Two NDR kinase–MOB complexes function as distinct modules during septum formation and tip extension in Neurospora crassa

Summary

NDR kinases are important for growth and differentiation and require interaction with MOB proteins for activity and function. We characterized the NDR kinases and MOB activators in Neurospora crassa and identified two NDR kinases (COT1 and DBF2) and four MOB proteins (MOB1, MOB2A, MOB2B and MOB3/phocein) that form two functional NDR–MOB protein complexes. The MOB1–DBF2 complex is not only essential for septum formation in vegetative cells and during conidiation, but also functions during sexual fruiting body development and ascosporogenesis. The two MOB2-type proteins interact with both COT1 isoforms and control polar tip extension and branching by regulating COT1 activity. The conserved region directly preceding the kinase domain of COT1 is sufficient for the formation of COT1–MOB2 heterodimers, but also for kinase homodimerization. An additional N-terminal extension that is poorly conserved, but present in most fungal NDR kinases, is required for further stabilization of both types of interactions and for stimulating COT1 activity. COT1 lacking this region is degraded in a mob-2 background. We propose a specific role of MOB3/phocein during vegetative cell fusion, fruiting body development and ascosporogenesis that is unrelated to the three other MOB proteins and NDR kinase signalling.

Introduction

Establishment of cell polarity and maintenance of cellular asymmetry are essential cellular properties that govern morphogenesis and development of uni- and multicellular organisms. Members of the conserved nuclear Dbf2p-related (NDR) kinase family are important for growth and differentiation in various organisms. In Drosophila melanogaster, the NDR kinase Tricornered is required for controlling cell proliferation as well as for neuronal morphogenesis (Justice et al., 1995; Xu et al., 1995; Geng et al., 2000; Emoto et al., 2004; 2006; Wei et al., 2007), while the Caenothabditis elegans homologue SAX-1 regulates aspects of neuronal cell shape and has been proposed to be involved in cell spreading, neurite initiation and dendritic tiling (Zallen et al., 2000; Gallegos and Bargmann, 2004). Further work has resulted in an emerging NDR signalling network (Kanai et al., 2005; Nelson et al., 2003 summarized in Hergovich et al., 2006), in which NDR kinase activity is controlled by its binding partner MOB (Bichsel et al., 2004; Hergovich et al., 2005) and an upstream germinal centre kinase of the Ste20 superfamily that controls NDR phosphorylation status and is required for its full activation (Stegert et al., 2005; Wei et al., 2007).

Although NDR pathway elements are highly conserved among eukaryotes, impairing their functions can result in highly divergent cellular responses, indicating that the detailed wiring of these components is critical for NDR signalling in a specific organism. This is best illustrated by the comparison of Cbk1p pathway mutants in Saccharomyces cerevisiae and Cryptococcus neoformans (called RAM mutants in these organisms for ‘regulation of Ace2p activity and cellular morphogenesis’). Mutations in RAM components in these two organisms result in either loss of cell polarity or hyperpolarized growth respectively (Nelson et al., 2003; Walton et al., 2006).

Phylogenetic analyses support the presence of a second group of NDR kinases in all eukaryotes called Dbf2p and SID2 in budding and fission yeast respectively. The activity of both kinases requires binding of MOB1, several upstream protein kinases and a Ras-superfamily GTPase (Walther and Wendland, 2003; Wolfe and Gould, 2005; Krapp and Simanis, 2008). The function of this network, called septation initiation network (SIN) in fission yeast or mitotic exit network (MEN) in budding yeast, is the co-ordination of nuclear division with cytokinesis. The unifying feature of mutants in SIN components is defective septum formation and the results are aseptate strains. S. cerevisiae MEN mutants behave slightly different and arrest as dumbbell-shaped cells, indicative of late telophase arrest. Thus, each NDR kinase pathway in the two yeasts has clearly defined and separate functions and either connects mitotic exit with cytokinesis or is involved in polarity and cellular morphogenesis respectively (Nelson et al., 2003; Kanai et al., 2005; Krapp and Simanis, 2008).

This clear separation between cell cycle and morphogenetic functions of NDR signalling is not conserved in animals, and the number of NDR kinases and of MOB adapter proteins and thus of potential MOB–NDR kinase interaction pairs has increased with the increasing complexity of the organism. Filamentous fungal genomes contain two ndr and up to four mob genes (Galagan et al., 2003; 2005), while four NDR kinases (representing two kinases in each major subgroup) and six MOB proteins are present in vertebrates (Bichsel et al., 2004; Devroe et al., 2004). Each kinase has been found to be able to associate with several MOB proteins in higher eukaryotes, and the distinction between cell cycle control and morphogenesis is less strict than in yeasts (Bichsel et al., 2004; He et al., 2005b; Hergovich et al., 2005; Lai et al., 2005).

Despite the relevance of an apically growing tip cell for most members of the fungal kingdom, the key components that are required for tip extension and for colonization of substrates are poorly understood. Furthermore, NDR pathway components have been shown to be essential for pathogenicity and virulence in all fungal pathogens analysed so far (Durrenberger and Kronstad, 1999; McNemar and Fonzi, 2002; Scheffer et al., 2005; Walton et al., 2006). To date, the protein kinase COT1 of Neurospora crassa, the founding member of the NDR kinase family (Yarden et al., 1992), and POD6, a germinal centre kinase, which is associated with COT1 (Seiler et al., 2006; Maerz et al., 2008), are among the best-characterized components that specifically regulate tip growth and branch formation, but not cell polarity per se. Temperature-sensitive mutants of these two kinases cease hyphal elongation with a needle-shaped apex at restrictive temperature and produce massive amounts of extension-arrested new tips along the entire cell. Strains in which these genes have been deleted are viable and form compact colonies with growth-arrested tips, indicating that both kinases are essential for tip extension and for restricting supernumerary branch formation, but are not required for establishing new sites of growth (Collinge and Trinci, 1974; Collinge et al., 1978; Yarden et al., 1992; Seiler and Plamann, 2003; Seiler et al., 2006). cot-1 encodes for two transcripts, whose ratio of expression is photo-regulated, but the functional significance of the two generated COT1 isoforms that differ only in their N-terminus remains undetermined (Lauter et al., 1998; Gorovits et al., 1999).

The importance of NDR signalling for fungal growth and the subtle differences in the wiring of these elements has prompted us to dissect the function and nature of interactions between NDR kinases and MOB proteins in more detail. We show that the MOB1–DBF2 and MOB2–COT1 complexes function as distinct modules during septation and tip growth. MOB3/phocein, however, is specific for vegetative cell fusion and fruiting body development and is unrelated to NDR kinase signalling.

