Activated Calcineurin Confers High Tolerance to Ion Stress and Alters the Budding Pattern and Cell Morphology of Yeast Cells*

The PP2B protein phosphatase, also known as cal- cineurin, is a regulator of ion homeostasis in yeast cells. We have investigated the physiological consequences of constitutive expression of a recombinant form of calcineurin in which the Ca 2 (cid:49) /calmodulin-binding and autoinhibitory domains of the catalytic subunit were de- leted. The concomitant expression of the regulatory subunit along with the truncated catalytic subunit re- sulted in high tolerance to toxic levels of Na (cid:49) and Li (cid:49) . This activated form of calcineurin substituted for the Na (cid:49) stress signal to promote the expression of the ENA1 gene, encoding a P-ATPase pump, and to induce the transition of the K (cid:49) uptake system to the high affinity mode that restricts influx of Na (cid:49) and Li (cid:49) . In addition, the transcriptional responsiveness of ENA1 to Na (cid:49) stress was enhanced. These results demonstrate that calcineurin has a pivotal role in a signaling cascade activated by ion stress in yeast. Moreover, we found that changes in the level of calcineurin activity affected budding pattern and cell morphology. Cells expressing the truncated calcineurin were elongated and budded in an unipolar pattern, whereas calcineurin-deficient mu- tants budded randomly. These results suggest that calcineurin

Calcineurin (CaN) 1 is a Ca 2ϩ -and calmodulin-dependent protein phosphatase consisting of a catalytic subunit (CNA) and a regulatory subunit (CNB). CNA is inactive in the absence of the Ca 2ϩ -binding CNB subunit and the heterodimer CNA/ CNB is activated by Ca 2ϩ and calmodulin (CAM) binding (1). The regulation of CaN activity by Ca 2ϩ is dual because Ca 2ϩ / CAM binding increases the V max and Ca 2ϩ /CNB increases the affinity of CaN for its substrate (2,3). At least four functional domains have been identified in the CNA subunit. The aminoterminal catalytic core, which has extensive similarity to type 2A protein phosphatases, is followed by a CNB-binding domain, a CAM-binding site, and an autoinhibitory domain ( Fig.  1) (4). The autoinhibitory domain likely functions as a competitive inhibitor that binds at or near the active site of the enzyme (5). The binding of Ca 2ϩ /CAM to CNA apparently activates CaN by displacing the autoinhibitory domain. Proteolytic removal of the calmodulin-binding and autoinhibitory domains of CaN yielded a core CNA/CNB heterodimer that was active in vitro in the absence of Ca 2ϩ /CAM, yet retained some responsiveness to free Ca 2ϩ , presumably through the Ca 2ϩbinding CNB subunit (2,6). Transfection of Jurkat T-cells with a recombinant CNA subunit lacking the CAM-binding and autoinhibitory domains induced Ca 2ϩ -independent activation of the interleukin-2 promoter, suggesting that the carboxylterminal truncation was sufficient to activate CaN in vivo (5).
Although an increasing number of CaN-dependent cellular processes are being identified in different eukaryotic systems, regulation of ion fluxes is clearly a principal function of CaN. In the yeast Saccharomyces cerevisiae, CaN is essential for tolerance to Na ϩ , Li ϩ , and Mn 2ϩ (7)(8)(9), regulates Ca 2ϩ transport into the vacuole (10,11), and participates in the regulation of the K ϩ uptake system (8). Further, simultaneous inactivation of CaN and the vacuolar H ϩ -ATPase are synthetically lethal, and regulation of the plasma membrane H ϩ -ATPase by CaN has been suggested (12,13).
The monovalent cations K ϩ and Na ϩ share a common uptake system in S. cerevisiae. When challenged with growth inhibitory concentrations of Na ϩ , yeast cells alter the kinetic properties of the K ϩ transport system increasing its affinity for K ϩ to restrict the influx of Na ϩ (14). This response is dependent on a functional TRK1 gene, that encodes a putative K ϩ transporter involved in high affinity K ϩ uptake (15,16). In addition, Na ϩ induces the expression of ENA1, a gene encoding a P-type ATPase that mediates Na ϩ efflux, to restore low cytosolic Na ϩ levels (17). We have shown that S. cerevisiae cells deficient in CaN activity become hypersensitive to Na ϩ and Li ϩ (an analog of Na ϩ ) because of an insufficient induction of ENA1 that results in low net ion efflux (8). Moreover, although CaN mutants display a normal transition to the high affinity mode of K ϩ transport induced by K ϩ starvation, these cells fail to enter the high affinity mode in response to Na ϩ stress. These results imply that there are two signaling pathways controlling appropriate K ϩ homeostasis in S. cerevisiae . One of these signaling cascades responds to K ϩ starvation by an, as yet unknown, CaN-independent mechanism. The other signaling pathway is CaN-dependent and coordinates gene expression and activity of monovalent cation transporters in response to Na ϩ stress (8).
