Identification of a dual-specificity protein phosphatase that inactivates a MAP kinase from Arabidopsis


*For correspondence (fax +1 510 642 4995; e-mail


Mitogen-activated protein kinases (MAPKs) play a key role in plant responses to stress and pathogens. Activation and inactivation of MAPKs involve phosphorylation and dephosphorylation on both threonine and tyrosine residues in the kinase domain. Here we report the identification of an Arabidopsis gene encoding a dual-specificity protein phosphatase capable of hydrolysing both phosphoserine/threonine and phosphotyrosine in protein substrates. This enzyme, designated AtDsPTP1 ( Arabidopsis thaliana dual-specificity protein tyrosine phosphatase), dephosphorylated and inactivated AtMPK4, a MAPK member from the same plant. Replacement of a highly conserved cysteine by serine abolished phosphatase activity of AtDsPTP1, indicating a conserved catalytic mechanism of dual-specificity protein phosphatases from all eukaryotes.


Studies have demonstrated that mitogen-activated protein kinase (MAPK) signalling is ubiquitous among eukaryotes from yeast to human ( Guan 1994;Su & Karin 1996;Wurgler-Murphy & Saito 1997). In metazoans and in yeast, MAPK pathways regulate growth, development and the response to stress. In Drosophila, Caenorhabditis elegans and vertebrates, receptor tyrosine kinases, acting through Ras and Raf, activate MAPKs and regulate growth and development. In mammals, the JNK, SAPK and p38 MAPKs regulate stress responses ( Karin & Hunter 1995). In Saccharomyces cerevisiae, six different MAPK signalling pathways regulate mating, pseudohyphal growth, invasiveness, cell wall biosynthesis, the response to osmotic stress and spore wall formation ( Sprague 1992;Wurgler-Murphy & Saito 1997).

Concerning the function of MAPKs in higher plants, studies in the past two years have revealed a large array of signalling pathways that are mediated by MAPKs ( Machida et al. 1997 ;Mizoguchi et al. 1997 ). For example, a specific MAPK activity is rapidly activated by cold, drought, mechanical stimuli and wounding ( Bogre et al. 1997 ;Jonak et al. 1996 ;Seo et al. 1995 ;Usami et al. 1995 ). MAPK activation was shown to be an early step in the wounding response because systemin, but not jasmonic acid, can mimic the effect of wounding by inducing MAPK activity ( Stratmann & Ryan 1997). In both tobacco and parsley cells, a MAPK is rapidly activated in response to pathogen elicitors ( Adam et al. 1997 ;Ligterink et al. 1997 ;Suzuki & Shinshi 1995). Salicylic acid, a signalling molecule in the pathogen response, also transiently activates a MAPK in tobacco cells ( Zhang & Klessig 1997). In addition to stress and pathogen responses, MAPKs also participate in developmental processes such as pollen development ( Wilson et al. 1997 ). In the ethylene signalling pathway, both histidine kinase and Raf-like serine/threonine protein kinase have been shown to play a role, although the MAPK component of the pathway has yet to be identified ( Chang & Meyerowitz 1995;Ecker 1995;Kieber 1997). Taken together, these findings demonstrate that MAPK activation is a common and early step in the response to stress, hormonal and developmental signals. A number of MAPK isoforms may be involved in pathways initiated by different signals ( Hare et al. 1997 ;Machida et al. 1997 ;Mizoguchi et al. 1997 ;Seo et al. 1997 ).

In mammalian and yeast systems, extracellular signals trigger activation of MAPK by a protein kinase cascade. The immediate upstream regulator is referred to as MAPK kinase, a dual-specificity kinase capable of phosphorylating both tyrosine and threonine residues ( Ahn et al. 1992 ;Guan 1994). Indeed, full activation of MAPK requires dual phosphorylation on threonine and tyrosine residues within the kinase domain. Conversely, dephosphorylation of activated MAPK by protein phosphatases inactivates the kinase.