Results

Three types of MOB proteins with distinct functions are present in filamentous fungi

Database searches identified four MOB proteins in the genome of N. crassa that were most similar to fungal MOB1 and MOB2 and to the more distantly related MOB family member MOB3/phocein (Fig. 1). Based on these sequence similarities, NCU01605 was designated mob-1, NCU03314 and NCU07460 mob-2a and mob-2b, respectively, and NCU07674 mob-3. Sequence comparisons within the available fungal genomes revealed the presence of at least one MOB protein of each type in all filamentous growing members of the fungal kingdom. MOB3/phocein, however, was detected only in filamentous fungi and higher eukaryotes, but not in unicellular yeasts.

Fig. 1.

Phylogenetic distribution of the MOB protein family. MOB proteins of budding and fission yeast (named ScMOB and SpMOB respectively), N. crassa (BROAD Accession No. NCUxxxx) and selected animal MOBs (NCBI Accession No.) were aligned using the MegAlign program from DNAStar (Lasergene) to generate a phylogenetic tree based on the clustal v (PAM 250) method. MOB1 and MOB3/phocein proteins form distinct subgroups, while MOB2-type proteins separate into two animal-specific clusters and a fungal-specific group. The C. elegans protein NP_502248 is likely a distinct member of the MOB1 group. Schematic diagrams of individual MOB proteins depict the conserved MOB domain. On the right, the percentage of sequence identity and similarity between the MOB domains and the respective N. crassa MOB protein in each subgroup is indicated.

Strains harbouring deletions of the four mob genes were provided by the N. crassa genome project (Dunlap et al., 2007) and were used to determine cellular functions of the different MOB proteins (Fig. 2). Δmob-1 was characterized by a growth rate that was reduced to 40% of wild type, increased branch formation and a strong cell lysis defect. The generation of aerial mycelium was abolished and conidiation was reduced to < 1% of wild type. Furthermore, Δmob-1 was unable to generate female reproductive structures. This inability to form protoperithecia may be seen as a secondary consequence of the described vegetative defects, but fertilization of wild type with heterokaryotic Δmob-1 + mob-1⁺ conidia resulted in defective ascosporogenesis and the frequent formation of asci containing only a single, large, ascospore. These giant ascospores produced colonial growth similar to wild type on selective medium, indicating that the deletion phenotype was sheltered by mob-1⁺ in these progeny and that MOB1 is directly involved in meiosis and ascosporogenesis.

Fig. 2.

MOB proteins have distinct cellular function in N. crassa.

A. Phenotypic characterization of the indicated mob deletion strains with regard to colony morphology (upper panel; growth for 5 days on minimal medium; bar = 1 cm), hyphal morphology on minimal medium (middle panel; bar = 50 μm), and protoperithecia formation (lower panel; growth for 7 days on cornmeal medium; bar = 300 μm).

B. Ascus development 3 weeks after fertilization with the indicated male partner in squeezed perithecia of the indicated crosses (bar = 100 μm); note the presence of large banana-shaped ascospores in the Δmob-1 cross.

C. Production of conidiospores was quantified by counting conidia generated in slants grown at room temperature for 5 days (n = 5; standard deviations are indicated as bars).

Δmob-2a and Δmob-2b displayed similar defects, which were less pronounced than those of Δmob-1. The deletions resulted in slightly reduced growth rates accompanied by increased branching frequencies, with Δmob-2a being more compromised than Δmob-2b (70% and 92% of the wild type growth rate respectively). Furthermore, we observed altered aerial hyphae formation and the generation of reduced amounts of conidia (11% and 54% of wild type respectively). The sexual development of Δmob-2a and Δmob-2b was not affected, and the strains were female fertile and generated abundant, normally shaped ascospores in Δ × wild type and in Δ × Δ crosses with ascospore germination rates that were indistinguishable from wild type.

The growth rate of Δmob-3 was almost as high as the wild type (89%). Conidial production was only mildly affected (76% of wild type) and probably a consequence of reduced aerial mycelium formation. This contrasted with an approximately 30-fold reduction in the number of protoperithecia produced by Δmob-3. Furthermore, the few protoperithecia produced were much smaller and less developed than in the wild type. When these Δmob-3 protoperithecia were fertilized with wild type or Δmob-3 conidia, further development was blocked and only empty perithecia were formed that lacked ascogenous hyphae and developing asci. Wild type protoperithecia that were fertilized with Δmob-3 conidia developed further, but also resulted in only very few viable ascospores. Such defects in fruiting body formation and sexual development have been frequently connected with a failure of vegetative cell fusion (Wei et al., 2003; Fleissner et al., 2008; Maerz et al., 2008). When we compared the ability to undergo cell fusion in germlings and mature hyphae of Δmob-3 and wild type (Fig. 3), cell fusion was readily visible in wild type under both conditions using conventional light microscopy, but we were unable to detect fusion events in Δmob-3, indicating that cell fusion is dependent on MOB3 function. These assays do not rule out that cell fusion may occur at significantly lower frequencies, which may be suggested by the limited formation of fertile perithecia in wild type × Δmob-3 crosses. However, it is worth noting that the capability for and mechanistic nature of self–self fusion (which was assayed in the microscopic tests) may not be identical to self–non-self fusion (detected in the cross).

Fig. 3.

MOB3 is required for vegetative cell fusion. Hyphal fusion (left images; bar = 10 μm) and germling fusion (right images; bar = 5 μm) in wild type and Δmob-3 cultures was assessed by light microscopy. Fusion events are indicated by arrows. Cell fusion was not observed in Δmob-3.

MOB1–DBF2 and MOB2A/2B–COT1 complexes function as distinct modules during septation and tip growth

To test for potentially redundant functions of the four MOB proteins, we generated double mutants. No obvious synthetic interaction was observed in Δmob-1;Δmob-2a and Δmob-1;Δmob-2b, and both strains displayed lysing hyphae as their most characteristic phenotype (data not shown). Δmob-2a;Δmob-2b, however, formed tight hyper-branching colonies with extension-arrested tips, a phenotypic trait highly reminiscent to conditional or deletion mutants of both kinases of the COT1 complex (cot-1 and pod-6) germinating at restrictive temperature (Fig. 4A). We did not detect synthetic defects in any mob double mutant combinations containing Δmob-3. Thus, we propose MOB3 has a specific role during vegetative cell fusion and sexual development that is unrelated to the functions of the three other MOB proteins.

