Báo cáo khoa học: Puriﬁcation of phosphoproteins by immobilized metal afﬁnity chromatography and its application to phosphoproteome analysis

Puriﬁcation of phosphoproteins by immobilized metal afﬁnity chromatography and its application to phosphoproteome analysis Mitsuyo Machida1, Hidetaka Kosako1, Kyoko Shirakabe1, Michimoto Kobayashi1, Masato Ushiyama2, Junichi Inagawa2, Joe Hirano2, Tomoyo Nakano3, Yasuhiko Bando3, Eisuke Nishida4 and Seisuke Hattori1,5

1 Division of Cellular Proteomics (BML), Institute of Medical Science, University of Tokyo, Japan 2 GE Healthcare Bio-Sciences KK, Tokyo, Japan 3 AMR Incorporated, Tokyo, Japan 4 Graduate School of Biostudies, Kyoto University, Japan 5 School of Pharmaceutical Sciences, Kitasato University, Tokyo, Japan

Keywords Akt; extracellular signal-regulated kinase (ERK); immobilized metal afﬁnity chromatography; phosphoproteome; two- dimensional gel electrophoresis

Correspondence S. Hattori, Division of Cellular Proteomics (BML), Institute of Medical Science, University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108-8639, Japan Fax ⁄ Tel: +81 3 5449 5314 E-mail: hattoris@ims.u-tokyo.ac.jp

(Received 1 August 2006, revised 6 January 2007, accepted 17 January 2007)

doi:10.1111/j.1742-4658.2007.05705.x

Prefractionation procedures facilitate the identiﬁcation of lower-abundance proteins in proteome analysis. Here we have optimized the conditions for immobilized metal afﬁnity chromatography (IMAC) to enrich for phos- phoproteins. The metal ions, Ga(III), Fe(III), Zn(II), and Al(III), were compared for their abilities to trap phosphoproteins; Ga(III) was the best. Detailed analyses of the pH and ionic strength for IMAC enabled us to determine the optimal conditions (pH 5.5 and 0.5 m NaCl). When whole cell lysates were fractionated in this way, about one-tenth of the total protein was recovered in the eluate, and the recovery of phosphorylated extracellular signal-regulated kinase (ERK) was more than 90%. Phosphor- ylated forms of ribosomal S6 kinase (RSK) and Akt were also enriched efﬁciently under the same conditions. Our Ga(III) IMAC and a commer- cially available puriﬁcation kit for phosphoproteins performed similarly, with a slight difference in the spectrum of phosphoproteins. When phos- phoproteins enriched from NIH3T3 cells in which ERK was either activa- ted or suppressed were analyzed by two-dimensional ﬂuorescence difference gel electrophoresis, phosphorylated ERK was detected as discrete spots unique to ERK-activated cells, which overlapped with surrounding spots in the absence of prefractionation. We applied the same technique to search for Akt substrates and identiﬁed Abelson interactor 1 as a novel potential target. These results demonstrate the efﬁcacy of phosphoprotein enrichment by IMAC and suggest that this procedure will be of general use in phos- phoproteome research.

Abbreviations Abi-1, Abelson interactor 1; ATM, ataxia-telangiectasia mutated; 2-DE, two-dimensional gel electrophoresis; 2-D DIGE, two-dimensional ﬂuorescence difference gel electrophoresis; ERK, extracellular signal-regulated kinase; 4-HT, 4-hydroxytamoxifen; IMAC, immobilized metal afﬁnity chromatography; MEK, MAP kinase or ERK kinase; PAS, phospho-Akt-substrate; PDGF, platelet-derived growth factor; pERK, pRSK, pAkt, phosphorylated forms of ERK, RSK, Akt, respectively; RSK, ribosomal S6 kinase; PI3-kinase, phosphatidyl-3-kinase.

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1576

Protein phosphorylation regulates fundamental cellular processes such as growth, cell division, differentiation, signal transduction and gene expression [1]. Protein kinases attach bulky and strongly ionic phosphate groups to their substrates, causing signiﬁcant conform- ational changes in the phosphorylated proteins. Thus,

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

subcellular [19] that does not maintain enzyme activity was used. In addition, the effect of ionic strength and pH on the performance of IMAC was not studied in detail.

protein phosphorylation controls enzymatic activity, protein–protein interaction, localization, protein stability, and other important properties of the substrates. There are 518 protein kinase genes within the human genome [2], a number that clearly reﬂects the importance of protein phosphorylation. For each cellular process, a set of speciﬁc protein kinases are involved. To understand the physiological functions of kinases, it is necessary to identify their substrates and elucidate the effects of phosphorylation on their func- tions. Therefore, it is of much interest to establish a method to comprehensively analyze total cellular phos- phoproteins, which together constitute the phospho- proteome.

In this study, we used an afﬁnity adsorbent com- posed of iminodiacetic acid-chelated Ga(III) ion, and we optimized conditions to enrich for phosphorylated signaling molecules such as extracellular signal-regula- ted kinase (ERK), Akt, and ribosomal S6 kinase (RSK). We evaluated our established technique against a commercial kit. We then applied the same conditions to prefractionation for phosphoproteome analysis by 2-DE. This enables resolution of the phosphorylated form of ERK (pERK) as discrete spots, which overlap with numerous surrounding protein spots without prefractionation. Moreover, by comparing phospho- protein proﬁles from phosphatidyl-3-kinase (PI3-kin- ase)-activated and -suppressed cells, we detected many spots that were more abundant in the PI3-kinase-acti- vated cells. Among them, we identiﬁed Abelson inter- actor 1 (Abi-1) as a potential target of Akt. These results clearly show the efﬁcacy of our prefractionation procedure, and the conditions described here for prefr- actionation will be of general use in conducting a com- prehensive phosphoproteome analysis. Two-dimensional gel electrophoresis (2-DE) has been widely used to systematically analyze total cellular pro- teins [3–5]. In combination with metabolic labeling of phosphoproteins using radioactive inorganic phos- phate, 2-DE has successfully identiﬁed phosphoproteins [6]. However, nonphosphorylated proteins comigrating with the radioactive spots have hindered MS identiﬁca- tion of phosphorylated proteins. Identiﬁcation of tyro- sine-phosphorylated proteins by 2-DE and western blot with antibody to phosphotyrosine [7] has faced similar technical problems.

Results

2-DE also has another problem: owing to insufﬁ- cient resolution power, the dense spots of cytoskeletal components or house-keeping metabolic enzymes, such as factors involved in signal transduction, have often obscured low-abundance proteins. To overcome this problem, prefractionation of total cellular proteins is highly desirable [8–14]. Prefractionation includes con- ventional chromatography [8], subcellular fractionation [9], puriﬁcation of organelles [10], fractionation of pro- teins according to their isoelectric points [11], isolation of speciﬁc protein complexes [12,13], and puriﬁcation of glycoproteins by lectins [14].

