Direct Binding of Lgl2 to LGN during Mitosis and Its Requirement for Normal Cell Division*

The Drosophila tumor suppressor protein lethal (2) giant larvae (l(2)gl) is involved in asymmetric cell division during development and epithelial cell polarity through interaction with the aPKC·Par-6 complex. We showed here that Lgl2, a mammalian homolog of l(2)gl, directly bound to LGN, a mammalian homolog of Partner of inscuteable in HEK293 cells. The C-terminal tail of Lgl2 bound to LGN with a Kd value of about 56 nm. Endogenous Lgl2 formed a complex with aPKC, Par-6, and LGN. This complex formation was enhanced in metaphase of the synchronized cells by treatment with thymidine and nocodazole. Immunofluorescence staining of the complex was the strongest at the cell periphery of the metaphase cells. Overexpression of the C-terminal tail of Lgl2 induced mis-localization of the nuclear mitotic apparatus protein NuMA and disorganization of the mitotic spindle during mitosis, eventually causing formation of multiple micronuclei. Knockdown of endogenous Lgl (Lgl1 and Lgl2) also induced disorganization of the mitotic spindle, thereby causing formation of multiple micronuclei. The binding between Lgl2 and LGN played a role in the mitotic spindle organization through regulating formation of the LGN·NuMA complex. These results indicate that Lgl2 forms a Lgl2·Par-6·aPKC·LGN complex, which responds to mitotic signaling to establish normal cell division.

The gene product of Drosophila Lethal (2) giant larvae (D-Lgl) 1 is essential for development of polarized epithelia and for cell polarity associated with asymmetric cell division of neuroblasts during fly development (1)(2)(3)(4)(5). D-Lgl is in the same genetic pathway as Discs-large (Dlg) and Scribble, which are necessary for establishing and maintaining the basolateral membrane domain and basal protein targeting (3,5,6), and functions competitively with Crumbs/Stardust and Par-3⅐Par-6⅐aPKC protein complexes that are necessary for the apical membrane domain (7)(8)(9).
Mammals have two Lgl homologs, Lgl1 (Mlgl, Hugl) and Lgl2. Lgl1 and Lgl2 directly interact with the Par-6⅐aPKC protein complex (7,10). Loss of Lgl1 in mice results in formation of neuroepithelial rosette-like structures, similar to the neuroblastic rosettes in human primitive neuroectodermal tumors. The newborn Lgl1 knock-out mice develop severe hydrocephalus and die neonatally. Due to loss of mitotic spindle orientation, a large proportion of neural progenitor cells fail to exit the cell cycle and differentiate, and instead, continue to proliferate (11). These results have revealed a critical role for mammalian Lgl1 in the regulation of mitotic spindle orientation, proliferation, differentiation, and tissue organization of neuroepithelial cells. Although there is growing information on the protein partners of Lgl, the molecular details of the roles of Lgl in mitotic spindle orientation and cell cycle, however, remain obscure.
In Drosophila neuroblasts, the rotation of the spindle requires the activities of Par-3, Inscuteable (Insc), and Partner of inscuteable (Pins) (12). Although no mammalian gene similar to Insc is known, mammalian Par-3 participates in the establishment of epithelial cell polarity and several mammalian proteins related to Pins have been identified (13,14). The mammalian Pins homolog, LGN, binds the nuclear mitotic apparatus protein NuMA and plays a key role in spindle pole organization during mitosis in mammalian cells (15,16). NuMA is a large coiled-coil and microtubule (MT)-binding protein that organizes the spindle poles orientation (17,18). The binding of NuMA to LGN is necessary for the binding of LGN to the ␣ subunits of heterotrimeric G-protein (G␣) at the cell cortex (19). The ternary NuMA⅐LGN⅐G␣ complex regulates the interaction of aster MTs with the cell cortex, thereby causing the chromosome segregation during mitosis (19).
In this study, we demonstrated that Lgl2 bound LGN directly. Lgl2 formed a complex with aPKC, Par-6, and LGN favorably in mitotic cells. The inhibition of the binding between Lgl2 and LGN caused the mitotic spindle disorganization and the chromosome mis-segregation through affecting the formation of the LGN⅐NuMA complex.