Fig. 4.

MOB1–DBF2 and MOB2A/2B–COT1 function as distinct modules during septation and tip growth.

A. Asco- or conidiospores of the indicated mutants were germinated for 2 days or 1 day, respectively, on minimal medium supplemented with hygromycin at 37°C (bar = 5 μm).

B. Calcofluor White staining indicated aberrant septation and the lack of functional cross-walls in Δmob-1 and Δdbf-2 (bar = 10 μm). Stainable cell wall material accumulated in a patchy manner in both mutants, but was more prominent in Δmob-1 than in Δdbf-2.

C. Co-purification of myc-DBF2- and myc-COT1-associated proteins from the indicated strains.

D. Immunoprecipitation (IP) experiments with anti-HA and anti-MYC antibodies and subsequent Western blot analysis (WB) from the indicated strains.

The prominent cell lysis phenotype of Δmob-1 was indicative of a cell wall defect and was supported by the presence of Calcofluor White-stainable material that accumulated in a patchy manner along the hyphal cortex (Fig. 4B). However, the formation of functional septa was completely abolished. Similar defects were also observed in Δdbf-2 (purified to the homokaryotic deletion via uninuclear microconidia) that harboured a deletion in the second ndr gene present in the N. crassa genome. The vegetative growth defects in both deletion strains led to the frequent appearance of suppressor mutations that regained the ability to generate septa and that subsequently regained the ability to conidiate. When the sexual capability of Δdbf-2 was analysed, we also found that protoperithecia formation was completely abolished. Interestingly, only giant, but no normal, shaped ascospores germinated from wild type × Δdbf-2 + dbf-2⁺ crosses on hygromycin-containing media. However, we were able to isolate normal pea-shaped ascospores from wild type × Δmob-1 + mob-1⁺ crosses that segregated in hygromycin-resistant and aseptate progeny and hygromycin-sensitive clones with a wild type phenotype. Thus, it is possible that, in contrast to MOB1, DBF2 is essential for proceeding through meiosis and ascosporo-genesis, and only sheltered Δdbf-2 + dbf-2⁺ progeny can survive the sexual cycle.

The mutant characteristics of mob and ndr deletion strains suggested specific interactions between COT1 and MOB2 proteins and DBF2 and MOB1. To further test these possible interactions, we generated strains harbouring tagged versions of both NDR kinases to identify co-purifying proteins. A myc-dbf-2-containing construct ectopically integrated in Δdbf-2;his-3 complemented the septation defect, indicating functionality of the fusion protein. In addition, we generated a strain, in which a 6×myc tag was inserted in frame at the second ATG of the endogenous cot-1 locus, which allowed the simultaneous detection of both COT1 isoforms (designated myc-cot-1). The wild type growth of this strain demonstrated the functionality of the modified endogenous cot-1 allele. Products of protein immunoprecipitation (IP) were resolved by SDS-PAGE, and specific bands were excised from gels and analysed by mass spectrometry (Fig. 4C). We observed two bands of proteins of c. 30 and 38 kDa that were consistently associated with myc-COT1 and identified them as MOB2A and MOB2B respectively. The interaction of MOB2A with COT1 was tighter than that with MOB2B, as we repeatedly detected reduced amounts of MOB2B, but never of MOB2A in our IPs. MOB1 could not undoubtedly be identified in myc-DBF2 precipitants by LC-MS (see Experimental procedures for criteria of LC-MS identification). Other co-purifying proteins were also detected, in varying amounts, in IPs using untagged wild type, and are, most likely, contaminants.

The identification of DBF2-associated MOB1 by LC-MS analysis was challenging, as the antibody light chain used for the IP experiments had the same molecular weight as predicted for MOB1. To overcome this difficulty and to further test for the specificity of the NDR–MOB interactions, we expressed HA-tagged versions of mob-1 and mob-2a in a nic-3 strain. Furthermore, myc-cot-1 was crossed into a his-3 background and myc-dbf-2 ectopically integrated in a his-3 strain. Various combinations of forced heterokaryons, capable of growing on minimal medium by complementation of the individual strains' auxotrophies, were then generated and subsequently used for co-IP experiments. We detected interactions of COT1 with MOB2A and of DBF2 with MOB1, but not between COT1 and MOB1 or DBF2 and MOB2A (Fig. 4D). Thus, the interaction profile of NDR kinases and MOBs and the phenotypic characteristics of the individual ndr and mob deletion strains indicated the existence of two distinct NDR–MOB complexes consisting of MOB1 associated with DBF2 and MOB2A and MOB2B bound to COT1.

MOB2 proteins affect kinase activity and COT1 stability

To determine the relative impact of individual MOB2 proteins on COT1 function, we generated double mutants of mob deletions with conditional cot-1(ts) and pod-6(ts) strains and assayed them for synthetic growth defects at different temperatures (Table 1). Strong synthetic interactions were detected for Δmob-2a;cot-1(ts) and Δmob-2a;pod-6(ts). Their growth rates were reduced to 53% and 55% at permissive temperature, and to 4% and 43% under semi-permissive conditions, respectively, compared with the slower-growing parental strains. Moreover, Δmob-2a;cot-1(ts) developed the typical cot-like hyper-branched colonies at semi-permissive conditions, indicating the importance of MOB2A for COT1 function. Synthetic interactions were also detected for double mutants of Δmob-1 and Δmob-2b with both conditional kinase strains at semi-permissive temperatures. We did not detect any synthetic interactions of cot-1(ts) or pod-6(ts) with Δmob-3, a further indication for the lack of related functions between MOB3 and NDR signalling pathways.

Table 1.

Radial growth rates of COT1 complex single and double mutants (n ≥ 3).

