The Saccharomyces cerevisiae 14-3-3 proteins
Bmh1 and Bmh2 directly influence the DNA
damage-dependent functions of Rad53
Takehiko Usui* and John H. J. Petrini*†‡
*Laboratory of Chromosome Biology, Memorial Sloan–Kettering Cancer Center, 1275 York Avenue, New York, NY 10021; and†Weill Medical College,
Cornell University Graduate School of Medical Sciences, 445 East 69th Street, New York, NY 10021
Communicated by Thomas J. Kelly, Sloan–Kettering Institute, New York, NY, December 18, 2006 (received for review September 27, 2006)
In this study, we mutated autophosphorylation sites in Rad53
based on their conservation with previously identified autophos-
phorylation sites in the mammalian Rad53 ortholog, Chk2. As with
wild-type Rad53, the autophosphorylation mutant, rad53-TA, un-
dergoes Mec1/Tel1-dependent interactions with Rad9 and Dun1 in
response to genotoxic stress. Whereas rad53-TA in vitro kinase
activity is severely impaired, the rad53-TA strains are not com-
pletely deficient for cell-cycle checkpoint functions, indicating that
the mutant kinase retains a basal level of function. We describe a
genetic interaction among Rad53, Dun1, and the 14-3-3 proteins
Bmh1 and Bmh2 and present evidence that 14-3-3 proteins directly
facilitate Rad53 function in vivo. The data presented account for
the previously observed checkpoint defects associated with 14-3-3
mutants in Saccharomyces pombe and Saccharomyces cerevisiae.
The 14-3-3 functional interaction appears to modulate Rad53
activity, reminiscent of 14-3-3’s effect on human Raf1 kinase and
distinct from the indirect mode of regulation by 14-3-3 observed
for Chk1 or Cdc25.
autophosphorylation ? DNA damage checkpoint ? kinase ?
integrity in the face of various genotoxic stresses (reviewed in
refs. 1 and 2). The checkpoints are activated via signal trans-
duction networks that are schematically comparable to other
cellular signal transduction pathways (e.g., MAP kinase path-
ways). Once damage is detected, PIKKs (PI3-kinase like kinases
Mec1 and Tel1 in Saccharomyces cerevisiae and ATM and ATR
in mammals) are responsible for the initial transduction of the
DNA damage signal.
PIKK-dependent phosphorylation of downstream targets fa-
cilitates a series of physical interactions that involve the phos-
phopeptide-binding domains, FHA (forkhead-associated) and
BRCT (Brca1 C terminus), found in the numerous DNA dam-
age-response proteins (3, 4). Foremost among such targets are
mediator proteins such as Rad9 and Mrc1 as well as effector
kinases such as Rad53, which require PIKK phosphorylation for
subsequent activation of the DNA damage-response pathway
(5–7). Upon DNA damage or DNA replication stress, Rad53 and
its orthologs in Saccharomyces pombe (Cds1) and mammals
(Chk2) are phosphorylated by PIKKs and then subsequently
undergo autophosphorylation to become fully active (5, 8–13).
Autophoshorylation of Rad53 depends on FHA-dependent
interaction with its mediator proteins Rad9 and Mrc1, which is
potentiated by PIKK-phosphorylation (12–15). Evidence sug-
gests that the Chk2 and Cds1 FHA domains also mediate
self-oligomerization in the course of autophosphorylation (11,
13, 16, 17). Finally, the Rad53 substrate Dun1 also has a FHA
domain that engages Rad53 through PIKK-dependent phos-
phorylation sites (18, 19). The formation of these PIKK-
the DNA damage or replication stress signal is amplified.
