In vitro regulation of budding yeast Bfa1/Bub2 GAP activity by Cdc5.

The Cdc5 protein of budding yeast is a polo-like kinase that has multiple roles in mitosis including control of the mitotic exit network (MEN). MEN activity brings about loss of mitotic kinase activity so that the mitotic spindle is disassembled and cytokinesis can proceed. Activity of the MEN is regulated by a small GTPase, Tem1, which in turn is controlled by a two-component GTPase-activating protein (GAP) formed by Bfa1 and Bub2. Bfa1 has been identified as a regulatory target of Cdc5 but there are conflicting deductions from indirect in vivo assays as to whether phosphorylation inhibits or stimulates Bfa1 activity. To resolve this question, we have used direct in vitro assays to observe the effects of phosphorylation on Bfa1 activity. We show that when Bfa1 is phosphorylated by Cdc5, its GAP activity with Bub2 is inhibited although its ability to interact with Tem1 is unaffected. Thus, in vivo inactivation of Bfa1-Bub2 by Cdc5 would have a positive regulatory effect by increasing levels of Tem1-GTP so stimulating exit from mitosis.

The Cdc5 protein of budding yeast is a polo-like kinase that has multiple roles in mitosis including control of the mitotic exit network (MEN). MEN activity brings about loss of mitotic kinase activity so that the mitotic spindle is disassembled and cytokinesis can proceed. Activity of the MEN is regulated by a small GTPase, Tem1, which in turn is controlled by a two-component GTPase-activating protein (GAP) formed by Bfa1 and Bub2. Bfa1 has been identified as a regulatory target of Cdc5 but there are conflicting deductions from indirect in vivo assays as to whether phosphorylation inhibits or stimulates Bfa1 activity. To resolve this question, we have used direct in vitro assays to observe the effects of phosphorylation on Bfa1 activity. We show that when Bfa1 is phosphorylated by Cdc5, its GAP activity with Bub2 is inhibited although its ability to interact with Tem1 is unaffected. Thus, in vivo inactivation of Bfa1-Bub2 by Cdc5 would have a positive regulatory effect by increasing levels of Tem1-GTP so stimulating exit from mitosis.
In budding yeast, the final stages of mitosis are controlled by the mitotic exit network (MEN). 1 Once the chromosomes are partitioned equally between mother and daughter cells, the MEN eliminates mitotic kinase activity, a prerequisite for spindle disassembly and cytokinesis (reviewed in Refs. 1 and 2). In the MEN, the key element controlling the pathway is Tem1 (3), a GTPase with biochemical properties similar to those of the Rho family of small GTPases (4). The MEN is negatively regulated by Bub2 and Bfa1 (5)(6)(7)(8), which together act as a two-component GTPase-activating protein (GAP) for Tem1. Presumably, Bfa1/ Bub2 reduce MEN activation in vivo by reducing Tem-GTP levels (4). The positive regulation of Tem1, and thus the MEN, has been linked to the activity of Lte1 (9, 10), a putative guanosine nucleotide exchange factor for Tem1 (11).
Following Tem1 activation, the functional sequence of the MEN involves activities of Cdc15 kinase followed by Dbf2 kinase with its associated binding protein, Mob1 (12). Ultimately, the MEN triggers the long term release of Cdc14 from the nucleolus, which, through a series of dephosphorylation events, reduces mitotic kinase activity (13). Recently, the Cdc14 early anaphase release (FEAR) pathway has been described which controls an initial release of Cdc14 from the nucleolus, which may contribute to optimal activity of the MEN although the precise role of the FEAR remains unclear (14 -16).
