Thioredoxin h regulates calcium dependent protein kinases in plasma membranes


  • [the copyright line has been changed since first publication on 14 Jun 2013]


Thioredoxin (Trx) is a key player in redox homeostasis in various cells, modulating the functions of target proteins by catalyzing a thiol–disulfide exchange reaction. Target proteins of cytosolic Trx-h of higher plants were studied, particularly in the plasma membrane, because plant plasma membranes include various functionally important protein molecules such as transporters and signal receptors. Plasma membrane proteins from Arabidopsis thaliana cell cultures were screened using a resin Trx-h1 mutant-immobilized, and a total of 48 candidate proteins obtained. These included two calcium-sensing proteins: a phosphoinositide-specific phospholipase 2 (AtPLC2) and a calcium-dependent protein kinase 21 (AtCPK21). A redox-dependent change in AtCPK21 kinase activity was demonstrated in vitro. Oxidation of AtCPK21 resulted in a decrease in kinase activity to 19% of that of untreated AtCPK21, but Trx-h1 effectively restored the activity to 90%. An intramolecular disulfide bond (Cys97–Cys108) that is responsible for this redox modulation was then identified. In addition, endogenous AtCPK21 was shown to be oxidized in vivo when the culture cells were treated with H2O2. These results suggest that redox regulation of AtCPK21 by Trx-h in response to external stimuli is important for appropriate cellular responses. The relationship between the redox regulation system and Ca2+ signaling pathways is discussed.

Structured digital abstract


Arabidopsis thaliana calcium-dependent protein kinase


Arabidopsis thaliana phosphoinositide-specific phospholipase


calcium-dependent protein kinase




poly(ethylene glycol)




Thioredoxin (Trx) is a small, ubiquitous, redox-responsive protein that possesses a pair of cysteine residues at the active site in the highly conserved motif Trp-Cys-[Gly/Pro]-Pro-Cys [1-3]. Trx modulates the function of the target proteins by catalyzing the thiol–disulfide exchange reaction. Higher plants contain multiple Trx isoforms in the cytosol, chloroplasts and mitochondria; nine genes encoding Trx-h have been identified in Arabidopsis thaliana [4, 5], although differences in the function and substrate specificity of these isoforms remain to be elucidated [6]. Among them, cytosolic Trx-h isoforms are known to be reduced by cytosolic NADPH-thioredoxin reductase, which uses NADPH as a source of reducing equivalent [7]. However, the NADP+/NADPH ratio in the cytosol of pea leaves (Pisum sativum), for example, changes only from 1 to 1.3 following a light/dark transition [8]. Thus, in contrast to chloroplast Trx, the redox state of cytosolic Trx-h isoforms may be maintained at a certain level during light/dark transitions. To examine how Trx-h isoforms interact with their target proteins under certain conditions, it is important to understand the physiological significance of Trx-h as a transducer of reducing equivalents or a cytosolic regulator.

Several studies to identify and characterize the cytosolic targets of Trx-h isoforms have been performed [6, 9-13]. However, although functionally important molecules such as signal receptors and transporters exist in membranes, little is known about Trx-h targets in plasma membranes. In order to identify Trx-h1 target proteins, we screened plant plasma membrane proteins using Trx mutant affinity chromatography, in which the mutant associates more stably with target proteins than the wild-type [14], and identified the acquired proteins by using MALDI-TOF/TOF MS. Within the isolated interacting proteins, two calcium-sensing proteins, a phosphoinositide-specific phospholipase 2 (AtPLC2) and a calcium-dependent protein kinase 21 (AtCPK21), were identified, which were anchored to the plasma membrane as Trx-h1 targets.

Calcium-dependent protein kinase (CDPK) is a member of the family of calcium-sensing proteins. Protein phosphorylation by this enzyme plays a pivotal role in amplifying and diversifying the action of calcium-mediated signals. CDPKs are Ser/Thr protein kinases that are only found in plants and some protozoans. They consist of four distinct domains in a single polypeptide: an N-terminal variable domain, a catalytic domain, an auto-inhibitory domain and a C-terminal calmodulin-like domain. The CDPK protein directly binds calcium on the calmodulin-like domain, and this calcium binding induces activation of the enzyme [15, 16].

The model plant Arabidopsis thaliana possesses 34 genes that encode CDPKs (AtCPKs) [4, 17]. Although some of the AtCPKs were found not to be membrane-integral proteins, 24 are predicted to be modulated by N-terminal N-myristoylation and/or C-palmitoylation. The lipid modifications explain their targeting to multiple subcellular sites, including the nucleus, plasma membrane, endoplasmic reticulum, peroxisome, mitochondrial outer membrane and oil body, as well as the cytosol [15, 17, 18]. CDPKs have been reported to be involved in diverse physiological processes of plants. However, a thorough understanding of the distinct functions of each individual CDPK remains elusive. The functional specificities of each CDPK require clarification by analyzing regulation at both transcriptional and post-translational levels as well as their subcellular localization, calcium and lipid sensitivity, and substrate recognition. Following capture of an AtCPK by Trx affinity chromatography, redox regulation of AtCPKs by a Trx was examined in order to better understand the functional aspects of the molecule.