CaN has also been implicated to function in the acquisition of cell polarity in eukaryotes. In the fission yeast Schizosaccharomyces pombe, a null CaN mutant lost cell polarity and pro-duced branched cell divisions. Overexpression of CaN caused abnormal cell and nuclear shapes and spindle pole body positioning (18). During the initial outgrowth of neurites, CaN was localized to the tips of growth cones associated with the neuronal cytoskeleton, and CaN inhibitors prevented axonal elongation (19). A strikingly similar spatial distribution has been described for CAM in S. cerevisiae. In unbudded cells, CAM concentrated at the site of bud formation before bud emergence, then concentrated at the tip of the growing bud, and finally localized to the neck region before cytokinesis (20). These results suggest that CAM-regulated proteins may participate in polarized cell growth. No data on the localization of CaN in yeast cells are yet available.
Here we show that constitutive activation of CaN results in an increased tolerance to toxic levels of Na ϩ and Li ϩ . The activation of CaN can both substitute for and enhance Na ϩ stress signaling that mediates ion tolerance. Moreover, increased CaN activity elicits a developmental switch in the budding pattern of the yeast S. cerevisiae, whereas CaN depletion may interfere with acquisition of cell polarity.

Genetic and Recombinant DNA Methods-
The entire open reading frame of the CNA2 gene was amplified by PCR using a pair of specific oligonucleotides that annealed to the start codon (CCCTCGAGGCAAT-GTCTTCAGACGCT, sense strand) and to the stop codon region (CTC-GAGCTATTTGCTATCATTCTT, antisense strand). Carboxyl-terminal truncated CNA2 alleles were also produced by PCR by combining the ATG-specific primer with either the antisense oligonucleotides AACTC-GAGTTAGGTGTCGTTTTCCAGCTC or CTCGAGTTACATAACAT-TATTTTC to introduce stops codons (underlined) after the amino acid residues Thr 459 or Met 407 , respectively (amino acid numbering refers to CNA2p). All oligonucleotides contained XhoI sites (CTCGAG) to facilitate cloning. PCR products were cloned and partially sequenced to assess the fidelity of the amplification reaction. A expression cassette containing the promoter and transcriptional termination regions of the plasma membrane H ϩ -ATPase gene PMA1 was cloned as a HindIII fragment into YEp351 (for LEU2 selection) ( Fig. 1) or YEp352 (for URA3 selection). These plasmids only differ in the selective marker for yeast transformation. A single XhoI site between the PMA1 promoter and termination signal was used to insert the PCR-amplified CNA2 gene and truncated alleles. For the expression of CNBp, a 1.3-kilobase EcoRV containing the native CNB1 gene was inserted at the SmaI site in the YEp351 or YEp352 polylinker. Functionality of all constructs was determined after transformation of isogenic yeast strains MCY100 and YP9, lacking either the catalytic subunits of CaN (cna1 cna2 mutant) or the regulatory one (cnb1), respectively (Table I). An internal fragment of the FKS2 gene spanning amino acids 219-1338 of the FKS2 protein was amplified by PCR using the synthetic oligonucleotides CCGAAT-TCCCCTCGCAACAC (sense) and GACGGAGTTGCTCAGGTT (antisense). The PCR product was partially sequenced to assess its identity with the FKS2 sequence. After insertion of a 1.3-kilobase XhoI-BamHI fragment containing the HIS3 gene between the SalI and BglII sites of the FKS2 fragment, it was used for one-step gene disruption (21). Disruption of the FKS2 gene was confirmed by Southern blot hybridization.
Biochemical Procedures-For measurement of ion fluxes, cells were grown in AP medium (8 mM phosphoric acid, 10 mM L-arginine, 2 mM MgSO 4 , 0.2 mM CaCl 2 , 2% glucose, plus vitamins and trace elements, brought to pH 6.5 with arginine) (22). The AP medium is essentially free of K ϩ and other alkali cations and was supplemented with KCl as required (indicated in parentheses). Net Li ϩ uptake was measured in cells grown in AP(2.5 mM K ϩ ) to A 550 ϭ 0.3, harvested, and resuspended in MES/Ca 2ϩ buffer (10 mM MES brought to pH 6.0 with Ca(OH) 2 and supplemented with 2% glucose) containing 1 mM KCl and 50 mM LiCl. At the indicated intervals, samples were harvested by filtration, washed, and the Li ϩ content analyzed by atomic absorption spectrophotometry (22). The kinetics of Li ϩ influx into RH16 and RH16(pCAtrB) cells was calculated from the initial rates of Li ϩ uptake at LiCl concentrations ranging from 25 to 200 mM in MES/Ca 2ϩ buffer containing 1 mM K ϩ . Values of Li ϩ content were always normalized to the dry weight of the sample.
