Báo cáo khoa học: Analysis of the CK2-dependent phosphorylation of serine 13 in Cdc37 using a phospho-specific antibody and phospho-affinity gel electrophoresis doc

Using this antibody, we showed that complexes of Hsp90 with its client signaling kinases, Cdk4, MOK, v-Src, and Raf1, contained the CK2-phosphorylated form of Cdc37 in vivo.. Our analyse

serine 13 in Cdc37 using a phospho-specific antibody

and phospho-affinity gel electrophoresis

Yoshihiko Miyata and Eisuke Nishida

Department of Cell and Developmental Biology, Graduate School of Biostudies, Kyoto University, Japan

Protein kinases play pivotal roles in cellular signal

transduction systems Reversible protein

phosphoryla-tion is one of the major mechanisms used to control

the function, localization and stability of proteins

inside cells [1] Therefore, the analysis of protein kinase

activity and the phosphorylation level of their

sub-strates are important for understanding signal

trans-duction pathways at a molecular level Many methods

have been described for determining protein kinase activity and protein phosphorylation levels In vitro, phosphorylation reactions can be monitored by incu-bating a protein kinase and a substrate in the presence

of radioactive ATP ([32P]ATP[cP]), followed by SDS⁄ PAGE and autoradiography to quantify radio-activity in the substrate When a peptide substrate

is used, the amount of radioactivity incorporated into

Keywords

Cdc37; CK2; gel electrophoresis; Hsp90;

protein kinase

Correspondence

Y Miyata, Department of Cell &

Developmental Biology, Graduate School of

Biostudies, Kyoto University, Kitashirakawa

Oiwake-cho, Sakyo-ku, Kyoto 606-8502,

Japan

Fax: +81 75 753 4235

Tel: +81 75 753 4231

E-mail: ymiyata@lif.kyoto-u.ac.jp

(Received 22 March 2007, revised 9 August

2007, accepted 4 September 2007)

doi:10.1111/j.1742-4658.2007.06090.x

The CK2-dependent phosphorylation of Ser13 in cell division cycle pro-tein 37 (Cdc37), a kinase-specific heat shock propro-tein 90 (Hsp90) cochaper-one, has previously been reported to be essential for the association of Cdc37 with signaling protein kinases [Bandhakavi S, McCann RO, Hanna

DE & Glover CVC (2003) J Biol Chem 278, 2829–2836; Shao J, Prince T, Hartson SD & Matts RL (2003) J Biol Chem 278, 38117–38220; Miyata Y

& Nishida E (2004) Mol Cell Biol 24, 4065–4074] Here we describe a new phospho-specific antibody against Cdc37 that recognizes recombinant puri-fied Cdc37 only when incubated with CK2 in the presence of Mg2+ and ATP The replacement of Ser13 in Cdc37 by nonphosphorylatable amino acids abolished binding to this antibody The antibody was specific for phosphorylated Cdc37 and did not crossreact with other CK2 substrates such as Hsp90 and FK506-binding protein 52 Using this antibody, we showed that complexes of Hsp90 with its client signaling kinases, Cdk4, MOK, v-Src, and Raf1, contained the CK2-phosphorylated form of Cdc37

in vivo Immunofluorescent staining showed that Hsp90 and the phosphory-lated form of Cdc37 accumuphosphory-lated in epidermal growth factor-induced mem-brane ruffles We further characterized the phosphorylation of Cdc37 using phospho-affinity gel electrophoresis Our analyses demonstrated that the CK2-dependent phosphorylation of Cdc37 on Ser13 caused a specific gel mobility shift, and that Cdc37 in the complexes between Hsp90 and its cli-ent signaling protein kinases was in the phosphorylated form Our results show the physiological importance of CK2-dependent Cdc37 phosphoryla-tion and the usefulness of phospho-affinity gel electrophoresis in protein phosphorylation analysis

Abbreviations

Cdc37, cell division cycle protein 37; EGF, epidermal growth factor; ERK, extracellular signal-regulated kinase; FKBP52, FK506-binding protein 52; GST, glutathione S-transferase; HRP, horseradish peroxidase; Hsp90, heat shock protein 90; TBB,

4,5,6,7-tetrabromobenzotriazole.

the peptide can be quantified by scintillation counting

after separating the radioactive peptide from free

ATP⁄ ADP using phosphocellulose filters Protein

phosphorylation in vivo can be examined in several

ways The phosphorylation level of a substrate can be

determined by isolating the substrate from radiolabeled

cells or tissues by immunoprecipitation followed by

SDS⁄ PAGE and autoradiography For a

phosphoryla-tion site with a known sequence, it is possible to

obtain a phospho-specific antibody that reacts with the

substrate only in its phosphorylated form by

immuni-zation and affinity purification with the corresponding

phosphopeptide A phospho-specific antibody can then

be used to directly quantify the site-specific

phosphory-lation of a substrate in vivo by western blot analysis

Recently, MS has become a powerful technology for

large-scale detection and quantification of in vivo

pro-tein phosphorylation [2,3]

