Notably, Ser258 phosphorylation of Bud32p does not alter the catalytic activity of the protein kinase per se, but positively regulates its ability to interact with Grx4p and thus to phos
Trang 1glutaredoxin by the Bud32p kinase unveils a novel
signaling pathway involving Sch9p, a yeast member
of the Akt / PKB subfamily
Caterina Peggion1,*, Raffaele Lopreiato1,*, Elena Casanova1, Maria Ruzzene1,2, Sonia Facchin1, Lorenzo A Pinna1,2, Giovanna Carignani1and Geppo Sartori1
1 Dipartimento di Chimica Biologica dell’Universita` di Padova, Italy
2 Venetian Institute of Molecular Medicine (VIMM), Padova, Italy
The Bud32p protein of Saccharomyces cerevisiae
belongs to the piD261 family of atypical Ser⁄ Thr
pro-tein kinases, which has representatives in virtually all
eukaryotic and archaeal organisms Unlike the
major-ity of eukaryotic protein kinases, the protein
preferen-tially recognizes acidic substrates [1–3] Several different approaches have shown that Bud32p is able
to interact with many yeast proteins [4–6] Particularly remarkable is its tight association with the still unchar-acterized Kae1p, as the two proteins make up a single
Keywords
Bud32p kinase; EKC ⁄ KEOPS complex;
Grx4p glutaredoxin; Sch9p kinase
Correspondence
G Sartori, Dipartimento di Chimica Biologica
dell’Universita` di Padova, viale G Colombo,
3-35121 Padova, Italy
Fax: +39 049 8073310
Tel: +39 049 8276141
E-mail: geppo.sartori@unipd.it
*These authors contributed equally to this
work
(Received 31 July 2008, revised
26 September 2008, accepted
1 October 2008)
doi:10.1111/j.1742-4658.2008.06721.x
The Saccharomyces cerevisiae atypical protein kinase Bud32p is a member
of the nuclear endopeptidase-like, kinase, chromatin-associated⁄ kinase, endopeptidase-like and other protein of small size (EKC⁄ KEOPS) complex, known to be involved in the control of transcription and telomere homeo-stasis Complex subunits (Pcc1p, Pcc2p, Cgi121p, Kae1p) represent, however, a small subset of the proteins able to interact with Bud32p, suggesting that this protein may be endowed with additional roles unre-lated to its participation in the EKC⁄ KEOPS complex In this context, we investigated the relationships between Bud32p and the nuclear glutaredoxin Grx4p, showing that it is actually a physiological substrate of the kinase and that Bud32p contributes to the full functionality of Grx4p in vivo We also show that this regulatory system is influenced by the phosphorylation
of Bud32p at Ser258, which is specifically mediated by the Sch9p kinase [yeast homolog of mammalian protein kinase B (Akt⁄ PKB)] Notably, Ser258 phosphorylation of Bud32p does not alter the catalytic activity of the protein kinase per se, but positively regulates its ability to interact with Grx4p and thus to phosphorylate it Interestingly, this novel signaling pathway represents a function of Bud32p that is independent from its role
in the EKC⁄ KEOPS complex, as the known functions of the complex in the regulation of transcription and telomere homeostasis are unaffected when the cascade is impaired A similar relationship has already been observed in humans between Akt⁄ PKB and p53-related protein kinase (Bud32p homolog), and could indicate that this pathway is conserved throughout evolution
Abbreviations
Akt ⁄ PKB, protein kinase B; EKC ⁄ KEOPS, endopeptidase-like, kinase, chromatin-associated ⁄ kinase, endopeptidase-like and other protein of small size; HA, hemagglutinin; PRPK, p53-related protein kinase; pSer, phosphorylated Ser; Ni-NTA, Ni 2+ -nitrilotriacetate–agarose.
