| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Pharmacology, Tokyo Women’s Medical University, Shinjuku-ku, Tokyo, Japan
1Correspondence: Department of Pharmacology, Tokyo Women’s Medical University, 8–1 Kawada-cho, Shinjuku-ku, Tokyo 162-8666, Japan. E-mail: ymaru{at}research.twmu.ac.jp
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-tubulin, MKLP1, and p80 katanin, and is involved in cytokinesis. Moreover, RNAi-mediated disruption of LAPSER1, which is accompanied by the mislocalization of p80 katanin, results in malformation of the central spindle. Significantly, the enhanced expression of LAPSER1 induces binucleation and renders the cells resistant to oncogenic transformation. In cells transformed by the v-Fps oncogene, overexpressed LAPSER1 induces abortive cytokinesis, followed by mitotic catastrophe in a p80 katanin-dependent manner. Cells that are rescued from this apoptotic pathway with Z-VAD-fmk display karyokinesis. These results suggest that LAPSER1 participates in cytokinesis by interacting with p80 katanin, the disruption of which may potentially cause genetic instability and cancer.—Sudo, H., Maru, Y. LAPSER1 is a putative cytokinetic tumor suppressor that shows the same centrosome and midbody subcellular localization pattern as p80 katanin.
Key Words: tetraploid • central spindle • microtubule • Fps • MKLP1
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The LAPSER1 and LZTS1 genes have been mapped to 10q24.3 and 8p21, respectively, and loss of heterozygosity (LOH) in these two regions is frequently found in patients with prostate adenocarcinomas as well as other cancers (1)
. A rat homologue of LZTS1 is highly expressed in the cerebrum and is reported to play a role in synaptic plasticity; the C-terminal leucine-zipper motifs are known to be required for intracellular localization of the protein (2)
. Alterations in expression and somatic mutations of the human LZTS1 gene have been found in some cancers. Forced expression of LZTS1 in LZTS1-negative breast cancer MCF7 cells results in the accumulation of cells that have been arrested in late S-G2/M. The association of LZTS1 with microtubules and cdc2 also suggests its involvement in the regulation of mitosis (3)
.
LAPSER1 is expressed in most normal tissues, with the highest abundance in prostate. Although no mutations have been reported so far in the LAPSER1 gene, loss of expression has been observed in some prostate cancer cell lines (1)
. Moreover, similar to LZTS1, the overexpression of LAPSER1 in various cancer cell lines results in reduced cell growth (1)
, but the molecular mechanisms underlying this suppression are unknown. In recent years, several studies have revealed a link between the centrosome and cytokinesis, and it has been shown that acentrosomal cells fail to complete cytokinesis (4)
. In another study, it has been demonstrated that the maternal centriole migrates to the intercellular bridge during cytokinesis (5)
. Centriole repositioning also correlates with bridge narrowing and microtubule depolymerization, whereas movement of the centriole away from the bridge is associated with cell cleavage or abscission.
Polo-like kinase-1 (PLK1), regarded as a novel drug target for prostate cancer (6)
; it is essential for mitotic spindle formation and is associated with docking proteins such as mitotic kinesin-like protein-1 (MKLP1) on the central spindle. Coordinated assembly and disassembly of microtubules at the kinetochores and centrosomes, respectively, are required for the organization of the spindle structure. In addition, the microtubule-severing protein katanin has been reported to localize to the centrosome and to play a role in mitotic spindle function (7)
.
The predominant phenotype that results from the functional abrogation of most proteins involved in cytokinesis is binucleation, which is a form of tetraploidy. Tetraploid cells are genetically unstable and have been shown to promote tumorigenesis in p53-null cells (8)
. Hence, there is increasing interest in the function of tumor suppressors that operate during cytokinesis (9
, 10)
. In our present study, we first demonstrate such tumor-suppressive properties for LAPSER1 on its overexpression and provide direct evidence that it influences cytokinesis by colocalizing and interacting with the essential mitotic regulators in the centrosome and midbody. Second, we reveal that the apoptosis-inducing activity of LAPSER1 in v-Fps-transformed cells is at least partially dependent on these interactions. Third, although the RNAi-mediated knockdown of LAPSER1 did not readily transform cells, it resulted in a disturbance of cytokinesis that potentially precedes genomic instability.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
3'): LAPSER1F NotI (gcccgcggccgcgatatcgccaccatggccattgtgcagact) and LAPSER1R BamHI (cgggggatccgatctcagtggcagt), followed by cloning into PCR TOPO 2.1 (Clontech) in the sense direction, digestion with BamHI, and subcloning into pCMV-Tag1 (pCMV-LAPSER1). C-terminally flag-tagged LAPSER1 was cloned into an adenoviral vector. For GFP-LAPSER1 construction, the insert was excised from PCR TOPO 4.0 with EcoRI and subcloned into pQBI25fc-3 in the sense direction. This N-terminally GFP-fused LAPSER1 was then transferred into both an adenoviral and a tetracycline-regulated vector. For GFP-LAPSER1-
LZ construction, the LAPSER1 PCR TOPO 4.0 insert was partially digested with HindIII, and the resulting 1012 bp fragment was inserted into pQBI25fc-3 in the sense direction. This construct was then digested with NotI, blunt ended, and self-ligated for adjustment of the open reading frame (ORF). The resulting N-terminal GFP-LAPSER1-LZ was then cloned into an adenoviral vector using the TAKARA adenovirus Expression Vector kit (TaKaRa, Otsu, Shiga, Japan). The titers of adenovirus expressing LacZ, LAPSER-flag, GFP-LAPSER1, and GFP-LAPSER1-LZ were 2.4 x 1010, 6.1 x 1010, 4.4 x 1010, and 1.0 x 1010 cfu, respectively. The LAPSER1-flag fragment was also cloned using EcoRI into pcDNA3.1 in the sense direction. For GFP-LAPSER1-N construction, the LAPSER1 insert in PCR TOPO 4.0 was partially digested with HindIII and EcoRI, and the resulting 1 kbp fragment was cloned into pQBI25fc-3. The mutant ORF was generated as described above using NotI.
