profiling substrate phosphorylation at the phosphopeptide level
TRANSCRIPT
Profiling substrate phosphorylation at the phosphopeptide level
Andrea Gatti*
Department of Biochemistry, University of California, Riverside, CA 92521, USA
Received 18 June 2002
Abstract
The identification of substrates is a key aspect in the study of the biological function of protein kinases. The procedure here
described is aimed at profiling substrate phosphorylation at the phosphopeptide level by sequentially involving (i) the assessment of
the in vitro activity of individual protein kinases on a complex mix of immobilized proteins, (ii) the fractionation of the phos-
phopeptides being released upon proteolysis of substrates, and (iii) the final identification of the targeted sequences. In particular,
the protein sample is spotted onto nitrocellulose membrane and then subjected to a solid-phase kinase assay in the presence of
[32P]ATP, prior to solid-phase proteolytic digestion and two-dimensional phosphopeptide mapping. Radiolabeled phosphopeptides
are subsequently isolated and sequenced to identify the substrates being targeted by the examined protein kinase. Using the c-isotypeof p21-activated protein kinase (c-PAK) and its known in vitro substrates, I verified that both the specificity of substrate phos-
phorylation and its efficiency are similar upon solid- and liquid-phase conditions. To demonstrate the feasibility of the overall
experimental system, I then employed a fairly crude cell extract as a source of candidate substrates and successfully identified the
sequence of a putative substrate of c-PAK.� 2003 Elsevier Science (USA). All rights reserved.
Keywords: PAK; Solid-phase kinase assay; Substrate specificity
The large number of protein kinases and the evenlarger spectrum of the respective substrates involved in
cell signaling reflects the high level of complexity of the
overall phenomenon of protein phosphorylation. To
better understand the molecular basis underlying the
specificity of protein phosphorylation, I investigated the
possibility of generating 2-D1 peptide maps to profile
substrate phosphorylation at the phosphopeptide level.
Traditionally, the search for relevant substrates ofprotein kinases has been conducted by comparative 2-D
gel electrophoresis of protein samples frommetabolically
radiolabeled cells, followed by amino acid sequencing of
selected phosphoproteins. New methodologies to iden-
tify substrates of protein kinases by screening cDNAexpression libraries with appropriately chosen molecular
probes such as antibodies and binding proteins have re-
cently become available [1]. Notably, many of these re-
cent applications employ nitrocellulose membrane to
immobilize the expression clones. In some cases, the
screening assay is based on solid-phase phosphorylation
of plaque proteins by exogenously added protein kinases
[2]. The idea of using a solid-phase kinase assay in thepresent studywas also encouraged by the conclusions of a
report of Valtorta et al. [3], according to which SDS–
PAGE-fractionated and nitrocellulose-immobilized pro-
teins are effectively phosphorylated by soluble protein
kinases with a specificity similar to that obtained in a
conventional (i.e., liquid-phase) kinase assay. Note that
in the present study the use of immobilized substrates was
chosen to facilitate the overall flow of operations prior topeptide profiling (e.g., by allowing a direct solid-phase
proteolytic cleavage of substrates).
As a model protein kinase, I employed the c-isotype ofp21-activated protein kinase (c-PAK) [4,5], also known
as PAK-2 [6], a serine/threonine kinase whose activity is
Analytical Biochemistry 312 (2003) 40–47
www.elsevier.com/locate/yabio
ANALYTICAL
BIOCHEMISTRY
* Present address: Via Stringher 27, Rome 00191, Italy. Fax: 39-06-
36300832.
E-mail address: [email protected] Abbreviations used: 2-D, two-dimensional; DEAE, diethylamino-
ethyl cellulose; H4, histone 4; HPLC, high-performance liquid chro-
matography; IMAC, immobilized metal affinity chromatography;
MBP, myelin basic protein; MS, mass spectrometry; PAK, p21-
activated protein kinase; PRS, postribosomal supernatant; TDX1,
thioredoxin peroxidase 1; TFA, trifluoroacetic acid.
0003-2697/02/$ - see front matter � 2003 Elsevier Science (USA). All rights reserved.
PII: S0003 -2697 (02 )00426-8
stimulated by binding to small G proteins such as Cdc42and Rac [7]. Although little information on the in vivo
activity of c-PAK is available, proteins such as myelin
basic protein (MBP) and histone 4 (H4) are known to be
phosphorylated by c-PAK in vitro [8,9]. To profile the
liquid- and the solid-phosphorylation of MBP and H4 at
the phosphopeptide level, the respective tryptic digests
were fractionated by 2-D peptide PAGE, a procedure
recently developed to map radiolabeled [10] and fluo-rescent [11] peptides. Note that such a 2-D gel system has
been employed in the analysis of in vitro [12] and in vivo
[13] phosphorylation of c-PAK. By combining manual
and automated amino acid sequencing, the specificities of
phosphorylation of MBP and H4 were demonstrated to
be independent of whether the respective kinase assays
were performed under liquid- or solid-phase conditions.
