Epithelial organs are made of tubes and cavities lined by a monolayer of polarized cells that enclose the central lumen. Lumen formation is a crucial step in the formation of epithelial organs. The Rho guanosine triphosphatase (GTPase) Cdc42, which is a master regulator of cell polarity, regulates the formation of the central lumen in epithelial morphogenesis. However, how Cdc42 is regulated during this process is still poorly understood. Guanine nucleotide exchange factors (GEFs) control the activation of small GTPases. Using the three-dimensional Madin–Darby canine kidney model, we have identified a Cdc42-specific GEF, Intersectin 2 (ITSN2), which localizes to the centrosomes and regulates Cdc42 activation during epithelial morphogenesis. Silencing of either Cdc42 or ITSN2 disrupts the correct orientation of the mitotic spindle and normal lumen formation, suggesting a direct relationship between these processes. Furthermore, we demonstrated this direct relationship using LGN, a component of the machinery for mitotic spindle positioning, whose disruption also results in lumen formation defects.
During development, epithelial cells develop apicobasal polarity, a specific type of constitutive cell polarity which is regulated by different mechanisms, including membrane transport, cytoskeleton organization, and cellular junction formation (Bryant and Mostov, 2008). The deregulation of apicobasal polarity is associated with major diseases such as polycystic kidney disease and cancer (Lee and Vasioukhin, 2008). Rho GTPases are molecular switches that control a wide variety of signaling pathways critical for the acquisition of the polarized phenotype. For instance, the orientation of epithelial cell polarity is controlled by the opposing actions of Rac1 and RhoA (O’Brien et al., 2001; Yu et al., 2003, 2005, 2008). Cdc42, which is a master regulator of cell polarity, controls the formation of a single lumen in MDCK cells. For this, Cdc42 is activated at the apical plasma membrane in a pathway regulated by annexin2 (Anx2) and PTEN, which mediate the enrichment of phosphatidylinositol-4,5-bisphosphate (PtdInsP2) at the apical domain (Martín-Belmonte et al., 2007). Additionally, Cdc42 activity has also been shown to regulate epithelial morphogenesis by controlling other processes such as vesicle traffic and exocytosis (Kroschewski et al., 1999; Müsch et al., 2001; Rojas et al., 2001; Wu et al., 2008) and, more recently, the mitotic spindle orientation (Jaffe et al., 2008). Therefore, Cdc42 appears to control different pathways and/or mechanisms for epithelial morphogenesis. How Cdc42 is regulated during these processes is currently unknown.
Rho guanine nucleotide exchange factors (GEFs) constitute the main activators of Rho GTPases. Rho GEFs are multidomain proteins modulated by different signals, whose number in the human genome greatly exceeds the number of Rho family proteins. This suggests that different Rho GEFs could be regulating where and how a Rho GTPase is activated to control different cellular processes (Jaffe and Hall, 2005). Intersectin (ITSN) is a multimodular protein that is mainly expressed in two splicing variants: ITSN short (ITSN-S) and ITSN long (ITSN-L). Only ITSN-L contains the Dbl domain at the C-terminal region of ITSN and functions as a specific GEF for Cdc42 (Hussain et al., 2001). ITSN-L has two isoforms in mammals, ITSN1-L, which is differentially expressed in brain, and ITSN2-L, which is ubiquitously expressed (Pucharcos et al., 2000). ITSN interacts with Wiskott-Aldrich syndrome protein (WASP) and neural WASP (N-WASP) through its SH3 domains to trigger actin polymerization together with Cdc42 (Hussain et al., 2001; McGavin et al., 2001; Irie and Yamaguchi, 2002). ITSNs have been proposed to be a connection between endocytosis and exocytosis because they bind to multiple endocytic and exocytic proteins such as dynamin and SNAP25 and -23 (Okamoto et al., 1999). In fact, the ability of ITSNs to interact with multiple components suggests that they might act as scaffolding proteins necessary for the formation of signaling platforms.
In this work, we have characterized ITSN2 as a specific GEF for Cdc42 activation in epithelial morphogenesis using the organotypic 3D MDCK cell system. We have found out that ITSN2 localizes to centrosomes and is required for the correct orientation of the mitotic spindle and for correctly positioning the apical surface during epithelial morphogenesis. In addition, we have demonstrated a direct relation between lumen formation and spindle orientation. Disruption of LGN, a component of the machinery which regulates spindle movements and orientation, interferes with lumen formation in MDCK cells forming cysts.
