In yeast, the pseudohyphal phenotype induced by isoamyl alcohol results from the operation of the morphogenesis checkpoint

The yeast Saccharomyces cerevisiae can grow either in a unicellular budding form or in branching chains of elongated cells known as pseudohyphae (Gimeno et al.,1992). Pseudohyphae form in diploids in response to nitrogen-limited growth, and in other stress and starvation conditions. It is thought that pseudohyphae formation is a strategy by which a normally sessile organism can forage for microenvironments that are more favourable. The pathways by which pseudohyphae form in response to nitrogen-limitation have been extensively studied (for a review, see Rua et al., 2001). Two signal transduction pathways are required involving the mating pheromone MAP kinase module and cAMP/PKA, respectively. These pathways converge on elements in the complex promoters of genes such as FLO11. Filamentous growth requires an extended G2 and a delay in the switch from polarised to isotropic growth of the bud (Kron et al., 1994). This involves inhibition of the activity of the Clb2-Cdc28 kinase, but the way in which this comes about is currently unclear. However, it is known that it does not depend upon inhibitory tyrosine phosphorylation by Swe1, because pseudohypha formation occurs normally during nitrogen-limited growth in a swe1Δ mutant (Ahn et al.,1999).

Pseudohyphae can also form upon exposure to branched-chain or`fusel' alcohols such as isoamyl alcohol (IAA) and 1-butanol(Dickinson, 1996; La Valle and Wittenberg, 2001; Lorenz et al., 2000). Fusel alcohols are formed by the catabolism of branched-chain amino acids such as leucine. Reliance on such a poor nutrient source may also act to trigger pseudohyphae formation as a foraging mechanism to search for better conditions. The mechanism by which fusel alcohols trigger pseudohyphae formation is poorly understood, but it is clearly different from that operating during nitrogen-limited growth because it has a distinct set of genetic requirements. For example, butanol-induced pseudohyphal formation is dependent on SWE1 (La Valle and Wittenberg, 2001) but is independent of genes such as FLO8 and FLO11 that are required for pseudohyphal induction during nitrogen-limited growth (Lorenz et al., 2000). Recently, it has been shown that 1-butanol causes a rapid cessation of translation by targeting the eIF2B translation initiation factor (Ashe et al., 2001). This effect is strain specific, affecting only certain samples of the W303-1A strain. The difference between the strains was tracked down to a P180S variation in Gcd1. As both strain subtypes reacted to 1-butanol by forming filaments, the relevance of butanol-induced inhibition of translation to filament formation is currently unclear.

We show that shortly after exposure to IAA, a series of small buds form accompanied by the appearance of multiple ectopic septin rings in the absence of nuclear division. These events are wholly dependent on Swe1, which inhibits Clb2-Cdc28 by phosphorylation of tyrosine 19. We further show that this process is dependent on Slt2 (Mpk1), the cell integrity MAP kinase, and that Slt2 is activated upon exposure to IAA. However, we show that tyrosine phosphorylation of Cdc28 is not dependent on Slt2, instead Slt2 acts as a negative regulator of Mih1, the tyrosine phosphatase that reverses the inhibitory phosphorylation applied by Swe1. Taken together these observations show that IAA acts by inducing the Swe1-dependent morphogenesis checkpoint.

Table 1 shows strains used in this study, all were congenic to Σ1278b. All experiments were carried out at 26°C.

Gene deletions and the Cdc3-GFP fusion, were constructed as described(Longtine et al., 1998), using the plasmids pFA6a-kanMX6 and pFA6a-GFP-(S65T)-kanMX6, respectively. The integrity of all constructs was confirmed by PCR. A full list of oligonucleotides used is presented in Table 2.

