Impaired Manganese Metabolism Causes Mitotic Misregulation*

Background: The P-type ATPase Pmr1 provides a major route for cellular detoxification of manganese. Results: Disregulation of Mn2+ homeostasis impairs genome replication and cell cycle progression. Conclusion: Genome instability and endomitosis can be triggered by alterations in cytosolic or Golgi Mn2+ levels. Significance: The Mn2+-dependent cell cycle defects might explain disease phenotypes observed in Hailey-Hailey patients having mutations in the human PMR1 orthologue ATP2C1. Manganese is an essential trace element, whose intracellular levels need to be carefully regulated. Mn2+ acts as a cofactor for many enzymes and excess of Mn2+ is toxic. Alterations in Mn2+ homeostasis affect metabolic functions and mutations in the human Mn2+/Ca2+ transporter ATP2C1 have been linked to Hailey-Hailey disease. By deletion of the yeast orthologue PMR1 we have studied the impact of Mn2+ on cell cycle progression and show that an excess of cytosolic Mn2+ alters S-phase transit, induces transcriptional up-regulation of cell cycle regulators, bypasses the need for S-phase cell cycle checkpoints and predisposes to genomic instability. On the other hand, we find that depletion of the Golgi Mn2+ pool requires a functional morphology checkpoint to avoid the formation of polyploid cells.

Manganese is an essential trace element, whose intracellular levels need to be carefully regulated. Mn 2؉ acts as a cofactor for many enzymes and excess of Mn 2؉ is toxic. Alterations in Mn 2؉ homeostasis affect metabolic functions and mutations in the human Mn 2؉ /Ca 2؉ transporter ATP2C1 have been linked to Hailey-Hailey disease. By deletion of the yeast orthologue PMR1 we have studied the impact of Mn 2؉ on cell cycle progression and show that an excess of cytosolic Mn 2؉ alters S-phase transit, induces transcriptional up-regulation of cell cycle regulators, bypasses the need for S-phase cell cycle checkpoints and predisposes to genomic instability. On the other hand, we find that depletion of the Golgi Mn 2؉ pool requires a functional morphology checkpoint to avoid the formation of polyploid cells.
Correct ion homeostasis is essential for the control of biochemical processes in eukaryotic cells. This is the case of the trace element Mn 2ϩ that is required as a cofactor for a wide range of enzymes located in every cellular compartment (1,2). In addition to serving as an essential enzymatic co-factor, Mn 2ϩ can also be toxic and heavy exposure to this ion has been shown to result in a central nervous system disorder resembling parkinsonism, more specifically known as manganism (3,4).
The Saccharomyces cerevisiae gene PMR1 encodes a Golgilocalized ATPase that transports Ca 2ϩ and Mn 2ϩ ions from the cytosol into the Golgi lumen (5)(6)(7). PMR1 is evolutionarily conserved from yeast to humans and disruption of one allele of the PMR1 ortholog ATP2C1 leads to Hailey-Hailey disease in humans (8). Single point mutations that change the ion selec-tivity of Pmr1 have been described (9,10), making it possible to distinguish between Ca 2ϩ -and Mn 2ϩ -specific phenotypes. While Ca 2ϩ is required for protein sorting, Mn 2ϩ serves as an essential co-factor for N-and O-linked protein glycosylation in the secretory pathway (5,11). Other pmr1⌬-dependent phenotypes are based on the accumulation of Mn 2ϩ in the cytosol that alter reverse transcriptase and telomerase activities (12,13), activate target of rapamycin (TOR) 3 signaling (14) or permit the scavenging of oxidative radicals (7).
In S. cerevisiae, the budding cycle is tightly coupled to the central events of the cell cycle to ensure that genetic information is correctly transferred from mother to daughter. The cyclin-dependent kinase (CDK) Cdc28 is the key component of the mechanism that controls the timing of different cell cycle events. Cdc28 is activated by the alternate association of different cyclins, whose expression is cell cycle regulated, and thereby permits the phosphorylation of different substrates at different times (15). G1-phase cyclins (Cln1-3) are involved in the G1 to S-phase transition promoting bud emergence, spindle pole body duplication and activation of B-type cyclin expression. The early expressed B-type cyclins Clb5 and Clb6 are required for the initiation of DNA replication and progression through S-phase. M-phase cyclins (Clb1-4) are required for spindle formation and the initiation of mitosis. M-phase cyclins also prevent mitotic exit and cytokinesis, and therefore, their activity must be eliminated to allow polarized protein secretion to the bud neck and cell division to take place (16). In addition to being regulated by the association with different cyclins, Cdc28 can also be regulated by post-translational modifications mediated by the morphology checkpoint protein Swe1 (Wee1 in Schizosaccharomyces pombe). Swe1 inhibits the kinase activity of Cdc28 complexed with certain B type cyclins including Clb2 and, to a lesser extent, Clb3 and Clb4 by phosphorylation of tyrosine 19 (Y19) causing a delay in G2/M transition (17,18). The essential coordination between cell cycle progression and the budding cycle is ultimately maintained by the presence of a checkpoint that blocks Swe1 degradation, thus delaying mitosis in response to defects in growth or bud formation. This checkpoint appears to be triggered by abnormalities in the actin cytoskeleton (19,20) as well as cell size (21).