Although the precise molecular mechanism remains

to be elucidated, in some cases protein

phosphoryla-tion causes a mobility shift of the protein band on

SDS⁄ PAGE, often but not always decreasing the

mobility However, the mobility shifts are generally

not very large, and in most cases protein

phosphoryla-tion does not induce a mobility shift at all Moreover,

achieving optimal band shifts often requires special gel

compositions (such as a low concentration of

bis-acryl-amide), which can only be determined by somewhat

hit-and-miss experimentation A more reproducible

and reliable method for discriminating phosphorylated

and nonphosphorylated forms of a broad range of

proteins by gel electrophoresis has long been sought

Recently, Kinoshita et al identified alkoxide-bridged

dinuclear metal complexes as novel phosphate-binding

compounds that preferentially capture

phosphomono-ester dianions bound to serine, threonine and tyrosine

residues in proteins [4] They also reported that these

compounds could be used to separate phosphorylated

and unphosphorylated proteins in SDS⁄ PAGE [5]

Protein kinase activity in cells is regulated in many

different ways Releasing an inhibitory subunit from a

catalytic subunit can activate a kinase By contrast,

binding an activating regulatory coprotein to an

inac-tive catalytic subunit can activate a kinase In many

signal-transducing protein kinases, site-specific

phos-phorylation by an upstream protein kinase (a

kinase-kinase) activates them Before these activation steps,

the protein kinases must be in the correct structural

conformation, to be activated by the appropriate

stim-uli However, the activation-ready structures of

signal-ing protein kinases are relatively unstable in nature

and require additional proteins called ‘molecular

chap-erones’ to stabilize them within cells Among the

molecular chaperones, heat shock protein 90 (Hsp90) and cell division cycle protein 37 (Cdc37) have been shown to be specifically required for the stability and function of many signaling protein kinases, including Raf1 [6,7], Cdk4 [8–10], MOK [11], IKK [12], and v-Src [13] Hsp90 is an important molecular chaperone whose ATP-dependent function is essential for the folding and function of many signaling molecules, including protein kinases and steroid hormone recep-tors [14–16] Cdc37 both acts as a molecular chaperone

by itself and is also required for the efficient recruit-ment of Hsp90 to protein kinase complexes [17,18] Therefore, Cdc37 activity is crucial for many signaling protein kinases to function correctly in vivo

Protein kinase CK2 is a ubiquitous and highly con-served protein kinase that is known to be involved in many physiological functions by phosphorylating a plethora of substrates [19–21] CK2 is elevated in many types of tumor, and its overexpression is tumorigenic

in experimental models in animals, suggesting that CK2 is involved in both cell cycle control and neoplas-tic cell growth [22] Although CK2 was one of the ear-liest protein kinases identified, its regulatory mechanism remains largely unknown CK2 is com-posed of two catalytic subunits (a and⁄ or a¢) and two noncatalytic subunits (b) The catalytic subunits of CK2 are constitutively active, whether or not they are associated with noncatalytic b-subunits [23], CK2 activity is independent of any known second messen-gers, and no upstream ‘kinase-kinase’ has been identi-fied that activates CK2 [24]

We and others have previously identified Cdc37 as

a pivotal substrate for CK2 and reported that Cdc37 phosphorylation by CK2 is essential for Cdc37 to act

as a molecular chaperone for many signaling protein kinases [25–28] Moreover, the molecular chaperone functions of Hsp90 and Cdc37 are required for CK2 itself to be optimally active [25,29,30], suggesting that CK2 and Cdc37 together constitute a positive feed-back mechanism to control various signaling protein kinases [25,31] Therefore, analyzing the level of CK2-dependent Cdc37 phosphorylation in vivo should

be crucial for understanding the regulatory mecha-nisms of Cdc37, CK2, and other signaling protein kinases

In this report, we describe a new antibody that rec-ognizes Cdc37 only when phosphorylated by CK2 We also demonstrate that the phosphorylation of Cdc37 can be analyzed by phospho-affinity gel electrophore-sis Together, the phospho-specific antibody and phos-pho-affinity gel electrophoresis enabled us to directly study CK2-dependent Cdc37 phosphorylation in vitro and in vivo

Phospho-specific antibody against Cdc37

We synthesized the phosphopeptide

VWDHIEVpSD-DEDETHC, including amino acid residues 6–20 of

mammalian Cdc37, in which Ser13 was

phosphory-lated This sequence is in the most N-terminal region

of Cdc37, where the amino acid sequence conservation

between species is highest In fact, the amino acid

sequence of the peptide we used is identical in human,

chimpanzee, rhesus monkey, cattle, pig, rat, mouse

and chick sequences The serine at position 13 of

Cdc37, which was included in the synthetic peptide,

has been reported to be phosphorylated by CK2 and

important for the functional regulation of Cdc37 [25–

28] We raised rabbit antiserum against this

phospho-peptide and examined the specificity of the purified

antibody Purified recombinant Cdc37 was incubated

with the recombinant catalytic subunit of CK2 (CK2a)