Based on the phospho-amino acid specificity, protein phosphatases are generally divided into serine/threonine and tyrosine phosphatases (PPases and PTPases). A growing number of phosphatases have recently been identified from mammalian cells for their ability to dephosphorylate both phosphotyrosine and phosphoserine/threonine residues ( Keyse 1995). These dual-specificity phosphatases have been shown to regulate MAPK activity by dephosphorylating both the threonine and the tyrosine residues (reviewed by Sun & Tonks 1994). Such phosphatases all contain the active site motif VXVHCXXGXSRSXTXXXAY(L/I)M. This particular motif shares some homology with PTPases but not with PPases. As a result, these phosphatases are often categorized as a special group in the PTPase family and referred to as dual-specificity protein tyrosine phosphatases (DsPTPases). In fact, DsPTPases do not share sequence similarity with PTPases beyond the active site motif of about ten amino acids. In addition, DsPTPases often have rather strict substrate specificity and each member usually acts on just one or a limited number of MAPKs ( Keyse 1995). The DsPTPases are also present in yeast, with MSG5 involved in the mating pathway and YVH1 in regulation of nitrogen metabolism ( Guan et al. 1992 ;Zhan et al. 1997 ). In the green algae Chlamydomonas eugametos, a DsPTPase has recently been identified and shown to be regulated by cell-cycle and oxidative stress ( Haring et al. 1995 ). Taken together, these studies suggest that a wide range of eukaryotic organisms contain DsPTPases.

Several studies in higher plants have shown that MAPK activation is accompanied by tyrosine phosphorylation of the enzyme protein ( Stratmann & Ryan 1997;Suzuki & Shinshi 1995;Zhang & Klessig 1997). Protein kinases similar to MAPK kinases that activate MAPK have also been identified from plant sources (reviewed by Mizoguchi et al. 1997 ). However, the mechanism underlying inactivation of MAPK is not understood. In particular, it is not known whether higher plants contain DsPTPases, a major group of protein phosphatases that target MAPK in other systems. In this study, we have identified from Arabidopsis a DsPTPase that dephosphorylates and inactivates a MAPK from the same plant.


Isolation and characterization of a cDNA encoding AtDsPTP1

In yeast and mammalian systems, a key component for inactivation of MAPK are the DsPTPases that dephosphorylate both phosphothreonine and phosphotyrosine residues. Using this as a paradigm for MAPK regulation in higher plants, we aimed at identifying a DsPTPase homologue from a plant source, which will provide a stepping stone for understanding the molecular mechanism underlying regulation of MAPK pathways. In a database search, we identified one expressed sequence tag (EST 80F1T7) that showed significant sequence identity to DsPTPases from mammalian systems. Using this EST as a probe to screen an Arabidopsis cDNA library resulted in isolation of AtDsPTP1 cDNA (see Experimental procedures).

The nucleotide and deduced amino acid sequence of AtDsPTP1 are shown in Fig. 1(a). The cDNA is 1080 bp long and contains 156 bp of 5′ untranslated region (UTR) followed by a 600 bp open reading frame and 313 bp 3′ UTR. This cDNA encodes a protein of 199 amino acids with a calculated molecular mass of 22 kDa. The amino acid sequence of the AtDsPTP1 protein shows significant homology (25–35% identical) to a number of DsPTPases from various organisms including human, rat, Drosophila, yeast and Chlamydomonas eugametos. The sequence alignment of AtDsPTP1 with several DsPTPases from other organisms is shown in Fig. 1(b). As expected, the active site signature motif in these sequences is highly conserved ( Fig. 1b, underlined region).

Figure 1.

Sequence analysis of the AtDsPTP1 cDNA.

(a) Nucleotide and deduced amino acid sequence of the AtDsPTP1 cDNA. The untranslated regions are shown in lower case. The conserved catalytic domain characteristic of DsPTPases is underlined. The asterisks indicate in-frame stop codons.

(b) Amino acid sequence alignment of AtDsPTP1 with DsPTPases from other organisms. The amino acid sequences were deduced from nucleotide sequences of cDNAs from Arabidopsis (AT-DSPTP1), Chlamydomonas eugametos (VH-PTP13, Haring et al. 1995 ), human (HS-VHR, Ishibashi et al. 1992 ; HS-MKP1, Keyse & Emslie 1992) and yeast (SC-YVH1, Guan et al. 1992 ). The numbers on the top indicate the amino acid residue position in AT-DSPTP1 protein. Dashes represent gaps introduced to improve the alignment. Dots indicate the conserved amino acids and asterisks indicate identical amino acid residues in all sequences. The highly conserved catalytic domain is underlined.

To confirm that AtDsPTP1 is from the Arabidopsis genome, we performed a DNA gel blot analysis ( Fig. 2). Based on this high-stringency blot, AtDsPTP1 is a single-copy gene. BglII digestion generated two hybridizing bands because a BglII site is present in the AtDsPTP1 cDNA ( Fig. 2, lane 2).

