Mdt1, a Novel Rad53 FHA1 Domain-Interacting Protein, Modulates DNA Damage Tolerance and G₂/M Cell Cycle Progression in Saccharomyces cerevisiae

Abstract

The Rad53 kinase plays a central role in yeast DNA damage checkpoints. Rad53 contains two FHA phosphothreonine-binding domains that are required for Rad53 activation and possibly downstream signaling. Here we show that the N-terminal Rad53 FHA1 domain interacts with the RNA recognition motif, coiled-coil, and SQ/TQ cluster domain-containing protein Mdt1 (YBl051C). The interaction of Rad53 and Mdt1 depends on the structural integrity of the FHA1 phosphothreonine-binding site as well as threonine-305 of Mdt1. Mdt1 is constitutively threonine phosphorylated and hyperphosphorylated in response to DNA damage in vivo. DNA damage-dependent Mdt1 hyperphosphorylation depends on the Mec1 and Tel1 checkpoint kinases, and Mec1 can directly phosphorylate a recombinant Mdt1 SQ/TQ domain fragment. MDT1 overexpression is synthetically lethal with a rad53 deletion, whereas mdt1 deletion partially suppresses the DNA damage hypersensitivity of checkpoint-compromised strains and generally improves DNA damage tolerance. In the absence of DNA damage, mdt1 deletion leads to delayed anaphase completion, with an elongated cell morphology reminiscent of that of G₂/M cell cycle mutants. mdt1-dependent and DNA damage-dependent cell cycle delays are not additive, suggesting that they act in the same pathway. The data indicate that Mdt1 is involved in normal G₂/M cell cycle progression and is a novel target of checkpoint-dependent cell cycle arrest pathways.

The integrity of eukaryotic genomes is monitored by remarkably conserved checkpoint signaling pathways that arrest the cell cycle in the presence of DNA damage, replication blocks, and mitotic spindle defects and activate repair processes in order to ensure precise DNA replication, chromosome segregation, and subsequent formation of viable progenitor cells. Defects in checkpoint mechanisms can lead to genomic instability and predispose multicellular organisms to cancer (53). Human genetic disorders that have an increased propensity for cancer and that are linked to mutations in checkpoint genes include ataxia telangiectasia (AT), caused by mutations in the AT-mutated (ATM) kinase (1); Nijmegen breakage syndrome, caused by mutations in the BRCT and FHA domain-containing Nbs1 protein (6); and Li-Fraumeni multicancer syndrome, caused by mutations in the tumor suppressor p53 or Chk2 kinase (4).

Studies with Saccharomyces cerevisiae have been instrumental to the understanding of molecular mechanisms of checkpoint pathways. For example, the tumor suppressor Chk2 kinase was only rather recently identified in searches for human orthologs of Rad53 (32), which plays a central role in S. cerevisiae checkpoint responses (reviewed in reference 53). Rad53 activation through phosphorylation depends on three major pathways that act in concert with the Mec1/Lcd1 complex (orthologs of human ATM/ATR and ATRIP) (7, 8, 40). Two partially overlapping Rad53 activation pathways, one consisting of Rad17, Rad24, Mec3, and Ddc1 (24) and the other containing the Rad9 protein (9), operate throughout the cell cycle, whereas a third pathway, involving Pol2, Dpb11, Drc1, Rfc5, and Mrc1, specifically monitors the progress of DNA replication and damage during S phase (2, 34). Damage-induced Rad53 activation leads to G₁/S delay through inhibition of the Swi6 transcription factor involved in CLN1/2 cyclin expression (45), inhibition of Dbf4/Cdc7 kinase-dependent late replication origin “firing” to delay S-phase progression (41), and G₂/M arrest through inhibition of the Polo-like kinase Cdc5 and the anaphase-promoting complex (39). In addition, Rad53 is involved in the transcriptional induction of DNA repair genes, such as ribonucleotide reductase (RNR) subunits (21); the direct regulation of repair proteins, such as Rad55 (3); chromatin assembly (15); and redistribution of the Ku70/80 and Sir2/3 silencing factors to sites of DNA damage (31). Rad53 is also essential for growth in unperturbed cell cycles through the regulation of deoxynucleoside triphosphate levels during S phase, but rad53Δ lethality can be suppressed by overexpression of RNR subunits or deletion of the RNR inhibitor and Rad53 target Sml1 (52). Conversely, as Rad53 is crucial for checkpoint-induced cell cycle arrest, downregulation of its activity is a target for antagonistic pathways that facilitate cell cycle reentry in the adaptation to limited DNA damage (35) or the recovery from arrest after successful DNA damage repair (27).

Rad53 belongs to the conserved family of FHA domain-containing checkpoint kinases, including mammalian Chk2, S. cerevisiae Dun1 and Mek1, and Schizosaccharomyces pombe Cds1. FHA domains are modular protein-protein interaction domains that preferentially bind to phosphothreonine (pThr) residues in target peptides (reviewed in references 12, 18, and 28). Rad53 contains two FHA domains, flanking the protein kinase catalytic domain. Both FHA domains are required for DNA damage-dependent Rad53 activation and possibly subsequent downstream signaling in vivo (37, 43). Interaction of the C-terminal FHA domain (FHA2) with phosphorylated Rad9 is critical for Rad53 activation (42, 46). A number of proteins—Rad9 (13), Dbf4 (11), Asf1 (43), and Ptc2/3 (27)—have been proposed as candidate ligands for the N-terminal Rad53 FHA1 domain, but it is likely that additional FHA domain-interacting proteins are involved in the wide range of Rad53 functions.