or with CK2 holoenzyme (CK2a2b2) or alone, in the

presence or absence of Mg2+ and⁄ or ATP As shown

in Fig 1A, the antibody (anti-[pSer13]-Cdc37

hereaf-ter) recognized Cdc37 only when incubated with either

CK2a or CK2 holoenzyme in the presence of ATP and

Mg2+ (lanes 6 and 9) The absence of either CK2

(lane 3), Mg2+ (lanes 4 and 7) or ATP (lanes 5 and 8)

completely abolished antibody binding (Fig 1A),

indi-cating that anti-[pSer13]-Cdc37 specifically recognized

the CK2-phosphorylated form of Cdc37 The presence

of equal amounts of Cdc37 in the phosphorylation mixtures was checked by probing western blots with

an antibody to Cdc37 (Fig 1B)

The binding of anti-[pSer13]-Cdc37 was completely abolished when Ser13 in Cdc37 was replaced by a non-phosphorylatable amino acid Recombinant wild-type Cdc37 and two Cdc37 mutants in which Ser13 was replaced by alanine [Cdc37(13SA)] or aspartic acid [Cdc37(13SD)] were incubated with CK2a or CK2 holo-enzyme or alone, in the presence of Mg2+–ATP Wild-type Cdc37 bound anti-[pSer13]-Cdc37 in the presence

of CK2a or CK2 holoenzyme (Fig 1C, lanes 4 and 7)

By contrast, neither Cdc37(13SA) nor Cdc37(13SD) were recognized by anti-[pSer13]-Cdc37, even after incu-bation with CK2a or CK2 holoenzyme (Fig 1C, lanes 5, 6, 8 and 9) The amounts of Cdc37(WT), Cdc37(13SA) and Cdc37(13SD) in all the incubation mixtures were approximately the same, as shown by Coomassie brilliant blue (CBB) staining (Fig 1D) These results showed that anti-[pSer13]-Cdc37 recog-nized Cdc37 only when Ser13 was phosphorylated by CK2, and that isolated CK2a phosphorylated the same site in Cdc37 (Ser13) as purified CK2 holoenzyme

We next examined the specificity of anti-[pSer13]-Cdc37 for anti-[pSer13]-Cdc37 in comparison to other CK2-phos-phorylated proteins CK2 phosphorylates serine or threonine residues followed by a stretch of acidic amino acids [19,21], which constitutes a consensus

Fig 1 Phospho-specific antibody against an Hsp90 cochaperone, Cdc37 (A, B) Recombinant Cdc37 was incubated at 30 C for 30 min alone (lanes 1–3), with CK2a (lanes 4–6), or with purified CK2 holoenzyme (lanes 7–9), in the presence (+) or absence (–) of Mg 2+ and ⁄ or ATP as indicated above the track Western blots of these mixtures with anti-[pSer13]-Cdc37 (A) or with anti-Cdc37 (B) are shown The posi-tions of molecular weight markers and Cdc37 are shown (C, D) Wild-type protein [Cdc37(WT), lanes 1, 4 and 7] as well as two mutant pro-teins [Cdc37(13SA), lanes 2, 5 and 8, and Cdc37(13SD), lanes 3, 6 and 9] were incubated at 30 C for 30 min alone (lanes 1–3), with CK2a (lanes 4–6) or with CK2 holoenzyme (lanes 7–9), and the mixtures were analyzed by western blotting with anti-[pSer13]-Cdc37 (C) CBB staining is shown in (D) The positions of molecular weight markers and Cdc37 are shown.