Knockdown of Lgl and LGN by the RNA Interference Method-The mammalian expression vector, pBS-H1 (23), was used for expression of small interfering RNA in HEK293 cells. The following inserts were used: human LGN gene-specific insert was a 19-nucleotide (nt) sequence corresponding to nts 124 -142 (GCTGCAGTTCAAGTTGGAA) of human LGN cDNA; human Lgl1 gene-specific insert was a 19-nt sequence corresponding to nts 2160 -2178 (CCTATACTTTGCCGACACA) of human Lgl1 cDNA; human Lgl2 gene-specific insert was a 19-nt sequence corresponding to nts 605-623 (ACCAGATCCTGATCGGCTA) of human Lgl2 cDNA; and a control (Scramble) insert was a 19-nt sequence (TCTAACAGTGTCCGAGCCA) with no significant homology to any mammalian gene sequence, all of which were separated by a 9-nt noncomplementary spacer (TTCAAGAGA) from the reverse complement of the same 19-nt sequence.
Antibodies-The GST fusion fragment of LGN (aa 360 -677) was produced in Escherichia coli, purified, and used as antigen to raise a polyclonal antibody (pAb) in rabbit. The rabbit anti-LGN pAb was affinity-purified using MBP-LGN-C (aa 360 -677) immobilized on Amino-link agarose beads (Pierce). A mouse anti-HA monoclonal Ab (mAb) was purchased from Babco. A rat anti-HA mAb was purchased from Roche Applied Science. A mouse anti-aPKC mAb was purchased from Transduction Laboratories. A goat anti-Par-6 pAb was purchased from Santa Cruz Biotechonology. A mouse anti-NuMA mAb was purchased from Calbiochem. A mouse anti-␣-tubulin mAb was purchased from Sigma. A mouse anti-actin mAb was purchased from Chemicon. Rabbit anti-Lgl1 and Lgl2 pAbs were made as described previously (10).
Pull-down Assay-To determine the binding site of Lgl2 to LGN, HEK293 cells were transfected with pCMV-HA-Lgl2-Full, -N, and -C. After 48-h incubation, the cells were harvested and suspended in Buffer A (20 mM Tris-HCl at pH 8.0, 1 mM EDTA, 1 mM dithiothreitol, 150 mM NaCl, 1% Triton X-100, 10 M ␣-phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin). The cell extract was obtained by centrifugation at 100,000 ϫ g for 15 min. The extract was applied onto the full-length or truncated form of GST-LGN (500 pmol) immobilized on glutathione-Sepharose beads (Amersham Biosciences) (100 l wet volume). After the beads were extensively washed with Buffer A, the bound proteins were eluted by boiling in the SDS sample buffer (60 mM Tris-HCl at pH 6.7, 3% SDS, 2% 2-mercaptoethanol, and 5% glycerol) for 10 min. The samples were then subjected to SDS-PAGE, followed by Western blotting with the anti-HA mAb.
Direct Binding of Lgl2 to LGN-MBP-Lgl2-C or MBP-Lgl2-Full was incubated with GST-LGN-Full or GST alone (20 pmol each) immobilized on 40 l (wet volume) of glutathione-Sepharose beads in 400 l of Buffer B (20 mM Tris-HCl at pH 8.0, 27.5 mM NaCl, 25 mM KCl, and 0.1% Triton X-100) at 25°C for 2 h. After the beads were extensively washed with Buffer B, the bound proteins were eluted by boiling in the SDS sample buffer. The samples were then subjected to SDS-PAGE, followed by staining with Coomassie Brilliant Blue. The amount of bound MBP-Lgl2 was determined by comparing the intensity of its band with those of various amounts of bovine serum albumin using a densitometer Fluorchem TM (Alpha Innotech Corp.). The K d value was calculated by Scatchard analysis.
Cell Culture, Cell Cycle Synchronization, and Transfection-HEK293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum. To synchronize the cells in mitosis, thymidine and nocodazole treatment was performed as described (24). Briefly, HEK293 cells were cultured in DMEM containing 5 mM thymidine (Sigma) for 16 h and incubated in fresh DMEM containing 10% fetal calf serum for 8 h. After the incubation, nocodazole (Sigma) was added at a final concentration of 0.33 M, and the incubation was continued for another 4 h. HEK293 cells were transfected with the full-length or truncated form of pCMV-HA-Lgl2 using the Lipofectamine 2000 (Invitrogen). After 48-h incubation, the cells were seeded at a density of 2.5 ϫ 10 5 cells per dish onto 35-mm collagen type I-coated dish. At 24 h after the seeding, the cells were fixed and stained with DAPI (Nacalai Tesque) and the rat anti-HA mAb.
