Orientation of the cell division axis offers a critical mechanism

CYLD regulates spindle orientation by stabilizing astral microtubules and promoting dishevelled-numa-dynein/ dynactin complex formation Yunfan Yang a,1, Min Liu a,1, Dengwen Li a,1, Jie Ran a, Jinmin Gao a, Shaojun Suo a, Shao-Cong Sun b, and Jun Zhou a,2 a State Key Laboratory of Medicinal Chemical Biology, College of Life Sciences, Nankai University, Tianjin 300071, China; and b Department of Immunology, The University of Texas MD Anderson Cancer Center, Houston, TX 77030 Edited* by Vishva M. Dixit, Genentech, San Francisco, CA, and approved January 3, 2014 (received for review October 12, 2013) Oriented cell division is critical for cell fate specification, tissue organization, and tissue homeostasis, and relies on proper orientation of the mitotic spindle. The molecular mechanisms underlying the regulation of spindle orientation remain largely unknown. Herein, we identify a critical role for cylindromatosis (CYLD), a deubiquitinase and regulator of microtubule dynamics, in the control of spindle orientation. CYLD is highly expressed in mitosis and promotes spindle orientation by stabilizing astral microtubules and deubiquitinating the cortical polarity protein dishevelled. The deubiquitination of dishevelled enhances its interaction with nuclear mitotic apparatus, stimulating the cortical localization of nuclear mitotic apparatus and the dynein/dynactin motor complex, a requirement for generating pulling forces on astral microtubules. These findings uncover CYLD as an important player in the orientation of the mitotic spindle and cell division and have important implications in health and disease. cell cortex cell cycle ubiquitin 3D cell culture knockout mouse Orientation of the cell division axis offers a critical mechanism for the control of cell type choices and the specification of tissue/organ architecture; this is achieved through accurate orientation of the mitotic spindle relative to the cell cortex (1). Spindle orientation is exquisitely regulated during development as well as in adult life, and defects in this process may have severe consequences, such as developmental disorders and tumor formation (1, 2). A dividing cell can orient its spindle along the planar axis or the apicobasal axis of the tissue, depending on the tissue environment and cell geometry. In most epithelia, such as the intestine crypt epithelium, planar spindle orientation is common to produce two daughter cells side by side. By contrast, apicobasal spindle orientation is frequently associated with asymmetric cell divisions, which result in two daughter cells of distinct identities (2). Astral microtubules play a key role in spindle orientation by linking the spindle to the cell cortex (3). The localization of cell polarity proteins such as dishevelled (Dvl) at the cell cortex is also important for spindle orientation by transmission of extrinsic signals or providing the intrinsic cues. Cortical polarity proteins can recruit the nuclear mitotic apparatus (NuMA) protein and then the microtubule minus end-directed dynein/ dynactin motor complex, which can generate pulling forces on astral microtubules to rotate the spindle (3). Therefore, the dynamic interaction of astral microtubules with the cell cortex via diverse protein complexes constitutes an essential part of the mechanism for spindle orientation. However, it remains elusive how the protein complexes controlling spindle orientation are assembled and activated to make a connection between astral microtubules and the cell cortex. As a posttranslational modification, protein ubiquitination is critical for diverse cellular and biological events, and it is a reversal process mediated by E3 ubiquitin ligases and deubiquitinases, respectively (4, 5). E3 ubiquitin ligases von Hippel Lindau (VHL) and parkin have recently been demonstrated to participate in the control of spindle orientation (6, 7), suggesting that protein ubiquitination may regulate the formation of the spindle orientation machinery. In this study, we provide the first evidence that cylindromatosis (CYLD), a deubiquitinase specifically removing lysine 63 (K63)-linked polyubiquitin chains and a regulator of microtubule dynamics (8, 9), stabilizes astral microtubules and stimulates the formation of the Dvl-NuMA-dynein/dynactin complex at the cell cortex, thereby promoting proper spindle orientation. Results CYLD Is Highly Expressed in Mitosis and Is Important for Oriented Cell Division. In this study, we analyzed the role of CYLD mainly using HeLa cells, which have been widely used to investigate various aspects of cell division, including spindle orientation (6, 10). Consistent with previous findings (11), immunoblotting of lysates from synchronous HeLa cells revealed that CYLD was highly expressed in mitosis (Fig. 1A). To further examine the role of this protein during mitotic progression, we inhibited its expression by using two sirnas, one targeting the coding sequence and another targeting the 3 UTR (Fig. 1B). Time-lapse microscopy of HeLa cells stably expressing YFP-tagged histone 2B (referred to as HeLa-H2B) showed that CYLD depletion delayed mitotic progression due to a prolongation of