	Growth rate (cm day−1 ± SD)			Relative growth (% of slower parental strain)
	Permissivea	Semi-permissivea		Permissivea	Semi-permissivea
		30°C	34°C
cot-1(ts)	3.0 ± 0.3	2.5 ± 0.1		–	–
pod-6 (ts)	3.1 ± 0.1		3.7 ± 0.2	–	–
Δmob-1	1.4 ± 0.1	2.7 ± 0.2	2.6 ± 0.2	–	–
cot-1(ts);Δmob-1	1.3 ± 0.1	0.6 ± 0.3		93	25
pod-6(ts);Δmob-1	1.4 ± 0.1		0.8 ± 0.1	100	30
Δmob-2a	2.1 ± 0.2	3.9 ± 0.2	4.5 ± 0.1	–	–
cot-1(ts);Δmob-2a	1.1 ± 0.1	0.1 ± 0		53	4
pod-6(ts);Δmob-2a	1.2 ± 0.1		1.6 ± 0.2	55	43
Δmob-2b	3.2 ± 0.2	5.2 ± 0.2	5.4 ± 0.2	–	–
cot-1(ts);Δmob-2b	2.6 ± 0.3	2 ± 0.3		87	82
pod-6(ts);Δmob-2b	2.1 ± 0.2		2.5 ± 0.3	68	68
Δmob-3	3.1 + 0.2	4.7 + 0.1	5.6 ± 0.2	–	–
cot-1(ts);Δmob-3	2.8 ± 0.1	2.1 ± 0.2		95	86
pod-6(ts);Δmob-3	3.1 ± 0.1		3.2 ± 0.2	100	89

^aPermissive conditions for all strains were 20°C, while 30°C and 34°C are semi-permissive conditions for strains harbouring cot-1(ts) and pod-6(ts) mutations respectively.

Next, we generated Δmob strains, in which cot-1 was myc-tagged at the endogenous locus, to quantify the impact of individual MOBs on COT1's kinase activity by crossing the deletion strains with myc-cot-1. In contrast to myc-COT1 precipitated from myc-cot-1, which displayed robust in vitro kinase activity, myc-COT1 purified from myc-cot-1;Δmob-2a and myc-cot-1;Δmob-2b showed activities that were reduced to 45 ± 3% and 80 ± 17% (n = 3) respectively (Fig. 5A). The co-purification of both MOB2 proteins and their double mutant phenotype suggested overlapping functions of the two MOBs. Thus, we performed kinase assays with precipitants of a Δmob-2a;Δmob-2b;myc-cot-1 strain. We were able to precipitate myc-COT1 from these poorly growing cultures. The detected in vitro activity was only barely above background (1.0 ± 0.5% of myc-COT1; n = 3), indicating that the presence of the two MOB2 proteins is a prerequisite for in vitro COT1 activity and in vivo COT1 function.

Fig. 5.

MOB2 proteins are essential for COT1 function.

A. Kinase activity of myc-COT1 purified from the indicated mutants and the catalytically inactive strain cot-1(D337A). Data represent means of at least four independent experiments with ≥ 3 independent clones of each mutant (standard deviations are indicated as bars).

B. The abundance of both COT1 isoforms in cell extracts of the indicated strains was determined by Western blot analysis with anti-myc antibody and probed with an anti-actin antibody as control.

C. The kinase dead cot-1(D337A) strain has morphological defects indistinguishable from Δcot-1.

We found that the small myc-COT1 isoform in the precipitant was almost absent from Δmob-2a;Δmob-2b;myccot-1 precipitant. To determine if this was due to the loss of kinase activity or to the absence of the MOB proteins, we examined the presence of the large and small COT1 isoforms in cell extracts of Δmob-2a;Δmob-2b;myc-cot-1 (Fig. 5B). As a control, we generated a kinase dead strain, in which the catalytic aspartate 337 was mutated to alanine. myc-cot-1(D337A) displayed highly compact growth and hyperbranching defects, which were undistinguishable from Δcot-1 or Δpod-6 cells (Fig. 5C). When we compared these extracts, we found that the degradation of COT1's smaller isoform was due to the absence of MOB2 proteins and not due to the lacking kinase activity.

MOB2 binding is required, but not sufficient for COT1 activation

The presence of two MOB2-type proteins that interact with myc-COT1, and the fact that two isoforms of COT1 are expressed in N. crassa that differ by a 118-amino-acid N-terminal extension of unknown function (Gorovits et al., 1999) led us to ask if a specific MOB2 protein interacted with each COT1 isoform. We generated two cot-1 constructs, in which a 3×myc–6×his tag was fused with the first or second ATG of cot-1 respectively. Expression of myc-COT1(long) or myc-COT1(short) was controlled by the inducible qa-2 promoter, and the constructs were ectopically inserted at the his-3 locus of a cot-1(ts);his-3 strain. Both constructs complemented the temperature-sensitive growth defects in the presence of 10 mM quinic acid, indicating that each isoform did substitute for all COT1 functions defective in the temperature-sensitive strain. These strains were used for single-step anti-myc co-IP and tandem his-myc affinity purification experiments (Fig. 6A). Both purification conditions resulted in the co-purification of both MOB2 proteins with the long and the short COT1 isoform, indicating that the N-terminal extension did not specify the interaction with MOB2A or MOB2B. Although we observed variable amounts of co-precipitated MOB2B from both extracts, this variability was not significantly different from control-IPs from myc-cot-1 cultures, and the total amount of co-precipitated MOB corresponded to the level of precipitated COT1 (n = 3). We then used IP products of cultures expressing the two isoforms to determine their in vitro kinase activities (Fig. 6B). myc-COT1(long) displayed 45 ± 4% of myc-COT1 activity, which was consistent with the presence of only one isoform and thus of 56 ± 19% precipitated myc-COT1(long) compared with myc-COT1 (n = 3). myc-COT1(short) activity, however, was reduced to 33 ± 4% of myc-COT1(long) and to 15 ± 2% of myc-COT1, despite the fact that its protein level was only reduced to 94 ± 31% and 45 ± 5% of myc-COT1(long) and myc-COT1 respectively. Thus, the N-terminal extension of myc-COT1 is involved in the stimulation of the kinase.

Fig. 6.

Functional characterization of the two COT1 isoforms.

A. Single myc and double myc–his purification experiments of the individual COT1 isoforms from the indicated strains (note that endogenously tagged myc-cot-1 is modified only by a 5×myc-tag, while the ectopically integrated constructs are expressed as myc–his fusion proteins).

B. Quantification of in vitro kinase activities (left diagram) and amounts of precipitated myc-COT1 (right diagram) from the strains expressing the individual isoforms (n = 3; standard deviations are indicated as bars).