Trang 2polypeptide in some archaeans, and their human
homologs are also able to interact [4] Two recent
papers have highlighted the importance of this
associa-tion, by describing a novel and highly conserved
pro-tein complex named endopeptidase-like, kinase,
chromatin-associated⁄ kinase, endopeptidase-like and
other protein of small size (EKC⁄ KEOPS) [7,8],
com-posed of Bud32p, Kae1p and three additional small
proteins, lacking a known biochemical signature
(Pcc1p, Pcc2p, Cgi121p) The EKC⁄ KEOPS complex
is essential for yeast viability, and is functionally
related to telomere homeostasis and transcription
con-trol, as its mutations cause transcriptional impairment
in the expression of specific gene groups, as well as
relevant shortening of telomeres Although the
molecu-lar mechanism of EKC⁄ KEOPS activity remains
elusive, it has been proposed that the complex might
promote the accessibility to chromatin, at telomeres as
well as elsewhere on the genome, and regulate the
recruitment of specific factors to their site of action
[7,8] Although the kinase activity of Bud32p is
rele-vant for the functions of the EKC⁄ KEOPS complex in
yeast, it is actually unknown whether
Bud32p-depen-dent phosphorylation of other subunits of the complex
could directly regulate its activity
In addition to the components of the EKC⁄ KEOPS
complex, many other proteins have been identified as
Bud32p interactors [4–6], suggesting that this protein
kinase may have additional roles by specifically
phos-phorylating other substrates Among these Bud32p
interactors, our attention has been drawn to the
glut-aredoxin Grx4p, which is an in vitro substrate of the
protein kinase, being readily phosphorylated by
recom-binant, purified Bud32p at Ser134 [4], suggesting that
Grx4p may be one of the physiological substrates of
Bud32p in yeast cells
Grx4p belongs to the subfamily of yeast monothiolic
glutaredoxins, together with Grx3p and Grx5p [9,10]
Whereas the function of Grx5p in mitochondrial Fe–S
cluster assembly has been extensively investigated, the
role of the nuclear glutaredoxins Grx3p and Grx4p is
less well characterized The single deletion of either
GRX3or GRX4 leads to weak growth defects, but the
double deletion strongly affects cellular growth and the
response to oxidative stress As the two proteins
dis-play relevant sequence similarity, they might have
overlapping or redundant functions Accordingly, it
has been shown that both Grx3p and Grx4p are
involved in the transcriptional modulation of the iron
regulon, by controlling the nucleo-cytoplasmic shuffle
of the transcriptional activator Aft1p [11–13]
In this work, we demonstrate that Grx4p is a
physio-logical substrate of Bud32p in yeast cells, and show
that this relationship is influenced by the phosphoryla-tion state of Bud32p In fact, Bud32, as well as its human homolog p53-related protein kinase (PRPK) [14,15], displays a highly conserved C-terminal sequence, rich in basic amino acids, that fulfils the con-sensus recognized by protein kinase B (Akt⁄ PKB) (RxxRxS⁄ THy) [16] Interestingly, the activity of PRPK on its known substrate (Ser15-p53) mainly (but not exclusively) depends on the phosphorylation of its Ser250 residue by Akt⁄ PKB [17] This prompted us to investigate whether the activity of Bud32p could also
be modulated by phosphorylation of its Ser258 residue, possibly mediated by Sch9p, which is considered to be
a yeast homolog of mammalian Akt⁄ PKB Sch9p is an AGC kinase [18] involved in a number of cellular pro-cesses, including the response to nutrient-mediated stimuli and the regulation of replicative and chronolog-ical lifespan [19–23] Recently, Sch9p has been identi-fied as a transcriptional activator that is recruited, only
in stress conditions, to the chromatin of genes induced
by osmotic stress [24] Sch9p is also regulated by TOR complex 1, which phosphorylates several amino acids situated in its C-terminal sequence [25] Recent data have implicated the PAS kinase Rim15p and the Rps6p protein as substrates of yeast Sch9p [25–28]
Here, we identify a novel phosphorylation cascade implicating Sch9p, Bud32p and Grx4p, which appar-ently does not affect the telomeric and transcriptional activities of the EKC⁄ KEOPS complex, suggesting an additional function for Bud32p in yeast cells
Results and Discussion
Grx4p is an in vivo substrate of Bud32p The characterization of Bud32p as a protein kinase was achieved by using a recombinant form of the enzyme, purified from Escherichia coli [1] Bud32p was shown to be able to autophosphorylate and to phos-phorylate in vitro the Ser⁄ Thr residues of acidic sub-strates, such as casein As is the case with many other protein kinases, autophosphorylation of Bud32p on its activation loop correlated with an increased activity on substrates, whereas all the mutant forms unable to autophosphorylate were also inactive on substrates [1– 3] A subsequent search for Bud32p-associated proteins
in yeast identified Grx4p as a Bud32p interactor, and the observation that recombinant Bud32p was able to phosphorylate recombinant Grx4p in vitro [4] sug-gested that Grx4p might be an in vivo substrate of the protein kinase
To verify this assumption, we constructed several yeast strains in which Bud32p (wild-type or mutant) is
Trang 3expressed from its chromosomal location in fusion
with the hemagglutinin (HA) epitope at the
C-termi-nus The two mutations analyzed here substitute
respectively Asp161 and Lys52, two amino acids that
are essential for the catalytic activity of the
recombi-nant protein [2] The two bud32 mutants are
character-ized by a slow growth phenotype that is, however, less