phosphorylation sequence for CK2 Thus, the amino

acid sequences surrounding CK2 phosphorylation sites

are similar in most CK2 substrates We examined two

other molecular chaperones, Hsp90 and

FK506-bind-ing protein 52 (FKBP52), which are also known to be

phosphorylated by CK2, to find whether they were

rec-ognized by anti-[pSer13]-Cdc37 when phosphorylated

In fact, the amino acid sequences around the known

CK2 phosphorylation sites in Cdc37 (Ser13) [26,28],

FKBP52 (Thr143) [32], and Hsp90 (Ser231 and

Ser263) [33] are highly homologous (Fig 2A) Purified

recombinant FKBP52 and Hsp90 were incubated with

or without CK2 in the presence of Mg2+–

[32P]ATP[cP] Cdc37(WT) and Cdc37(13SA) were

included as positive and negative controls Analysis of

the phosphorylation mixtures by SDS⁄ PAGE and

autoradiography clearly showed that Cdc37(WT),

FKBP52 and Hsp90 were heavily phosphorylated by

CK2 in vitro (Fig 2B) Western blot analysis of the

same phosphorylation mixtures probed with

anti-[pSer13]-Cdc37 showed that the antibody only

recog-nized CK2-phosphorylated Cdc37 (Fig 2C, lane 5),

and not CK2-phosphorylated FKBP52 (Fig 2C,

lane 7) or CK2-phosphorylated Hsp90 (Fig 2C,

lane 8) The presence of equal amounts of Cdc37,

FKBP52 and Hsp90 in the phosphorylation mixtures

were checked by CBB staining (Fig 2D) These results

indicated that anti-[pSer13]-Cdc37 specifically

recog-nized the CK2-phosphorylated form of Cdc37, and did

not crossreact with a generic CK2 phosphorylation

consensus sequence

Phosphorylated Cdc37 associates with signaling

protein kinases

We previously reported that replacing Ser13 in Cdc37

with a nonphosphorylatable amino acid abolished the

binding of Cdc37 to Hsp90 client signaling protein

kinases [26,27] Using the phospho-specific antibody

to Cdc37, we next investigated whether the CK2-phos-phorylated form of Cdc37 associated with signaling protein kinases in vivo Four typical Hsp90 client kin-ases, Cdk4, MOK, v-Src, and Raf1, were expressed

in COS7 cells as FLAG-tagged proteins, and kinase– Hsp90–Cdc37 complexes were immunopurified on anti-FLAG affinity resin As controls, two nonclient kinases for Hsp90, CK1 and DYRK2, were included

A

B

C

D

Fig 2 Substrate specificity of the antibody against

CK2-phosphory-lated Cdc37 (A) Alignment of the amino acid sequences

surround-ing the CK2 phosphorylation sites of Cdc37 (rat), FKBK52 (rabbit),

and Hsp90 (human) The position of the CK2-catalyzed

phosphoryla-tion sites is indicated by an arrow (B) Recombinant purified

Cdc37(WT) (lanes 1 and 5), Cdc37(13SA) (lanes 2 and 6), FKBP52

(lanes 3 and 7) or Hsp90 (lanes 4 and 8) was incubated alone

(lanes 1–4) or with purified CK2 (lanes 5–8) in the presence of

Mg 2+ -[ 32 P]ATP[cP], and phosphorylated proteins were visualized by

autoradiography after SDS ⁄ PAGE (C) The same protein mixtures as

shown in (B) were analyzed by western blotting with

anti-[pSer13]-Cdc37 (D) The same protein mixtures shown in (B) and (C) were

stained with CBB after SDS ⁄ PAGE The positions of molecular

weight markers, as well as Hsp90, FKBP52, and Cdc37, are shown.

The amounts of Hsp90, total Cdc37,

CK2-phosphory-lated Cdc37 and immunoprecipitated kinases were

examined by western blot analysis, using the

corre-sponding antibodies As shown in Fig 3, each kinase,

Cdk4 (lane 4), MOK (lane 5), v-Src (lane 6), and Raf1

(lane 7), associated specifically with Hsp90 (Fig 3A)

and with Cdc37 (Fig 3B) Importantly,

anti-[pSer13]-Cdc37 recognized anti-[pSer13]-Cdc37 in all Hsp90–client protein

kinase complexes (Fig 3C), indicating that

CK2-phos-phorylated Cdc37 was present in these kinase

com-plexes The control kinases, CK1 (lane 2) and DYRK2

(lane 3), did not bind to Hsp90 (Fig 3A), Cdc37

(Fig 3B), or phospho-Cdc37 (Fig 3C), although

com-parable amounts of protein were immunoprecipitated

for each kinase, as shown on western blots obtained

using anti-FLAG (Fig 3D)

Intracellular distribution of CK2-phosphorylated

Cdc37

As shown above, CK2-phosphorylated Cdc37

associ-ates with signaling protein kinases, so intracellular

regions where signaling protein kinases accumulate

might also be expected to contain high concentrations

of phosphorylated Cdc37 Growth factors such as

epi-dermal growth factor (EGF) and insulin-like growth

factor-I are known to induce membrane ruffling, and

actin cytoskeleton and signaling molecules, such as

Rho family G-proteins and protein kinases, are known

to accumulate in these areas We therefore examined

the intracellular distribution of CK2-phosphorylated

Cdc37 in EGF-stimulated KB cells, a cell line known

to show prominent membrane ruffling in response to

growth factors [34] KB cells were serum starved by

incubating them in medium containing only 1% fetal

bovine serum for 4 h and then incubated with or

with-out 30 nm EGF for 5 min Anti-Hsp90, anti-Cdc37

and anti-[pSer13]-Cdc37 were used to

immunofluores-cently stain the cells, to examine the intracellular

distri-butions of the proteins Before EGF stimulation,

Hsp90 was mainly localized in the cytoplasm, whereas

Cdc37 and phospho-Cdc37 were present throughout

the KB cells, both in the cytoplasm and in the nucleus

(Fig 4A–C) EGF induced membrane ruffling and

Hsp90 was localized to the membrane ruffles (Fig 4D,

arrowheads) as previously reported [35] Cdc37 also

accumulated in areas of membrane ruffling in

EGF-stimulated KB cells (Fig 4E, arrowheads), as did

CK2-phosphorylated Cdc37 (Fig 4F, arrowheads)

