with the proteasome inhibitor Z-L3VS leads to an accumulation of CPAP at elongated daughter centrioles.

Figure 5.

Alteration of CPAP localization after Z-L3VS treatment. Immunofluorescence microscopic analysis of CPAP in U-2 OS/centrin-GFP cells after treatment with 1 μM Z-L3VS for 72 h or 0.1% DMSO as control. Nuclei stained with DAPI. Bar, 10 μm.

Although depletion of C-Nap1, hSAS-6, Cep170, and ninein decreased Z-L3VS–mediated abnormal daughter centriole elongation in our siRNA screen, the localization of these proteins did not change in cells treated with any of our proteasome inhibitors (Figures 2 and 6). This suggests that, although treatment with Z-L3VS does not promote a significant change in phenotype of these proteins, the presence of C-Nap1, hSAS-6, Cep170, and ninein is necessary to promote Z-L3VS–mediated abnormal daughter centriole elongation.

Figure 6.

Expression C-Nap1 and hSAS-6 after Z-L3VS treatment. (A and B) Immunofluorescence microscopic analysis for C-Nap1 (A) or hSAS-6 (B) in U-2 OS/centrin-GFP cells after a 72-h treatment with 1 μM the proteasome inhibitor Z-L3VS or lactacystin (both 1 μM). Treatment with 0.1% DMSO was used as control. Nuclei stained with DAPI. Bar, 10 μm.

We then determined the localization of CP110 and Cep97, two proteins whose depletion we found to increase the percentage of Z-L3VS–treated cells that contained long daughter centrioles. Immunofluorescence microscopic analysis of CP110 and Cep97 in DMSO-treated control cells showed centriolar localization of both proteins (Figure 7, A and B, top) in accordance with previously published findings (Chen ; Kleylein-Sohn ; Spektor ; Schmidt ). When cells were treated with Z-L3VS to induce abnormally elongated daughter centrioles, both CP110 and Cep97 localized to the tips of elongated daughters (Figure 7, A and B, middle). Treatment of cells with lactacystin at concentrations that did not induce elongated daughter centrioles, resulted in a similar CP110 and Cep97 localization pattern as detected DMSO-treated controls (Figure 7, A and B, bottom).

Figure 7.

CP110 and Cep97 are retained at the tips of elongated daughter centrioles induced by Z-L3VS. (A and B) Immunofluorescence microscopic analysis of CP110 (A) and Cep97 (B) in U-2 OS/centrin-GFP cells after a 72-h treatment with 1 μM of the proteasome inhibitors Z-L3VS or lactacystin (all 1 μM). Treatment with 0.1% DMSO was used as control. Note the localization of both CP110 and Cep97 to the distal ends of elongated daughter centrioles. Nuclei stained with DAPI. Bar, 10 μm.

Together, Z-L3VS–induced abnormal daughter centriole elongation was associated with prominent changes in FOP, CAP350, and CPAP protein abundance and changes in localization at long daughter centrioles. Not all proteins identified in our siRNA screen followed this pattern, although most proteins showed an accumulation at centrioles by immunofluorescence microscopy (C-Nap1, hSAS-6, CEP170, ninein, and Cep97; data not shown). Clearly, Z-L3VS–induced abnormal centriole elongation was not due to displacement of CP110 and Cep97 from the distal tips of daughter centrioles (Figure 7). However, the lack of increased CP110 protein levels in Z-L3VS–compared with DMSO-treated cells, as measured by immunofluorescence analysis, is in agreement with the role of CP110 as a negative regulator of centriole elongation (Kohlmaier ; Schmidt ; Tang ). Together, these results suggest that some proteins may play a direct role in aberrant microtubule elongation along the length of daughter centrioles, whereas others may not function directly at daughter centrioles but are yet required for this process.

DISCUSSION

This report presents the idea that daughter centriole elongation is not restrained by structural constraints but is in fact regulated by proteolysis. Until recently, it was unknown how centriole length was regulated, one possibility being that daughter centrioles were prevented from elongating to longer lengths than the maternal centriole due to the inability of centriolar structural components to stably form a longer structure. Here, together with several other recent reports, we show that this is not the case (Vidwans ; Kohlmaier ; Schmidt ; Tang ). We demonstrate that by inhibiting cellular proteolytic processes, daughter centrioles are capable of elongating to much longer lengths than previously thought, even in the presence of a maternal centriole of normal length.

Because we only detected abnormally long daughter centrioles under our assay conditions, our findings suggest that mother and daughter centriole elongation are, to a certain degree, separable processes and can be distinguished by proteolytic inhibition. The basis for this differential regulation is unknown, but it is possible that cell cycle-dependent events play a role. Daughter centriole elongation predominantly occurs during S and G2 phases of the cell division cycle. Hence, the treatment with proteasome inhibitors Z-L3VS or MG132 may render cells competent for aberrant daughter centriole elongation through prolongation of a cell cycle stage that is permissive for this process. In line with this notion is our previous finding that Z-L3VS–treated cells accumulate in G2 (Duensing ). However, a cell cycle arrest per se is not sufficient to induce abnormal daughter centriole elongation because cells treated with epoxomicin contained a significant proportion of cells that were arrested in G2 phase of the cell division cycle (data not shown), but they lacked the ability to produce abnormally long daughter centrioles (Figure 1B).