stringent than that exhibited by cells in which BUD32
is deleted (Fig 1A)
First, we checked the activity of Bud32p upon
immunoprecipitation in an in vitro assay on
recombi-nant, purified Grx4p Preliminary experiments,
per-formed in the conditions used in the biochemical
characterization of recombinant Bud32 (i.e in the
presence of Mn2+ as bivalent ion) [1–4], showed that
the kinase activity of immunoprecipitated Bud32p was
extremely low Conversely, the substitution of Mn2+
with Mg2+significantly improved the enzymatic
activ-ity of native Bud32p This behavior could be related to subtle differences between the structures of the recom-binant and the native protein kinase (e.g protein fold-ing and⁄ or post-translational modifications) Even in the presence of Mg2+, the catalytic efficiency detected for the native kinase remains low, the results, however, being consistent with those already reported for the recombinant protein [4] It is worth noting that very recent data [29] have indicated that the activity of Bud32p is inhibited by its functional partner Kae1p, suggesting that the protein(s) coprecipitating with Bud32p might reduce its activity also in our in vitro assays Altogether, native Bud32p displays specific kinase activity on recombinant Grx4p: in fact, the wild-type protein, which undergoes autophosphoryla-tion, is able to phosphorylate Grx4p, whereas the mutants D161A and K52A almost completely fail to autophosphorylate and exhibit a significantly lower activity on Grx4p (Fig 1B)
The partial phosphorylation of Grx4p still observed
in the case of the two mutants (corresponding to 30–40% of that of the wild-type) could be explained either by the intervention of another (contaminant) kinase, copurified with Bud32p, or, alternatively, by residual activity of the mutant proteins, which would still be able to catalyze the kinase reaction in the pres-ence of an excess of exogenous Grx4p substrate Together, these results demonstrate that native Bud32p
is able to phosphorylate Grx4p, further suggesting that the same relationship may exist in yeast cells
In a further approach, we investigated whether endogenous Grx4p was phosphorylated by Bud32p Native, myc-tagged Grx4p was immunoprecipitated from yeast, and the precipitate was analyzed for the presence of coprecipitated Bud32p The ability of Bud32p to interact in vivo with Grx4p has already been reported, although coimmunoprecipitation of the two proteins depends on the experimental conditions [4,7]
In our present experiments, a weak but specific signal was detected when the Grx4p immunoprecipitates were analyzed for the presence of wild-type Bud32p Interestingly, this signal was clearly stabilized in the presence of a catalytic mutant of Bud32p, such as D161A (Fig 2A) This observation is consistent with Grx4p being an in vivo substrate of Bud32p, based on the assumption that the transient interaction between the enzyme and its substrate is strengthened if the course of the catalytic reaction is hampered In accor-dance with this, the low amount of wild-type Bud32p present in the precipitate is still active on native Grx4p when subjected to in vitro phosphorylation, whereas the reaction is impaired in the case of coimmunoprecipitation of Grx4p with D161A mutant
A
B
Fig 1 (A) The kinase activity of Bud32p is relevant for its in vivo
functionality The wild-type [W303 and BUD32-HA (WT)], the bud32
mutants K52A-HA (K52A), D161A-HA (D161A) and the BUD32
deleted strain (bud32D) were grown until stationary phase in YPD
medium, and diluted to 3 · 10 7 cellsÆmL)1; 10-fold serial dilutions
were then spotted onto solid YPD medium Growth was observed
after 2 days at 28 C (B) Immunoprecipitated yeast endogenous
Bud32p autophosphorylates and phosphorylates recombinant Grx4p
in vitro HA-tagged Bud32p, wild-type or catalytically inactive, was
immunoprecipitated from lysates (500 lg of total protein) of strains
BUD32-HA (WT), D161A-HA (D161A) and K52A-HA (K52A) grown
in complete glucose medium (YPD) until exponential phase
Immu-nocomplexes were subjected to an in vitro phosphorylation reaction
in the presence of [ 33 P]ATP[cP] and 500 ng of recombinant, purified
Grx4p After SDS ⁄ PAGE, proteins were blotted onto a filter that
was autoradiographed (left panel) and then revealed with specific
antibodies for Bud32p or Grx4p (right panel).
Trang 4(Fig 2B) To be sure that the observed Grx4p
phos-phorylation was catalyzed by coimmunoprecipitated
Bud32p, we performed a similar experiment in a
bud32D⁄ GRX4-myc mutant strain; the result, shown in
Fig 2C, clearly indicates that Bud32p is responsible
for the radioactivity incorporated into Grx4p In this
assay, the phosphorylation reaction occurs by addition
of the [33P]ATP[cP] directly to the resin containing the
native, immunoprecipitated Grx4p and proteins
associ-ated with it Any phosphotransferase activity on
Grx4p therefore requires the presence in the
immuno-precipitate of (at least) one protein kinase Our results
confirmed the presence of Bud32p in the Grx4p
immu-noprecipitate, supporting the idea that Bud32p is
responsible (or coresponsible) for the kinase activity
observed on Grx4p Accordingly, in the case of the
bud32D⁄ GRX4-myc mutant strain, in which Bud32p is
lacking (Fig 2C), the phosphorylation of Grx4p
com-pletely disappeared However, the possibility cannot be
excluded that another, unidentified kinase may act on
Grx4p, but in this case it should be associated with
Bud32p rather than with Grx4p, as no activity was
detected when the immunoprecipitation was performed
in the absence of Bud32p Finally, we showed
(Fig 2D) that the phosphorylation state of Grx4p, as
revealed by antibody against phosphorylated serines