Immunodetection of the ENA1 protein in purified plasma membrane preparations was as described (23). For the ENA1 promoter activity analysis, an in frame ENA1-lacZ fusion was obtained by cloning the SalI-EcoRI fragment from ENA1 (Ϫ1384 to ϩ40) into YIp356R (24). The resulting plasmid, pFR70i, was linearized with NcoI to direct integration into the ura3 locus. ␤-Galactosidase activities were measured in cell-free extracts (25) and normalized to the protein concentration. Protein concentrations were determined by the method of Bradford (26).
Flow Cytometry and Microscopy-For flow cytometry analysis, cells in exponential growth were fixed in 80% ethanol, RNase-treated in 50 mM sodium citrate, pH 7.5, and stained with 2 g/ml propidium iodide in the same buffer. For each sample, the fluorescence intensity of 10,000 cells was scored as a measure of relative DNA content. Exponentially growing cultures at similar cell densities were always used for microscopy studies. Intact cells were visualized with Nomarski optics. Cell wall chitin was stained with Calcofluor White after fixation with 4% formaldehyde and visualized by excitation with UV light (27). The budding pattern was determined in cells showing at least three bud scars according to the criteria of Chant and Pringle (28). Cells with bud scars scattered across the surface were considered to have a random budding. Cell dimensions were calculated from micrographs of cells fixed with 4% formaldehyde. Cell length and width were measured at the longest axis of the cell and at the point of maximum width along the cell, respectively. The length/width axial ratio was calculated for each individual cell. Mean values were 1.18 (S.E. ϭ 0.007, n ϭ 123) for control DBY746 cells, and 1.20 (S.E. ϭ 0.008, n ϭ 84) for W303-1A. Since more than 98% of control cells had a length/width ratio under 1.4, cells exceeding this threshold value were considered elongated cells.

Requirements for the Activation of CaN in Vivo-Since
CaN is essential for the Na ϩ and Li ϩ tolerance of yeast cells (7,8), we reasoned that the constitutive expression of an activated form of CaN would increase the tolerance of wild-type cells to these ions. A recombinant CNA subunit (CNAtr1) truncated after the threonine at position 459, to delete both the CAMbinding and autoinhibitory domains, was constructed by PCR. The CNA2 gene was used as a template for PCR, because CNA2p is 4-fold more abundant than CNA1p in yeast cells and deletion of CNA2 gives a more severe salt-sensitive phenotype than deletion of CNA1 (7,29). CNAtr1 was cloned between the  (45) promoter and the transcriptional termination signal of the plasma membrane (H ϩ )-ATPase gene, PMA1. As a control, the full-length open reading frame of CNA2 was also amplified and cloned in the same manner. Deletion of both the CAM-binding and inhibitory domains yielded a functional catalytic subunit since it complemented the NaCl sensitivity of a cna1 cna2 strain (Table II). However, CNAtr1 was unable to rescue the sensitivity of a cnb1 strain (Table II), suggesting that CNAtr1 still required CNB for activity in vivo. Moreover, a truncation of CNA2 (CNAtr2) up to Met 409 that produced a catalytic core retaining the region of high similarity between CNA2 and PP2A and PP1 phosphatases, but where the putative CNBbinding domain of CNA2 had been removed ( Fig. 1), was unable to recover the NaCl sensitivity of a double cna1 cna2 mutant. This failure might be due to the inability of CNAtr2 to bind CNB. These results indicate that the catalytic subunit must bind the regulatory subunit for activity in vivo in spite of the deletion of the CAM-binding and autoinhibitory domains. Therefore, the native CNB1 gene was also cloned in the vector plasmid for the concomitant expression of the regulatory subunit ( Fig. 1). As anticipated, the simultaneous overexpression of the truncated CNAtr1 and the CNB subunits increased the Li ϩ tolerance of wild-type cells by almost 5-fold (Table II). Overexpression of CNAtr1 alone had a modest effect on the Li ϩ tolerance and CNB alone had no effect (Table II). Thus, unlike T-cells, expression of a truncated