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1577

It has been shown that phosphopeptides can be puri- ﬁed efﬁciently by IMAC. However, IMAC has been rarely used for phosphoprotein puriﬁcation. Therefore, we tried to optimize the conditions for IMAC puriﬁca- tion of phosphoproteins. As a ﬁrst step, the metal ions, Ga(III), Fe(III), Zn(II) and Al(III), were compared for their abilities to trap phosphoproteins (Fig. 1). We used whole lysates from NIH3T3 cells expressing the kinase domain of B-Raf fused to the ligand-binding domain of estrogen receptor [NIH3T3(DB-Raf:ER)] [21], in which ERK activity could be activated by an estrogen antagonist, 4-hydroxytamoxifen (4-HT). As a negative control, the lysate from cells treated with U0126, an inhibitor of mitogen-activated protein (MAP) kinase (ERK kinase; MEK) was used. Lysates from ERK-activated (4-HT-treated) cells were subjec- ted to IMAC in which a metal-chelating resin and one of the four ions was used. After extensive washing of the resin, bound proteins were eluted with 0.2 m sodium phosphate buffer (pH 8.0). The same amount of protein from the lysates before IMAC, the eluates, and unbound fractions (except for eluates from Al and Zn IMAC; see the legend to Fig. 1) were subjected to SDS ⁄ PAGE, and the gel was stained for total proteins (SYPRO Ruby) and phosphoproteins (Pro-Q Dia- mond). The proteins eluted from Ga(III)-activated Taking these backgrounds into consideration, we attempted to establish conditions for enriching cellular phosphoproteins using immobilized metal afﬁnity chro- matography (IMAC) [15]. IMAC has been used to purify phosphopeptides, but only rarely to purify phos- phoproteins. In most studies, Fe(III) was used as the immobilized ion. However, Posewitz & Tempst [16] demonstrated that Ga(III) is superior to Fe(III) in its binding speciﬁcity for phosphopeptides over nonphos- phopeptides. With regard to phosphoprotein puriﬁca- tion, Andersson & Porath [17] showed that IMAC with Fe(III) could separate nonphosphorylated, parti- ally phosphorylated, and highly phosphorylated oval- bumin. Recently, several studies have reported the efﬁcacy of IMAC puriﬁcation of phosphoproteins [18– 20]. However, denaturing conditions [18] or low pH

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

A

Eluate

Unbound

Lysate

-

+ -

- +

- - + + + + Ga Fe Al Zn

(kDa)

contained less phosphoprotein. Western blot using antibodies against the phosphorylated forms of ERK (pERK) and RSK (pRSK) showed that both phospho- proteins were enriched efﬁciently by Ga(III) IMAC, whereas they did not bind during IMAC with the other three metal ions (Fig. 1B).

U0126 4-HT Metal ion 250 150 100

75

50 y b u R O R P Y S

37 (kDa)

250 150 100

75

50 d n o m a i D Q - o r P

37

The binding of phosphoproteins in IMAC is based on a coordination bond between an immobilized metal ion and electron-donor groups of the solutes. Ga(III) ion has six sites for the coordination bond, and func- tions as an acceptor of electron pairs. Two of these sites bind to iminodiacetic acid groups of the metal chelate resin, and the remaining sites serve as binding sites for electron-donor groups such as phosphate groups present on the protein. As the interaction between Ga(III) and phosphate may be affected by the pH of the buffer, we evaluated the effect of pH on the puriﬁcation of pERK from total cell lysate (Fig. 2). The lysates were diluted with 50 mm Mes buffer con- taining 0.5 m NaCl at different pH values ranging from 5.0 to 7.0, and the mixture was incubated with Ga(III)-activated resin for 2 h. After the resin had been washed, bound proteins were eluted with phos- phate buffer (pH 8.0). The eluates were subjected to western blot analysis using antibody to pERK.

B

pERK

pRSK

Fig. 1. Performance of IMAC with different metal ions. (A) Lysates (100 lg, 1 mgÆmL)1) from NIH3T3(DB-Raf:ER) cells treated with 1 lM 4-HT for 30 min were diluted with 1 mL Mes buffer (pH 5.5) containing 0.5 M NaCl and subjected to IMAC with Ga(III) (Ga), Fe(III) (Fe), Al(III) (Al) and Zn(II) (Zn) as described in Experimental procedures. Lysates, unbound fractions and eluates (0.8 lg each) were subjected to SDS ⁄ PAGE (9% gel). As Al(III) and Zn(II) IMAC bound very little protein, all the recovered samples in the ‘Eluate’ fraction were used (eightfold greater volume than for the Ga(III) IMAC ‘Eluate’). The gel was stained ﬁrst with Pro-Q Diamond, then with SYPRO Ruby. As a negative control, a lysate from the same cells treated with 10 lM U0126 was also used. (B) The same amounts of the samples were subjected to western blotting using antibodies speciﬁc for the phosphorylated forms of ERK (pERK) and RSK (pRSK).

As shown in Fig. 2A, pERK bound tightly to the resin between pH 5.0 and 5.7, and was efﬁciently elut- ed with the phosphate buffer. The recovery of pERK decreased as the pH increased above 6.0. Consistent with this result, the amount of pERK detected in the unbound fractions was very low at pH 5.0–5.7, and increased as the pH increased above 6.0. A nonphos- phorylated protein, a-tubulin, was not detected at all in the eluates (Fig. 2A). Aliquots of the same fractions were subjected to SDS ⁄ PAGE, and the gel was silver- stained to visualize protein proﬁles (Fig. 2A). Total protein in the eluate fractions also decreased with increasing pH. We measured the protein content of the eluates and calculated protein recovery based on the total applied (Fig. 2B, closed circles). The recovery of pERK as estimated by densitometric analysis of the blot is also shown in the ﬁgure (open boxes). On the basis of these numbers, the calculated enrichment of pERK at different pH values is plotted in the same ﬁg- ure (triangles). At pH 5.5, nearly 10-fold puriﬁcation was achieved. When contact time with the resin was varied, 2 h mixing gave the best results (data not shown).

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1578

resin (‘Eluate’) showed much stronger staining with Pro-Q Diamond, suggesting that phosphoproteins were enriched in this fraction (Fig. 1A). Although Fe(III) IMAC has been used for enrichment of phosphopro- teins [17–19], Ga(III) was more efﬁcient under our experimental conditions, and the other two metals did not adsorb signiﬁcant amounts of the proteins. Con- sistently, the unbound fraction from Ga(III) IMAC As other negatively charged groups may also bind to Ga(III), we examined the effect of ionic strength on the enrichment of pERK by IMAC. Lysates were dilu- ted with 50 mm Mes buffer (pH 5.5) containing var- ious concentrations of NaCl, and the samples were

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

C

A

Lysate

Eluate

Unbound

Lysate

Eluate

Unbound

+ -

- +

- - + +

- - - + + +

- - + +

- +

+ -

- +

- - + +

- - + + 5.0 5.2 5.5 5.7 6.0 6.2 6.5 6.7 7.0 5.0 5.2 5.5 5.7 6.0 6.2 6.5 6.7 7.0

U0126 4-HT pH

U0126 4-HT NaCl (M)

- + 0

- - - + + + 0.1 0.5 1.0 1.5 2.0

- + 0

- - - + + + 0.1 0.5 1.0 1.5 2.0

(kDa)

100 (kDa)

100

75

50 pERK

-tubulin

Lysate

Eluate

Unbound

Eluate

Unbound

Lysate

- +

- + +

- +

+ +

- +

- - + +

- +

LY294002 + PDGF NaCl (M)

- + 0

- - - + + + 0.1 0.5 1.0 1.5 2.0

- + 0 0.1 0.5 1.0 1.5 2.0

LY294002 PDGF pH

- - - + + + 5.0 5.2 5.5 5.7 6.0 6.2 6.5 6.7 7.0 5.0 5.2 5.5 5.7 6.0 6.2 6.5 6.7 7.0

pAkt

D

B

15

100

10

100 l

)

) d o f (

75

57. )

75 %

10 i

( ) (

i

( ) (

i

50

05.