Evidence of a link between Mn 2ϩ , DNA replication, and cell cycle progression came from different observations. Mutants coding for temperature-sensitive (ts) alleles of the Mn 2ϩ -dependent, putative ER-localized lipid phosphatase Cdc1, exhibit a small-bud growth arrest that is identical to the phenotype displayed by Mn 2ϩ -depleted cells (22)(23)(24). On the other hand, PMR1 mutants are sensitive to the replication inhibitor hydroxyurea (HU) (25) and exhibit negative genetic interactions with mutants impaired in replication initiation and progression (26,27). We explored the molecular impact of altered Mn 2ϩ homeostasis and dissected the influence of cytoplasm Mn 2ϩ overload versus depletion of Golgi-hosted Mn 2ϩ . Our results indicate that excess of cytosolic Mn 2ϩ challenges the replication machinery and is responsible for impaired DNA synthesis, transcriptional up-regulation of G1/S cyclins, bypass of Rad53/Mec1 checkpoint functions and increased DNA damage. In contrast, Golgi-linked cell polarity defects appear to activate the Swe1-dependent morphology checkpoint to prevent polyploidy. Our results clearly demonstrate that Pmr1 is needed to buffer alterations in Mn 2ϩ levels, which would otherwise cause loss of cell cycle control, genetic instability, and multinucleation, primary events in tumor formation in higher eukaryotic cells.

EXPERIMENTAL PROCEDURES
Yeast Strain, Plasmids, and Growth Conditions-Yeast strains used in this study are listed in supplemental Table S1. Gene deletions were constructed by PCR-based methods using pFA6a-kanMX6 (28), pAG32 (EUROSCARF), and pFA6a-klLEU2MX6 (kindly provided by B. Pardo) as template plasmids. Regulated P GAL expression of Pmr1 was constructed by PCR-based methods using template plasmid pFA6a-kanMX6-PGAL1 (28). Point mutants pmr1-Q783A and pmr1-D53A were constructed using an in vivo site-directed mutagenesis as described previously (29). VCX1-M1 was expressed from highcopy yeast-expression plasmid p2UGpd (gift of K. Hirschi), containing the strong constitutive GPD promoter. ATX2 was cloned into the BamHI and SacI sites of p2UGpd. Standard yeast growth conditions and genetic manipulations were used. All experiments were carried out in the corresponding Synthethic Complete (SC) medium, except for microarray experiments and the drop test shown in supplemental Fig. S1C (left) that were carried out in YPAD medium.
Cell Cycle Analysis-Samples for flow cytometry were prepared following standard procedures and analyzed on a FACScalibur (Becton Dickinson). For G1 synchronization, cells were incubated for 1 h in 1 g/ml ␣-factor (RP-4582, Biomedical S.L.) and supplemented with the same amount of fresh ␣-factor for another hour. Cells were released from ␣-factor treatment by washing three times in pre-warmed, fresh media.
Analysis of Replication Intermediates-Cells were arrested with sodium azide (0.1% final concentration) and cooled down on ice. Total DNA was isolated, digested with restriction enzymes, and resolved by two-dimensional gel electrophoresis as described previously (30), then transferred onto a Hybond-XL membrane and subsequently hybridized with specific 32 P-labeled probes. Signals were quantified using a PhosphorImager Fujifilm FLA-5100 and the ImageGauge program. The relative intensity of replication intermediates was normalized to the signal intensity obtained in the 1n-spot (non-saturating exposure).
Microarray Analysis-Gene expression profiles were determined by using the 3Ј-expression microarray technology by Affymetrix platform at the Genomics Unit of CABIMER (Seville, Spain). Total RNA from yeast cells grown on YPD at 30°C to mid-log-phase was isolated using the RNeasy Midi kit (Qiagen). Synthesis, labeling, and hybridization of cRNA to GeneChip Yeast Genome 2.0 Arrays was performed with RNA from 3 independent cultures of each strain. The resulting data were reprocessed using the Robust Multi-array Average (RMA) method. The following statistical data analyses were performed using the limma package (affylmGUI interface) of the R Bioconductor Project. More than 1.8-fold changes with 95% confidence levels (FDR-adjusted p valuesϽ0.05) were considered to be significant. The microarray data from this publication have been submitted to the GEO database and assigned the identifier GSE29420.