NA damage and DNA replication checkpoints are highly
conserved cellular surveillance systems to maintain genome
The activation-loop autophosphorylation sites in Chk2, T383/
T387, are conserved in yeast orthologues Rad53 and Cds1 (11,
13) [see supporting information (SI) Fig. 7], and Rad53 T354 has
been confirmed as a bona fide autophosphorylation sites by mass
spectrometric analysis (12). We hypothesized that rad53 auto-
phosphorylation site mutants would retain PIKK phosphoryla-
tion-dependent basal activity but would be impaired in signal
amplification. The rad53 mutant proteins would thus have a
higher level of activity than a rad53-K227A kinase-dead protein
(rad53-KD) (8), but would nevertheless be hypomorphic. Anal-
ysis of autophosphorylation-site mutants would thus allow us to
parse the functional significance of autophosphorylation from
the initial Mec1/Tel1 (PIKK)-dependent transphosphorylation.
We further hypothesized that phenotypic and molecular analyses
of rad53 autophosphorylation-site mutants might lead to iden-
tification of mediators of Rad53 function involved in the signal
We mutated both T354 and T358 or T354 alone to alanines
and examined the rad53-T354AT358A (rad53-TA) or rad53-
T354A mutants to test the hypothesis. Although in vitro and in
vivo assessments revealed that autophosphorylation mutants
specify limited activity, the PIKK phosphorylation-induced pro-
tein interaction of rad53-TA and rad53-T354A with Rad9 and
Dun1 were unaffected. rad53-TA and rad53-T354A exhibited
sensitivity to hydroxyurea (HU) and methyl methanesulfonate
(MMS), but were more proficient in checkpoint signaling than
rad53-KD. We screened for high-copy suppressors that could
compensate for impaired Rad53 autophosphorylation. We rea-
soned that two general categories of gene products would
suppress the rad53-TA phenotype when present at elevated
levels: Proteins that are regulated by Rad53 (i.e., Rad53 sub-
strates) and, when overexpressed, would compensate for re-
duced Rad53 activity; second, proteins that interact with Rad53
and direct the active kinase to particular targets. When overex-
pressed, this class of proteins might potentiate the interaction of
the partially active kinase with its targets. This category could
include proteins that govern the transition to, and stabilize the
active conformation of, the kinase. Alternatively, mediators that
function downstream of Rad9, or proteins that enhance inter-
action with mediator proteins could also emerge from this
J.H.J.P. analyzed data; and T.U. and J.H.J.P. wrote the paper.
The authors declare no conflict of interest.
Freely available online through the PNAS open access option.
Abbreviations: HU, hydroxyurea; IP, immunoprecipitation; MMS, methyl methanesulfo-
nate; PIKK, PI3-kinase-like kinase; WB, Western blot.
‡To whom correspondence should be addressed at: Laboratory of Chromosome Biology,
Memorial Sloan–Kettering Cancer Center, 1275 York Avenue, RRL 901C, New York,
NY 10021. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2007 by The National Academy of Sciences of the USA
www.pnas.org?cgi?doi?10.1073?pnas.0611259104 PNAS ?
February 20, 2007 ?
vol. 104 ?
no. 8 ?
We obtained 14-3-3 paralogs, BMH1 and BMH2, multiple
times upon screening for resistance of rad53-TA and rad53-
T354A to HU. The data obtained support a functional interac-
tion of Bmh1/2 in the Rad53 checkpoint pathway and suggest
that these proteins may be mediators of Rad53 activity.
In Vitro Properties of rad53-TA and rad53-T354A.Weestablishedtwo
rad53 alleles in which autophosphorylation sites in the activation
loop were altered, rad53-T354A and the doubly mutant allele
dead rad53-K227A D339A gene product demonstrate that PIKK-
dependent protein interactions are unaffected by the lack of
Rad53 autophosphorylation (19). To examine whether these
interactions were also unaffected in rad53-TA and rad53-T354A,
a Flag epitope was targeted to the 3? end of rad53-TA and
rad53-T354A loci. Immunoprecipitation (IP) from extracts of
wild-type, rad53-KD-, rad53-TA-, and rad53-T354A-expressing
cells was carried out. rad53-TA, rad53-T354A, and rad53-KD
were coprecipitated with Rad9 after MMS treatment in a
manner indistinguishable from wild type (Fig. 1A). These data
suggest that the alterations in Rad53 do not affect Mec1/Tel1
phosphorylation of Rad9.