When we first tested different proteasome inhibitors for their ability to induce abnormal centriole elongation, we used a drug concentration that resulted in a cell viability of at least 50%. We repeated these experiments with higher concentrations of proteasome inhibitors, and we found that, in principal, all proteasome inhibitors used were able to induce abnormally long centrioles, albeit associated with reduced cell viability and in an overall lower percentage of cells that survived the treatment. Remarkably, abnormally elongated centrioles were virtually absent in DMSO-treated control cells.

All proteasome inhibitors tested here have been shown to inhibit 26S proteasome activity. However, several of our inhibitors, most notably Z-L3VS and MG132, also have nonproteasomal activities against cellular proteases. In contrast, other inhibitors such as lactacystin and epoxomicin are more potent and selective inhibitors of the 26S proteasome with less nonproteasomal inhibitory activities. In fact, our results support the notion that protease inhibition may be more relevant for the observed phenotype because some of the most potent proteasome inhibitors used lacked the ability to induce centriole elongation and at the same time failed to prevent the cleavage of the protease substrate casein (Figure 1E).

To determine what proteins were involved in abnormal daughter centriole elongation induced by Z-L3VS, we used an siRNA screen based on a previous study, in which 127 centrosomally associated proteins were identified (Andersen ). We did not see a complete abolishment of abnormal centriole length in our siRNA screen and believe this is explained by the discrepancy between the rapid effects of proteasome inhibition on centriole biogenesis and the slower kinetics of maximum siRNA efficacy. Z-L3VS–associated alterations of centriole biogenesis become visible already at 6 h after treatment and affect ∼25% of cells after 24 h of treatment (our unpublished data). It is generally accepted that siRNA-mediated knockdown of protein expression usually results in maximum protein depletion at ∼48–72 h posttransfection. The residual extra long daughter centrioles that we observed are likely to have formed before maximum knockdown of the targeted proteins.

In addition, the abnormal centriole elongation phenotype reported here mostly arises in the context of centriole overduplication, specifically centriole multiplication, during which a single maternal centriole nucleates the concurrent formation of multiple daughter centrioles (Duensing ). These supernumerary centrioles were very stable and persisted for prolonged time intervals (Duensing ). In line with this notion, we found that Z-L3VS–induced abnormally elongated centrioles persisted in a significant proportion of cells for up to 72 h after removal of the drug (data not shown). It is hence possible that abnormally elongated daughter centrioles that form in response to proteasome inhibition are particularly stable and less dynamic, a possibility that is currently under investigation.

Nonetheless, results from this siRNA screen confirm and extend several previous studies. Recently, overexpression of CPAP was shown to induce the abnormal elongation of both mother and daughter centrioles (Kohlmaier ; Schmidt ; Tang ). CPAP protein level was shown to be cell cycle regulated through the action of the anaphase-promoting complex/cyclosome and the 26S proteasome (Tang ). Our screen confirms this role of CPAP in centriole elongation by showing that CPAP depletion reduces the number of cells that exhibit centriole elongation following Z-L3VS treatment.

CPAP overexpression was found to synergize with depletion of another centriole component, CP110, in the formation of extra long centrioles (Kohlmaier ; Tang ). CP110 is a distal-end capping protein, and knockdown of this protein alone can lead to centriole elongation and formation of primary cilia (Spektor ). These results have suggested a simple model in which CPAP and CP110 act as positive and negative regulators, respectively, of centriole elongation. In line with this model, we found that depletion of either CP110, or a CP110 interacting partner Cep97, enhanced Z-L3VS–mediated daughter centriole elongation.

However, our results significantly extend these previous studies by showing that many additional proteins besides CPAP and CP110 are involved in centriole length control. We found that FOP and CAP350 were also necessary for abnormal daughter centriole elongation, and significantly accumulate in Z-L3VS– or MG132–treated cells. FOP and CAP350 have been shown to interact and form a complex at the centrosome, with CAP350 necessary for FOP localization (Yan ). The FOP–CAP350 complex was shown to play a role in microtubule-anchoring at the centrosome. In addition, we show that hSAS-6, a structural centriolar component that is essential for daughter centriole synthesis (Kleylein-Sohn ; Strnad ), is also necessary for abnormal centriolar elongation induced by Z-L3VS. Moreover, C-Nap1 a protein known to play a role in maintaining centrosome cohesion, as well as Cep170 and ninein, both proteins known to associate with maternal centriolar appendages, also were found to play a role in centriole length control. Depletion of Cep170 may interfere with microtubule-dependent trafficking to the centrosome (Guarguaglini ), whereas depletion of ninein may result in the loss of microtubule anchoring at the centrosome (Delgehyr ).

Our results thus suggest that centrosome cohesion and microtubule-dependent processes may indirectly contribute to abnormal centriole elongation induced by Z-L3VS and MG132 treatment. Support for our notion comes from previous work suggesting that both centriolar and pericentriolar components are necessary for centriole assembly, possibly by concentrating the recruitment of components necessary for daughter centriole synthesis around the maternal centriole (Dammermann ; Loncarek ).

In conclusion, we provide evidence that daughter centriole elongation is regulated by proteolysis. Our results suggest that maintaining the balance between positive- and negative-regulatory proteins is key for daughter centriole elongation control. We show that several additional proteins besides CPAP and CP110 are involved in this process, including proteins known to play a role in centrosome cohesion and microtubule anchoring. This report illustrates the complex circuitry of centriolar proteins that regulate centriole biogenesis, including daughter centriole length control.