A-Like Faker (ALF) Assay-The formation of a-mating cells from MAT␣ strains was scored as described (33) with some modifications. Briefly, MAT␣ strains were grown on YPAD plates for 3 days to obtain single colonies. Each ALF frequency value was obtained by the average of at least two fluctuation tests of four independent colonies each. A-like faker cells were selected by growing on YPAD plates overnight at 30°C. Cells were transferred onto a mating tester lawn of MAT␣ by replica plating followed by incubation at 30°overnight. The mated lawn was then replica plated to Synthetic Dextrose (SD) medium, and colonies were counted. Total cells were grown on YPAD plates.
Microscopy-Rad52-YFP foci levels were analyzed in midlog-phase cells bearing plasmid pWJ1213 (R. Rothstein). Approximately 300 cells derived from three independent transformation experiments were analyzed for each strain. Metaand anaphase entry were assessed by spindle (tubulin staining) and nuclear morphology (DAPI) as described (34). For visualization of chromosome IV tagged with GFP, cells were grown at an A 600 ϭ 0.2 and fixed as described (35). Approximately 200 cells from at least three independent experiments were analyzed for each strain. Microscope images were obtained at 25°C by projection of a series of focal plane images derived from wide-field fluorescence microscopy (DM-6000B, Leica) at 100ϫ magnification using A4 and L5 filters and a digital chargecoupled device camera (DFC350, Leica). Pictures were processed with LAS AF (Leica). For bud emergence analysis, cells were synchronized with ␣-factor and released in fresh medium for 90 min. Samples were taken every 15 min, fixed (3,7% formaldehyde), and observed under a light microscope. For each time point, 200 cells were counted and the fraction of cells with no bud, a small bud (smaller than one half of the yeast cell) or a large bud (equal or larger than one half of the yeast cell) was documented. Chitin deposition was visualized using calcofluor white (F3543, Sigma) as described previously (36). Filamentous actin was visualized using rhodamine-phalloidin (P2141, Sigma) as described (37). For septin ring visualization, cells were transformed with the plasmid pLP17 (38) expressing Cdc12 tagged with GFP. At least 200 cells derived from two independent transformation experiments were analyzed for each strain.

Excess of Cytosolic Mn 2ϩ Impairs DNA Synthesis and S-phase
Progression-Pmr1 is a key regulator of the intracellular Mn 2ϩ levels, thus offering the possibility to investigate the impact of impaired Mn 2ϩ homeostasis on cell cycle progression. First, we determined the fate of S-phase progression by FACS analysis in pmr1⌬. Prior to S-phase release, cells were synchronized in G1/S phase with ␣-factor and HU. Note that HU provides complete activation of early replication origins but limits fork progression (39). Interestingly, pmr1⌬ cells were slightly delayed in S-phase progression and remained in G2/M-phase longer than wild type (WT) cells ( Fig. 1, A, ϪMnCl 2 ). Addition of extracellular MnCl 2 ( Fig. 1, A, ϩMnCl 2 ) exacerbated this delay in pmr1⌬ cells (S-phase peaked after 80 min) but did not affect WT cells (S-phase peaked after 40 min), whereas the addition of extracellular CaCl 2 did not affect S-phase progression (supplemental Fig. S1A) indicating a Mn 2ϩ -dependent delay. To further confirm the specific role of Mn 2ϩ in S-phase defects we generated two previously described separation-of-function mutants, pmr1-Q783A and pmr1-D53A, which transport almost exclusively either Ca 2ϩ or Mn 2ϩ , respectively (9, 10). Only Mn 2ϩ -transport-deficient mutant (pmr1-Q783A) showed a delay in S-phase progression in the presence of MnCl 2 (supplemental Fig. S1B). It is important to note that pmr1-Q783A is not a full loss-of-function mutant since it is not as sensitive to MnCl 2 as the pmr1⌬ mutant (supplemental Fig. S1C). To rule out that impaired Mn 2ϩ supply of the secretory pathway is the cause of S-phase defects, we transformed pmr1⌬ cells with a plasmid containing a mutant allele of the vacuolar membrane Ca 2ϩ /H ϩ exchanger VCX1 (VCX1-M1). Vcx1-M1 has enhanced Mn 2ϩ /H ϩ exchange and was previously found to suppress the Mn 2ϩ toxicity phenotype of pmr1⌬ (40)(supplemental Fig. S1D) and, indeed, high-dosage expression of VCX1-M1 in pmr1⌬ partially suppressed S-phase transition defects caused by MnCl 2 addition (Fig. 1B). Finally, by overexpression of the Golgi membrane protein coding ATX2 gene, we further proved that an increase in cytosolic Mn 2ϩ would lead to an S-phase delay (Fig. 1C). Atx2 works in opposite directions to Pmr1 to control manganese homeostasis and ATX2 overex-pression has been shown to resemble the Mn 2ϩ sensitivity phenotype of pmr1⌬ (41) (supplemental Fig. S1E). Importantly, in the presence of 3 mM MnCl 2 , ATX2 overexpression in WT cells reconstituted the same S-phase delay pattern as observed in pmr1⌬ cells. In previous studies, cytosolic Mn 2ϩ levels were determined to be 5-20-fold higher in pmr1 mutants (7,42), and therefore we assessed whether the extracellular addition of high MnCl 2 concentrations to WT cells would mimic the S-phase transition delay observed in pmr1⌬ mutants (supplemental Fig.  S1F). Addition of 10 mM MnCl 2 caused a profound delay in S-phase transit and a further increase to 25 mM MnCl 2 even prevented WT cells to enter into S-phase. Taken together, these results strongly suggest that cytosolic Mn 2ϩ overload delays S-phase transit.