The in vitro redox regulation of AtCPK21 by Trx-h1 and identification of the intramolecular disulfide bond responsible for the modulation are described here. In addition, AtCPK21 was shown to be oxidized in A. thaliana cell cultures when the cells are treated with hydrogen peroxide for several hours, suggesting a response of AtCPK21 to external oxidative stimuli.


Screening of Trx-h1 targets in the plasma membrane

To identify potential target proteins of Trx-h1, plasma membrane proteins of A. thaliana were screened using an immobilized Trx-h1C43S mutant whose active site Cys43 was substituted with Ser [6]. Proteins in the plasma membranes were then solubilized using the detergent n-dodecyl-β-d-maltoside, and incubated with Trx-h1C43S immobilized on Sepharose beads. After thorough washing of the gel, the proteins bound to the Trx-h1C43S mutant via intermolecular disulfide bonds were eluted from the gel by reduction with dithiothreitol (DTT). The proteins included in the eluate showed very different band patterns on SDS/PAGE from those in the solubilized and flow-through fractions, suggesting specific acquisition of particular proteins from the solubilized fraction by this method (Fig. 1). The protein bands separated by SDS/PAGE were analyzed by MALDI-TOF/TOF MS. In total, 48 proteins were identified as candidates (Table S1). Within these identified proteins, we focused on AtCPK21, which is known to be a calcium-sensing protein (Fig. 1 and Table 1).

Table 1. Identification of Trx-h1 target proteins related to calcium signaling
NomenclatureaGene codebMolecular weight (kDa)cConserved Cys/total CysdDeterminantse
  1. a The nomenclature, b gene code and c molecular weight are from the Arabidopsis Information Resource (

  2. d Number of the conserved cysteine residues per total number of the cysteine residues in each protein.

  3. e Determinants are amino acid sequences of the representative tryptic peptides determined by MALDI-TOF/TOF MS.

Phosphoinositide-specific phospholipase C2 (AtPLC2)AT3G0851066.12/7GKDLGDEEVWGR(265–276)
Calcium-dependent protein kinase 21 (AtCPK21)AT4G0472059.84/6SIPINPVQTHVVPEHR(19–34)
Figure 1.

Screening of Trx-h1 targets in the plasma membrane of Arabidopsis thaliana. Proteins captured using Trx mutant-immobilized resin were separated by SDS/PAGE (7.5–15%) and stained with Coomassie Brilliant Blue G-250. Protein bands were identified by MALDI-TOF/TOF MS. The positions of the protein bands assigned to a phosphoinositide-specific phospholipase 2 (AtPLC2) and a calcium-dependent protein kinase 21 (AtCPK21) are indicated.

AtCPK21 contains four conserved cysteine residues, and has been reported to be anchored to the plasma membrane by an N-terminal lipid modification [15, 18]. Here we examined modulation of AtCPK21 by Trx-h1 in vitro because CDPKs were implicated in oxidative signal transduction in plants [19, 20].

Trx-h1 interacts with AtCPK21 in yeast

To assess interaction between Trx-h1 and AtCPK21, a yeast two-hybrid system for membrane-associated proteins was used (Fig. 2). As this split-ubiquitin system functions without transportation of the protein of interest to the nucleus, interaction of proteins of interest should be detected at cellular membranes. The yeast co-expressing AtCPK21 and Trx-h1 grew on selective plates (SC–His) but not on severely selective plates (SC–Ade) (Fig. 2). Co-expression of AtCPK21 with Trx-h1C43S resulted in better growth on selective SC–His plates than with the wild-type, suggesting that Trx-h1C43S interacts more stably with AtCPK21 than the wild-type does. This result confirms the interaction of AtCPK21 with AtTrx-h1.

Figure 2.

Interaction between AtCPK21 and AtTrx-h1 in yeast. In vivo protein–protein interaction was assayed using a yeast two-hybrid membrane protein system. The combinations of bait and prey were Alg5-Cub only (negative control), Alg5-Cub and Alg-NubI (positive control), AtCPK21 and AtTrx-h1 wild-type (test 1) and AtCPK21 and the AtTrx-h1C43S mutant (test 2). Dilution series of an overnight culture of the transformants (lanes 1–11, 20–210-fold dilution) and the dilution medium only (lane 12) were spotted on selection plates and incubated for 2 days at 30-C. The strength of the interaction was monitored using the HIS3 and ADE2 reporter genes. (A) Synthetic complete medium minus histidine (-His); (B) casamino acids medium minus adenine (-Ade); (C) casamino acids medium plus adenine (Control).