Figure 2.

DNA gel blot analysis of the AtDsPTP1 gene.

Restriction enzymes are BamHI (lane 1), BglII (lane 2), EcoRI (lane 3) and HindIII (lane 4). The positions of the DNA size markers are shown on the right.

To determine whether the AtDsPTP1 gene is normally expressed in Arabidopsis plants, total RNA was isolated from root, stems, leaves and flowers. The mRNA levels of AtDsPTP1 were probed with its cDNA insert. As shown in Fig. 3, the AtDsPTP1 gene is expressed in all tissues examined.

Figure 3.

Expression of the AtDsPTP1 gene in different organs of Arabidopsis.

(a) RNA gel blot analysis of total RNA (10 μg) from roots (lane 1), stem (lane 2), leaves (lane 3) and flowers (lane 4).

(b) Ethidium bromide-stained ribosomal RNAs in the corresponding agarose gel showing the relative amount of RNA in each lane.

In animal systems and in the green algae Chlamydomonas eugametos, expression of some DsPTPase members has been shown to respond to extracellular signals including stress factors ( Haring et al. 1995 ;Keyse & Emslie 1992;Sun & Tonks 1994). We also analysed the expression pattern of AtDsPTP1 in plants under various stress conditions including heat shock, cold, wounding and high salt. None of these stress factors seemed to significantly alter the expression levels of AtDsPTP1 mRNA (data not shown).

The AtDsPTP1 gene encodes a dual-specificity protein phosphatase

Based on the amino acid sequence of AtDsPTP1 protein, we speculated that this protein may function as a dual-specificity protein phosphatase. However, further functional analyses of this protein rely on the confirmation of this hypothesis. In addition, no previous study has shown that higher plants contain dual-specificity protein phosphatases. It is thus critical to examine the enzymatic properties of AtDsPTP1 protein.

To produce recombinant AtDsPTP1 protein, we amplified the coding region of AtDsPTP1 cDNA by PCR and cloned the insert into the pGEX-4T-3 vector. The recombinant AtDsPTP1 protein was then expressed as a glutathione-S-transferase (GST) fusion in Escherichia coli. Figure 4 shows that overexpression of the GST–AtDsPTP1 fusion protein (48 kDa) was induced by isopropyl β- d-thiogalactopyranoside (IPTG) (lane 2). Affinity purification by glutathione-Sepharose beads generated a major protein band that can be cleaved into GST (26 kDa) and AtDsPTP1 (22 kDa) ( Fig. 4, lane 3 and 4). Further purification of the AtDsPTP1 protein was performed by ‘on-bead’ cleavage ( Fig. 4, lane 5). Protein purified by this procedure (90% or higher purity) was used for the enzymatic analyses.

Figure 4.

Expression and purification of the recombinant AtDsPTP1 protein.

Lane 1, total bacterial proteins before induction; lane 2, total bacterial proteins after induction with IPTG; lane 3, GST–AtDsPTP1 fusion protein; lane 4, partial cleavage of the fusion protein by thrombin; lane 5, purified AtDsPTP1 protein. The molecular mass of the proteins is shown on the right.

In addition to the ‘wild-type’ AtDsPTP1 protein, we also produced a ‘mutant’ protein based on previous studies on DsPTPases and PTPases from various organisms. As discussed earlier, the cysteine in the catalytic core of the enzyme is essential for the enzymatic activity. This is true for both tyrosine-specific and dual-specificity PTPase ( Denu et al. 1996 ;Flint et al. 1997 ;Neel & Tonks 1997). As shown in Fig. 1(b), the cysteine 135 in AtDsPTP1 is located in the active site signature motif. To determine whether this conserved residue is important for the catalytic activity of AtDsPTP1 enzyme, we replaced Cys135 by Ser resulting in a mutant with a single atom change in the protein. This mutant, named C135S, was used in most of the phosphatase assays.

After the recombinant AtDsPTP1 proteins were purified, we performed phosphatase assays using pyronitrophenyl phosphate (pNPP) as a substrate. Most phosphatases are capable of cleaving the phosphate from pNPP resulting in a yellow nitrophenol product that can be conveniently quantified by absorbance at 405 nm (A405). The AtDsPTP1 protein had very low specific activity against pNPP as compared to AtPTP1, a tyrosine-specific protein phosphatase ( Xu et al. 1998 ). As shown in Figs 5, 8 μg ml–1 of AtDsPTP1 enzyme generated 0.2 OD units during the 3 h incubation, whereas 0.02 μg ml–1 AtPTP1 enzyme generated 0.8 OD during a 15 min incubation ( Xu et al. 1998 ).