(pSer), was much lower when Grx4p was
immuno-precipitated from the bud32D mutant strain, as com-pared to the wild-type However, the detectable presence of pSer on Grx4p, even in the absence of Bud32p, indicates that the glutaredoxin is phosphory-lated also by other kinases in yeast cells Taken together, these results demonstrate that Bud32p partic-ipates in the in vivo phosphorylation status of Grx4p
Phosphorylation of Ser134 of Grx4p by Bud32p contributes to the functionality
of the glutaredoxin in yeast cells
As indicated by previous in vitro data, phosphorylation
of Grx4p by wild-type Bud32p occurs mainly at Ser134 and, more weakly, at Ser133, two residues embedded in a highly acidic stretch of the protein [4] This sequence is situated in the linker region between the thioredoxin-like and the glutaredoxin domains of Grx4p [10], and its modification would be likely to influence, directly or indirectly, the activity of the enzyme To evaluate the contribution of either Ser134
or Ser133 phosphorylation to the biological compe-tence of Grx4p in yeast cells, we created a series of unphosphorylatable mutants (S134A, S133A, S133A⁄ S134A), as well as the phospho-mimic S134D, in order
to compare their behavior with that of the wild-type sequence S cerevisiae, however, possesses another
Fig 2 Bud32p coimmunoprecipitates with and phosphorylates native Grx4p Myc-tagged Grx4p was immunoprecipitated from 500 lg of total protein from the following strains: GRX4-myc ⁄ BUD32-HA (WT), GRX4-myc ⁄ bud32-D161A-HA (D161A), BUD32-HA (no tag) and GRX4-myc ⁄ bud32D (bud32D), grown as in Fig 1 Immunocomplexes were either directly subjected to SDS ⁄ PAGE and immunoblotting (A, D)
or subjected to an in vitro phosphorylation reaction in the presence of [ 33 P]ATP[cP] and then to SDS ⁄ PAGE and western blotting (B, C) The band corresponding to Grx4p in (C) is indicated (*) In (A), the starting amounts of Bud32p present in the cell lysates were equivalent, as revealed by antibody against HA (Input) In (B), the amounts of Bud32p and Grx4p present in the kinase reaction after immunoprecipitation are revealed by antibody against HA or Myc, respectively.
Trang 5nuclear monothiolic glutaredoxin, Grx3p, which is very
similar to Grx4p; the two proteins cooperate and show
interchangeable roles, e.g in the transcriptional
regula-tion of iron-dependent genes [11–13] Therefore, to
specifically investigate in vivo the effect of mutating
Grx4p, we created the double null strain grx3D⁄ grx4D
Surprisingly, we noticed that, unlike what was
observed with other commonly used yeast strains (such
as BY4742 and CML128), cells containing the double
mutations are nonviable in the W303 genetic
back-ground (Fig S1), indicating that in the W303 strain
the functions of nuclear monothiolic glutaredoxins are
essential This may reflect the subtle differences
exist-ing between yeast laboratory strains, in particular with
regard to the responses to environmental changes or
stresses involving these oxidoreductases
The different GRX4 coding sequences were inserted
in galactose-inducible yeast plasmids (see Experimental
procedures), which were used to transform
hetero-zygous diploid grx3D⁄ grx4D cells After sporulation
and tetrad dissection, we isolated a complete set of
haploid grx3D⁄ grx4D strains containing wild-type or mutant GRX4 plasmids We then compared their growth in glucose medium, where the plasmidic alleles are weakly expressed, and observed (Fig 3A, left panel) that, in these conditions, wild-type GRX4 was able to fully restore yeast growth, similarly to the bona fide positive control (wild-type W303 cells carry-ing the empty plasmid) Although the substitution of Ser133 by Ala (S133A) did not affect the in vivo func-tionality of Grx4p, the mutation of Ser134 (S134A) slightly affected its function, this impairment, however, being completely restored by the phospho-mimic sub-stitution by Asp (S134D) Accordingly, the double mutation of Ser133 and Ser134 (SS-AA) showed the same effect as the single S134A mutation, confirming that Ser134 of Grx4p has a major role in vivo with respect to Ser133 In addition, we checked the effects
of Grx4p overexpression, by growing the yeast strains
in galactose medium, in which the expression of plas-mid-carried genes is strongly induced (Fig 3A, right panel) We observed that overexpression of wild-type
A
B
Fig 3 Ser134 phosphorylation of Grx4p
by Bud32p contributes to its functionality
in vivo (A) The wild-type W303 strain
carrying the empty plasmid and the mutant
strain grx4D ⁄ grx3D, carrying the plasmids
coding for either wild-type or mutant Grx4p
(S134A, S133A, SS-AA, S134D) were grown
until stationary phase in SD selective
med-ium and diluted at 3 · 10 7 cellsÆmL)1
Ten-fold serial dilutions were spotted either onto
solid SD (Glucose) or SG (Galactose) plates.
Growth was observed after 3 days at 28 C.
See text for details (B) Total protein lysates
(500 lg) of yeast cells expressing wild-type,
HA-tagged Bud32p (Bud32–HA) were used
to immunoprecipitate Bud32p as in Fig 1.
Immune complexes were subjected to an
in vitro phosphorylation reaction in the
pres-ence of [ 33 P]ATP[cP] and 25 ng of
recombi-nant wild-type Grx4p (WT) or 50 ng of the
Grx4p double mutant S133A ⁄ S134A
(SS-A-A) After SDS ⁄ PAGE and blotting, filters
were autoradiographed (upper panels), and
then visualized with specific antibodies
against Bud32p or Grx4p (lower panels) The
radiolabeled bands (*) are produced by an
unidentified contaminant of the purified
Grx4p proteins The strong signals in
western blots (**) are from the IgG light
chains released by the resin used for
Bud32–HA precipitation.