DNA damage also induces the interaction of Rad53 with its
substrate, Dun1 (18, 19). Dun1-Myc and Flag-Rad53, -rad53-KD,
-rad53-TA or -rad53-T354A-expressing cells were tested. Myc IPs
from their extracts showed that rad53-TA and rad53-T354A as
well as rad53-KD and wild type coprecipitated with Dun1
following treatment with MMS (Fig. 1B). This interaction takes
place through the Dun1 FHA domain, which preferentially binds
phosphorylated residues (3, 18, 19), supporting the view that
rad53-TA and rad53-T354A are phosphorylated by Mec1/Tel1.
The residual activity of the rad53 autophosphorylation mu-
tants was assessed in vitro. Rad53 (Flag) IP from extracts of
MMS-treated cells were performed, and the ability of rad53
mutants to autophosphorylate was measured by using in vitro
kinase assays. No autophosphorylation was detected in
of autophosphorylation seen in wild-type extracts of MMS-
treated cells (Fig. 1C). The residual activity observed may reflect
the contribution of additional autophosphorylation sites in
Rad53 (12, 20). Purified GST-dun1-KD proteins were also used
as a substrate to assess the activity of rad53 mutant proteins (18).
No appreciable increase of Dun1 phosphorylation was seen in
rad53-TA or rad53-KD IPs, whereas 4.7-fold and 2.1-fold in-
creases were observed in the wild-type and rad53-T354A IPs,
respectively (Fig. 1C). When an artificial substrate, histone H1,
was used, significant phosphorylation of histone H1 by wild type
but not rad53-T354A was detected (data not shown).
Collectively, these data indicate that the rad53 autophosphor-
ylation mutants behave essentially as wild type with regard to
DNA damage-dependent protein interactions and that the in
vitro activities of rad53-TA and, to a lesser degree, rad53-T354A
are markedly reduced. These observations suggest that intracel-
lular disposition of the rad53 mutants is largely normal and that
the mutant proteins do not exhibit gross structural aberrations.
rad53-TA and rad53-T354A Exhibit Limited Activity in Vivo. To fully
address the significance of Rad53 autophosphorylation in check-
point signaling, we assessed the in vivo functionality of rad53-TA
and rad53-T354A in response to HU-, MMS-, and cdc13-induced
DNA damage. Response to each of these genotoxic stresses
requires Rad53, but involves distinct sensors and mediators (1,
2). We found that their in vivo functions were compromised, but
to a significantly lesser degree than rad53-KD. To assess the
response of Rad53 autophosphorylation mutants to DNA rep-
lication stress, rad53-TA, rad53-T354A, rad53-KD, and RAD53
were synchronized in G1with ?-factor and released into media
containing 200 mM HU. Viability was tested by colony forma-
tion on plates without HU after 0, 2 and 4 h of HU treatment.
After 4 h treatment, viability of rad53-KD and rad53-TA was
1.0%, whereas rad53-T354A was 2.5% (Fig. 2A), consistent with
rad53-T354A’s higher level of activity in vitro (Fig. 1C).
MMS sensitivity was also assessed. Asynchronous cultures of
rad53-TA, rad53-T354A, rad53-KD, and RAD53 were treated for
20 min with varying concentrations of MMS and plated to assess
viability by colony formation (Fig. 2B). rad53-T354A was 5-fold
more resistant, and rad53-TA was 3-fold more resistant than
rad53-KD at 0.9% MMS, indicating that both mutants retain
greater activity in vivo than rad53-KD.