Figure 5.

The AtDsPTP1 protein is an active phosphatase.

(a) The time course of AtDsPTP1 phosphatase activity.

(b) The enzyme activity is proportional to enzyme concentration.

In (a) and (b), the activity of both AtDsPTP1 (○) and C135S mutant protein (•) was measured using pNPP substrate. Sodium vanadate (200 μm) inhibited the AtDsPTP1 activity (□).

Several results indicated that the low phosphatase activity was indeed an intrinsic function of AtDsPTP1. Firstly, the activity was linear with the duration of reaction ( Fig. 5a). Secondly, pNPP hydrolysis increased proportionally with increasing amounts of AtDsPTP1 added to the reaction ( Fig. 5b). Thirdly, sodium vanadate, a specific inhibitor for DsPTPases and PTPases, inhibited the activity of AtDsPTP1 against pNPP ( Fig. 5). Finally, the mutant protein with active site cysteine replaced by serine (C135S) had no detectable phosphatase activity in this assay ( Fig. 5). Taken together, these results show that AtDsPTP1 protein is an active phosphatase.

To determine whether the AtDsPTP1 gene product is a dual-specificity protein phosphatase, we performed phosphatase assays using protein substrates labelled by either tyrosine kinase or serine/threonine kinase. Myelin basic protein (MBP) was used as substrate for both kinases in the labelling process so that the phosphatase activity of AtDsPTP1 protein against phosphoserine/threonine or phosphotyrosine (in the same substrate) could be compared. As shown in Fig. 6, the AtDsPTP1 protein dephosphorylated both phosphotyrosine and phosphoserine/threonine in a time- and enzyme concentration-dependent manner. It was noted that AtDsPTP1 displayed a higher activity against phosphotyrosine compared with phosphoserine/threonine in MBP. The C135S mutant of AtDsPTP1 showed no detectable activity against either substrate ( Fig. 6). These assays demonstrated that the AtDsPTP1 gene encodes a functional dual-specificity protein phosphatase. Cys135 is essential for the phosphatase activity of AtDsPTP1.

Figure 6.

The AtDsPTP1 protein is a dual-specificity protein phosphatase.

(a) Both P-Tyr and P-Ser/Thr in MBP were hydrolysed by the AtDsPTP1 protein in a time-dependent manner.

(b) Dephosphorylation of P-Tyr and P-Ser/Thr in MBP at various enzyme concentrations.

In both (a) and (b), (○) and (•) represent the phosphatase activity of AtDsPTP1 against P-Tyr and P-Ser/Thr, respectively, while (□) and (▪) indicate the dephosphorylation by the C135S mutant protein.

To identify potential inhibitors and activators of AtDsPTP1, we examined the effect of various chemicals on the activity of AtDsPTP1 ( Table 1). Most of the chemicals used did not show a significant effect on phosphatase activity but a few strongly inhibited activity. Sodium vanadate and sodium tungstate have been shown to inhibit all DsPTPases and PTPases, and they indeed abolished the phosphatase activity of AtDsPTP1 against all tested substrates including pNPP, phosphotyrosine and phosphoserine/threonine. NaF, a generic serine/threonine phosphatase inhibitor, inhibited AtDsPTP1 activity against pNPP and phosphoserine/threonine but not the activity against phosphotyrosine. Spermidine, a polyamine, strongly inhibited AtDsPTP1 activity against phosphoserine/threonine but was a less potent inhibitor when pNPP or phosphotyrosine was used as a substrate ( Table 1). As one of the polyamines, spermidine plays a vital role in plant growth and developmental processes ( Evans & Malmberg 1989;Galston et al. 1997 ). Inhibition of DsPTPases may account for part of the polyamine action in higher plants, although further studies are required to test this notion. These data may provide useful tools for studying the physiological function of AtDsPTP1 in the plant, although in vivo studies using these reagents need careful interpretation.

Table 1. Effect of various chemicals on the phosphatase activity of AtDsPTP1
% activity of control
ChemicalConcentrationpNPPMBP 32P-labelled at Ser/Thr MBP 32P-labelled at Tyr
  1. Phosphatase activity was determined using pNPP and MBP 32P-labelled at Ser/Thr or Tyr as a substitute as described in Experimental procedures.