50

5

25

52.

25 ) d o f ( ) ( y t i v i t c a c i f i c e p S

) ( y t i v i t c a c i f i c e p S

d n a ) ( g n d n b n i e t o r p f o o i t a R

K R E p f o y r e v o c e r

d n a ) ( g n d n b n i e t o r p f o o i t a R

K R E p f o y r e v o c e r

0

5.0

5.5

6.5

7.0

0

0.5

1.5 0 2.0

6.0 pH

1.0 NaCl (M)

Fig. 2. Effect of pH and ionic strength on the puriﬁcation of pERK and pAkt by IMAC. (A) Lysates (100 lg, 1 mgÆmL)1) were diluted with 1 mL Mes buffer, at various pH values, containing 0.5 M NaCl and fractionated by IMAC as described in Experimental procedures. Amounts equivalent to one-ﬁfth of the applied lysate, the recovered unbound fractions (concentrated by trichloroacetic acid precipitation), and the elu- ates were analyzed by SDS ⁄ PAGE (10% gel stained with silver, upper panel) or by western blots with antibodies to pERK (middle), a-tubulin (middle) and pAkt (bottom). The positions of molecular mass standards (in kDa) are shown on the left. (B) On the basis of the pERK recovery (h) and total proteins (d) determined for each fraction, the relative enrichment for pERK (m) under the pH condition was calculated along with standard deviation (n ¼ 3). (C, D) The lysates (100 lg, 1 mgÆmL)1) were diluted with 1 mL Mes buffer (pH 5.5) containing various con- centrations of NaCl and subjected to IMAC puriﬁcation. The result of western blot with anti-pERK, anti-a-tubulin (middle) and anti-pAkt (C), and recovery of pERK (h) and proteins (d), and fold puriﬁcation of pERK (triangles) (D) are shown (n ¼ 3). For pAkt puriﬁcation, cell lysates were prepared from NIH3T3 cells treated with 10 ngÆmL)1 PDGF for 30 min in the presence or absence of 25 lM LY294002.

(PDGF) were

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1579

mixed with the activated resin. Unbound fractions and eluates were analyzed by western blot using antibodies to pERK and a-tubulin. As shown in Fig. 2C, pERK bound tightly to the resin at NaCl concentrations below 0.5 m. At higher NaCl concentrations, the recovery of pERK gradually decreased with increasing NaCl concentration. a-Tubulin was recovered in the unbound fractions but not in the eluates at NaCl con- centrations over 0.5 m. At 0.1 m NaCl, the total amount of protein bound to the resin decreased greatly, and at 0.5 m, bound protein decreased still fur- ther (Fig. 2D). When applied in 0.5 m NaCl, about 10% of the total protein bound to and was eluted from the resin, yielding about ninefold relative enrich- ment of pERK (Fig. 2D). To investigate whether other phosphoproteins behave similarly on IMAC, lysates of NIH3T3 cells treated with platelet-derived growth fac- examined. Phosphorylated Akt tor (pAkt, Fig. 2A,C) and proteins phosphorylated by Akt [phospho-Akt-substrate (PAS)] [22] (Fig. 3) were also

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

A

B

Ga/IMAC

Commercial kit

l

e t a s y L

e t a u E

d n u o b n U

e t a u E

d n u o b n U

e t a s y L

e t a u E

d n u o b n U

e t a u E

pAkt

(kDa) 250

150

100

75

50

37 PAS

Pro-Q Diamond

SYPRO Ruby

Fig. 3. Comparison of different phosphopro- tein enrichment techniques. (A) Phosphopro- teins were enriched from lysates of PDGF-treated NIH3T3 cells using our stand- ard Ga IMAC protocol or a commercial kit (Qiagen). Lysates and fractionated samples were precipitated with trichloroacetic acid ⁄ acetone and analyzed (1.5 lg per lane) by SDS ⁄ PAGE, using Pro-Q Diamond and SYPRO Ruby staining dyes. (B) The gels were further analyzed by western blot with antibodies to pAkt and PAS.

puriﬁed under the same conditions. General phos- phoproteins visualized by Pro-Q Diamond staining behaved similarly (data not shown). Acid phosphatase treatment of the lysates before IMAC diminished the amount of bound protein to almost undetectable level, indicating that most of the bound proteins were phos- phoproteins (data not shown). Taking all these data into consideration, we established the optimum condi- tions for IMAC resin loading and washing as pH 5.5 in the presence of 0.5 m NaCl.

antibody to pERK (Fig. 4B, bottom panel) is shown. Proteins of similar abundance in both samples should appear as yellow spots (an equal mix of red and green), whereas proteins more abundant in 4-HT-treated cells should give rise to red or orange spots according to the extent of the relative abundance. Anti-pERK immuno- reactive spots (Fig. 4B, bottom panel) matched per- fectly the red spots in 2-D DIGE in their shapes, relative intensities and positions, strongly suggesting that these red spots constitute pERK. We identiﬁed proteins in spots 1 and 2 by peptide mass ﬁngerprint- ing using MASCOT software (http://www.matrix- science.com), conﬁrming that spots 1 and 2 were ERK1 and ERK2, respectively (scores 190 and 216). When lysates were directly compared by 2-D DIGE, total spots of pERK were obscured by overlapping with sur- rounding spots (Fig. 4B, upper panel). This result clearly indicates the efﬁcacy of prefractionation in IMAC.

high-molecular-mass

As several products for phosphoprotein enrichment have recently become commercially available, we com- pared the performance of our Ga(III) IMAC with one of these products (PhosphoProtein Puriﬁcation Kit; Qiagen) using lysates from NIH3T3 cells treated with PDGF (Fig. 3). The principles and ingredients of these products are generally not speciﬁed. About 10% of the applied protein was recovered by both puriﬁcation procedures (Fig. 3A), with relatively good recovery for pAkt and substrates phosphorylated by Akt (Fig. 3B). Interestingly, phosphoproteins were more efﬁciently enriched with Ga(III) IMAC compared with the Qiagen product (Fig. 3A,B).