Cells were harvested by centrifugation, washed in PBS and snap-frozen in liquid nitrogen. After thawing, cells were washed once in STOP solution(1× PBS + 10 mM NaN₃ + 50 mM NaF) and once in 20%trichloroacetic acid (TCA). Pellets were resuspended in 200 μl of 20% TCA with glass beads. Cells were disrupted by three 10-second cycles of agitation in a Ribolyser (Hybaid) set at a speed of 6.5. Extracts were separated from glass beads by piercing a hole at the bottom of the Eppendorf tube and centrifuging at 2000 g for 5 minutes. TCA precipitated proteins were then obtained by centrifuging 5 minutes at 28,000 g and by discarding supernatant. The pellet was resuspended in 200 μl of 2× electrophoresis sample buffer containing 250 mM Tris pH 8 and boiled for 5 minutes. Electrophoresis gels were loaded with 45 μl of this TCA extract. Antibodies used for detection of proteins were as follows:anti-phospho-tyro-Cdc28 (Y19) (Cdc2-Tyr15; Cell Signaling Technology);anti-Cdk1/Cdc2 (PSTAIR) (Upstate Biotechnology); anti-phospho p44/42 (Slt2)(New England Biolabs); anti-rabbit IgG (H+L) (Jackson ImmunoResearch Labs);anti-mouse IgG-HRP and anti-rabbit IgG-HRP (BabCO). Detection of phospho-Cdc2(Tyr15) and of diphospho-Slt2 (using a three antibody protocol to enhance sensitivity), were done as described previously(Harrison et al., 2001). Bands were visualised with ECL solution (Pharmacia) using the western blot reader GeneGnome (Syngene Bio Imaging). Images were acquired with GeneSnap 4.00.00(Synoptics Ltd), and analysed using GeneTools 3.00.22 (Synoptics Ltd). Loading controls were either Cdc11, detected using rabbit anti-Cdc11 polyclonal antisera (Santa Cruz), or Cdc28 detected by rabbit anti-PSTAIR polyclonal antisera (Upstate Biotechnology). Each experimental value was normalised with respect to the signal from the appropriate loading control.

Cell suspensions were sonicated for three seconds to separate cells that remained associated despite having completed cytokinesis. When cells were to be examined for GFP and DAPI fluorescence, they were mixed with an equal volume of mounting medium containing 0.1 mg/ml DAPI. Cells that were only to be examined for DAPI staining were fixed in 70% ethanol for 1 hour and then washed and resuspended in 1× PBS buffer and 1 μl of DAPI (0.05 mg/ml)was added to 50 μl of cell suspension and incubated at room temperature for 1 hour. Actin was visualised using TRITC-conjugated phalloidin (Sigma)following the published protocol (Lee et al., 1998).

Cells were washed twice with 500 μl of 0.1 M K₂HPO₄, 0.1 M KH₂PO₄ and 1.2 M sorbitol, resuspended in 500 μl of the same solution plus 1.2 μlβ-mercaptoethanol and 20 μl of 5 mg/ml zymolyase 20-T, and incubated for 40 minutes at 37°C

One ml samples of cell culture were centrifuged at 4000 gfor 3 minutes and washed in 200 μl of distilled water; they were then fixed with 1 ml of ice-cold 70% ethanol and stored at 4°C. The fixed cells were collected by centrifugation at 4000 g for 3 minutes, and then resuspended in 1 ml of 50 mM sodium citrate (pH 7.0). RNA was digested by the addition of 25 μl of 20 mg/ml RNase A, followed by incubation for 3 hours with gentle agitation at 37°C. Then, 500μl of 5 μg m^–1 propidium iodine was added and the samples were kept at 4°C overnight. Before flow cytometry, cells were briefly sonicated to disrupt clumps. DNA content was analysed in a Beckton-Dickinson cell sorter (Franklin Lakes, NJ, USA). After fixation,haploid cells were treated with zymolyase, as described above, to separate cells in filaments.

Cells were examined with a Leica DMLB fluorescence microscope. Digital images were acquired with a cooled CCD camera (Princeton instruments, model RTE) linked to an Apple Macintosh G4 computer running Open Lab software version 2.2.5 (Improvison, Warwick, UK). Images were exported as TIF files and edited in Adobe Photoshop version 5.5.