None 100100100
CaCl21 mM9410196
EDTA10 mM1078896
MgCl25 mM957970
NaF20 mM7374101
Okadiac acid1 μM9391103
Spermidine2 mM703470
Sodium tungstate200 μM353
Sodium vanadate200 μM311

Inactivation of an Arabidopsis MAPK by AtDsPTP1

As discussed earlier, an important family of substrates for DsPTPases are MAPKs ( Keyse 1995). To date, MAPKs are the only proteins that have been shown to be regulated by tyrosine phosphorylation in higher plants. To determine whether AtDsPTP1 recognizes MAPK as a substrate, we expressed and purified one of the Arabidopsis MAPKs, AtMPK4 ( Mizoguchi et al. 1993 ), and examined its regulation by AtDsPTP1. As described in Experimental procedures, different amounts of AtDsPTP1 were added to the AtMPK4 protein which had been phosphorylated by the Arabidopsis MAPK kinase At-MEK. The phosphorylation status and kinase activity of AtMPK4 were then monitored by SDS–PAGE and autoradiography.

As shown in Fig. 7(a), the AtMPK4 protein was phosphorylated by At-MEK in the presence of 32P-ATP (lane 1). Addition of increasing amounts of AtDsPTP1 protein to the phosphorylated AtMPK4 reduced the amount of 32P-label in the AtMPK4 protein, indicating dephosphorylation of AtMPK4 by AtDsPTP1 ( Fig. 7a, lanes 3–5). In a parallel experiment, we measured the kinase activity of AtMPK4 using MBP as a substrate. A high level of MBP phosphorylation was detected when AtMPK4 was not treated with AtDsPTP1 ( Fig. 7b, lane 1). After treatment with various amounts of AtDsPTP1 protein, AtMPK4 had reduced kinase activity, as indicated by less phosphorylation of MBP ( Fig. 7b, lanes 3–5). Consistent with the earlier phosphatase activity assays, the C135S mutant of AtDsPTP1 displayed little activity against AtMPK4 in dephosphorylation analysis ( Fig. 7a, lane 2). As a result, AtMPK4 treated with the C135S mutant of AtDsPTP1 was still fully active against MBP ( Fig. 7b, lane 2). This in vitro analysis using purified enzymes showed that AtDsPTP1 is capable of dephosphorylating and inactivating a MAPK from the same plant.

Figure 7.

Dephosphorylation and inactivation of AtMPK4 by AtDsPTP1.

(a) Dephosphorylation of activated AtMPK4 by AtDsPTP1.

Lane 1, phosphorylation level of AtMPK4 in the absence of AtDsPTP1; lane 2, phosphorylation level of AtMPK4 after treatment with 1 μg of C135S mutant protein; lanes 3–5, phosphorylation levels of AtMPK4 after treated with 0.2, 0.5 and 1.0 μg of AtDsPTP1.

(b) Phosphorylation of MBP by AtMPK4 was reduced by AtDsPTP1 treatment. Lane 1, phosphorylation of MBP by AtMPK4. lane 2, phosphorylation of MBP by AtMPK4 treated with 1 μg C135S mutant; lanes 3–5, phosphorylation of MBP by AtMPK4 treated with 0.2, 0.5 and 1.0 μg of AtDsPTP1.


Signal transduction by tyrosine phosphorylation has not been established in higher plants. The recent discovery of MAPK function in a number of stress and pathogen signalling pathways has opened up a new window to understanding the role of tyrosine phosphorylation in plant systems ( Machida et al. 1997 ;Mizoguchi et al. 1997 ). Regulation of MAPK activity is accompanied by tyrosine phosphorylation and dephosphorylation. Accumulating evidence in yeast and mammalian systems supports the notion that DsPTPases play an important role in MAP kinase inactivation ( Keyse 1995;Neel & Tonks 1997). However, it is not known whether such DsPTPases exist in higher plants. In this report, we have identified from Arabidopsis a DsPTPase that dephosphorylated and inactivated a MAPK member from the same plant, providing a molecular basis for a further understanding of MAPK regulation in plant cells by extracellular signals.