An important regulator of these activities is the polo-like kinase, Cdc5. Cdc5 has several essential roles in mitosis (12,(17)(18)(19) as well as regulation of the MEN and FEAR pathways. Even in the MEN there is evidence for more than one role for Cdc5. Net1, the protein that sequesters Cdc14 to the nucleolus, is phosphorylated in vitro and in vivo by Cdc5 so facilitating release of Cdc14 during anaphase (20,21). Cdc5 also has a positive but unidentified role at some stage between Tem1 and activation of Dbf2 (12). Furthermore, Cdc5 phosphorylates Bfa1 both in vitro and in vivo during mitosis in normally growing cells (15,22,23). Two opposite interpretations have been made on the role of this Bfa1 phosphorylation. Hu and co-workers (23) observed that Bfa1 was unphosphorylated in a nocodazole arrest, a condition where the MEN is kept inactive. For this and several other reasons, they equated phosphorylation of Bfa1 with an active MEN and hence proposed that phosphorylation inhibited Bfa1 GAP activity. However, Lee et al. (12,22) reached the opposite conclusion. They observed that Bfa1 was highly phosphorylated in a nocodazole arrest and that overexpression of Cdc5 inhibited the Dbf2 activity that normally signifies an active MEN (12,22). 2 By linking phosphorylation of Bfa1 with arrest of the MEN, they concluded that phosphorylation of Bfa1 increased GAP activity to negatively control mitotic exit.
It is important to note that these opposite conclusions on the effects of phosphorylation on Bfa1 GAP activity were based on downstream assays of the MEN and not on direct biochemical analyses of Bfa1 behavior. To resolve this important question of how Cdc5 and Bfa1 regulate the MEN, we have directly examined how phosphorylation affects the in vitro properties of Bfa1 in combination with Tem1 and Bub2. This purely in vitro approach was possible as the sites of phosphorylation in Bfa1 by Cdc5 in vivo are largely equivalent to those produced in vitro (23). In vitro, Bfa1 binds to Tem1 but has different activities depending on whether or not Bub2 is also present. Without Bub2, Bfa1 inhibits GTP hydrolysis and GTP dissociation from Tem, while Bfa1 acts as a two-component GAP in the presence of Bub2 (4). Here we show that in the absence of Bub2, phosphorylation of Bfa1 by Cdc5 does not affect dissociation and hydrolysis of GTP bound to Tem1. However, in the presence of Bub2, phosphorylation of Bfa1 by Cdc5 reduces GAP activity. We therefore conclude that one role of Cdc5 in the MEN is to reduce the hydrolysis of GTP by Bfa1/Bub2, so increasing Tem1-GTP levels and promoting exit from mitosis. However, Bfa1 phosphorylation cannot account for the complete control * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by Association for International Cancer Research Grant 99-008.
of GAP activity and so other regulatory mechanisms for the MEN are also discussed.

MATERIALS AND METHODS
Plasmid and Strain Construction-⌬70N-CDC5 was amplified by PCR using as template a plasmid containing a CDC5 allele harboring two point mutations, S165D and T238D (kindly provided by F. Uhlmann and M. Sullivan). The amplified sequence starting at codon 71 of CDC5 was cloned in-frame with GST in a URA3-based plasmid under the control of the GAL1 promoter. The URA3 was then replaced by a TRP1 marker, and the plasmid was introduced into a strain deleted for CDC5 and harboring a CDC5-URA3 plasmid (KKY902-2BTa). The CDC5-URA3 plasmid was eliminated by growth on 5-fluoro-orotic acid to generate strain MGY29. A ⌬70N-CDC5 kinase-dead plasmid expressing GST-Cdc5KD was constructed using a similar strategy but the template was a CDC5 allele harboring an N209A point mutation that inactivates the kinase (23).
Phosphorylation of Bfa1-Phosphorylation reactions using radiolabeled ATP were carried out in 30 l of kinase buffer (50 mM Tris/HCl, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol) containing 1-10 g of substrate. 10 -50 ng of kinase were added followed by 50 M ATP and 0.1 l of [␥-32 P]ATP (Amersham Biosciences, 370 MBq/ml, 3000 Ci/ mmol). After incubation for 15 min at 30°C, reactions were stopped by addition of 10 l of 4ϫ Laemli buffer and heating at 100°C for 2 min. After SDS-PAGE electrophoresis, radiolabeling was visualized by autoradiography. For preparative reactions of phosphorylated Bfa1, the protein was incubated in the same reaction mixture with 10 mM nonradioactive ATP instead of [␥-32 P]ATP. Reaction volumes were increased to accommodate up to 100 g of Bfa1. Incubation was for 4 h. After dialysis against 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM dithiothreitol, 25% glycerol, the phosphorylated Bfa1 (1 g/l) was stored at Ϫ80°C. An identical reaction on Bfa1 was carried out using kinase dead GST-Cdc5KD to provide non-phosphorylated control material.