The use of the solid-phase kinase assay with a fairly crudecell extract and purified c-PAK, followed by 2-D peptide
PAGE and standard peptide purification steps, was
subsequently shown to allow the identification of the
target sequence of a novel substrate of c-PAK.
Methods
Preparation of protein samples
Postribosomal supernatant (PRS) was obtained from
rabbit reticulocytes as previously described [14]. PRS was
diluted 10� with 10mM Tris–HCl, pH 7.4 (buffer A),
prior to being loaded onto a 5-ml diethylaminoethyl cel-
lulose (DEAE; Whatman) column preequilibrated with
the same buffer. After washing the column with 15 vol ofbuffer A, proteins were eluted with 3 vol of buffer A sup-
plemented with 0.5M NaCl. The DEAE eluate was dia-
lyzed at 4 �C against double-distilled water for 24 h prior
to being used.When indicated, purifiedMBP frombovine
brain (Fluka) or proteins from the dialyzed DEAE eluate
were immobilized on nitrocellulose (0.45 lm; Schleicher& Schuell) by spotting a 0.01- to 10-ml aliquot of the
sample onto amembrane square (0.25–100 cm2) placed ona heating block set at 50 �C. Note that the binding ca-
pacity of the employed nitrocellulose is approximately
0.1mg protein/cm2, as indicated by the supplier.
In vitro phosphorylation
Liquid-phase phosphorylation was carried out in a
50-ll reaction mixture containing 20mM Tris–HCl, pH7.4, 10mM MgCl2, 30mM b-mercaptoethanol, and
0.2mM [c-32P]ATP (1000 cpm/pmol), in the presence of
0.1 lg of GST–c-PAK and 0.5 lg of GTP–Cdc42 [15].
The reaction was terminated after 30min incubation at
30 �C by addition of 2� sample buffer (50mM Tris–
HCl, pH 6.8, 2% SDS, 10% glycerol, and 30mM b-mercaptoethanol), prior to SDS–PAGE on 12.5%
polyacrylamide gel and subsequent electrotransfer ontonitrocellulose. Solid-phase phosphorylation was carried
out in 0.05–5ml of the above-described reaction buffer
in the presence of 0.1–10 lg of GST–c-PAK, 0.5–50 lgof GTP–Cdc42 [15], and the additional nitrocellulose-
bound protein sample. The reaction was terminated by
extensive washing of the nitrocellulose square in double-
distilled water. When indicated, preactivation of GST–
c-PAK was carried out in presence of GTP–Cdc42 andunlabeled ATP, prior to solid-phase kinase assay with
[32P]ATP. This step was included to reduce the incor-
poration of radioactivity on GST–c-PAK and enhance
the detection of labeled substrates.
Tryptic digestion and phosphopeptide mapping
Nitrocellulose-bound proteins were digested with10 ng/ll trypsin (DPCC-treated; Sigma) in 50mM am-
monium bicarbonate and 1% (w/v) Zwittergent 3-16
(Fluka) at 30 �C for 24 h. I have previously reported [10]
that under similar conditions of proteolytic digestion the
vast majority (approximately 90%) of the radiolabeled
phosphopeptides of immobilized substrates can be ex-
tracted from the nitrocellulose.
Tryptic digests were subjected to Sep-Pak cartridges(50mg; Waters) prior to 2-D peptide PAGE, as previ-
ously described [10]. The 2-D peptide PAGE consists of
nondenaturing isoelectric focusing in a gel tube as the
first dimension (native IEF) and electrophoresis on a
40% polyacrylamide alkaline slab gel as the second
dimension (40% PAGE). After 2-D peptide PAGE,
radiolabeled phosphopeptides were visualized by auto-
radiography of the dried 40% gel.
Isolation and purification of phosphopeptides
When using a purified protein as substrate of c-PAK,
the respective phosphopeptides were extracted from the
40% gel and directly subjected to a cleanup procedure
through Sep-Pak cartridges prior to manual and auto-
mated amino acid sequencing [10]. In contrast, two addi-tional steps of purification were required when the source
of substrates was a complex mixture of proteins such as
the DEAE eluate of PRS. Prior to the cleanup through a
Sep-Pak cartridge, the sample eluted from the 40% gel
was subjected to an immobilized metal affinity chroma-
tography (IMAC), essentially as described by [16]. No-
tably, the inclusion of the IMAC step involved a minimal
loss of radioactivity (less than 20%), thus confirming theefficiency of such type of chromatography in the isolation
of phosphopeptides from relatively complex protein di-
gests [17]. Briefly, 200 ll of Chelating Sepharose (Phar-
macia) was packed onto a column and washed with
double-distilled water. The column was then activated
with 30mM FeCl3 until metal appeared in the eluate.