Trx-h1 reactivates the oxidized AtCPK21 in vitro

The effect of oxidation and reduction on AtCPK21 activity was examined. A calcium-dependent protein phosphorylation activity assay was performed using calf histone III-S as an artificial substrate. The initial specific activity of the reduced form (0.22 μmol/min/mg) was comparable to those of other CDPKs [21, 22]. The activity in the presence of EGTA was negligibly low under all conditions (< 0.002 μmol/min/mg). Oxidation by 5 μm CuCl2 decreased the kinase activity to 19% of that of the reduced form, and reduction of the oxidized form using 10 mm DTT restored the activity to 90% (Fig. 3A). These results clearly indicate that oxidation suppresses the enzyme activity, and this suppression is reversibly recovered by reduction. The ability of Trx-h1 to assist reactivation of the oxidized enzyme by reduction of intramolecular disulfide bonds was then examined. Although the activity of the oxidized form gradually recovered with increasing concentrations of DTT, addition of 6 μm Trx-h1 greatly enhanced recovery of the activity (Fig. 3B). For example, the activity recovered by addition of both Trx-h1 and 50 μm DTT reached 90%, but was only 42% when Trx-h1 was omitted. The concentration of Trx-h1 required for 50% recovery was < 1 μm in the presence of 10 μm DTT, suggesting high affinity of Trx-h1 for the AtCPK21 protein (Fig. 3C).

Figure 3.

Effect of Trx-h1 and DTT on the kinase activity of AtCPK21. (A) Kinase activity of the oxidized and reduced AtCPK21 was measured with 0.5 mg/mL calf histone III-S as substrate in the presence of 100 μm CaCl2 or 100 μm EGTA. The reduced form of AtCPK21 (Red) was oxidized by incubation with 5 μm CuCl2 (Ox), and then reduced using 10 mm DTT for 30 min at 25-C (Ox/DTT) prior to assay. (B) Effect of DTT concentration on recovery of the activity. The oxidized AtCPK was incubated with DTT at the concentrations indicated with 6 μm Trx-h1 (closed circle) or without (open circle) for 30 min at 25-C and then used in the assay. (C) Effect of various concentrations of Trx-h1 on recovery of the calcium-dependent protein kinase activity. The oxidized enzyme was treated with Trx-h1 in the presence of 10 μm DTT for 30 min at 25-C prior to the assay. Values are means of two independent experiments.

Cys97 and Cys108 form a disulfide bond in oxidized AtCPK21

AtCPK21 contains six cysteine residues in total, allowing formation of multiple inter- or intramolecular disulfide bonds under oxidizing conditions. Peptide mapping analysis was performed in order to specify which cysteine residues form disulfide bonds relevant to redox regulation of AtCPK21. Tryptic peptides obtained from oxidized AtCPK21 were separated by reversed-phase HPLC and analyzed by MALDI-TOF/TOF MS. A single redox-sensitive fragment with a molecular mass of 2301.07 m/z [Mr+H]+ was identified (Fig. 4A), which was specifically found in the oxidized form sample but not in the reduced form. When tryptic peptides containing this 2301.07 Da fragment were reduced using the reducing reagent Tris-(2-carboxyethyl) phosphine hydrochloride, the signal intensity of this fragment was attenuated by less than one-tenth of the initial intensity. In contrast, the signal intensity of fragments with molecular masses of 1147.58 and 1156.59 m/z [Mr+H]+ increased (Fig. 4B). The two fragments (1147.58 and 1156.59 m/z) were therefore analyzed using MS/MS spectrometry, and the internal sequences of these polypeptides were determined to be GQFGITYMCK(89–98) (molecular mass 1146.52 Da) and EIGTGNTYACK(99–109) (molecular mass 1155.52 Da), respectively (Fig. 4C,D), indicating the Cys97–Cys108 disulfide bond (Fig. 4D). This combination is quite feasible as these two cysteine residues are located in the catalytic domain. The predicted 3D structure model of the catalytic domain (Tyr80–Ile398) calculated using swiss-model software [23] on the basis of the crystal structure of Toxoplasma gondii calmodulin-domain protein kinase 1 (Fig. S1) clearly indicates location of these cysteines at the ATP-binding site. The distance between the sulfur atoms of the two cysteine residues in the predicted structure was calculated to be 3.89 Å using molmol software [24], i.e. they are close enough to each other to form a disulfide bond.

Figure 4.