Nucleotide Exchange and GTPase Assays-Conditions for [␥-32 P]GTP exchange and hydrolysis assays were as described elsewhere (4).

RESULTS
Phosphorylation of Bfa1 by Cdc5-The aim of this work was to characterize the effects of Cdc5-dependent phosphorylation on Bfa1. A major technical requirement was to produce Cdc5 in sufficient quantities and with sufficient activity to fully phosphorylate Bfa1 so that any effects of phosphorylation would be reproducible and have their maximal impact. Cdc5 was initially prepared from Escherichia coli and Baculovirus expression systems but was unable to convert Bfa1 completely to an electrophoretically slower, phosphorylated form (data not shown). Eventually, we developed a yeast-based expression system for a GST-Cdc5 derivative, which is activated by S165D and T238D mutations and stabilized by deletion of 70 N-terminal residues (24,25). In vivo, the activity of this GST-Cdc5 fusion protein was indicated by complementation of a cdc5 deletion mutant under low expression conditions (use of glu-cose with a GAL promoter) (data not shown). GST-Cdc5KD provided a kinase-dead control and has a N209A mutation (23) and an equivalent 70-residue N-terminal deletion. GST-Cdc5 and GST-Cdc5KD were readily purified (Fig. 1A). The differential activities of affinity-purified GST-Cdc5 and GST-Cdc5KD were confirmed using myelin basic protein as a substrate (Fig. 1B). Conditions for efficient phosphorylation of Bfa1 were established using increasing amounts of GST-Cdc5 (Fig. 1C). With suboptimal levels of Cdc5, multiple isoforms of Bfa1 were visible by protein staining in agreement with the presence of multiple phosphorylation sites in the molecule. However, with 5 ng or more GST-Cdc5 per assay, a single slower migrating form of phosphorylated Bfa1 was detected. No change in the mobility of Bfa1 was produced by treatment with equivalent amounts of GST-Cdc5KD. The efficient and specific phosphorylation of Bfa1 GST-Cdc5 was further confirmed by the differential radiolabeling produced by GST-Cdc5 or GST-Cdc5KD (Fig. 1C). Note that in the following examination of the effects of phosphorylation on Bfa1, optimally phosphorylated Bfa1 was prepared with GST-Cdc5 as described above, while non-phosphorylated Bfa1 was prepared by mock-treatment with catalytically inactive GST-Cdc5KD. Thus, our data are specific to Cdc5 kinase and cannot be attributed to a passive Cdc5-Bfa1 interaction or to unspecified contaminating material in the kinase preparation.

Phosphorylation of Bfa1 by Cdc5 Has No Effect on Inhibition of Tem1 GTP Hydrolysis and Dissociation-
The effects of phosphorylation of Bfa1 on Tem1 activity were first examined in the absence of Bub2. Under these conditions unphosphorylated Bfa1 inhibits both the intrinsic GTP hydrolysis and GTP dissociation activities of Tem1 (4). The inhibitory effects of Bfa1 on GTP hydrolysis and GTP dissociation were monitored by loading Tem1 with [␥-32 P]GTP in the presence of varying amounts of non-phosphorylated or phosphorylated Bfa1. Amounts of [␥-32 P]GTP bound to Tem1 were assayed before and after 5-min Cdc5 Phosphorylation Reduces Bfa1/Bub2 GAP Activity 14592 incubation at 30°C. In the absence of any Bfa1, 50 -55% of bound [␥-32 P]GTP was lost from Tem1 in 5 min (Fig. 2). This reflects the loss of [␥-32 P]GTP by the intrinsic GTP hydrolysis and GTP dissociation activities of Tem1 (4). In the presence of non-phosphorylated Bfa1 there was a progressive increase in the amount of [␥-32 P]GTP remaining bound to Tem1 resulting from the inhibition of GTP hydrolysis and GTP dissociation as seen previously (4). When phosphorylated Bfa1 was used, there was a similar inhibition of GTP loss from Tem1 (Fig. 2). Thus, phosphorylation by Cdc5 does not impair the ability of Bfa1 to interact with Tem1 and inhibit GTP dissociation and hydrolysis.
