Novel Type of Receptor-like Protein Kinase from a Higher Plant ( Catharanthus roseus ) cDNA, GENE, INTRAMOLECULAR AUTOPHOSPHORYLATION, AND IDENTIFICATION OF A THREONINE IMPORTANT FOR AUTO- AND SUBSTRATE PHOSPHORYLATION*

We characterize CrRLK1, a novel type of receptor-like kinase (RLK), from the plant Catharanthus roseus (Madagascar periwinkle). The protein (90.2 kDa) de- duced from the complete genomic and cDNA sequences is a RLK by predicting a N-terminal signal peptide, a large extracytoplasmic domain, a membrane-spanning hydrophobic region followed by a transfer-stop signal, and a C-terminal cytoplasmic protein kinase with all 11 conserved subdomains. It is a novel RLK type because the predicted extracytoplasmic region shares no similarity with other RLKs. The autophosphorylation was investigated with affinity-purified proteins expressed in Escherichia coli . The activity was higher with Mn 2 (cid:49) than with Mg 2 (cid:49) and achieved half-maximal rates at 2–2.5 (cid:109) M ATP. The phosphorylation was predominantly on Thr, less on Ser, and not on Tyr. In contrast to other plant RLK, the kinase used an intra- rather than an intermolecular phosphorylation mechanism. After protein cleavage with formic acid, most of the radioactivity was in a 14.1-kDa peptide located at the end of the kinase domain. Mutagenesis of the four Thr residues in this peptide identified Thr-720 in the subdomain XI as important for autophosphorylation and for phosphoryl- ation of (cid:98) -casein. enriched and cDNA synthesis kits from Amersham Corp. (cDNA Synthesis RPN1256Y) and Pharmacia Inc. (You-Prime cDNA Synthesis 27-9273-01). After addition of Eco RI linkers and digestion with Eco RI, the cDNAs were ligated to Eco RI-digested phage (cid:108) NM1149 (13) and packaged with a kit from Amersham Corp. (Lambda In Vitro Packaging Kit No. N334L). The screening procedures have been described (13–15). The 5 (cid:57) end of the mRNA was obtained with the cRACE technique that uses the circularization of single-stranded cDNA and polymerase chain reaction amplifications with two sets of primers (16). The position of the primers in the first cDNA were 396 3 375 (first strand cDNA synthesis), 215 3 198 and 254 3 272 (first amplification of circularized cDNA), and 103 3 81 and 327 3 348 (nested primers for second amplification). The DNA fragments were cloned blunt-end into the Sma I site of vector pTZ19R. Genomic Southern Blot and Genomic Clones— The Southern blots were performed with published procedures (15). They identified a 8.3- kbp Eco RI and a 2.5-kbp Asp 718 fragment (see “Results”), and both were analyzed. For the Eco RI fragment, the fragments in the range from 7–10 kbp were eluted from the gel and cloned into the Eco RI site of (cid:108) NM1149 (13). The fragment was recloned into vector pTZ19R after its identification by screening with the cDNA. With the Asp 718 fragment, the 5 (cid:57) end of the gene was obtained by inverse PCR (17) with the circularized 2.5-kbp genomic Asp 718 fragment, using oligonucleotides corresponding to the positions 348 3 327 and 969 3 990 of the cDNA. The blunt-ended fragment was cloned into vector pTZ19R.

Animal receptor protein kinases (RPK) 1 located in the membranes play an important role in the perception and transmittance of external signals. All RPK contain an extracellular domain connected by a membrane-spanning amino acid (AA) stretch to the intracellular kinase domain. The proteins can be subdivided into two groups that autophosphorylate on Ser/Thr or on Tyr residues, and oligomerization appears to play an important role in the regulation of enzyme activity (1).
Several plant cDNAs and genes for RPK homologs sharing the same basic domain structure have been described. They were called receptor-like kinases (RLK), because the ligands are unknown and their function as receptors has not yet been demonstrated. Several groups are presently distinguished, and the classification is usually based on the sequences in the extracellular domain (reviewed in Ref. 2): (a) the S-domain type, with similarities to the S-locus glycoproteins in Brassica; these proteins also contain ten or more cysteines in conserved positions; (b) the leucine-rich repeat type, with 9 -26 Leu-rich repeats; and (c) the epidermal growth factor (EGF)-like type that contains several EGF-like repeats. Very recent data suggest two additional types, with the extracellular region related either to plant defense proteins (3) or to lectins (4), but the investigations have not been extended much beyond the sequences.