The mammalian expression vectors were used directly in COS7 cell transfection experiments using Superfect (Qiagen, Valencia, CA, USA). In Rat1 cells, the Cell Phect Transfection Kit (Amersham Biosciences, Arlington Heights, IL, USA) was used for calcium phosphate transfection, followed by G418 selection. We obtained several independent clones from the Rat1 transfections, as well as a mixed population of cells that stably expressed LAPSER1-flag (Rat 1/LAPSER1-flag). These clones behaved in a similar fashion, and a representative population was used in subsequent gene transfer experiments. In the tetracycline-regulated system, we used the pUHD 10–3 IRES GFP expression vector kindly provided by Dr. Owen Witte at UCLA (11)
. Stable pUHD 10–3 (tetracycline regulator) -transfected Rat 1 cells were established by puromycin selection. GFP-LAPSER1 was cloned into a modified version of pUHD 10–3 IRES GFP vector in which IRES-GFP was eliminated. We obtained several independent clones from these transfections in which GFP-fused LAPSER1 was expressed under the control of the tetracycline regulator (GFP-LAPSER1/Tet) by selection with G418, after the cotransfection of pUHD10–3/GFP-LAPSER1 and pSV2NEO into stable pUHD10–3-transfected Rat 1 cells. A representative population of cells was then used in all subsequent experiments.
Cell culture and flow cytometry
The human osteoblast-derived sarcoma cell line Saos-2 was obtained from the Institute of Development, Aging and Cancer, Tohoku University (Sendai-shi, Japan). Cells were maintained in D-MEM (NISSUI) supplemented with 10% fetal bovine serum, 0.2% NaHCO3, 4 mM glutamine, 100 U/ml penicillin, and 100 µg/ml streptomycin. For flow cytometry analysis using propidium iodide (PI), 1 x 107 cells were fixed in cold 70% ethanol, washed in phosphate-buffered saline (PBS), then stained in PI at 50 µg/ml in PBS supplemented with RNase A for 30 min at 37°C. Data were collected using EPICS XL (Beckman Coulter, Fullerton, CA, USA).
RNAi technology
To generate human LAPSER1 siRNAs, the oligomers sense: gcagcaggagaagcggcaauTT and antisense: auugccgcuucuccugcugcAG were annealed. For rat katanin siRNA, the oligomers sense: caaccugugguccauaaacaaTT and antisense: auuguuuauggaccacagguugAC, and for human MKLP1 siRNA, the oligomers sense: ccgaaauggagacuauaagTT and antisense: cuuauagucuccauuucggTT, were synthesized (lowercase indicates ribonucleic acids, capitals denote deoxyribonucleic acids). Only sense oligomers were annealed for the control. Approximately 3.0 x 105 cells/well were then plated on collagen-coated 6-well plates the day before transfection. On the day of transfection, medium was replaced with DMEM/10% FCS and 100 nM of annealed oligomers was transfected using TRANS IT TKO reagent (Mirus Bio Corporation, Madison, WI, USA). Analyses were performed 48 h later.
Antibodies and reagents
Rabbit polyclonal antibodies against Akt (Cell Signaling, Danvers, MA, USA), phospho-Akt (S473) (New England Biolabs, Beverly, MA, USA), cleaved caspase-3 (Cell Signaling), Cdc2 (Santa Cruz, Santa Cruz, CA, USA), and phospho-Cdc2 (Y15) (Santa Cruz) were used. Goat polyclonal antibodies were used against p53 and MKLP1 (both from Santa Cruz), as were mouse monoclonal antibodies against flag (M2) (Sigma, St. Louis, MO, USA), Xpress (Invitrogen, Carlsbad, CA, USA), actin (Chemicon, Temecula, CA, USA), phosphotyrosine (PY-20) (ICN Biomedicals, Irvine, CA, USA), tubulin (NeoMarkers, Suffolk, UK), ß-tubulin (Sigma, St. Louis, MO, USA), -tubulin (Sigma), cyclinB1 (NeoMarkers), acetylated tubulin (Sigma), phosphorylated histone H3 (S10) (Upstate, Lake Placid, NY, USA), and Aurora B (BD Biosciences, Bedford, MA, USA). Chicken polyclonal antibodies were used against p80 katanin (GenWay Biotech, San Diego, CA, USA). Nocodazole and Z-VAD-fmk were obtained from Sigma, purvalanol A from Calbiochem, and bongkrekic acid from Biomol (Plymouth Meeting, PA, USA). Rabbit polyclonal anti-LAPSER1 antibodies were raised against a bacterially generated GST fused to the C-terminal 220 amino acids of rat LAPSER1. We also used this construct for the in vitro binding assays.