The AtDsPTP1 protein is among the smallest DsPTPases found from various organisms and contains only the essential phosphatase domains in the prototype enzymes such as VH1 and VHR from vaccinia virus and human, respectively ( Guan et al. 1991 ;Ishibashi et al. 1992 ). The predicted AtDsPTP1 protein sequence exhibited the highest degree of similarity to the mammalian members of DsPTPases ( Keyse 1995). The most conserved region is the signature motif, VXVHCXXGXSRSXTXXXAY(L/I)M, near the carboxyl end of the protein ( Fig. 1). This sequence motif, unlike the tyrosine-specific PTPases, extended to contain the AYLM module that is conserved among all members of the VH1 subfamily ( Denu et al. 1996 ). Although the importance of AYLM residues is yet to be elucidated, they serve as a convenient marker for DsPTPases. Additional regions of sequence similarity were also found between AtDsPTP1 and DsPTPases from other organisms, but these conservations are less consistent compared with the active site ( Fig. 1b). The cysteine residue in the active site has been shown to be essential for the formation of phosphoenzyme reaction intermediate in catalysis by PTPases and DsPTPases ( Guan & Dixon 1990). In all studies reported to date, mutations at the cysteine residue abolish the enzyme activity, which provides a criterion for studying the catalytic mechanism of a new PTPase or DsPTPase. As shown in our study, replacement of Cys135 by serine also eliminated the phosphatase activity of AtDsPTP1, suggesting that a similar phosphoenzyme intermediate is involved in the catalytic process of all PTPases and DsPTPases.

Comparing DsPTPases with the tyrosine-specific PTPases, only 5% of sequence identity is found ( Denu et al. 1996 ). This similarity is close to the homology between any two random proteins. Despite of this limited identity, recent structural analyses of the two groups of enzyme reveal that they retain highly similar structural fold. Detailed analyses of the enzyme structures suggested that the substrate specificity may be determined by a highly charged region upstream of the catalytic domain ( Denu et al. 1996 ;Neel & Tonks 1997). Based on the sequence analysis, the AtDsPTP1 gene may encode a DsPTPase. Dephosphorylation assays using the recombinant protein clearly demonstrated that AtDsPTP1 was capable of hydrolysing both phosphoserine/threonine and phosphotyrosine in a protein substrate, although it may prefer phosphotyrosine over phosphoserine/threonine as a substrate ( Fig. 6).

Recently, we have characterized a phosphotyrosine-specific PTPase, AtPTP1, from Arabidopsis ( Xu et al. 1998 ), which shares little sequence similarity with the AtDsPTP1 protein. Besides the major difference in substrate specificity (tyrosine-specific versus dual-specificity), AtPTP1 displayed significantly higher activity towards both pNPP and phosphotyrosine-containing proteins ( Xu et al. 1998 ) compared with AtDsPTP1. For example, the phosphatase activity of AtPTP1 is about 1600-fold higher against pNPP and 80-fold higher against protein substrate than the activity of AtDsPTP1. This may suggest that AtDsPTP1 prefers protein substrates over pNPP as compared with AtPTP1. In addition, AtDsPTP1 readily dephosphorylated the activated form of AtMPK4, a MAPK from Arabidopsis, consistent with the fact that MAPKs are physiological substrates for DsPTPases in other systems ( Sun & Tonks 1994). This in vitro study provides critical initial information for further studies on the in vivo substrates for AtDsPTP1 in Arabidopsis.

In addition to DsPTPases, other protein phosphatases have also been implicated in MAPK inactivation. These include tyrosine-specific PTPases ( Wurgler-Murphy & Saito 1997) and serine/threonine phosphatases ( Alessi et al. 1995 ;Shiozaki & Russell 1995) that have been identified in yeast. Along these lines, a recent study ( Meskiene et al. 1998 ) reported identification of an alfalfa PP2C homologue that serves as a potential regulator of SAMK, a stress-responsive MAPK ( Bogre et al. 1997 ;Jonak et al. 1996 ). Following the characterization of both tyrosine-specific and dual-specificity PTPases ( Xu et al. 1998 ; this study), further functional analysis, using molecular and reverse genetics approaches, will generate exciting new information on the function of tyrosine phosphorylation in higher plants.