rad53-TA and rad53-T354A were more proficient than
rad53-KD in the activation of the S/M checkpoint, which inhibits
mitosis when DNA replication is blocked. Tubulin was immu-
nostained to monitor spindle morphology 3 h after release from
G1arrest into HU-containing media. rad53-KD, rad53-TA, and
rad53-T354A cells, 29.3%, 22.7%, and 15.0%, respectively,
showed elongated spindle, indicative of mitosis, whereas 2.3% of
wild-type cells did (Fig. 2C). Thus, the S/M checkpoint defect in
The ability of rad53-TA and rad53-T354A to mount the G2/M
DNA damage cell-cycle checkpoint response was tested in the
cdc13–1 cdc15–2 (cdc13 cdc15) strain (Fig. 2D). In checkpoint-
proficient (i.e., RAD53) cdc13 cdc15 cells, cdc13 is inactivated at
the restrictive temperature, causing telomere damage and cell
cycle arrest at G2/M, whereas checkpoint mutants do not arrest
at G2/M but divide nuclei and arrest at telophase because of the
cdc15 (21, 22). As expected, upon shift to the restrictive tem-
perature after release from G1arrest, RAD53 cdc13 cdc15 cells
arrested at G2/M over the course of the experiment. rad53-KD
cells started dividing at least 1 h earlier than rad53-TA cells. At
6 h at the restrictive temperature, 50–60% of the rad53-KD and
rad53-TA cells had undergone nuclear division, whereas only
indicated strains. Asynchronous cells (?) and 90 min after 0.03% MMS treatment (?) were examined. (C) Rad53 kinase activity was examined in the indicated
strains. GST-dun1-KD was used as a substrate. (i)32P-Flag-Rad53, (ii) Flag (Rad53) WB, (iii)32P-GST-dun1-KD, and quantifications of Rad53 auto- (open bars) and
transphosphorylation (filled bars) activity are shown.
Examination of rad53-T354A and rad53-TA proteins. (A and B) Western blot (WB) analyses of input cell extracts and Rad9 (A) and Myc (B) IP’s from the
www.pnas.org?cgi?doi?10.1073?pnas.0611259104Usui and Petrini
20% of rad53-T354A had done so (Fig. 2D). These data indicated
that the autophosphorylation mutants retained in vivo function,
with the checkpoint proficiency of rad53-T354A greater than
rad53-TA or rad53-KD, whereas rad53-TA was more proficient
14-3-3 Palalogs BMH1 and BMH2 Are High-Copy Suppressors of rad53-
TA. The data above demonstrate that rad53-TA and rad53-
T354A retain residual activity, which we propose is conferred by
Mec1/Tel1 phosphorylation. To identify proteins that function-
suppressor screen for genes that could alleviate the HU sensi-
tivity of rad53-TA and rad53-T354A. The rad53-TA and rad53-
T354A strains were transformed with a high-copy yeast genomic
library, and transformants were tested for growth on 20 mM and
40 mM HU-containing plates, respectively (Table 1). Among the
suppressors obtained was the histone chaperone ASF1 (three
independent clones), previously identified as a high-copy sup-
pressor of rad53–21 HU sensitivity (23), and the WD40-
containing transcription modulator WTM2 (24) (one clone)
(Table 1). WTM2 was recently identified as a high-copy sup-
pressor of mec1? HU sensitivity and implicated in ribonucle-
otide reductase (RNR) regulation (25).
The most commonly obtained suppressors were S. cerevisiae
each), in the rad53-TA screen and in 10 independent clones (five
each) in the rad53-T354A screen (Table 1). The 14-3-3 proteins
are highly conserved eukaryotic proteins that regulate diverse
biological processes by governing the disposition of phosphor-
ylated proteins (26, 27). In S. cerevisiae, complete 14-3-3 defi-
ciency is lethal, but single bmh? mutants are unaffected (28).