To test whether the delayed S-phase transit in pmr1⌬ mutants was due to impaired origin firing and/or slow DNA synthesis, we analyzed the fate of replication intermediates (RIs) at the molecular level by two-dimensional gel analysis (Fig. 1D). To do so, we compared the molecular pattern and appearance of RIs from origin ARS305 (probe A) and progressing along chromosome III (close to ARS305, probe B; further downstream of ARS305, probe C) (43). Upon release from HU arrest, within 20 min the amount of RIs at and close to ARS305 dropped 2-3-fold in the WT and replication was completed within 60 min (Fig. 1D, probes A and B). However, consistent with slower replication in pmr1⌬ cells, the amount of RIs remained constant within 40 min and replication was completed within 80 min. No obvious differences were visible comparing the two-dimensional pattern of RIs (bubble and simple-Y shaped structures) nor did we detect the appearance of unusual RIs structures in pmr1⌬ cells. From these results we conclude that the increase in intracellular Mn 2ϩ contributes to slower replication fork (RF) progression in pmr1⌬ cells. In addition to defects in fork progression, we noticed that the relative amount of RIs at the onset of replication appeared to be lower in pmr1⌬ cells (see Fig. 1D, probe A). This finding prompted us to analyze the timing of replication initiation in pmr1⌬ cells (supplemental Fig. S1G). Upon ␣-factor release into low levels of HU to limit fork progression, samples were taken every 4 min and subjected to two-dimensional gel analysis. While in WT cells the first RIs in ARS305 were present within 8 min, in pmr1⌬ cells RIs were visible about 12 min after S-phase entry. Thus, replication initiation and fork progression are affected in pmr1⌬ cells.
Late Origin Activation Is Required to Prevent Genome Instability under Condition of Increased Cytosolic Mn 2ϩ -To determine if the fork progression delay has an impact on the onset of late origin firing activation, we compared the replication timing between the early firing origin ARS607 and the late firing origin ARS603 ( Fig. 2A), both located in chromosome VI. As for ARS305, ARS607 showed a defect in replication initiation, as seen by the lower amount of RIs after HU release, and a slowdown of RF progression in pmr1⌬ cells suggesting that this is a general feature. Interestingly RF progression and the temporal activation of late origin ARS603 firing were delayed by 20 min in pmr1⌬ mutants. Importantly, pmr1⌬ cells are capable of initiating origin-firing at ARS603, indicating that the pmr1⌬-de-pendent S-phase delay is not mediated by the inhibition of late origin activation.
CLB5 encodes a B-type cyclin that activates Cdc28p to promote initiation of DNA synthesis and clb5⌬ cells have been shown to present a prolonged S-phase as a result of failure to activate late origins (44,45). We reasoned that if pmr1⌬ cells were not affected in late origin activation, pmr1⌬ clb5⌬ mutant might show a synergistic effect in S-phase progression. For this purpose, we generated pmr1⌬ clb5⌬ double mutants and followed cell cycle progression by FACS analysis (Fig. 2B). As expected, while pmr1⌬ and clb5⌬ single mutants were slightly delayed in S-phase progression, the double mutant showed an additive delay (S-phase peaked after 80 min).  ). An empty vector (p2UGpd) was used as control. MnCl 2 (1 mM) was added after release from ␣F into fresh medium. C, FACS analysis of WT cells (BY4741) transformed with an ATX2 overexpressing plasmid leading to a cytosolic Mn 2ϩ increase. An empty vector (p2UGpd) was used as control. MnCl 2 (3 mM) was added after release from ␣F. D, two-dimensional gel analysis of RF progression along chromosome III in WT (BY4741) and pmr1⌬ (NGY051) cells. Cells were synchronized at G1/S with ␣F and HU (200 mM) prior to S-phase release. DNA samples were digested with EcoRV (E) and HindIII (H) as indicated (top). Replication is initiated at ARS305 (bubble-shaped molecules; probe A) and passive fork progression is detected toward the left arm of chromosome III (simple-Y shaped molecules; probes B and C). Note that origin ARS304 (gray) remains dormant. Quantification of total RIs is shown to the right.