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1580

(2-D DIGE) To evaluate the validity of our established IMAC as a prefractionation procedure in proteome research, NIH3T3(DB-Raf:ER) cells were treated with either 4-HT or U0126 for 30 min, and lysates from these cells were subjected to IMAC. The recovery of ERK and pERK were evaluated by western blot, conﬁrming the selective enrichment of pERK (Fig. 4A). The eluates from IMAC were separately labeled with ﬂuorescent Cy5 and Cy3 dyes, and subjected to two-dimensional difference gel electrophoresis (Fig. 4B, middle panel). A part of the area that reacted with the We used the same method to identify substrates of Akt. As Akt is activated by PDGF downstream of PI3-kinase, NIH3T3 cells were either untreated or trea- ted with PDGF in the presence or absence of an inhib- itor of PI3-kinase (LY294002). Phosphoproteins were enriched from these cell lysates and probed with Pro-Q Diamond staining and western blot with antibody to PAS (Fig. 5A). Whereas Pro-Q Diamond staining showed no marked differences among the three sam- ples, the anti-PAS blot detected a signiﬁcant differ- ence between untreated and PDGF-treated cells. However, there was little difference between cells trea- ted with PDGF alone and PDGF plus LY294002 except two major bands (50 kDa and 36 kDa) unique to the PDGF-treated cells. These two bands are

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

Eluate

Lysate

A

+ -

- +

+ -

U0126 4HT

Lysate - - + - + +

Eluate - - - +

LY294002 PDGF

Eluate - - - +

+ +

Lysate - - - +

+ +

ERK

pERK

(kDa) 150 100 75

B

50 (kDa)

37 PAS

Pro-Q Diamond

Total Lysate

pAkt

U0126 + 4-HT

pGSK3

37 pS6

B

1 U0126 + 4-HT

2

37

2 IMAC Eluate

3

1

75

5

4 anti-pERK

7

6

50

37

25

8

lysates; middle panel,

Fig. 4. 2-D DIGE of IMAC-puriﬁed proteins from ERK-stimulated and suppressed cells. (A) Lysates from NIH3T3(DB-Raf:ER) cells treated with 4-HT or U0126 or IMAC-puriﬁed samples from the same cell lysates (1 lg per lane) were subjected to western blot with antibodies to pERK and ERK. (B) These samples were labeled with Cy5 (4-HT-treated cells, 50 lg) or Cy3 (U0126-treated cells, 50 lg), and the samples were mixed and run on the same 2-DE gel (upper panel, cell IMAC-puriﬁed samples). The ﬂuorescence images were obtained separately at different wavelengths, and are represented using pseudo-colors (red for 4-HT-treated and green for U0126-treated cells). A nonlinear pH gradient (3–10) provided the ﬁrst dimension of separation and SDS ⁄ PAGE (11% gel) the second. Parts of the areas corresponding to the anti-pERK-positive region in the western blot (bottom panel) are shown.

Fig. 5. Comparative 2-D DIGE of IMAC-puriﬁed proteins from PI3- (A) Lysates from NIH3T3 kinase-activated and suppressed cells. cells treated with PDGF alone or PDGF + LY294002 and IMAC-puri- ﬁed samples from the same cell lysates were stained with Pro-Q Diamond (1.5 lg per lane, left panel). These samples were subjec- ted to western blotting with antibodies to PAS, pAkt, phosphor- ylated glycogen synthase kinase-3b (pGSK-3b) and phospho-S6 ribosomal protein (pS6) (1.5 lg per lane, right panel). (B) The same IMAC-puriﬁed samples were analyzed on 2-D DIGE as in Fig. 4. Cy3 (PDGF + LY294002) and Cy5 (PDGF) signals are expressed by green and red pseudo-colors, respectively. Positions of standard proteins in kDa and isoelectric points are shown at the left and top of the gel, respectively. The representative result of three inde- pendent experiments is shown.

considered to correspond to glycogen synthase kinase- 3b (GSK-3b) (50 kDa, Fig. 5A, middle panel) [23] and ribosomal S6 protein [24] (36 kDa, Fig. 5A, bottom panel). These results suggest that most of the anti- PAS-reactive molecules could be phosphorylated by kinases via PI3-kinase-independent pathways under the experimental conditions. The phosphoprotein-enriched samples

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1581

from cells treated with PDGF and PDGF plus LY294002 were labeled with Cy5 (PDGF) and Cy3 (LY + PDGF), then subjected to 2-D DIGE (Fig. 5B). The gel image analysis by decyder software detected 1799 spots. A number of reddish to orange spots were observed, indi- cating that proteins in these spots are more abundant in PDGF-treated cells. We selected spots listed in Table 1 by their volume ratios (> 1.2) and spot sizes (indicated by arrows). To determine proteins in these spots, increased amounts of IMAC-puriﬁed sample (350 lg) from PDGF-treated NIH3T3 cells were run

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

Table 1. Proteins preferentially phosphorylated in PI3-kinase-activated cells. Proteins in the spots were determined by a LC-MS ⁄ MS system as described in Experimental procedures. Spot numbers correspond to those in Fig. 5. Accession number, probability, score, sequence cov- erage as a percentage, and theoretical molecular mass (MW) and isoelectric point (pI) were obtained from NCBInr database with the SEQUEST search engine. E-n, 10)n.

Spot. no.

Candidate proteins

Accession no.

Probabilitya

Volume ratiod

Scoreb

Coveragec

Theoretical MW ⁄ pI

1 2 3 4 5 6 7 8

Abelson interactor 1 Kinesin-like protein KIF3A Cytidine-5’-triphosphate synthase 2 Aconitate hydratase, mitochondrial Lamins C ⁄ C2 Heat-shock 70-kDa protein 4 Chaperonin subunit 5 (epsilon) Calcium-regulated heat-stable protein (24kD)

16225952 125403 13879437 18079339 1346414 12805195 6671702 13385290

3.99E-10 3.23E-08 5.55E-08 8.47E-07 7.40E-09 1.61E-09 1.19E-10 9.68E-09

10.24 140.22 70.21 100.24 50.14 80.26 50.25 40.26

3.2 23.1 16.2 22.7 9.2 22.6 9.8 20.3

1.23 1.25 1.42 1.48 1.48 1.36 1.24 1.46

51589.3 ⁄ 7.3 80167.2 ⁄ 6.2 66682.1 ⁄ 6.1 85462.9 ⁄ 8.1 65445.6 ⁄ 6.5 54677.0 ⁄ 5.2 59623.8 ⁄ 6.0 16062.2 ⁄ 8.4

aThe probability indicates that the observed match is random. bProtein scores are calculated by the addition of identiﬁed peptide scores. cThe coverage indicates the sequence coverage according to identiﬁed peptides. dSpot volume ratios (Cy5 ⁄ Cy3) were measured using DECYDER software.