The physiological functions of plant RLKs are unknown, except for the S-domain type RLKs from Brassica that appear to be involved in the self-incompatibility phenotype (5)(6)(7). There is considerable interest also in the other RLK because they are thought to be involved in extracellular signal perception, in particular in plant development, plant/microbe interactions, and disease resistance phenomena (reviewed in Ref. 8).
Biochemical studies after heterologous protein expression in Escherichia coli showed that autophosphorylation of the plant RLK is on Ser and/or Thr, with the possible exception of a Petunia RLK that uses Ser and Tyr residues (9). Autophosphorylation appears to occur predominantly by intermolecular phosphorylation (trans), and it has been suggested that oligomerization may be involved in the regulation of the kinase activity (10). The sites of autophosphorylation have not yet been identified in any RLK.
CrRLK1, the protein kinase from Catharanthus roseus described in this work, is interesting for several reasons. It represents a novel RLK type, and in contrast to other plant RLK it autophosphorylates predominantly with an intra-rather than with an intermolecular mechanism. We also report the first identification of a Thr residue in a RLK that is important for both autophosphorylation and the activity of the kinase to phosphorylate other proteins. The Thr is conserved in some other related plant kinases, suggesting that these may follow the same activation mechanism.

EXPERIMENTAL PROCEDURES
Plant Material-The cell suspension culture of Madagascar periwinkle (C. roseus L. G. Don, line CP3a), its maintenance on MX medium, and the nutritional downshift by incubating the cells in an 8% sucrose solution have been described (11,12).
RNA, cDNA Library, and cRACE for the 5Ј End of the mRNA-Poly(A)-enriched RNA was isolated (13) from C. roseus cell cultures treated for 7.5 h with a change from MX medium to 8% sucrose (14). The Northern blots were performed according to published procedures (15). The cDNA library was constructed using 5 g of poly(A)-* This work was supported by Fonds der Chemischen Industrie. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) Z73295.
The 5Ј end of the mRNA was obtained with the cRACE technique that uses the circularization of single-stranded cDNA and polymerase chain reaction amplifications with two sets of primers (16). The position of the primers in the first cDNA were 396 3 375 (first strand cDNA synthesis), 215 3 198 and 254 3 272 (first amplification of circularized cDNA), and 103 3 81 and 327 3 348 (nested primers for second amplification). The DNA fragments were cloned blunt-end into the SmaI site of vector pTZ19R.
Genomic Southern Blot and Genomic Clones-The Southern blots were performed with published procedures (15). They identified a 8.3kbp EcoRI and a 2.5-kbp Asp718 fragment (see "Results"), and both were analyzed. For the EcoRI fragment, the fragments in the range from 7-10 kbp were eluted from the gel and cloned into the EcoRI site of NM1149 (13). The fragment was recloned into vector pTZ19R after its identification by screening with the cDNA. With the Asp718 fragment, the 5Ј end of the gene was obtained by inverse PCR (17) with the circularized 2.5-kbp genomic Asp718 fragment, using oligonucleotides corresponding to the positions 348 3 327 and 969 3 990 of the cDNA. The blunt-ended fragment was cloned into vector pTZ19R.
DNA Sequence and Computer Analysis-The cDNA and the gene were sequenced on both strands by the dideoxynucleotide chain termination technique using vectors and phages as described in Ref. 18. The pTZ18R and pTZ19R system, helper phage M13K07, E. coli strain JM109 (Pharmacia LKB Biotechnology Inc.), and the reverse sequencing and the universal primers (Boehringer Mannheim) or custom-synthesized oligonucleotides were used with subcloned cDNA fragments. DNA polymerization reactions were performed with [ 35 S]dATP␣S (37 TBq/mmol, Amersham Corp.) and modified T7 DNA polymerase (Sequenase, Biochemical Corp.). The protein motifs were identified according to Refs. 19 and 20. TBLASTN (21) was employed for similarity searches in the data bases.