Identification of disulfide bonds in oxidized AtCPK21. Tryptic peptides of the AtCPK21 oxidized by 5 μm CuCl2 were separated by HPLC and analyzed by MALDI-TOF/TOF MS. The indicated values are monoisotopic masses of the [Mr+H]+ ions. (A) Mass spectrum of the fraction containing a disulfide-linked peptide. (B) The fraction containing the disulfide-linked peptide (2301.07 m/z) was treated with a reducing reagent (50 mm Tris-(2-carboxyethyl) phosphine hydrochloride) and analyzed. (C, D) The resulting peptides with masses of 1147.58 m/z and 1156.59 m/z were fragmented. Attribution of the y ions is indicated by dashed lines, and the corresponding sequence of the peptide is shown at the top. (E) Cys97 and Cys108 form an intramolecular disulfide bond. The values in parentheses correspond to the theoretical monoisotopic mass of each peptide. Positions of the conserved cysteine residues in the secondary structure are indicated: A, auto-inhibitory domain; striped box, calmodulin-like Ca2+ -binding domain.

Endogenous CDPK is oxidized by H2O2 treatment in vivo

Finally, the ability of an endogenous AtCPK21 to be oxidized or reduced in vivo in response to external stimuli was examined. A. thaliana cell cultures were incubated with 10 mm H2O2 for 4 h in the dark, and proteins were directly extracted using trichloroacetic acid to avoid artificial oxidation of free thiol groups in the proteins during extraction [25]. The extracted proteins were labeled by PEG-maleimide (poly(ethylene glycol)- maleimide) to monitor in vivo redox states. In order to visualize the redox response of the endogenous AtCPK21, an IgG against AtCPK21 was prepared using the whole recombinant protein and affinity-purified as described in 'Experimental procedures'. The endogenous AtCPK21 was detected as a double band with apparent molecular mass of 61 kDa in the cell extracts by immunoblotting (Fig. 5B). An IgG against the N-terminal variable region of NtCDPK1 that recognized an ortholog of AtCPK21 in tobacco Nicotiana tabacum [26] also detected a doublet band in the cell extract in the same way (data not shown). The recombinant AtCPK21 was detected as a double band on SDS/PAGE and migrated either by autophosphorylation or by dephosphorylation (Fig. S2), suggesting possible autophosphorylation and/or phosphorylation of the endogenous protein. These results confirm that the anti-AtCPK21 IgG recognizes the endogenous AtCPK21. A redox response of endogenous AtCPK21 in A. thaliana cell cultures was detected. A band with an apparent molecular mass of 75 kDa was detected without H2O2 supplementation, representing addition of six molecules of PEG-maleimide (2 kDa) to the fully reduced AtCPK21 (increase in apparent molecular mass of 12 kDa), suggesting that the endogenous AtCPK21 is usually present in the reduced form (Fig. 5A). By oxidation with H2O2, multiple bands emerged gradually below the position of the fully reduced form, which represented addition of one to six molecules of PEG-maleimide per single protein, indicating that AtCPK21 was oxidized.

Figure 5.

Effect of H2O2 on the redox states of AtCPK21 in cultured cells. Five-day-old T87 cell cultures were incubated with 10 mm H2O2 under dark conditions for the indicated time period, and total proteins were extracted using trichloroacetic acid. (A) Free thiol groups of the proteins were labeled with PEG-maleimide (2 kDa), and the labeled proteins were subjected to immunoblotting using the anti-AtCPK21 IgG. (B) The proteins without PEG-maleimide labeling were subjected to immunoblotting.

The amount of the fully reduced form decreased by less than half after H2O2 treatment for 2 h (Fig. 5A). In contrast, the total amounts of endogenous AtCPK21 did not alter during H2O2 treatment for 4 h (Fig. 5B), suggesting that H2O2 treatment did not cause degradation of the enzyme molecule. In contrast, we found only weak and transient oxidation of AtCPK21 proteins by H2O2 treatment in cells maintained under light conditions (Fig. S3). We did not detect the oxidized form when using < 10 mm H2O2.

To confirm the oxidation of AtCPK21 in vitro, the recombinant protein was labeled with PEG-maleimide. Oxidation by 5 μm CuCl2 resulted in multiple bands below the position of fully reduced form on SDS/PAGE (Fig. S4). When the oxidized sample was reduced by Trx-h1 in the presence or absence of 50 μm DTT, most of the multiple bands migrated close to the fully reduced form, although the recovery was not always complete.


In this study, the potential targets of cytosolic Trx-h in plasma membranes of plant cells were initially investigated in order to uncover the role of Trx-h. A total of 48 proteins were identified as potential Trx-h1 target candidates in plasma membranes by Trx affinity chromatography [27] using Trx-h1C43S mutant-immobilized resin. All of the captured proteins contained multiple Cys residues in the molecule, although their relationship with Trx is still largely uncharacterized. Among these, a calcium-sensing protein AtCPK21 was of particular interest as such proteins are known to be important for the cellular signaling process.