Trang 6Grx4p was toxic to yeast cells, whereas overexpression
of either the single S134A mutant or the double
S133A⁄ S134A mutant was less detrimental, indicating
that these substitutions somehow impaired the activity
of the glutaredoxin, rendering it less toxic to the cell
Remarkably, the phospho-mimic substitution S134D
was able to restore the toxicity of the glutaredoxin,
supporting the relevance of Ser134 phosphorylation
for the biological properties of Grx4p
Taken together, our data indicate that under normal
growth conditions, Ser134 phosphorylation may be
almost dispensable for Grx4p functionality, although it
could be relevant in the regulation of specific
path-way(s) upon environmental changes, allowing yeast
cells to respond appropriately to these stimuli
Finally, to evaluate the specific contribution of
Bud32p to Grx4p phosphorylation at Ser134-Ser133,
we used a mutant version of recombinant, His-tagged
Grx4p (S133A⁄ S134A) as substrate for an in vitro
phosphorylation reaction by native Bud32p
immuno-precitated from yeast cells Despite several purification
attempts, the recovery of wild-type and mutant Grx4p
was low, and the phosphorylation assays were run with
quantities of substrate consistently lower than in other
experiments (such as the one shown in Fig 1)
How-ever, as shown in Fig 3B, the results demonstrated
that Bud32p was able to phosphorylate the minimal
amount of wild-type Grx4p present in the reaction
(left, upper panel), whereas the phosphorylation of
mutant Grx4p completely failed to do so (right, upper
panel), despite the presence of a higher amount of
recombinant protein, as revealed by the western blot
(lower panels) Accordingly, a parallel experiment,
per-formed with the same Grx4p substrates and an aliquot
of recombinant Bud32p, showed that the recombinant
protein was also unable to phosphorylate the mutant
Grx4p (not shown), thus confirming that the
phos-phorylation of Grx4p by Bud32p specifically involves
Ser134
Activity of Bud32p on Grx4p is regulated through
phosphorylation by Sch9p
As recently demonstrated [17], the kinase activity of
PRPK (the human homolog of Bud32p) on its known
substrate Ser15-p53 is positively regulated in vitro and
in vivo by phosphorylation of Ser250, which is
specifi-cally mediated by Akt⁄ PKB The observation that
Bud32p at the homologous residue (Ser258) also
displays the consensus sequence for Akt⁄ PKB
(Rxx-RxS⁄ THy), and the existence in yeast of a functional
homolog of Akt⁄ PKB, the protein kinase Sch9p,
prompted us to look for the presence in yeast of a
similar enzyme–substrate relationship and to determine whether the activity of Bud32p could be modulated by Sch9p
Synthetic genetic interaction between BUD32 and SCH9
In order to find whether a functional relationship existed between Sch9p and Bud32p, we took advantage
of a genetic approach, easily carried out in yeast, look-ing for a possible genetic interaction between sch9 and bud32 mutants Figure 4A shows that the combination
of the SCH9 and BUD32 deletions is nonviable, and that deletion of SCH9, when combined with a bud32 catalytically inactive mutant (D161A), affects the growth of yeast cells more severely than each of the two single mutations (Fig 4B) These results supported the hypothesis that Sch9p and Bud32p are functionally related, and prompted us to examine their relationship
in depth
In vivo phosphorylation of Ser258 of Bud32p is strongly reduced in an sch9D mutant strain Phosphorylation of Bud32p at Ser258 can be identified
by the use of antibodies (anti-pSer258) that recognize the phosphorylated target site for Akt⁄ PKB present at the C-terminus of the protein [17] We first established that wild-type, His-tagged Bud32p, when overexpres-sed in yeast and purified by Ni2+–nitrilotriacetic acid agarose (NiNTA), is recognized by the antibodies, indicating that the protein is phosphorylated in vivo at Ser258 The antibodies are specific, as a similarly over-expressed and purified Bud32p mutant, in which the Ser258 residue is replaced with Ala (S258A), is recog-nized very weakly (Fig 5A)
Next, to ascertain whether phosphorylation of Ser258 of Bud32p is due to Sch9p, we immunoprecipi-tated endogenous HA-tagged Bud32p (expressed from its chromosomal location) from a wild-type strain (BUD32-HA) and from a mutant strain in which SCH9 had been deleted (sch9D⁄ BUD32-HA) Fig-ure 5B shows that Bud32p was recognized by the anti-bodies against pSer258 when immunoprecipitated from the wild-type strain, but not (or very poorly) when immunoprecipitated from the sch9D mutant, support-ing the idea that Sch9p is implicated in the phosphory-lation of Bud32p at Ser258
In order to investigate the impact of Ser258 substitu-tion in vivo, we compared the growth of the S258A mutant with that of the wild-type strain and of other bud32 mutants (D161A, K52A) and deletion mutants (bud32D), and observed that cells were almost
Trang 7unaffected when compared to catalytically inactive or
null mutants (not shown) Although we cannot
com-pletely rule out the possibility that Sch9p
phosphory-lates Bud32p also at other Ser⁄ Thr residues (which
would be embedded in sequences different from the consensus recognized by Akt⁄ PKB), we suppose that Ser258 phosphorylation could affect cell growth only in specific situations, and not in the (normal) conditions
Fig 5 Bud32p is phosphorylated at Ser258 in the wild-type but not in an sch9D mutant strain (A) Antibodies against pSer258 (anti-pSer258) recognize ectopically expressed wild-type Bud32p and not the S258A mutant The W303 wild-type strain, transformed with galactose-induc-ible vectors carrying the BUD32 sequence (WT or S258A mutant) fused to a His epitope, was grown in YPGal until exponential phase, when Bud32–His (WT or S258A) was purified with the NiNTA resin from the cell lysate The resin was subjected to SDS ⁄ PAGE and immuno-blotting (B) Endogenous HA-tagged Bud32p is phosphorylated at Ser258 when immunoprecipitated from the wild-type but not from an sch9D mutant strain The wild-type BUD32-HA strain (WT) and the sch9D ⁄ BUD32-HA mutant (sch9D), both expressing BUD32-HA from its chromosomal location, were grown in YPD medium (until exponential phase) Native Bud32p was immunoprecipitated from cell lysate with the anti-HA resin, and processed by SDS⁄ PAGE and immunoblotting.