Immunocytochemistry and fluorescent imagery
For staining of -tubulin in Saos-2 cells, 2.8 x 105 cells were plated onto 3.5 cm polylysine-coated glass-bottom dishes (Matsunami, Tokyo, Japan). This was followed by 70% methanol fixation for 20 min at –20°C, rinsing for 10 min with PBS, and incubation with anti--tubulin antibodies (diluted 1:300 in 4% bovine serum albumin (BSA)/PBS) for 60 min at 37°C. After further washing with PBS, cells were stained with secondary antibody, and in some cases with 2 µM TO-PRO-3 (Molecular Probes, Carlsbad, CA, USA) for 60 min. The cells were washed again in PBS, and in other cases stained with DAPI (4',6-diamidino-2-phenylindole dihydrochloride hydrate). Fluorescence signals were detected using a confocal laser scanning microscope, LSM 510 Meta (Carl Zeiss, Thornwood, NY, USA). The same procedure as above was performed for staining with anti-LAPSER1 or anti-p80 katanin antibodies. For staining with anti-flag, anti-acetylated tubulin, or anti-tubulin, we fixed Rat 1/LAPSER1-flag cells or Saos-2 cells with 4% paraformaldehyde in PBS for 1 h at room temperature. Samples were then processed as described for -tubulin staining. Fluorescence images of GFP-fused LAPSER1 constructs were also observed as described above. We used the in situ apoptosis detection kit (TaKaRa) for the TUNEL assay. For double immunostaining experiments, we used FITC-, TRITC-, or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). Image quantification was performed using NIH Image.
Immunoblotting
Approximately 3.0 x 105 cells were plated in 6-well plates and harvested 24 h after adenoviral infection. For knockdown studies, 1.0 x 105 cells/well were plated on collagen-coated 6-well plates, transfected, and harvested 48 h later. Lysates were extracted in 50 µl of Nonidet P-40 lysis buffer (50 mM HEPES, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, and 0.5% Nonidet P-40) containing 2.5 µg/ml of leupeptin, 5 µg/ml of aprotinin, 0.2 mM phenylmethylsulfonyl fluoride, and 50 µl of 2 x sample buffer (2% sodium dodecyl sulfate (SDS), 20% glycerol, 1% bromphenol blue, 2% mercaptoethanol, and 1.52% Tris base). Western blot was then performed as described previously (12)
.
Cell growth assay
For cell growth analysis, 5 x 104 cells were plated in triplicate in 6-well plates and cell numbers were determined at different time points after staining with Trypan blue. For focus formation assays, we plated 1 x 105 Rat 1/LAPSER1-flag or mock-transfected cells in 6 cm dishes. Twenty-four hours later, these cells were transfected with 3 µg each of viral oncogene expression constructs using a Cell Phect Transfection Kit, followed by glycerol shock (13)
. The culture medium was changed every 3 or 4 days; 3 wk after transfection, foci were fixed with 3.7% formaldehyde and stained with crystal violet. Focus-forming unit values were calculated per 1 µg of transfected DNA. Soft agar assays were performed as described previously (14)
.
Yeast two-hybrid screening
For bait vector construction, the full-length untagged LAPSER1 fragment was excised from pCMV-LAPSER1 by a BamHI/EcoRV double digestion and subcloned into the pGBKT7 bait vector (Clontech). This construct was confirmed by sequencing. Approximately 1 x 106 clones were screened as described previously (12)
. The identified katanin con80 (con80) was then His6-tagged, bacterially expressed, and purified using an Ni column (Ni-NTA Spin Kit, Qiagen). The protein was refolded by sequential dilutions in urea (final: 1M) and dialysis with dilution buffer (pH:7.7, containing 4, 2, and 1M urea, 10% glycerol, 40 mM Tris, 5 mM EDTA, 1% triton X100, and 1 µM PMSF) at 1.5 h intervals. Con80 expression in COS7 cells was performed by transfection with the pEF-HisA (Invitrogen) mammalian expression vector.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
80 kDa in both total lysates and anti-LAPSER1 immunoprecipitates from Rat 1 cells, but preincubation of the antibody with the purified GST-LAPSER1 C terminus eliminates this band. Thus, we confirmed that the 80 kDa species is endogenous LAPSER1. We addressed the question of whether rat LAPSER1 displayed tumor suppressive activity, as reported previously (1)
. Stable overexpression of flag-tagged LAPSER1 in Rat 1 cells was detected by Western blot with anti-flag or anti-LAPSER1 antibodies (Fig. 1
A). Rat 1 cells were utilized because of their sensitivity to oncogene expression. To test whether authentic oncogenes could transform these cells when LAPSER1 is overexpressed, both Rat 1/LAPSER1-flag and control cells (Rat 1/Mock) were transfected with v-Ras and v-Src, and subjected to focus-forming assays. The efficiency of transformation was 1.6 (v-Ras) and 2.5 (v-Src) times lower in Rat 1/LAPSER1-flag than in the control (Fig. 1B
), suggesting that LAPSER1 may indeed confer resistance to oncogenic transformation.
|
We next employed an adenovirus-mediated LAPSER1 expression system in three transformed cell lines: rat 3Y1 cells transformed by v-Fps (3Y1/v-Fps), Rat 1 cells transformed by v-Abl (Rat 1/v-Abl), and Saos-2 cells. LAPSER1 was successfully overexpressed in these cells (Fig. 1C
). Using this transient expression system, we then tested the effects of LAPSER1 overexpression on colony formation in soft agar (Fig. 1D
). In the case of 3Y1/v-Fps and Rat 1/v-Abl, LAPSER1 expression decreased the colony numbers compared with the LacZ control. In the Saos-2 cells, however, we could not detect any macroscopic colony formation in either the LAPSER1- or LacZ-expressing cells. We therefore tested the effects of LAPSER1 expression on cell proliferation in liquid culture (Fig. 1E
) and found that it indeed suppresses cell growth. Furthermore, we established Rat 1 cells in which GFP-LAPSER1 was expressed under the control of a tetracycline (tet) -regulated CAG promoter (Rat 1/GFP-LAPSER1/Tet) and confirmed by Western blot with anti-GFP and anti-LAPSER1 antibodies that GFP-LAPSER1 is expressed only in the absence of tetracycline (Fig. 1F
). Although we found no differences in cell growth in liquid cultures between the tet (+) and tet (–) conditions, the results of our focus-forming assay support the idea that LAPSER1 expression suppresses transformation (Fig. 1G, H
). The expression levels of exogenous LAPSER1 in three different expression systems were calculated to be 3- to 15-fold higher than the endogenous protein, as shown in supplemental Fig. 1B. Hence, our data strongly suggest that LAPSER1 functions as a tumor suppressor.