AtCPK21 is one of 34 AtCPK isoforms that are encoded by a multiple gene family in A. thaliana [4, 15]. The amino acid sequences of AtCPKs are highly conserved, and their function, expression pattern and intercellular distribution are redundant [15, 16]. Interestingly, AtCPK21 was the only isoform of calcium-dependent protein kinase isolated in our experiments, and the reason for this remains elusive. This may be due to the abundance of AtCPK21 in the plasma membrane prepared from the cell cultures or easier ionization of the peptides derived from this protein molecule by mass spectrometry. However, we cannot exclude the possibility that other calcium-dependent protein kinase isoforms were detected by MS analysis, as three of the four peptides (Table 1) were common to multiple isoforms. It is also possible that AtCPK isoforms in general are potential targets of Trx-h1 because four Cys residues (Cys97, Cys108, Cys161 and Cys195) in the catalytic domain are conserved in almost all AtCPKs.

Of these four conserved Cys residues, the disulfide bond between Cys97 and Cys108 was suggested to be responsible for the redox sensitivity of AtCPK21 (Fig. 4). How does the Cys97–Cys108 disulfide bond modulate the kinase activity of AtCPK21? The predicted 3D structure also shows that Cys97 and Cys108 are located very close to the ATP-binding site, and this combination is therefore the most feasible (Fig. S1). The distance between the sulfur atoms of these two Cys residues is close enough to form a disulfide bond (3.89 Å) in the model, and conformational changes should make it possible to form a disulfide bond. However, formation of the disulfide bond in the molecule did not affect ATP binding itself (Fig. S5), although the Vmax value decreased by 18%, as in a chloroplast ATP synthase that was reversibly inactivated by formation of a disulfide bond in the γ subunit and reactivated by reduction by chloroplast Trx-f [28, 29]. The cores of eukaryotic protein kinases have common folds: the N-terminal lobe comprising five strands of antiparallel β-sheet and one α-helix, and the C-terminal lobe comprising a four α-helix bundle [30]. Because the ATP-binding site is situated at the interface of the two lobes, disulfide bond formation may interfere with the conformational change of the catalytic domain induced by ATP hydrolysis rather than ATP binding itself. Furthermore, the possibility cannot be excluded that additional intra- or intermolecular disulfide bonds in oxidized AtCPK21 may affect activity, as PEG-maleimide labeling resulted in multiple bands of the oxidized form on the immunoblot (Fig. 5). Further studies are required to clarify the relationship between these additional disulfide bonds and the redox regulation of AtCPK21.

An understanding of whether the inactivation of AtCPK21 actually occurs in vivo under oxidative stress conditions or other physiological conditions is critical in determining the physiological significance of the role of Trx in regulation of this protein kinase. For this purpose, A. thaliana cell cultures were treated with H2O2, because CDPKs have been implicated in the cellular response to external oxidative stimuli in soybean (Glycine max) and tobacco [19, 20]. As shown in Fig. 5A, oxidation of AtCPK21 was detected in cells treated with H2O2 for 2 h, as visualized by PEG-maleimide labeling. In contrast, no degradation of the desired protein was observed for at least 4 h under our experimental conditions (Fig. 5B), although 10 mm H2O2 caused cell death in soybean cells by 8 h [31]. In addition, only weak and transient oxidation of AtCPK21 was detected in cell cultures maintained in the light and treated with 10 mm H2O2 for 24 h under light conditions (Fig. S3). As photosynthesis in the chloroplasts produces reducing equivalents in the light, and a proportion of the reducing equivalents are transported from chloroplasts to the cytosol, they may contribute to maintaining cellular redox homeostasis, which is important in maintaining redox-sensitive enzymes as the reducing forms.

CDPKs are highly regulated molecules in the cell whose function is modulated not only by calcium but also by autophosphorylation and several phosphatidylinositols [17]. The data presented here indicate that redox modulation of AtCPK21 may be one of the diverse regulation systems of calcium signaling, which is especially sensitive to external stimuli such as oxidative stress in the dark, when cells do not have sufficient reducing equivalents to maintain redox enzymes. For example, StCDPK5 of potato (Solanum tuberosum) was reported to phosphorylate and activate NADPH oxidase in plasma membranes in response to pathogen attack. The activated NADPH oxidase produces reactive oxygen species using cytosolic NADPH, and the reactive oxygen species result in a plant immune response [20]. In addition, excess reactive oxygen species may result in irreversible oxidation and inactivation of thiol enzymes in the cytosol, including CDPK. Oxidation and inactivation of CDPKs may then act as a negative feedback loop in the oxidative signal pathway, depending on the cytosolic redox state or cytosolic NADP+/NADPH ratio. As the reduction level of cytosolic Trx-h is directly affected by the cytosolic NADP+/NADPH ratio, the redox balance in the cytosol may coordinate the activity of CDPKs as appropriate cellular responses to external stimuli such as oxidative stress.