After column equilibration with 0.1% trifluoroacetic
A. Gatti / Analytical Biochemistry 312 (2003) 40–47 41
acid (TFA), the sample eluted from the 40% gelwas loaded and the column was washed with 0.1%
TFA (10 vol), prior to elution of phosphopeptides
in 100mM phosphate buffer, pH 10. After IMAC and
a sub-sequent cleanup through a Sep-Pak cartridge,
the isolated phosphopeptide was finally purified by re-
versed-phase high-performance liquid chromatography
(HPLC). The elution from HPLC was monitored by
checking optical density and radioactivity of the fractionsto isolate the radiolabeled phosphopeptide.
Manual and automated amino acid sequencing
Prior to amino acid sequencing, each phosphopeptide
was immobilized on a disk of arylamine membrane (Se-
quelon-AA; Millipore), as described by the supplier. The
disk was routinely cut into two parts; approximately 3/4of the disk was applied to an amino acid sequencer
(Procise 492; Applied Biosystems), while the remaining 1/
4 was subjected to manual Edman degradation as de-
scribed by others [18]. The computer program BLAST
was utilized to match the identified phosphopeptide with
sequences retrieved from the protein databases.
Results
Phosphorylation of known substrates
A comparative analysis of conventional and solid-
phase phosphorylation of an established in vitro sub-
strate of c-PAK (i.e., MBP) was carried out to validate
the choice of using the nitrocellulose-immobilized pro-teins as source of substrates of the purified c-PAK. Note
that to minimize the impact of c-PAK autophosphory-
lation on the total incorporation of radioactivity, a
preincubation of c-PAK with Cdc42 and unlabeled ATP
(preactivation) was routinely carried out prior to the
addition of ½32P]ATP.When performing the liquid-phase kinase assay, a
25-lg aliquot of purified MBP was incubated with pre-activated c-PAK in the presence of [32P]ATP. After in-
cubation, samples were subjected to SDS–PAGE,
followed by protein transfer onto nitrocellulose. Auto-
radiography of the resulting blot membrane allowed
excision of the nitrocellulose band containing the
phosphorylated substrate, prior to solid-phase tryptic
digestion as previously described [10]. When performing
the solid-phase kinase assay, 25 lg of MBP was spottedonto a 0.25-cm2 nitrocellulose square placed on a heat-
ing block set at 50 �C. The nitrocellulose square was thensubjected to a brief washing with double-distilled water
and incubated with preactivated c-PAK in the presence
of [32P]ATP. After incubation, the nitrocellulose square
was extensively washed to remove free radioactivity and
then subjected to solid-phase tryptic digestion [10].
Tryptic digests were subjected to 2-D peptide PAGEto fractionate the radiolabeled phosphopeptides, as
previously described [10]. As shown in Fig. 1, the 2-D
phosphopeptide maps resulting from liquid- (Fig. 1A)
and solid-phase (Fig. 1B) phosphorylation of MBP by
c-PAK are indistinguishable. Manual and automated
sequencing of the isolated phosphopeptides demonstrate
identical specificity and similar efficiency of MBP
phosphorylation (Fig. 2). In particular, microsequencinganalysis of phosphopeptides 1 and 2 resulting from both
liquid- and solid-phase kinase assay showed that Thr-33
and Ser-114 are the major sites of MBP being phos-
phorylated by c-PAK, regardless of the kinase assay
conditions. Incidentally, Thr-33 and Ser-114 have been
described as major phosphorylation sites of MBP fol-
lowing conventional kinase assay with the catalytic do-
main of c-PAK [19].To verify that a similar conclusion could also apply to
other c-PAK substrates, a comparative analysis of solid-
and liquid-phase phosphorylation was also carried out
with H4. As in case of MBP, the specificity of H4
phosphorylation by c-PAK was found to be indepen-
dent of the type of kinase assay (A. Gatti, unpublished
observation). In summary, the overall analysis of the in
vitro phosphorylation of two established substrates of c-PAK validates the choice of using the solid-phase assay
as an alternative to the conventional, liquid-phase
in vitro kinase assay.
Phosphorylation of unknown substrates and identification
of the respective phosphopeptides
The possibility of employing a solid-phase kinaseassay to screen a complex mixture of proteins was next
examined. In particular, the DEAE eluate of PRS
Fig. 1. Comparison between liquid-phase and solid-phase kinase as-
says. A 25-lg aliquot of purified MBP from bovine brain was either
subjected to liquid-phase kinase assay prior to 12.5% SDS–PAGE and
subsequent transfer to nitrocellulose (A) or directly spotted onto a
nitrocellulose square prior to solid-phase kinase assay (B). After
tryptic proteolysis of the nitrocellulose-bound MBP, the same amount
of tryptic digest was loaded onto the 2-D peptide PAGE. The frac-
tionated 32P-labeled phosphopeptides were visualized upon autoradi-
ography of the dried peptide gel. Only the relevant sections of the
autoradiograms with the major phosphopeptides are shown. Similar
results were obtained in two additional experiments.
