An Arabidopsis Protein Phosphorylated in Response to Microbial Elicitation, AtPHOS32, Is a Substrate of MAP Kinases 3 and 6*

Although mitogen-activated protein kinases (MAPKs) have been shown to be activated by a wide range of biotic and abiotic stimuli in diverse plant species, few in vivo substrates for these kinases have been identified. While studying proteins that are differentially phosphorylated upon treatment of Arabidopsis suspension cultures with the general bacterial elicitor peptide flagellin-22 (flg22), we identified two proteins with endogenous nickel binding properties that become phosphorylated after flg22 elicitation. These highly related proteins, AtPHOS32 and AtPHOS34, show similarity to bacterial universal stress protein A. We identified one of the phosphorylation sites on AtPHOS32 by nanoelectrospray ionization tandem mass spectrometry. Phosphorylation in a phosphoSer-Pro motif indicated that this protein may be a substrate of MAPKs. Using in vitro kinase assays, we confirmed that AtPHOS32 is a substrate of both AtMPK3 and AtMPK6. Specificity of phosphorylation was demonstrated by site-directed mutagenesis of the first phosphorylation site. In addition, immunosubtraction of both MAPKs from protein extracts removed detectable kinase activity toward AtPHOS32, indicating that the two MAPKs were the predominate kinases recognizing the motif in this protein. Finally, the target phosphorylation site in AtPHOS32 is conserved in AtPHOS34 and among apparent orthologues from many plant species, indicating that phosphorylation of these proteins by AtMPK3 and AtMPK6 orthologues has been conserved throughout evolution.

MAMP perception by defined host cell surface receptors initiates MAMP-triggered immunity, the first line of defense contributing to the arrest of microbial growth. The best characterized MAMP perception system in plants is the recognition of bacterial flagellin or flg22, the bioactive peptide from a highly conserved 22-amino-acid region of the protein (3), by the receptor kinase flagellin-sensitive 2 (FLS2) (4). Recognition of flg22 by FLS2 initiates intracellular signaling that results in a number of rapid responses thought to contribute to defense(s) against the invading microbe (5).
Among the most studied of these early responses is the rapid activation of mitogen-activated protein kinases (MAPKs). MAPKs are the final components of signaling cascades, activated by upstream MAPK kinases (MAPKKs), which are also activated by other upstream kinases, MAPKK kinases (MAPKKKs) (6 -8). In Arabidopsis, flg22 has been shown to activate two MAPKs, AtMPK3 and AtMPK6 (9,10). Moreover, plants with reduced levels of AtMPK6 displayed increased susceptibility to infection by virulent and avirulent strains of the bacterial pathogen, Pseudomonas syringae, indicating that signaling via AtMPK6 plays a critical role during defense responses (11).
Although activation of MAPKs in response to a wide range of biotic and abiotic stimuli has been well characterized (8), relatively little is known about the substrates of these kinases. The only in vivo substrate identified for AtMPK6 is 1-aminocyclopropane-1-carboxylate synthase 6, the rate-limiting enzyme in the production of the stress hormone ethylene (12). In a large scale screen for potential substrates of AtMPK3 and AtMPK6, a protein microarray strategy used denatured bacterially expressed proteins as substrates for bacterially expressed MAPKs (13). This screen identified 44 putative substrates of AtMPK3 and 38 putative substrates of AtMPK6, 26 of which were common for both MAPKs. However, no experiments were performed to determine whether these candidates were phosphorylated in planta. Here, we describe the identification of AtPHOS32 as a protein that becomes phosphorylated in response to flg22 treatment of Arabidopsis cell suspension cultures. Sequencing the first phosphorylation site of AtPHOS32 by nanoelectrospray ionization tandem mass spectrometry (nanoESI-MS/MS) indicated that this protein may be a sub-strate of MAPKs. Using in vitro kinase assays together with site-directed mutagenesis of the phosphorylated residue, we confirmed that AtPHOS32 is a substrate of both AtMPK3 and AtMPK6.