IP: anti-FLAG WB: anti-FLAG

IP: anti-FLAG WB: anti-PAS

+ + -

- + -

+ - -

+ + -

- - -

- + -

+ - -

FLAG-Abi-1 BD110 LY294002

+ + +

(kDa) 100

75

50

Fig. 6. Phosphorylation of Abi-1 in response to PI3-kinase activa- tion. HEK293 cells were transfected with a plasmid encoding FLAG-tagged Abi-1 and ⁄ or a plasmid encoding constitutively active PI3-kinase (BD110). The indicated cells were treated with 25 lM LY294002 for 45 min. Anti-FLAG immunoprecipitates were probed with either anti-PAS (left panel) or anti-FLAG (right panel) (SDS ⁄ PAGE, 8% gel). An arrowhead indicates anti-PAS reactivity comigrating with FLAG-Abi-1 (arrow). Positions of standard proteins in kDa are shown on the left.

on a separate 2-D gel (see Experimental procedures). Proteins in these spots were digested with lysy- lendopeptidase, and the peptides were analyzed by liquid chromatography ⁄ tandem mass spectrometry (LC-MS ⁄ MS) (Table 1). One of the identiﬁed proteins, calcium-regulated heat-stable protein (spot 8), was indeed a previously reported Akt substrate [25]. Also, a ScanSite search [26] revealed that two of the identi- ﬁed proteins, Abi-1 and lamins C ⁄ C2, contain Akt phosphorylation motifs. Interestingly, according to the same search, lamins C ⁄ C2 and heat-shock 70-kDa pro- tein 4 have motifs phosphorylated by ataxia-telangiec- tasia mutated (ATM) protein kinase. As LY294002 also inhibits ATM activity [27], it is possible that these proteins are substrates of ATM rather than of Akt.

Discussion

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1582

Among the proteins identiﬁed, we further character- ized Abi-1, because it has an Akt phosphorylation motif and is known to interact with Abl and inhibit Abl-induced ERK activation [28]. To further charac- terize Abi-1 in more detail, HEK293 cells were trans- fected with a plasmid expressing FLAG-tagged Abi-1 and ⁄ or a plasmid expressing a constitutively activated form of PI3-kinase (BD110) [29] (Fig. 6). FLAG-Abi-1 was detected as a (cid:2) 65-kDa protein by antibody to FLAG in the immunoprecipitates of FLAG-Abi-1- transfected cells (indicated by an arrow, right panel). On the other hand, FLAG-Abi-1 was detected by anti- body to PAS only in the immunoprecipitates of the cells cotransfected with FLAG-Abi-1 and BD110 (arrowhead, this anti-PAS left panel). In addition, reactivity of FLAG-Abi-1 was abolished by pretreat- ment of the cells with LY294002. These results suggest that Abi-1 is phosphorylated in response to PI3-kinase activation, probably by Akt. In this study, we have shown that Ga(III) IMAC is very effective in the puriﬁcation of phosphoproteins. Although this technique has often been used to purify phosphopeptides, its application to phosphoprotein puriﬁcation has been rare. In most studies, proteins with multiple phosphorylation sites, such as ovalbu- min, casein and phosvitin, were used, as they bind tightly to the chelated metal ion [17,30]. We estab- lished the best conditions for puriﬁcation of a wide

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

experimental conditions

Recently, puriﬁcation kits for phosphoproteins have become commercially available. We examined one of these and found it to as effective as our procedure for purifying phosphoproteins. However, the buffer ingredients and chelated metal ions are not speciﬁed, and therefore cannot be modiﬁed. This may be a signiﬁcant drawback when other biochemical fractionations are to be combined with phosphoprotein puriﬁcation. Our method pro- vides a set of known conditions that are compatible with other procedures, but which could be modiﬁed as needed, guided in part by the data provided here in Fig. 2. is Prefractionation of samples range of phosphoproteins by optimizing the pH and ionic strength. The phosphorylated forms of ERK, RSK, Akt (Figs 1 and 2) and proteins phosphorylated by Akt (Fig. 3) were successfully puriﬁed by IMAC with good recovery. Using the procedure we estab- lished, phosphoproteins were enriched, and comparat- ive 2-DE of puriﬁed samples detected pERK as isolated spots that were unique to ERK-stimulated cells. We also identiﬁed several proteins that were more abundant in the phosphoprotein fraction from PI3-kinase-activated cells than PI3-kinase-suppressed cells, among which we identiﬁed Abi-1 as a novel potential target of Akt. These results demonstrate that our established procedure is applicable to a wide range of phosphoproteome analyses. Fe(III) ion is the most widely used metal

[31], that our Ga(III) shows

recommended for analyzing less-abundant proteins by 2-DE [8,9,11–14]. Biochemical fractionation of subcellular compartments, separation of proteins according to their isoelectric points, immunoprecipitation of tyrosine-phosphorylat- ed proteins using antibodies to phosphotyrosine, and other procedures have been shown to be effective. We puriﬁed phosphoproteins from ERK-stimulated and ERK-suppressed cells by IMAC, and compared the protein proﬁles by 2-D DIGE. We detected pERK as discrete and unique spots in a sample from ERK- stimulated cells (Fig. 4B), whereas such discrete spots could not be detected in the absence of prefractiona- tion (Fig. 4B). We conﬁrmed by peptide mass ﬁnger- printing that these spots indeed contained ERK. This result IMAC procedure reduced the total number of protein spots on 2-DE, greatly facilitating the identiﬁcation of the remaining spots of interest.

ion for purifying phosphopeptides using IMAC. However, Posewitz & Tempst [16] found Ga(III) to be most spe- ciﬁc for phosphopeptide binding. We evaluated several metal ions, and found that Ga(III) binds phosphopro- teins most effectively under our conditions. For the puriﬁcation of phosphopeptides by IMAC, a pH range of 2.5–3.0 is generally used [16]. However, this range is not suitable for proteins, as some proteins precipitate and most enzymes lose their activity in this the optimal pH pH range. Therefore, we sought between 5.0 and 7.0. As described by Muszynska et al. the interaction between phosphoproteins and chelated metal ions weakens as the pH increases (Fig. 2A). We chose pH 5.5, a level at which the enrichment estimated for pERK with respect to total protein reached its peak (Fig. 2B). The enrichment of pAkt (Fig. 2) and PAS (data not shown) was similar. Pro-Q Diamond staining also showed that the condi- tions were optimal for phosphoprotein enrichment (data not shown).

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1583

Ionic strength was another parameter examined with regard to reducing the undesirable binding of negat- ively charged groups (e.g. acidic amino acids). The amount of protein bound during IMAC decreased dramatically with increasing NaCl concentration in the binding solution (Fig. 2C). Again, taking the increase in the speciﬁc activity of pERK as the criterion, we chose 0.5 m NaCl as the standard condition. With a binding buffer containing 0.5 m NaCl at pH 5.5, (cid:2) 10% of total protein bound to the IMAC column. pAkt, pRSK and anti-PAS reactive molecules were also enriched with relatively good yields, whereas the nonphosphoprotein, a-tubulin, was not recovered at all (Figs 1–3). Under these conditions, phosphatase treat- lysates almost completely abolished ment of the cell protein binding (data not shown), indicating that most of the bound proteins were phosphoproteins. When phosphoprotein-enriched fractions from lysates of ERK-activated and ERK-suppressed NIH3T3(DB- Raf:ER) cells (Fig. 4), or PDGF-treated and untreated NIH3T3 cells (data not shown) were analyzed by 2-D DIGE, there was a large difference between the pairs. However, when PDGF-treated NIH3T3 cells in the presence and absence of a PI3-kinase inhibitor were compared, no spots with a marked difference were observed, as illustrated in Fig. 5B. Western blot with antibody to PAS showed a similar trend (Fig. 5A). Even with a known Akt substrate, calcium- regulated heat-stable protein, the volume ratio was 1.46. Activation of Akt was completely suppressed by LY294002 (Fig. 2), suggesting that these substrates were also phosphorylated by kinases in other signa- ling pathways under these conditions. Potential ATM substrates which contain ATM phosphorylation motifs (lamins C ⁄ C2 and heat-shock 70-kDa pro- tein 4) were also identiﬁed, because LY294002 inhib- its ATM as well as PI3-kinase [27]. As ATM is activated by DNA damage [32], it is probably reason-