A
B
Fig 4 Genetic interaction between BUD32 and SCH9 (A) The double deletion of BUD32 and SCH9 is lethal for yeast Heterozygotic diploid bud32D ⁄ sch9D cells were transformed with a centromeric plasmid (pFL38) carrying the wild-type BUD32 sequence and the URA3 marker [counterselectable on 5-fluoroorotic acid (5-FOA)-containing medium) After tetrad dissection, haploid spores were recovered and genotypes were determined Yeast cells containing the plasmidic URA3 marker and coming from some complete tetrads were plated on 5-FOA medium Only cells containing the double deletion bud32D ⁄ sch9D cannot lose the plasmid and cannot grow on 5-FOA plates Two represen-tative complete tetrads are shown DB, DS, DD are for bud32D, sch9D, and bud32D ⁄ sch9D, respectively (B) The slow-growth phenotype of the bud32-D161A mutant is exacerbated by the additional deletion of SCH9 Wild-type BUD32-HA (WT) and mutant strains D161A-HA (D161A), sch9D ⁄ BUD32-HA (sch9D) and sch9D ⁄ D161A-HA (sch9D ⁄ D161A) were grown until stationary phase in YPD medium, and diluted
to 3 · 10 7 cellsÆmL)1; 10-fold serial dilutions were then spotted onto solid YPD medium Growth was observed after 3 days at 28 C.
Trang 8here tested In this case, the S258A substitution should
not affect the main properties of the kinase, but may
operate as a regulatory site for Bud32p, e.g by
allow-ing⁄ promoting its association with specific partners (or
substrates)
However, the absence of a growth phenotype for
the S258A mutant does not explain the strong
genetic interaction observed between BUD32 and
SCH9 (see above and Fig 4); we must, then, infer
that the two genes, besides being interrelated via
phosphorylation of the Bud32p Ser258 residue by
Sch9p, have overlapping functions in a still
unidenti-fied pathway
Conditions that regulate Sch9p abundance
influence Bud32p phosphorylation at Ser258
The abundance of Sch9p, which is involved in the
con-trol of numerous nutrient-sensitive processes, is in fact
modulated by nutrients [21] To verify whether the
onset of conditions that modify the amount of Sch9p
has an effect on the phosphorylation of Bud32p at
Ser258, we compared the level of Bud32p
phosphoryla-tion in wild-type cells grown on a fermentable carbon
source (glucose) to that in cells grown on glycerol To
this end, we used a yeast strain expressing Bud32p and
Sch9p fused to different epitopes (SCH9-myc⁄
BUD32-HA), and first verified the abundance of Sch9p:
Fig 6A shows that the amount of Sch9p in the cell
lysates (as revealed by western blot) was higher in the
case of cells grown in glucose than in the case of cells
grown in glycerol Accordingly, phosphorylation at
Ser258 of Bud32p, immunoprecipitated from the same
lysates, is more extensive in cells grown in glucose than
in cells grown in glycerol (Fig 6B) These results
suggest that Bud32p might be a physiological target of
Sch9p, representing one of the effectors of this protein
kinase known to be involved in multiple cellular
processes
Sch9p interacts with Bud32p and phosphorylates
it in vitro at Ser258 The results described above are consistent with a role for Sch9p in the phosphorylation of Bud32p at Ser258, but do not prove that Bud32p is a direct substrate of Sch9p To clarify this point, we checked the ability of Sch9p to interact with Bud32p and phosphorylate it
in vitro
First, a pull-down experiment was performed by incubating recombinant, purified Bud32-His, previ-ously bound to the NiNTA resin, with a yeast cellular lysate in which Sch9p was HA-tagged The results shown in Fig 7A indicate that the two proteins are able to interact In fact, a western blot analysis revealed the presence of Sch9p associated with the NiNTA resin only where recombinant Bud32p had been previously bound to the resin The different bands recognized by the antibodies against HA are noteworthy, as they probably highlight different phos-phorylation states of Sch9p, as already demonstrated [21]
Next, we investigated whether Bud32p is a substrate
of Sch9p The HA-tagged Sch9p was immunoprecipi-tated by the use of the anti-HA affinity matrix: Fig 7Ba confirms the immunoprecipitation of Sch9p (lane 1); as controls, the anti-HA matrix was incubated either with a cellular lysate of the untagged strain, or with no lysate (lanes 2 and 3) The resin was subjected
to an in vitro phosphorylation reaction in the presence
of recombinant, purified Bud32-His In the autoradio-graph (Fig 7Bb), it can be seen that, when immuno-precipitated Sch9p was present (lane 1), radioactivity incorporation was greatly increased (2.5-fold) with respect to the background levels (lanes 2 and 3), which are due to autophosphorylation of recombinant Bud32-His Subsequent detection of the filter with anti-bodies against pSer258 (Fig 7Bc) confirmed that this phosphorylation takes place at Ser258