EXPERIMENTAL PROCEDURES
Maintenance and Treatment of Suspension Cultures-Suspension cultures of Arabidopsis thaliana ecotype Landsberg were maintained as described previously (9). Cultures were grown on a rotary shaker at 120 rpm in a 16/8-h light/darkness cycle at 24°C. Six days after subculturing, 5 ml of culture was removed to a 20-ml beaker and shaken at room temperature for 45 min prior to treatment with 100 nM flg22 peptide (3). Radioactive pulse labeling of cells was performed as described previously (14).
Protein Extraction, Two-dimensional Gel Analysis, and Ni-NTA Enrichment-For analytical two-dimensional gel analysis, proteins were isolated and separated by two-dimensional gel electrophoresis as described previously (14). For nickel-nitrilotriacetic acid (Ni-NTA) enrichment of AtPHOS32 and AtPHOS34, proteins were extracted from suspension-cultured cells using a homogenization buffer lacking chelators or reducing agents (100 mM HEPES-KOH, pH 7.5, 5% glycerol, 50 mM sodium pyrophosphate, 1 mM sodium molybdate, 25 mM sodium fluoride, 25 mM glycerophosphate, 0.5% polyvinyl pyrrolidine, 1 mM phenylmethylsulfonyl fluoride, 10 M leupeptin, and 1 nM calyculin A). Protein extracts were incubated with Ni-NTA resin on a rotating wheel for 2-4 h at 4°C. The resin was collected by centrifugation at 3,000 ϫ g for 5 min at 4°C and washed three times with 5 bed volumes of homogenization buffer. The resin was transferred to a Micro-Bio spin chromatography column (Bio-Rad), and the remaining homogenization buffer was removed by centrifugation at 1,000 ϫ g for 10 s. Proteins bound to the Ni-NTA resin were eluted three times in 70 l of elution buffer (50 mM phosphate buffer, pH 7.5, 100 mM imidazole). The pooled eluted proteins were concentrated by methanol/chloroform precipitation. In brief, 3 volumes of methanol were added followed by 1 volume of chloroform and 4 volumes of water, with mixing by vortexing after each addition. After centrifugation at 10,000 ϫ g for 1 min at room temperature, the aqueous (upper) phase was discarded, and proteins were precipitated from the organic phase with 4 volumes of methanol at Ϫ20°C for at least 40 min. After collecting the precipitate by centrifugation at 13,000 ϫ g for 10 min at 4°C, the pellet was washed in 80% acetone and stored at Ϫ20°C until gel separation.
Mass Spectrometry-The two-dimensional gel separated proteins were prepared for and analyzed by nanoESI-MS/MS as described (14). MALDI-TOF analysis was performed as described (15). Data base searches were performed using MS-Fit with a mass tolerance of Ϯ50 ppm. Modifications included allowing for two missed cleavages in the trypsin digest, cysteine modification by acrylamide, and oxidized methionine in the initial search and subsequently phosphorylation of Ser, Thr, or Tyr.
Bacterial Expression-For bacterial expression of AtPHOS32, the open reading frame was amplified from a cDNA library and directionally cloned into the NdeI-BamHI-digested pET-3a vector (Novagen). Individual clones were resequenced prior to transformation into BL21(DE3) bacteria for expression according to standard procedures. Proteins were isolated from bacteria under native conditions using the Ni-NTA isolation procedure described above.