catalytic subunit lacking the inhibitory domain does not suffice to activate constitutively the calcineurin-dependent pathway in yeast, and the co-expression of a functional regulatory CNB subunit is required. Surprisingly, expression of the full-length CNA2 protein along with the regulatory CNB subunit did not increase substantially the Li ϩ tolerance of wild-type cells (Table II), suggesting that there may be factor(s) that limit the activity of the native CaN but become dispensable in the recombinant CNAtr1/CNB. Effects of Constitutive Activation of CaN on Ion Homeostasis-Previously, we have shown that CaN is an essential component of a signaling pathway that mediates the homeostasis of monovalent alkaline cations in yeast through the coordinate regulation of influx (TRK1) and efflux transporters (ENA1) (8). Presumably, the high Li ϩ tolerance of cells expressing the recombinant CNAtr1/CNB CaN resulted from an increased activity of the ENA1-ATPase since ion efflux is the major determinant for Li ϩ tolerance in S. cerevisiae (14). The expression of the ENA1 gene was examined in YP4-7 cells containing an integrated copy of a ENA1-lacZ translational fusion. These cells were transformed with plasmid pCAtrB and the ␤-galactosidase activity was determined before and after induction with NaCl. As shown in Fig. 2, NaCl increased the transcriptional activity of the ENA1 promoter by 5-fold in wild-type cells. Overexpression of CNAtr1/CNB resulted in a constitutive 3-fold increase in ENA1 promoter activity in the absence of stress, and an 8-fold higher ␤-galactosidase activity when cells were challenged with NaCl. The amount of ENA1 protein accumulated from the wild-type ENA1 gene in these cells was also examined. Immunodetection of ENA1 in plasma membrane fractions by Western blotting showed protein levels in agreement with the transcriptional activities measured (Fig. 2).
To measure ENA1-and TRK1-dependent ion fluxes, we used strain DBY746 and its isogenic derivatives RH16, in which the ENA1-ENA4 tandem array has been deleted, RH2, a ⌬trk1 mutant unable to influx K ϩ in the high affinity mode, and the congenic strain TE12, lacking functional ENA1-4 and TRK1 Full-length and truncated alleles of CNA2 were expressed using an expression cassette containing the promoter and transcriptional termination signal of the plasma membrane H ϩ -ATPase PMA1 gene. The native CNB1 gene was used for expression of the regulatory subunit. These constructs were cloned in the multicopy plasmid YEp351 to express different combinations of calcineurin subunits as indicated. The functionality of all constructs was evaluated by complementation of the NaCl and LiCl sensitivity of isogenic strains MCY100 (cna1 cna2 double disruptant) and YP9 (cnb1 disruptant). The LiCl concentrations that reduced the growth rate of YP4 -7 cells transformed with these plasmids by 50% are indicated as IC 50 . The IC 50 of cells transformed with pCAtr2 was not determined since functional complementation did not occur. Values are the mean Ϯ S.E. of at least two measurements.
Regulatory and full-length catalytic 51.5 Ϯ 6.5 ϩ ϩ pCAtrB Regulatory and catalytic truncated at Thr 459 217.5 Ϯ 1.5 ϩ ϩ F IG. 1. Gene constructs for functional expression of calcineurin subunits. A, organization of functional domains in the catalytic subunit of calcineurin. The amino-terminal part of the protein contains the phosphatase catalytic domain that shares high similarity to PP2A phosphatases. The regulatory CNB subunit-and Ca 2ϩ /CAMbinding sites are shown as striped and dotted boxes, respectively. The carboxyl-terminal part of the protein contains the autoinhibitory domain. The relative positions of truncations to the CNA subunit that eliminate the CAM-binding and autoinhibitory domains alone (CNAtr1) or together with the CNB-binding domain (CNAtr2) are indicated. B, the full-length and truncated alleles of CNA2 were cloned between the promoter and the transcription termination region of the PMA1 gene for constitutive expression. When required, the CNB1 gene was inserted to express the CNB subunit in its native form as shown. The plasmid backbone is YEp351.