In general, the plant intracellular environment is thought to be maintained under highly reduced and rarely oxidized conditions, because plant cells possess diverse antioxidant systems such as superoxide dismutase, ascorbate peroxidase, glutathione, peroxiredoxin and Trx [32]. Recently, however, small molecules known to cause oxidative stress have been shown to act as signal transducers in the cells, and small fluctuations in the redox state may exert regulatory effects in some cellular processes. For example, NPR1, a transcriptional coactivator of the plant defense response, was converted from an inactive oligomeric form, which was formed via intermolecular disulfide bonds, to an active monomeric form following pathogen attack and a change in the intracellular redox state [33]. In this way, changes in the intercellular redox state appear to be important for regulation of some cellular signaling pathways. Taken together, cytosolic redox states may fluctuate in response to external stimuli or depending on the metabolic states of the cell, and the fluctuation of cytosolic redox states may exert regulatory effects on signaling molecules such as AtCPK21.

Experimental procedures

Plant materials

T87 cells of Arabidopsis thaliana, ecotype Columbia [34], were obtained from the RIKEN BioResource Center (Yokohama, Japan), and maintained in suspension as described previously [35]. The cells were grown in 100 mL JPL medium [35] in a 300 mL flask at 22-C under dark conditions, and aliquots of the cells were transferred to fresh medium every 2 weeks. To prepare plasma membranes, 100 mL of 2-week-old cells were added to 2 L of medium in a 3-liter flask, and cultivated with stirring using a magnetic stirrer and aeration for 2 weeks. Approximately 150–200 g fresh weight cells were obtained from 2 L of culture.

Preparation of plasma membranes

Cell cultures were suspended into 0.25 m sorbitol, 50 mm Tris-acetate (pH 7.5), 1% polyvinylpyrrolidone, 2 mm EGTA, 2 mm DTT and protease inhibitor cocktail (Roche Basel, Switzerland), and disrupted using a BeadBeater (BioSpec Products, Bartlesville, OK). After differential centrifugation, a post-mitochondrial fraction was subjected to aqueous two-phase partitioning [36] in which the phase solution contained 10 mm NaCl. The resulting plasma membrane fraction was suspended in 0.15 m NaCO3 and centrifuged at 100 000 g for 30 min and the resulting precipitate was suspended in 0.25 m sorbitol and 20 mm Tris-acetate (pH 7.5).

Preparation of recombinant Trx-h1 and the mutants

Trx-h1 and the mutant Trx-h1C43S were expressed in Escherichia coli using plasmids constructed for A. thaliana Trx-h1 [6]. The expressed proteins were purified as described previously [37].

Screening of Trx-h1 target proteins in the plasma membrane

The mutant Trx-h1C43S was immobilized on CNBr-activated Sepharose 4B (GE Healthcare, Buckinghamshire, UK) according to the manufacturer's instructions. The plasma membranes (40 mg) were solubilized using 2% n-dodecyl-β-d-maltoside in 20 mm Tris/HCl (pH 8.0), 0.5 mm EGTA and protease inhibitor cocktail, and centrifuged at 100 000 g for 30 min. The supernatant was mixed with Trx-h1C43S mutant-immobilized resin and incubated for 1 h at 25-C. The resin was washed in turn with buffer A containing 20 mm Tris/HCl (pH 8.0), 0.5 mm EGTA, 0.05% n-dodecyl-β-d-maltoside, with buffer A plus 500 mm NaCl, and with buffer A plus 0.1% SDS. The resin was centrifuged at 3000 g for 5 min at 25 °C in all washes. The resin was finally suspended in buffer A including 0.1% SDS and 20 mm DTT, and incubated for 1 h. Trichloroacetic acid (5%) was added to the eluate, followed by centrifugation at 10 000 g for 15 min at 4-C. The precipitate was washed twice with ice-cold acetone, air-dried and solubilized in 5% SDS, 0.1 M NaOH, and SDS sample buffer composed of 2% SDS, 10 mm Tris-HCl, pH 6.8, 6% glycerol, 10 mm DTT was added. No proteins were present in the eluate when using a resin without immobilized Trx-h1 (data not shown). It was also demonstrated that a Trx mutant in which both Cys residues in the active site were substituted by Ser did not trap any proteins [14].