Here we describe a novel pThr-containing protein that specifically interacts with the pThr-binding site of the Rad53 FHA1 domain. Based on DNA damage response phenotypes associated with its deletion, we have termed this protein Mdt1 (for modifier of damage tolerance). Mdt1 is hyperphosphorylated in a checkpoint-dependent manner after DNA damage, and the MDT1 gene exhibits a number of genetic interactions with checkpoint components. Mdt1 is required for normal G₂/M cell cycle progression in the absence of DNA damage. Altogether, our data indicate that Mdt1 is a novel target of cell cycle arrest checkpoints.

MATERIALS AND METHODS

Yeast strains and plasmids.

All yeast strains used in functional experiments (Table 1) were derived from W303-1A strain U952-3B (that is RAD5) containing the sml1Δ mutation (52), except for a W303-1A MEC1-myc strain that was SML1 (38). rad53 alleles with FHA1 and kinase catalytic domain mutations were described before (36, 37), and MDT1-myc and mdt1-T305A alleles were generated by using similar PCR-based site-directed mutagenesis procedures. Gene disruptions were constructed by using standard PCR-based strategies (5) and confirmed by colony PCR. The MDT1-myc mec1Δ tel1Δ strain was obtained by sporulation and tetrad dissection of a heterozygous diploid W3030 strain (MATa/MATα MDT1-myc/mdt1::LEU2 MEC1/mec1::klURA3 TEL1/tel1::KAN SML1/sml1::HIS3). Yeast two-hybrid analyses were performed with PJ69-4A (MATa trp1-901 leu2-3,112 ura3-52 his3-200 gal4Δ gal80Δ LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ) (22). For overexpression under the control of the GAL1 promoter, the MDT1 open reading frame was fused by PCR to a C-terminal myc tag and cloned into p416-GAL1. For expression under the control of its own promoter from a centromeric plasmid, PCR products containing positions −500 to −1 of MDT1 and the full-length open reading frame fused to a C-terminal myc tag were cloned into the SacI and SmaI sites of pRS416. Correct sequences were confirmed by cDNA sequence analysis. Yeast transformations were performed by using lithium acetate-polyethylene glycol 3350. Untransformed strains were grown in YPD (1% yeast extract, 2% peptone, 2% glucose); plasmid-transformed strains were grown in synthetic complete medium lacking uracil. For repression of the GAL1 promoter, the medium contained 2% sucrose; for induction, the medium contained 2% sucrose plus 4% galactose. All incubations were performed at 30°C.

TABLE 1.

Yeast strains used in this study^a

Strain	Genotype	Reference or source
Y53	MATaade2-1 can1-100 leu2-3,112 trp1-1 ura3-1 sml1::HIS3	52
Y54	Y53 rad53::HIS3 sml1-1	52
Y57	Y53 rad53-R70A	36
Y59	Y53 rad53-K227A	36
Y60	Y53 rad53::URA3	36
Y109	Y53 mdt1::LEU2	This study
Y110	Y53 rad53-R70A mdt1::LEU2	This study
Y112	Y53 rad53::URA3 mdt1::LEU2	This study
Y117	Y53 rad53-K227A mdt1::LEU2	This study
Y122	Y53 rad9::LEU2	This study
Y123	Y53 rad53-R70A rad9::LEU2	This study
Y125	Y53 rad17::TRP1	This study
Y160	Y53 rad17::TRP1 mdt1::LEU2	This study
Y185	Y53 rad9::URA3 mdt1::LEU2	This study
Y186	Y53 rad9::URA3 mdt1::LEU2 rad53-R70A	This study
Y187	Y53 lcd1::LEU2	38
Y188	Y53 lcd1::LEU2 mdt1::TRP1	This study
Y189	Y53 MDT1-myc	This study
Y198	Y53 mdt1-T305A	This study
Y208	Y53 MDT1-myc mec1::klURA3	This study
Y218	Y53 mdt1-T305A rad17::TRP1	This study
Y272	Y53 MDT1-myc chk1::klURA3	This study
Y274	Y53 MDT1-myc rad53::klURA3	This study
Y343	Y53 MDT1-myc mec1::klURA3 tel1::KAN	This study

^aY53 is the wild type. kl, Kluyveromyces lactis.

HU, MMS, and UV sensitivity assays.

For drop tests, 2 μl of 10-fold serial dilutions (starting A₆₀₀, 0.5) of yeast cultures were spotted onto plates containing hydroxyurea (HU) or methylmethane sulfonate (MMS) or spotted onto YPD, UV irradiated by using a Uvitec CL-508 cross-linker (λ, 254 nm) with a single light bulb at various doses, and incubated for 3 days. For liquid assays, overnight cultures were diluted to an A₆₀₀ of 0.15 and grown for 3 h before the addition of MMS or HU. Aliquots were removed immediately before and at 3 h after the addition of MMS or at 2-h intervals over a 12-h period after the addition of 100 mM HU and plated on YPD. Colonies were counted after 3 days, and survival rates were determined as colonies formed in damage-treated samples as a percentage of colonies formed in relevant untreated samples in ≥3 independent experiments.