Next, we asked if replication forks might be more prone to DNA damage due to delayed replication fork progression. To assess this possibility, we measured replicative damage by monitoring the formation of DNA repair centers that are dependent on the homologous recombination protein Rad52 (46) (Fig. 2C,  left). An increase of Rad52-YFP foci was evident in pmr1⌬ and clb5⌬ simple mutants. However, in the double mutant, the formation of DNA repair centers was boosted with more than 60% of the cells containing Rad52-YFP foci. To assess whether DNA damage in pmr1⌬ clb5⌬ mutants is mediated by cytosolic Mn 2ϩ overload, this mutant was transformed with plasmid pVCX1-M1 to decrease the cytosolic Mn 2ϩ levels. Interestingly, we observed a statistically significant reduction in the formation of Rad52-YFP foci in cells overexpressing VCX1-M1 (Fig. 2C, right), indicating that cytosolic Mn 2ϩ overload is indeed promoting genetic instability. To preserve RF integrity, intra-S checkpoint activation leads to the phosphorylation of the checkpoint sensor Rad53 (43). Interestingly, we found that Rad53 phosphorylation was reduced in HU-treated clb5⌬ mutants, possibly due to reduced number of replicons per cell (Fig. 2D). Thus, the combination of having less replicons and slow replication forks might account for the extreme delay in S-phase progression and the massive formation of DNA repair centers in pmr1⌬ clb5⌬ mutants.

Deletion of PMR1 Leads to Transcriptional Up-regulation of Factors Involved in Cell Cycle and Polarity-Eukaryotic cell
cycle is carefully regulated to ensure that genetic material is duplicated precisely before cell division takes place. In S. cerevisiae, a single Cdk, Cdc28 associates with multiple cyclins to regulate the cell cycle. To investigate whether defects in S-phase progression might lead to an alteration in the cellular gene expression profile in an attempt to readjust the cell cycle, we analyzed total mRNA levels in pmr1⌬ mutants by microarray analysis (Fig. 3A, accession number GSE29420 at GEO database). In pmr1⌬ cells, a significant number of genes involved in sugar (22 of 241) and iron metabolism (5 of 241) were downregulated, while genes associated with the cell cycle and stress response (30 of 129), cell polarity (18 of 129), and the regulation of kinase activity (13 of 129) were up-regulated. Importantly the transcriptional up-regulation of Cdc28 as well as of cyclins involved in the G1/S transition of the cell cycle such as CLB5/6, CLN1/2, and PCL1/2 appear to be relevant, because this finding is in concordance with the previously reported loss of viability of pmr1⌬ cln1⌬ cln2⌬ triple mutant (47). Remarkably, one of the up-regulated genes was SRL3 (suppressor of rad53 Lethality) (48), a potential substrate of the Cdc28 kinase. To verify SRL3 up-regulation, we tested by tetrad analysis (Fig. 3B) if the PMR1 deletion could suppress rad53⌬ and mec1⌬ lethality, mutants known to be suppressed by SRL3 overexpression (48). As this appeared to be true, we further characterized the lack of Rad53-dependent Sml1 protein degradation in response to MMS-induced DNA damage by Western blot analysis of pmr1⌬ rad53⌬ mutants (Fig. 3C). An assay based on counterselection of a plasmid harboring RAD53 revealed that only cells lacking the Mn 2ϩ -(pmr1-Q783A) but not Ca 2ϩ -pump function (pmr1-D53A) suppressed rad53⌬ lethality (Fig. 3D). In agreement with this observation, lowering cytosolic Mn 2ϩ by transport to the vacuolar space by Vcx1-M1 restored lethality of pmr1⌬ rad53⌬ cells (Fig. 3E). Finally, we determined that pmr1⌬ rad53⌬ srl3⌬ triple mutants were unviable. 4 It is thus conceivable that the bypass of G1/S checkpoint functions occurs in response to G1/S phase transition constraints mediated by cytosolic Mn 2ϩ overload.