MAPK Immunoprecipitation and in Vitro Kinase Assays-For immunoprecipitation of AtMPK3 and AtMPK6, antibodies (␣-MPK3 and ␣-MPK6, respectively) were raised in rabbits (Eurogentec, Seraing, Belgium) against synthetic peptides corresponding to the N terminus of the AtMPK3 kinase (MNT-GGGQYTDFPAVE) or the AtMPK6 kinase (MDGGSGQPAA-DTEMT). The AtMPK3/6 kinases were immunoprecipitated from flg22-elicited Arabidopsis suspension culture cells using ␣-MPK3 or ␣-MPK6 antibodies as follows. 10 min after the addition by 100 nM flg22, the elicited cells were harvested by filtration and lysed in immunoprecipitation buffer (100 mM Tris-HCl, pH 7.5, 5% glycerol, 50 mM sodium pyrophosphate, 1 mM sodium molybdate, 25 mM sodium fluoride, 25 mM glycerophosphate, 0.5% polyvinylpyrrolidine, 150 mM NaCl, 1 mM phenyl methyl sulfonyl fluoride, 10 l of leupeptin, 1 nM calyculin-A, and 1 mM Na 3 VO 4 ). The soluble protein extract was cleared by centrifugation at 10,000 ϫ g for 10 min at 4°C. Antibodies were added to the protein extract, and the mixture was incubated on a rotating wheel for 2 h at 4°C. Protein A-Sepharose (pre-equilibrated in immunoprecipitation buffer) was added to the immunoprecipitation mixture, and incubation was continued for another 2 h at 4°C. The mixture was washed three times in immunoprecipitation buffer, once in kinase buffer (50 mM Tris-HCl, pH 7.5, 10% glycerol, 150 mM NaCl, 5 mM MnCl, 10 mM MgCl), and once in kinase buffer containing 1 mM dithiothreitol. In vitro kinase assays were performed for 10 min at 30°C in a total volume of 20 l of kinase buffer containing the immunoprecipitated resin, substrate, and ATP mix (10 M ATP, 0.1 l of [␥-32 P]ATP). Reactions were stopped by adding SDS loading buffer and heating for 15 min at 65°C. After separation of proteins on a 14% SDS-polyacrylamide electrophoresis gel, radioactive incorporation was visualized by phosphorimaging using a Fuji FLA5000 scanner.
Site-directed Mutagenesis-Site-directed mutagenesis of bacterial expression plasmids to change the in vivo phosphorylated serine residue of AtPHOS32 to alanine or aspartic acid residues was performed commercially (BioS&T, Montreal, Canada), and the mutated expression plasmid was resequenced prior to use.
MAPK Mutant Plants-Insertion lines used in this study were mpk3-1 (SALK_051594), mpk3-2 (GABI 697F07), mpk6-3 (SALK_027507, the same line as used by Liu and Zhang (12)), and mpk6-4 (SALK_062471). Numbering of mpk6 alleles continues according to Liu and Zhang (12). Genomic PCR with primers designed by the SIGnAL iSect tool using standard settings was used to isolate homozygote plants. The GABI-Kat lines are described in Rosso et al. (16), and the SALK lines are described in Alonso et al. (17).
MAPK Immunodepletion-For immunodepletion experiments, 5 l of MAPK specific antibodies was prebound to protein A-Sepharose for 2 h. An equal amount of protein extract from flg22-treated Arabidopsis cell suspension cultures was incubated for 2 h with protein A-Sepharose with and without the MAPK antibodies as indicated. After removing the Sepha-rose by centrifugation, the remaining immunodepleted supernatant was used for in vitro kinase assays carried out as described above for 20 min at 30°C.

RESULTS
Isolation and Identification of AtPHOS32 and AtPHOS34-Previously, we described a two-dimensional gel-based phosphoproteomic screen to identify proteins undergoing rapid changes in phosphorylation in suspension cell cultures of Arabidopsis in response to treatment with the bacterial elicitor peptide, flg22 (14,15). Using radioactive pulse labeling of cells with [ 32 P]orthophosphate, a number of proteins showed reproducible changes in radioactive incorporation within 4 min of elicitation (indicated in Fig. 1A by circles and the large square). Proteins that did not change in the level of radioactive incorporation (indicated in Fig. 1A by arrows) served as internal controls for both radioactive incorporation and loading. Most of the radioactive images corresponding to differentially phosphorylated proteins did not align with stainable proteins even when using 18-cm single pI unit first dimension gradients (data not shown). The failure to visualize the differentially phosphorylated protein may result from low levels of putative signaling components, low stoichiometry of phosphorylation on these components, or a combination of these factors. Therefore, other enrichment methods were necessary to identify the target proteins.