Yeast two-hybrid analyses.

PJ69-4A cells were cotransformed with plasmid pAS2 (36), encoding a Gal4 DNA-binding domain (BD)- Rad53 FHA1 domain fusion protein (residues 20 to 164), and an S. cerevisiae cDNA library fused to the Gal4 activation domain (AD) in pGADGH (17). Transformation mixtures were plated on synthetic complete medium lacking Leu and Trp to select for the two plasmids as well as His and Ade to select for interacting clones that restored Gal4 activity and activated the transcription of the GAL1-HIS3 and GAL2-ADE2 reporter genes. All reporter plates also contained 10 mM 3-amino-1,2,4-triazole to provide tighter control of the HIS3 reporter. A positive interaction was confirmed by retransformation assays of plasmid pGADGH-MDT1. pAS2 plasmids containing single-residue substitutions of the FHA1 pThr-binding site were described before (36). Additional constructs were generated by PCR and cloned into pGADGH, and correct sequences were confirmed by cDNA sequence analysis.

Cell cycle analyses.

Cells were synchronized by treatment of log-phase cultures with 15 μg of nocodazole (Calbiochem)/ml for 90 min plus another 10 μg/ml for 60 min or by treatment with two doses of 20 μg of α-factor/ml over 3 h. Synchrony (>90%) was assessed by microscopy. For MMS checkpoint experiments, cells were treated with 0.05% MMS for the last 30 min of the nocodazole arrest and washed with and released into YPD containing 0.05% MMS. For UV experiments, cells were irradiated with 80 J/m² and released into YPD after three washes. Aliquots were fixed in 70% ethanol and stained with 4′,6′-diamidino-2-phenylindole (DAPI), and ≥100 cells were scored per sample at a magnification of ×400 with a Zeiss Axiovert 25 inverted phase-contrast and fluorescence microscope (37). Flow cytometry analyses were performed as described previously (36).

Immunofluorescence microscopy.

Five-milliliter cultures were grown to 1 × 10⁷ to 2 × 10⁷ cells/ml, initially fixed with 3.7% formaldehyde for 10 min, and resuspended in PEM buffer (100 mM PIPES, 1 mM EGTA, 1 mM MgSO₄ [pH 6.9]) containing 3.7% formaldehyde for overnight fixation at 4°C. Cells were washed twice with PEM buffer, incubated with 150 mM glycine in phosphate-buffered saline (PBS) for 10 min, washed with PEMS (PEM buffer, 1.2 M sorbitol), resuspended in 1 ml of PEMS containing 0.25 mg of Zymolyase 20T (Seikagaku), and incubated at 30°C for 10 to 30 min until cell walls were digested. Cells were immediately placed on ice, washed with PEMS, and resuspended in 0.5 ml of PEMS. Cell suspensions (20 μl) were dried on polylysine-coated slides, washed twice with PBS, and submerged in −20°C methanol for 6 min and −20°C acetone for 30 s. Slides were washed twice with PBS and blocked with PBS-T-B (PBS, 0.1% Tween 20, 1% bovine serum albumin) for 1 h in a humid chamber. Primary antibody incubation with 1/100-diluted antitubulin TAT1 (generously provided by Matthew O'Connell) in PBS-T-B was performed overnight at 4°C, followed by four PBS-0.1% Tween 20 (PBS-T) washes at room temperature. Alexa 594-conjugated goat anti-mouse immunoglobulin G (Molecular Probes) in PBS-T-B was added at room temperature. After 1 h, DNA was counterstained with 1.7 μg of Hoechst 33258/ml in PBS-T for 10 min, and slides were washed four times with PBS-T. Coverslips were mounted with fluorescence microscopy mounting medium (Dako) and sealed with nail polish. Photomicrographs were taken on Kodak 320-T color reversal film at an original magnification of ×250 by using a 100× oil immersion objective.

Protein blots, kinase assays, and immunoprecipitation.

Immunoblotting was performed as previously described (36, 37). For immunoprecipitation, clarified yeast extracts were prepared with native lysis buffer (PBS containing 10% glycerol, 1% NP-40, 0.5% Triton X-100, 5 mM EDTA, 5 mM EGTA, 50 NaF, 10 mM β-glycerophosphate, 5 mM sodium pyrophosphate, 2 mM phenylmethylsulfonyl fluoride, and 1× protease inhibitor mix [Sigma]) by bashing with glass beads and were incubated for 2 h at 4°C with 20 μg of mouse anti-myc monoclonal antibody 9E10 (generously provided by Andy Poumbourios)/ml and 20 μl of protein G-Sepharose (Amersham), washed with lysis buffer, and eluted into sodium dodecyl sulfate loading buffer. Rabbit polyclonal anti-pThr antibody was obtained from Cell Signaling, and mouse monoclonal antiactin antibody was obtained from Chemicon.

Kinase assays were carried out as described previously (20) for 20 min at 30°C in a 20-μl reaction mixture containing 20 mM Tris (pH 7.4), 10 mM magnesium acetate, 4 mM manganese chloride, 0.1 mM dithiothreitol, 0.05% (wt/vol) Tween 20, 50 μM ATP, and 1 μCi of [γ-³²P]ATP with anti-myc antibody immunoprecipitates from 10-ml log-phase cultures treated with 0.05% MMS. A recombinant Mdt1 (positions 531 to 668)-His₆ fragment was purified by Ni²⁺-agarose chromatography essentially as described previously (19) and added at ∼20 ng/μl. Reactions were stopped by boiling in sample buffer, and 15 μl from a 40-μl total volume was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography.