Golgi Mn 2ϩ Pool Depletion Activates the G2/M Morphology Checkpoint to Avoid Polyploidy-As previously mentioned, FACS analysis of S-phase progression in pmr1⌬ cells revealed a substantial delay in G2/M phase entry or exit (see Fig. 1A). We suspected that this delay might be related to the observed transcriptional up-regulation of Swe1 (see Fig. 3A), a protein kinase involved in regulating G2/M transition (17). To test if this was the case, we generated pmr1⌬ swe1⌬ double mutants and followed cell cycle progression by FACS analysis (Fig. 4A). Interestingly, the absence of Swe1 liberated pmr1⌬ cells from the G2/M delay (see Fig. 4A, after 120 min). To further confirm this observation, we placed the expression of PMR1 under the control of the repressible GAL1 promoter and followed progression though meta-and anaphase (Fig. 4B). Upon repression of PMR1 transcription, we observed a delay in meta-and anaphase progression (Fig. 4B, red circles), which did not occur in a swe1⌬ background (Fig. 4B, blue triangles) confirming that Swe1 mediates the G2/M delay.
The bypass of the G2/M delay in pmr1⌬ might lead to uncontrolled chromosome segregation resulting in loss of heterozygosity (LOH). To assess this possibility, we measured LOH frequency by the formation of a-like faker cells (33) (ALF, Fig. 4C). Unexpectedly, while pmr1⌬ and swe1⌬ single mutants did not show a statistically significant change relative to the WT, LOH was suppressed in pmr1⌬ swe1⌬ mutants. Previous studies have concluded that Swe1 does not play a role during the cell cycle in normal, unperturbed conditions (17). In contrast, Swe1 is part of the morphogenesis checkpoint and, in response to perturbations that prevent bud formation, inhibits mitotic progression through negative regulation of Clb/Cdk (49). Overexpression of the mitotic cyclin Clb2 has been shown to result in the accumulation of multinucleated cells in swe1⌬ background (50). Multinucleation has also been observed in pmr1⌬ as well as swe1⌬ during meiosis (51,52). Therefore, the presence of more than one copy of chromosome III (chr. III) could account for LOH suppression. We examined the nuclear morphology of pmr1⌬ swe1⌬ mutants (Fig. 5A). Microscopical analysis revealed that about 41% of the cells contained multiple nuclei (DAPI staining) and multiple copies of chr. IV (GFP-tagged) and that the cellular diameter exceeded 8 m in at least 35% of the total cell population. Importantly, we noticed the formation of giant cells, which were prone to undergo spontaneous lysis. A very gentle handling allowed us to confirm the presence of multinucleated cells by FACS (Fig. 5A, right). Multinucleated cells were evident in pmr1-Q783A but not pmr1-D53A swe1⌬ mutants (supplemental Fig. S2), suggesting that multinucleation was mediated by an Mn 2ϩ but not Ca 2ϩ transport defect. Moreover, lowering cytosolic Mn 2ϩ levels by transport into the vacuole using the mutant allele VCX1-M1 did not restore normal cell size nor suppress multinucleation, instead these phe-notypes appeared to be exaggerated, as seen by a statistically significant increase in cell diameter (Fig. 5B). Together these data demonstrate that Golgi Mn 2ϩ depletion leads to polyploidy in the absence of an active morphology checkpoint.
Mn 2ϩ Depletion Causes Cell Polarity Defects Linked to Protein Glycosylation-Mn 2ϩ is needed as a co-factor for the proper action of Golgi hosted mannosyltransferases (53)(54)(55) and impaired Golgi Mn 2ϩ transport has been shown to cause protein underglycosylation (11). Other studies have shown that mannosyltransferase mutants affected in protein glycosylation display a pronounced delay in bud formation (56,57). Moreover, SWE1 mRNA levels are elevated in cells that cannot make a bud (58). To test whether pmr1⌬ cells were affected in bud formation, we synchronized cells with ␣-factor at G1-phase prior to release into S-phase, finding that pmr1⌬ cells showed a delay of 15 min in bud emergence relative to the WT (Fig. 6A). We conclude from this experiment that Golgi Mn 2ϩ was required for timely bud emergence and that the morphogenesis checkpoint was essential to delay mitosis until a bud was formed.  and polarity and bypasses rad53⌬ and mec1⌬ lethality. A, microarray analysis of mRNA levels in pmr1⌬ mutants (NGY035) compared with the WT mRNA levels (BY4742) from cells grown in YPAD medium. Genes whose expression was more than 1.8-fold down or up-regulated and that were significantly enriched for a GO-term are indicated (p valueϽ0.002). B, tetrad analysis crossing pmr1⌬ (NGY003) with rad53⌬ sml1⌬ (W2105-17B) and mec1⌬ sml1⌬ (U963-61A). The genotype of the relevant spores is indicated. C, Western blot analysis of Rad53 and Sml1 proteins upon MMS treatment (0.03%) in WT (W303-1A), pmr1⌬ (NGY003), and pmr1⌬ rad53⌬ (W1887-5C). The phosphorylated form of Rad53 is indicated (*). Adh1 protein was used as a loading control. D, deficient Golgi Mn 2ϩ pump function suppresses the lethality associated with rad53 deletion. pmr1-Q783A rad53⌬ (NGY180), pmr1-D53A rad53⌬ (NGY181), and rad53⌬ (NGY182) mutants containing plasmid pJA92 (kindly provided by S. Elledge) expressing RAD53 under its own promoter were streaked onto SC-ura and SC supplemented with 500 mg/ml of 5-fluorotic acid (5-FOA). Plasmid loss was deduced by the formation of 5-FOA-resistant cells. E, lowering cytosolic Mn 2ϩ by transport to the vacuolar space confers lethality to pmr1⌬ rad53⌬ cells. Shown is the growth of pmr1⌬ rad53⌬ (W1887-5C) transformed with a control vector (p2UGpd) or a plasmid containing a mutant allele of VCX1 (pVCX1-M1) on SC-ura.