While attempting to use Ni-NTA columns to isolate other candidate phosphoproteins that had been epitope-tagged with polyhistidine, we noticed that two additional proteins binding to the Ni-NTA columns were consistently phosphorylated after elicitation with flg22 (data not shown). These additional proteins were also enriched by Ni-NTA columns in untransformed Arabidopsis suspension cells, indicating that these proteins had endogenous nickel binding properties. Therefore, we used Ni-NTA to enrich for nickel-binding proteins after radioactive pulse labeling with [ 32 P]orthophosphate and separated these proteins by two-dimensional gel electrophoresis (Fig. 1B). The silver-stained image shows that after elicitation with flg22, newly appearing proteins with pIs more acidic than those seen in the 0-min controls (Fig. 1B, left panels) co-migrated with the radioactive spots observed in the phosphorimage (Fig. 1B, right  panels). These results indicate that these proteins are differentially phosphorylated in response to elicitation with flg22. The apparent pI and molecular weight of these proteins were similar to the series of radioactive spots indicated in Fig. 1A in the box.
The pattern of newly appearing acidic proteins led us to hypothesize that these proteins may be phosphorylated isoforms of the spots in Fig. 1B, left panels, indicated by the oval and box. However, the non-linearity of silver staining made it difficult to conclude whether the level of putative unphosphorylated forms decreased concurrently with the increase of the more acidic forms.
To identify each protein, we performed Ni-NTA enrichment using a liter of cell culture as starting material. This preparative protein isolation resulted in colloidal Coomassie Blue-stainable amounts of each apparent isoform ( Fig. 2A). We isolated a portion of the gel from the center of each spot and performed an in-gel trypsin digest followed by peptide mass fingerprinting using MALDI-TOF MS ( Fig. 2B; the complete peak list for each spot is provided as supplemental Table 1). The lower molecular weight series of proteins were all identified with 35-65% protein coverage as an Arabidopsis protein (At5g54430) (Fig. 2B). Remaining peaks not matching At5g54430 were searched independently against the Arabidopsis data base to determine whether there may be a second, co-migrating protein in the isolated gel piece, but no additional protein was identified in any of the spots (data not shown). We named this protein AtPHOS32 for PHOSphorylated protein with an apparent molecular mass of 32 kDa.
A similar analysis of the series of slightly higher apparent molecular weight proteins identified all proteins in the series as only containing At4g27320, a protein that is highly related (Ͼ80% amino acid identity) to AtPHOS32. Because of the lower sequence coverage for this protein (26 -33%), we performed nanoESI-MS/MS to confirm the identity (Fig. 2C). Two peptide sequence tags clearly showed single amino acid substitutions that differentiated this protein from At5g54430 (Fig. 2D, underlined), conclusively identifying the protein as At4g27320. Because of the slightly higher apparent molecular weight, we named this protein AtPHOS34.
Similarity to USPA Proteins from Bacteria-A function for AtPHOS32 or AtPHOS34 was not immediately obvious from the amino acid sequence, but both proteins contain a universal stress protein A (USPA) domain named after a bacterial protein from Escherichia coli that accumulates in response to a plethora of stimuli including nutrient starvation and exposure to many  dotted box), other differentially phosphorylated proteins (circles), and constitutively phosphorylated proteins that serve as internal controls (arrows). B, two-dimensional gels of AtPHOS32 and AtPHOS34 after enrichment using Ni-NTA chromatography. Samples were prepared at 0, 4, and 8 min after elicitation as described for A prior to the incubation of protein extracts with Ni-NTA resin. Eluted proteins were precipitated and separated by two-dimensional gels as described above. The panels on the left are silver-stained images of the gel. The panels to the right are phosphorimages. The circles and squares mark the unphosphorylated form of AtPHOS32 and AtPHOS34, respectively.
stresses (18). The function of this family of proteins found in many bacteria is not known. In Arabidopsis, 44 proteins are predicted to contain a USPA domain (19). The amino acid sequence identity between the plant and bacterial proteins in this domain is relatively low (ϳ20%; see alignment in Fig. 2D). However, the overall structural fold of the domain is predicted to be conserved in plant proteins (19).