Phosphorylation of Bfa1 by Cdc5 Reduces GAP Activity with Bub2-The effect of phosphorylation of Bfa1 was next examined in assays of GAP activity where Bub2 is also present. These assays were performed by loading Tem1 with [␥-32 P]GTP in the presence of phosphorylated or non-phosphorylated Bfa1. Reactions were then incubated with or without Bub2. In a kinetics experiment, Tem-[␥-32 P]GTP and Bfa1 were incubated with a single amount of Bub2, and the amount of radioactive GTP bound to Tem1 was assayed at frequent intervals during incubation at 30°C. The addition of Bub2 with non-phosphorylated Bfa1 caused the expected loss of radioactivity (Fig. 3A), which is due to stimulation of Tem1 GTP hydrolysis by Bfa1/ Bub2 GAP activity (4). When phosphorylated Bfa1 was used in parallel assays with Bub2, the kinetics of the GAP activity were clearly reduced by ϳ50%. This level of inhibition was repeatedly detected in multiple assays of GAP activity where levels of filter binding were highly reproducible and varied by no more than 5%. The possibility that the phosphorylation status of Bfa1 was affected by the GAP assay itself was excluded by the observation that there was no change in the level of Bfa1 phosphorylation during the GAP reaction (data not shown). In the absence of any Bub2, ϳ85-90% of the radioactive [␥-32 P]GTP remained bound to Tem1 throughout (Fig. 3A). This high level of Tem-[␥-32 P]GTP results from the inhibition of the intrinsic GTP dissociation and GTPase activities of Tem1 by Bfa1 as described above (Fig. 2).
To further characterize the influence of Cdc5 phosphorylation on Bfa1, GAP activity was assayed after a fixed time with increasing amounts of Bub2 (Fig. 3B). As seen above, no loss of [␥-32 P]GTP from Tem1 occurred from control reactions incubated with no Bub2, whether or not Bfa1 was phosphorylated.
However, addition of Bub2 resulted in the expected GAP activity as judged by the progressive loss of [␥-32 P]GTP from Tem1 as Bub2 concentrations increased. Significantly, when Bfa1 had been phosphorylated by Cdc5 approximately twice the amount of Bub2 was required to produce the same GAP activity compared with non-phosphorylated Bfa1. Thus, both kinetic and a concentration dependence assays demonstrate that phosphorylation of Bfa1 by Cdc5 reduces the ability of Bfa1 to function as a GAP in association with Bub2. DISCUSSION We have used an in vitro approach to understand the effects of Cdc5 kinase on the regulation of Tem1 and the MEN. We demonstrate that when Bfa1 is phosphorylated by Cdc5, its GAP activity with Bub2 for Tem1 is reduced. The equivalent result of this activity in vivo would be for Cdc5 to increase After addition of excess cold nucleotide, radioactive label bound to Tem1 was assayed at time 0 and after 5 min. q, phosphorylated Bfa1; f, non-phosphorylated Bfa1. The data represent the percentage of initial radioactivity remaining after incubation. Data points are the average of at least two assays, which typically differed by less than 5%.
FIG. 3. Phosphorylated Bfa1 has reduced GAP activity. A, Tem1 was incubated in the presence of 1.6 M phosphorylated (q, E) or non-phosphorylated Bfa1 (f, Ⅺ). Then Bub2 (100 ng) was added (q, f) or not (E, Ⅺ), and the time course of GTP hydrolysis was followed in filter binding assays. B, increasing amounts of Bub2 were added to Tem1-[␥-32 P]GTP in the presence of 1.6 M phosphorylated (q) and non-phosphorylated Bfa1 (f). Amounts of radioactive label bound to Tem1 were assayed before and after 15-min incubation at 30°C. The data represent the percentage of initial radioactivity remaining after incubation. Data points are the average of at least two assays, which typically differed by less than 5%.