42 A. Gatti / Analytical Biochemistry 312 (2003) 40–47
obtained from rabbit reticulocytes was chosen as thesource of candidate substrates of c-PAK because such
sample was previously reported to be rich in various
substrates of c-PAK [8]. Initially, a 0.2-ml aliquot
(200 lg of protein) of the DEAE eluate was spotted onto
a 1-cm2 nitrocellulose square prior to a solid-phase ki-
nase assay with a preactivated c-PAK in the presence of
[32P]ATP. Subsequent solid-phase tryptic digestion was
followed by 2-D peptide PAGE, whose autoradiographyrevealed multiple phosphopeptides (Fig. 3C).
To verify that the detected phosphopeptides derived
from the cleavage of substrates, a parallel solid-phase
kinase assay was carried out in the absence of DEAE
eluate (Fig. 3B). The resulting 2-D profile of tryptic digest
resulting from such incubation of preactivated c-PAKwith control nitrocellulose (i.e., not loaded with proteins)
showed few barely detectable spots, presumably dueto the autophosphorylation of nitrocellulose-bound
c-PAK. To verify that the assigned phosphopeptides
shown in Fig. 3C resulted from the kinase activity of
c-PAK rather than from the activity of kinases present in
the DEAE eluate, a parallel sample of DEAE eluate
of PRS was subjected to a solid-phase kinase assay in
the absence of exogenously added c-PAK (Fig. 3A).
After 2-D peptide PAGE of the resulting tryptic digest,
no radiolabeled peptides were detected within the routine
exposure time, thus ruling out the relevance of nonspe-
cific incorporation of radiolabel under the describedconditions.
Next, I tested the possibility of scaling up this
experimental approach to identify the target sequences
of unknown substrates of c-PAK. To this aim, the
Fig. 2. Manual and automated sequencing of phosphopeptides 1 and 2 resulting from liquid- and soild-phase phosphorylation of MBP. After 2-D
peptide PAGE, the radiolabeled phosphopeptides were isolated and bound to Sequelon-AA disks, prior to manual and automated sequencing. The
cycle at which 32P was released by manual Edman degradation and the amino acid sequence identified by automated sequencing are indicated,
together with the relevant sequence of MBP. X designates amino acid residues that could not be unambiguously identified by automated sequencing.
The underlined residues are those found to be phosphorylated.
Fig. 3. Solid-phase phosphorylation of a complex mix of proteins. A
200-lg aliquot of proteins from DEAE eluate of PRS from rabbit
reticulocytes was directly spotted onto a 1-cm2 nitrocellulose square
prior to solid-phase kinase assay in the presence of preactivated c-PAK(C). Parallel assays were also carried out either without the exoge-
nously added c-PAK (A) or without the DEAE eluate (B). Results
shown are representative of two independent experiments.
A. Gatti / Analytical Biochemistry 312 (2003) 40–47 43
procedure consisted of the eight steps depicted in Fig. 4:(1) preparation of crude partially purified cell extract,
(2) immobilization of such cell extract onto nitrocellu-
lose membrane, (3) solid-phase phosphorylation by pu-
rified and activated c-PAK in the presence of [32P]ATP,
(4) solid-phase tryptic digestion, (5) cleanup and con-
centration of peptides by using a Sep-Pak cartridge, (6)
peptide fractionation on 2-D peptide PAGE, (7) ex-
traction and purification of individual phosphopeptidesby sequential IMAC/Sep-Pak/HPLC chromatographies,
and (8) manual and automated amino acid sequencingof the HPLC-purified phosphopeptides.
A 10-ml aliquot (10mg of proteins) of the DEAE el-
uate of PRS from rabbit reticulocytes was the starting
material used in the above operations. A major 32P-la-
beled peptide, corresponding to the phosphopeptide a of
Fig. 3C, was extracted from the second dimensional gel of
the preparative 2-D peptide PAGE (data not shown) and
isolated via the described means of peptide purificationprior to automated andmanual Edman degradation. The
Fig. 4. Schematics of the overall procedure. At the end of the operations here outlined, phosphopeptide a was identified as the labeled sequence of
c-PAK-phosphorylated TDX-1 (see Fig. 5).