The crystal structure of Methanococcus jannaschii USPA protein, MJ0577 (PDB ID 1MJH), was determined (20), and this analysis revealed that the protein binds ATP (20). Only five of the USPA domain proteins from Arabidopsis contain the residues required for binding ATP (19), and both AtPHOS32 and AtPHOS34 are among these five (Fig. 2D, asterisks indicate residues coordinating ATP in MJ0577).
Identification of One Phosphorylation Site on AtPHOS32-Using the tryptic digest of Fig. 2A, Spot 2 of AtPHOS32, we sequenced one of the phosphorylated peptides from AtPHOS32 by nanoESI-MS/MS. Evidence that this peptide is phosphorylated is clearly demonstrated by the neutral loss on the doubly charged parent ion ((Mϩ2H) 2ϩ to (Mϩ2H)⌬ 2ϩ ) (Fig. 3). The y-ion series identified the only serine residue being phosphorylated. Note that a mixed y-ion series was detected corresponding to fragments containing the phosphorylated serine residue or containing dehydroalanine as would occur after neutral loss of the phosphate. Interestingly, the serine residue in this motif is conserved in related proteins throughout a variety of plant species including monocots (Fig. 4), indicating that phosphorylation of this residue is evolutionarily conserved among plants. However, the location of this phosphorylation site is within an N-terminal extension that is not found within the bacterial USPA proteins (Fig. 2D).
AtPHOS32 Is Phosphorylated by AtMPK3 and AtMPK6-The phosphorylated serine residue identified in AtPHOS32 is followed by a proline. This Ser-Pro motif is common in substrates of MAP kinases, among other kinases (21). One of the well characterized responses of Arabidopsis elicited with flg22 is the rapid activation of two MAP kinases, AtMPK3 and AtMPK6 (9,10). Therefore, we tested whether AtPHOS32 was a substrate of these MAP kinases.
To generate substrates for in vitro kinase assays, we cloned the open reading frame into a vector for bacterial expression without any epitope tag. The bacterially expressed AtPHOS32 bound to Ni-NTA columns under native or denaturing condi-  Table 1). Peptide mass mapping analysis identified AtPHOS32 (At5g54430) and AtPHOS34 (At4g27320) as highly related proteins. C, direct sequencing of AtPHOS34 by nanoESI-MS/MS confirmed the accuracy of the peptide mass mapping experiment. D, sequence alignment of AtPHOS32 (At5g54430), AtPHOS34 (At4g27320), and a bacterial USPA for which the first structure was obtained (MJ0577; Zarembinski et al. 20). The serine residue in bold indicated by the arrow is the phosphorylation site identified in Fig. 4. The underlined peptides indicate the sequences obtained for AtPHOS34 by nanoESI-MS/MS that differentiate AtPHOS34 from AtPHOS32. Shaded residues are conserved between AtPHOS32, AtPHOS34, and MJ0577. Asterisks indicate the residues from MJ0577 involved in binding ATP.
tions, demonstrating that the nickel binding activity by which these proteins were initially identified is intrinsic to the protein.
As controls for kinase specificity, we similarly produced mutant proteins with either alanine (S21A) or aspartate (S21D) substituted for the phosphorylated serine residue.
To perform immunoprecipitation kinase assays, we produced peptide antibodies recognizing AtMPK3 and AtMPK6. We first tested the specificity of these antibodies using mpk3 (Fig. 5A) or mpk6 (Fig. 5B) T-DNA insertion mutants. The ␣-MPK3 antibody recognized a band of the expected size in Col-0 plants but not in two independent mpk3 mutants (Fig.  5A, upper panel). Perhaps more importantly, we demonstrated that kinase activity phosphorylating the general substrate, myelin basic protein, was immunoprecipitated only from elicitortreated Col-0 plants but not from either of the mpk3 mutants (Fig. 5A, lower panel). Similar experiments using the ␣-MPK6 antibody showed the absence of a cross-reacting protein (Fig. 5B,  upper panel) or immunoprecipitated kinase activity (Fig. 5B, lower  panel) in two independent mpk6 mutants. Therefore, these antibodies are specific for their respective kinases in immunoprecipitation kinase assays.