Expression of Kinase Fusions in Escherichia coli and Protein Purification-We used translational fusions of the 42.7-kDa maltose-binding protein (MBP, expressed from vector pMAL-c2; New England Biolabs) (22) with polypeptides encoded in the first cDNA. The expression was carried out with the complete coding region in the cDNA (MCPK0, fusion ϭ 126 kDa), with the N-terminal 55.3-kDa (MCPK2, fusion ϭ 98 kDa), and with the 43.2-kDa catalytic domain (MCPK1, fusion ϭ 86 kDa) (see Fig. 1C for overview). To express MCPK0, the cDNA was excised with EcoRI, the single-stranded ends were filled in, and the blunt-end fragment was inserted into the XmnI site of pMAL-c2. MCPK2 was obtained by deletion of the sequences downstream of the Asp718 site. MCPK1 was constructed by using the EcoRV site in the putative membrane-spanning region (position 1071 in the cDNA). The MBP part in the fusion proteins allowed a one-step affinity purification on amylose-resin columns (22). The procedures followed the protocol provided by the manufacturer (New England Biolabs). In some cases, the E. coli strain PR745 (New England Biolabs) was used for protein expression because it provided a higher yield of the fusion proteins and less degradation products.
Site-directed Mutagenesis-The mutagenesis was performed with suitable subfragments in vector pTZ19R and single-stranded DNA obtained with helper phage M13K07 in E. coli strain RZ1032 (23). In all cases, the original fragments in the pMAL-c2 expression constructs were exchanged against the mutated fragments after verification of the changes by DNA sequence analysis. The expression of the correct size fusion proteins was confirmed, and the proteins were affinity-purified as described above.
The essential Lys (position 499) in the phosphotransfer site was changed to Arg with the oligonucleotide 5Ј GTA GCT GTG AGA AGA GGG 3Ј (mutated base underlined), and the proteins containing the mutation were called MCPK0* and MCPK1* (Fig. 1C). A subclone containing the 3Ј-terminal Asp718 fragment was used to exchange the four Thr residues in the C-terminal 14.1-kDa polypeptide (positions 720, 734, 767, 795) to Ala with appropriate custom-synthesized oligonucleotides.
Identification of Phosphorylated AA-The incubations followed the standard assays except for the following modifications. The amount of MBP-protein kinase fusion was raised to 10 g, the [␥-32 P]ATP was increased to 10 Ci, and the incubations were carried out at 37°C for 40 min. The radioactive proteins from 10 assays were analyzed by gel electrophoresis, blotted to a nitrocellulose membrane, and stained with Ponceau S red (0.5% solution in 10% acetic acid). The 126-kDa bands were excised and treated with 6 ml of 70% formic acid for 48 h with vigorous shaking at 37°C to cleave the protein between Asp-Pro residues (24) and to solubilize the polypeptides (25). The nitrocellulose membrane was removed, and the extract was dried by vacuum centrifugation. The residue was dissolved in 70 l of sample buffer and subjected to Tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis (26). Labeled polypeptides were detected by autoradiography.
For phosphoamino acid analysis, the polypeptides were blotted to Immobilon membranes. The most heavily labeled peptide (14.1 kDa) was excised and hydrolyzed under nitrogen for 90 min at 110°C in 6 N HCl (27). The hydrolysate was dried in a vacuum centrifuge and washed twice with distilled water. The residue was dissolved in 10 l of distilled water, spotted onto a thin layer cellulose plate, and analyzed by ascending chromatography with a solvent consisting of propionic acid, 25% ammonium hydroxide, and isopropyl alcohol (45:17.5:17.5; v/v/v) (procedure modified from Ref. 28). The 32 P-labeled AA were detected by autoradiography. Standards with authentic phosphothreonine, phosphoserine, and phosphotyrosine (Sigma) were run in separate lanes and visualized with a ninhydrin spray (Sigma).
Antiserum and Immunoblots-A fragment containing the complete 2667-bp cDNA, but with the Lys in the phosphotransfer site mutated to Arg, was inserted into the NcoI site of the expression vector pQE-6 (29), thus providing a start codon and the signals necessary for regulated protein expression. The plasmid was maintained in E. coli strain M15[pRep4] (29). The cultures were grown in Luria-Bertani medium (1% Bacto-tryptone, 0.5% Bacto-yeast extract, 0.5% NaCl, 0.2% glycerol, 0.02% MgSO 4 (pH 7.5)), and protein expression was induced with 2 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C. The protein (CPK0*) was isolated with the techniques detailed in Strebel et al. (30) and then used to raise antibodies in a hen. The immunoblots (14) were performed with the IgY fraction (BioTools, Denzlingen, Germany) and a secondary antibody (rabbit anti-chicken) coupled with alkaline phosphatase (Sigma).

