Regulation of a LATS-homolog by Ras GTPases is important for the control of cell division

Identification of NdrC as a Ras GTPase interacting protein

Dictyostelium discoideum NdrC (DDB0349842) belongs to the LATS/NDR kinase family, which constitutes a subgroup of AGC (protein kinase A/G/C-related) kinases [18, 20]. The Dictyostelium LATS/NDR family consists of four kinases, two ‘shorter’ NDR kinases (NdrA/B), and two ‘larger’ LATS/NDR-related kinases (NdrC/D) that are characterized by an extended N-terminus [19]. Similarly, the mammalian LATS/NDR kinase family is subdivided into two ‘larger’ LATS kinases (LATS1/2) and two ‘shorter’ NDR kinases (NDR1/2) (Additional file 1: Figure S1A). The Dictyostelium NdrC kinase is made up of 1,312 amino acids (147 kDa), and its protein sequence comprises the general features described for other LATS/NDR kinases, which include an N-terminal regulatory domain (NTR), an insert sequence (I) between subdomain VII and VIII of the catalytic domain, an activation segment (AS) as well as a conserved hydrophobic motif (HM) (Figure 1A, Additional file 1: Figure S1B).

When the Dictyostelium GTPases RasG, RasC and Rap1 were employed as bait in a yeast two-hybrid screen, NdrC was identified as a novel interacting protein that exhibited strong positive interactions with all three proteins. This screen also revealed that RasG and Rap1 exhibited strong positive interactions with the previously described Ras-binding domain (RBD) containing proteins PL3K and Rip3. In contrast, NdrC was the only protein that bound RasC in the yeast two-hybrid screen. The RBD of NdrC was localized between amino acid residues 107 and 284, the sequence that interacted with all three Ras proteins in the yeast two-hybrid assays (Figure 1A and data not shown). The minimal Ras-binding domain of NdrC was not defined further by additional experiments. A bioinformatics analysis did not reveal evidence of a RBD sequence in this region, but this is not surprising since the RBD sequences of a number of other RBD proteins were not detected by bioinformatic analyses [21].

Ras protein binding to NdrC was confirmed by pull-down assays using the identified RBD of NdrC tagged to GST and Dictyostelium lysates (Figure 1B). Bound Ras proteins were detected by Western blot analysis using specific anti-Ras polyclonal antibodies. The experiment revealed that the NdrC-RBD interacted not only with RasG, RasC and Rap1, but also with RasB (Figure 1B). This result was confirmed and extended by pull-down experiments using GST-tagged NdrC-RBD in combination with His-tagged Ras subfamily proteins that allowed a more quantifiable comparison of the binding. These interaction studies showed that NdrC bound to the activated Ras proteins but not to the wild type (Figure 1C), although the wild-type proteins did bind in the presence of GTPγS (data not shown). NdrC bound best to activated RasG and RasB, with lower levels of binding to activated Rap1 and RasC (Figure 1C). NdrC did not interact with either RasD or RasS (Figure 1C).

ndrC-null cells exhibit a severe defect in cytokinesis and aberrant numbers of centrosomes

To investigate the cellular function of NdrC, a knockout strain of the ndrC gene was generated by homologous recombination using a disruption construct carrying an inverted blasticidin-S resistance cassette within the partial ndrC gene sequence (Figure 2A). The most striking phenotype of the ndrC-null cells when compared to the wild type was that the majority of mutant cells were larger and multinucleated (Figure 2B). When wild-type and ndrC-null cells were fixed and stained with TO-PRO-3 to visualize the nuclei, and counterstained for actin using Alexa Fluor 488-coupled phalloidin, the increased number of nuclei in the mutant was clearly revealed (Figure 2C, D), and this increase was quantified by nuclei counts of fixed, DAPI-stained cell preparations (Figure 2E). The percentage of mono-nucleated cells in wild type was 95% compared to only 29% in ndrC- null cells (Figure 2E). The number of nuclei per cell in the ndrC- nulls peaked at 1, 2, 4 and 8 indicating concerted mitosis (Figure 2E).