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

Experimental procedures

able that the spot volume ratios for these proteins were low. Cell culture and preparation of cell lysates

this pathway. Therefore,

phosphoprotein product for

NIH3T3 cells were maintained in DMEM containing 10% heat-inactivated bovine serum and antibiotics. NIH3T3 cells expressing the kinase domain of B-Raf fused to the lig- and-binding domain of estrogen receptor [NIH3T3(DB- Raf:ER)] [21] were maintained in phenol red-free DMEM containing 10% heat-inactivated fetal bovine serum and antibiotics. NIH3T3 cells (4 · 106 cells per 90 mm dish) were pretreated with or without 25 lm LY294002 (Sigma, St Louis, MO, USA) for 45 min before stimulation with PDGF (Sigma; 10 ngÆmL)1) for 30 min to examine PI3-kin- ase ⁄ Akt signaling components. Similarly, NIH3T3(DB- Raf:ER) cells were treated with either 10 lm U0126 (Promega, Madison, WI, USA) or 1 lm 4-HT (Sigma) for 30 min to analyze factors involved in ERK signal-transduc- tion pathways. Cells were lysed with 1 mL per dish of a lysis buffer containing 15 mm Mes ⁄ NaOH (pH 5.5), 0.15 m NaCl, 1% Nonidet P40, 5 mm EDTA, 1 mm Na3VO4, 10 mm NaF, 100 UÆmL)1 aprotinin, 10 mgÆmL)1 leupeptin, 1 mm phenylmethanesulfonyl ﬂuoride and 25 mm b-glycero- phosphate. The lysates were centrifuged at 15 000 g at 4 (cid:2)C for 10 min using a TOMY TMA-22 rotor, and the superna- tants were retained. Protein concentrations of cell lysates were determined as described by Bradford [35].

Regardless of this difﬁculty, we identiﬁed Abi-1 as a novel potential Akt target (Fig. 5). Abi-1 was react- ive with antibody to PAS in cells constitutively expressing active PI3-kinase, and this reactivity was abolished after LY294002 treatment (Fig. 6). As Abi-1 binds to Abl tyrosine kinase and inhibits the Abl-induced ERK signaling pathway [28], our ﬁnd- ings suggest the interesting possibility that Akt regu- lates the phosphoprotein puriﬁcation procedure provided in this study should be of general use in the enrichment of phosphopro- teins for 2-DE analysis. Using a commercially avail- able enrichment combined with the 2-D DIGE system, we were able to identify candidates potentially involved in p38 MAP kinase pathways [33].

Poros MC resin (10 mg slurry; Applied Biosystems, Foster City, CA, USA) in a 1.5-mL microtube was charged with 100 lL 100 mm metal ion solutions (GaCl3, FeCl3, AlCl3 and ZnCl2). The resin was washed three times with 500 lL 0.1% acetic acid to remove unbound metal ion. Then, 100 lL cell lysate (1 mgÆmL)1) diluted with 1 mL 50 mm Mes ⁄ NaOH buffer containing NaCl (buffer pH and NaCl concentration as speciﬁed in each ﬁgure) was added, and the mixture was incubated at 4 (cid:2)C for 2 h. The resin was washed with 500 lL of the same buffer to remove nonspe- ciﬁcally bound proteins. Finally, phosphoproteins were elut- ed with 100 lL 0.2 m sodium phosphate buffer (pH 8.0) at 60 (cid:2)C for 15 min. Proteins in the eluates and unbound frac- tions were concentrated by trichloroacetic acid precipita- for phosphoprotein tion. A commercially available kit puriﬁcation (Qiagen, Valencia, CA, USA) was used to com- pare its performance with the Ga(III) IMAC described above.

Puriﬁcation of phosphoproteins by IMAC

Proteins were separated by SDS ⁄ PAGE (9% gel). The gel was stained using Pro-Q Diamond phosphoprotein gel stain (Molecular Probes, Eugene, OR, USA) according to the

Detection of phosphorylated proteins and total proteins

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1584

Recently, an IMAC protocol using Ga(III)-activated and Fe(III)-activated resins was reported for urea-solu- bilized extracts from mouse synapses [18]. Taking tryp- the tic digests of phosphoprotein-enriched samples, authors further enriched for phosphopeptides by con- ventional peptide IMAC and identiﬁed more than 300 phosphorylation sites by LC-MS ⁄ MS. Two other studies [19,20] used Fe(III)-activated resins to purify phosphoproteins and identiﬁed proteins in the spots on 2D gels that corresponded to 32P-labeled spots. Most of the proteins identiﬁed in the latter study were repor- ted to be phosphorylated, indicating the speciﬁcity of the puriﬁcation procedure for phosphoproteins [20]. However, the effect of pH and ionic strength on the recovery of phosphoproteins and the increase in the speci- ﬁc activity was not examined in detail. In contrast with Ga(III) and Fe(III) IMAC, two zinc ions trapped in an organic compound with an appropriate distance was found to speciﬁcally bind to phosphoproteins [34]. How- ever, cellular responses were not characterized in detail. We established the puriﬁcation procedure and further examined dynamic changes in phosphoprotein proﬁles that were elicited by PI3-kinase activation by 2-D DIGE. Protein phosphorylation is a ubiquitous reaction involved in virtually all biological phenomena. From a medical viewpoint, the deregulated activity of protein kinases has been observed in the development of various human diseases such as cancer. Identiﬁcation of protein kinase substrates advances the elucidation of kinase functions, and provides novel targets for pharmaceutical drugs useful in the treatment of such diseases. Therefore substrate identiﬁcation is a key issue. Given a speciﬁc inhibitor of a kinase of interest, substrates of that kinase can be identiﬁed by the strat- egy described in this study.

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

the gel,

the manufacturer’s protocol. After scanning of same gel was stained for total proteins with SYPRO Ruby protein gel stain (Molecular Probes). In some experiments, total proteins were stained with silver.