metaphase (Fig. 1 C and D). In addition, CYLD-depleted cells displayed uneven timing of daughter cell adhesion to the substratum, indicative of misoriented cell division (Fig. 1 C and E). Significance Orientation of the mitotic spindle relative to the cell cortex is known to control the orientation of the cell division plane, thereby contributing to cell fate specification and tissue organization. The molecular mechanisms of how spindle orientation is regulated during mitosis remain poorly defined. In this paper, we demonstrate that cylindromatosis (CYLD) regulates spindle orientation via its dual functions as a microtubuleassociated protein and deubiquitinase. CYLD stabilizes astral microtubules, hence ensuring microtubule extension to the cell cortex and interaction with cortical sites. The deubiquitinase activity of CYLD, however, catalyzes the removal of the polyubiquitin chain from dishevelled and thereby promotes the dishevelled-numa-dynein/dynactin complex formation at the cell cortex, a requirement for generating pulling forces on astral microtubules. Author contributions: Y.Y. and J.Z. designed research; Y.Y., M.L., D.L., J.R., J.G., and S.S. performed research; S.-C.S. contributed new reagents/analytic tools; Y.Y. analyzed data; and J.Z. wrote the paper. The authors declare no conflict of interest. *This Direct Submission article had a prearranged editor. 1 Y.Y., M.L., and D.L. contributed equally to this work. 2 To whom correspondence should be addressed. E-mail: junzhou@nankai.edu.cn. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1319341111/-/DCSupplemental. 2158 2163 PNAS February 11, 2014 vol. 111 no. 6 www.pnas.org/cgi/doi/10.1073/pnas.1319341111

Fig. 1. CYLD is highly expressed in mitosis and is important for oriented cell division. (A) Immunoblots showing the levels of CYLD, phosphorylated histone H3 (p-h3), and β-actin in HeLa cells synchronized at different phases. (B) Immunoblots for CYLD and β-actin expression in control and CYLD sirna-treated HeLa-H2B cells. (C) Time-lapse images showing prolonged metaphase and misoriented cell division (uneven timing of daughter-cell adhesion to the substratum) in CYLD sirna-treated HeLa-H2B cells, compared with control. Dashed lines indicate misoriented cell divisions. (Scale bars, 10 μm.) (D) Duration of mitotic phases in cells treated as in C (n = 10 mitotic cells per group). (E) Quantification of normal and misoriented cell divisions in cells treated as in C (n = 12 mitotic cells per group). Student t test for all graphs. *P < 0.05; ns, not significant. Error bars indicate SEM. CYLD Depletion Leads to Spindle Misorientation. We then investigated whether the loss of CYLD disrupts spindle orientation, by measuring the angle between the spindle axis and the substratum (Fig. 2A). As shown by time-lapse microscopy, the spindle angles in CYLD-depleted HeLa cells underwent more dramatic changes before anaphase onset, compared with control CELL BIOLOGY Fig. 2. CYLD depletion leads to spindle misorientation. (A) Scheme depicting spindle angle (α) measurement. (B) Immunoblots for CYLD and β-actin expression in control and CYLD sirna-treated HeLa cells. (C and D) Time-lapse images (C, zx projection) and spindle angles (D) in mitotic HeLa cells transfected with DsRed-histone 2B (red), GFP α-tubulin (green), and control or CYLD sirnas. Anaphase onset was set at 0 min. (Scale bars, 5 μm.) (E) Immunofluorescence images (E I ), spindle angle distribution (E II ), average spindle angle (E III ), and average spindle diameter (E IV ) of control and CYLD sirna-treated metaphase HeLa cells stained with anti α-tubulin (green) and anti γ-tubulin (red) antibodies and DAPI (blue). The position of the z stage is indicated in micrometers; 3D, xy projection (n = 70 cells per group). (Scale bars, 5 μm.) (F) Immunofluorescence images (F I Left, z sections; F I Right, zx projection) and average spindle angle (F II ) of metaphase HeLa cells transfected with CYLD sirna and GFP, GFP-CYLD, GFP-CYLD-C/S, or GFP-CYLD-ΔCG1/2 (green), followed by staining with anti γ-tubulin antibody (red) and DAPI (blue). C/S, mutation of cysteine 601 to serine; ΔCG1/2, without the two amino-terminal CAP-Gly domains (n = 40 cells per group). (Scale bars, 5 μm.) Student t test for all graphs. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. Error bars indicate SEM. Yang et al. PNAS February 11, 2014 vol. 111 no. 6 2159

(Fig. 2 B D). Immunofluorescence microscopy of fixed cells showed that CYLD depletion greatly broadened the distribution of the spindle angles (Fig. 2 E I and E II ). The average spindle angle in CYLD sirna-treated cells was above 20, whereas it was less than 10 in control cells (Fig. 2E III ). The loss of CYLD did not obviously affect the gross morphology or length of the mitotic spindle (Fig. 2 E I and E IV ). Similar results were obtained in shrna-mediated stable CYLD-knockdown HeLa cells (referred as HeLa-shCYLD; Fig. S1). To examine the selectivity of CYLD in spindle orientation, we analyzed whether other deubiquitinases, such as A20 and USP37, are involved in the above process. Like CYLD, A20 is able to cleave K63-linked polyubiquitin chains and plays an important role in the down-regulation of NF-κB signaling (9). sirna-mediated