COT1 dimerization and interaction with MOB requires overlapping regions

The binding of MOB to NDR requires the N-terminus of the kinase (Bichsel et al., 2004; He et al., 2005b; Song et al., 2008). In addition to this proposed MOB binding interface that is conserved in fungal and animal NDRs, most fungal NDR kinases contain an uncharacterized N-terminal extension of varying length. Sequence analysis of the available fungal proteins revealed the existence of homology groups that differ in this N-terminal extension (Fig. S1). Schizosaccharomyces pombe ORB6 is a representative of an NDR kinase that, similar to animal counterparts, has no N-terminal extension. This contrasts with S. cerevisiae Cbk1p that has a 260-amino-acid extension lacking sequence homology to other proteins. However, a conserved feature of the Cbk1p extension that is shared with NDR kinase extensions in filamentous ascomycetes is its high content of asparagine and glutamine residues. In addition, NDR kinases of filamentous ascomycetes share conserved sequence motifs. For example, a methionine corresponding to the start codon of the short COT1 isoform is present in several (but not all) species, suggesting that the existence of multiple isoforms may be beneficial for filamentous growth.

Thus, we analysed the sequence requirements for the COT–MOB2 interaction in more detail. First, we confirmed the interaction of COT1 with both MOB2 proteins by yeast two-hybrid assays (Fig. 7A). Furthermore, we detected two-hybrid interactions between COT1 and POD6 and observed dimerization of COT1. No interaction of POD6 with any of the MOB proteins and between the two MOB proteins was detected in these assays. We then determined which domain of COT1 is required for MOB binding by analysing the N-terminal regions of the two COT1 isoforms (amino acids 1–212 and 119–212; designated long and short respectively; Fig. 7B). Specific interactions were detected between both COT1 fragments and MOB2A, indicating that region 119–212 of COT1 is sufficient to interact with MOB proteins. In addition, we found that the long fragment also interacted with MOB2B. Moreover, growth curves of two-hybrid cultures under selective conditions revealed a stronger interaction of MOB2A with the COT1 region 1–212 than with the region 119–212 (Fig. 7C). These data are consistent with our biochemical purifications that showed tight binding of MOB2A with myc-COT1 and a more variable interaction between myc-COT1 and MOB2B (Figs 4C and 6A).

Fig. 7.

Homo- and heterodimerization of COT1 requires overlapping regions.

A. Yeast two-hybrid analysis of COT1 complex components. Genes cloned into pGBKT7 and pGADT7 (mentioned as first or second fusion respectively) were coexpressed as fusions with the GAL4 DNA-binding domain and activation domains respectively. Plasmids expressing the indicated proteins either as prey or as bait alone were used as negative controls. pGBKT7-53 (murine p53) and pGADT7-recT (SV40 large T antigen) fusions were used as positive control.

B. Interaction analysis of two N-terminal COT1 fragments.

C. Growth curves of yeast harbouring the indicated two-hybrid plasmids in liquid medium.

Because COT1 dimerized in the two-hybrid assays, we also examined which region of COT1 is required for self-association (Fig. 7B and C). Both N-terminal COT1 fragments interacted with full-length COT1, yet the interaction of COT1 was stronger with fragment 1–212 than with fragment 119–212. Self-association was also detected between the two long COT1 fragments, but not when the two short fragments were used, indicating that region 119–212 is sufficient for homodimerization, but also that the interaction is stabilized by amino acids 1–118.

Discussion

As part of our comparative characterization we have assigned distinct cellular functions to the four members of the MOB family. The two NDR kinases present in N. crassa interact with a specific subset of MOB adaptors. COT1 is regulated by the combined function of MOB2A and MOB2B. This is indicated by the co-purification of COT1 with these two MOBs, the cot-like synthetic phenotype of the Δmob-2a;Δmob-2b double mutant and the deletion's impact on the in vitro kinase activity of COT1. Based on our findings that include the reduction in kinase activity in mob mutants and synthetic interaction of mob mutants with cot-1(ts), we believe that MOB2A contributes more to COT1 function than MOB2B. Currently, we do not understand why two close paralogues have evolved in N. crassa. We have, however, indications that several splice variants of MOB2 proteins are expressed in N. crassa (S. Maerz and S. Seiler, unpublished), which may suggest additional levels of COT1 regulation through the two MOB2 proteins.

The cot-1 locus allows the translation of two isoforms. The length of the N-terminus of the short isoform corresponds to the N-terminus of animal NDR kinases, while the long version contains a fungal-specific extension of 118 amino acids. We show that the region directly preceding the kinase domain [amino acids 119–212 of COT1(long)] is sufficient for homodimerization of COT1. Furthermore, this region is also responsible for the interaction of COT1 with MOB2 proteins, suggesting the presence of either COT1 homo- or COT1–MOB2 heterodimers in the cell and a potential for regulating NDR function at the level of dimmer formation. The sequence of the N-terminal extension is poorly conserved, but its presence is a feature of most fungal NDR kinases and characterized by a high abundance of the amino acids asparagine and glutamine (e.g. 24% and 36% asparagine and glutamine residues in N. crassa COT1 and S. cerevisiae Cbk1p respectively). This region is required for stabilization of kinase homo- and of COT1–MOB2 heterodimers and is also involved in the stimulation of the kinase activity. The fact that the level of MOB binding to COT1(long) and COT1(short) in vivo is similar, yet their in vitro kinase activities are different, indicates that kinase activation is not a mere consequence of MOB binding, but requires additional level of regulation. This may include additional interacting proteins and/or post-translational modifications of the kinase.

The reduced stability of COT1(short) homodimers may be a reason for the degradation of COT1(short) in the mob-2 double deletion background. This prediction is consistent with data from budding and fission yeast. The size of ORB6 corresponds to the short COT1 isoform, and shut-off experiments of mob-2 in S. pombe result in degradation of ORB6. Budding yeast Cbk1p, however, has a long N-terminal extension and deletion of mob-2 does not affect Cbk1p stability in this yeast (Hergovich et al., 2006; Jansen et al., 2006). The presence of a second ATG and thus the potential for the presence two expressed isoforms in N. crassa and related species (e.g. Podospora anserina, Magnaporthe grisea and Sclerotinia sclerotiorum) allows the prediction that the stability of animal NDR kinases is also regulated through their interaction with MOB proteins. However, no data are currently available regarding animal knock-down experiments of MOB proteins and their impact on endogenous NDR levels.

We compared the expression profiles of the four mob and the two ndr kinase genes within a growing colony (whole genome microarray data are available at http://bioinfo.townsend.yale.edu/index.jsp; Kasuga and Glass, 2008). These data indicate that cot-1 and mob-2a were expressed at higher levels in the peripheral region of the colony, while dbf-2 displayed equal expression levels in young and older regions of the colony. mob-1, mob-2b and mob-3 mRNAs were not even detected in the youngest section, but were constant throughout the remaining colony (Fig. S2A). These expression profiles are consistent with the observed defects in the respective mutants, and support functions of COT1 and MOB2A during tip growth, while the remaining proteins may primarily function in subapical regions. The expression profiles of dbf-2 and mob-1 may suggest a function of DBF2 in the apical region of the colony that is independent of MOB1.