RESULTS

Identification of Mdt1 as a Rad53 FHA1 domain-interacting protein.

In order to identify novel Rad53 FHA1-binding proteins, a yeast two-hybrid screen of a yeast cDNA library fused to the Gal4 AD was performed with a Gal4 BD-FHA1 domain fusion vector as bait. In this system, protein-protein interactions of AD-containing and BD-containing fusion proteins in cotransformed yeast cells restore Gal4 transcriptional activity and lead to the expression of HIS3 and ADE2 reporter genes from distinct promoters, allowing growth on plates lacking both histidine and adenine (16). This approach led to the isolation of an FHA1-interacting clone expressing an in-frame AD fusion protein with residues 302 to 668 encoded by the previously uncharacterized yeast open reading frame YBL051C, which we have termed MDT1 (Fig. 1). Full-length Mdt1 contains 668 residues, with an N-terminal RNA recognition motif (RRM) domain, a nuclear localization signal, a coiled coil, and a total of 15 SQ/TQ motifs; 14 of the latter motifs are concentrated within 288 residues, including a 60-residue supercluster containing 1 TQ motif and 8 SQ motifs (Fig. 1A). Such SQ/TQ cluster domains (SCDs), which represent potential phosphorylation sites for ATM-like kinases, are a hallmark of numerous checkpoint proteins, including Rad53, Rad9, and Mrc1 (2, 26, 42); the Rad9 SCD (containing six SQ/TQ motifs between residues T390 and Q458) was recently shown to be critical for the interaction with Rad53 (42).

FIG. 1.

Mdt1 interaction with the Rad53 FHA1 domain. (A) Schematic diagram of Mdt1 domain topology and positions of SQ/TQ motifs (balls). Bars and numbers indicate Mdt1 fragments used in interaction assays. NLS, nuclear localization signal; CC, coiled coil. (B to D) Yeast two-hybrid analysis of Mdt1 (positions 302 to 668) with a wild-type or mutated FHA1 domain or an empty vector (B), of the indicated Mdt1 fragments with a wild-type FHA1 domain (C), or of pGADGH-MDT1 with the FHA1 and FHA2 domains (D). Cotransformed cultures were streaked on plates lacking Leu and Trp (−LW) or lacking Leu, Trp, His, and Ade but containing 10 mM 3-amino-1,2,4-triazole (−LWHade). (E) Rad53 Western blot of anti-myc antibody immunoprecipitates (IP) from the indicated cultures (four left lanes) and the corresponding lysates (3% of input; four right lanes) without or with MMS treatment. (F) Western blot analysis of immunoprecipitated myc-tagged Mdt1 expressed from the GAL1 promoter in the indicated strains probed with anti-myc antibody (top panel) or reprobed with anti-pThr antibody (bottom panel). Arrows indicate the positions of the myc-tagged Mdt1 bands. Numbers at right indicate mass standards (kilodaltons). WT, wild type.

Interestingly, the FHA1-Mdt1 interaction was abolished by several single-residue substitutions that disrupt the FHA1 phosphopeptide-binding site (36), including the very subtle R70K and S105T substitutions (Fig. 1B); this result indicates that Mdt1 interacts specifically with the pThr-binding site of the Rad53 FHA1 domain. Truncation analyses revealed that residues 297 to 375 of the Mdt1 SCD are sufficient and critical for its interaction with the FHA1 domain, as their deletion abolished reporter gene activation, whereas the C-terminal SQ/TQ motif supercluster was dispensable for this interaction (Fig. 1C). Importantly, substitution of the single Thr residue (T305) of the Mdt1 fragment from residues 297 to 375 by Ala inhibited its interaction with the FHA1 domain (Fig. 1C). In contrast to its interaction with the Rad53 FHA1 domain, Mdt1 did not interact with the Rad53 FHA2 domain (Fig. 1D). The reciprocal dependence of the Rad53 FHA1 domain-Mdt1 interaction on the integrity of the FHA1 pThr-binding site and the presence of T305 in Mdt1 strongly indicates that Mdt1 interacts with Rad53 in a pThr-specific manner.

Constitutive Mdt1 phosphorylation and Rad53 interaction in vivo.

Small amounts of Rad53 could be coimmunoprecipitated with an anti-myc antibody from a yeast strain expressing myc-tagged Mdt1 at essentially physiological levels (under the control of its own promoter on a low-copy-number centromeric plasmid in an mdt1Δ strain) but not from the wild-type strain expressing untagged Mdt1, confirming the two-hybrid interaction of the two proteins (Fig. 1E). The myc tag did not interfere with Mdt1 function (see Fig. 5B). Surprisingly, this interaction had already occurred in the absence of MMS treatment and seemed to involve predominantly unphosphorylated or underphosphorylated Rad53 (Fig. 1E). However, immunoprecipitated myc-tagged Mdt1 consistently appeared as a doublet in Western blots, and the more slowly migrating Mdt1 form already reacted with an anti-pThr antibody under basal conditions (Fig. 1F). Together with the results of the two-hybrid analysis, these data suggest that Mdt1 is constitutively phosphorylated on Thr as a prerequisite for its interaction with the Rad53 FHA1 domain.