To date, investigations on the action of isoamyl alcohol (IAA) have reported the effects of prolonged exposure in liquid culture or the appearance of micro-colonies on agar plates. We investigated the early events upon exposure to IAA. Strain CMS2 was constructed containing a heterozygous CDC3/CDC3-GFP fusion in the Σ1278b background, which shows the strongest response to IAA. Preliminary experiments established that 0.5% IAA was the optimum concentration to give a clear response without affecting viability.

CMS2 was exposed to 0.5% IAA in liquid culture, and samples were withdrawn at 1-hour intervals for 8 hours. The samples were examined by DIC microscopy to record their overall appearance, fluorescence microscopy to visualise Cdc3-GFP and DAPI-stained to visualise the nuclei. Fig. 1A shows representative cells at various times during the course of the experiment Within 1 hour of IAA treatment, small ectopic patches of septin appeared at the tips of pre-existing buds (arrows, 1- and 2-hour time points). This was followed by the appearance of further unipolar buds forming chains of up to five small buds projecting from the mother cell by 8 hours. The formation of each bud was accompanied by the appearance of a bright septin ring at its neck, which often co-existed with the fainter ring that remained after the previous round of budding (e.g. 2-hour time point). Quantification showed that the interval between the formation of each bud was approximately 60 minutes(Fig. 1B), which is faster than the 110-minute doubling-time of the untreated control culture(Fig. 1D). By 8 hours, 50% of the cells had two or more buds (Fig. 1B). In about 20% of the filaments, large round cells were observed at either end of the chain (Fig. 1A,C). Some filaments produced side branches (arrow, Fig. 1A).

Initially, the nucleus in the mother cell remained undivided. However,after 5 hours mitosis began in some cells, and filaments with more than one nucleus appeared (Fig. 1A, 6 and 8-hour time points). By 8 hours approximately 28% of the filaments contained two nuclei (Fig. 1C). The majority (22.5% of the total population) of these nuclei were in a large cell at each end of a chain of small buds. One of these round cells often displayed a double septin ring at the junction with the filament (open arrow, Fig. 1A, 8-hour time point),suggesting that events leading to septum formation were resuming. In a smaller fraction of filaments at 8 hours, one of the nuclei was located in the chain of small buds (5% of the total). Filaments with more than two nuclei appeared after 6 hours, but such filaments had disappeared from the population by 8 hours. The proportion of cells with two nuclei did not increase after 24 hours incubation. Consistent with the delay of nuclear division, cell number increase was immediately halted for 4 hours after IAA addition before showing a slow increase, possibly due to the resumption of nuclear division and cytokinesis (Fig. 1D). Overall,these data show that the filaments produced upon IAA treatment arise through repeated rounds of budding, in the absence of nuclear division.

Haploid cells also responded to the IAA exposure by producing chains of anucleate buds, although the process took longer than in diploid cells. After 8 hours exposure to IAA, only one or two anucleate buds had formed (data not shown). However, by 24 hours, chains of up to four buds had formed in 100% of the cells (Fig. 2). These cells showed considerable heterogeneity in the width of the filaments: some were the same width as the mother cell (barbed arrows Fig. 2B), others were much narrower (open arrows Fig. 2B),while some contained a mixture of wide and narrow compartments (filled arrows with barbed tails Fig. 2B). After 24 hours exposure, the filaments become populated with nuclei: of 173 filaments examined, 56 (32%) remained anucleate, 59 (34%) had one nucleus in the filament, 29 (17%) had nuclei at both ends, and 29 (17%) had more than one nucleus within the filament. There was no obvious pattern in the way that the compartments of the filaments acquired nuclei as mitosis was observed at both the neck of the filament and the mother cell or wholly within the filament. As was the case in diploids, where there was a nucleus at both ends, the compartments containing the nuclei were large and round.