The 14-3-3 proteins function in hetero- and homodimeric com-
plexes and have been implicated in DNA damage-dependent cell
cycle regulation in S. cerevisiae and S. pombe as well as in
The possibility that suppression was attributable to bypass of
effect is allele-specific. Whereas BMH1/2 overexpression in-
creased viability of rad53-TA at 20 mM HU, it had no effect on
the HU sensitivity of rad53-KD, rad53?, or rad53–21 (Fig. 3 A
and B). In contrast, rad53–21 HU sensitivity was suppressed by
WTM2 and ASF1 overexpression (Fig. 3B) (23).
Second, the strength of suppression was correlated with the
level of residual Rad53 activity in rad53-TA and rad53-T354A.
BMH2 could not suppress the lethality of rad53-TA in 40 mM
HU, whereas it did increase resistance of the more active
rad53-T354A allele (Fig. 3C).
Finally, BMH1/2 suppression was not restricted to HU sensi-
tivity. After 6-h incubation in 0.03% MMS, the viability of
rad53-TA cells carrying a BMH2 high-copy plasmid (0.1%) was
5-fold higher than that of rad53-TA cells carrying a WTM2
high-copy or empty plasmid (SI Fig. 8). The toxicity of HU, but
not MMS, arises from perturbation of nucleotide pools; it is
indirectly through an effect on nucleotide levels. Consistently,
sml1? did not suppress rad53-TA (data not shown, Fig. 4 A and
B). These data indicate that the mechanism of BMH1 and BMH2
suppression is direct and appears to be distinct from that of
WTM2 and ASF1.
Genetic Requirements of BMH2 Suppression. The specificity of
influence Rad53 functions in the normal physiological context
and predicts that the genetic requirements of suppression should
mirror those of Rad53 function itself. Activation of Rad53
depends on its phosphorylation by Mec1 and Tel1, an event that
requires the mediator proteins, Mrc1 and Rad9 in response to
DNA replication stress and DNA damage, respectively (5–7).
After 6-h treatment with 20 mM HU in liquid culture, BMH2
overexpression increased viability of rad53-TA, mrc1? rad53-TA,
sensitivity (A), MMS sensitivity (B), S/M (C), and G2/M cell cycle checkpoints (D)
of the indicated strains were examined. Error bars represent standard devia-
tion. (A) Cells arrested at G1by ?-factor were released into 200 mM HU and
accessed for viability. (B) Asynchronous cells were treated with the indicated
indicated cells that showed elongated spindle was determined by tubulin
staining when treated with HU for 3 h as in A. The numbers in parentheses
temperature (37°C) to examine the cdc13-induced G2/M DNA damage check-
point. Nuclear division was monitored by DAPI staining (?200 cells counted).
The y axis represented the percentage of cells that arrested at telophase
Phenotypic analyses of rad53-T354A and rad53-TA mutants. HU
Table 1. Summary of high-copy suppressor screenings
The numbers in parentheses represent the numbers of independent clones
obtained from the screenings.
rad53-TA, rad53-KD and rad53? transformed with single-copy RAD53, high-
copy BMH2, and empty plasmids on 0 mM and 20 mM HU plates was assessed.
Tenfold serial dilutions were plated. (B and C) Growth of rad53–21 (B),
mM or 40 mM HU plates was tested as in A.
BMH1/2 are high-copy suppressors of rad53-TA. (A) Growth of the
Usui and Petrini PNAS ?
February 20, 2007 ?
vol. 104 ?
no. 8 ?