genes. DBY746 cells transformed with pCAtrB acquired high tolerance to NaCl and LiCl (results not shown) and showed a substantially lower rate of Li ϩ uptake than control cells (Fig.  3A). Expression of CNAtr1/CNB in ENA1-4 ⌬trk1 cells also reduced Li ϩ uptake (Fig. 3B). In a ENA1-4 ⌬trk1 background, the rate of Li ϩ uptake is determined by the activity of the ENA1p only, since deletion of the TRK1 gene precludes changes in selectivity of K ϩ over Li ϩ (8,14). Moreover, ⌬ena1-4 ⌬trk1 cells had similar rates of Li ϩ uptake regardless of whether they expressed activated calcineurin or not (Fig.  3D). Together, these results indicate a higher rate of Li ϩ efflux mediated by the ENA1 pump in cells expressing the activated form of CaN, in agreement with the enhanced expression of the ENA1 gene (Fig. 2). Similarly, ⌬ena1-4 TRK1 cells expressing CNAtr1/CNB accumulated Li ϩ at a 2-fold lower rate than control cells when transferred to 50 mM Li ϩ and 1 mM K ϩ (Fig. 3C). Since ⌬ena1-4 cells lack significant Li ϩ efflux, this reduction in Li ϩ influx might depend on the transition of the K ϩ transport system to the high affinity mode of K ϩ uptake, with a higher selectivity for K ϩ over Na ϩ and Li ϩ (14,15). In wildtype cells, this switch requires TRK1 and is triggered by Na ϩ stress in a calcineurin-dependent manner (8). To demonstrate a reduced affinity for Li ϩ , TRK1-dependent Li ϩ uptake at several external concentrations of LiCl was measured. RH16 cells were used to avoid interference by the constitutive expression of the ENA1-ATPase in cells transformed with the recombinant CNAtr1/CNB calcineurin. At all LiCl concentrations tested, RH16 cells expressing the activated calcineurin accumulated Li ϩ at a lower rate than control RH16 cells. A doublereciprocal plot of initial rates of Li ϩ influx indicated that cells containing the plasmid pCAtrB had a reduced affinity for Li ϩ without previous induction by NaCl treatment of the high affinity mode of K ϩ transport (Fig. 4). In the presence of 1 mM K ϩ , the apparent K m for Li ϩ was 51.8 mM in control cells and 130.8 mM in cells transformed with pCAtrB. This change in Li ϩ affinity allowed the growth of strain RH16(pCAtrB) at Li ϩ concentrations that prevented the growth of RH16 (data not shown). Besides ENA1 and TRK1, no other significant CaN-dependent Li ϩ transport could be identified since ⌬ena1 ⌬trk1 cells showed similar rates of Li ϩ uptake (Fig. 3D) and LiCl tolerance (data not shown) whether they expressed activated CaN or not.
Taken together, these results indicate that continued activation of CaN results in a permanent state of adaptation to ion stress, i.e. the K ϩ uptake system is in the high affinity mode of transport before induction by Na ϩ stress and the ENA1 protein is expressed constitutively. Furthermore, the transcriptional responsiveness of the ENA1 gene to ion stress is enhanced. A consequence of this precocious and enhanced adaptation is a dramatic increase in survival when the cells overexpressing CNAtr1/CNB are challenged with otherwise lethal concentrations of LiCl and NaCl. Thus, over 90% of YP4-7 cells trans-

FIG. 2. Enhanced expression of ENA1 in cells expressing recombinant CNAtr1/CNB calcineurin.
A, wild-type YP4-7 strain containing an integrated ENA1-lacZ translational fusion was transformed with plasmid pCAtrB for the concurrent expression of the truncated catalytic subunit and the regulatory subunit of calcineurin or with the vector plasmid without insert. ␤-Galactosidase activities were measured in these transformants after incubation in 0.5 M NaCl for 60 min (Na) and in control, non-stressed cells (C). The histogram illustrates the mean and S.D. of ␤-galactosidase activities from triplicate samples. B, plasma membrane fractions of cells treated as described above were isolated, the proteins separated by SDS-PAGE, and blotted to nitrocellulose. The ENA1p was immunologically detected with rabbit anti-ENA1 antiserum and CIAP-conjugated secondary antibody. Constitutive Activation of CaN Alters Budding Pattern and Cell Morphology-Besides a substantial increase in Na ϩ and Li ϩ tolerance, reduced growth rate and abnormal cell shapes were also brought about by the constitutive activation of CaN. Wild-type cells of the DBY746 background had an average length/width axial ratio of 1.18 (S.E. ϭ 0.007, n ϭ 123), with 98.4% of them having an axial ratio below 1.4 (see "Materials and Methods"). However, cells expressing CNAtr1/CNB had an average axial ratio of 1.78 (S.E. ϭ 0.026, n ϭ 335) and 74.9% of the cells displayed an elongated shape with an axial ratio ranging from 1.4 to 3.2 (Fig. 5). Overexpression of wild-type calcineurin had a small effect on the cell shape, since only 13.8% of the cells transformed with pCAB had an axial ratio higher than 1.4 (average value was 1.27 Ϯ 0.01, n ϭ 217). Often, elongated mother and daughter cells did not separate completely, thus producing chains of elongated cells. Staining of the chitin ring formed during cell division with Calcofluor White demonstrated that arrays of elongated cells were composed of clonally related siblings (Fig. 5). These arrays, however, were easily disrupted by mild sonication, indicating that cytokinesis and cell separation were completed. Older cells in the arrays of elongated cells appeared highly vacuolated and were poorly stained with vital dyes (data not shown). Neither the slow growth nor the elongated cell shape in cells expressing CNAtr1/CNB were related to altered homeostasis of Na ϩ or K ϩ because the plasmid pCAtrB induced growth retardation and aberrant morphology both in the DBY746 derivatives RH16 (⌬ena1-4), and TE12 (⌬ena1-4 ⌬trk1) that lack calcineurin-dependent Na ϩ and K ϩ ion transporters (Fig. 5).