Molecular Analysis of CrRLK1, a Novel Type of Plant RLK-
The first cDNA was identified fortuitously during a screen of a library from cell cultures transferred from the standard growth medium to an aqueous 8% sucrose solution. This nutritional downshift was originally used for the induction of alkaloid biosynthesis (11,12) and more recently for the cloning of induced cytochrome P450 (14). The cDNA contained 2667 bp and an open reading frame of 750 AA that started in the EcoRI site and terminated at position 2251 (see scheme in Fig. 1A). Similarity searches in the data bases indicated that the C-terminal 300 AA represented the catalytic domain of a protein kinase that was separated from the large N-terminal region by a putative membrane-spanning stretch. This suggested that the cDNA encoded a RLK.
The expected N-terminal hydrophobic signal sequence was absent in the cDNA, and the size of the mRNA (3 kb, Fig. 2B) suggested that the cDNA lacked the 5Ј end of the mRNA. The RNA blots also showed that the mRNA was of very low abundance and that it was in fact not induced by the nutritional downshift of the cultures. The 5Ј end was obtained with the cRACE technique, and several independent clones indicated that the mRNA started at least 336 bp upstream of the EcoRI site. The first AUG was 159 bases upstream of the restriction site, and this increased the size of the protein by 53 AA.
The genomic blot revealed a simple pattern of strongly hybridizing fragments ( Fig. 2A), suggesting a single copy gene. The 8.3-kbp EcoRI fragment was cloned and analyzed. The kinase sequence was located at the end of the fragment, terminating in an EcoRI site that corresponded to that at the 5Ј end of the cDNA. No introns were detected in the protein coding region, but the cDNA and genomic sequences diverged in the last 109 bp of the 3Ј noncoding region in the cDNA. A repeated data base search showed that the sequences in the cDNA were from E. coli (x in Fig. 1A). The genomic DNA contained at the point of divergence an almost perfect direct repeat of 43 bp (rr in Fig. 1B). The junction with the E. coli sequences was pre-cisely between the repeats, suggesting that the repeat was responsible for the recombination. The 5Ј end of the gene was obtained by inverted PCR with the 2.5-kbp Asp718 fragment indicated by the genomic blot ( Fig. 2A). It also contained no introns.
The deduced protein CrRLK1 (Fig. 3) contained 803 AA (calculated size 90.2 kDa). Its predicted domain structure followed the definition of the RLK type kinases (2): a hydrophobic N-terminal putative signal peptide, a large N-terminal region that is thought to be extracytoplasmic, a putative hydrophobic membrane-spanning stretch of 24 AA followed by a membrane transfer-stop signal (31), and a C-terminal kinase catalytic domain with all 11 subdomains described for protein kinases (20). The highest similarity scores for the kinase domain (58% and 54% identity, respectively) were obtained with the Lycopersicon pimpinellifolium kinases Pto (32) and Fen (33); interestingly, these do not belong to the RLK class because they consist of little more than the catalytic kinase domain (Fig. 3).  (32,33). In CrRLK1: , putative hydrophobic signal sequence, membrane-spanning region, and membrane transfer stop signal (RRRKR).
[PK], start of the catalytic domain; I to XI, subdomains conserved in protein kinases (20). Protein kinase signatures (19): ATPbinding, S/T kinase. The essential Lys in the phosphotransfer site is marked with K (position 499). N in bold print, putative N-glycosylation sites. DP between subdomain IX and X, cleavage site for the C-terminal 14.1-kDa peptide obtained after treatment with formic acid. T, the four Thr in the C-terminal 14.1-kDa peptide. In Fen and Pto: dots, sequence identity; underlined, sequence similarity; dashes, gaps introduced for optimal alignment. The identity scores between the catalytic domain of CrRLK1 and those of the other RLKs were 35-42%, and these values appear to be typical (2). CrRLK1 revealed 14 putative N-glycosylation sites, and 11 were located in the N-terminal domain (Fig. 3). This distribution is characteristic for RLK, and there is evidence that they are glycoproteins (34). The N-terminal 450 AA of CrRLK1 revealed no similarity to the extracellular regions in the S-domain type RLK, it contained no leucine-rich repeats or similarities to the EGF-like class RLK, and it also revealed no relation to pathogenesis-related proteins or lectins. CrRLK1 therefore represents a novel RLK type.