ITSN2 is required for normal lumen morphogenesis
In our previous work, we have described that Cdc42 must be activated to induce the formation of the apical domain and the central lumen in 3D MDCK cysts (Martín-Belmonte et al., 2007). A candidate screening for Cdc42 GEFs using RNAi was performed to identify Cdc42-specific GEFs associated with epithelial lumen formation using the apical marker gp135/podocalyxin and the actin cytoskeleton integrity as readout to detect luminal defects in the 3D MDCK model. Using this system, ITSN2 emerged as a candidate for regulating the Cdc42 activity controlling lumen formation (unpublished data). To confirm the function of ITSN2, we designed three siRNA heteroduplexes to deplete endogenous ITSN2 levels. The siRNA directed against the C-terminal region (ITSN2-3) specifically reduced the levels of the long isoform, whereas siRNAs ITSN2-1 and ITSN2-2, directed against the middle and N-terminal region of ITSN2, respectively, were able to reduce the expression of both isoforms (Fig. 1 A). The pool of the three siRNAs dramatically reduced both isoforms as well (Fig. 1 A) and was used for all of the following experiments to silence ITSN2. When siRNA-transfected cells were plated to form cysts, most control cells formed normal lumens at 72 h (70%; Fig. 1 B); MDCK cysts with reduced ITSN2 formed cysts of similar size, but only 41% of them had normal lumens at 72 h. Instead, most cysts with reduced ITSN2 had multiple small lumens, although the basolateral marker β-catenin and tight junctions marker ZO-1 localized normally (Fig. 1, B and C). This phenotype very closely resembles that obtained by silencing Cdc42 in MDCK cysts (Fig. 1 C; Martín-Belmonte et al., 2007). To further confirm these results, we prepared RNAi pPRIME lentiviral system carrying short hairpin RNAs targeted specifically to silence the canine isoform of ITSN2 and obtained consistent results (Fig. S1). To investigate the GEF function of ITSN2 in 3D MDCK cysts, we analyzed GTP-bound Cdc42 levels in ITSN2 and Cdc42 knockdown (KD) cells by GST pull-down assays. Cells silenced for ITSN2 showed a significant reduction in Cdc42-GTP levels, which might explain the aberrant phenotype observed in cyst morphogenesis (Fig. 1, D and E; and Fig. S1 D).
To confirm the specificity of ITSN2 function on Cdc42 activation, we prepared MDCK cells expressing the constitutively active form of Cdc42, Cdc42V12-GFP under the control of a tetracycline-repressible promoter (tet-off; Fig. 1 F), and we tested the ability for lumen formation in ITSN2- and Cdc42-silenced cells. The expression of Cdc42V12 was able to restore the normal formation of the lumen in cells silenced for ITSN2 and also, although to lesser extent, of cells silenced for Cdc42 (Fig. 1, G and H). Together, these results suggest that ITSN2 is required for the activation of Cdc42 to form the lumen.
ITSN2 localizes at centrosomes
Next, we characterized the localization of ITSN2 in MDCK cysts. Staining with several antibodies available for ITSN2 did not serve to identify the localization of the endogenous protein (unpublished data). Therefore, we prepared MDCK cells stably expressing a fusion of the fluorescent protein Venus and human ITSN2 (vhITSN2). In interphase cells, vhITSN2 localized to intracellular clusters close to the apical membrane, which was stained with the apical marker gp135 in MDCK cysts (Fig. 2 A, arrows). vhITSN2 colocalized with the centrosomal markers pericentrin (Fig. 2 B, arrows) and γ-tubulin (not depicted). In mitotic cells, vhITSN2 also distributed at the edge of the spindle poles, stained with acetylated tubulin (Fig. 2 C, arrows), and colocalized with the centrosomal marker pericentrin (not depicted). Because vhITSN2 is an siRNA-resistant form of ITSN2 (Fig. 2 D), we used these cells to validate the specific effect of ITSN2 silencing on lumen formation. We observed that vhITSN2 expression restored the normal phenotype in lumen formation in cells silenced for endogenous ITSN2 (Fig. 2, E and F).
To confirm that the localization of ITSN2 to centrosomes was not an artifact caused by the ectopic expression of vhITSN2, we obtained centrosome-enriched fractions of MDCK cells using centrifugation in Ficoll gradients (Blomberg-Wirschell and Doxsey, 1998). We detected endogenous ITSN2 in the sediments positive for γ-tubulin/pericentrin and negative for α-tubulin/gp135 (Fig. 2 G, left lane, F2-P). The quantitative analysis showed that both ITSN isoforms (ITSN2-L and ITSN2-S) were highly enriched in the centrosomal fraction together with other centrosomal proteins (Fig. 2 H). To characterize whether the association of centrosomes with ITSN2 is dependent on its association with microtubules or, conversely, whether ITSN2 associates with centrosomes independently of the microtubules, we performed depolymerization/repolymerization experiments using nocodazole (Hung et al., 2000). Microtubules were depolymerized using nocodazole for 4 h, and the subsequent pattern of microtubule regrowth was determined (Fig. S2 A). In nocodazole, interphase microtubules were depolymerized. However, compared with untreated cells, no significant decrease in the concentration of ITSN2 was observed in nocodazole-treated cells (Fig. S2 A). To ensure that ITSN2 is indeed located at the centrosomes where microtubule growth is initiated, the nocodazole-treated cells were washed out to permit microtubule regrowth. After increasing times of regrowth (0–45 min), cells were fixed and stained to detect α- and γ-tubulin, and confocal images were analyzed. As shown in Fig. S2 A, microtubule growth started at the centrosomes, where ITSN2 is located.