Identification of the acquired proteins

The acquired proteins were separated by SDS/PAGE (7.5–15%) and stained using Coomassie Brilliant Blue G-250. The protein bands were excised from the gel, diced, de-stained in 30% acetonitrile, 50 mm ammonium bicarbonate and dried in 70% acetonitrile, 50 mm ammonium bicarbonate, finally in 100% acetonitrile. The gel pieces were treated with 0.1% of an acid-labile surfactant (Waters, Milford, MA) for 10 min at 60-C as described previously [38, 39]. The gel pieces were then dried and subjected to in-gel digestion [40]. The tryptic peptides were extracted from the gel, dried, dissolved in 0.1% trifluoroacetic acid, and centrifuged at 10 000 g for 30 min to remove hydrolyzed surfactants. The supernatant was desalted using OMIX (R) pipette tips C18MB (Varian, Palo Alto, CA, USA), and eluted using 10 mg/mL α-cyano 4-hydroxy cinnamic acid (Fluka, Milwaukee, WI, USA) in 50% acetonitrile, 0.1% trifluoroacetic acid and 10 mm monobasic ammonium phosphate as described previously [41]. The eluate was spotted on a sample plate, and analyzed by MALDI-TOF/TOF MS using an ABI 4700 proteomics analyzer (Applied Biosystems, Carlsbad, CA, USA). The data were analyzed using mascot software (Matrix Science, Boston, MA) and GPS Explorer 3.0 Workstation (Applied Biosystems).

Cloning, expression and purification of A. thaliana AtCPK21

An AtCPK21 cDNA encoding 531 amino acids (NM 116710, accession numbers in MEDLINE) was amplified from an Athaliana cDNA library by PCR using the primers 5′-GGCATATGGGTTGCTTCAGCAGTAAAC-3′ (containing an NdeI site, underlined) and 5′-CCGGATCCTCAATGGAATGGAAGCAGTTTC-3′ (containing an XhoI site, underlined). The PCR product was confirmed by sequencing and sub-cloned into the NdeI/XhoI sites of pCold I (Takara, Otsu, Japan), thereby adding a hexahistidine (6 × His) tag and a Factor Xa cleavage site to the N-terminus of the AtCPK21. The 6xHis-tagged AtCPK21 was induced in E. coli BL21 by adding 0.5 mm isopropyl-β-d-thiogalactopyranoside for 24 h at 15-C. The expressed protein was purified using Ni-NTA Superflow resin (Qiagen, Hilden, Germany) and the 6 × His tag was removed by Factor Xa digestion (Novagen, Madison, WI, USA). The digested proteins were separated by gel filtration using a Superdex 200 10/300 GL column (GE Healthcare) in 10 mm Tris-acetate (pH 7.5) and 0.2 m NaCl. Approximately 2 mg of purified protein preparation was obtained from 1 L of culture.

Protein–protein interaction assay in yeast

Interaction between AtCPK21 and AtTrx-h1 was assessed using a yeast two-hybrid membrane protein system 3 (Dualsystems Biotech, Schlieren, Switzerland) according to the manufacturer's instructions. To construct plasmids for the system, the AtCPK21 full-length cDNA encoding 531 amino acids was amplified by PCR using primers 5′-ATAACAAGGCCATTACGGCCAAAAATGGGTTGCTTCAGCAGTA-3′ and 5′- AACTGATTGGCCGAGGCGGCCCCATGGAATGGAAGCAGTTTCC-3′. Both AtTrx-h1 wild-type and the AtTrx-h1C43S mutant were amplified using the primers 5′-ATAACAAGGCCATTACGGCCAAAAATGGCTTCGGAAGAAGGAC-3′ and 5′-AACTGATTGGCCGAGGCGGCCCCAGCCAAGTGTTTGGCAATG-3′. The SfiI restriction sites are underlined. The PCR product was confirmed by sequencing. The AtCPK21 and AtTrx-h1 cDNAs were sub-cloned into the SfiI site of pBT3-C and pPR3-C, respectively. These vectors were included in the two-hybrid membrane protein system. The resulting bait and prey plasmids were co-transformed into a NMY51 yeast strain, which contains the HIS3 and ADE2 reporter genes. Control experiments were performed in which plasmid pCCW-Alg5 with or without pAI-Alg5 was transformed into NMY51. To monitor the strength of the protein–protein interaction, each transformant was inoculated into liquid synthetic complete medium (SC) lacking Leu and Trp (SC–Leu–Trp) including 0.67% yeast nitrogen base without amino acids, 2% glucose, 5 mm MES/KOH (pH 5.5), 0.067% –Leu/–Trp drop-out supplement (Clontech, Mountain View, CA, USA) and pre-cultured overnight at 30-C. A negative control strain (NMY51 pCCW-Alg5) was pre-cultured in SC–Leu. The pre-culture was serially diluted 20–210-fold using dilution medium including 0.67% yeast nitrogen base without amino acids, 2% glucose, and spotted on various selection plates using a 96-pin spotter (Tokken, Chiba, Japan), and incubated for 2 days at 30-C. The SC–His plate comprised 0.67% yeast nitrogen base without amino acids, 2% glucose, 2% agar, 0.062% CSM-His (Q-BIOgene, Irvine, CA, USA), 0.0016% Cys and 0.0016% Met. The casamino acids medium minus Ade contained 0.67% yeast nitrogen base without amino acids, 2% glucose, 2% agar, 1% casamino acids, 0.004% Ura, 0.003% Trp, 0.0016% Cys and 0.0016% Met. The casamino acids medium plus Ade also contained 0.008% adenine hemisulfate.