Overexpression of LAPSER1 results in central spindle defects
An important question that arose from our expression analyses regarded the mechanism(s) underlying its tumor suppressor activity. To gain insight into these mechanistic aspects of LAPSER1, we analyzed the phenotypes that resulted from its overexpression in Saos-2 cells, as inhibitory effects of LAPSER1 on the proliferation of these cells were observed in our adenovirus overexpression experiments. The most striking phenotype that resulted from this experiment was binucleation (Fig. 2
A); we confirmed this by flow cytometry, which showed a 15% increase in the 4N DNA content compared with the LacZ-expressing control cells (Fig. 2B
). We also expressed GFP-tagged LAPSER1 in this adenoviral system and obtained similar results (supplemental Fig. 2A–C). Determination of the number of binucleates also revealed a >10% increase in LAPSER1-expressing cells compared with the control (Fig. 2C
), consistent with our flow cytometry results. This phenotype was accompanied by increased levels of phosphorylated histone H3 at serine 10, a well-known mitotic marker (Fig. 2D
), suggesting that the binucleates were generated after mitosis and not by fusion or endoreplication. We thus focused our subsequent observations on mitosis.
|
Confocal microscopic analysis of the LAPSER1-expressing cells using antibodies against either tubulin (data not shown) or acetylated tubulin (a stable microtubule marker) revealed almost no differences between the LAPSER1-flag- and LacZ-infected cells in prometa/meta- and anaphase. Surprisingly, however, we found that the cells expressing LAPSER1-flag had few or very small central spindles during cytokinesis, resulting in no separation of the cytoplasm compared with the LacZ control (Fig. 2E
). This binucleation was enhanced by nocodazole in a synergistic fashion, suggesting that the common target is the central spindle (Fig. 2C
). We also observed a mild accumulation of both cyclin B1 and a tyrosine 15-dephosphorylated form of cdc2 (Supplemental Fig. 2D) and no significant differences in the mitotic index between the overexpressing and control cells (2.9% and 3.2%, respectively). It is noteworthy that all of the phenotypes shown in Fig. 2A-E
were found to be dependent on the MOI of the adenoviral infection and were not observed with an MOI of <100.
Subcellular localization of endogenous LAPSER1
Based on the phenotypes observed in the LAPSER1-overexpressing cells, we examined the subcellular distribution of the endogenous protein, particularly during mitosis. During cytokinesis, LAPSER1 was found to localize at the center of the midbody (Fig. 3
A-1, 2), and in some cells at both the centrosomes and the midbody (Fig. 3A
-3). During anaphase (Fig. 3A
-4) and prometa/metaphase, LAPSER1 localizes at the centrosomes but never in the midline or chromosomes. To confirm these observations, we next performed double staining with anti--tubulin and anti-LAPSER1 antibodies. During interphase, LAPSER1 was observed to colocalize with -tubulin, thus confirming its centrosomal distribution (Fig. 3B
-1). During cytokinesis, LAPSER1 signals were evident at the poles (Fig. 3B
-2), where they colocalized with -tubulin. At later stages, these signals were found in the central midbody between the -tubulin signals at the minus ends of the central spindles (Fig. 3B
-3). This staining pattern was not observed when the antibody was preincubated with purified GST-LAPSER1 C terminus at all phases of the cell cycle (Supplemental Figs. 3, 4). Anti-LAPSER1 RNAi decreased levels of endogenous LAPSER1 proteins in >90% in Saos-2 cells (Fig. 3C
, upper panels) and dramatically reduced the LAPSER1 signals in centrosomes during interphase (Fig. 3C
, lower panels, centrosome) and at the midbody during cytokinesis (Fig. 3C
, lower panels, midbody). We analyzed the intracellular localization of LAPSER1 relative to that of other central spindle molecules, and found positive signals around the midbody region, a structure that is deficient in acetylated tubulin staining in these cells (microtubule masked region) (Fig. 3D
). This deficiency is called epitope masking and is known to be caused by the midbody matrix (15)
.
|
We further performed double staining and found that cells in cytokinesis carrying midbody LAPSER1 signals also had strong phospho-histone H3 signals in their nuclei (data not shown). We next investigated the position of LAPSER1 relative to that of Aurora B, another well-known cytokinetic regulator. When Aurora B signals were close to the contractile ring (Fig. 3E
-1), but not when they were distant from the ring (Fig. 3E
-2), LAPSER1 was observed to colocalize with Aurora B. This suggested that an equivalent final destination for both of these proteins is achieved via different routes. Finally, we studied the effects of an MKLP1 knockdown on LAPSER1. As shown in Fig. 3F
(left panel), we could efficiently knock down MKLP1, and midbody LAPSER1 signals were almost completely absent from such cells (Fig. 3F
, right panel).
The subcellular distribution of exogenously expressed LAPSER1 was also examined by confocal microscopy. In each cell used in the in vitro transformation assays, exogenous and endogenous LAPSER1 showed similar subcellular localizations (supplemental Fig. 5). Our data thus suggest that LAPSER1 localizes at the centrosomes during interphase and migrates to the midbody during cytokinesis, where it is integrated in the cytokinetic machinery.