FIG. 5.

Aberrant morphology of mdt1Δ cells. (A) Phase-contrast and fluorescence photomicrographs of MDT1 (WT) or mdt1Δ cells grown in YPD without DNA-damaging agents and stained for nuclear DNA (Hoechst) and tubulin. (B) Phase-contrast micrographs of mdt1Δ cells containing pRS416 or pRS416-MDT1-myc.

Mec1/Tel1-dependent Mdt1 hyperphosphorylation in response to DNA damage.

For unknown reasons, we found highly variable levels of expression of myc-tagged Mdt1 from plasmid vectors in various checkpoint mutants (for example, note the higher levels of plasmid-expressed myc-tagged Mdt1 in rad53-K227A; Fig. 1F). As this fact complicated Mdt1 phosphorylation analysis after DNA damage, we generated a stable MDT1-myc allele by integrating a C-terminal myc tag into the genomic MDT1 locus, resulting in essentially equal protein levels in all strains tested (Fig. 2 and data not shown). Interestingly, Mdt1 was shifted to more slowly migrating forms in response to MMS treatment (Fig. 2A). This DNA damage-dependent hypershift (as well as the constitutive pThr shift) could be reversed by pretreatment of immunoprecipitated myc-tagged Mdt1 with λ phosphatase (Fig. 2B), indicating that Mdt1 is hyperphosphorylated in response to DNA damage.

FIG. 2.

DNA damage-dependent Mdt1 hyperphosphorylation. (A) Anti-myc antibody Western blot analysis of chromosomally myc-tagged Mdt1 in wild-type (top) and mec1Δ (bottom) cells treated with 0.1% MMS for the indicated times. (B) Western blot of immunoprecipitated myc-tagged Mdt1 expressed from a centromeric plasmid under the control of its own promoter in mdt1Δ cells in the presence or absence of 0.1% MMS. Some samples were incubated with λ phosphatase before electrophoresis. (C) Anti-myc antibody Western blot analysis of chromosomally myc-tagged Mdt1 without or with 0.1% MMS treatment for 2 h in wild type, mec1Δ, and mec1Δ tel1Δ cells. (D) Kinase assays of myc-tagged Mec1 or control immunoprecipitates (IP). The left panels show kinase reactions with Mdt1 (positions 531 to 668) as a substrate; the right panels show no-substrate controls of the corresponding gel areas. Phosphorimaging autoradiographs are shown in the top panels; the relevant sections of the gels stained with Coomassie brilliant blue (CBB) are shown in the bottom panels.

Mdt1 was still noticeably hypershifted after DNA damage in mec1Δ cells (Fig. 2A, bottom panel) as well as in rad53Δ and chk1Δ cells (data not shown). S. cerevisiae contains a second ATM-like kinase, Tel1, which has an overlapping substrate specificity and can sometimes substitute for Mec1 (14, 30, 49). We therefore generated a mec1Δ tel1Δ double-deletion strain to test whether the DNA damage-dependent Mdt1 hyperphosphorylation is checkpoint dependent. Importantly, MMS-induced Mdt1 hyperphosphorylation was essentially abolished in the mec1Δ tel1Δ strain (Fig. 2C), demonstrating that Mdt1 is a checkpoint target. Furthermore, a recombinant Mdt1 fragment containing the supercluster of nine SQ/TQ motifs was readily phosphorylated in myc immunoprecipitates from a strain containing a myc-tagged MEC1 allele but not from the untagged wild-type strain (Fig. 2D). Although we cannot exclude the possibility that the Mec1 immunoprecipitates contained another Mec1-associated protein kinase, this experiment strongly suggests that Mdt1 is a direct Mec1 kinase substrate. Altogether, these experiments indicate that Mdt1 is an in vivo substrate of the yeast ATM-like kinase Mec1/Tel1.

Genetic interactions of MDT1 with checkpoint components.

To further evaluate the roles of MDT1 in DNA damage, we tested MDT1 for genetic interactions with other checkpoint genes. MDT1 showed a strong genetic interaction with rad53Δ in overexpression experiments with a pGAL1-MDT1-myc plasmid compared to the empty vector control on medium containing 4% galactose. Under these conditions, MDT1 overexpression had no effect on wild-type cells but reduced colony formation of rad53Δ cells in the absence of exogenous DNA-damaging agents by ∼100-fold (Fig. 3A). Because rad53Δ cells are severely compromised in checkpoint functions and are kept viable only by higher deoxynucleoside triphosphate levels resulting from the sml1Δ mutation (51, 52), these data suggest that elevated Mdt1 levels hypersensitize rad53Δ mutants to endogenous DNA damage.

FIG. 3.

Genetic interactions of MDT1 with checkpoint components. (A) Drop test analysis of serial 10-fold dilutions of the indicated yeast strains transformed with pGAL1-MDT1-myc (M) or an empty vector control (V) plated on medium lacking uracil but containing 2% sucrose (−Gal) or 2% sucrose and 4% galactose (+Gal). WT, wild type. (B) Survival analysis of the indicated yeast strains treated with 0.04% MMS for 3 h. Data are the mean and standard error of three experiments. (C) Drop test analysis of serial 10-fold dilutions of the indicated yeast strains on YPD (top panels) or YPD containing 0.02% MMS (bottom panels).