Altogether, these results indicate that ITSN2 localizes to centrosomes, and it could activate Cdc42 in this region to form the lumen. Because Cdc42 regulates cell polarity through different effectors that control vesicle trafficking, actin organization, and microtubule stabilization (Etienne-Manneville, 2004), we next focused on characterizing the process in which ITSN2 activates Cdc42.
The exocytosis of apical gp135/podocalyxin and apical junction formation are both dependent on Cdc42 but independent of ITSN2
Activated Cdc42 can associate with numerous effectors that regulate vesicle traffic, mainly through the regulation of the actin cytoskeleton. Accumulated evidence supports the idea that ITSNs are adaptors that coordinate Cdc42-dependent membrane trafficking events in different cells (Hussain et al., 2001; Irie and Yamaguchi, 2002; Malacombe et al., 2006). To evaluate whether ITSN2 functions in exocytosis during lumen formation, we analyzed apical vesicle exocytosis using calcium switch experiments. For these experiments, we used MDCK stably expressing GFP-tagged Anx2 (Anx2-GFP), a peripheral membrane protein which associates with the plasma membrane in a calcium-dependent manner (Rescher and Gerke, 2004), and the apical marker gp135/podocalyxin. Control or MDCK cells silenced for ITSN2 or Cdc42 were polarized in monolayers. After 48 h, the cells were treated overnight with normal (Fig. 3 A) or low-calcium medium (Fig. 3 B). Low-calcium conditions induced the internalization of the apical protein gp135 and Anx2-GFP in a vesicular apical compartment (Fig. 3 B), as described previously (Vega-Salas et al., 1987, 1988). The restitution of normal calcium levels induced the exocytic translocation of gp135 and Anx2-GFP to the plasma membrane in control cells (Fig. 3 C). In contrast, Cdc42 KD cells did not restore the apical compartment after returning to normal calcium conditions (Fig. 3 C), inducing a phenotype very similar to that produced by silencing Cdc42 in the 3D MDCK cyst model (Martín-Belmonte et al., 2007, 2008). Importantly, the restitution of normal calcium levels in ITSN2 KD cells resulted in normal exocytosis of gp135 and Anx2-GFP to the plasma membrane (Fig. 3 C). Previous experiments have shown that activated Cdc42 is implicated in the formation of normal tight junctions at epithelial cell–cell contacts (Joberty et al., 2000; Hurd et al., 2003; Wells et al., 2006). To test whether ITSN2 could activate Cdc42 to form apical junctions, we performed a calcium switch experiment and analyzed the recovery of transepithelial resistance in vivo using an electrical cell-substrate impedance system (Lo et al., 1995). Again, the recovery of normal transepithelial resistance levels was delayed in cells knocked down for Cdc42, whereas cells knocked down for ITSN2 showed a recovery profile similar to that of control cells (Fig. 3 D). Furthermore, the staining of the tight junction protein ZO-1 also showed a specific effect of Cdc42 in the recovery of tight junction integrity in the calcium switch experiment (Fig. 3, A–C). Finally, we also observed that ITSN2 and Cdc42 were not implicated in the orientation of the microtubule-organizing center (MTOC; Fig. S3, A and B) or in the formation of the primary cilium (not depicted). These results suggest that ITSN2 activation of Cdc42 does not affect either vesicle trafficking or tight junction formation in MDCK cell morphogenesis.
With the goal of identifying the possible role of Cdc42 in which ITSN2 is involved during the process of lumen formation, we further investigated the phenotypes of ITSN2 KD and Cdc42 KD in MDCK cysts. First, we quantified the proportion of internal vesicles with apical markers and intercellular lumens in siRNA-transfected cells. For this quantification, we assumed that the intracellular lumens were similar to the vesicular apical compartments that appeared in the calcium switch experiments (Fig. 3 B) and, therefore, could indicate flaws in the process of exocytosis. ITSN2 KD cysts had a significantly reduced proportion of intracellular gp135-loaded vesicles compared with Cdc42 KD cysts, which contained both intercellular lumens and intracellular gp135-loaded vesicles (Fig. 3, E and F). Collectively, these results indicated that ITSN2 does not detectably function to activate Cdc42 for vesicular trafficking or the formation of tight junctions, but it could be necessary for other processes such as mitotic spindle orientation, in which defects in the activation of Cdc42 results in multiple lumen phenotypes.