LAPSER1 knockdown phenotypes
We hypothesized from our initial data that endogenous LAPSER1 plays a role in regulating the central spindle, and tested this possibility in Saos-2 cells. Surprisingly, we observed that during cytokinesis, LAPSER1 knockdown Saos-2 cells show weakened acetylated tubulin or tubulin signals in their central spindles (Fig. 4
A). This is consistent with our current findings that anti-MKLP1 RNAi eliminates the LAPSER1 signals in the midbody (Fig. 3F
) and with a previous report that MKLP1 knockdown results in the lack of a central spindle (15)
. This phenotype was rescued by adenovirus-mediated mild GFP-LAPSER1 expression with an MOI of 50, which is resistant to anti-LAPSER1 RNAi because of species differences within the target sequences (a difference of three nucleotides). We quantified this phenotype using the acetylated tubulin signal intensities and the area occupied by these signals. Among the population of LAPSER1 knockdown cells in early cytokinesis (with a rounded morphology; see below), >90% of the cells were below 4000 pixels and 50% were <150 arbitrary units in intensity, whereas 50% of the control cell population showed >4000 pixels and >90% were above 150 arbitrary units in intensity (Fig. 4B
).
|
We next performed phase-specific phenotype analysis by determining the cell number in prometa/metaphase, anaphase, cytokinesis/telophase. According to our quantification method shown in Fig. 4B, we
assigned the central spindles in anaphase and cytokinesis into two categories: "strong" defining cells with a product of intensity and pixel levels of >800,000, and "weak" defining cells with a value less than this. We separated cytokinesis into two further and more detailed phases: "round" cells, defined as having a rounded morphology representing early cytokinesis, and "extended" cells, showing an extended periphery and late cytokinesis. This classification has proved useful in distinguishing between building and abscission of the central spindle phenotypes. The clearest differences were evident for round cytokinesis. Compared with control cell populations showing 3-fold higher levels of strong cytokinesis than weak cytokinesis, the LAPSER1 knockdown cells showed quite weak cytokinetic characteristics (Fig. 4C
, white bars). Furthermore, there was a tendency for extended weak cells, which are representative of cells without abscission of the central spindle, to be increased in number among the knockdown cells (18.5%, SD=3.8, and 23.7%, SD=3.7 for control and knockdown, respectively). The exogenous expression of GFP-LAPSER1 in the knockdown cells also reversed the phenotype in this assay, confirming that endogenous LAPSER1 plays a role in these outcomes and that they were not simply the result of off-target effects. There were no significant differences in the mitotic index between the knockdown cells and the control (1.13% and 1.03%, respectively). However, we repeatedly observed binucleated cells in a small but appreciable population of these knockdown cells (2.9%, SD=0.7, control: 1.1%, SD=0.2). These results support our hypothesis that LAPSER1 plays a role in the formation of the central spindle.
The binding partners of LAPSER1
We screened for binding partners of LAPSER1 using anti--tubulin Western blot analysis of anti-flag immunoprecipitates from Saos-2 cells infected with adeno-LAPSER1-flag virus, and found a physical interaction between -tubulin and LAPSER1-flag (Fig. 5
A). Their centrosomal colocalization was also confirmed (Fig. 3B
). Second, we performed yeast two-hybrid screening using full-length rat LAPSER1 and a human placenta cDNA library. We identified a 1.2 kb fragment containing 1258–1566 nucleotides in-frame with p80 katanin cDNA, encoding residues 420–522 of the p80 katanin protein and forming a part of its conserved 80 domain (con80). Katanin is a well-known centrosomal heterodimeric protein (16)
that consists of p60 and p80 subunits. The p80 subunit has been shown to target p60 to the centrosomes (17)
. From interphase to early mitosis (prometa/metaphase), p80 katanin has also been reported to localize to centrosomes in mammalian cells. Anti-flag immunoprecipitates from COS7 cells transfected with plasmids for the expression of katanin con80 tagged with Xpress, and infected with adeno-LAPSER1-flag virus contained katanin fragments detectable using anti-Xpress antibody (Fig. 5B
).
|
We next expressed and purified the GST-tagged LAPSER1 C terminus and His-tagged con80 products using Escherichia coli, and confirmed their physical interaction by GST pulldown (Fig. 5C
). However, since both the GST-tagged full-length and N-terminal LAPSER1 products aggregate in E. coli, it was not possible to elucidate the binding domains. The Kd value of the interaction between con80 and the C-terminal domain was found to be 30 nM, thus showing relatively tight binding. Hence, the C terminus of LAPSER1 is sufficient to bind con80 (Fig. 5D
). Anti-flag immunoprecipitates from HeLa cells infected with the adeno-LAPSER1-flag virus contained the full-length endogenous p80 katanin detected by anti-katanin antibody (Fig. 5E
). We also detected an interaction between endogenous LAPSER1 and endogenous p80 katanin in Rat 1 cells (supplemental Fig. 1A-3). Anti-flag immunoprecipitates from PC3 cells infected with the adeno-LAPSER1-flag virus contained endogenous MKLP1, supporting our earlier data showing the localization of LAPSER1 at the midbody, but not endogenous Aurora B (Fig. 5F
). We found an increase in total MKLP1, but not Aurora B, levels in the LAPSER1-flag-expressing cells compared with the control. Furthermore, these effects are inhibited by 10 µM purvalanol A, a relatively specific cdc2 inhibitor (Fig. 5G
). Taken together, our results show that LAPSER1 physically binds to the centrosomal proteins, -tubulin and p80 katanin, and to the midbody protein MKLP1.