We next tested whether mdt1Δ affects DNA damage sensitivity by treating liquid cultures with 0.04% MMS for 3 h. In these assays, mdt1Δ generally improved DNA damage tolerance, even in the wild-type strain, and was particularly efficient in suppressing the DNA damage hypersensitivity of checkpoint-compromised rad9Δ and rad17Δ strains, resulting in a four- to fivefold increase in viability (Fig. 3B). The mdt1Δ effect was even more dramatic in a rad9Δ rad53-R70A double-mutant background. rad9Δ rad53-R70A were considerably more hypersensitive to MMS than either single mutation (Fig. 3C), but this hypersensitivity was completely suppressed by mdt1Δ, resulting in ∼100-fold more efficient colony formation (Fig. 3C, compare lanes 6 and 8). However, mdt1Δ had no effect on rad53Δ, indicating that it acts in a Rad53-dependent manner. Altogether, these various genetic interactions of MDT1 (Fig. 3) are consistent with its direct interaction with Rad53 (Fig. 1) and its Mec1/Tel1-dependent phosphorylation after DNA damage (Fig. 2) and strongly support its connection with the checkpoint machinery.

Role of Mdt1 in the normal G₂/M transition and DNA damage-dependent cell cycle delay.

A major reason for the DNA damage hypersensitivity of rad9Δ and rad17Δ strains is their inability to delay anaphase in response to DNA damage (50). We therefore investigated the role of MDT1 in G₂/M cell cycle transitions by scoring nuclear division kinetics after the release of cells from nocodazole arrest. Interestingly, under control conditions, the completion of mitosis after nocodazole release was delayed by about 20 min in mdt1Δ cells compared to wild-type cells (Fig. 4A), with a much lower content of binucleate cells (wild type, 40.2% ± 3.2%; mdt1Δ, 28% ± 3.6%) after 60 min. Similar results were obtained for several checkpoint-deficient strains, where the mdt1Δ allele consistently caused delayed binucleation kinetics compared to the relevant MDT1 control (Fig. 4A and data not shown). For all of these strains, mdt1Δ cultures (in rich media, irrespective of DNA-damaging agents) contained a considerable number of cells with aberrant morphology. mdt1Δ cells appeared elongated and multibudded, both in mononucleated premitotic cells (Fig. 5A) and in binucleated postmitotic cells (Fig. 5A); as a consequence, mitotic spindles in mdt1Δ cells were considerably longer than those in the corresponding wild-type cells (Fig. 5A, bottom panel). The aberrant morphology was reversed by the expression of MDT1-myc under the control of its own promoter on a centromeric plasmid (Fig. 5B). This mdt1Δ phenotype is reminiscent of that of cell cycle mutants that maintain high levels of G₁ cyclin-dependent kinase activity (Cdc28-Cln) but insufficient levels of G₂/M cyclin-dependent kinase activity (Cdc28-Clb) required for switching from apical to isotropic bud growth during G₂/M phase, for example, mutations in CLB1/2 (47), CDC34 (44), or FKH1/2 forkhead transcription factors (25). Therefore, the aberrant morphology of mdt1Δ cells (Fig. 5A) and their delayed binucleation kinetics after release from nocodazole arrest in the absence of exogenous DNA damage (Fig. 4A) indicate that Mdt1 has a physiological function in contributing to the G₂/M transition during normal cell cycles.

FIG. 4.

Role of MDT1 in G₂/M cell cycle progression. (A to C) Percentages of binucleated cells (mean and standard error of ≥3 independent experiments) in the indicated strains after release from nocodazole arrest under control conditions (A) and in response to MMS (B) and UV (C) damage. (D) Flow cytometry analysis of similar samples. The main peak at 20 min represents G₂/M cells with a 2n DNA content; the peak with a 1n DNA content at later time points represents G₁ cells. noc, nocodazole; WT, wild type. (E and F) Western blot analyses of Rad53 in the indicated strains from untreated cultures (0) (E), from unsynchronized cultures treated with 150 mM HU (H) or 0.1% MMS (M) for 1 h (E), or from nocodazole-arrested cultures without or with 0.1% MMS treatment (F).

We next tested how the delayed binucleation kinetics of mdt1Δ cells were affected by DNA damage of cells prior to the nocodazole release. As expected, wild-type cells had significantly delayed binucleation kinetics after pretreatment with MMS (Fig. 4B) or UV (Fig. 4C), in contrast to rad17Δ cells, which showed barely delayed progression through mitosis. Interestingly, in response to DNA damage, mdt1Δ cells had binucleation kinetics very similar to those of wild-type cells, in contrast to their delayed kinetics under control conditions. If DNA damage-dependent and mdt1Δ-dependent cell cycle delays are independent, then mdt1Δ cells should progress more slowly through mitosis than wild-type cells irrespective of damage; however, in response to MMS they had essentially identical nuclear division kinetics (Fig. 4B). Thus, the finding that mdt1Δ-dependent and DNA damage-dependent cell cycle delays are not additive suggests that they are epistatic, genetically placing them in the same pathway.