ITSN2 and CDC42 are both required for normal positioning of the mitotic spindle in 3D MDCK cultures
Recent work performed in Caco-2 epithelial cells suggested that Cdc42 is required for the orientation of the mitotic spindle to position the apical surface in cells forming 3D cysts (Jaffe et al., 2008). We analyzed the orientation of the mitotic spindle in dividing cells using cysts of MDCK cells with reduced levels of Cdc42 or ITSN2 and compared them with control cells. We transfected MDCK cells with control or siRNA heteroduplexes targeting Cdc42 or ITSN2 (Fig. 4 A) and then measured the angle formed between the apicobasal axis and the spindle pole axis in three dimensions, which was calculated as described in the schematic drawing of Fig. 4 B. Most of the mitotic spindles analyzed in control cells were normal (Fig. 4 B, right). However, cysts with reduced levels of both Cdc42 and ITSN2 had a significant increase in abnormally positioned spindle poles, the effect being stronger in cells silenced for Cdc42 (Fig. 4, B [left and middle] and C). These results suggest that the defect in mitotic spindle rotation and/or positioning in cells silenced for Cdc42 might generate the defects observed in lumen formation. To test this hypothesis, we analyzed mitotic spindle dynamics in metaphase by confocal time-lapse microscopy using 3D MDCK cells expressing GFP–α-tubulin. We observed that in control cells, there was an ∼90° rotation of the spindle from the apicobasal axis to the plane of the epithelium (Fig. 4 D, top; and Video 1), which was described previously in MDCK cysts (Yu et al., 2003). However, in cells silenced for Cdc42 or ITSN2, there was a disruption of this spindle rotation, which resulted in the translocation of one of the dividing cells to the center of the cyst, narrowing the luminal space (Fig. 4 D, middle and bottom; and Videos 2 and 3). Moreover, the impairment in spindle rotation also induced a delay in the metaphase to anaphase transition, which we characterized in MDCK 2D monolayers for statistics (Fig. S3 C). Metaphase time (MT) was affected in cells silenced for Cdc42 (MT = 27.3 ± 7.7 min) and cells silenced for ITSN2 (MT = 31.2 ± 11.5 min), as compared with control cells (MT = 16.9 ± 4.6 min; Fig. S3, C and D). These results indicate that ITSN2 and Cdc42 are both required for normal mitotic spindle orientation during cell division. In particular, they seem to control the normal rotation of the mitotic spindle in mitotic cells.
In summary, we have shown that ITSN2 activates Cdc42, which in turn controls the orientation of the mitotic spindle during mitosis. Although our results can explain the phenotype of multiple lumens observed in cells silenced for ITSN2 or Cdc42, it sheds no light on the subjacent molecular mechanism. To identify the possible mechanism of action of ITSN2, we characterized the domain responsible for targeting ITSN2 to centrosomes. ITSN2 contains two EH (Eps15 homology) domains, a central coiled-coil region, and five consecutive SH3 domains. Additionally, ITSN-L presents an extended C-terminal region containing a DH (Dbl homology), a PH (pleckstrin homology), and a C2 domain (Fig. 5 A). We prepared GFP-fused constructs comprising different domains of ITSN2 and characterized their subcellular localization in MDCK cells (Fig. 5, A and B). We observed that when overexpressed, the EH domains of ITSN2 targeted GFP to the centrosomes in MDCK cells (Fig. 5 B, left; arrows show the colocalization of EH domains with pericentrin). In contrast, the rest of the domains analyzed distributed in the cytoplasm, forming clumps that did not colocalize with centrosomal markers (Fig. 5 B, bottom right). Previous results have shown the ability of ITSNs to interact with multiple components of the endocytic machinery, which suggests that ITSNs might act as scaffolding proteins necessary for the formation of signaling platforms (Yamabhai et al., 1998; Okamoto et al., 1999; Pucharcos et al., 2000; Hussain et al., 2001). To characterize the possible role of ITSN2 as a scaffolding protein at centrosomes, we performed pull-down assays using GST fusions of the various aforementioned domains of ITSN2 (Fig. 5 A). ITSN2 was found to interact through the SH3 domains with p150Glued, one of the subunits of the dynactin complex, and with γ-tubulin but not with other members of the complex such as dynein intermediate chain (DIC; Fig. 5 C). As a control, we analyzed the interaction of ITSN2 with N-WASP through the SH3 domains of ITSN2 (Fig. 5 C), as described previously for ITSN2 and WASP in T cells (McGavin et al., 2001). The dynein–dynactin complex is an essential regulator of spindle orientation and cell division (Quintyne et al., 1999). Previous studies have indicated that Cdc42 activation is involved in the reorientation of the MTOC during cell migration through a pathway dependent on the dynein–dynactin complex (Palazzo et al., 2001; Gomes et al., 2005). We observed that vhITSN2 colocalized with p150Glued and pericentrin at the centrosome in 3D cysts (Fig. 5 D). These results suggest that ITSN2, which is targeted to the centrosome through its EH domains, mediates the activation of Cdc42 in this region, which in turn would control the orientation of the spindle through the activity of other centrosomal proteins such as the dynein–dynactin complex.