An analysis of living cells lacking NdrC revealed that they were able to perform mitosis, but an impairment of cell division resulted in the formation of multinuclear cells. Frequently, putative daughter cells were still connected by thin cellular bridges suggesting division by cytofission rather than mitosis-coupled cytokinesis (Figure 2B, right). The generation times, measured for ndrC-null and wild-type cells were almost identical, regardless of whether cells were grown in shaking culture in rich axenic medium or on a solid substrate with bacterial lawns (Additional file 2: Figure S2). Since ndrC- null cells are considerably larger than wild-type cells, their cell masses increase at considerably faster rates indicating an additional defect in the control of cell growth.

Inspection of multinucleated ndrC-null cells, fixed and stained with TO-PRO-3 for visualization of nuclei and immunostained with anti-α-tubulin antibodies to visualize microtubules and centrosomes, revealed a high percentage of mutant cells carrying increased numbers of centrosomes (Figure 3). Frequently, the supernumerous centrosomes were not connected to the nuclei (Figure 3B). To quantify the disruption of centrosome integrity, the ratio of centrosomes visualized by anti-α-tubulin staining to nuclei visualized by DAPI staining, was determined for wild-type and ndrC-null cells (Figure 3D). About one third (31%) of the ndrC- null cells had more than one centrosome per nucleus, compared to only 0.6% in wild-type cells. ndrC-null cells with evenly distributed centrosomes underwent mitosis in a synchronized manner (concerted mitosis) (Figure 3E), but although the increased ratio of centrosomes per nuclei did not affect the ability of cells to perform concerted mitosis, the unattached surplus centrosomes were not capable of forming mitotic spindles (Figure 3F).

Aberrant numbers of centrosomes have also been observed for knockout strains of mouse LATS2 [22, 23] and human NDR1/2 [3, 24]. The cell division defect in mammalian cells has been attributed to a disruption of the signaling cascade controlling cytokinesis on the one hand [25], and a control of nuclear division on the other [26]. Human LATS1 was also reported to act as mitotic exit network kinase [27], responsible for regulation of the G2/M-arrest [28]. In addition, the Drosophila LATS-homolog Warts regulates mitotic progression [29]. However, despite these clear indications for a role of NDR/LATS kinases in regulation of the cell cycle and/or cell division, the underlying mechanisms are largely unknown.

Subcellular localization of NdrC

Gene expression profiling data indicated that NdrC is expressed at very low levels throughout the developmental cycle of Dictyostelium (http://dictyexpress.biolab.si) [30]. In order to specify the cellular localization of NdrC, polyclonal antibodies were raised against NdrC. Immunolabeling experiments with fixed Dictyostelium cells indicated that NdrC is preferentially enriched at the centrosome (Figure 4A). This localization was confirmed by co-staining with a monoclonal antibody directed against a genuine centrosomal component, the corona protein CP224 [31]. The centrosomal localization of NdrC was detectable only in a subfraction of cells, suggesting a cell cycle-dependent enrichment. Expression of full-length NdrC with a N-terminal GFP-tag confirmed the localization at the centrosome. NdrC had not been detected in a previous proteome screen of the Dictyostelium centrosome [32], but was found to co-purify with centrosomes that were isolated from Dictyostelium wild-type and cells expressing GFP-NdrC (Additional file 3: Figure S3).

Over-expression of a fusion protein comprising GFP and the N-terminal 300 amino acids of NdrC containing the RBD domain, showed a very similar localization as that reported previously for the Ras-binding domain of human Raf1 that relocates upon stimulation to active zones of the cell cortex [33, 34] (Figure 4B). This finding indicates a preferential binding of RBD of NdrC to membrane-associated Ras proteins. A GFP-fusion protein comprising the catalytic domain and the C-terminus of NdrC (amino acid residues 435 to 1312) localized to the cytoplasm and showed no specific enrichment (Additional file 4: Figure S4).

Putative control of NdrC by Ras GTPase family members

We have shown previously that rasG- null cells or cells overexpressing RasB-(G12T) exhibit a multinucleated cell phenotype that is similar to the one exhibited by the ndrC- null cells [35, 36]. In addition, severe defects in cytokinesis in cells overexpressing the exchange factor RasGEF-Q, which acts specifically upstream of RasB have been described [37]. Given that NdrC contains a RBD, it is conceivable that RasB or RasG exert regulatory functions on NdrC.