Co-immunoprecipitation Experiments-HEK293 cells treated with or without the thymidine and nocodazole treatment were extracted by the addition of Buffer C (25 mM Tris-HCl at pH 8.0, 27.5 mM NaCl, 25 mM KCl, 10% sucrose, 0.1% Nonidet P-40, 50 mM NaF, 2 mM Na 3 VO 4 , 10 mM EDTA, 10 M ␣-phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin). The cell extract was obtained by centrifugation at 100,000 ϫ g for 60 min. The extract was incubated with the anti-LGN pAb (5 g) or the anti-Lgl2 pAb (5 g) at 4°C for 18 h. Immunocomplexes were then precipitated with protein A-Sepharose CL-4B beads (Amersham Biosciences). After the beads were extensively washed with Buffer C, the bound proteins were eluted by boiling in the SDS sample buffer. The samples were then subjected to SDS-PAGE, followed by Western blotting with the anti-Lgl2, anti-aPKC, anti-Par-6, or anti-LGN Abs.
Other Methods-Immunofluorescence microscopy of cultured cells was done as described (25). Protein concentrations were determined with bovine serum albumin as a reference protein (26). SDS-PAGE was done as described by Laemmli (27).

RESULTS AND DISCUSSION
To identify a Lgl2-binding protein(s), we performed the yeast two-hybrid screening using a human brain library with fulllength Lgl2 as a bait. We isolated full-length of LGN (Fig. 1A).
LGN has a modular domain structure consisting of seven tetratricopeptide repeats (TPR) and four G-protein regulatory (GPR) or GoLoco domains. The TPR domain may serve as a regulatory domain for the interaction between the GPR domain and G␣ (28) or may play an important role in the localization of LGN (28). The GPR domain interacts with G␣ and stabilizes the GDP-bound form of G␣ (29,30).
We next confirmed the binding of Lgl2 to LGN and determined their binding regions by a pull-down assay. HA-tagged full-length Lgl2 expressed in HEK293 cells bound the N-terminal fragment of LGN (LGN-N, aa 1-365) but not the C-terminal fragment (LGN-C, aa 360 -677) (Fig. 1B). HA-tagged C-terminal fragment of Lgl2 (Lgl2-C, aa 636 -1015), but not N-terminal fragment of Lgl2 (Lgl2-N, aa 1-640), bound the N-terminal fragment of LGN (Fig. 1C). These results indicate that the C-terminal region of Lgl2 binds the N-terminal region of LGN.
To confirm the direct binding of Lgl2 to LGN, we performed affinity chromatography using the pure recombinant proteins of Lgl2 and LGN. MBP-Lgl2-C bound to immobilized GST-LGN-full-length but not GST ( Fig. 2A). MBP-Lgl2-full-length bound to GST-LGN-full-length but less efficiently than MBP-Lgl2-C. There might be a conformational change required for efficient binding between Lgl2-full-length and LGN-full-length. As Lgl2 has been shown to bind to Par-6 through its N-terminal WD repeats and aPKC through its phosphorylation sites (10), these bindings might help to open the C-terminal tail of Lgl2 for its binding to LGN. The binding of MBP-Lgl2-C to GST-LGN-full-length was dose-dependent and saturable (Fig.  2B). Scatchard analysis demonstrated a single class of binding site with a K d value of 56 nM. These results indicate that Lgl2 directly binds LGN.
To examine the in vivo binding of Lgl2 to LGN, a co-immunoprecipitation assay was performed using the HEK293 cell lysate. When Lgl2 was immunoprecipitated with the anti-Lgl2 pAb, LGN was co-immunoprecipitated (Fig. 3A). Lgl2 was reciprocally co-immunoprecipitated with the anti-LGN pAb. In addition to LGN, Par-6 and aPKC were co-immunoprecipitated with Lgl2, consistent with the earlier observations that Lgl2 forms a complex with Par-6 and aPKC (7,10). Thus, Lgl2 forms a novel complex with Par-6, aPKC, and LGN.
LGN has been shown to play an important role in spindle pole orientation and to localize at cell periphery during mitosis (29). Therefore, we next examined by a co-immunoprecipitation assay whether the amount of the Lgl2⅐Par-6⅐aPKC⅐LGN complex is changed during mitosis. To produce large numbers of mitotic cells, HEK293 cells were treated with thymidine and nocodazole. Over 80% of the cells showed mitotic cell phenotype (data not shown). The Lgl2⅐Par-6⅐aPKC⅐LGN complex was more predominantly formed in the mitotic cells than in the control cells (Fig. 3A). We furthermore examined by immunostaining the localization of these proteins in the interphase and the metaphase of HEK293 cells. As shown previously (29), LGN localized in the cytoplasm in the interphase and at the cell periphery in the metaphase (Fig. 3B). aPKC co-localized with LGN at the cell periphery in the metaphase. Lgl2 and aPKC co-localized in the perinuclear structure and faintly at the cell-cell contact sites in the interphase. In the metaphase, Lgl2 and aPKC as well as LGN localized at the cell periphery. These immunostaining data are consistent with the biochemical data shown in Fig. 2A and suggest that Lgl2 forms a complex with LGN, aPKC, and Par-6 during mitosis.