44 A. Gatti / Analytical Biochemistry 312 (2003) 40–47
automated analysis revealed the amino acid sequence ofphosphopeptide a, which exhibits a phosphorylatable
residue in the position of the single phosphorylation site
detected by manual sequencing (Fig. 5). Underscoring
the high purity of the HPLC-purified peptide, the auto-
mated sequencing indicated that the noise due to the
contaminating amino acid residues represented less than
15% of the signal of the identified residues.
Database searching revealed the high degree of ho-mology between the identified amino acid residues of
phosphopeptide a and a stretch of sequence of human
thioredoxin peroxidase 1 (TDX1), a protein with per-
oxidase activity expressed in red blood cells in a variety
of isoforms. Given that the sequence of the rabbit iso-
form of TDX1 is not available in the current databases,
the exact degree of homology between the phospho-
peptide a and the corresponding sequence of rabbitTDX1 could not be calculated. However, the identity of
9 of the first 11 amino acid residues provides a signifi-
cant level of confidence. Hence, rabbit TDX1 is likely to
represent an in vitro substrate of c-PAK. The presence
of a tryptic site (Arg-110) in TDX1 in a position im-
mediately upstream of the identified sequence further
supports this conclusion. The two sites that could not be
unequivocally identified by automated sequencing con-sist of the radiolabeled residue and the glutamic acid
residue, both of which are routinely underdetected bythis type of analysis [10].
Discussion
To better understand the molecular basis of the ex-
traordinary specificity underlying cell signaling, it is
important to investigate how the activation of a par-ticular signaling pathway can selectively recruit multiple
effectors. Due to the difficulty of deciphering the com-
plex network of interactions between the protein kinases
and their respective substrates, an experimental system
that facilitates the profiling of substrate phosphoryla-
tion is in high demand. The use of antibodies raised
against phosphorylation sites has recently emerged as an
elective tool in the analysis of the in vivo activity ofprotein kinases. In fact, phosphorylation site-specific
antibodies against specific amino acid sequences allowed
the identification of many substrates. However, the po-
tential use of such antibodies is somewhat limited to
those instances in which critical information on the
nature of the phosphorylation sites is available.
On the contrary, the method presented here does not
depend on any preliminary knowledge, thereby allowingan unbiased search for substrates. With respect to this,
another method based on the unbiased screening of un-
known proteins as potential substrates of individually
examined protein kinases has been recently developed
[20]. Different from the procedure here described, in such
study the target sequences are fractionated and purified
prior to, rather than after, their proteolytic cleavage.
Considering that most recently developed approaches toprotein identification are based on enzymatic fragmen-
tation of the examined proteins [21], the possibility of
profiling substrate phosphorylation at the phosphopep-
tide level—rather than at the phosphoprotein level—is of
clear interest. With respect to this, note that here I opted
for the use of trypsin as protease because trypsin effi-
ciently cleaves the majority of proteins at many sites,
thereby leading to the generation of conveniently sizedpeptides (in relation to the following peptide fraction-
ation/sequencing). However, proteases other than trypsin
(e.g., V8 and Lys-C endoproteases) may be employed to
tailor the size of the resulting peptides to meet specific
requirements.
The serine/threonine kinase PAK has been impli-
cated in a number of biochemical scenarios, involving
cell proliferation, growth arrest [22–24], and responseto stress and apoptotic conditions [25–28]. In the
course of the study of how multisite autophospho-
rylation of the ubiquitous c-PAK correlates to its
activation, a 2-D peptide PAGE system was recently
developed to fractionate the phosphopeptides resulting
from solid-phase digestion of radiolabeled proteins [10,
12]. The high degree of resolution and reproducibility
Fig. 5. Microsequencing of phosphopeptide a. A 10-mg aliquot (10ml)
of proteins from DEAE eluate of PRS was spotted onto a 100-cm2
nitrocellulose square and subjected to solid-phase kinase assay in the
presence of preactivated c-PAK. After tryptic digestion, peptides wereconcentrated by using a Sep-Pak cartridge and fractionated by 2-D
peptide PAGE. Phosphopeptide a was isolated and purified by se-
quential IMAC, Sep-Pak, and a reversed-phase HPLC, prior to
manual and automated sequencing. The cycle at which 32P was re-
leased by manual Edman degradation and the amino acid sequence
identified by automated sequencing are indicated, together with the
sequence of human TDX1 retrieved from a Blast search as the best
matching protein. X designates amino acid residues that could not be
unambiguously identified by automated sequencing. Results shown are
representative of two independent experiments.
A. Gatti / Analytical Biochemistry 312 (2003) 40–47 45
by which 2-D profiles of protein digests are generatedby this experimental system, combined with the possi-
bility of identifying the fractionated phosphopeptides,
prompted me to use this peptide mapping system to
profile the kinase activity of c-PAK on unknown
substrates.