Autophosphorylation Properties-Initial attempts to express the cDNA-encoded protein in E. coli failed because no transformants were obtained, and this indicated that the protein was toxic to the bacteria. We therefore constructed fusion proteins in which the maltose-binding protein (MBP) from E. coli was fused in-frame with various parts of the kinase cDNA. The constructs are summarized in Fig. 1C. These fusions had no obvious toxic effect on E. coli, and they also had the advantage that they could be purified by affinity chromatography for autophosphorylation assays.
Most of the experiments characterizing the autophosphorylation properties were performed with MCPK0 (Fig. 1C). The results in the standard assay with 10 mM divalent cations showed that the autophosphorylation was completed after about 35-45 min (Fig. 4A), and, therefore, 25-min incubations were used routinely. The autophosphorylation with MnCl 2 was about 10-fold higher than with MgCl 2 , and no activity was observed with CaCl 2 (Fig. 4B). About half-maximal activity was reached with 0.5-0.75 mM MnCl 2 (Fig. 4D), and a linear in-crease was observed with MgCl 2 up to the highest concentration tested (Fig. 4E). The temperature dependence revealed the highest values at 37°C and no activity at 0°C (Fig. 4C). The kinase used ATP at about a 5-fold higher rate than GTP, and unlabeled GTP inhibited the autophosphorylation with labeled ATP (not shown). The autophosphorylation with ATP revealed half-maximal rates at 2-2.5 M ATP (Fig. 4F).
The question of whether the protein phosphorylated strictly itself (intramolecular) and/or other kinase molecules (intermolecular) was investigated with the purified MBP-fusion proteins described in Fig. 1C. Fig. 5 shows the stained proteins used in the assays (A) and the autoradiographies of the phosphorylated proteins (B). MCPK0 (126 kDa, complete cDNA-encoded protein) and MCPK1 (86 kDa, only the catalytic domain) were both active in autophosphorylation (lanes 1 and  2), and there was no obvious difference between the two proteins. MCPK1* was a mutant in which the essential Lys (Lys-499 in the complete protein) in the phosphotransfer site of MCPK1 had been mutagenized to Arg, and the protein was inactive (lane 3). This protein was then co-incubated with MCPK0 to test whether the active kinase phosphorylated the inactive mutant. The result showed no significant phosphorylation of MCPK1* (lane 5, the weak band at 80 kDa is a degradation product from MCPK0, see lanes 1 and 6). The same result was observed in the combination of MCPK1 (active) with MCPK0* (inactive mutant) (not shown). The kinase also did not trans-phosphorylate MCPK2 (98 kDa) that contained the Nterminal part of the kinase plus a small part of the catalytic domain (lane 6); as expected, this protein did not autophosphorylate (lane 4). Other experiments (not shown) confirmed the finding with MCPK2 that the MBP-part of the fusion proteins was no substrate for the kinase. The two active kinase fusions (MCPK0, MCPK1) both produced radioactive bands in co-incubations (lane 8), indicating that they did not hinder each other in activity. In summary, these results showed that the C. roseus kinase predominantly performed an intramolecular autophosphorylation, and an intermolecular phosphorylation of other kinase molecules was only detectable after very long exposures.