The fact that pmr1⌬ cells were delayed in bud emergence suggested that they might be defective in polarized growth during the cell cycle. This type of phenotype is often associated with alteration in chitin deposition and inability to properly organize the actin cytoskeleton. Accordingly, analysis of chitin deposition in pmr1⌬ mutant, which is normally found in the neck region of budded cells and in bud scars, revealed that chitin was delocalized and deposited at elevated levels (Fig. 6B). In the case of actin, which is normally organized with actin patches found exclusively in the buds and actin cables oriented toward the tips of the buds, pmr1⌬ showed a slight defect in the organization with some mislocalized patches (Fig. 6C). Interestingly, although chitin deposition was normal in swe1⌬ cells, a massive accumulation of chitin at the bud emergence site was evident in pmr1⌬ swe1⌬ mutants. Moreover, the actin cytoskeleton was clearly depolarized in giant pmr1⌬ swe1⌬ cells, showing actin patches randomly distributed. Impaired bud neck formation was even more evident looking at septin rings between mother and daughter cells (Fig. 6D). Septins were thickened and mislocalized in a substantial fraction of pmr1⌬ swe1⌬ cells. These results suggest that Golgi Mn 2ϩ is essential to establish or maintain cell polarity, making a functional morphogenesis checkpoint crucial to coordinate budding with G2/M entry in pmr1⌬ mutants.

DISCUSSION
In this work, we have uncovered a remarkable impact of manganese on DNA synthesis and nuclear segregation (Fig. 7, see figure legend for explanation). Cytosolic Mn 2ϩ excess could interfere with the activity of Mg 2ϩ -dependent enzymes. It is known that high Mn 2ϩ concentration can compromise e.g. the in vitro activity of the flap endonuclease Rad27 (59) as well as the fidelity of DNA polymerases (60). Mn 2ϩ has also been shown to decrease the catalytic activity of reverse transcriptases leading to reduced Ty1 retrotransposition in pmr1⌬ mutants (12) as well as to alter the fidelity of the telomerase in vitro and to induce telomere shortening in vivo (13). Moreover, cytosolic Mn 2ϩ accumulation in pmr1⌬ mutants has been shown to increase viral recombinants by affecting RNA-dependent RNA polymerases (61). Although we do not know all enzymatic activities that may be altered in the presence of Mn 2ϩ , the impaired function of proteins that take part in DNA synthesis or even DNA polymerases themselves, are likely to contribute to replication slow-down. It is tempting to speculate that cytosolic manganese overload might slow-down the nucleotide incorporation during DNA synthesis causing a reduction in the consumption of dNTPs. MEC1 and RAD53 play the essential role of maintaining an adequate nucleotide supply during G1/S transition (48) and a reduced need for dNTPs during DNA synthesis could explain the pmr1⌬-dependent suppression of rad53⌬ and mec1⌬ lethality. Despite this possibility, suppression of rad53⌬ lethality was linked to the up-regulation of SRL3. Little is known about the function of Srl3, but apparently trancriptional up-regulation of SRL3 takes part in the activation of the cell wall integrity (CWI) pathway (62). Cytosolic Mn 2ϩ overload could resemble osmotic stress conditions by which the stress-activated protein kinase (SAPK) Hog1 is activated leading to replication constrains (63). Notably, we find that some of the known Hog1 targets (e.g. Hsl1 and Swe1, (64)) were up-regulated in pmr1⌬ mutants.