The mdt1Δ-dependent delay was fully maintained and even enhanced in response to DNA damage in the rad17Δ mdt1Δ strain compared to the parental rad17Δ strain, in a manner that almost approached the checkpoint-induced delay in wild-type cells (Fig. 4B and C). Similar results were obtained for other checkpoint mutants carrying the mdt1Δ allele (rad9Δ mdt1Δ and lcd1Δ mdt1Δ; data not shown) and were confirmed in flow cytometry experiments, where the reemergence of G₁ cells (with a 1n DNA content) was not delayed by MMS treatment in rad17Δ cells but was delayed by ∼40 min in wild-type cells and by more than 20 min in rad17Δ mdt1Δ cells (Fig. 4D). Because such delays would allow more time for DNA damage repair, these results provide an explanation for the DNA damage-protective effect of mdt1Δ (Fig. 3).

mdt1Δ had no effect on Rad53 activation for the various strains analyzed in asynchronous (Fig. 4E and data not shown) or nocodazole-arrested (Fig. 4F) cultures, indicating that the delayed cell cycle kinetics are not a consequence of restored Rad53 activation. These results suggest that Mdt1 acts downstream of or in parallel with Rad53.

Because residue T305 is critical for the interaction of Mdt1 with the Rad53 FHA1 domain (Fig. 1C), we generated an mdt1-T305A allele to test its role in the DNA damage response. This allele did not obviously affect DNA damage sensitivity in an otherwise wild-type background but, interestingly, led to slightly but reproducibly reduced hypersensitivity to MMS and UV in rad17Δ cells (Fig. 6), similar to mdt1Δ. These results indicate that T305 does not function solely to target Mdt1 to the checkpoint machinery via the Rad53 FHA1 domain, in which case it would be expected to result in damage hypersensitivity similar to that of the rad53-R70A allele, but rather that it has additional functions in cell cycle control or the DNA damage response.

FIG. 6.

Partial suppression of rad17Δ DNA damage hypersensitivity by mdt1-T305A. Equal serial 10-fold dilutions of the indicated strains (starting optical density at 600 nm, ∼0.5) were spotted on control plates after being untreated (top) or treated for 3 h with 0.04% MMS (middle) or 150 J of UV light per m² (bottom). WT, wild type.

DISCUSSION

Cell cycle function of Mdt1.

In this study, we have identified Mdt1 as a novel protein involved in G₂/M progression during normal cell cycles as well as the cellular response to DNA damage. mdt1Δ cells are delayed in completing mitosis after release from a preanaphase arrest (Fig. 4A) and have a morphological phenotype characteristic of G₂/M cell cycle mutants (Fig. 5A), indicating that Mdt1 normally promotes progression through late phases of the cell cycle. The mechanism of Mdt1 function during normal cell cycles is currently unclear. In BLAST searches, the highest scores are restricted to the Mdt1 RRM domain, which is remarkably similar to RRM domains in two uncharacterized putative fission yeast proteins (on cosmids SPAC12G12 and SPBC16A3; >60% identity and ≥80% similarity) and reasonably similar to an RRM domain in mammalian mRNA cleavage-stimulating factors (31% identity and 51% similarity). Mdt1 therefore could have a function in RNA metabolism, possibly in the posttranscriptional regulation of genes involved in G₂/M cell cycle progression. However, RRM domains are not restricted to protein-RNA interactions, as some RRM domains mediate interactions with single-stranded DNA, for example, in the yeast telomeric DNA-binding protein Gbp2 (29) (30% RRM identity with Mdt1); these data could be relevant for interactions with single-stranded DNA resulting from the processing of DNA double-strand breaks. Interestingly, a novel human protein termed ASCIZ, which shares remarkable structural and functional similarities with Mdt1, was recently identified (C. J. McNees et al., unpublished data). The two proteins have similar domain topologies, with an N-terminal nucleic acid-binding domain (a double Zn²⁺ finger domain in ASCIZ instead of an RRM domain) followed by a nuclear localization signal and a C-terminal SCD, and a high level of sequence similarity (>17% identity and >36% strong similarity). Like Mdt1, ASCIZ interacts via its SCD with the pThr-binding site of the FHA domain of the human Rad53 ortholog Chk2, and it has roles during normal cell cycles and in the DNA damage response. These data indicate that Mdt1 and ASCIZ are the founding members of an evolutionarily conserved protein family.

While this work was in progress, MDT1 (YBL051C) was independently isolated in three unrelated genetic screens. In an automated high-throughput screen, mdt1Δ was found to be synthetic lethal with bni1Δ, involved in actin cable assembly, polarized cell growth, and preanaphase spindle orientation (48). Interestingly, while MDT1 or the mechanism of its genetic interaction with BNI1 was not further characterized, several other synthetically lethal interactions of bni1Δ in this screen involved genes functioning in mitosis and the transition from apical to isotropic bud growth (48); these data indirectly support our conclusion of a cell cycle function for Mdt1. Likewise, in a systematic analysis of cell size distribution in all available gene deletion strains, an mdt1Δ strain was found to be among the 5% most abnormally large strains (23), consistent with our morphological findings (Fig. 5). In another study, YBL051C was designated PIN4 because it is present on a plasmid that can induce prion formation in yeast cells when overexpressed at very high levels (10). However, it is unclear whether YBL051C is indeed responsible for the PIN⁺ phenotype, first because YBL051C was significantly truncated on the PIN4 plasmid and second because it was not the only open reading frame on this plasmid (10).