To further investigate the possible roles of RasG or RasB in NdrC function, we determined the subcellular localization of RasG and RasB. Wild-type cells overexpressing GFP-tagged RasB or RasG were normal in respect to cell size and nuclei number, and the GFP-tagged proteins predominately localized to the cortex (Figure 4C, D). The GFP-tagged activated forms of the proteins, RasB-(G12T) and RasG-(G12T), also localized predominately to the cortex (Figure 4E-G). However, whereas RasG-(G12T) overexpression did not affect the cell phenotype (Figure 4E), overexpression of RasB-(G12T) caused disturbed cytokinesis characterized by an impairment of cells to sever the connection between daughter cells (Figure 4F). This resulted in enlarged, multinucleated cells (Figure 4F), a defect similar to that described previously for cells expressing an untagged RasB-(G12T) [35].

Both, GFP-tagged RasB and GFP-RasG bound to the GST-NdrC-RBD, as shown by in vitro pull-down assays of wild-type AX2 cells expressing GFP-tagged Ras proteins (Figure 4H). Although NdrC also binds RasC and Rap1, cell division is not affected by either the loss or up-regulation of Rap1 [38], or by the loss or up-regulation of RasC [39]. Thus the significance of the binding of Rap1 and RasC to NdrC remains to be established.

Further evidence for a possible functional connection between the Ras-GTPases and NdrC was obtained by measuring the levels of activated Ras protein in the ndrC- null cells by Western blot analysis of GST-Byr2-RBD pull-downs using specific antibodies for the Ras family members. Levels of activated RasB and RasG were increased in the absence of NdrC compared to wild-type cells, indicating that NdrC acts as a negative regulator for activation of RasG and RasB (Figure 4I). In contrast, the levels of activated Rap1 and the total levels of all the Ras GTPases were found to be identical in lysates of wild-type and NdrC- null cells (Figure 4I). The mechanism involved in the feedback inhibition of RasG and RasB activation by NdrC, presumably at the level of RasGEF activity, is not known at this time, but it appears to be a general mechanism for downstream Ras effectors in Dictyostelium, since similar results have been observed for pik null and ripA null cells.

The findings that an absence of RasG and the overexpression of activated RasB both result in a multinucleate phenotype suggest the possibility that RasB and RasG play antagonistic roles in regulating NdrC during cytokinesis (Figure 5). The multinucleate phenotype of rasG-null cells [36] is consistent with a role for RasG in activating NdrC, whereas the multinucleate phenotype generated by over-expression of either activated RasB [35], or the RasB-specific exchange factor GefQ [37], indicates that RasB is a negative regulator of cytokinesis and cell division.

The regulation of NdrC by RasG and RasB implies that these Ras proteins would also be localized at the centrosome or in the nucleus. When cells were labeled with GFP-tagged Ras proteins, we find wild-type RasB and RasG, as well as constitutively active forms of the two proteins, are localized predominantly at the plasma membrane. This result contrasts somewhat with a previous study, using antibody staining, that revealed RasB clearly localized at the nucleus [35]. When we stained cells with an antibody highly specific to RasG, the majority of the stain was associated with the cell cortex, but some stain was also detected in the cytoplasm and in the nucleus (Additional file 5: Figure S5). These results imply that only a small proportion of the cellular RasG and RasB is localized in the nucleus, and are consistent with RasG and RasB both having other functions in the cell. In fact, RasG is not only involved in cytokinesis, but also in folate chemotaxis, in maintenance of cell shape, and in cell motility [40]. In addition, it has been shown recently that RasG, acting through PI3K, is important for pinocytosis [41]. Evidence for other specific functions for RasB is less direct, but gefQ- null cells exhibit myosin-II over-assembly and defects in polarity [37], implying a possible role for RasB in these functions, and rasB appears to be an essential gene [35], suggesting an additional vital function for RasB.

Cell culture, gene replacement and transformation of Dictyostelium

Cells of the Dictyostelium discoideum wild-type strain AX2-214, or mutant cells derived from it, were cultivated at 21°C in nutrient HL5 medium (Formedium), either in shaken culture at 150 rpm or in Petri dishes without shaking. To induce starvation, cells were washed twice in 17 mM Soerensen’s phosphate buffer (PB), pH 6.0, and shaken at a density of 10⁷ cells per ml in the buffer.

To generate ndrC-null mutants, wild-type AX2 cells were transformed with the gene replacement construct (Figure 2A) by electroporation using a Bio-Rad gene pulser at 0.8-0.9 kV and 3 μF, and 4-mm cuvettes. Independent transformants were selected by addition of 7.5 μg/ml of Blasticidin-S (ICN Biomedicals Inc.). Transformants were cloned by spreader dilutions on lawns of non-pathogenic Klebsiella aerogenes. Three independent NdrC-knockout clones were identified by testing genomic DNA for insertion of the resistance cassette into the ndrC gene by PCR.