These results indicated that phosphorylated Cdc37

colocalized with Hsp90 in the growth factor-induced

membrane ruffles, where signaling protein kinases also

accumulated, and are consistent with the results in

Fig 3, showing that phosphorylated Cdc37 forms complexes with Hsp90 client signaling protein kinases

We next investigated whether the accumulation of phosphorylated Cdc37 in the EGF-induced membrane ruffles was a result of an increase in the phosphoryla-tion of Cdc37 by activated CK2 Serum-starved KB cells were treated with EGF for up to 60 min, and the levels of phosphorylated Cdc37 in total cell extracts were determined by western blotting using

A

B

C

D

Fig 3 Association of Ser13-phosphorylated Cdc37 with various Hsp90 client protein kinases COS7 cells were transfected with empty vector DNA (lane 1, control), or plasmids encoding FLAG-tagged protein kinases CK1 (lane 2, control), DYRK2 (lane 3, con-trol), Cdk4 (lane 4), MOK (lane 5), v-Src (lane 6), or Raf1 (lane 7), and the kinase–chaperone complexes were immunopurified The amounts of Hsp90 (A), Cdc37 (B), phosphorylated Cdc37 (C) and protein kinase (D) in the kinase–chaperone complexes were assessed by western blotting with Hsp90, Cdc37, anti-[pSer13]-Cdc37, and anti-FLAG, respectively The positions of molecular weight markers are shown.

anti-[pSer13]-Cdc37 (Fig 5A) The levels of total

Cdc37 (Fig 5B), phosphorylated extracellular

signal-regulated kinase (ERK) (Fig 5C) and total ERK

(Fig 5D) were also measured by western blotting using

anti-Cdc37, anti-[pTEpY]-ERK (an antibody specific

for the dually phosphorylated, activated form of

ERK), and anti-ERK, respectively The results clearly

showed that the levels of phosphorylated Cdc37 as well

as total Cdc37 did not change after EGF stimulation

(Fig 5A,B) The activation of EGF-induced signaling

pathways under these conditions was confirmed by the

observation that EGF rapidly stimulated the dual

phosphorylation of ERK (Fig 5C) These results agree

with an earlier report showing that growth factors did

not significantly activate CK2 in cells [36] In addition,

we noted that anti-[pSer13]-Cdc37 recognized a single

protein band in western blots of whole cell lysates

(Fig 5A), reinforcing our conclusion that

anti-[pSer13]-Cdc37 specifically recognizes phospho-Cdc37

but not other CK2 substrates (Fig 2) Taking these

findings together, we concluded that

CK2-phosphory-lated Cdc37 accumulates in EGF-induced membrane

ruffles as a result of intracellular redistribution of

phospho-Cdc37 rather than a net increase in the

phos-phorylation of Cdc37 by CK2 within the cell

Analysis of Cdc37 phosphorylation by phospho-affinity gel electrophoresis

We wanted to establish a simple biochemical method for analyzing the phosphorylation of Cdc37 To this end, we investigated separating phosphorylated Cdc37 from nonphosphorylated Cdc37 by phospho-affinity gel electrophoresis, a technique recently developed by Kinoshita et al [5] Purified recombinant Cdc37 was incubated with CK2a or CK2 holoenzyme or alone, in the presence or absence of ATP and⁄ or Mg2+, and the mixtures were analyzed by phospho-affinity gel electro-phoresis CBB staining of the phospho-affinity gel showed that the mobility of Cdc37 was markedly decreased when it had been incubated with CK2a or CK2 holoenzyme in the presence of both ATP and

Mg2+(Fig 6A, lanes 6 and 9) This mobility shift was not a nonspecific effect due to ATP and⁄ or Mg2+, because no shift was observed in the absence of CK2 (Fig 6A, lanes 1–3) Nor was the mobility shift caused

by the physical association of CK2 and Cdc37, as CK2 did not induce the Cdc37 mobility shift in the absence of ATP or Mg2+ (Fig 6A, lanes 4, 5, 7 and 8) The CK2a- or CK2 holoenzyme-dependent phos-phorylation of Cdc37 in the presence of Mg2+–ATP

Fig 4 Subcellular localization of phosphorylated Cdc37 in growth factor-induced membrane ruffles KB cells incubated in low-strength serum (1%) were untreated (A–C) or treated with 30 n M EGF for 5 min (D–F) The intracellular localization of Hsp90 (A, D), total Cdc37 (B, E) or the phosphorylated form of Cdc37 (C, F) are shown in the left columns by immunofluorescent microscopy Corresponding phase contrast images are shown in the right columns For each panel, two typical images (top and bottom) from different fields are shown.

was confirmed by labeling with anti-[pSer13]-Cdc37

(Fig 1A) It should be noted that no band shift was

detected when the protein mixtures were analyzed by

normal SDS⁄ PAGE (Fig 1B,D)

We then examined the effect of Cdc37 mutations at

the CK2 phosphorylation site on phospho-affinity gel

electrophoresis mobility Recombinant purified

wild-type Cdc37 protein or Ser13 mutants of Cdc37,

Cdc37(13SA) and Cdc37(13SD) were incubated with CK2a or CK2 holoenzyme in the presence of