Mdt1 functions as a checkpoint target.

The following considerations strongly support the notion that Mdt1 is a novel checkpoint target. First, Mdt1 interacts physically with the Rad53 FHA1 domain, linking it to the checkpoint machinery (Fig. 1). Second, Mdt1 hyperphosphorylation in response to DNA damage in vivo is checkpoint dependent, and Mdt1 is a direct Mec1 kinase substrate in vitro (Fig. 2). Third, MDT1 exhibits several strong genetic interactions with other checkpoint components (Fig. 3).

Checkpoint-dependent phosphorylation either can activate targets to fulfill their DNA damage response functions, for example, Rad9 (49), Rad53 (40), or Rad55 (3), or can inhibit protein functions that antagonize the DNA damage response, for example, Sml1 (51) or Dbf4 (11). Given that the promitotic function of Mdt1 is intrinsically antagonistic to cell cycle arrest checkpoints, the most plausible explanation for Mec1/Tel1-dependent phosphorylation is that it serves to negatively regulate Mdt1 functions in order to facilitate cell cycle arrest in the presence of DNA damage (Fig. 7). This model is supported by the finding that the mdt1Δ-dependent and DNA damage-induced cell cycle delays are not additive (Fig. 4B), suggesting that both act in the same pathway. Because the negative regulation of Mdt1 functions by checkpoints would have the same effect as the deletion of Mdt1, we propose that the partial restoration of the G₂/M checkpoint in rad17Δ cells by mdt1Δ is not simply an indirect consequence of slower mitosis in the absence of Mdt1 but rather indicates that Mdt1 is an important downstream target of cell cycle arrest checkpoints. The fact that mdt1Δ does not fully restore the G₂/M arrest checkpoint in rad17Δ cells could be explained by the presence of parallel pathways, for example, the well-established Cdc5-anaphase-promoting complex pathway (39) (Fig. 7).

FIG. 7.

Model for the role of Mdt1 during normal cell cycles (grey lines) and in response to DNA damage (black lines). Mdt1 phosphorylation on T305 contributes to normal mitosis in unperturbed cell cycles, in a pathway parallel to that of Cdc5. Phosphorylated T305 also functions as a scaffold to link Mdt1 to the checkpoint machinery via the interaction of Mdt1 with the Rad53 FHA1 domain pThr-binding site. In response to DNA damage, Mdt1 is phosphorylated on multiple residues in a Mec1/Tel1-dependent manner, inhibiting its promitotic function. Together with checkpoint-dependent inactivation of parallel pathways, for example, Rad53-dependent phosphorylation of Cdc5, this event leads to G₂/M delays while DNA damage persists.

Our data establish Mdt1 as a member of a growing list of proposed Rad53 FHA1 domain ligands that includes Rad9 (13), Dbf4 (11), Asf1 (43), and Ptc2/3 (27), highlighting the versatility of this pThr-binding domain. Rad53 interacts with Mdt1 even in the absence of DNA damage, a finding which is most likely due to the fact that the critical Mdt1 residue, T305, is partially phosphorylated during normal cell cycles (Fig. 1). At this point, it remains to be determined whether the phosphorylated T305-FHA1 interaction functions to link Mdt1 to the checkpoint machinery in general (Fig. 7) or whether it facilitates a direct kinase-substrate interaction with Rad53. Mdt1 hyperphosphorylation in response to DNA damage results in a heterogeneous electrophoretic mobility pattern similar to those of other SCD-containing proteins, such as Rad9 (42), Rad53 (37), and Mrc1 (2). These data indicate that multiple sites are phosphorylated to variable degrees, reminiscent of the cell cycle-dependent phosphorylation of Sic1, where the biological effect depends not so much on the precise location of phosphorylation sites but rather on a threshold of a minimal number of phosphorylated residues (33). Mdt1 contains 15 SQ/TQ motifs, and our in vivo and in vitro data (Fig. 2) indicate that Mec1 and Tel1 are the major Mdt1 kinases. DNA damage-dependent Mdt1 phosphorylation was apparently unaffected in rad53Δ cells (data not shown), but a limited number of Rad53-specific phosphorylation events would be obscured easily by the heterogeneous banding pattern resulting from multiple Mec1/Tel1 phosphorylation sites. However, T305 seems to have additional functions beyond its role in linking Mdt1 to the checkpoint machinery, because the mdt1-T305A allele paradoxically acted as a partial suppressor of rad17Δ-associated DNA damage hypersensitivity (Fig. 6), similar to mdt1Δ. The simplest explanation for this observation is that the basal phosphorylation of T305 contributes to the normal cell cycle function of Mdt1 (Fig. 7), such that its replacement by nonphosphorylatable A305 results in an mdt1Δ-like DNA damage phenotype. This dual function of phosphorylated T305 could serve as a “go” signal during unperturbed cell cycles but at the same time provide an immediate target for “stop” signals in response to DNA damage by linking Mdt1 to the Mec1/Tel1 pathway.

In conclusion, we have identified Mdt1 as a novel protein involved in G₂/M cell cycle progression and as a novel checkpoint target. The results provide the basis for future studies of the precise mechanisms of Mdt1 functions during normal cell cycles and Mdt1 regulation by and roles in DNA damage checkpoints.