and rad9? rad53-TA up to 8-, 9-, and 20-fold (Fig. 4A); however,
BMH2 suppression was blocked in mrc1? rad9? rad53-TA cells
(Fig. 4B), a context in which Rad53 phosphorylation by Mec1/
Tel1 is abrogated (7). These data suggest that Bmh2’s functional
interaction with Rad53 requires phosphorylation of Rad53 by
Whereas suppression was unaffected by Chk1 deficiency (data
not shown), BMH2 overexpression failed to suppress HU sen-
sitivity in Dun1-deficient rad53-TA and rad53-T354A strains
(Fig. 4C and SI Fig. 9). Suppression appeared to require physical
interaction between Rad53 and Dun1. The HU sensitivity of
dun1-S74A H77A (dun1-fha) rad53-TA cells, in which Rad53-
Dun1 interaction is abolished (18), was unaffected by BMH2
overexpression (Fig. 4C). These data indicate that BMH2 sup-
pression requires DNA damage-induced formation of the
The effect of BMH1/2 on clastogen sensitivity was correlated
with the mitigation of rad53-TA checkpoint defects. rad53-TA
cdc13 cdc15 cells were transformed with RAD53, BMH2, and
empty plasmids (Fig. 4D). As expected, upon shift to the
restrictive temperature, RAD53-transformed rad53-TA cells ar-
rested at G2/M over the course of the experiment, whereas
nuclear division was observed in 60% of control rad53-TA
transformants. In contrast, nuclear division was substantially
reduced in BMH2-overexpression rad53-TA. This suppression of
the G2/M checkpoint defect was lost in the dun1? rad53-TA
mutant (Fig. 4D). These data suggest that Bmh2 modulates the
functional interaction of Rad53 and Dun1 required for the G2/M
DNA-damage cell-cycle checkpoint.
Physical Interaction of Bmh1/2 with Rad53. 14-3-3 proteins prefer-
entially bind phosphopeptide (34–36). The apparent require-
ment for Mec1/Tel1 phosphorylation of Rad53 in BMH2 sup-
pression raised the possibility that Bmh2 interacted with Rad53
in response to DNA damage. To test this possibility, Bmh2 IPs
were carried out from extracts of MMS-treated Flag-Rad53-,
-rad53-TA-, -rad53-KD-expressing cells. All three forms of
Rad53 were present in Bmh1/2 immunoprecipitates from ex-
tracts of MMS-treated cells but not asynchronously growing
untreated cells (Fig. 5A), indicating that Bmh1/2 interacts with
Rad53 in response to DNA damage. The data also suggest that
the Bmh1/2-Rad53 interaction is not dependent on Rad53
The lack of suppression in mrc1? rad9? rad53-TA suggested
that the Bmh1/2–Rad53 interaction was influenced by Mec1/
Tel1-dependent phosphorylation. The interaction was therefore
examined in mec1 tel1? strains. Coprecipitated Rad53 was
slightly reduced in Bmh2 immunoprecipitates from MMS-
treated mec1 cell extracts, but was substantially reduced in mec1
tel1? cell extracts (Fig. 5B). Phosphatase treatment of Bmh2 IPs
liberated Rad53 from immunoprecipitate, confirming that the
interaction depends on phosphorylation (SI Fig. 10). Collec-
tively, these data favor the interpretation that Bmh2 engages
basally activated (i.e., Mec1/Tel1-phosphorylated, nonautophos-
phorylated) Rad53 in response to DNA damage and replication
Phosphopeptide-Binding Site bmh2 Mutants.Asecondpredictionof
the hypothesis that Bmh1/2 mediate Rad53 functions under
physiological conditions is that bmh2 mutants impaired for
Rad53 interaction would phenocopy Rad53 hypomorphism.
These mutants should also fail to suppress rad53-TA when
the two phosphopeptide-binding sites of Bmh2, K51 and E185.
K51 is critical for formation of the phosphopeptide-binding
pocket, and E185 determines the binding specificity of phos-
phopeptides to 14-3-3 (35, 36). We identified five bmh2K51x
mutants (x ? F, G, L, P, R) and a bmh2-E185D mutant that
exhibited reduced suppression of HU sensitivity relative to
additional binding-site mutant, bmh2-L225Q, was constructed
based on the bmh1–170 allele (L225Q E244V E262X) which
exhibited HU sensitivity and genomic instability (32). Mutations
that abolished dimerization (37) were neither able to suppress
were not analyzed further. Conversely, the phosphopeptide-
binding site bmh2 mutants retained sufficient functionality to
support the viability of a bmh1? mutant but exhibited HU and
MMS sensitivity when expressed from the BMH2 genomic locus,
(Fig. 6B and data not shown). All bmh2-K51x and bmh2-E185D
mutants suppressed rad53-TA ?10-fold less than BMH2 at 10
mM HU but failed to suppress at 20 mM HU, whereas bmh2-
L225Q did not suppress at all (Fig. 6A). These data suggest that
phosphopeptide binding is required for the suppression, consis-
tent with the observation that Rad53-Bmh2 interaction depends
on Rad53 phosphorylation.