The idea of using a solid-phase kinase assay to profile
substrate phosphorylation was based on the seminal
report of Valtorta et al. [3], according to which thepatterns of protein phosphorylation resulting from the
use of nitrocellulose-immobilized substrates are similar
to those obtained with the conventional liquid-phase
kinase assay [3]. In the present study, the respective sites
of phosphorylation of MBP (Figs. 1 and 2) and H4 (data
not shown) were found to be equally targeted upon
solid- and liquid-phase kinase assay, thus ruling out the
possibility of an artifactual phosphorylation of nitro-cellulose-immobilized substrates.
Notably, the solid-phase kinase assay may circum-
vent several problems routinely associated with con-
ventional in vitro kinase assays. In liquid-phase kinase
assay, both intramolecular and intermolecular interac-
tions may mask relevant domains of potential sub-
strates. This may be the case for enolase, whose in vitro
phosphorylation by src takes place exclusively underacidic conditions [29]. Compared to soluble proteins,
immobilized proteins are more readily screened by
in vitro kinase assay. In fact, nitrocellulose-bound pro-
teins can be more concentrated per unit of reaction
volume and the same [32P]ATP-containing incubation
buffer can be used with multiple batches of nitrocellu-
lose-bound proteins. Compared to proteins in solution,
immobilized proteins are more effectively digested byendoproteases such as trypsin.
Note that in the present study the protein sample was
directly spotted onto nitrocellulose, rather than being
run on SDS–PAGE and blotted onto the membrane as
originally described by Valtorta et al. [3]. In addition to
allowing larger amounts of candidate substrates to be
screened, direct spotting of proteins onto nitrocellulose
avoids the typical problems of electroblotting such asthe limited recovery of high-MW proteins.
A systematic effort to optimize the overall procedure
for preparative purposes lay outside the scope of this
study. When selecting for the final step of peptide
identification, it is reasonable to take into account the
recent progress in the sensitivity and capability of mass
spectrometry (MS) technologies. Approximately 1 pmol
of a given phosphopeptide is required for automatedamino acid sequencing, while a sensitivity extending into
the low-femtomole range is becoming common for
MS-based technologies [30]. Regardless of whether the
final analysis is automated Edman degradation or a
MS-based technology, the sequencing of the isolated
phosphopeptides together with the finding of phosphory-
latable residues in the position of sites of phosphorylation
(being assessed by manual sequencing) is likely to besufficiently informative for the identification of the
respective substrates.
In the case of phosphopeptide a, a database search
revealed high homology with an antioxidant enzyme,
namely TDX1. Note that the presumed site of TDX1
phosphorylation is fully consistent with the recognition
determinant [(K/R)RX(S/T)] of c-PAK phosphorylation
previously described [9]. TDX1, originally termed nat-ural killer enhancing factor-B, belongs to a highly con-
served and widely expressed family of antioxidants
known to be present in red blood cells [31]. TDXs rep-
resent the enzymes that link the reduction of H2O2 to
thioredoxin, thereby protecting cells from apoptosis in-
duced by serum deprivation and etoposide [32]. Inter-
estingly, by using a modified protocol of the present
procedure, a peptide from an additional protein withperoxidase activity (i.e., catalase) was also identified as
in vitro target sequence of c-PAK (A. Gatti, unpub-
lished observation). Taking into consideration that c-PAK has been recently found to be activated by stimuli
leading to cytostatic and apoptotic conditions [33,34], a
role for c-PAK in the regulation of cellular redox can be
anticipated.
A number of substrates of protein kinases have beenidentified upon in vitro screening of cDNA expression
libraries and subsequent DNA sequencing of positive
clones [1]. Such strategy is based on the use of nitro-
cellulose to immobilize the expressed proteins to be
screened with appropriately chosen molecular probes
(e.g., antibodies, purified kinases, binding proteins, etc.).
Compared to the use of cDNA libraries to express large
amounts of candidate substrates, the major limitation ofthe present procedure is the limited abundance of can-
didate substrates. This problem can be circumvented
either by using selective chromatographic steps as pro-
tein enrichment steps [35,36] or by taking advantage of
more sensitive sequencing technologies (as discussed
above).
In conclusion, profiling substrate phosphorylation at
the phosphopeptide level together with sequence data-base searching may represent an interesting opportunity
in the forthcoming scenario of functional proteomics.
As a variety of protein array applications is progres-
sively becoming available, the adaptation of this exper-
imental approach to enable global analysis of substrate
phosphorylation is envisaged.
Acknowledgments
The experimental work for this project was carried
out in the laboratory of Dr. Jolinda A. Traugh. Many
thanks also to Barbara Walter for the purification of
baculovirus-expressed GST–c-PAK and to Kevin Orton
for the HPLC of phosphopeptide a.