Intracellular localization of p80 katanin
We next examined the subcellular localization of p80 katanin, which has been reported to localize to the centrosomes (17)
. However, our finding that katanin and LAPSER1 interact raised the question of whether p80 katanin also colocalizes with LAPSER1 at the midbody. Indirect immunostaining with anti-p80 katanin revealed a polar distribution of this protein in prometa/metaphase, as reported (Fig. 6
A-1) (16)
. Cells in cytokinesis showed localization of p80 katanin at the midbody (Fig. 6A
-2, 3). Although we could not knock down p80 katanin efficiently in human cells, we could do so in rat cells, and found that this significantly reduced p80 katanin signals in the midbody (see below). We conducted double staining with anti-LAPSER1 and anti-p80 katanin and cells in prophase and prometa/metaphase/then showed colocalization of these signals at the poles (Fig. 6B
-1, 2). During cytokinesis, both signals colocalized at the midbody (Fig. 6B
-3) and some cells exhibited split signals, as shown in Fig. 6B
-4, suggesting that the transition mode of both molecules is symmetrical. We repeatedly failed, however, to detect any chromosomal or spindle localization of either molecule during the transition phase, suggesting a diffuse translocation. Consistent with the physical interaction between MKLP1 and LAPSER1 (Fig. 5F
) and between LAPSER1 and p80 katanin (Fig. 5E
), we also detected colocalization of MKLP1 and p80 katanin in the midbody (Fig. 6C
). The split p80 katanin signals suggest a translocation event for this protein, after that of MKLP1, from the midline to the central midbody (Fig. 6C
, transition). In MKLP1 knockdown cells (Fig. 3F
), the midbody localization of both LAPSER1 (shown in Fig. 3F
) and p80 katanin is lost (Fig. 6D
).
|
In the LAPSER1 knockdown cells, localization of p80 katanin to the midbody is abrogated by up to 70%, whereas the centrosomal distribution remains intact (Fig. 6E
). In cytokinetic cells harboring a LAPSER1 knockdown, p80 katanin signals were found either beneath the plasma membrane of the intercellular bridge (Fig. 6E
-3) or were completely absent from the bridge (Fig. 6E
-4). On the other hand, localization of MKLP1 was unchanged in the LAPSER1 knockdown cells (Fig. 6F
). These data suggest that the interaction between LAPSER1 and p80 katanin initiates at the poles in early mitosis, forming a tight complex, and that the transition of p80 katanin to the midbody is dependent on the presence of LAPSER1. Furthermore, the midbody localization of both LAPSER1 and p80 katanin is dependent on MKLP1.
The significance of the C terminus of LAPSER1
p80 katanin binds to the C terminus of LAPSER1, which harbors three leucine zipper motifs (LZ). Generally, leucine zipper motifs provide an interface for protein-protein interactions, and so we next examined whether a full-length LAPSER1 and a C-terminally truncated form of this protein (GFP-LAPSER1-LZ) were equally active. We exogenously expressed GFP-LAPSER1, GFP-LAPSER1-LZ, and an N-terminally truncated form of LAPSER1 (GFP-LAPSER1-N) in COS7 cells as shown in Supplemental Fig. 6A. Both GFP-LAPSER1 and GFP-LAPSER1-LZ were found in the centrosomes (Supplemental Fig. 6B), but GFP-LAPSER1-N signals were not evident there. Similarly, the adenovirus-mediated expression of GFP-LAPSER1 and GFP-LAPSER1-LZ in Saos-2 cells showed that the products of both constructs were present in the centrosomes and at the midbody (Fig. 4A
, Supplemental Fig. 5B, and data not shown). Thus, GFP-LAPSER1-LZ retains the ability to migrate from the centrosomes to the midbody. Despite its apparently unaltered localization, GFP-LAPSER1-LZ expressing cells showed only a marginal increase in their 4N subpopulation (supplemental Fig. 2B, C). Furthermore, the expression of GFP-LAPSER1-LZ failed to rescue the LAPSER1 knockdown phenotype (Fig. 4)
. These results suggest that the C terminus is dispensable for the localization of LAPSER1, but indispensable for its function in the midbody.
LAPSER1-induces apoptosis in 3Y1/v-Fps cells
We have shown that adenoviral LAPSER1-flag overexpression counteracts the transformation of 3Y1 cells by the v-Fps oncogene, as determined by a soft agar colony assay (Fig. 1D
). Surprisingly, the infected cells underwent apoptosis when they were subsequently grown in liquid cultures instead of being plated onto soft agar (Fig. 7
A, B). Both flag- and GFP-tagged LAPSER1, but not GFP-LAPSER1-LZ, induce cell death. LAPSER1 fails to induce apoptosis in 3Y1 parental cells or in 3Y1 cells transformed by either v-Src or v-Ras (Fig. 7B
). Moreover, apoptosis was not induced in the human prostate carcinoma cell line PC3, which has been shown to express LAPSER1 at the mRNA level (1)
(Fig. 7B
). The prostatic carcinoma cell line Du145, which has negligible LAPSER1 expression, was found to be sensitive to GFP-LAPSER1 overexpression in an LZ-dependent manner.