Using these antibodies to immunoprecipitate the kinases from flgtreated cells, we found that both AtMPK3 and AtMPK6 phosphorylated wild-type AtPHOS32 (Fig. 5C). To directly test whether the kinases phosphorylated the serine residue identified in planta, we used the bacterially expressed proteins with alanine (S21A) or aspartate (S21D) residues substituted for the target serine residue. Either mutation greatly decreased the phosphorylation of AtPHOS32 by AtMPK3 or AtMPK6 (Fig. 5C).

AtMPK3 and AtMPK6 Are the Predominant Kinases Phosphorylating AtPHOS32 in Plant Extracts-A further test of specificity involved determining whether AtMPK3 and
AtMPK6 were the only kinases that could phosphorylate the target serine residue. First, we confirmed that total protein extracts were sufficient to phosphorylate AtPHOS32 and that mutating the single serine residue to alanine abolished this in vitro phosphorylation (Fig. 6A). Therefore, specific kinase activity was present within the total protein extracts. Next, we used the antibodies raised against AtMPK3 and AtMPK6 to immunodeplete the kinases from the protein extracts (Fig. 6B). Immunodepletion of either kinase individually had a modest effect on the phosphorylation of AtPHOS32. However, if both AtMPK3 and AtMPK6 were immunodepleted, little measurable kinase activity remained in the protein extract.

DISCUSSION
Although MAPKs in plants have been demonstrated to be activated in response to many biotic and abiotic stresses (8), we currently know very little about the substrates of this family of kinases. As one could consider the phosphorylation of the substrates as the functional output of MAPK activation, defining these downstream elements is essential to understanding MAPK function within the cell. In this work, we have identified AtPHOS32 as a new substrate of the stress-regulated MAPKs, AtMPK3 and AtMPK6.
AtPHOS32 and the highly related AtPHOS34 were initially isolated from plant extracts using Ni-NTA affinity columns. Because these proteins are phosphorylated, we initially considered whether the Ni-NTA enrichment may be a form of immobilized metal affinity chromatography used to enrich for phosphopeptides. However, further analysis indicated that the ability of these proteins to bind to nickel appears to be intrinsic to the protein sequence rather than related to their status as  phosphoproteins. First, we saw few other phosphoproteins coeluting from Ni-NTA, as indicated by the lack of additional radioactive images on the two-dimensional gels with AtPHOS32 and AtPHOS34. More importantly, however, we were able to isolate the unphosphorylated, bacterially ex-pressed protein under denaturing conditions (data not shown). Therefore, the nickel binding activity most likely can be attributed to the short polyhistidine stretch at the N and C termini of the proteins (seen in Fig. 2D). Whether this metal binding activity plays a role in the function of AtPHOS32 remains to be determined.
Substantial evidence supports the conclusion that AtPHOS32 is a substrate of AtMPK3 and AtMPK6. First, AtPHOS32 becomes phosphorylated within minutes after elicitation of cells with flg22 (Fig. 1), a similar time frame for the activation of the MAPKs (9, 10). Second, AtPHOS32 is phosphorylated in planta on a serine residue within a Ser-Pro context (Fig. 3), as would be expected for substrates of MAPKs (21). Third, activated AtMPK3 and AtMPK6 efficiently phosphorylate AtPHOS32, and most if not all of this phosphorylation is abolished by mutating the target serine residue (Fig. 5C). A final piece of evidence would have been to use genetic knockouts of AtMPK3 and AtMPK6 to examine phosphorylation of AtPHOS32 in mutants lacking both MAPKs. However, as was recently described, the mpk3/mpk6 double mutant is embryo lethal (22). Therefore, we instead used immunodepletion to reduce the levels of AtMPK3 and AtMPK6 from whole cell protein extracts. After immunodepletion, no detectable kinase activity targeting the serine residue of AtPHOS32 remained (Fig. 6B). Together, these data demonstrate that AtPHOS32 is a substrate of both AtMPK3 and AtMPK6. AtPHOS32 was not identified in a large scale screen for putative substrates of these MAPKs (13), demonstrating the importance of pursuing multiple lines of investigation when attempting to biochemically characterize signaling pathways. As was found with the Feilner et al. (13) study, we showed that AtMPK3 and AtMPK6 have overlapping specificity for substrates. Why plants maintain two kinases that are coordinately activated by most stresses and display overlapping substrate specificities remains an intriguing question that needs to be addressed if we are to understand the function(s) of these multifaceted signaling pathways.