4% Chaps, 1% Pharmalyte (3–10 for isoelectric focusing), and 2% dithiothreitol was added to the sample. Two labe- led samples were combined and diluted with rehydration buffer (7 m urea, 2 m thiourea, 4% Chaps and 0.2% dithio- threitol) to a ﬁnal volume of 450 lL. This mixture was immediately subjected to isoelectric focusing on an IPG- phor system using a 24-cm-long strip (nonlinear pH 3–10 gradient; GE Healthcare Bio-Sciences Corp.). Before the second-dimension separation, strips were reduced in 10 mL rehydration buffer with 100 mg dithiothreitol, and subse- quently alkylated in 10 mL rehydration buffer with 250 mg iodoacetamide. The strips were immediately applied to SDS ⁄ polyacrylamide gels, and the gels were run with an Ettan DALT system (GE Healthcare Bio-Sciences Corp.) following the manufacturer’s protocol. The protein proﬁle was obtained using a Typhoon 9400 variable mode imager (GE Healthcare Bio-Sciences Corp.). The Cy3 and Cy5 dye images were scanned with 532 and 633 nm lasers and with emission ﬁlters of 580 and 670 nm, respectively. Spot vol- ume ratios were measured using decyder software (GE Healthcare Bio-Sciences Corp.).

Western blotting, DNA transfection, and immunoprecipitation

Proteins were resolved with SDS ⁄ PAGE, and the proteins in the gel were electrophoretically transferred to a poly(vinylidene diﬂuoride) membrane (Millipore, Bedford, MA, USA). Membranes were incubated overnight with an appropriate dilution of primary antibodies [anti-pRSK, anti-pERK, anti-pAkt anti- PAS, anti-(phospho-S6 ribo- somal protein) (pS6), anti-a-tubulin (Cell Signaling Tech- nology, Danvers, MA, USA) and anti-ERK (Santa Cruz Biotechnology, Santa Cruz, CA, USA)] in 20 mm Tris ⁄ HCl (pH 7.5) ⁄ 150 mm NaCl ⁄ 0.1% Tween-20 containing 1% gel- atin. The membranes were washed three times with 20 mm Tris ⁄ HCl (pH 7.5) ⁄ 150 mm NaCl ⁄ 0.1% Tween-20 and then incubated with a horseradish peroxidase-conjugated anti- rabbit Ig antibody (GE Healthcare Bio-Sciences Corp, Pis- cataway, NJ, USA). Immunoreactive bands were detected by the enhanced chemiluminescence method using a kit purchased from PerkinElmer (Wellesley, MA, USA). Band intensities were quantiﬁed by using Image-J in NIH-Image (http://rsb.info.nih.gov/ij). HEK293T cells were transfected with an expression vector for FLAG-tagged Abi-1 (a gift from Dr S. Suetsugu, University of Tokyo) and ⁄ or a plas- mid expressing constitutively active PI3-kinase (BD110 [33], kindly provided by Y. Fukui, University of Tokyo) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). One day after transfection, the cells were treated with or without 25 lm LY294002 for 45 min, and cell lysates were prepared. FLAG-Abi-1 was immunoprecipitated with anti-FLAG M2-afﬁnity gel (Sigma), and the immunoprecipitates were subjected to western blot with antibody to FLAG M2 (Sigma) or PAS.

Identiﬁcation of proteins

For protein identiﬁcation, 350 lg phosphoprotein-enriched fraction from PDGF-stimulated NIH3T3 cells was run on a 2D gel [nonlinear pH 3-10 gradient (24cm)] 10% SDS-poly- acrylamide gel (3–10 NL, 24 cm). The proteins in the gel were electrophoretically transferred to a poly(vinylidene diﬂuoride) membrane (ProBlott; Applied Biosystems) and stained with Coomassie brilliant blue R-350. The DIGE proﬁle was overlaid on the protein-stained ﬁlter, and the spots of interest were excised and digested in situ with lysy- lendopeptidase [36]. Liberated peptides were analyzed by LC-MS ⁄ MS using RP-lLC and Finnigan LTQ linear ion trap mass spectrometer (Thermo Electron, Waltham, MA, USA) equipped with nano spray ionization sources (AMR Inc., Meguro, Tokyo, Japan) under the conditions des- cribed by Fujii et al. [37]. LC-MS ⁄ MS data were examined using the NCBInr database with a SEQUEST search engine (Thermo Electron). To computationally identify motifs within given proteins, a motif scan program in Scansite version 2.0 was used [24].

2-D DIGE

Acknowledgements

This was carried out using an Ettan DIGE system (GE Healthcare Bio-Sciences Corp.) according to the manufac- turer’s protocol. Phosphoproteins puriﬁed with IMAC were precipitated with 10% trichloroacetic acid and washed with cold acetone. The precipitated proteins were solubilized in solubilization buffer [7 m urea, 2 m thiourea, 4% CHAPS, 30 mm Tris ⁄ HCl (pH 8.0)], and then the protein concentra- tion was measured by the method of Bradford [35]. Two samples to be compared (50 lg protein per sample) were differentially labeled with 400 pmol ﬂuorescent cyanine dyes, Cy5 or Cy3, on ice for 30 min in the dark. The reac- tion was stopped by adding 1 lL 10 mm lysine, then the samples were further incubated on ice for 10 min. An equal volume of a solution containing 7 m urea, 2 m thiourea,

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1585

We thank Drs Shiro Suetsugu and Yasuhisa Fukui for providing plasmids, and Drs Naoyuki Iida and Ryoko Otsuka for valuable comments. This work was suppor- ted in part by Grants-in-Aid for Scientiﬁc Research from Japan Society for the Promotion of Science (to HK and MK), and Grants-in-Aid and the Encouraging Development of Strategic Research Centers, Special Coordination Funds for Promoting Science and Tech-

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

from the Ministry of Education, Culture, nology, Sports, Science and Technology (to SH). This work was also supported by grants from the Nakajima Foundation (to HK) and the NOVARTIS Foundation (Japan) for the Promotion of Science (to SH).

References

13 Cronshaw JM, Krutchinsky AN, Zhang W, Chait BT & Matunis MJ (2002) Proteomic analysis of the mamma- lian nuclear pore complex. J Cell Biol 158, 915–927. 14 Taniguchi N, Ekuni A, Ko JH, Miyoshi E, Ikeda Y, Ihara Y, Nishikawa A, Honke K & Takahashi M (2001) A glycomic approach to the identiﬁcation and characterization of glycoprotein function in cells trans- fected with glycosyltransferase genes. Proteomics 1, 239– 247.

1 Hunter T (2000) Signaling-2000 and beyond. Cell 100,

15 Porath J (1992) Immobilized metal ion afﬁnity chroma-

113–127.

tography. Protein Expr Purif 3, 263–281.

2 Manning G, Whyte DB, Martinez R, Hunter T &

Sudarsanam S (2002) The protein kinase complement of the human genome. Science 298, 1912–1934.

16 Posewitz MC & Tempst P (1999) Immobilized gallium (III) afﬁnity chromatography of phosphopeptides. Anal Chem 71, 2883–2892.

3 O’Farrell PH (1975) High resolution two-dimensional electrophoresis of proteins. J Biol Chem 250, 4007– 4021.

17 Andersson L & Porath J (1986) Isolation of phospho- proteins by immobilized metal (Fe3+) afﬁnity chroma- tography. Anal Biochem 154, 250–254.

18 Collins MO, Yu L, Coba MP, Husi H, Campuzano I,

Blackstock WP, Choudhary JS & Grant SG (2005) Pro- teomic analysis of in vivo phosphorylated synaptic pro- teins. J Biol Chem 280, 5972–5982.