Fig 6 Phosphorylation at Ser258 of Bud32p is related to Sch9p abundance (A) Sch9p levels in cell lysates Equal amounts (20 lg of total protein) of cell lysates obtained from strain SCH9-myc ⁄ BUD32-HA, grown in YPD (glucose) or in YP with glycerol (Glycerol) until exponential phase, were subjected to SDS ⁄ PAGE and immunoblotting to visualize Sch9p (B) Phosphorylation state of immunoprecipitated Bud32p HA-tagged Bud32p was immunoprecipitated from the same lysates used in (A) (500 lg of total protein), using the anti-HA resin Bound proteins were eluted with SDS ⁄ PAGE loading buffer, electrophoresed, and immunoblotted with the indicated antibodies.
Trang 9Taken together, the in vivo and in vitro results on
Bud32p phosphorylation by Sch9p indicate that
Bud32p is one of the downstream targets of Sch9p,
whose substrates are still largely unknown, with few
exceptions, e.g the Rps6p ribosomal protein, indicated
in a recent report as a probable Sch9p substrate, at
least in vitro [25] Our data confirm that Sch9p
recog-nizes the same target sequence as human Akt⁄ PKB,
further supporting the assumption of the similarity
between the two protein kinases This raises the
ques-tion of whether phosphorylaques-tion of Bud32p at Ser258
represents a way to regulate its activity
Ser258 phosphorylation of Bud32p promotes
recognition and phosphorylation of Grx4p
We next compared the two forms of Bud32p
(phos-phorylated, or not, at Ser258) for their ability to
phosphorylate recombinant, purified Grx4p The
HA-tagged Bud32p was immunoprecipitated from the
wild-type strain BUD32-HA (Fig 8A, lane 1) and from the
sch9D mutant strain sch9D⁄ BUD32-HA (lane 3), as
well as from a mutant in which wild-type SCH9 is
present but Ser258 of Bud32p is replaced by Ala
(S258A-HA) (lane 2); the precipitates were then
sub-jected to an in vitro phosphorylation reaction in the
presence of recombinant Grx4p Figure 8A shows that, when Bud32p was not Ser258-phosphorylated, i.e in the sch9D strain, or when it carried the S258A muta-tion, phosphorylation of Grx4p was reduced by up to 40% of the wild-type activity, similarly to what was seen with the K52A and D161A mutants The observ-ation that Grx4p phosphorylobserv-ation was comparable in both mutant strains (bud32-S258A and sch9D, lanes 2 and 3, respectively) rules out the possibility of Grx4p being a direct substrate of Sch9p
Nevertheless, unlike with catalytic-defective mutants, autophosphorylation of immunoprecipitated Bud32p was not affected with respect to the wild-type, strongly suggesting that Ser258 modification of Bud32p does not alter the catalytic activity of the protein kinase per se We further confirmed such evidence by check-ing in vitro the enzymatic activity of native Bud32p on the model substrate casein, and observed that both forms of Bud32p, phosphorylated or not, had similar catalytic properties, as they were able to phosphorylate casein (data not shown) These results may therefore indicate that Ser258 modification could modulate the ability of Bud32p to recognize the Grx4p substrate
To confirm this hypothesis, we performed a pull-down experiment in which His-tagged wild-type Bud32p and mutant S258A were bound to the NiNTA
Fig 7 Sch9p is able to interact with Bud32p and phosphorylate it in vitro (A) Interaction between Sch9p and Bud32p by pull-down assay Five micrograms of recombinant, purified Bud32–His (lane 1), bound to the NiNTA resin, were incubated with 500 lg of a cellular lysate of strain SCH9-HA The resin was washed with a buffer containing imidazole, and subjected to SDS ⁄ PAGE and immunoblotting to detect Sch9p or Bud32p As a reference, the same amount of resin, with no Bud32–His bound, was incubated with the same lysate, and then treated as described and loaded in lane 2 Western blot analysis revealed that Sch9p was retained by the resin only in lane 1 (B) Immuno-precipitated Sch9p phosphorylates recombinant Bud32p HA-tagged Sch9p was immunoImmuno-precipitated with the anti-HA resin from 500 lg (total proteins) of a lysate of strain SCH9-HA (lane 1) As a reference, the anti-HA resin was incubated with a lysate of wild-type strain W303, in which SCH9 has no tag (lane 2), or with no lysate (lane 3) The immunoprecipitation of Sch9p only in lane 1 is revealed in (a) The resins were subjected to a phosphorylation reaction in the presence of [33P]ATP[cP] (b) or cold ATP (c) and 100 ng of recombinant His-tagged Bud32p [quantified by antibodies against His in panel (d)] After SDS ⁄ PAGE and immunoblotting, Bud32p phosphorylation was detected as radioactivity incorporation (b) and by anti-pSer258 (c).