Next, we analyzed the localization of Cdc42 in different phases of the cell cycle using MDCK cells expressing Cdc42-GFP (Fig. 6). It has been reported that the localization of Cdc42 is associated with the plasma membrane and the Golgi in different cell types (Itoh et al., 2002; Yoshizaki et al., 2003). The stable expression of Cdc42-GFP caused a significant reduction of the endogenous Cdc42, so the final levels of total Cdc42 were similar to the endogenous protein in these cells (Fig. 6 B). Cdc42-GFP–expressing cells formed normal cysts as we showed previously (Martín-Belmonte et al., 2007). As expected, Cdc42-GFP localized mainly to the plasma membrane in interphase cells (Fig. 6 A, top); however, we detected an important fraction of Cdc42-GFP in close proximity to the mitotic spindles during different phases of cell division in MDCK cells (Fig. 6 A, middle and bottom). Interestingly, we observed that Cdc42 colocalized with ITSN2 at the mitotic spindles in dividing MDCK cells (Fig. 6 C, arrows). Therefore, during cell division, Cdc42 is localized to the region of the spindle pole and is activated by ITSN2 to control the orientation of the spindle. To investigate whether the presence of ITSN2 is necessary not only for the activation but also for the localization of Cdc42, we analyzed the effect of silencing ITSN2 on the localization of Cdc42 in MDCK cells in mitosis. In control cells, the localization of Cdc42 was concentrated at the spindle poles in cells in mitosis, but in cells silenced for ITSN2, Cdc42 was dispersed throughout the cytoplasm (Fig. 6, D [arrows] and E). These data suggest that ITSN2 is required for the localization and activation of Cdc42 at the spindle poles of mitotic cells, which in turn regulates the proper orientation of the mitotic spindle.
Disruption of LGN, a regulator of the mitotic spindle orientation, interferes with lumen formation
The data presented in this paper and previous results suggest a correlation between the orientation of the mitotic axis, regulated by Cdc42, and the formation of a single central lumen (Jaffe et al., 2008). However, to further demonstrate this relationship more convincingly, we investigated the effect of the alteration of another component of the machinery specific for mitotic spindle orientation on lumen formation. LGN, which is part of the machinery that controls the spindle orientation, contains two tetratricopeptide repeat motifs in its N-terminal region and four Gαi/o-Loco (GoLoco) motifs near the C terminus (Fig. 7 A). Tetratricopeptide repeat motifs are involved in protein–protein interactions, whereas GoLoco motifs have been implicated as inhibitors of GDP dissociation from heterotrimeric G protein α-subunits. The expression of the GoLoco motifs can act as a dominant-negative form of LGN (Ct-LGN; Fig. 7 A; Morin et al., 2007). LGN, which is expressed in MDCK cells, localizes at the cytoplasm in interphase but translocates to the cell cortex and spindle poles in mitosis (Du et al., 2001). We confirmed this localization for LGN-myc in MDCK cells (Fig. 7 B; arrows indicate LGN localization at the spindle poles). LGN regulates mitotic spindle movements and orientation, so interfering with its function randomizes the plane of division and disrupts the orientation of the mitotic spindle in epithelial and neuroepithelial cells (Du and Macara, 2004; Morin et al., 2007). To disrupt LGN function, we expressed Ct-LGN in MDCK cells under the control of an inducible promoter (tet-off; Fig. 7 C). We confirmed that Ct-LGN expression was sufficient to disperse the normal localization of LGN from the mitotic spindles (Fig. 7 D) and the normal orientation of the mitotic spindle in 3D MDCK cysts (abnormal spindles in control cells [+doxycycline (Dox)], 25.5%; abnormal spindles in Ct-LGN–induced cells [−Dox], 54.2%; Fig. 7 E), as was described previously in chick neuroepithelial cells (Morin et al., 2007). Interestingly, we also found that the expression of Ct-LGN disrupted the formation of normal lumens in MDCK cells forming cysts (Fig. 7, F and G). In sum, when we alter the orientation of the axis with the expression of a dominant-negative form of LGN, a protein essential for the orientation of the mitotic spindles in different models of cell division, the formation of the lumen results also altered. Therefore, with these experiments, we have demonstrated a direct relationship between the orientation of the mitotic axis and the formation of the lumen.
Collectively, the results presented in this study suggest that ITSN2 is involved in the activation of Cdc42 to regulate spindle orientation during mitosis. To perform this function, ITSN2 localizes to the centrosomes, both in interphase cells and in cells in mitosis. The EH domains of ITSN2 mediate the localization of ITSN2 at the centrosomes. Thus, ITSN2 operates as a spatial regulator for Cdc42 activity, which is also associated to the spindle poles during cell division. Then, Cdc42 is required somehow for the normal function of the spindle machinery, which is involved in regulating the orientation of the mitotic spindle at the centrosomes and for the normal formation of the lumen in epithelial morphogenesis.