Mg2+–ATP and analyzed by phospho-affinity gel electrophoresis In contrast to the CK2a- or CK2 holoenzyme-dependent mobility shift shown by wild-type Cdc37 (Fig 6B, lanes 4 and 7), the mobilities of Cdc37(13SA) and Cdc37(13SD) were not affected by CK2a or CK2 (Fig 6B, lanes 5, 6, 8 and 9), indicating that the mobility shift of Cdc37 in phospho-affinity gel electrophoresis was caused by its CK2-dependent phos-phorylation on Ser13 In addition, these results showed that CK2a phosphorylated Cdc37 only on Ser13, as in the case of CK2 holoenzyme, even in the absence of CK2b We therefore used CK2a to phosphorylate Cdc37 in vitro in subsequent experiments

Phospho-affinity gel electrophoresis can be combined with other detection systems, including autoradiogra-phy and western blotting Cdc37 was incubated with

or without CK2a in the presence of Mg2+–ATP, and separated by phospho-affinity gels The phospho-affin-ity gels were washed with a buffer containing EDTA

to remove Mn2+, and thereby remove the phosphate-binding activity of Phos-tag, so that the phosphopro-teins in the gel could be transferred to a membrane The membrane was probed with Cdc37 or anti-[pSer13]-Cdc37 Anti-Cdc37 labeled both the low-mobility and high-low-mobility Cdc37 bands (Fig 7A, lanes 1 and 2), whereas anti-[pSer13]-Cdc37 recognized only the low-mobility Cdc37 band (Fig 7B, lane 2), indicating that the band with the decreased mobility represented Cdc37 in the Ser13-phosphorylated form When we carried out the phosphorylation reaction in the presence of [32P]ATP[cP], analyzed the mixtures by phospho-affinity gel electrophoresis and autoradio-graphed the gels, radioactivity was detected only in the low-mobility band using wild-type Cdc37 (Fig 7C, lane 3) Neither the mobility shift nor the radioactivity could be detected when CK2a was omitted (Fig 7C,D, lane 2) or when the phosphorylation site mutant Cdc37(13SA) was used (Fig 7C,D, lane 4) We there-fore concluded that CK2 phosphorylates only one residue, Ser13, in Cdc37, and that this single phos-phorylation causes the Cdc37 mobility shift seen in phospho-affinity gel electrophoresis

Analysis of Cdc37 phosphorylation and dephosphorylation by phospho-affinity gel electrophoresis

To analyze the time course of Cdc37 phosphorylation

in vitro, mixtures of recombinant Cdc37 and CK2a, in the presence of Mg2+–ATP, were sampled at different time points, and the samples were analyzed by

A

B

C

D

EGF

WB: Anti-[pSer13]-Cdc37

WB: Anti-[pTEpY]-ERK

WB: Anti-ERK WB: Anti-Cdc37

min 62

47

33

1 2 3 4 5

1 2 3 4 5

1 2 3 4 5

1 2 3 4 5

62

47

33

62

47

33

62

47

33

Fig 5 Effect of EGF treatment on the phosphorylation of

Cdc37 KB cells incubated in low-strength serum (1%) were

untreated (lane 1) or treated with 30 n M EGF for 5 min (lane 2),

15 min (lane 3), 30 min (lane 4), or 60 min (lane 5) Cell extracts

were prepared, and phosphorylated Cdc37 (A), total Cdc37 (B),

dually phosphorylated and activated ERK (C) and total ERK

(D) were labeled on western blots with the corresponding

antibodies.

phospho-affinity gel electrophoresis Phosphorylation

was rapid and detected within 2 min (Fig 8A, lane 4)

and completed within 6 min at 30C (Fig 8A,

lanes 4–7) The transition in mobility from the higher

to lower bands was direct, with no bands of intermedi-ate mobility being detected, supporting our previous conclusion that CK2a phosphorylates only one site in Cdc37

We then studied the time course of dephosphoryla-tion by incubating CK2a-phosphorylated Cdc37 (Fig 8B, lane 2) with (Fig 8B, lanes 7–10) or without (Fig 8B, lanes 3–6) k-phosphatase in the presence of the CK2 inhibitor 4,5,6,7-tetrabromobenzotriazole (TBB) and analyzing the products by phospho-affinity gel electrophoresis and western blotting using anti-Cdc37 Incubating phospho-Cdc37 with k-phosphatase induced rapid dephosphorylation, with a high-mobility Cdc37 band appearing within 3 min (Fig 8B, lane 8) Again, the band shift induced was direct, from the low-mobility to high-mobility bands, with no inter-mediate bands being detected