Cdc5 Phosphorylation Reduces Bfa1/Bub2 GAP Activity 14593 phorylation on Bfa1 activity. One possibility is that the two groups observed different extents of phosphorylation and only complete phosphorylation was inhibitory. This idea stems from more recent observations of Pereira and co-workers (15) who showed that a phosphorylated form of Bfa1 is already present in a metaphase arrest. Upon progression into anaphase, Bfa1 is then further phosphorylated. Although it is not clear whether both changes are caused by Cdc5, there does appear to be two phosphorylation events on Bfa1: one before metaphase that does not inhibit the Bfa1/Bub2 GAP activity and a second during anaphase that could participate in activation of MEN by inhibiting its GAP activity.
In considering how phosphorylation affects Bfa1 and GAP activity, Hu et al. (23)proposed that phosphorylated Bfa1 has a lower affinity for Tem1 as judged by co-immune precipitation. However, our in vitro results show that phosphorylated and non-phosphorylated forms of Bfa1 have equal affinities for Tem1 as judged by their similar abilities to inhibit GTP hydrolysis and dissociation. An alternative mechanism may involve interference in the interaction of phosphorylated Bfa1with Bub2 rather than with Tem1. It is notable that a C-terminal deletion form of Bfa1 lacking the domain responsible for Bub2 interaction still has partial in vitro GAP activity in combination with Bub2 even though the two proteins can no longer interact (4). This might explain why we observe only a partial inhibition of Bfa1/Bub2 GAP activity after Bfa1 phosphorylation. Moreover, it has been recently shown that Bub2 is also phosphorylated in a cell cycle-dependent manner, and it was proposed that this modification could affect GAP activity (26). Since GAP activity is reduced after phosphorylation of the mammalian GAPs, RGS16 and p120-GAP (27,28), the complete inhibition of Bfa1/Bub2 GAP activity might only be achieved by phosphorylation of both components.
The partial inhibitory effect of Bfa1 phosphorylation on GAP activity might also reflect the multiple levels of regulatory mechanisms controlling mitotic exit. Indeed, cell cycle progression is apparently normal in ⌬bfa1 mutants, at least without spindle perturbations. Therefore, timely MEN activity is not only provided by inhibition of Bfa1 as ⌬bfa1 cells have a permanent rather than a temporal loss of GAP activity. Further regulatory input comes from Lte1, which is a putative positive regulator for Tem1 (11). Although activity of Lte1 is dispensable under normal growth conditions (11,29), disruptions elsewhere in the overall activation of the MEN can make Lte1 activity essential (14). Furthermore, recent data also implicate PAK (p21-activated kinase)-like kinases in the positive regulation of mitotic exit (30). Thus, there seems to be a high degree of redundancy in the overall control of mitotic exit which is governed by the antagonistic and opposing effects of a number of regulatory events, of which phosphorylation of Bfa1 is just one example.
There are numerous indications that MEN-like regulatory pathways are evolutionarily conserved. The MEN is closely related to the SIN (septation initiation network) pathway of Schizosaccharomyces pombe, where the Tem1-like small GTPase, Spg1, is again regulated by a two component GAP comprising Cdc16 and Byr4, which are homologous to Bub2 and Bfa1 (reviewed in Refs. 1 and 2). As with the MEN of budding yeast, the activity of the SIN depends on a polo kinase, Plo1, which is the S. pombe homologue of Cdc5. It is not yet known whether Byr4 is phosphorylated by Plo1, although considerable genetic evidence points to Cdc16/Byr4 being affected by Plo1 (31). In fission and budding yeasts the polo-like kinases may also stimulate SIN/MEN activity downstream of their respective two-component GAPs, although in neither species is it clear how this stimulation occurs (12,32,33). Polo kinase homologues extend from yeasts to humans and are ubiquitously implicated in mitotic regulation. Similarly, homologues of many other components of the MEN have been recognized in higher eukaryotes (summarized in Ref. 1) and roles for Rho-like GTPases, of which Tem1 is an example, have also been seen in late mitosis and cytokinesis in higher eukaryotes (34,35). Together, these observations suggest conservation in the MENlike pathways and their regulation by polo-like kinases by mechanisms such as the one described here.