The mechanisms and external factors that regulate spindle orientation in epithelial cells are poorly understood. The majority of these studies have been performed using asymmetric division models (for review see Siller and Doe, 2009). However, it has been postulated that the main players of this process in mammalian epithelial cells may be the same, including the dynein–dynactin complex, Gα proteins, LGN, NuMA (nuclear mitotic apparatus protein), the Par complex, and the Rho GTPase Cdc42 (for review see Siller and Doe, 2009). The molecular mechanism associated with the function of Rho GTPases is generally related to its subcellular localization (Iden and Collard, 2008). We have shown that Cdc42 localizes to the mitotic spindle in the centrosomal region, where it can activate downstream effectors that control the mitotic spindle machinery. An essential question is how Cdc42 associates with the spindle poles. We found that ITSN2 is localized in centrosomes, through the EH domains of ITSN2. Therefore, ITSN2 could recruit and activate Cdc42 at the spindle poles. In fact, silencing of ITSN2 significantly reduces the amount of active Cdc42 and that of the Cdc42 associated with the spindle poles. In addition, ITSN2 seems to mediate protein–protein interactions with other centrosomal proteins through the SH3 domains, and thus, it might scaffold signaling platforms for Cdc42 at this location. Therefore, a potential mechanism for Cdc42 function in the orientation of the mitotic spindles might be mediated by ITSN2, which would be responsible not only for activating Cdc42 in the centrosomal region but also for scaffolding proteins involved in downstream signaling.
In addition to the data presented in this study for the role of Cdc42 by controlling mitotic spindle orientation in the MDCK cells, Cdc42 activity was also recently described to regulate epithelial morphogenesis and spindle orientation in Caco-2 cells, a mammalian intestinal model of epithelial morphogenesis (Jaffe et al., 2008). However, the link between Cdc42, spindle orientation, and the formation of the lumen described previously was strictly correlative (Jaffe et al., 2008). To clarify this issue, we have shown that disruption of LGN, a known essential component of the spindle orientation machinery, also caused a defect lumen formation similar to the defect caused by the silencing of ITSN2 and Cdc42, which suggests a direct relationship between these two processes.
The mechanism of spindle rotation may involve centrosome movement directed by interactions between the astral microtubules and the cell cortex. Therefore, the connection of astral microtubules to centrosomes and the cell cortex must be fully coordinated. A role for the actin cytoskeleton and motors such as the dynein–dynactin complex has been proposed in this process (Schuyler and Pellman, 2001; Kunda and Baum, 2009). The actin cytoskeleton is one of the main downstream effects of Rho GTPase function. To stabilize the interaction between the cell cortex, centrosomes, and the astral microtubules, Cdc42 regulates actin remodeling via different effectors such as Pak (p21-activated kinase), N-WASP, and formins in mammalian cells (Narumiya and Yasuda, 2006). Moreover, in yeast, a Cdc42 GEF complex (Bem1–cdc24p, functionally similar to ITSN) interacts with PAK and regulates the polarization of the cell division machinery (Heil-Chapdelaine et al., 1999). A previous study has elucidated a route in which Cdc42 may regulate the activation of Pak2 at the centrosomes, which in turn regulates the spindle assembly machinery, including Aurora A, dynein–dynactin, and NuMA (Mitsushima et al., 2009). In other series of experiments, Cdc42 and its effector mammalian Dia3 have been shown to regulate the biorientation of the chromosomes, which involved attachment of the plus ends of microtubules to kinetochores, to ensure alignment of chromosomes during metaphase and their correct segregation during anaphase (Yasuda et al., 2004). Therefore, although the function of Cdc42 in spindle positioning is clear, the effectors involved and the molecular mechanism acting downstream of Cdc42 remain to be identified.
Different epithelial tissues undergo events of massive membrane trafficking in the initial steps of morphogenesis (Bryant and Mostov, 2008). Previous data have demonstrated a role for Cdc42 in vesicle trafficking in polarized epithelial cells (Müsch et al., 2001). We have observed a function for Cdc42 in the exocytosis of gp135-containing vesicles and the remodeling of tight junctions in calcium switch experiments. In addition, we have found a greater effect of silencing Cdc42 on lumen formation and mitotic spindle orientation than silencing ITSN2. Our present data suggest the existence of another GEF as regulator for Cdc42 during this process. Interestingly, we have detected ITSN2 at intracellular locations in MDCK cells, which might be endosomal membranes. Multiple studies have demonstrated an important function for ITSN2 during endocytosis (Pucharcos et al., 2000; McGavin et al., 2001). Recently, it has been shown that endocytosis can act to promote cell polarity in different models (Georgiou et al., 2008, Harris and Tepass, 2008; Leibfried et al., 2008). Even more, some proteins associated with endocytosis have been described to be required for spindle positioning in mitosis (Royle et al., 2005; Liu and Zheng, 2009). The possibility that ITSN2 regulates cell polarity through endocytosis remains to be investigated.