knockdown of A20 expression did not obviously affect spindle orientation (Fig. S2). In addition, the depletion of USP37, which has been implicated in cell cycle control due to its activity to regulate degradative ubiquitination (12), did not affect spindle orientation (Fig. S2). Collectively, the above data demonstrate an important and unique role for CYLD in the control of spindle orientation. To study the mechanisms underlying the regulation of spindle orientation by CYLD, HeLa cells were transfected with CYLD sirna and CYLD-expressing plasmids. Wild-type CYLD largely rescued CYLD sirna-induced spindle misorientation, as evidenced by the decrease of the spindle angle to a normal level (Fig. 2F). By contrast, only partial rescue was observed for the ΔCG1/2 mutant, which lacks the two amino-terminal cytoskeleton-associated protein glycine-rich (CAP-Gly) domains that mediate the interaction of CYLD with microtubules (8), and the catalytically inactive C/S mutant, in which cysteine 601 in the deubiquitinase domain is substituted by serine (13) (Fig. 2F). Thus, the microtubule-binding property and the deubiquitinase activity are both required for CYLD to regulate spindle orientation. CYLD Promotes the Stability of Astral Microtubules. The above findings prompted us to examine the effect of CYLD depletion on astral microtubules, which are essential for spindle orientation (3). Measurement of the relative astral microtubule fluorescence intensity revealed that the loss of CYLD resulted in a remarkable reduction of astral microtubules, and that this effect was rescued by the microtubule-stabilizing agent paclitaxel (Fig. 3 A C). By using purified proteins, we further found that the length of microtubules polymerized in vitro was increased by wild-type CYLD and the C/S mutant, but not the ΔCG1/2 mutant (Fig. 3 D F). CYLD and its mutants did not increase the number of microtubules (Fig. 3 E and G), implying that the microtubule nucleation process is not affected. We also found that wild-type CYLD and the C/S mutant, but not the ΔCG1/2 mutant, rendered microtubules resistant to cold-induced depolymerization in the purified system (Fig. 3 H and I). Moreover, CYLD sirnainduced reduction of astral microtubules was rescued by wild-type CYLD and the C/S mutant, but not the ΔCG1/2 mutant (Fig. 3J). These results reveal that CYLD promotes the stability of astral microtubules. Loss of CYLD Impairs the Localization of the Dynein/Dynactin Motor Complex at the Cell Cortex. To provide more mechanistic insight into the regulation of spindle orientation by CYLD, we analyzed the localization of the dynein/dynactin motor complex at the cell cortex, which is crucial for the production of pulling forces on astral microtubules to rotate the spindle (14). Immunofluorescence microscopy revealed that CYLD depletion in HeLa cells impaired the cortical localization of the dynactin component p150 Glued and dynein intermediate chain (DIC) 1 and 2 (Fig. 4 A and B). The loss of CYLD resulted in a more evident reduction in the cortical localization of NuMA (Fig. 4 C E), which is necessary for recruiting the dynein/dynactin complex to the cell cortex in response to cortical cues for spindle orientation (15). The decrease of NuMA at the cell cortex was also Fig. 3. CYLD promotes the stability of astral microtubules. (A) Immunostaining with anti α-tubulin antibody in HeLa-shControl and HeLa-shCYLD cells treated with control (DMSO) or 0.5 μm paclitaxel for 20 min. (Scale bars, 5 μm.) (B) Scheme depicting the method for measuring relative astral microtubule fluorescence. (C) Quantification of relative astral microtubule fluorescence in cells treated as in A (n = 30 cells per group). (D) Coomassie staining of GST and GST-CYLD fusion proteins purified from 293T cells with glutathione resin. (E G) Images (E), average length (F, n = 200 microtubules per group), and average number per field (G, n = 10 fields per group) of microtubules incorporated with rhodamine-labeled tubulin, polymerized in the presence of GST or GST-CYLD fusion proteins. (Scale bars, 5 μm.) (H and I) Images (H) and average length (I) of microtubules polymerized as in E and placed on ice for 30 min. (Scale bars, 5 μm.) (J) Immunofluorescence images (J I ) and quantification (J II ) of relative astral microtubule fluorescence in metaphase HeLa cells transfected with CYLD sirna and GFP or GFP-CYLD, GFP-CYLD-C/S, or GFP-CYLD-ΔCG1/2 (green), followed by staining with anti α-tubulin antibody (red) and DAPI (blue) (n = 25 cells per group). (Scale bars, 5 μm.) Student t test for all graphs. ***P < 0.001; ns, not significant. Error bars indicate SEM. 2160 www.pnas.org/cgi/doi/10.1073/pnas.1319341111 Yang et al.