As discussed below, the function of this new substrate is not known, which impedes immediate experiments on the functional significance of the phosphorylation. However, the fact that this phosphorylation site has been conserved between dicot and monocot plant species (Fig. 4) indicates that phos-  (12)), and mpk6-4 (SALK_062471). Specificity of this antibody in immunoprecipitation kinase assays (lower panel) was demonstrated by the presence of kinase activity phosphorylating the general kinase substrate, myelin basic protein, only in extracts from Col-0 plants treated with flg22 but not in extracts from the mpk6 mutants. C, activated AtMPK3 and AtMPK6 phosphorylation of AtPHOS32 is dependent upon the predicted serine residue. Activated MAPKs were immunoprecipitated from flg22-elicited cells for in vitro kinase assays using wild type (WT) or phosphorylation site mutants (S21A or S21D) of AtPHOS32. Only the wild-type substrate is phosphorylated efficiently. FIGURE 6. AtMPK3 and AtMPK6 are the predominant kinases phosphorylating AtPHOS32. A, kinase activity of whole cell extracts specifically phosphorylate the target serine residue. In vitro kinase experiments were performed using whole cell extracts from control and elicited cell cultures using GST-PHOS32 and S21A mutants of AtPHOS32. The radioactive image is labeled with 32 P, whereas the Coomassie Blue-stained gel is labeled with CBB. B, immunosubtraction of AtMPK3 and AtMPK6 removes detectable kinase activity from whole cell protein extracts. Extracts containing the kinase are indicated by ϩ, and the absence of the kinase after immunosubtraction by the corresponding antibody is indicated by Ϫ. Only immunodepletion of both AtMPK3 and AtMPK6 eliminated detectable phosphorylation of AtPHOS32. phorylation of this protein by MAPKs is involved in a sufficiently important process to be maintained over more than 200 million years of speciation. These results further support our previous observations from large scale phosphoproteomic studies that indicate specific protein phosphorylation in plants has been conserved throughout evolution (23).
From sequence comparison analysis, the possible functions of AtPHOS32 and AtPHOS34 in MAPK signaling remain unclear. Although the amino acid sequence conservation is less than 30% (ϳ20% identity), a central domain of 150 amino was found to be structurally related to universal stress protein A (USPA) from M. jannachii (19). USPA is a small protein in bacteria that accumulates in response to a plethora of stimuli including nutrient starvation and exposure to many stresses (18), but its function is not known. The crystal structures of the M. jannaschii MJ0577 protein revealed that this protein binds ATP (20). The secondary structure of the protein as well as the residues for binding ATP are conserved in AtPHOS32 and AtPHOS34 (19) (Fig. 2D), indicating that these proteins may also bind ATP. However, the functional relevance of ATP binding has not been elucidated in bacteria. In addition, the crystal structure of another bacterial USPA protein from Haemophilus influenza (Protein Data Bank code 1JMV 24) did not indicate that this protein binds ATP. Therefore, ATP binding is not intrinsic to the USPA structural domain. The function of AtPHOS32 and AtPHOS34 as well as the role of the predicted ATP binding will be the subject of future studies.
It should be noted that the phosphorylation site identified in this study is present in an N-terminal extension that lies outside of the USPA domain. Therefore, the potential phosphoregulation of these proteins by the MAPKs is a unique aspect of this family of plant proteins.