Experimental procedures

Isolation and sequence analyses of the AtDsPTP1 cDNA

Using MKP1 ( Keyse & Emslie 1992), a DsPTPase from human, as a query sequence, we searched the Arabidopsis database using the on-line BLAST program. An EST clone (80F1T7) was identified as a possible homologue of MKP1. Sequence alignment suggested that this EST represented a partial cDNA. Using the labelled cDNA insert as a probe, an Arabidopsis cDNA library ( Kieber et al. 1993 ) was screened at high stringency as described previously ( Gupta et al. 1998 ). Among the 106 plaque-forming units screened, four independent clones were identified as positive. The cDNA inserts were excised from the phage clones using helper phage. Partial sequencing of the cDNAs showed that all four clones were products of the same gene. The longest cDNA clone (named AtDsPTP1) was subcloned in pBlueScript vector (Stratagene, La Jolla, CA) as two smaller fragments, and the nucleotide sequences of both strands were obtained using fluorescent dNTPs on a ABI automated sequencer.

Plant materials, DNA and RNA gel blot analyses

Plants (Arabidopsis thaliana ecotype Columbia) were grown in a greenhouse under long-day conditions (16/8 h light/dark cycle) to flowering stage. Different tissues including roots, leaves, stems and flowers were harvested, frozen in liquid nitrogen and stored at –80°C if not used immediately.

For DNA gel blot analysis, genomic DNA (3 μg) isolated from Arabidopsis leaves was digested with 30 units of various restriction enzymes at 37°C for 4 h. The restriction fragments were resolved in a 0.8% agarose gel, transferred to GeneScreen Plus nylon membrane (Du Pont/New England Nuclear, Boston, MA) and probed with 32P-labelled AtDsPTP1 cDNA as described by Gupta et al. (1997) .

For RNA gel blot analysis, total RNA (10 μg) from roots, stems, leaves and flowers was resolved by electrophoresis in a 1.2% agarose gel, transferred to GeneScreen Plus nylon membrane, and hybridized as described for the DNA gel blot above.

Overexpression of the recombinant AtDsPTP1 protein

The coding region of AtDsPTP1 cDNA was amplified with Pfu polymerase (Strategene) and subcloned into the pGEX-4T-3 vector (Pharmacia, Piscataway, NJ). The resulting clones were sequenced to ensure in-frame fusion of AtDsPTP1 with GST and to avoid clones that contain PCR-introduced mutations. The construct was transformed into Escherichia coli strain BL21 DE3 (Novagen, Madison, WI). Overexpression of the GST–AtDsPTP1 fusion protein was performed as described previously ( Luan et al. 1996 ;Xu et al. 1998 ). Briefly, bacterial culture was initiated and grown to mid-log phase. Expression of the GST fusion protein was induced by 0.2 m m IPTG at 30°C for 3–4 h and the bacterial cells were pelleted and resuspended in a buffer containing 100 m m NaCl, 50 m m Tris–HCl pH 8.0, 2 μm phenylmethylsulphonyl fluoride, 1 m m benzamidine and 2 m m EDTA. Cells were lysed by sonication followed by centrifugation at 15 000 g to produce a the supernatant containing the fusion protein. The GST fusion protein was purified by glutathione-Sepharose 4B (Pharmacia). The AtDsPTP1 protein was cleaved from the GST fusion using thrombin (Sigma) following the ‘on-bead’ cleavage protocol (Pharmacia) and analysed by SDS–PAGE.

Site-directed mutagenesis

Two PCR primers P38 (5′-CGGTAGTGTTCTTGTTCATTCCTTTGTTGGC-3′) and P39 (5′-GCCAACAAAGGAATGAACAAGAACACTACCG-3′) were designed to mutate Cys135 (TGC) to serine (TCC) in AtDsPTP1 protein. The site-directed mutagenesis was performed using Quick-change kit (Stratagene) according to the manufacturer’s instructions. The mutation was confirmed by nucleotide sequencing before protein expression and the mutant protein was produced as described for the wild-type protein.

Phosphatase assays

To study the generic phosphatase activity of AtDsPTP1, pNPP was used as a substrate. A total of 8 μg of AtDsPTP1 or C135S mutant protein was added to the phosphatase buffer (50 m m Tris–HCl pH 7.0, 2 m m dithiothreitol) containing 1 m m pNPP in 1 ml volume and the reaction was incubated at 30°C for various time periods before measuring A405 using a Shimazu 7000 spectrophotometer. To analyse the phosphatase activity at various enzyme concentrations, different amounts (0.5, 1, 2, 4 and 8 μg) of AtDsPTP1 or C135S mutant protein were included in the reaction mixture and incubated at 30°C for 3 h before measuring A405 using the Shimazu 7000 spectrophotometer.