observed in CYLD shrna-transfected HaCaT keratinocytes (Fig. 4 F and G). CYLD Stimulates the Formation of the Dvl-NuMA-Dynein/Dynactin Complex Through Deubiquitinating Dvl. NuMA localization at the cell cortex is known to involve its interaction with multiple proteins important for spindle orientation, including the family of Dvl proteins (15 17). In addition, hyperubiquitination of Dvl has been observed in CYLD-deficient cells to mediate enhanced Wnt/β-catenin signaling (18). We thus investigated whether CYLD modulates the interaction between NuMA and Dvl via deubiquitinating Dvl. Immunoprecipitation revealed that depletion of CYLD reduced the interaction of NuMA with Dvl1, Dvl2, and Dvl3 in HeLa cells (Fig. 5A). The interaction of NuMA with Dvl3, but not DIC1/2, was also reduced in HeLa-shCYLD cells, compared with control (Fig. 5B). In addition, CYLD sirnainduced decrease of the NuMA-Dvl3 interaction was rescued by wild-type CYLD and CYLD-ΔCG1/2, but poorly by CYLD-C/S (Fig. 5C). Using a mutant form of ubiquitin that contains only a single lysine (K63), with all of the other lysines mutated to arginine, we found that depletion of CYLD enhanced K63-linked ubiquitination of Dvl3 in both interphase and mitotic cells (Fig. 5D). To provide additional insight into how CYLD regulates the NuMA- Dvl interaction, we mutated all of the seven lysines in the DIX domain of Dvl3, which are involved in Dvl ubiquitination (18), to arginines (7KR). The loss of CYLD significantly increased Dvl3 ubiquitination in cells transfected with wild-type Dvl3, but not Dvl3-7KR (Fig. 5E). The 7KR mutation remarkably enhanced the interaction of Dvl3 with NuMA (Fig. 5F). In addition, although CYLD depletion dramatically inhibited the interaction of NuMA with wild-type Dvl3, it did not affect the interaction of NuMA with Dvl3-7KR (Fig. 5G). These data thus demonstrate that CYLD-mediated deubiquitination of Dvl3 promotes its interaction with NuMA. By immunofluorescence microscopy, we further found that CYLD sirna-induced decrease of NuMA and p150 Glued localization at the cell cortex was remarkably rescued by wild-type CYLD and CYLD-ΔCG1/2, but not CYLD-C/S (Fig. 5H). We also observed that Dvl3-7KR, but not wild-type Dvl3, could significantly rescue CYLD sirna-induced decrease of NuMA and p150 Glued localization at the cell cortex (Fig. 5H). In addition, CYLD sirna-induced increase of the spindle angle was rescued significantly by wild-type CYLD and Dvl3-7KR, but modestly by CYLD-C/S and CYLD-ΔCG1/2 and poorly by wildtype Dvl3 (Figs. 2F and 5I). Collectively, these results suggest that CYLD stimulates the formation of the Dvl-NuMA-dynein/ dynactin complex at the cell cortex through deubiquitinating Dvl and subsequently promotes spindle orientation. In an attempt to identify the enzyme promoting K63-linked polyubiquitination of Dvl3, we tested the effects of selected E3 ubiquitin ligases, including tumor necrosis factor receptor-associated factor 2 (Traf2), Traf3, Traf5, Traf6, VHL, parkin, Trim21, WW domain-containing protein 2 (WWP2), and atrophin-1 interacting protein 4 (AIP4). Among these E3 ubiquitin ligases, only AIP4 significantly enhanced Dvl3 ubiquitination (Fig. S3). Using the ubiquitin-k63 mutant, we further found that AIP4 dramatically promoted K63-linked ubiquitination of Dvl3 (Fig. 5 J and K). CYLD Is Required for Spindle Orientation in 3D Cell Culture and Mice. To investigate the physiological relevance of these findings, we examined whether the loss of CYLD affects spindle orientation in the 3D culture system of Caco-2 cells, an in vitro model of intestinal epithelium. As shown in Fig. S4, in the control group, CELL BIOLOGY Fig. 4. Loss of CYLD impairs the localization of the dynein/dynactin motor complex at the cell cortex. (A) Immunostaining with anti-p150 Glued and anti-dic1/2 antibodies in HeLa-shControl and HeLa-shCYLD cells. (Scale bars, 5 μm.) (B) Quantification of the cortical localization of p150 Glued and DIC1/2 in cells treated as in A (n = 20 cells per group). (C) Immunostaining with anti-numa and anti α-tubulin antibodies and DAPI in HeLa-shControl and HeLa-shCYLD cells. (Scale bars, 5 μm.) (D) Enlargements (Upper) and intensity plots (Lower) of the areas outlined by rectangles in C. (E) Quantification of the cortical localization of NuMA in cells treated as in C (n = 20 cells per group). (F) Immunostaining with anti-numa antibody and DAPI in metaphase HaCaT cells transfected with DsRed-histone H2B and CYLD shrnas or control vector. (Scale bars, 5 μm.) (G) Quantification of the cortical localization of NuMA in cells treated as in F (n = 20 cells per group). Student t test for all graphs. *P < 0.05, **P < 0.01, ***P < 0.001. Error bars indicate SEM. Yang et al. PNAS February 11, 2014 vol. 111 no. 6 2161

Fig. 5. CYLD stimulates the formation of the Dvl-NuMA-dynein/dynactin complex through deubiquitinating Dvl. (A) Immunoprecipitation (IP) and immunoblotting (IB) showing that CYLD sirna treatment compromises the interactions of the Dvl family of proteins with NuMA in HeLa cells. (B) Immunoblots showing that the loss of CYLD inhibits NuMA interaction with Dvl3, but not with DIC1/2, in HeLa-shControl and HeLa-shCYLD cells. (C) Immunoblots showing that transfection of GFP-CYLD or its ΔCG1/2 mutant, but not its C/S mutant, rescues the NuMA-Dvl3 interaction in CYLD-depleted 293T cells. (D) Examinationof K63- linked ubiquitination of Dvl in 293T cells transfected with control or CYLD sirnas together with HA-Dvl3 and His-Myc-ubiquitin-K63 and synchronized ininterphase or mitosis. The ubiquitin-k63 mutant contains only a single lysine (K63), with all of the other lysines mutated to arginine. (E) Examination of Dvl ubiquitinationin 293Tcells transfected with control orcyldsirnas, together with