We observe that early and late origin activation is clearly delayed but not impaired in pmr1⌬ mutants, an observation that is in concordance with the previously reported synthetic sick interaction of pmr1⌬ with mutants of the ORC complex (27). Moreover, lack of Pmr1 invokes the transcriptional upregulation of Clb5/6, both cyclins needed for the full activation of early and late origins (44,45). Interestingly, the additive S-phase delay and increase in DNA damage in pmr1⌬ clb5⌬ mutants points to the possibility that replisome composition and integrity is challenged by manganese. This might explain the enhanced sensitivity of pmr1⌬ mutants to DNA damaging agents such as camptothecin, MMS, and 4-NQO (65). 4 Our two-dimensional gel analysis of RIs did not reveal evidence for increased replication fork break down, nor an increase in Holliday junction formation (see Figs. 1D, 2A, and supplemental Fig. S1F). However, replication fork slow-down in pmr1⌬ cells could mediate a transient uncoupling of leading and lagging strand synthesis, thus increasing the formation of gapped DNA. Evidence for such a possibility is given by the synthetic growth defect of pmr1⌬ in combination with a lack of proteins that have a role in gap repair such as Rad18, Rad27, or the members of the Rad52 epistasis group (26), as well as the transcriptional up-regulation of the cell-cycle checkpoint serine-threonine kinase Dun1, a regulator of postreplicative DNA repair (66).
Cell cycle progression and the morphogenesis (budding) cycle are tightly coupled to ensure a successful cell proliferation. Dramatic changes in cell polarity that occur in G1 (polarization at the bud site), G2 (depolarization within the bud), and mitosis (repolarization to the bud neck) are triggered by changes in the kinase activity of Cdc28, the universal regulator of cell cycle progression (67). We present data suggesting that the coordination of cell cycle progression and morphogenesis by the morphology checkpoint Swe1 is essential to avoid polyploidy in PMR1 mutants lacking the Golgi Mn 2ϩ pump func-tion. Delayed bud emergence and alterations in chitin deposition indicate that pmr1⌬ cells are defective in polarized growth. Accordingly, G1-specific cyclin-dependent kinases Cln1/2-FIGURE 6. Pmr1 is essential to establish a proper cell polarity. A, WT (NGY124) and pmr1⌬ (NGY134) cells were synchronized at G1 with ␣-factor and released in fresh medium for 90 min. For each time point, the fraction of cells with no bud (blue circle), small bud (red square), and large bud (green triangle) was scored (n Ͼ200 per point). B, chitin and C, actin localization in WT (NGY124), pmr1⌬ (NGY134), swe1⌬ (NGY131), and pmr1⌬ swe1⌬ (NGY126) cells stained with calcofluor white and rhodamine-phalloidin, respectively. Bar, 5 m. Magnification and exposures time were the same for all strains. D, WT (W303-1A), pmr1⌬ (NGY003), swe1⌬ (NGY120-B), and pmr1⌬ swe1⌬ (NGY120-A) cells were transformed with the plasmid pLP17 expressing Cdc12 tagged with GFP and analyzed by fluorescence microscopy. For each strain, the fraction of cells with no septin ring (gray), aberrant (white), and bud-neck (black) localization was documented (n Ͼ200). Bar, 5 m.
Cdc28 and Pcl1/2-Pho85, which regulate essential events during the G1/S transition and morphogenesis events, such as polarized growth and bud emergence, appear to be up-regulated in cells lacking Pmr1, as a plausible consequence of dealing with difficulties to establish a proper cell polarity. Moreover, cells lacking Pmr1 show genetic interactions with the cell polarity GTPase, Cdc42 (68). Several reports have shown that mannosyltransferase mutants affected in N-glycosylation are defective in polarized growth and become dependent on an intact mitotic checkpoint for survival, which provides time to produce a bud (56,57). Overriding the morphogenesis checkpoint by Clb2 overexpression in these mutants resembles the multinucleation phenotype of pmr1⌬ swe1⌬ mutants (56,57). The fact that some components of the cis-Golgi mannosyltransferase complex are Mn 2ϩ -dependent enzymes suggests that the function of these enzymes is impaired if the Golgi is not supplied with Mn 2ϩ (53)(54)(55). Accordingly, PMR1 mutants are deficient in protein N-glycosylation and this phenotype can be partially suppressed by Mn 2ϩ addition (11). From these findings, it becomes apparent that Pmr1 acts upstream of cis-Golgi mannosyltransferases suggesting that their multinucleation phenotype is epistatic.
The regulation of Mn 2ϩ homeostasis has been shown to be highly conserved between yeast and higher eukaryotes (69). Mutations in the human PMR1 ortholog ATP2C1 cause Hailey-Hailey disease, an autosomal dominant blistering skin disorder (8). Deficient Mn 2ϩ homeostasis might be linked to tumor formation in some Hailey-Hailey patients (70) as well as in haploinsufficient ATP2C1 Ϫ/ϩ mice (71), given the evidence that cytosolic overload and depletion of the Golgi-hosted Mn 2ϩ increase the incidence of genetic instability and multinucleation in yeast. Thus, it will be interesting to see whether Mn 2ϩ -dependent alterations in replication and checkpoint activities indeed contribute to cancer formation in affected Hailey-Hailey patients.