|
The 3Y1/v-Fps cells expressing LAPSER1-flag are TUNEL-positive (Fig. 7C
) and show caspase-3 activation (Fig. 7D
) but no changes in the p53, Akt, or phosphorylated Akt protein levels (Fig. 7D
). We also found that this apoptotic response was partially inhibited by purvalanol A, a relatively specific cdc2 inhibitor, suggesting an involvement of cdc2 activity in this pathway (Fig. 7E
). On the other hand, the mitochondrial membrane transition pore inhibitor bongkrekic acid had no effect (Fig. 7E
). Flow cytometric analysis also revealed an increase in the 4N population or G2/M arrest in these apoptotic populations (Fig. 7F
) and a dependency of this response on the expression of the C terminus of LAPSER1. Furthermore, DAPI staining of 3Y1/v-Fps cells after infection with the wild-type but not LZ LAPSER1 adenoviral vectors revealed the presence of micronuclei in 5.6% of the cells (Fig. 7G
). These results strongly suggest that the apoptosis caused by LAPSER1 in 3Y1/v-Fps cells could be due to mitotic catastrophe. We therefore examined whether this response was inhibited by the pan-caspase inhibitor Z-VAD-fmk. In the presence of Z-VAD-fmk (100 µM), we could detect only basal levels of TUNEL-positive 3Y1/v-Fps cells, confirming a successful inhibition of apoptosis. Under these conditions, we further observed an expanded binucleated cell population. We also found an increase in haploidy instead of the typical sub-G1 population, specifically in the presence of Z-VAD-fmk (Fig. 7F
).
Microscopic analysis revealed a massive emergence of cytokinesis without a central spindle in cells untreated with Z-VAD-fmk, the same phenotype we had observed earlier and shown in Fig. 2E
. The combined ratio (%) of no central spindle plus binucleated cells in the presence of Z-VAD-fmk (8% in each case) was similar to the no central spindle ratio (21%) detected without Z-VAD-fmk. The rate of binucleated cells in the absence of Z-VAD-fmk was 1.5%, lower than the rate in the presence of this agent. These data show that there was a shift in the distribution of the cell population from cytokinesis without a central spindle in the absence of Z-VAD-fmk to binucleation in the presence of Z-VAD-fmk. We therefore speculate that in the presence of Z-VAD-fmk, cells bypass apoptosis during no central spindle cytokinesis and form binucleates. Consistent with this prediction, we observed dead cells during no central spindle cytokinesis with TUNEL-positive nuclei in the absence of Z-VAD-fmk (Fig. 7H
-2, 3). The question emerged, however, as to why haploids were generated in the presence of Z-VAD-fmk. It has been reported that haploids are generated after the formation of binucleates (18)
, and we observed that in 55% of cells in cytokinesis, a portion of their genome migrates to the cell periphery (Fig. 7I
). Taken together, our current data and earlier findings suggest that generation of haploid cells in the presence of Z-VAD-fmk is a potential cell fate for binucleated cells, which escape apoptosis during cytokinesis.
The relationship between katanin and LAPSER1 in 3Y1/v-Fps cells
We further assessed the relationship between LAPSER1 and p80 katanin in 3Y1/v-Fps cells by first examining their subcellular localization. Confocal microscopic observations revealed that in 3Y1/v-Fps cells, the endogenous LAPSER1 and p80 katanin proteins are detectable in the midbody (Fig. 8
A), similar to Saos-2 cells (Fig. 3A
). We knocked down endogenous p80 katanin by siRNA transfection and confirmed this by Western blot analysis (Fig. 8B
). Confocal microscopic observations subsequently revealed that most of the siRNA-transfected cells had no p80 katanin signals in their midbodies. However, these cells did show midbody-localized LAPSER1 signals (Fig. 8C
). Finally, we examined whether LAPSER1-induced 3Y1/v-Fps cell death is dependent on the presence of endogenous p80 katanin. We infected p80 katanin knockdown cells with the LAPSER1-flag adenovirus. As shown in Fig. 8D
, partial inhibition of the cell death was evident in the p80 katanin knockdown cells. This indicates that the death inducer activity of LAPSER1 is partly dependent on the presence of p80 katanin.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Our current findings show that LAPSER1 is present in both the centrosome and midbody and plays a role in cytokinesis. We also show that LAPSER1 knockdown abrogates central spindle midbody formation, confirming that it plays a role in central spindle organization and in the progression of cytokinesis. This central spindle phenotype has also been observed after the functional knockdown of ZEN4/MKLP1, a well-known cytokinetic molecule harboring microtubule bundling activity that forms centralspindlin together with MgcRacGAP (15)
. We also found that LAPSER1 binds to MKLP1 and regulates its levels in a cdc2 inhibitor-sensitive manner, but does not bind Aurora B. These results suggest that one pathway for organization of the central spindle by LAPSER1 is via MKLP1 binding. Phosphorylation of MKLP1 by cdc2 is known to prevent its association with microtubules (19)
. Therefore, LAPSER1 might specifically bind to the phosphorylated inactive form of MKLP1 and increase its stability, resulting in its functional inhibition. Furthermore, given the elimination of LAPSER1 localization to the midbody by anti-MKLP1 RNAi, MKLP1 could also serve as a docking protein for LAPSER1, as in the case of PLK1.
Another pathway for the organization of the central spindle by LAPSER1 may be via the microtubule depolymerizer katanin. We observed that LAPSER1 physically binds to p80 katanin and that these two molecules migrate from the mitotic centrosomes to the midbody in a diffuse manner that is dependent on MKLP1. An analogous mechanism was recently reported for centriolin and Cep55, which also diffusely translocate from mitotic centrosomes to the midbody (20
, 21)
. Significantly, the translocation of centriolin is also dependent on MKLP1, and Cep55 has been shown to be regulated via phosphorylation by cdc2 and PLK1 (20)
. Rat LAPSER1 contains PLK1 phosphorylation motifs, two sites in both the N- and C-terminal regions. Moreover, one site in the N terminus is a consensus motif for cdc2 phosphorylation. The disruption of centrosomal targeting seen in LAPSER1-N might therefore be due to the loss of the cdc2 and PLK1 motifs in the N terminus.