Besides ion homeostasis, other cellular processes regulated by CaN have been identified in S. cerevisiae. The FKS1 and FKS2 genes encode similar subunits of 1,3-␤-D-glucan synthases. A fks1 mutation is synthetically lethal when combined with CaN mutations because the expression of FKS2 is calcineurin-dependent and fks1 fks2 cells are nonviable (30). Since defective cell wall synthesis may cause abnormal morphology, cell aggregation, and growth arrest (31), we investigated whether CaN activation was distorting cell shape and preventing cell separation through altered expression of FKS2. A DNA fragment internal to FKS2 was amplified by PCR and used to construct fks2::HIS3 disruptants. Cells with a disrupted fks2 gene still showed elongated shapes after transformation with the plasmid pCAtrB (data not shown). Furthermore, cells were similarly stained with Calcofluor White (a stain for cell wall chitin) regardless of whether they expressed CNAtr1/CNB or not (Fig. 6). These results indicated that altered expression of the FKS2 protein or the chitin content was not causing the elongated morphology of CNAtr1/CNB cells. The fks2 gene disruption did not suppress the LiCl tolerance conferred by the plasmid pCAtrB either (data not shown).
The ability of cells expressing the recombinant CNAtr1/CNB to form arrays of elongated cells suggested an unipolar budding pattern, a cell division program specific to the pseudohyphal growth of S. cerevisiae (32). Haploid yeast cells divide with an axial pattern, in which daughter cells form new buds adjacent to their own birth pole (Fig. 6). Diploid cells follow a dipolar pattern, where the first bud of a mother cell often emerges opposite to its birth pole and subsequent buds emerge at either pole. In the unipolar pattern, all buds emerge opposite to the birth pole of the mother cell. Calcofluor White was used to stain the chitin rings of cells that remained attached to each other (to assess clonal relationship) and the bud scars marking previous cell divisions. Haploid cells of the DBY746 background displayed a canonical axial budding pattern in 92,2% of the cells (n ϭ 192). However, elongated cells expressing the activated calcineurin switched to an unipolar budding pattern (Figs. 5 and 6). Thus, 77.9% of elongated cells budded unipolarly, 18% of cells had bud scars at both poles, and only 4.1% of elongated cells showed an axial pattern (n ϭ 244). This developmental switch was dependent on the transition to the elongated cell morphology because those transformed cells that remained normally shaped divided with an axial pattern, indicating that both phenomena (elongated cell shape and unipolar budding) were linked. Additional evidence for the involvement of CaN on bud site selection came from the analysis of budding pattern in cnb1 mutants. Thus, while DBY746 haploid cells budded almost exclusively with an axial pattern (see above), isogenic cnb1 cells showed an undefined pattern with buds emerging randomly in 79.5% of mother cells (2.7% of cells budded axially and 17.8% budded polarly; n ϭ 219) (Fig. 6).