For expression of fluorescently tagged Ras GTPases, AX2 wild-type and NdrC-null cells were transformed by electroporation with pDEX derived plasmids enabling the expression of GFP-RasB, GFP-RasB-G12T, GFP-RasG and GFP-RasG-G12T. Transformants were selected by addition of 10 μg/ml of G418 (Sigma-Aldrich) or 7.5 μg/ml Blasticidin-S (ICN Biomedicals Inc.), and cloned.

Live-cell microscopy

Cells were transferred into an open chamber and washed twice with Soerensen phosphate buffer. Confocal images were taken using an inverted LSM 510 Meta confocal microscope (Zeiss) equipped with a 63× Neofluar 1.4 or a 100× Neofluar 1.3 oil immersion objective. For excitation, the 488-nm argon ion laser line and the 543-nm as well as the 633-nm helium neon laser lines were used, and emission was collected using 505–530 nm, 585–615 nm band-pass or a 650 nm long-pass filter.

Immunofluorescence

For immunolabeling, AX2 wild-type or ndrC-null cells, settled onto glass coverslips, were fixed with a mixture of 15% (v/v) of saturated picric acid and 2% (w/v) paraformaldehyde in 10 mM Pipes-HCl, pH 6.0, at room temperature for 20 min, and post-fixed with 70% ethanol for 10 min. α-tubulin was detected using monoclonal rat antibodies (YL1/2) [43] and Alexa Fluor 568-conjugated goat anti-rat IgG (Invitrogen). For visualization of filamentous actin, cells were stained with Alexa Fluor 488-phalloidin (Molecular Probes). DNA was visualized either with TO-PRO-3 or with DAPI.

Polyclonal antibodies were generated against GST-tagged NdrC-RBD. The sequence encoding amino acid residues 1 to 299 of NdrC was cloned into the bacterial expression vector pGEX-6P1 (GE Healthcare). Purification of the GST-fusion protein was performed using standard procedures, and the recombinant protein was used to immunize a female white New Zealand rabbit together with the adjuvant Gerbu100 (Gerbu Biochemicals).

GST-RBD pull-down assay of endogenous Ras proteins

The binding of endogenous activated Ras proteins to the GST-Ras Binding Domain (GST-RBD) of Dictyostelium NdrC was determined as described previously [39]. 350 μl of Dictyostelium wild-type AX2 or ndrC-null cell suspension (5 × 10⁷ cells/ml) were lysed by mixing with an equal volume of 2× lysis buffer. 400 μg of protein lysate were incubated with 100 μg of GST-RBD and glutathione sepharose beads (GE Healthcare) at 4°C for 1 hour. 50 μl of 1× SDS gel loading buffer was added to the pelleted beads, and the suspension was boiled for 5 min. Samples were subjected to SDS-PAGE, and Western blots were probed with polyclonal antibodies directed against RasB, RasG, RasC, or Rap1. Equal sample loading was verified by staining of a duplicate gel with Coomassie Blue.

Interaction of His-tagged Ras and GST-NdrC

To produce 6xHis-tagged Ras-GTPases, wild-type RasG, RasC , RasB, RasS, Rap1 and RasD, and the constitutively activated forms of RasG (G12T), RasC (G13T), RasB (G15T), RasS (G12T), Rap1 (Q65E) and RasD (G12T) were cloned into pET-21a (Novagen) [41]. To express GST-NdrC, a fragment encoding amino acid residues 1–299 of NdrC was cloned into pGEX6P-1. Recombinant proteins were expressed in Escherichia coli BL21 (DE3) codonplus-RIL (Stratagene), and protein concentrations were determined using the DC Protein Assay (Bio-Rad). His-Ras/RBD-NdrC-GST interaction was examined as described previously [41]. In brief, 400 μg of purified His-tagged Ras protein were incubated with 100 μg of GST-RBD on glutathione sepharose beads and tumbled in binding buffer at 4°C overnight. Beads were washed three times and analyzed by SDS-PAGE, and Western blots were probed with anti-His tag monoclonal antibody (Santa Cruz Biotechnology).