The translocation of p80 katanin depends on the presence of LAPSER1, whereas the translocation of LAPSER1 is independent of p80 katanin; however, a tight association is evident even in the transitional state (Fig. 6B
-4). These findings suggest that LAPSER1 is a carrier protein for p80 katanin. In addition, the C terminus of LAPSER1 has p80 katanin binding ability, and much of its biological activity is lost after deletion of the C terminus. Furthermore, katanin has inherent microtubule-severing activity. Taken together, one simple hypothesis for this interaction is that LAPSER1 carries katanin into the central midbody to perform its microtubule-severing activity and organizes the central spindle microtubules. It is noteworthy that Spastin, which is highly homologous to p60 katanin and has microtubule-severing activity, has also been reported to localize at the midbody (22)
, supporting our idea that microtubule depolymerizers colocalize in the midbody with bundling proteins.
The similar central spindle phenotypes seen for both the overexpression and knockdown of LAPSER1 seem to be paradoxical. However, there are differences in the response of astral microtubules to the alteration of LAPSER1 expression levels (Fig. 2E
, Fig. 4A
), indicating the existence of an intracellular region-specific microtubule response. Both the overexpression and knockdown of katanin have been shown to prevent axonal microtubule bundles from developing in neurons (23
, 24)
, suggesting common mechanisms. Alternatively, the major molecule affected may be different between the two events. The lack of a central spindle and the increased protein levels of MKLP1 found in LAPSER1 overexpressing cells suggest that the main effects of this overexpression may be to inhibit the function of MKLP1. On the other hand, the central spindle malformation and mislocalization of p80 katanin found in the LAPSER1 knockdown cells suggest that the silencing of LAPSER1 may primarily inhibit the function of katanin.
In a pathological context, three of our current findings support our hypothesis that LAPSER1 is a tumor suppressor. First, we examined the tumor-suppressing properties of LAPSER1 in transformation assays using three different systems and found, in each case, that LAPSER1 inhibits transformation. LAPSER1 also inhibits cell proliferation and generates binucleated cells. Moreover, the lack of a central spindle in LAPSER1-overexpressing cells may cause cytokinetic defects and eventually inhibit cell growth. The binucleation events that follow the exogenous expression of LAPSER1 can be interpreted as the mode by which LAPSER1 inhibits tumor generation, since binucleates have less proliferative potential. We call this "regulated" binucleation. There is a report that the formation of binucleated hepatocytes caused by a failure of cytokinesis is a physiological response to metabolic load (25)
. Furthermore, chromosome nondisjunction-induced binucleation was recently demonstrated even in the absence of chromosome bridges (26)
, indicating that binucleation occurs by design rather than mechanical failure (27)
. This also suggests that there is a benefit of binucleation to the whole organism. The molecular characterization that distinguishes these adaptive binucleation responses from "deregulated" binucleation, which threatens the survival of the whole organisms (8)
, has not been elucidated. There also seems to be no requirement for p53 or Rb during LAPSER1-induced binucleation since Saos-2 cells, which lack both of these genes, show this phenotype when LAPSER1 is overexpressed.
Second, we observed LAPSER1-induced apoptosis in 3Y1/v-Fps cells. This is partly dependent on p80 katanin, but presumably is independent of p53 or Akt. Fps/Fes is reported to be involved in the microtubule network (28)
. It has been observed that activated Fps strongly colocalizes with microtubules, then binds and phosphorylates tubulin. Also, Fes was previously reported to localize at the centrosomes (29)
. Hence, the apoptotic response of cells transformed with v-Fps, but not v-Ras or v-Src, provides further evidence that LAPSER1 functions via the microtubule system.
The apoptotic response we observed in LAPSER1-overexpressing 3Y1/v-Fps cells showed characteristics of mitotic catastrophe, but also showed TUNEL-positive cytokinesis (Fig. 7H
-2, 3). This indicates that apoptosis occurs during cytokinesis in these cells. In addition, blocking these effects of LAPSER1 with Z-VAD-fmk revealed an unexpected phenotype: the generation of haploid cells. Recently, a parasexual somatic reduction division, termed neosis, has been reported. Polyploids evade death via mitotic catastrophe by attempting to undergo neosis, and the resultant cells are referred to as Raju cells (30)
. There are similarities between neosis and our current observations of polyploids, as well as evidence of mitotic catastrophe. The cells that survived mitotic catastrophe had a reduced DNA content, indicating the occurrence of reduction division. Consistent with this, we actually observed "sequential cytokinesis"-type neosis (Fig. 7I
) (30)
.
Third, anti-LAPSER1 RNAi induces central spindle defects and the appearance of binucleates in a small but appreciable fraction of the cells. We assume that this might cause genomic instability. To more precisely address this issue, long-term experiments that will analyze the genetic events as well as the progression of the tumors are necessary. The effects of LAPSER1 on central spindles, its interaction with katanin, and the requirement of its C-terminal region for biological activity are all commonly observed in different physiological and pathological settings. Thus, we propose that LAPSER1 is a novel and biologically significant participant in the cytokinetic phase of the cell cycle.
|
ACKNOWLEDGMENTS |
|---|
Received for publication October 2, 2006. Accepted for publication February 1, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Sudo and Y. Maru LAPSER1/LZTS2: a pluripotent tumor suppressor linked to the inhibition of katanin-mediated microtubule severing Hum. Mol. Genet., August 15, 2008; 17(16): 2524 - 2540. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
-flag), immunoprecipitates with anti-flag antibody. Heavy chain and light chain, antibodies. B) Lysates from COS7 cells transfected with the katanin fragment (Con80) expression construct and infected with LAPSER1-flag viruses were immunoprecipitated with anti-flag antibodies and subjected to Western blot with anti-Xpress antibodies. LYSATE, whole-cell lysate; IP (