We found that in a haploid Σ1278h strain, the Cdc3-GFP fusion protein was mis-localised into large cytoplasmic bars, even in cells not treated with IAA. We therefore used immunocytofluorescence with polyclonal antisera raised against Cdc11 to investigate whether multiple septin rings formed in haploids(Fig. 2C,D). As with diploids,septin rings were also observed in the absence of nuclear division. Immunocytofluorescence requires the cell wall to be removed with zymolyase,however, the compartments remain associated showing that they are coenocytic(Fig. 2C,D). After 16 hours culture, the compartments that contained nuclei separated upon zymolyase treatment, showing that they were physiologically autonomous(Fig. 2E).

In S. cerevisiae, the morphogenesis checkpoint has been shown to delay nuclear division when bud formation has been disrupted. It is triggered in two different situations. First, when the actin cytoskeleton has become depolarised because of genetic lesions such as cdc24 or cdc42 mutations or by treatment with latrunculin A(Lew and Reed, 1995a; Lew and Reed, 1995b; McMillan et al., 1998). Second, when septin ring formation has been disrupted by mutations affecting the component septins or proteins such as Elm1, Cla4 and Gin4 that are required for the proper organisation of the septins(Barral et al., 1999). The delay in nuclear division observed above led us to ask whether Swe1 was involved in the effects seen with IAA. We found that a swe1Δmutation abrogated the response to IAA in haploid(Fig. 3A) and diploid strains(data not shown). This observation has recently been independently reported elsewhere (La Valle and Wittenberg,2001). Swe1 phosphorylates Cdc28 at tyrosine 19(Sia et al., 1996). Consistent with this, IAA treatment resulted in tyrosine phosphorylation of Cdc28 in a Swe1-dependent manner (Fig. 6). Such inhibitory phosphorylation of Cdc28 would be expected to delay mitosis. Indeed, FACS analysis showed an increase in the number of cells in G2(Fig. 3B), indicating that mitosis was delayed, although there was not a complete block. The peak of phosphotyro-Cdc28 levels at 3 hours (Fig. 6) corresponds to start of the increase in the number of cells in G2 (Fig. 3), whereas the fall after 5 hours may be responsible for the resumption of nuclear division(Fig. 1C).

The morphogenesis checkpoint is induced by defects in actin polarisation or in the septin cytoskeleton (see above). Since septin rings appear normal during IAA treatment, we investigated whether there were any defects in actin polarisation. As shown in Fig. 4, actin was highly polarised towards the tip of the growing bud chains during IAA treatment. Therefore, a failure to polarise actin is unlikely to be responsible for the induction of the morphogenesis checkpoint.

The cell integrity Pkc1/Slt2 MAP kinase pathway has been shown to be required for the operation of the morphogenesis checkpoint, although tyrosine phosphorylation of Cdc28 persists in an slt2Δ mutant(Harrison et al., 2001). We determined whether Slt2 is required for the IAA-induced response and found that an slt2Δ mutant failed to respond to IAA by producing filaments (Fig. 5A). Slt2 is activated by dual threonine and tyrosine phosphorylation by Mkk1 and Mkk2. Upon IAA treatment of a wild-type cell, such phosphorylation is detected with polyclonal antisera specific to the active diphospho form of Slt2 within 30 minutes (Harrison et al., 2001; Martin et al., 2000)(Fig. 5B). Tyrosine phosphorylation of Cdc28 persisted in an slt2Δ mutant(Fig. 6), indicating that although Slt2 is required for the response to IAA, it does not act upstream of Cdc28 tyrosine phosphorylation. We also investigated whether tyrosine phosphorylation of Cdc28 acted upstream of Slt2 activation by examining the effect of a swe1Δ mutation. The membrane that was probed with anti-phosphotyro-Cdc28 was also probed with anti-active Slt2 antisera. Although there were some differences compared with wild type, phosphorylation of Slt2 still occurred in a swe1Δ mutant, indicating that Slt2 activation is not downstream of Cdc28 tyrosine phosphorylation. Interestingly,activation of Slt2 peaks after 1 hour while the peak of Cdc28 phosphorylation occurs after 3 hours (Fig. 6). Thus, Slt2 activation occurs much more rapidly than Cdc28 tyrosine phosphorylation.