Analysis of signaling kinase–Hsp90–Cdc37 com-plexes by phospho-affinity gel electrophoresis Finally, we analyzed complexes between signaling pro-tein kinases and Hsp90–Cdc37 molecular chaperones

by phospho-affinity gel electrophoresis The Hsp90 cli-ent kinases Cdk4, MOK, v-Src and Raf1 were expressed as FLAG-tagged fusion proteins in COS7 cells, and the kinase–chaperone complexes were immu-nopurified using anti-FLAG agarose The immuno-complexes were separated by phospho-affinity gel electrophoresis, transferred to western blots, and labeled with anti-Cdc37 or anti-[pSer13]-Cdc37 Spe-cific associations between Cdc37 and Cdk4, MOK, v-Src and Raf1 were observed (Fig 9A), consistent with the result shown in Fig 3 Interestingly, Cdc37 in the protein kinase complexes was detected as a single band in phospho-affinity gels (Fig 9A), and the band was recognized by anti-[pSer13]-Cdc37 (Fig 9B), indicating for the first time that all of the Cdc37 in the signaling kinase–Hsp90 complexes was in its Ser13-phosphorylated form in vivo

Discussion

In this study, we have produced a phospho-specific antibody against Cdc37, which recognized recombinant purified Cdc37 only when incubated with CK2 in the presence of Mg2+ and ATP The specificity of this antibody was demonstrated by showing that it did not recognize mutant Cdc37 in which the CK2 phosphory-lation site, Ser13, had been replaced with nonphosph-orylatable amino acids, and it did not recognize other CK2-phosphorylated proteins such as Hsp90 and

A

B

Fig 6 Analysis of Cdc37 phosphorylation by phospho-affinity gel

electrophoresis (A) Recombinant Cdc37 alone (lanes 1–3), with

CK2a (lanes 4–6) or with CK2 holoenzyme (lanes 7–9), in the

pres-ence (+) or abspres-ence (–) of Mg 2+ and ⁄ or ATP as indicated, was

incu-bated at 30 C for 30 min, analyzed by phospho-affinity gel

electrophoresis, and then stained with CBB (B) Wild-type protein

(lanes 1, 4 and 7) and two Cdc37 mutant proteins, 13SA (lanes 2, 5

and 8) and 13SD (lanes 3, 6 and 9), were incubated alone

(lanes 1–3), with CK2a (lanes 4–6) or with CK2 holoenzyme

(lanes 7–9) in the presence of Mg2+–ATP for 30 min at 30 C The

phosphorylation mixtures were analyzed by phospho-affinity gel

electrophoresis and stained with CBB.

FKBP52 Thus, the antibody specifically recognizes

Cdc37 only when it is phosphorylated on Ser13 by

CK2 Using this antibody, we have shown that

com-plexes between Hsp90 and its client signaling kinases

Cdk4, MOK, v-Src and Raf1 contain

CK2-phosphory-lated Cdc37 in vivo These results are consistent with

previous reports that CK2 phosphorylates Cdc37 on

Ser13 both in vivo and in vitro, and that this

phosphor-ylation is essential for the binding of Cdc37 to multiple

Hsp90 client protein kinases [25,26,28]

Immunofluorescence staining with the antibody

showed that phosphorylated Cdc37 accumulated in

growth factor-induced membrane ruffles in KB cells

This accumulation was the result of a redistribution of

phospho-Cdc37 within the cells rather than an

upregu-lation of total Cdc37 phosphoryupregu-lation by CK2 after

EGF stimulation Previously, Hsp90 has been reported

to bind to polymerized actin and to accumulate in

growth factor-induced membrane ruffles, where actin

filaments are abundant [35] In fact, Hsp90 has been

suggested to be involved in the intracellular

distribu-tion and trafficking of many signaling molecules by

interacting with its client proteins and the cytoskeletal

architecture [37,38] Membrane ruffling is one of the

morphological responses rapidly induced in cells by

growth factor stimulation, and the accumulation of a

variety of signaling molecules, including receptor tyro-sine kinases and G-proteins, in membrane ruffles has been reported [39] This might be why Hsp90 and phosphorylated Cdc37 accumulate in membrane ruf-fles, as only the phosphorylated form of Cdc37 is active in recruiting Hsp90 to its client signaling kinases [26] The Hsp90–Cdc37–kinase complexes might then interact with the actin cytoskeleton via Hsp90 in the membrane ruffling region Membrane ruffling has been related to the metastatic status of tumor cells, and it has been suggested as an indicator of tumor cell motil-ity and metastatic potential [40] Therefore, the inhibi-tion of Hsp90, Cdc37 or CK2 might influence, not only signaling affecting cell growth, but also the metas-tasis of neoplastic cells

In this study, we exploited the recently developed compound Phos-tag, which has specific phosphate-binding activity [4,5], to separate phosphorylated and nonphosphorylated forms of Cdc37 by phospho-affin-ity gel electrophoresis We have demonstrated a spe-cific mobility shift in Cdc37 only when it is incubated with CK2 in the presence of Mg2+and ATP Replac-ing the CK2 phosphorylation site in Cdc37 with non-phosphorylatable amino acids completely abolished this mobility shift The low-mobility Cdc37 band was both the only band recognized by the phospho-specific