His-Myc-ubiquitin and HA-Dvl3 orha-dvl3-7kr. 7KR, mutation oflysines 5, 20, 34, 43, 47, 57, and 66 to arginines. (F) Comparison of the interaction of GFP-NuMA with HA-Dvl3 and HA-Dvl3-7KR in 293T cells. (G) Comparison of the interaction of GFP-NuMA with HA-Dvl3 and HA-Dvl3-7KR in 293T cells treated with control or CYLD sirnas. (H I III ) Immunofluorescence images (H I ) and quantification of the cortical localization of NuMA (H II ) and p150 Glued (H III ) of metaphase HeLa cells transfected with CYLD sirna and plasmids expressing the indicated proteins, followed by staining with anti-numa or anti-p150 Glued antibodies and DAPI. Cells transfected with HA-Dvl3 and HA-Dvl3-7KR are costained with anti-ha antibody (n = 20 cells per group). (Scale bars, 5 μm.) (I) Average spindle angle in metaphase HeLa cells transfected with CYLD sirna and plasmids expressing the indicated proteins (n = 20 cells per group). (J and K) Examination(J) and quantification (K) of K63-linked ubiquitination of Dvl3 in 293T cells transfected with Flag or Flag- AIP4, together with HA-Dvl3 and His-Myc-ubiquitin-K63. Student t test for all graphs. *P < 0.05, ***P < 0.001; ns, not significant. Error bars indicate SEM. dividing Caco-2 cells oriented their spindles largely parallel to the apical and basal surfaces of the cyst, with spindle angles relative to the surface less than 30. By contrast, cysts formed from CYLDdepleted Caco-2 cells displayed spindle misorientation, with a broader distribution of spindle angles. To corroborate the role of CYLD in the control of spindle orientation in physiological settings, we examined intestinal crypt tissues of CYLD +/+ and CYLD / mice with immunofluorescence microscopy. In CYLD +/+ mice, intestinal epithelial cells divided with spindle angles relative to the apical and basal surfaces less than 30 (Fig. S4). By contrast, the loss of CYLD resulted in a clear increase in the percentage of cells displaying spindle angles above 30 (Fig. S4). We also examined epidermal tissues of newborn CYLD +/+ and CYLD / mice. In CYLD +/+ mice, the majority of epidermal basal cells divided largely perpendicularly to the basal lamina, whereas the depletion of CYLD significantly increased the proportion of basal cells with spindle angles relative to the basal lamina less than 60 (Fig. S4). These data demonstrate that CYLD is required for tissue organization by ensuring oriented cell division. Discussion Our data, especially those obtained by rescue experiments with the CYLD-ΔCG1/2 and CYLD-C/S mutants (Figs. 2F, 3J, and 5 C and H), establish a crucial role for CYLD in the control of spindle orientation via its dual functions as a microtubule-associated protein and deubiquitinase (Fig. 6). The stabilization of astral microtubules by CYLD ensures microtubule extension to the cell cortex and hence interaction with cortical sites, a prerequisite for orientation of the spindle. The deubiquitinase activity of CYLD, however, catalyzes the removal of the polyubiquitin chain from Dvl and thereby increases the interaction of Dvl with NuMA, promoting the cortical localization of NuMA and the dynein/dynactin motor complex (Fig. 6). These bifunctional effects of CYLD, together with the previously reported actions of proteins such as adenomatous polyposis coli, VHL, huntingtin, and intraflagellar transport protein 88 in regulating spindle orientation (6, 10, 19, 20), indicate an exquisite interplay between astral microtubule stabilization and the assembly of cortical polarity protein complexes in fine-tuning of spindle orientation. Previous studies have demonstrated E3 ubiquitin ligases VHL and parkin as important regulators of spindle orientation (6, 7). The present study identifies a critical role for the deubiquitinase CYLD in this process. Together, these findings suggest that reversal ubiquitination may offer a regulatory mechanism for the formation of various protein complexes mediating spindle orientation. It is worthy of note that although our data establish Dvl as a substrate of CYLD in the control of spindle orientation, it would not be surprising if other cortical proteins were identified 2162 www.pnas.org/cgi/doi/10.1073/pnas.1319341111 Yang et al.

synergize with other tumor-associated alterations to cause epithelial tissue disorganization and genomic instability, thereby stimulating tumor development and progression. Fig. 6. Molecular model for CYLD function in spindle orientation. in the future fulfilling this function, given that the recruitment of NuMA to the cell cortex could be mediated by a number of different mechanisms (15). Our preliminary study of lateral geniculate nucleus and resistance to inhibitors of cholinesterase 8 homolog A, two polarity proteins known to associate with NuMA (15), reveals that they are also modified by ubiquitination; however, the depletion of CYLD does not significantly change their ubiquitination level (Fig. S5). It will be interesting to investigate in the future whether other NuMA-associated proteins are involved in the action of CYLD in regulating spindle orientation. Our findings may have implications in tumor development and progression. CYLD is well known as the tumor suppressor protein mutated in familial cylindromatosis and multiple