Since cells expressing CNAtr1/CNB displayed both an elongated cell shape and a putative unipolar cell division pattern that resulted in a pseudophyphae-like phenotype, we investi- gated whether overexpression of CaN was promoting pseudophyphal growth. The standard laboratory strain W303 has been described to undergo a dimorphic transition to a pseudohyphal growth (33). Expression of activated calcineurin in W303-1A cells also promoted the accumulation of elongated cells, with 74.1% of cells having a length/width axial ratio higher than 1.4 (n ϭ 301). Isogenic MATa and MAT␣ haploids and MATa/MAT␣ diploid cells of the W303 background displayed cell elongation with the same intensity after transformation with pCAtrB (data not shown). Haploid W303-1A and homozygous W303 diploid cells expressing activated CaN were cultured in SLAD medium for nitrogen starvation, a condition that induces the transition to a filamentous pseudohyphal growth in diploid strains (32). Nitrogen starvation in either liquid or solid SLAD media failed to show synergism with the expression of CNAtr1/CNB with regard to cell morphology when compared to parallel cultures in nitrogen-rich YNB medium. Furthermore, neither haploid nor diploid CNAtr1/CNB cells invaded the agar after culturing in SLAD plates for up to 4 weeks, a distinctive property of pseudohyphae known as foraging (32). Pseudohyphae of S. cerevisiae also show a distinctive cell cycle with a shortened G 1 phase and a population dominated by cells in the G 2 phase, because DNA replication takes place but mitosis is delayed until the daughter cell grows to the size of the mother cell (34). However, DAPI staining of nuclear masses in elongated CNAtr1/CNB cells indicated that nuclear division occurred before the daughter cells reached the same size than the mother cells (data not shown), and flow cytometry analysis of exponentially growing haploid W303 cells demonstrated that CNAtr1/CNB cells had a lengthened G 1 phase rather than the G 2 phase (data not shown).

DISCUSSION
Work on Jurkat T-cells showed that cells transfected with a truncated CNA subunit lacking the CAM-binding and autoinhibitory domains did not require a Ca 2ϩ signal to activate the transcription of the interleukin-2 gene (5), leading to the suggestion that truncation of the catalytic subunit results in constitutive CaN activity in vivo. However, expression of a similarly truncated CNAtr1 subunit in yeast cells had no significant effect on Li ϩ tolerance or cell morphology, unless accompanied by the co-expression of the regulatory B subunit. These results, together with the inability of CNAtr1 to complement a cnb1 mutant, demonstrate that the binding of the regulatory subunit is essential for the activity of the truncated catalytic subunit in vivo. The difference between the mammalian and the yeast system may arise from the availability of endogenous regulatory subunits for binding and activation of the ectopically expressed catalytic subunit. Thus, CNBp might be limiting in yeast cells, whereas there is evidence that the endogenous B subunit is in excess of the catalytic A subunit in Jurkat cells (35). Remarkably, co-expression of a full-length CNA2p along with the regulatory CNBp subunit had little effect on the Li ϩ tolerance, growth rate, or morphology of yeast cells, suggesting that other factor(s) might be limiting the activity of wild-type CaN. Obviously, one such factor might be calmodulin which is no longer required for the activity of the CNAtr1/CNB complex. However, the failure of native CaN to induce tolerance is not likely the result of insufficient CAM since CaN is one of the proteins with highest affinity for CAM and CAM is most probably in excess of CaN in yeast. Cunningham and Fink (10) have shown that elevated cytosolic Ca 2ϩ in pmc1 and pmr1 mutants lacking Ca 2ϩ -ATPases inhibits yeast growth and distorts cell shape through permanent activation of CaN. Here, we have shown abnormal cell morphology through the concurrent overexpression of CNAtr1 and CNB subunits in cells with otherwise normal Ca 2ϩ homeostasis. The finding that similar consequences can be attained either by an increase in cytosolic Ca 2ϩ in cells with wild-type CaN, or by the expression of a recombinant CNAtr1/CNB calcineurin, suggests that the truncation of CaN sets a lower threshold for the Ca 2ϩ -dependent activation of CaN. This conclusion is consistent with the recent demonstration that deletion of the regulatory region of the A subunit reduced the Ca 2ϩ -dependence of CaN (3). Presumably, this lowered Ca 2ϩ dependence arises because the B subunit of proteolytically truncated CaN (analogous to the recombinant CNAtr1/CNB) has a higher affinity for Ca 2ϩ than the native enzyme, allowing activation at Ca 2ϩ levels (0.3 M) below those found in stimulated cells (0.5-1 M) (2, 3). Moreover, it has been recently proposed that CaN represses Ca 2ϩ sequestration into an intracellular compartment, presumably to amplify Ca 2ϩ signaling (11). Thus, the expression of a truncated CaN with a reduced Ca 2ϩ requirement for activity might further increase the cytosolic concentration of free Ca 2ϩ , thereby exacerbating the activation of CaN and other Ca 2ϩ -dependent signaling proteins.