Dysfunction of LGN has been shown to have the phenotype of multiple micronuclei due to chromosomal mis-segregation and defect in cell division through mis-localization of mitotic spindle regulator NuMA, a LGN binding partner (15,16). Therefore, to investigate the functional significance of the binding between Lgl2 and LGN, we examined the effect of overexpression of Lgl2-C, the LGN-binding fragment, on cell division. About 8% of the cells overexpressing Lgl2-C showed the presence of multiple micronuclei (Fig. 4A). Less than 0.5% of the cells overexpressing HA-control, Lgl2-full-length, and Lgl2-N contained micronuclei. We next examined whether the Lgl-Cinduced formation of multiple micronuclei is due to chromosomal mis-segregation and defect in mitotic spindle organization through dysfunction of NuMA. Both NuMA and MTs of mitotic spindle were mis-localized in the cells overexpressing Lgl-C.
LGN, but not aPKC and Par-6, was not localized at the cell cortex during mitosis in the cells overexpressing Lgl-C ( Fig.  4B and data not shown). To further confirm this result, we performed a loss-of-function analysis by use of the RNA interference (RNAi) method. We knocked down endogenous Lgl (Lgl1 and Lgl2) and LGN by the RNAi method (Fig. 4C). The Lgl-knockdown cells had apparently a similar multinucleate phenotype as that of the LGN-knockdown cells (Fig. 4A). The percentage of the Lgl-knockdown cells showing the multinucleate phenotype was 6% and slightly less than that of the LGNknockdown cells. Both NuMA and MTs of mitotic spindle were mis-localized in the Lgl-knockdown cells similarly to those of the LGN-knockdown cells. In addition, LGN, but not aPKC and Par-6, was not localized at the cell cortex during mitosis in the Lgl-knockdown cells. These results suggest that Lgl plays a role in chromosome segregation through regulating the LGN and NuMA localization during mitosis. In addition, to examine the Lgl-C effect on binding between endogenous LGN and NuMA in the cells, we examined the amount of NuMA bound to endogenous LGN by a co-immunoprecipitation assay using the anti-LGN pAb. The LGN⅐NuMA complex was more predominantly formed in the cells overexpressing Lgl-C than in the control cells (Fig. 4D). Lgl2-C indeed bound to LGN more predominantly than Lgl2-full-length. These results indicate that the binding of Lgl2 to LGN is important for the formation of the LGN⅐NuMA complex to regulate chromosome segregation.
Previous studies have shown that Lgl2 is largely expressed at the cell-cell contact sites in epithelial cells (10). However, there have been no analysis of potential change in the pattern of Lgl2 complex formation during mitosis. We have first presented here that Lgl2 forms a novel complex with LGN, Par-6, and aPKC and that this complex formation is enhanced during mitosis. This complex shows cortical localization in mitotic cells. The functional significance of the cell cycle-dependent cortical localizaiton of this complex is not yet clear, but the localization of LGN at the cell cortex during mitosis is somewhat similar to what is described for Drosophila Pins. In Drosophila, Pins is normally found in the lateral cortex of epithelial cells and only become asymmetrically localizes upon the expression of inscuteable in neuroblasts, for which a mammalian homolog has not been found so far (31). Pins is also dependent on heterotrimeric G-protein activity for its localization (30). Furthermore, Pins and LGN have been shown to bind G␣ through their C-terminal GPR domain (29,30). They behave as GDP dissociation inhibitors of the G␣. For the efficient binding of LGN to G␣ at the cell cortex, the binding of LGN to NuMA seems to be prerequisite (19). LGN normally folds back onto itself and shows a closed conformation. The binding to NuMA triggers a partially open state of LGN for further interaction with G␣. The formation of ternary NuMA⅐LGN⅐G␣ complex regulates the interaction of aster MTs with the cell cortex, thereby causing the chromosome segregation during mitosis. The binding of Lgl2 to LGN enhances the formation of NuMA-LGN complex. Lgl2 seems to induce the full activation of LGN in cooperation with NuMA at the cell cortex during mitosis. Formation of the Lgl2⅐Par-6⅐aPKC⅐LGN complex might activate the activity of LGN during mitosis. Therefore, the Lgl2⅐Par-6⅐aPKC⅐LGN complex may play a role in coupling the Par signaling pathway to LGN and the heterotrimeric G-proteins, to ensure that bipolar spindles are correctly organized and oriented for normal cell division.