During the operation of the morphogenesis checkpoint, Cdc28 tyrosine phosphorylation is controlled both by the rate of phosphorylation by Swe1 and dephosphorylation by Mih1 (Sia et al.,1996). Harrison et al.(Harrison et al., 2001) showed that during the operation of the checkpoint induced by actin depolarisation,an mih1Δ mutation restored the checkpoint function that was defective in an slt2Δ mutant, suggesting that Slt2 acts as a negative regulator of Mih1, not an upstream activator of Swe1. This hypothesis predicts that an mih1Δ mutation should rescue the filamentation defect of an slt2Δ mutant. We determined whether this was the case in the response to IAA. Fig. 7 shows that an mih1Δslt2Δ double mutant does filament in response to IAA in a manner identical to that seen in a wild-type strain.

We have shown here that the effect of exposure of S. cerevisiaecells to IAA is the appearance of a series of small buds that form at a rate faster than the doubling-time of the parent strain. At the same time, nuclear division is delayed. As a result, linear chains of small anucleate buds extend from the mother cell after 3 hours. In many cells, the first bud appears on the surface of a very small bud that has formed just prior to the IAA exposure, and before the occurrence of mitosis and cytokinesis that would normally be required to permit the daughter cell to bud. This loss of budding control is accompanied by loss of both temporal and spatial control of septin localisation. Patches of septin appear before each successive bud, and in some cells, multiple septin patches can be seen. After 5 hours nuclear division resumes and approximately 28% of the filaments eventually become binucleate. The majority of binucleate filaments were found in cases where large round cells were located at either end of chains of between three and five anucleate buds. A similar process occurs in haploid cells, although the process is a little slower so that chains of buds were only fully formed after 24 hours. These filaments contain a higher proportion of nucleated cells than was observed in diploids and there was also considerable heterogeneity in the width of the filaments. It is not clear why the process should take longer in haploids; however, it is clear that a similar mechanism is operating.

The delay in cell division and continued budding observed upon IAA treatment suggested to us that the morphogenesis checkpoint has been induced. This checkpoint ensures that mitosis only takes place when bud formation is proceeding normally. It has been extensively studied in two situations. Firstly, when the actin cytoskeleton has been depolarised by latrunculin A or by cdc24 or cdc42 mutations (Lew and Reed, 1995a; Lew and Reed,1995b; Lew and Reed,1993; McMillan et al.,1998). Secondly, where mutations have disrupted the organisation of the septin ring (Barral et al.,1999; Bouquin et al.,2000; Longtine et al.,2000; Sreenivasan and Kellogg,1999). Both of these situations cause a delay of mitosis. In addition, disruption of the septin ring prolongs polarised growth of the bud,resulting in cells that are similar in appearance to those treated with IAA. The checkpoint is mediated by the inhibitory tyrosine 19 phosphorylation of Cdc28 by Swe1 and thus cannot occur in swe1Δ mutant. The checkpoint also requires the action of the Pkc1/Slt2 MAP kinase pathway(Harrison et al., 2001). However, checkpoint function is restored in a slt2Δ mih1Δ mutant.

We have shown that filament formation following IAA treatment in both haploids and diploids requires Swe1 and that Cdc28 becomes tyrosine-phosphorylated in a Swe1-dependent fashion. Furthermore, the IAA response is also dependent on Slt2 and we have shown that Slt2 is rapidly activated by phosphorylation upon IAA treatment. Moreover, as is the case when the checkpoint is induced by actin depolarisation, Cdc28 phosphorylation is not dependent on Slt2, but the loss of checkpoint function in an slt2Δ mutant is rescued by an mih1Δ mutation. Taken together these observations strongly suggest that the same Swe1-dependent, morphogenesis checkpoint operating upon latrunculin A treatment also operates to delay mitosis in response to IAA treatment.