familial trichoepithelioma, genetic conditions associated with the development of skin-appendage tumors (21). In addition, the loss of CYLD has been implicated in several other malignancies, such as colon and hepatocellular carcinomas, multiple myeloma, and melanoma (21). It has been proposed that CYLD deficiency may instigate tumor growth by increasing cell proliferation due to inappropriate activation of signaling pathways such as NF-κB and Wnt/β-catenin (9, 18). In this study, our data show that the loss of CYLD leads to misoriented cell division in epithelial cells. Interestingly, CYLD-deficient mice do not spontaneously develop tumors, suggesting that spindle misorientation alone is unlikely to be tumorigenic. However, CYLD-deficient mice are more susceptible to chemically induced colon and skin tumors than wild-type mice (22, 23). It is therefore tempting to speculate that spindle misorientation due to the disruption of CYLD may Materials and Methods Fluorescence Microscopy. Cells were fixed with 4% (wt/vol) paraformaldehyde/ PBS for 30 min followed by permeabilization in 0.5% Triton X-100/PBS for 20 min, or fixed with methanol at 20 C for 5 min for experiments involving the visualization of microtubules. Cells were blocked and incubated with primary antibodies and then rhodamine- or fluorescein-conjugated secondary antibodies followed by staining with DAPI. Cell cysts were fixed with acetone/ methanol at 20 C for 5 min and incubated with antibodies or rhodamine phalloidin followed by staining with DAPI. Mouse tissues were fixed in 4% (wt/vol) paraformaldehyde/pbs, embedded in Tissue-Tek OCT (Sakura), and snap-frozen in liquid nitrogen. Sections were then stained with antibodies or fluorescein phalloidin and subsequently with DAPI. For time-lapse microscopy, cells were cultured in a 37 C chamber, and mitotic progression was recorded. Immunoblotting and Immunoprecipitation. Proteins were resolved by SDS/PAGE and transferred onto polyvinylidene difluoride membranes (Millipore). The membranes were blocked and incubated with primary antibodies and then with horseradish peroxidase-conjugated secondary antibodies. Specific proteins were visualized with enhanced chemiluminescence detection reagent (Thermo Fisher Scientific). For immunoprecipitation, cell lysates were incubated with antibody-coated agarose beads at 4 C for 2 h. The beads were washed and boiled in the SDS loading buffer, and the proteins were detected by immunoblotting. Microtubule Assembly and Stability Assays. Microtubule assembly assay was performed using 5 mg/ml microtubule-associated protein-free tubulin spiked with 10% (wt/wt) rhodamine tubulin (Cytoskeleton) and 20 μm purified GST or GST-CYLD proteins. GTP (1 mm) was then added and the mixture was incubated at 37 C for 20 min to allow microtubule polymerization. To analyze microtubule stability, microtubules assembled as described above were placed on ice for 30 min before examination. Statistics. Analysis of statistical significance was performed by the Student t test capability in Microsoft Excel. ACKNOWLEDGMENTS. We thank Xueliang Zhu for comments on the manuscript. This work was supported by National Basic Research Program of China Grants 2012CB945002 and 2010CB912204, and National Natural Science Foundation of China Grants 31130015, 31271437, 31371382, and 91313302. 1. Morin X, Bellaïche Y (2011) Mitotic spindle orientation in asymmetric and symmetric cell divisions during animal development. Dev Cell 21(1):102 119. 2. Pease JC, Tirnauer JS (2011) Mitotic spindle misorientation in cancer out of alignment and into the fire. J Cell Sci 124(Pt 7):1007 1016. 3. Lu MS, Johnston CA (2013) Molecular pathways regulating mitotic spindle orientation in animal cells. Development 140(9):1843 1856. 4. Pickart CM (2001) Mechanisms underlying ubiquitination. Annu Rev Biochem 70: 503 533. 5. Nijman SM, et al. (2005) A genomic and functional inventory of deubiquitinating enzymes. Cell 123(5):773 786. 6. Thoma CR, et al. (2009) VHL loss causes spindle misorientation and chromosome instability. Nat Cell Biol 11(8):994 1001. 7. Sun X, et al. (2013) Parkin deficiency contributes to pancreatic tumorigenesis by inducing spindle multipolarity and misorientation. Cell Cycle 12(7):1133 1141. 8. Gao J, et al. (2008) The tumor suppressor CYLD regulates microtubule dynamics and plays a role in cell migration. J Biol Chem 283(14):8802 8809. 9. Harhaj EW, Dixit VM (2011) Deubiquitinases in the regulation of NF-κB signaling. Cell Res 21(1):22 39. 10. Delaval B, Bright A, Lawson ND, Doxsey S (2011) The cilia protein IFT88 is required for spindle orientation in mitosis. Nat Cell Biol 13(4):461 468. 11. Stegmeier F, et al. (2007) The tumor suppressor CYLD regulates entry into mitosis. Proc Natl Acad Sci USA 104(21):8869 8874. 12. Huang X, et al. (2011) Deubiquitinase USP37 is activated by CDK2 to antagonize APC(CDH1) and promote S phase entry. Mol Cell 42(4):511 523. 13. Trompouki E, et al. (2003) CYLD is a deubiquitinating enzyme that negatively regulates NF-kappaB activation by TNFR family members. Nature 424(6950):793 796. 14. Laan L, et al. (2012) Cortical dynein controls microtubule dynamics to generate pulling forces that position microtubule asters. Cell 148(3):502 514. 