4 Lewis TS, Hunt JB, Aveline LD, Jonscher KR, Louie DF, Yeh JM, Nahreini TS, Resing KA & Ahn NG (2000) Identiﬁcation of novel MAP kinase pathway sig- naling targets by functional proteomics and mass spec- trometry. Mol Cell 6, 1343–1354.

5 Celis JE & Gromov P (1999) 2D protein electrophoresis:

can it be perfected? Curr Opin Biotechnol 10, 16–21.

19 Guerrera IC, Predic-Atkinson J, Kleiner O, Soskic V & Godovac-Zimmermann J (2005) Enrichment of phos- phoproteins for proteomic analysis using immobilized Fe (III)-afﬁnity adsorption chromatography. J Proteome Res 4, 1545–1553.

20 Dubrovska A & Soucheinyskyi S (2005) Efﬁcient enrich- ment of intact phosphorylated proteins by modiﬁed immobilized metal-afﬁnity chromatography. Proteomics 5, 4678–4683.

21 Pritchard CA, Samuels ML, Bosch E & McMahon M

6 Prestayko AW, Crane PM & Busch H (1976) Phosphor- ylation and DNA binding of nuclear rat liver proteins soluble at low ionic strength. Biochemistry 15, 414–421. 7 Cooper JA, Bowen-Pope DF, Raines E, Ross R & Hun- ter T (1982) Similar effects of platelet-derived growth factor and epidermal growth factor on the phosphoryla- tion of tyrosine in cellular proteins. Cell 31, 263–273.

8 Lescuyer P, Hochstrasser DF & Sanchez JC (2004)

(1995) Conditionally oncogenic forms of the A-Raf and B-Raf protein kinases display different biological and biochemical properties in NIH 3T3 cells. Mol Cell Biol 15, 6430–6442.

Comprehensive proteome analysis by chromatographic protein prefractionation. Electrophoresis 25, 1125–1135. 9 Cordwell SJ, Nouwens AS, Verrills NM, Basseal DJ &

Walsh BJ (2000) Subproteomics based upon protein cel- lular location and relative solubilities in conjunction with composite two-dimensional electrophoresis gels. Electrophoresis 21, 1094–1103.

22 Zhang H, Zha X, Tan Y, Hornbeck PV, Mastrangelo AJ, Alessi DR, Polakiewicz RD & Comb MJ (2002) Phosphoprotein analysis using antibodies broadly reac- tive against phosphorylated motifs. J Biol Chem 277, 39379–39387.

10 Takahashi N, Yanagida M, Fujiyama S, Hayano T &

23 Cross DAE, Alessi DR, Cohen P, Andjelkovich M & Hemmings BA (1995) Inhibition of glycogen synthase kinase-3 by insulin mediated by protein kinase B. Nat- ure 378, 785–789.

Isobe T (2003) Proteomic snapshot analyses of preribo- somal ribonucleoprotein complexes formed at various stages of ribosome biogenesis in yeast and mammalian cells. Mass Spectrom Rev 22, 287–317.

11 Zuo X & Speicher DW (2000) A method for global

24 Saitoh M, Pullen N, Brennan P, Cantrell D, Dennis PB & Thomas G (2002) Regulation of an activated S6 kinase 1 variant reveals a novel mammalian target of rapamycin phosphorylation site. J Biol Chem 277, 20104–20112.

analysis of complex proteomes using sample prefractionation by solution isoelectrofocusing prior to two-dimensional electrophoresis. Anal Biochem 284, 266–278.

12 Rigaut G, Shevchenko A, Rutz B, Wilm M, Mann M & Seraphin B (1999) A generic protein puriﬁcation method for protein complex characterization and proteome exploration. Nat Biotechnol 17, 1030–1032.

25 Auld GC, Campbell DG, Morrice N & Cohen P (2005) Identiﬁcation of calcium regulated heat-stable protein of 24 kDa (CRHSP24) as a physiological substrate for PKB and RSK using KESTREL. Biochem J 389, 775– 783.

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1586

M. Machida et al.

IMAC puriﬁcation of phosphoproteins

32 Banin S, Moyal L, Shieh S, Taya Y, Anderson CW,

26 Obenauer JC, Cantley L & Yaffe MB (2003) Scansite 2.0: proteome-wide prediction of cell signaling interac- tions using short sequence motifs. Nucleic Acids Res 31, 3635–3641.

Chessa L, Smorodinsky NI, Prives C, Reiss Y, Shiloh Y, et al. (1998) Enhanced phosphorylation of p53 by ATM in response to DNA damage. Science 281, 1674–1677.

33 Ueda K, Kosako H, Fukui Y & Hattori S (2004) Pro-

27 Goodarzi AA & Lees-Miller SP (2004) Biochemical char- acterization of ataxia-telangiectasia mutated (ATM) pro- tein from human cells. DNA Repair (Amst) 3, 753–767. 28 Fan PD & Goff SP (2000) Abl interactor 1 binds to Sos

teomic identiﬁcation of Bcl2-associated athanogene 2 as a novel MAPK-activated protein kinase 2 substrate. J Biol Chem 279, 41815–41821.

34 Kinoshita E, Yamada A, Takeda H, Kinoshita-Kikuta

and inhibits epidermal growth factor- and v-Abl- induced activation of extracellular signal-regulated kinases. Mol Cell Biol 20, 7591–7601.

E & Koike T (2005) Novel immobilized zinc (II) afﬁnity chromatography for phosphopeptides and phosphory- lated proteins. J Sep Sci 28, 155–162.

35 Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein util- izing the principle of protein-dye binding. Anal Biochem 72, 248–254.

36 Iwamatsu A & Yoshida-Kubomura N (1996) Systematic

29 Yokogawa T, Nagata S, Nishio Y, Tsutsumi T, Ihara S, Shirai R, Morita K, Umeda M, Shirai Y, Saitoh N, et al. (2002) Evidence that-3¢-phosphorylated po- lyphosphoinositides are generated at the nuclear sur- face: use of immunostaining technique with monoclonal antibodies speciﬁc for PI 3,4-P (2). FEBS Lett 473, 222–226.

30 Muszynska G, Andersson L & Porath J (1986) Selective adsorption of phosphoproteins on gel-immobilized ferric chelate. Biochemistry 25, 6850–6853.

peptide fragmentation of polyvinylidene diﬂuoride (PVDF)-immobilized proteins prior to microsequencing. J Biochem (Tokyo) 120, 29–34.

31 Muszynska G, Dobrowolska G, Medin A, Ekman P & Porath JO (1992) Model studies on iron (III) ions with phosphorylated amino acids, peptides and proteins. J Chromatogr 604, 19–28.

37 Fujii K, Nakano T, Kanazawa M, Akimoto S, Hirano T, Kato H & Nishimura T (2005) Clinical-scale high- throughput human plasma proteome analysis: lung ade- nocarcinoma. Proteomics 5, 1150–1159.

FEBS Journal 274 (2007) 1576–1587 ª 2007 The Authors Journal compilation ª 2007 FEBS

1587