In summary, we have identified ITSN2 as an activator of Cdc42 at the centrosomes to regulate spindle orientation during mitosis. It has been recently described that the apical membrane and the lumen originate from the place where the midbody is formed during cytokinesis (Schlüter et al., 2009). Therefore, the regulation of the orientation of the mitotic spindle, which determines the location of the midbody, would be essential to ensure the formation of a single lumen.
Materials and methods
Antibodies against ITSN2 (Novus Biologicals), N-WASP (Santa Cruz Biotechnology, Inc.), Cdc42 (BD), pericentrin (Covance), mLGN/GPSM2/AGS3 (Abnova), α-tubulin (Sigma-Aldrich), γ-tubulin (Sigma-Aldrich), p150Glued (BD), GFP (Sigma-Aldrich), β-catenin (Sigma-Aldrich), and acetylated tubulin (Sigma-Aldrich) were used as primary antibodies. Gp135 antibody was a gift from the Ojakian laboratory (State University of New York Downstate Medical Center, Brooklyn, NY). Peroxidase-conjugated donkey anti–mouse IgG and anti–rabbit IgG were used as secondary antibodies for Western blots (Jackson ImmunoResearch Laboratories, Inc.). Alexa Fluor–conjugated secondary antibodies (Alexa Fluor 488, 555, or 647; Invitrogen) and TOPRO-3 (for nuclear/DNA staining) were used in microscopy protocols. Nocodazole (Sigma-Aldrich) and cytochalasin D (Sigma-Aldrich) depolymerized tubulin and actin cytoskeleton, respectively.
N-WASP–GFP was cloned in a pEGFP-C1 vector (Takara Bio Inc.). GFP-tagged ITSN2 constructs (EH/EH, coiled-coil, SH3, and C terminus) were cloned in a pEGFP-C2 vector (Takara Bio Inc.). Human ITSN2-Cherry was a gift from S. de la Luna (Centre de Regulació Genòmica, Barcelona, Spain). Human ITSN2-L–GFP, GST-EH/EH, GST–coiled-coil, GST-5xSH3, and GST–C terminus were gifts from K. Kaibuchi (Nagoya University, Chikusa-ku, Nagoya, Japan). vhITSN2 vector was a contribution from I. Macara (University of Virginia School of Medicine, Charlottesville, VA). Myc-tagged Ct-LGN inducible and LGN-myc vector were a gift from X. Morin (École Normale Supérieure, Paris, France).
MDCK cells were grown as described previously (Martín-Belmonte et al., 2007). MDCK cells stably expressing β-tubulin–GFP, Cdc42-GFP, vhITSN2-L, inducible Cdc42V12-myc or inducible Ct-LGN–myc were made by cotransfection with blasticidin-resistant gene and selection for 10 d with 20 ng/ml Dox and 0.5 mg/ml blasticidin. To prepare cysts in Matrigel, cells were trypsinized to a single cell suspension of 2 × 104 cells/ml in 2% Matrigel and plated in coverglass chambers (Thermo Fisher Scientific) covered with Matrigel. Cysts were grown for 3–5 d.
Immunofluorescence of cysts was previously described (Martín-Belmonte et al., 2007). Cysts were analyzed on a confocal microscope (510 or 710 LSM; Carl Zeiss, Inc.) using a 63× NA 1.4 oil Plan-Apochromat objective and a 63× NA 1.2 water C-Apochromat Corr (for live cell and cyst imaging). The imaging acquisition system used was ZEN software suite (Carl Zeiss, Inc.). For image processing, we used ImageJ (National Institutes of Health) and ZEN software suite. For videomicroscopy and 3D reconstitutions, we processed maximum projections of all stacks and then reduced background using ImageJ software. Fixed cyst imaging was performed in PBS medium or mounted using ProLong Gold antifade reagent. Fixed cells in monolayers were analyzed in Fluoromount medium. Cysts with actin/gp135 staining at the interior surface and β-catenin facing the ECM were identified as normal lumens (interior apical pole). Cysts that had actin/gp135 absent, in small multiple lumens, or at the periphery were considered as abnormal lumens. Per condition, >100 cysts/experiment were analyzed, SD was calculated, and statistical significance was determined by paired Student’s t test. For spindle orientation analysis, cysts were stained with acetylated tubulin, and the angle formed by the spindle and the apicobasal axis was measured. Divisions with angles <45° were considered abnormal. Per condition, >30 cysts/experiment were analyzed. For videomicroscopy, cells stably expressing β-tubulin–GFP were analyzed using an LSM 510 with incubation chamber at 5% CO2 and 37°C. Stacks of five images were taken every 2 min, and videos were rendered at 2 frames/s from centrosome duplication until cells achieved telophase.