Because Rad53 and Dun1 govern the G2/M DNA damage-
checkpoint defect (18, 38), we tested the prediction that the
nonbinding mutants would phenocopy Rad53 hypomorphism in
rad53-TA strains transformed with the high-copy BMH2 (squares) and empty
(triangles) vectors were incubated in 20 mM HU for 0, 3, and 6 h and accessed
for viability. The dashed lines were RAD53, BMH2, and empty plasmid trans-
formants of rad53-TA (A) and rad53-TA sml1? (B and C) from top. Note that
affect HU sensitivity of rad53-TA. (A) mrc1? rad53-TA (black symbols), and
dun1-fha rad53-TA sml1? (filled symbols). (D) G2/M DNA damage checkpoint
was assessed in rad53-TA or dun1? rad53-TA sml1? cdc13 cdc15 transformed
with the indicated plasmids as in Fig. 2D.
Genetic requirement for BMH2 suppression. (A–C) The indicated
Bmh2 IP (iv and v) from the indicated strains. WBs of Flag (Rad53) (i and iv),
tubulin (ii) and Bmh1/2 (iii and v) were shown. PI, preimmune.
www.pnas.org?cgi?doi?10.1073?pnas.0611259104 Usui and Petrini
that function. At 6 h, at the restrictive temperature after G1
arrest release, 25% of the bmh2-K51L and bmh2-K51R cells
underwent nuclear division in bmh1? cdc13 cdc15 (Fig. 6C),
approximately 3-fold more than wild-type cells.
Consistent with the phosphopeptide-binding functions of K51
and E185, the G2/M checkpoint defects observed were correlated
with reduced Rad53 interaction. Bmh2 IPs were performed in the
genomic phosphopeptide-binding site bmh2 mutants in the bmh1?
background. Rad53 was significantly reduced in IPs from MMS-
wild-type cells (Fig. 6D). These data strongly support the interpre-
tation that Bmh2-Rad53 interaction is important for Rad53-
dependent checkpoint functions and suggest that Bmh1/2 function
as mediators of Rad53 and Dun1 activity in vivo.
In this study, we examined the functional outcomes of Rad53
autophosphorylation and Rad53 phosphorylation by Mec1/Tel1.
rad53 mutations that impair autophosphorylation markedly reduce
in vitro activity; however they retain a significantly greater degree
of checkpoint function in vivo than rad53-KD mutants. These
observations support the view that Mec1/Tel1-phosphorylation of
damage-induced interactions of rad53-KD, rad53-TA, and rad53-
T354A with mediator proteins and the Rad53 substrate Dun1 were
unaffected. The experiments performed here do no exclude the
possibility that Rad53 may have inherent basal activity in vivo and
that the primary function of PIKK-mediated phosphorylation is to
promote physical interaction of the basally active kinase with other
effectors and mediators of the DNA damage response. In either
case, these associations culminate in Rad53 autophosphorylation
and its full activation.
The suppression of rad53-TA and rad53-T354A HU sensitivity
by BMH1/2 overexpression revealed a previously unrecognized
aspect of Rad53 regulation. Several lines of evidence support the
interpretation that Bmh1/2 facilitate Rad53 functions in vivo.
Bmh1/2 physically interact with Rad53 in a DNA damage-
dependent manner; accordingly, this interaction is promoted by
Mec1/Tel1 phosphorylation of Rad53. bmh2 mutants in which
physical interaction with Rad53 is lost fail to suppress rad53-TA
when overexpressed and phenocopy rad53 hypomorphism when
expressed normally in a RAD53 wild-type strain. Finally, the
genetic requirements for BMH1/2 suppression mirror those of
Rad53 activity itself.