Harrison et al. interpreted the rescue of checkpoint function in a slt2Δ mih1Δ mutant by placing Slt2 as a negative regulator of Mih1. This model satisfactorily explains the genetic data. However, we reproducibly failed to see any decrease of phophotyro-Cdc28 in an slt2Δ mutant. Moreover, significant levels of phosphotyro-Cdc28 clearly persisted in an slt2Δ mutant upon latrunculin A treatment (Harrison et al.,2001). One explanation for this inconsistency is that the western blots using the anti-phosphotyro Cdc28 antibody do not detect the residual levels of active Cdc28 and that in the slt2Δ mutant, sufficient active Cdc28 remains to abrogate the checkpoint. An alternative explanation is that the Western blots report the total cellular level of phosphotyro Cdc28,while the Slt2/Mih1 pathway may act to dephosphorylate Cdc28 at a specific location where its action is critical for cell cycle progression. These explanations are not mutually exclusive.

La Valle and Wittenberg (La Valle and Wittenberg, 2001) also showed that Swe1 was required for the production of pseudohyphae in response to 1-butanol and we have found that pseudohyphae do not develop in response to 1-butanol in swe1Δand slt2Δ mutants. Moreover, the response was restored in an slt2Δ mih1Δ double mutant (data not shown). Thus the induction of filamentous growth in response to 1-butanol is also likely to be due to the morphogenesis checkpoint. In the light of these results, we suggest that previous reports of alcohol-induced filamentous growth(Lavalle and Wittenberg, 2001;Lorencz et al., 2000) should be re-interpreted as acting through the morphogenesis checkpoint.

It is not clear why IAA induces the morphogenesis checkpoint. Polarisation of the actin cytoskeleton, bud formation and septin ring formation are not affected by IAA treatment. It is possible that the initial formation of ectopic septin patches triggers the checkpoint. However, the septin rings that subsequently form appear normal. It is more likely that IAA is acting to trigger the morphogenesis checkpoint for some other reason. If this is the case, then IAA forms a third trigger of the morphogenesis checkpoint in addition to defects in the actin and septin cytoskeletons. The induction of the checkpoint when both the actin and septin cytoskeletons are functioning normally, shows that operation of the checkpoint does not simply result in prolonged polarised growth, but apparently drives new rounds of bud formation including the formation of new septin rings.

Both IAA and nitrogen-limited growth induce the formation of pseudohyphae that look superficially similar and share some common requirements, such as components of the mating pheromone-response MAP-kinase pathway. However, our results presented here show that the production of pseudohyphae occur differently in each case. In the case of pseudohyphae induced by nitrogen-limited growth, G2 is extended producing elongated cells, but no anucleate buds form. Moreover, although Swe1 has been reported to contribute to the formation of nitrogen-limited pseudohyphae in certain conditions(La Valle and Wittenberg,2001), it is not essential for the formation of pseudohyphae during nitrogen-limited growth because normal pseudohyphae are generated in a homozygous swe1Δ/swe1Δ mutant(Ahn et al., 1999) (our unpublished observations).

At this stage it is not clear whether the induction of the morphogenesis checkpoint by IAA is a response to a pathology or whether the morphogenesis checkpoint plays a role in the normal physiological response to an agent that is commonly encountered by yeast cells in their natural environment. If the response is physiological, then the elongated cell chains resulting from IAA treatment, may still be regarded as pseudohyphae. The Swe1-dependent pathway inducing this phenotype would thus be an alternative route to the production of pseudohyphae in addition to the pathways operating in response to nitrogen-limited growth. In this connection it is interesting to note that La Valle and Wittenberg (La Valle and Wittenberg, 2001) found that Swe1 was required for the residual pseudohyphal formation observed in the tec1Δ mutant in response to nitrogen-limited growth. Thus the Swe1 pathway may contribute to pseudohyphal formation in response to nitrogen-limiting growth, but does not play an essential role under normal circumstances.

C.M.A. was supported by a fellowship from CONACYT (Mexico). We thank Ray Wightman for critical reading of the manuscript.