We have previously shown that the mRNA levels of the two COT1 transcripts are photo-regulated. In dark-grown cultures the large transcript is favoured, while light induced the expression of the small transcript (Lauter et al., 1998). To extend this analysis, we used our endogenously tagged myc-cot-1 strain and assayed for COT1 expression in cells that were grown under vegetative conditions either in the light or in the dark as well as under conditions favouring asexual or sexual development. COT1 protein level decreased during the sexual development, but COT1 abundance was constant in the other conditions (Fig. S2B). More importantly, the ratio of the two isoforms did not change significantly during any growth condition. Thus, the light-dependent differences in mRNA abundance do not translate into significantly different protein levels of the two COT1 isoforms, and the significance of these mRNA profiles is currently unclear.

The mutant characteristics and our co-precipitation data indicate that the primary adaptor of the NDR kinase DBF2 is MOB1. This NDR–MOB pair is important for connecting mitotic exit with septum formation in both yeasts and in Aspergillus nidulans (Kim et al., 2006;Krapp and Simanis, 2008), and mutations in components of the SIN network result in aseptate strains. A full set of SIN components can be detected in the genome of N. crassa (e.g. CDC7/NCU01335; SPG1/NCU08878;CDC14/NCU06636; SID1/NCU04096; CDC11/NCU03545; SID2/NCU09071; MOB1/NCU01605), suggesting that this network operates in a similar manner in N. crassa. This is further supported by the fact that all mutants in these components display defects in septum formation (S. Maerz and S. Seiler, unpublished). However, in contrast to unicellular fungi (Schweitzer and Philippsen, 1991; Luca and Winey, 1998; Salimova et al., 2000; Krapp et al., 2004), SIN components are not essential for vegetative viability in N. crassa, and mutants still allow filamentous growth and colony formation. However, conidiation, which resembles a budding-type growth programme and also requires septum formation, is abolished in Δmob-1 and Δdbf-2 and only regained in strains harbouring suppressor mutations. In addition to their function in septation, our data also indicate a function of the two SIN components during sexual development. The fact that we were unable to isolate viable haploid progeny harbouring only Δdbf-2, along with the presence of viable hygromycin-resistant dbf-2⁺ progeny and the abnormal shape and size of the generated ascospores, suggests that DBF2 may have an essential function during meiosis. In contrast to dbf-2, the deletion of mob-1 also impaired meiosis, but functional ascosporogenesis was still possible, albeit at a reduced rate. A possible function of MOB1/DBF2 in controlling the mitotic cell cycle and co-ordinating mitotic exit with septation was not analysed here, but may be suggested by the cognate yeast and A. nidulans mutants.

It was interesting that double mutants of conditional cot-1(ts) or pod-6(ts) strains with Δmob-1 displayed synthetic defects that were more prominent than the growth defects observed in mutant combinations with Δmob-2b. This suggests that the deletion of mob-1 has more impact on the function of COT1 than the absence of the direct COT1-interactor MOB2B. Indirect data from budding yeast suggest an involvement of MEN function in regulating the localization of Mob2p and Cbk1p (Weiss et al., 2002). In addition, the activity of the SIN components CDC7 and SID1 has been shown to control ORB6 activity in interphase fission yeast cells (Weiss et al., 2002; Kanai et al., 2005). Thus we speculate that a similar, currently poorly understood connection between the two NDR kinase pathways also exists in N. crassa that may involve the presence of MOB1-bound DBF2 activity. A function of COT1 and POD6 as potential SIN effectors during septation is further supported by the localization of both proteins at the forming septum (Seiler et al., 2006).

MOB3 is required for vegetative cell–cell fusion and during sexual development. Based on these defects and the lack of synthetic interactions, it has a function unrelated to NDR signalling and the other MOB proteins. The mammalian homologue phocein has been shown to interact with the multidomain protein striatin, which has been suggested as a scaffolding protein linking cell signalling and endocytosis (Benoist et al., 2006). A functional conservation of mammalian and fungal striatin genes has been demonstrated in Sordaria macrospora by complementation of the striatin/pro11 mutant with a mouse striatin cDNA (Poggeler and Kuck, 2006). Interesting is that S. macrospora pro11 and N. crassa Δmob-3 (and also the N. crassa striatin deletion strain; S. Maerz and S. Seiler, unpublished) display highly similar developmental defects resulting in arrested protoperithecial development and subsequently sterility. This suggests the presence of a conserved signalling complex required for developmental decisions in filamentous ascomycetes and higher eukaryotes, yet not in unicellular yeasts, which lack detectable mob-3 and striatin genes.

Experimental procedures

Strains, media and growth conditions

Strains used in this study are listed in Table 2 (see also McCluskey, 2003). General genetic procedures and media used in the handling of N. crassa have been described (Davis and DeSerres, 1970; Davis, 2000) or are available through the Fungal Genetic Stock Center (http://www.fgsc.net). Microscopic documentation of fungal hyphae or colonies was performed with an SZX16 stereomicroscope, equipped with a Colorview III camera and Cell^D imaging software (Olympus, Japan) or an ORCA ER digital camera (Hamamatsu, Japan) mounted on an Axiovert S100 microscope (Zeiss, Germany). Image acquisition was performed using the Openlab 5.01 software (Improvision, UK) and images were further processed using Photoshop CS2 (Adobe, USA).

Table 2.

Neurospora crassa strains used in this study.