15. Radulescu AE, Cleveland DW (2010) NuMA after 30 years: The matrix revisited. Trends Cell Biol 20(4):214 222. 16. Ségalen M, et al. (2010) The Fz-Dsh planar cell polarity pathway induces oriented cell division via Mud/NuMA in Drosophila and zebrafish. Dev Cell 19(5):740 752. 17. Kikuchi K, Niikura Y, Kitagawa K, Kikuchi A (2010) Dishevelled, a Wnt signalling component, is involved in mitotic progression in cooperation with Plk1. EMBO J 29(20):3470 3483. 18. Tauriello DV, et al. (2010) Loss of the tumor suppressor CYLD enhances Wnt/betacatenin signaling through K63-linked ubiquitination of Dvl. Mol Cell 37(5):607 619. 19. Yamashita YM, Jones DL, Fuller MT (2003) Orientation of asymmetric stem cell division by the APC tumor suppressor and centrosome. Science 301(5639):1547 1550. 20. Godin JD, et al. (2010) Huntingtin is required for mitotic spindle orientation and mammalian neurogenesis. Neuron 67(3):392 406. 21. Massoumi R (2011) CYLD: A deubiquitination enzyme with multiple roles in cancer. Future Oncol 7(2):285 297. 22. Massoumi R, Chmielarska K, Hennecke K, Pfeifer A, Fässler R (2006) Cyld inhibits tumor cell proliferation by blocking Bcl-3-dependent NF-kappaB signaling. Cell 125(4): 665 677. 23. Zhang J, et al. (2006) Impaired regulation of NF-kappaB and increased susceptibility to colitis-associated tumorigenesis in CYLD-deficient mice. J Clin Invest 116(11): 3042 3049. CELL BIOLOGY Yang et al. PNAS February 11, 2014 vol. 111 no. 6 2163

Supporting Information Yang et al. 10.1073/pnas.1319341111 SI Materials and Methods Antibodies and Chemicals. Antibodies against α-tubulin, disheveled 3 (Dvl3), nuclear mitotic apparatus (NuMA), and lateral geniculate nucleus (LGN; Abcam); USP37 and resistance to inhibitors of cholinesterase 8 homolog A (RIC8A; Proteintech); A20 (ebioscience); dynein intermediate chain 1 and 2 (DIC1/2), Dvl1, Dvl2, cylindromatosis (CYLD), Flag, β-actin, and γ-tubulin (Santa Cruz Biotechnology); p-h3 (Millipore); p150 Glued (BD Bioscience); Myc and HA (Sigma-Aldrich); and GFP (Roche) were purchased from the indicated sources. Horseradish peroxidase-conjugated antimouse and anti-rabbit secondary antibodies were from Amersham Biosciences. Rhodamine- or fluorescein-conjugated secondary antibodies were obtained from Jackson ImmunoResearch Laboratories. Paclitaxel, nocodazole, DAPI, and rhodamine- or fluoresceinconjugated phalloidin were from Sigma-Aldrich. Matrigel and collagen I were from BD Bioscience. sirnas. Control sirna (5 -CGUACGCGGAAUACUUCGA-3 ), CYLD sirnas (1: 5 -CGAAGAGGCUGAAUCAUAA-3 ; 2: 5 - CUGCAAUAGUGGUCAGAAA-3 ), A20 sirnas (1: 5 -GAA- GCUCAGAAUCAGAGAU-3 ; 2: 5 -AUCUGCUUGAACUG- AAAGC-3 ), and USP37 sirnas (1: 5 -CCAAGGAUAUUUCA- GCUAA-3 ; 2:5 -GCAUACACUUGCCCUGUUA-3 ) were synthesized by Dharmacon. Plasmids. Mammalian expression plasmids for GFP-CYLD, GST- CYLD, HA-Dvl3, and His-Myc-ubiquitin were cloned using pegfpc1, pebg, pcmv-ha, and pcdna3 vectors, respectively, and various mutants were generated by PCR and sitedirected mutagenesis. The DsRed-histone 2B expression plasmid was generated using the pdsredn1 vector, and the pegfpc1- NuMA plasmid was from S. Dong (Baylor College of Medicine, Houston; Addgene plasmid 28238). The plasmids pcdna3-flag- WWP2 (from C. Chen, Kunming Institute of Zoology, Yunnan, China), pcmv-ha-vhl (from C. Kang, Tianjin Medical University, Tianjin, China), pcdna3-flag-traf2, pcdna3-flag-traf3, pcdna3-flag-traf5, pcdna3-flag-traf6 and pcdna3-flag- AIP4 (from D. Chen, Peking University, Beijing), and pcdna3- Flag-Trim21 (from C. Wang, Shanghai Institute of Biochemistry and Cell Biology, Shanghai) were obtained from the indicated sources. CYLD shrna expression plasmids were generated using the psuper vector as described previously (1), with the following double-stranded oligonucleotides (1: sense strand, 5 -gatcccccctcatgcagttctctttgttcaagagacaaaga- GAACTGCATGAGGtttttggaaa-3 ; antisense strand, 5 -agcttttc- caaaaacctcatgcagttctctttgtctcttgaacaaagagaa- CTGCATGAGGggg-3. 2: sense strand, 5 -gatccccctgcaat- AGTGGTCAGAAAttcaagagaTTTCTGACCACTATTGCAGtttttggaaa-3 ; antisense strand, 5 -agcttttccaaaaactgcaatag- TGGTCAGAAAtctcttgaaTTTCTGACCACTATTGCAGggg-3 ). Mice. CYLD knockout mice (in C57BL6/DBA mixed genetic background) were generated and genotyped as described previously (2, 3). CYLD +/ mice were intercrossed to generate CYLD +/+ and CYLD / littermates. Animal experiments were performed in accordance with protocols approved by the Animal Care and Use Committee of Nankai University. Cell Culture, Synchronization, and Transfection. HeLa, 293T, and Caco-2 cells (from the American Type Culture Collection), HeLa-H2B cells (from E. Griffis, University of Dundee, Dundee, United Kingdom) and HaCaT cells (from J. Dong, Nankai University) were grown in DMEM supplemented with 10% (vol/vol) FBS at 37 C in 5% (vol/vol) CO 2. HeLa-shControl and HeLa-shCYLD stable cell lines were established by transfection of HeLa cells with control or CYLD shrna-expressing lentiviral particles (Santa Cruz Biotechnology) followed by selection in puromycin. To produce cysts, Caco-2 cells were trypsinized and mixed with Hepes (20 mm), collagen I (1 mg/ml), and Matrigel [40% (vol/vol)] to achieve a final concentration of 5 10 4 cells/ ml. The mixture (100 μl) was plated in each well of eight-well chamber slides, allowed to solidify for 30 min, and then overlaid with 400 μl of media. For synchronization, HeLa cells were subjected to double thymidine block (2 mm, 18 h), followed by release for 2 h to obtain S-phase cells and release for 8 h to obtain G2-phase cells. Alternatively, cells were treated with thymidine (2 mm) for 16 h, released for 3.5 h, and treated with nocodazole (100 ng/ml) for 12 h, followed by release for 0.5 h to obtain mitotic cells and release for 3 h to obtain G1-phase cells. All plasmids were transfected to cells using polyethyleneimine (Sigma-Aldrich), and sirnas were transfected using Lipofectamine 2000 (Invitrogen). 