25 nucleotide stealth siRNA duplexes targeting mRNA sequences of canine Cdc42 and ITSN2 were purchased from Invitrogen. Sequences were submitted to BLAST search to ensure targeting specificity. The specificity of Cdc42 and ITSN2 was further checked by Western blot analysis. MDCK cells were trypsinized and then nucleofected (Lonza) with either Cdc42 or ITSN2 siRNA duplexes or scrambled siRNA. After 24-h incubation, cells were resuspended and plated in 12-well plates and in coverglass chambers to grow cysts. Total cell lysates were analyzed by Western blotting as described previously (Martín-Belmonte et al., 2007) to confirm siRNA efficiency.
Calcium switch and microtubule repolymerization assays
Calcium switch was performed by incubating cells overnight in a calcium-free MEM medium (supplemented with dialyzed FCS) and then restoring normal calcium conditions with complete MEM for 120 min. Microtubule depolymerization was performed by incubating cells in 20 µg/ml nocodazole-supplemented MEM for 240 min. Then, nocodazole was washed out by incubating cells in normal MEM for 15–45 min.
Centrosomes were purified using fast isolation of centrosomes (Blomberg-Wirschell and Doxsey, 1998). 5 × 108 cells were treated during 180 min with 20 µg/ml nocodazole and 1 µg/ml cytochalasin D. Cells were washed in cold PBS, 8% sucrose, 0.1× PBS, and then 8% sucrose and finally lysed using 0.5% Triton X-100 and 0.1% 2-mercaptoethanol in Tris-HCl buffer, pH 8.0, for 10 min at 4°C. Cell debris and nuclei were removed by fast centrifugation at 1,500 g for 5 min at 4°C and 33-µm filters. The lysate was loaded on top of a 1.5-ml 20% Ficoll 400 cushion and centrifuged at 25,000 g for 20 min at 4°C. Five 1-ml fractions were collected, and then fractions were diluted 1:4 in Pipes-EDTA buffer and centrifuged at 25,000 g for 20 min at 4°C. The sediment was resuspended in Laemmli buffer and loaded. Supernatants and top fractions were loaded as control.
Cysts were lysed in 500 µl of 4°C 2× gold lysis buffer (2% Triton X-100, 40 mM Tris HCl, pH 7.5, 150 mM NaCl, 20 mM MgCl2, 30% glycerol, 1 mM DTT, and EDTA-free protease inhibitors; Roche) and centrifuged at 15,000 rpm for 5 min. A 50-µl sample from the supernatant was set aside for determination of Cdc42 and tubulin levels in the total lysate, and GTP loading on Cdc42 was determined with GST-Pak3-CRIB pull-down.
ITSN2 pull-down assays
Cysts grown in 100,000 cells/ml on top of Matrigel-coated dishes were lysed in 0.1% SDS, 1% Triton X-100, 0.5 mM DTT, and 1× TBS buffer with protease inhibitor cocktail and NaOV. Cell debris was removed by fast cold centrifugation at 14,000 g for 2 min, and cleared lysates were incubated in rotation with protein-GST–loaded beads for 30 min. Beads were centrifuged and washed four times, dried using aspiration, and resuspended in Laemmli buffer.
Online supplemental material
Fig. S1 shows the effect of ITSN2 silencing using stable short hairpin RNA expression. Fig. S2 shows that ITSN2 localization to centrosomes is independent of microtubules. Fig. S3 shows that MTOC orientation is not affected by ITSN2 or Cdc42 silencing. Videos 1, 2, and 3 show spindle dynamics in control MDCK, ITSN2-silenced, and Cdc42-silenced cells, respectively.
We thank C.M. Ruiz-Jarabo and I. Correas for comments on the manuscript and members of the F. Martín-Belmonte, J. Millán, and M.A. Alonso laboratories for discussion. We also thank I. Macara for the plasmid of vhITSN2, X. Morin for the plasmids of LGN-myc and Ct-LGN–myc, S. de la Luna for the plasmid of hITSN2, and K. Kaibuchi for the plasmids of GST-tagged domains of ITSN2.
This work was supported by grants from the Human Frontier Science Program (HFSP-CDA 00011/2009) and Marie Curie (IRG-209382) to F. Martín-Belmonte, National Institutes of Health grants (R01 DK067153 and R01 DK074398) to K. Mostov, and grants from the Ministerio de Ciencia e Innovación to F. Martín-Belmonte (BFU2008-01916) and M.A. Alonso (BFU2006-01925) and to F. Martín-Belmonte and M.A. Alonso (CONSOLIDER CSD2009-00016). An institutional grant from the Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa is also acknowledged.
Note added in proof. In this issue, Qin et al. (2010. J. Cell Biol. doi:10.1083/jcb.201002097) have characterized the role of Tuba, another GEF specific for Cdc42, to be involved in spindle orientation and epithelial cyst formation, suggesting that Cdc42 must be regulated at different levels to control epithelial morphogenesis.
Also, during the editing process, Zheng et al. (2010. J. Cell Biol. doi:10.1083/jcb.200910021) characterized the role of LGN in spindle orientation and epithelial cyst formation, which corroborates the results presented in this paper.