We have previously demonstrated that CaN activity allowed homeostasis of Na ϩ and Li ϩ through the coordinate regulation of the efflux and influx transporters of these monovalent cations (8). Tanida et al. (11) have proposed that CaN could amplify the Ca 2ϩ signals in yeast cells by repressing the sequestration of Ca 2ϩ within an as yet unidentified internal store. A Ca 2ϩ spike in response to NaCl stress has been described in plant cells but not yet in yeast (36). However, since the Ca 2ϩdependent activity of CaN is required for tolerance to NaCl, a similar Ca 2ϩ signaling can be implied in yeast. Thus, CaN might be required only to modulate the intensity of the Ca 2ϩ signaling activated by Na ϩ stress. However, it is unlikely that an increased level of cytosolic Ca 2ϩ per se would suffice to trigger specifically the Na ϩ stress response in the absence of the inducer (i.e. Na ϩ ). Our data support a more pivotal role of CaN in the cellular response of Na ϩ stress. First, we have shown that constitutive activation of CaN not only enhances the expression of ENA1 after induction by NaCl but also allows the expression of ENA1 in the absence of inducer (Fig. 2). Second, CaN activity is essential for the transition of the K ϩ uptake system to the high affinity mode of K ϩ transport under Na ϩ stress (8), and we show now that the activation of CaN was able by itself, without induction by Na ϩ stress, to switch the K ϩ transport system to a state with higher discrimination of K ϩ over Li ϩ (and Na ϩ ) (Fig. 4). Such coordinated and Na ϩ stressspecific response of two different ion transporters probably requires the integration of different input signals for the activation of specific pathway(s). Since the activation of CaN alone can substitute for such integration, CaN is probably an integral component of the pathway(s) that regulates ion homeostasis in yeast. In other words, activation of CaN is both a necessary and sufficient signal for the cellular response to Na ϩ stress.
An elongated cell shape was found concurrent with a switch to a polar budding pattern in cells expressing the recombinant CNAtr1/CNB calcineurin. Often, elongated cells remained attached after polar budding, thereby producing chains of elongated cells that resembled pseudohyphae. The overexpression or constitutive activation of protein phosphatases 2A (PP2A) or PAM1, a protein of unknown function that suppresses the growth of PP2A-deficient cells, produced chains of elongated cells very similar to those shown in Fig. 5 (37-40). Interestingly, PP2A modulates morphogenesis during pseudophyphal differentiation in response to nitrogen starvation in S. cerevisiae (33). However, we failed to find synergism between CaN activation and nitrogen starvation in promoting a full dimorphic transition to pseudohyphal growth, or between the over-expression of truncated CaN and PP2A or PAM1. 2 Further, the timing of nuclear division and the length of G 1 and G 2 phases did not match those of pseudohyphae. However, the set of distinctive phenotypic characteristics of pseudohyphae are likely to be controlled independently. For instance, haploid elm mutants showed delayed cell separation and the ability to forage within the agar but not the unipolar budding pattern; instead, these strains displayed the axial budding pattern typical of wild-type haploids (33). Conversely, overexpression of PHD1, a putative transcriptional regulator, enhanced pseudohyphal growth in diploids but not in haploids despite the fact that it promoted the appearance of elongated haploid cells (41). Therefore, the role of CaN in the dimorphic transition of S. cerevisiae, if any at all, may be limited to the control of cell shape and budding pattern.
A defined pattern of budding requires spatial information and the establishment of cell polarity. Our data indicate that CaN may affect the acquisition of cell polarity and regulate bud site selection in S. cerevisiae. First, constitutive CaN activation induced a developmental switch from the axial budding pattern typical of haploid cells to a polar pattern in which elongated cells formed new buds preferentially at the pole opposite to the birth pole of the mother cell. Second, inactivation of CaN by disruption of the CNB1 gene randomized the budding of DBY746 haploid cells. However, since not all cells expressing the recombinant CNAtr1/CNB calcineurin were elongated, and CaN deficiency did not randomize the budding pattern in all genetic backgrounds examined, the interaction of CaN with cell polarity and budding pattern seems indirect. Mutations in proteins determining general bud site selection or polarity establishment result in random budding (see Ref. 42, and references therein). Among the proteins required for cell polarity is the putative Ca 2ϩ -binding protein CDC24. Certain cdc24 mutations cause a random budding and sensitivity to high levels of Ca 2ϩ , and the interaction between CDC24 and BEM1 (a budsite assembly protein) is affected by Ca 2ϩ (42,43). Moreover, CAM concentrates at the site of bud formation before bud emergence, suggesting that CAM-regulated proteins may participate in polarized cell growth (20). Whether CaN has a specific role in the control of cell morphology and budding pattern or the changes observed correlating with altered CaN activity are the result of perturbed Ca 2ϩ homeostasis remains to be determined.