46 A. Gatti / Analytical Biochemistry 312 (2003) 40–47
References
[1] F.A. Al-Obeidi, J.J. Wu, K.S. Lam, Protein tyrosine kinases:
structure, substrate specificity, and drug discovery, Biopolymers
47 (1998) 197–223.
[2] R. Fukunaga, T. Hunter, MNK1, a new MAP kinase-activated
protein kinase, isolated by a novel expression screening method
for identifying protein kinase substrates, EMBO J. 16 (1997)
1921–1933.
[3] F. Valtorta, W. Schiebler, R. Jahn, B. Ceccarelli, P. Greengard, A
solid-phase assay for the phosphorylation of proteins blotted on
nitrocellulose membrane filters, Anal. Biochem. 158 (1986) 130–
137.
[4] M. Teo, E. Manser, L. Lim, Identification and molecular cloning
of a p21cdc42/rac1-activated serine/threonine kinase that is
rapidly activated by thrombin in platelets, J. Biol. Chem. 270
(1995) 26690–26697.
[5] G.A. Martin, G. Bollag, F. McCormick, A. Abo, A novel serine
kinase activated by rac1/CDC42Hs-dependent autophosphoryla-
tion is related toPAK65andSTE20,EMBOJ. 14 (1995) 1970–1978.
[6] U.G. Knaus, S. Morris, H.J. Dong, J. Chernoff, G.M. Bokoch,
Regulation of human leukocyte p21-activated kinases through G
protein-coupled receptors, Science 269 (1995) 221–223.
[7] L. Lim, E. Manser, T. Leung, C. Hall, Regulation of phosphor-
ylation pathways by p21 GTPases. The p21 Ras-related Rho
subfamily and its role in phosphorylation signalling pathways,
Eur. J. Biochem. 242 (1996) 171–185.
[8] S.M. Tahara, J.A. Traugh, Cyclic Nucleotide-independent protein
kinases from rabbit reticulocytes. Identification and characteriza-
tion of a protein kinase activated by proteolysis, J. Biol. Chem.
256 (1981) 11558–11564.
[9] P.T. Tuazon, W.C. Spanos, E.L. Gump, C.A. Monnig, J.A.
Traugh, Determinants for substrate phosphorylation by p21-
activated protein kinase (gamma-PAK), Biochemistry 6 (1997)
16059–16064.
[10] A. Gatti, J.A. Traugh, A two-dimensional peptide gel electro-
phoresis system for phosphopeptide mapping and amino acid
sequencing, Anal. Biochem. 266 (1999) 198–204.
[11] A. Gatti, K.C. Menes, J.A. Traugh, Solid-phase labeling with a
fluorescent reagent to fingerprint nonradioactive proteins, Anal.
Biochem. 280 (2000) 189–192.
[12] A. Gatti, Z. Huang, P.T. Tuazon, J.A. Traugh, Multisite
autophosphorylation of p21-activated protein kinase c-PAKas a function of activation, J. Biol. Chem. 274 (1999) 8022–
8028.
[13] A. Gatti, B.N. Walter, J.A. Traugh, Identification of sites on p21-
activated protein kinase c-PAK phosphorylated in vivo, in
preparation.
[14] G.M. Hathaway, T.S. Lundak, S.M. Tahara, J.A. Traugh,
Isolation of protein kinases from reticulocytes and phosphoryla-
tion of initiation factors, Methods Enzymol. 60 (1979) 495–511.
[15] R. Jakobi, C.J. Chen, P.T. Tuazon, J.A. Traugh, Molecular
cloning and sequencing of the cytostatic G protein-activated
protein kinase PAK, J. Biol. Chem. 271 (1996) 6206–6211.
[16] L. Andersson, J. Porath, Isolation of phosphoproteins by immo-
bilized metal (Fe3þ) affinity chromatography, Anal. Biochem. 154
(1986) 250–254.
[17] M.C. Posewitz, P. Tempst, Immobilized gallium(III) affinity
chromatography of phosphopeptides, Anal. Chem. 71 (1999)
2883–2892.
[18] S. Sullivan, T.W. Wong, A manual sequencing method for
identification of phosphorylated amino acids in phosphopeptides,
Anal. Biochem. 197 (1991) 65–68.
[19] S.D. Yang, T.J. Huang, J.J. Huang, Autophosphorylation-
dependent protein kinase predominantly phosphorylates Ser115,
the in vivo site in brain myelin basic protein, J. Protein Chem. 13
(1994) 599–607.
[20] A. Knebel, N. Morrice, P. Cohen, A novel method to identify pro-
tein kinase substrates: eEF2 kinase is phosphorylated and inhibited
by SAPK4/p38?, EMBO J. 20 (2001) 4360–4369.
[21] H. Erdjument-Bromage, M. Lui, L. Lacomis, A. Grewal, R.S.