Fig 7 Characterization of Cdc37 phosphorylation by phospho-affinity gel electrophoresis Recombinant Cdc37 was phosphorylated with CK2a and analyzed by phospho-affinity gel electrophoresis (A) Non-phosphorylated (lane 1) and phosphorylated (lane 2) Cdc37 were sepa-rated by phospho-affinity gel electrophoresis and analyzed by western blotting with anti-Cdc37, for total Cdc37 (B) The same membrane as

in (A) was stripped and reprobed with anti-[pSer13]-Cdc37 (C) Wild-type protein [Cdc37(WT)] (lane 3) and a mutant [Cdc37(13SA)] (lane 4) of Cdc37 were phosphorylated with CK2a in the presence of [ 32 P]ATP[cP] and analyzed by phospho-affinity gel electrophoresis followed by autoradiography As controls, CK2a alone (lane 1) or Cdc37(WT) alone (lane 2) was incubated under the same conditions (D) CBB staining of the same gel as in (C).

antibody against Cdc37 and the only band that

became radioactive after incubation with CK2 in the

presence of [32P]ATP[cP] Using this new technique,

we have been able to strengthen our previous

conclu-sion that CK2 rapidly phosphorylates Cdc37 at only

one site, Ser13, and to demonstrate for the first time

that only phosphorylated Cdc37 is present in signaling

protein kinase complexes with Hsp90 These results

further support the previous proposal that the

phos-phorylation of Cdc37 by CK2 is essential for its

pro-tein kinase-binding activity [25,26,28] Future studies

will be needed to elucidate whether the

phosphoryla-tion of Cdc37 fluctuates dynamically within cells

according to cellular conditions

Phospho-affinity gel electrophoresis is a simple and easy method with the advantage that it uses no radio-active material Moreover, it is a technique that could be used for the analysis of virtually any phosphoprotein, as

a phosphorylation-dependent mobility shift in these gels could occur for any protein It should be noted, however, that large excesses of metal chelators such as EDTA and chemicals containing phosphate moieties, including sodium phosphate, b-glycerophosphate or ATP, may interfere with the assay In addition, crude protein mixtures such as tissue extracts that may contain

a large amount of phosphoproteins may not be easy to analyze directly with this technique Further improve-ments in phospho-affinity gel electrophoresis might be possible and might broaden its applicability

Previous studies have identified many substrates for CK2 [21], and demonstrated its involvement in many different cellular functions, including cell division, cell survival, and gene expression [19,20,24] Surprisingly, however, the regulatory mechanism of CK2 in cells remains largely unknown Observations that CK2 activ-ity was enhanced by growth factors have been chal-lenged [36], and this discrepancy arose partly from the lack of a reliable method for quantifying CK2 activity

in vivo Our results agree with those of previous studies [25,26,28] in demonstrating that CK2 phosphorylates only one site in Cdc37, so the phosphorylation state of Cdc37 should be an index of CK2 activity both in vivo and in vitro The phosphorylation of Cdc37 can now be readily monitored using the phospho-specific antibody and phospho-affinity gel electrophoresis, as described here Several lines of evidence suggest a critical role for CK2 in tumorigenesis [22] For example, CK2 is tumori-genic when overexpressed in a transtumori-genic mouse [41] As the CK2–Cdc37 module has been suggested to function

as a master switch with a positive feedback mechanism for various signaling protein kinases [25,31], the devel-opment of a simple method for determining CK2 activ-ity and Cdc37 phosphorylation is likely to be both biologically and clinically important in the future

Experimental procedures

Plasmids, proteins and antibodies

Plasmids that encode the FLAG-tagged protein kinases Cdk4, MOK, v-Src and Raf1 for expression in mammalian cells have been described previously [26] cDNA for Zebra-fish CK1a was a kind gift from J E Allende (Universidad

de Chile), and an EcoRI fragment of the coding region was inserted into an EcoRI site of pCMV–Tag2B to obtain an

expression plasmid for FLAG-tagged DYRK2 used as a

A

B

Fig 8 Time course of Cdc37 phosphorylation and

dephosphoryla-tion analyzed by phospho-affinity gel electrophoresis (A) Time

course of CK2-dependent Cdc37 phosphorylation was analyzed by

phospho-affinity gel electrophoresis, between 0 min (lane 1) and

10 min (lane 9) (B) Time course of the dephosphorylation of

CK2-phosphorylated Cdc37 by k-phosphatase was analyzed by

phospho-affinity gel electrophoresis Lane 1, nonphosphorylated Cdc37;

lane 2, Cdc37 phosphorylated with CK2a Phosphorylated Cdc37,

as in lane 2, was incubated with (lanes 7–10) or without (lanes 3–6)

k-phosphatase in the presence of the CK2 inhibitor TBB for up to

20 min The mixtures were analyzed by phospho-affinity gel

electro-phoresis followed by western blotting with anti-Cdc37 Samples

were analyzed after dephosphorylation for 0 min (lanes 3 and 7),

3 min (lanes 4 and 8), 10 min (lanes 5 and 9), and 20 min (lanes 6

and 10).