Strain	Genotype	Source
Wild type	74-OR23-1 Mat A	FGSC #987
Wild type	ORS-SL6 Mat a	FGSC #4200
cot-1(ts)	cot-1(C102t)	FGSC #4066
pod-6(ts)	pod-6(I310K)	Seiler et al. (2006)
Δmob-1 (heterokaryon)	hph∷mob-1Δ bar∷mus-51Δ + mob-1+ bar∷mus-51Δ	FGSC #11487
Δmob-1	hph∷mob-1Δ	FGSC #11487 × FGSC #987
Δmob-2a	hph∷mob-2aΔ	FGSC #11296
Δmob-2b	hph∷mob-2bΔ	FGSC #13575
Δmob-3	hph∷mob-3Δ	FGSC #12362
Δdbf-2	hph∷dbf-2Δ his-3− bar∷mus-51Δ	Microconidia of FGSC #12000
Δmob-1;cot-1(ts)	hph∷mob-1Δ cot-1(H351R)	This study
Δmob-2a;cot-1(ts)	hph∷mob-2aΔ cot-1(H351R)	This study
Δmob-2b;cot-1(ts)	hph∷mob-2bΔ cot-1(H351R)	This study
Δmob-3,cot-1(ts)	hph∷mob-3Δ cot-1(H351R)	This study
Δmob-1;pod-6(ts)	hph∷mob-1Δ pod-6(I310K)	This study
Δmob-2a;pod-6(ts)	hph∷mob-2aΔ pod-6(I310K)	This study
Δmob-2b;pod-6(ts)	hph∷mob-2bΔ pod-6(I310K)	This study
Δmob-3;pod-6(ts)	hph∷mob-3Δ pod-6(I310K)	This study
myc-cot-1(long)	cot-1(C102t) his-3∷pqa2-myc5-his6-cot-1(1-1842)	This study
myc-cot-1(short)	cot-1(C102t) his-3∷pqa2-myc5-his6-cot-1(433-1842)	This study
myc-cot-1	myc-cot-1(183-4)	This study
myc-cot-1;Δmob-2a	hph∷mob-2aΔ myc-cot-1(183-4)	This study
myc-cot-1;Δmob-2b	hph∷mob-2bΔ myc-cot-1(183-4)	This study
Δmob-2a;Δmob-2b	hph∷mob-2aΔ hph∷mob-2bΔ	This study
Δmob-2a;Δmob-2b;myc-cot-1	hph∷mob-2aΔ hph∷mob-2bΔ myc-cot-1(183-4)	This study
trp-1;his-3	trp-1− his-3−	FGSC#4050 × FGSC#6103
myc-cot-1;his-3	myc-cot-1(183-4) his-3−	This study
myc-dbf2(EC)	PgpdA-myc–his-dbf2∷nat(EC)	This study
myc-dbf2(EC);his-3	PgpdA-myc-his-dbf2∷nat(EC);his-3−	This study
HA-mob-1	his-3∷Pccg-1-HA-mob-1;nic-3−	This study
HA-mob-2a	his-3∷Pccg-1-HA-mob-2a;nic-3−	This study
myc-cot-1(D337A)	myc-cot-1(D337A)	This study

Tagged constructs

The myc-tagged cot-1 replacement cassette, pCZ18 (Seiler et al., 2006), was linearized with XhoI and co-transformed with a HindIII/BamHI genomic fragment containing the his-3 gene in cot-1;Δmus-51;his-3 strain. Transformants where screened for their ability to grow on minimal medium at 34°C. Five transformants were backcrossed to obtain the tagged cot-1 allele in a wild type background. Proper integration of the myc-cot-1 cassette at the cot-1 locus (and the replacement of the native cot-1 ORF) was verified by Southern blot analysis and by sequencing. To generate myc-cot-1(D337A) pCZ18 was mutagenized using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's protocol using the oligonucleotides CAC GGT TGT GCA TGC AGA GCT ATT AAG CCA GAC and its complement (not specified) for the mutagenesis.

For the expression of double-tagged cot-1 constructs, the pQA2-myc–his vector was used, which allowed the expression of N-terminally 3×myc–6×his-tagged proteins from the his-3 locus (He et al., 2005a). PCR fragments containing genomic DNA of the long or short cot-1 ORF, respectively, and their 3′-UTRs were cloned into pQA-myc–his. The primers used for myc–his-COT1(long) were AGA CAA GGT GAA TTC ATG GAC AAC and AGC AAG CGC TAG TTG TAT TT, and for myc–his-COT1(short) TAT CTG AGC GAA TTC ATG CCT TCG and CGT ATC CCG GGC ATA GTA TT. The resulting constructs were transformed by electroporation into a cot-1(ts);his-3 strain at the his-3 locus.

To generate tagged dbf-2 (NCU9071), the 3×myc-6×his tag was excised using DraI and Cfr9I from the pQA-myc–his plasmid and blunted using Klenow DNA polymerase. The obtained fragment was cloned between the Pgpd promoter and TrpC terminator of pEHN1nat (kindly provided by Stefanie Pöggeler), which was linearized with NcoI and blunted with Mung Bean nuclease. PCR-amplified fragments of the genomic dbf-2 ORF were obtained using the primers GGC GCG CCT ATG TCT AGC TAC TTG ACA AAC TTC and CTA GGA TCC CTA CAG CAT CGT ACC AAA ATT G. The integrity of the myc–his-dbf-2 expression vector was verified by sequencing prior to its transformation into wild type or his-3 protoplasts. Transformants were selected on minimal medium containing 30 μg ml⁻¹ nourseothricin.

PCR-amplified fragments of the respective mob ORF were cloned into pHAN1 (Kawabata and Inoue, 2007), allowing expression of HA-tagged proteins from the his-3 locus. The following primers were used: GGG ATC CAT GAG CTC CTT TCT TAC GAC C and GGG GCC CCT AGT CGC TGC GTA ACA TGC for mob-1 and GCC CGG GTA TGG ATC CCA ATA ATG GTT CG and GGA ATT CCTAGC TCG AGG GTG GGC C for mob-2a. For expression of HA-tagged MOB proteins, the HA constructs were transformed into nic-3;his-3 or trp-1;his-3 protoplasts, respectively, and were selected for complementation of the his-3 auxotrophy. IP was performed with cell extracts from fused, heterokaryotic, strains that were selected by their ability to grow on minimal media lacking supplements.