Trang 10resin and analyzed for their ability to coprecipitate
Grx4p Yeast strains expressing native, myc-tagged
Grx4p were transformed with a plasmid bearing
wild-type BUD32, or the bud32-S258A mutant, both fused
to a His-tag After growth in galactose medium (to
induce the expression of the plasmid inserted genes),
cells were lysed and treated with the NiNTA resin in
order to isolate His-tagged Bud32p together with the
associated proteins In this experiment, we analyzed
the ability of ectopic Bud32p-His, phosporylated or
not at Ser258, to compete with endogenous Bud32p
for binding to native Grx4p Figure 8B shows that
wild-type His (lane 2) efficiently substituted for the
endogenous kinase in the association with Grx4p,
whereas a similar amount of the S258A mutant
(lane 3) failed to bind Grx4p, as shown by a signal
comparable to the background level (lane 1) The
(unexpected) high signal of the Grx4p background
indicates that NiNTA resin may be somehow able to
aspecifically bind myc-tagged Grx4p Taken together,
these results indicate that phosphorylation of Bud32p
at Ser258 positively regulates the ability of the protein
kinase to associate with Grx4p and therefore to
phos-phorylate it, whereas this modification does not affect
the catalytic activity of the protein kinase per se
Phosphorylation of Bud32p at Ser258 is unrelated
to its functions within the EKC/KEOPS complex
The observed phosphorylation of Bud32p at Ser258,
besides having an effect on Grx4p, might also influence
in yeast cells the activity of the whole EKC⁄ KEOPS
complex, of which Bud32p is a crucial component
Notably, Bud32p-Ser258 modification might impact on both functions (transcription control and telomere homeostasis), in which the EKC⁄ KEOPS complex is involved [7,8]
We therefore investigated whether phosphorylation
of Bud32p-Ser258 is linked to these processes, first by analyzing telomere length in several wild-type and mutant strains As shown in Fig 9A, catalytically inac-tive or null bud32 mutations (K52A; D161A; bud32D) led to telomeres that were shortened in comparison to the wild-type, whereas the telomere length of the S258A mutant was unaffected, being almost identical
to that of the wild-type (W303 or BUD32-HA) Remarkably, deletion of either SCH9 or GRX4 did not impair telomere elongation We have then examined the effects of Ser258 phosphorylation of Bud32p on the transcriptional activity of the EKC⁄ KEOPS com-plex by analyzing the activation rate of the galactose-inducible gene GAL1, known to be regulated by the complex By using real-time RT-PCR and northern blot analyses (a representative northern blot is shown
in Fig 9B), we compared the levels of GAL1 mRNA
in wild-type and bud32 mutant strains upon transcrip-tion inductranscrip-tion, observing (as expected) a reductranscrip-tion of mRNA levels in kinase-dead or null mutants, but no difference between the wild-type and the S258A mutant strains, in accordance with the effects observed
on telomere elongation The results presented here thus indicate that the phosphorylation cascade involving Sch9p, Bud32p and Grx4p is apparently not relevant
to the telomeric or to the transcriptional function
of the Bud32p-associated EKC⁄ KEOPS complex Accordingly, Grx4p has never been isolated as a
Fig 8 Ser258 phosphorylation of Bud32p influences its interaction with Grx4p (A) Yeast endogenous Bud32p phosphorylates the Grx4p substrate only if activated by Ser258 phosphorylation Native, HA-tagged Bud32p was immunoprecipitated from the wild-type BUD32-HA strain (lane 1), the S258A-HA mutant strain (lane 2) and the sch9 null strain (sch9D ⁄ BUD32-HA, lane 3), and subjected to an in vitro phos-phorylation reaction in the presence of recombinant Grx4p (B) Grx4p coprecipitates with ectopically expressed wild-type Bud32p and not with the S258A mutant The GRX4-myc ⁄ BUD32-HA strain was transformed with an empty centromeric plasmid (lane 1) and with the same plasmid carrying the wild-type (lane 2) or the S258A mutant (lane 3) forms of the BUD32 sequence, both fused to the His epitope Cells were grown until exponential phase in galactose medium to induce the ectopic expression of the plasmid-borne genes, and aliquots of the lysates containing equal amounts of Grx4p (as revealed by anti-myc detection, shown in Input) were incubated with the NiNTA resin to bind His-tagged Bud32p The resin was finally subjected to SDS ⁄ PAGE and western blotting.