1. Brummelkamp TR, Nijman SM, Dirac AM, Bernards R (2003) Loss of the cylindromatosis tumour suppressor inhibits apoptosis by activating NF-kappaB. Nature 424(6950): 797 801. 2. Reiley WW, et al. (2007) Deubiquitinating enzyme CYLD negatively regulates the ubiquitin-dependent kinase Tak1 and prevents abnormal T cell responses. J Exp Med 204(6):1475 1485. 3. Reiley WW, et al. (2006) Regulation of T cell development by the deubiquitinating enzyme CYLD. Nat Immunol 7(4):411 417. Yang et al. www.pnas.org/cgi/content/short/1319341111 1of4

Fig. S1. shrna-mediated CYLD depletion leads to spindle misorientation. (A) Immunoblots for CYLD and β-actin expression in HeLa-shControl and HeLashCYLD cells. (B) Immunostaining with anti α-tubulin (green) and anti γ-tubulin (red) antibodies and DAPI (blue) in HeLa-shControl and HeLa-shCYLD cells. The position of the z stage is indicated in micrometers; 3D, xy projection. (Scale bars, 5 μm.) (C E) Spindle angle distribution (C), average spindle angle (D), and average spindle diameter (E) in cells treated as in B (n = 70 cells per group). Student t test for all graphs. ***P < 0.001; ns, not significant. Error bars indicate SEM. Fig. S2. Depletion of A20 or USP37 does not affect spindle orientation. (A) Immunoblots for A20 and β-actin expression in control or A20 sirna-treated HeLa cells. (B) Immunofluorescence images (z sections) of metaphase HeLa cells transfected with control or A20 sirnas followed by staining with anti γ-tubulin (red) and anti α-tubulin (green) antibodies and DAPI (blue). The position of the z stage is indicated in micrometers. (Scale bars, 8 μm.) (C and D) Average spindle angle (C) and average spindle diameter (D) in cells treated as in B (n = 50 cells per group). (E) Immunoblots for USP37 and β-actin expression in control or USP37 sirna-treated HeLa cells. (F) Immunofluorescence images (z sections) of metaphase HeLa cells transfected with control or USP37 sirnas followed by staining with anti γ-tubulin (red) and anti α-tubulin (green) antibodies and DAPI (blue). The position of the z stage is indicated in micrometers. (Scale bars, 8 μm.) (G and H) Average spindle angle (G) and average spindle diameter (H) in cells treated as in F (n = 50 cells per group). Student t test for all graphs; ns, not significant. Error bars indicate SEM. Yang et al. www.pnas.org/cgi/content/short/1319341111 2of4

Fig. S3. Effects of selected E3 ubiquitin ligases on Dvl3 ubiquitination. (A C) Examination of Dvl3 ubiquitination in 293T cells transfected with various E3 ligases and His-Myc-ubiquitin, together with HA-Dvl3 (A and C) or Flag-Dvl3 (B). (D) Experiments were performed as in A C, and the relative level of Dvl3 ubiquitination was quantified. Student t test for all graphs. ***P < 0.001; ns, not significant. Error bars indicate SEM. Yang et al. www.pnas.org/cgi/content/short/1319341111 3of4

Fig. S4. CYLD is required for spindle orientation in 3D cell culture and mice. (A) Immunoblots for CYLD and β-actin expression in control and CYLD sirnatreated Caco-2 cells. (B and C) Immunofluorescence images (B) and spindle angle distribution (C)(n = 25 mitotic cells per group) of cysts formed from control or CYLD sirna-treated Caco-2 cells, stained with anti α-tubulin antibody (red) and DAPI (blue). (B, Inset) Entire cysts. Yellow lines, lumen borders. Dashed lines, dividing cells. (Scale bars, 10 μm.) (D) Immunofluorescence images of intestine crypt tissues of CYLD +/+ and CYLD / mice, stained with anti α-tubulin antibody (red) and DAPI (blue). Yellow lines, apical surface; white lines, basal surface; dashed lines, dividing cells. (Scale bars, 10 μm.) (E) Spindle angle distribution in intestine crypt tissues described as in D (n = 70 mitotic cells per group). (F) Immunofluorescence images of epidermal tissues of CYLD +/+ and CYLD / mice, stained with fluorescein phalloidin (green) and DAPI (blue). White lines, basal lamina; dashed lines, dividing cells. (Scale bars, 10 μm.) (G) Spindle angle distribution in epidermal tissues described as in F (n = 40 mitotic cells per group). Fig. S5. Effects of CYLD sirnas on the ubiquitination of LGN and RIC8A. (A) Examination of LGN and RIC8A ubiquitination in 293T cells transfected with control or CYLD sirnas, together with His-Myc-ubiquitin. (B) Experiments were performed as in A, and the relative level of LGN and RIC8A ubiquitination was quantified. Student t test for all graphs; ns, not significant. Error bars indicate SEM. Yang et al. www.pnas.org/cgi/content/short/1319341111 4of4