Yeast two-hybrid assays

The Matchmaker Two-Hybrid system 3 (Clontech, USA) was used according to the manufacturer's instructions. cDNA of the indicated genes was amplified with primers spanning the ORFs from start to stop codons as annotated by the N. crassa database (http://www.broad.mit.edu/annotation/fungi/neurospora_crassa_7/index.html) and cloned either into the pGADT7 vector containing the GAL4 activation domain or into pGBKT7 containing the DNA-binding domain using the following primers: MOB1EcoRI-5′ (GGA ATT CAT GAG CTC CTT TCT TAC GAC C), MOB1BamHI-3′ (CGG GAT CCC TAG TCG CTG CGT AAC ATG C), MOB2aEcoRI-5′ (GGA ATT CAT GTC CAA CCT CTT TTC TGG AA), MOB2aEcoRI-3′ (GGA ATT CCT AGC TCG AGG GTG GGC C), MOB2bEcoRI-5′ (GGA ATT CAT GTC TTG GAG CTC AGC CAA C), MOB2bBamHI-3′ (CGG GAT CCT TAA GCC AGG CCT GCC ATC TG), COT1longEcoRI-5′ (GGA ATT CAT GGA CAA CAC CAA CCG CC), COT1BamHI-3′ (CGG GAT CCT TAT CGG AAG TTG TTG TCG AAA C), COT1shortEcoRI-5′ (CGG AAT TCA TGC CTT CGAATA CCC AGA CC), COT1-N-termBamHI-3′ (CGG GAT CCT TAC TCG GGC TTG TCC TTG GTT C), POD6NdeI-5′ (GAT CAG CAT ATG GCG ACC CTA TCG GTA TAC), POD6EcoRI-3′ (GGA ATT CCT ACC TCC CTC AGA CAC TCG TG). The fusion proteins were (co)expressed in S. cerevisiae AH109 and potential interactions determined by activation of lacZ or his3 and ade2 reporter constructs that discriminate positive interactions on the basis of colour in the presence of X-α-galactopyranoside or by selection for viability on SD medium lacking adenine and histidine.

NDR kinase purification and identification of co-purifying proteins

All buffers contained the following phosphatase and protease inhibitors: 20 mM β-glycerophosphate, 2 mM Na₃VO₄, 5 mM NaF, 0.5 mM PMSF, 1 mM benzamidine, 1 mM DTT, 1 μg ml⁻¹ pepstatin A, 5 μg ml⁻¹ aprotinin. Mycelial samples were frozen in liquid nitrogen, pulverized and suspended either in lysis buffer 1 for IP experiments (20 mM Tris pH 7.5, 150 mM NaCl, 10% glycerol, 0.5 mM EDTA, 0.1% NP-40) or in lysis buffer 2 for double-tag his-myc purifications (20 mM Tris pH 7.5, 150 mM NaCl, 10% glycerol, 25 mM glucose, 0.01% triton X-100). For the double-tag purification, 25 ml of cleared crude extract (10 min, 16 000 g) was incubated for 2 h at 4°C with 500 μl of Ni-NTA agarose beads (Quiagen, Germany). The beads were washed twice with washing buffer (20 mM Tris pH 7.5, 500 mM NaCl, 10% glycerol, 25 mM glucose, 0.01% triton X-100) and were eluted in 20 mM Tris pH 7.5, 137 mM NaCl, 10% glycerol, 25 mM glucose, 0.01% triton X-100, 200 mM imidazole. For the IP, the supernatant of cell extract (10 min; 16 000 g) or the elution fraction of the Ni-NTA purification were incubated with monoclonal 9E10 anti-myc (Santa Cruz, USA) antibody on a rotation device for 2 h at 4°C, and with Protein-A-Sepharose beads for additional 1 h at 4°C. The beads were washed three times (20 mM Tris pH 7.5, 500 mM NaCl, 10% glycerol, 0.5 mM EDTA, 0.1% NP-40). Immunoprecipitated proteins were recovered by boiling the beads for 10 min at 98°C in Laemmli buffer and separated by 10% SDS-PAGE. Quantification of precipitated proteins were performed by densiometry using the software AIDA Image Analyzer 4.2.2 (Raytest, USA) of Coomassie Brillant Blue-stained gels (Blum et al., 1987).

For protein identification by LC-MS, peptides of the in-gel trypsinated proteins (Shevchenko et al., 1996) were extracted from gel slices of silver-stained protein bands and separated on a Dionex NAN75-15-03-C18 PM column with an ultimate3000 HPLC system (Dionex, Amsterdam, the Netherlands) prior to mass analyses with an LCQ DecaXP mass spectrometer (Thermo Electron Corp., San Jose, USA). Cycles of MS spectra with m/z ratios of peptides and four data-dependent MS2 spectra were recorded by mass spectrometry. The ‘peak list’ was created with extractms provided by the Xcalibur software package (BioworksBrowser 3.3.1SP1). The MS2 spectra were analysed against the N. crassa genome protein database using the Turbo-SEQUEST program (Eng et al., 1994) of the Bioworks software (Version 3.1, Thermo Electron; Germany). Protein identification required at least two different high scoring peptides meeting the following criteria: (i) XCorr (1+, 2+, 3+) > 2.0, 2.5, 3.0, (ii) ΔCn > 0.4 and (iii) Sp > 500. MS2 spectra of the highest scoring peptides were manually verified.

COT1 activity assays

Mycelial samples were frozen in liquid nitrogen, pulverized and re-suspended in IP buffer (50 mM Tris pH 7.5, 100 mM KCl, 10 mM MgCl₂, 0.1% NP40, 20 mM β-glycerophosphate, 2 mM Na₃VO₄, 5 mM NaF, 0.5 mM PMSF, 1 mM benzamidine, 2 mM EGTA, 1 mM DTT, 1 μg ml⁻¹ pepstatin A, 5 μg ml⁻¹ aprotinin). The samples were homogenized, centrifuged at 4000 g for 15 min, and the supernatant subjected to a second centrifugation step for 15 min at 22 000 g. To equalize the protein concentration of the crude extracts protein content was measured by a Bradford assay using Roti-Quant (ROTH, Germany). For IP, 1 ml aliquots of crude extracts were incubated for 2 h at 4°C on a rotation device with 1.5 μg of anti-myc antibody 9E10 (Santa Cruz, USA). The antigen-antibody complexes were recovered using protein A-sepharose (Amersham, UK) and washed once with IP buffer, twice with IP buffer containing 1 M NaCl followed by two times with kinase reaction buffer (20 mM Tris pH 7.5, 10 mM MgCl₂, 1 mM DTT, 1 mM benzamidine, 1 mM Na₃VO₄, 5 mM NaF). Kinase assays were performed using a modified protocol described previously for Ndr kinase (Millward et al., 1998). Briefly, beads were re-suspended in 30 μl of kinase reaction buffer containing 2 mM synthetic substrate peptide (KKRNRRLSVA), 0.5 mMATP and 1 μCi [³²P]-ATP. After incubation for 1 h at 37°C, samples were centrifuged for 5 min at 16 000 g, the supernatant was spotted onto P81 phospho-cellulose paper circles (Whatman, UK). Dried circles were washed five times for 30 min with 1% phosphoric acid and once with acetone before incorporation of phosphate into the substrate peptide was measured by liquid scintillation counting. The remaining protein-sepharose pellet was boiled for 10 min in Laemmli buffer and the supernatant was used to determine the myc-COT1 concentration in the kinase reaction by SDS-PAGE and Western blot.