Oxidation and reduction of AtCPK21 in vitro

To oxidize AtCPK21, 3 μm recombinant protein was incubated with 5 μm CuCl2 in 10 mm HEPES/NaOH (pH 7.2) for 1 h at 25-C, and then the solution was exchanged into 10 mm HEPES/NaOH (pH 7.2) and 100 mm NaCl by gel filtration. The solution was concentrated using a Vivaspin 500 centrifugal filter unit (Sartorius, Goettingen, Germany), and the protein concentration was determined using a BCA protein assay kit (Pierce, Appleton, WI, USA). Oxidized AtCPK21 (2 μm) was incubated with DTT and Trx-h1 at various concentrations for 1 h at 25-C.

Measurement of kinase activity of AtCPK21

The activity of calcium-dependent protein phosphorylation was measured as described previously [42]. The assay was performed in 20 mm HEPES/NaOH (pH 7.2), 10 mm MgCl2, 200 μm [γ-32P]ATP (150 cpm/pmol), 0.5 mg/mL calf histone III-S (Sigma, St. Louis, MO, USA) and 100 nm AtCPK21 for 6 min at 30-C with 100 μm CaCl2 or 100 μm EGTA. The reaction was initiated by adding ATP and stopped by adding 20 mm HCl.

Peptide mapping analysis

AtCPK21 oxidized by CuCl2 was digested by trypsin for 20 h at 37-C in 10 mm HEPES/NaOH, pH 7.2, 0.1 m NaCl. The tryptic peptides were separated and directly spotted onto a metal plate for mass analysis using a DiNa LC-MALDI spotter system (KYA TECH, Tokyo, Japan), and their molecular mass was analyzed by MALDI-TOF/TOF MS. The fraction containing a disulfide-linked peptide was collected from the wells on the plate, dissolved in 50 mm Tris-(2-carboxyethyl) phosphine hydrochloride, and incubated for 1 h at 25-C. The eluted peptides were desalted and analyzed by MALDI-TOF/TOF MS. The resulting data were analyzed using proteinpilot software (Applied Biosystems).

Hydrogen peroxide treatment and extraction of T87 cells

Five-day-old T87 culture cells were collected by low-speed centrifugation at 1000 g for 1 min, and 10 mL of cells were transferred to fresh medium (final 100 mL) with or without 10 mm H2O2, and incubated for 4 h at 22-C in the dark. Aliquots of the cells were sampled at various time points, drained, washed six times with NaCl/Pi in a cell strainer (100 μm mesh, BD Falcon, Franklin Lakes, NJ, USA) and frozen in liquid nitrogen. Two volumes of 10% trichloroacetic acid and an equal volume of glass beads were added to the frozen cells, and the cells were disrupted using a MicroSmash homogenizer (Tomy, Tokyo, Japan). The homogenate was centrifuged at 860 g for 30 s at 4-C, and the supernatant was decanted into a new tube. After one more centrifugation and decantation, the proteins were precipitated and dissolved in 1% SDS and 40 mm HEPES/NaOH (pH 7.2).

Detection of endogenous AtCPK21

The proteins extracted from the cell cultures were separated by SDS/PAGE (7.5 or 10%) and analyzed by immunoblotting as described previously [43]. To assess the redox state of AtCPK21, the protein precipitates were dissolved in 10 mg/mL SUNBRIGHT® ME-020MA PEG-maleimide (2 kDa) (NOF, Tokyo, Japan), 1% SDS and 40 mm HEPES/NaOH (pH 7.2), and labeled by vigorous shaking for 30 min at 25-C [44]. The labeled proteins were separated by SDS/PAGE (7.5–15%) and immunostained. The polyclonal antibody was raised against whole recombinant AtCPK21 in rabbit. The antiserum was purified by affinity chromatography using recombinant proteins immobilized on CNBr-activated Sepharose 4B (GE Healthcare). The immunostained bands were visualized using a chemiluminescent substrate (Pierce) on an LAS-1000plus chemiluminescence image analyzer (Fuji, Tokyo, Japan).

Protein determination

Protein concentrations were determined using a Bio-Rad protein assay (Bio-Rad, Hercules, CA, USA) unless otherwise specified.


This work was supported by a grant-in-aid from the Japan Society for the Promotion of Science to H.U. -N. We thank Motohashi (Kyoto Sangyo University, Japan) for valuable suggestions, Shibahara (Japan Patent Office) and Matsumoto (Kyushu University, Japan) for instruction in mass spectrometry, and Yuasa (Hiroshima University, Japan) for kindly providing the anti-NtCDPK1 IgG. We also thank Shimizu-Sato and Yoko Yamamoto of Nagoya University and all other members of our laboratories for continuous encouragement and helpful suggestions.