Annan, D.E. McNulty, S.A. Carr, P.J. Tempst, Examination of
micro-tip reversed-phase liquid chromatographic extraction of
peptide pools for mass spectrometric analysis, J. Chromatogr. A
826 (1998) 167–181.
[22] M.L. Galisteo, J. Chernoff, Y.C. Su, E.Y. Skolnik, J. Schlessinger,
The adaptor protein Nck links receptor tyrosine kinases with the
serine-threonine kinase Pak1, J. Biol. Chem. 271 (1996) 20997–
21000.
[23] S. Dharmawardhane, L.C. Sanders, S.S. Martin, R.H. Daniels,
G.M. Bokoch, Localization of p21-activated kinase 1 (PAK1) to
pinocytic vesicles and cortical actin structures in stimulated cells,
J. Cell Biol. 138 (1997) 1265–1278.
[24] R.H. Daniels, P.S. Hall, G.M. Bokoch, Membrane targeting of
p21-activated kinase 1 (PAK1) induces neurite outgrowth from
PC12 cells, EMBO J. 17 (1998) 754–764.
[25] S. Zhang, J. Han, M.A. Sells, J. Chernoff, U.G. Knaus, R.J. Ulev-
itch, G.M. Bokoch, Rho family GTPases regulate p38 mitogen-
activated protein kinase through the downstreammediator Pak1, J.
Biol. Chem. 270 (1995) 23934–23936.
[26] B.N. Walter, Z. Huang, R. Jakobi, P.T. Tuazon, E.S. Alnemri, G.
Litwack, J.A. Traugh, Cleavage and activation of p21-activated
protein kinase gamma-PAK by CPP32 (caspase 3). Effects of
autophosphorylation on activity, J. Biol. Chem. 273 (1998) 28733–
28739.
[27] T. Rudel, G.M. Bokoch, Membrane and morphological changes
in apoptotic cells regulated by caspase-mediated activation of
PAK2, Science 276 (1997) 1571–1574.
[28] N. Lee, H. MacDonald, C. Reinhard, R. Halenbeck, A. Roulston,
T. Shi, L.T. Williams, Activation of hPAK65 by caspase cleav-
age induces some of the morphological and biochemical
changes of apoptosis, Proc. Natl. Acad. Sci. USA 94 (1997)
13642–13647.
[29] J. Cooper, F.S. Esch, S.S. Taylor, T. Hunter, Phosphorylation
sites in enolase and lactate dehydrogenase utilized by tyrosine
protein kinases in vivo and in vitro, J. Biol. Chem. 259 (1984)
7835–7841.
[30] S.P. Gygi, D.K.M. Han, A.C. Gingras, N. Sonenberg, R.
Aebersold, Protein analysis by mass spectrometry and sequence
database searching: tools for cancer research in the post-genomic
era, Electrophoresis 20 (1999) 310–319.
[31] O. Golaz, G.J. Hughes, S. Frutiger, N. Paquet, A. Bairoch, C.
Pasquali, J.C. Sanchez, J.D. Tissot, R.D. Appel, C. Walzer, D.F.
Hochstrasser, Plasma and red blood cell protein maps: update,
Electrophoresis 14 (1993) 1223–1231.
[32] P. Zhang, B. Liu, S.W. Kang, M.S. Seo, S.G. Rhee, L.M. Obeid,
Thioredoxin peroxidase is a novel inhibitor of apoptosis with a
mechanism distinct from that of Bcl-2, J. Biol. Chem. 272 (1997)
30615–30618.
[33] R.D. Rooney, P.T. Tuazon, W.E. Meek, E.J. Carroll, J.J. Hagen,
E.L. Gump, C.A. Monnig, T. Lugo, J.A. Traugh, Cleavage arrest
of early frog embryos by the G protein-activated protein kinase
PAK I, J. Biol. Chem. 271 (1996) 21498–21504.
[34] J. Roig, J.A. Traugh, p21-activated protein kinase gamma-PAK is
activated by ionizing radiation and other DNA-damaging agents.
Similarities and differences to alpha-PAK, J. Biol. Chem. 274
(1999) 31119–31122.
[35] K. Karlsson, N. Cairns, G. Lubec, M. Fountoulakis, Enrichment
of human brain proteins by heparin chromatography, Electro-
phoresis 20 (1999) 2970–2976.
[36] X. Wang, S. Bruderer, Z. Rafi, J. Xue, P.J. Milburn, A. Kramer,
P.J. Robinson, Phosphorylation of splicing factor SF1 on Ser20
by cGMP-dependent protein kinase regulates spliceosome assem-
bly, EMBO J. 18 (1999) 4549–4559.
A. Gatti / Analytical Biochemistry 312 (2003) 40–47 47