For the phosphatase assays using protein substrate, MBP was labelled either at tyrosine (P-Tyr MBP) or serine/threonine (P-Ser/Thr MBP) using specific protein kinase in the labelling reactions. The procedure was based on instructions from enzyme manufacturers. For tyrosine labelling, 25 units of human c-Src tyrosine kinase (Upstate Biotechnology Inc., Lake Placid, NY) were incubated with 50 μg substrate protein and 50 μCi γ-32P-ATP (Du Pont/New England Nuclear, Boston, MA) in 100 μl of reaction buffer (25 m m Tris–HCl, pH 7.2, 5 m m MnCl2, 0.5 m m EGTA, 0.05 m m Na3VO4, 25 m m Mg-acetate) for 2 h at 30°C. The reaction mixture was loaded to a Centri-Spin-10 column (Princeton Separation Inc., Adelphia, NJ) to remove the free 32P-ATP and salts from the reaction. The purified labelled protein substrate was collected in the fluent after spinning at 750 g for 3 min. For 32P-labelling of serine/threonine residues in MBP, 50 units of bovine heart protein kinase A (Sigma) were used in a similar reaction as described for Src kinase, except for the reaction buffer (25 m m Tris–HCl, pH 7.5, 1 m m DTT, 100 m m NaCl, 12 m m MgCl2). Purification of labelled protein was as described above for the Src-labelled substrate.

Protein phosphatase activity was determined by measuring the release of free 32P from labelled substrates. Assays were performed as described previously ( Xu et al. 1998 ), with some modifications. For the time-course analysis, 200 ng of wild-type AtDsPTP1 or C135S protein was mixed with 2 × 104 cpm of 32P-labelled substrate in the phosphatase buffer (50 m m Tris–HCl pH 7.0, 2 m m DTT) in a volume of 150 μl. After incubating the reaction mixture at 30°C for various periods of time, the reaction was stopped by addition of 2 vol 25% TCA. Then 20 μg of bovine serum albumin (Sigma) was added as a carrier protein to facilitate the protein precipitation. Proteins were pelleted by centrifugation at 11 000 g for 5 min, and the supernatant was subjected to scintillation counting. Blank incubations were carried out without enzyme protein. The phosphatase activity of AtDsPTP1 protein is presented as percentage dephosphorylation of the protein substrate after subtracting blank readings.

To determine catalytic activity at different enzyme concentrations, 25, 50, 100, 200 and 400 ng of AtDsPTP1 or C135S were used in the phosphatase reaction. The reactions were stopped after 20 min and percentage dephosphorylation was calculated as described above.

AtMPK4 dephosphorylation and kinase activity assays

Arabidopsis MEK1 and AtMPK4 were expressed as maltose-binding protein fusions according to the manufacturer’s instructions (New England Biolabs, Beverly, MA, USA). For the phosphorylation and activation of AtMPK4, AtMEK1 was added to AtMPK4 at a 1:2 ratio in the kinase buffer (30 m m HEPES pH 7.5, 40 m m KCl, 4 m m MgCl2, 0.06 m m ATP, 5.2% glycerol) with 5 μCi of γ-32P-ATP (specific activity 3000 Ci/mmol) and incubated at 30°C for 1 h. The reaction mixture was loaded to a Centri-Spin-10 column to separate the free 32P-ATP and the kinases. To dephosphorylate the AtMPK4 protein various amounts of AtDsPTP1 (0.2, 0.5 and 1.0 μg) or 1.0 μg of C135S mutant protein was mixed with 1.0 μg of radiolabelled AtMPK4 and incubated at 30°C for 15 min. An aliquot of this reaction was analysed by SDS–PAGE ( Sambrook et al. 1989 ) and autoradiographed to monitor the dephosphorylation of AtMPK4 by AtDsPTP1. To the rest of the reaction mixture, 1 m m sodium vanadate was added to inactivate the AtDsPTP1 and this was used for kinase assay as described below.

To examine AtMPK4 activity, 1.0 μg of protein substrate MBP (Sigma) and 5 μCi of γ-32P-ATP were added to the reaction mixtures containing AtMPK4 or AtMPK4 treated with AtDsPTP1. The reaction was incubated for 10 min at 30°C before stopping with 2× SDS loading buffer. The samples were separated on 12% SDS–polyacrylamide gel and the dried gel was exposed to X-ray film for autoradiography. The developed X-ray film was scanned by a Umax Power look II scanner and processed by Abode Photoshop 4.0.