The functional interaction of Rad53 and Bmh1/2 in S. cerevi-
siae demonstrates that 14-3-3 proteins can directly influence
checkpoint signaling. This finding contrasts with the mechanism
of 14-3-3 protein checkpoint functions in S. pombe and human
cells. In those settings, 14-3-3 proteins indirectly regulate Cdk1
which removes inhibitory phosphorylation of Y15 (33, 39–43),
although the different mode of 14-3-3 functions on Cdk1 regu-
lation has not been excluded (44). This event is initiated when
DNA damage induces Chk1- and Cds1-dependent phosphory-
lation of Cdc25, which creates a 14-3-3-binding site. The 14-3-3
then associates with Cdc25, leading to its sequestration in the
cytoplasm, effectively blocking its ability to activate Cdk1. In S.
cerevisiae checkpoint signaling, the Y19 in Cdc28 kinase (equiv-
alent to Y15 of Cdc2), does not appear to be the nexus of DNA
damage-dependent Cdk regulation (45, 46). Consistent with this
fact, we found that Mih1, the S. cerevisiae Cdc25 ortholog, is not
required for BMH2 suppression of rad53-TA (data not shown).
proteins in S. pombe and mammals during the response to DNA
damage (47–49). Given that DNA damage-induced protein inter-
actions of Rad53 autophosphorylation mutants are indistinguish-
able from wild-type Rad53, it is unlikely that Bmh1/2’s effect is
attributable to regulation of Rad53 localization. Chk1 deficiency
does not exacerbate the HU sensitivity of rad53-TA, nor does it
impair BMH2 suppression (data not shown), further supporting the
view that the mechanisms of 14-3-3 interaction with Rad53 are
distinct from those operative with Chk1.
The data presented here demonstrate that Bmh1/2 mediate
Rad53 kinase activity in a direct manner, and suggest an analogy
to the role of 14-3-3 proteins in the regulation of Raf-1 kinase
(37) (reviewed in refs. 27 and 50). Like Rad53, the activity of
Raf-1 kinase critically depends on phosphorylation in the acti-
vation loop (51). Available evidence suggests that 14-3-3 inter-
acts with Raf1 at four distinct phosphorylated residues. As some
of the interactions enhance and others inhibit its activity, the
14-3-3 engagement of Raf1 appears to stabilize either inactive or
active conformers of the kinase (50). We propose that the
binding of Bmh1/2 to Rad53 may similarly stabilize an active
form of Rad53. Implicitly, this hypothesis predicts that BMH2
overexpression may stabilize an active conformation in rad53-TA
and rad53-T354A, allowing access of Rad53 substrates (e.g.,
Dun1) to the catalytic site. Because BMH2 overexpression does
not suppress rad53–21, this E365K mutation (52) may preclude
the required conformational changes.
A small fraction of Rad53 DNA damage-induced interaction
with Bmh1/2 was Mec1/Tel1-independent, raising the possibility
that Rad53 may have two Bmh1/2-binding sites; one is the Mec1/
Tel1 phosphorylation site, and another is the Mec1/Tel1-
independent interaction site. The putative second site may depend
on a different kinase, but it is noteworthy that 14-3-3 protein
binding to nonphosphorylated peptides has been observed (e.g.,
refs. 53 and 54).
as in Fig. 2D. bmh2-K51L and K51R are in bmh1?. (D) WB analyses of input cell extracts and Bmh2 IP’s from the wild-type and indicated bmh2 mutant cells in
Analyses of phosphopeptide-binding site bmh2 mutants. (A) HU sensitivity of rad53-TA transformed with the high-copy BMH2 and indicated bmh2
Usui and Petrini PNAS ?
February 20, 2007 ?
vol. 104 ?
no. 8 ?