The results after total hydrolysis of both MCPK0 and MCPK1 and subsequent analysis of the radioactive AA showed that the autophosphorylation was mostly on Thr (80%) and much less in Ser residues (20%), and radioactive Tyr was not detectable. Identification of a Thr Residue Important for Auto-and Substrate Phosphorylation-MCPK0 was used to identify the region of the protein that was autophosphorylated. The radioactive protein was treated with formic acid to cleave between Asp and Pro residues, and the fragments were separated by electrophoresis. Fig. 6A shows that most of the radioactivity was in a 14.1-kDa polypeptide. Counting from the N-terminal, the MCPK0 protein sequence indicated that polypeptides of 34.6 (MBP-part), 26.1 (partly MBP), 6.5, 41.2, 3.7, and 14.1 kDa were to be expected from the formic acid cleavage. Due to the large size differences, the labeled 14.1-kDa peptide could be unambiguously located to the C-terminal of the kinase, more precisely to the region containing the subdomains X and XI of the kinase catalytic domain and the C-terminal end of the protein (see Fig. 3, the cleavage site is marked with DP). Most of the other radioactive polypeptides could be explained as partial digests containing the 14.1-kDa peptide. The exception was a 32.6-kDa protein (Fig. 6A) that could represent the sum of the 26.1-and the 6.5-kDa peptides, and this suggested a low level of phosphorylation at the N-terminal. The 14.1-kDa peptide was isolated, and the analysis after total hydrolysis demonstrated about 90% of the radioactivity in Thr, the rest in Ser, and none in Tyr (Fig. 6B).
The 14.1-kDa peptide contained four Thr residues (positions 720, 734, 767, and 795 in CrRLK1; marked by T in Fig. 3). All four were mutagenized in MCPK0 individually to Ala to investigate their role in the autophosphorylation. Assays with the affinity-purified proteins showed that the Thr-720 mutant was nearly inactive (Fig. 7B), although the protein was present in the same amount as the unmodified protein (Fig. 7A). Only long exposures of the film revealed a clear radioactive band, but it was too weak to attempt an identification of the phosphorylated AA. It may represent the Ser autophosphorylation that was detected after total hydrolysis of MCPK0. The mutated DNA was completely resequenced, but no other differences from MCPK0 were detected, indicating that the single Thr-720 3 Ala exchange was responsible for the effect. The other Thr mutants were not significantly affected in autophosphorylation, except for the Thr-795 mutant that showed a somewhat weaker activity (Fig. 7B). The ability of the proteins to phosphorylate other polypeptides was tested with ␤-casein, and Fig.  7C shows that the Thr-720 mutant was incapable of phosphorylating this substrate. The other proteins did not reveal such drastic effect, except that the Thr-767 mutant showed an approximately 5-fold increase in ␤-casein phosphorylation (Fig. 7C).
These results identified Thr-720 in the subdomain XI of the catalytic kinase domain as a residue important for both auto-and substrate phosphorylation, and they showed that Thr-767 influenced the extent of substrate phosphorylation.

CrRLK1
Represents a Novel RLK Type-The C. roseus protein belongs to the RLK group of plant kinases that has attracted much attention recently because of the likely roles in the perception and transduction of external signals (2). The ligands and the receptor-function, however, have not yet been demonstrated, and the physiological roles are in most cases a matter of speculation. CrRLK1 is no exception, and its function in metabolism remains to be explored.
The molecular data indicate a single copy gene that is expressed at very low levels under all conditions that have been tested so far. The absence of introns is unusual, and the only other known example for an RLK gene is ZmPK1, a S-locus type RLK from maize (35). All other RLK genes contain 1 (36, 37), 2 (9), 5-6 (3, 5, 38, 39), or even 26 introns (40).
Intramolecular Autophosphorylation and Properties-The only other RLK investigated in detail (RLK5 from A. thaliana) used predominantly trans-phosphorylation, suggesting that oligomerization may play a role in activation (10). The experiments with RLK5 also employed fusion proteins (with MBP and glutathione S-transferase), and therefore it seems very unlikely that a different approach was responsible for the difference to CrRLK1. The two RLK5 fusion proteins had very similar apparent K m values for ATP (17.8 and 15.2 M), suggesting that the MBP or the glutathione S-transferase parts of the proteins had no differential effect on the enzyme activity. The experiments with the C. roseus MBP-fusion indicated half-maximal rates at 2-2.5 M ATP, and the higher affinity is likely to reflect one of the differences between intra-and intermolecular phosphorylation. This property is interesting because it suggests that CrRLK1 activation does not require oligomerization.
A higher autophosphorylation activity with Mn 2ϩ than with Mg 2ϩ has also been described for other RLK (10,34), but the significance in vivo is not clear. The preferential phosphorylation of Thr residues was also reported for a Leu-rich (37) and a S-domain type RLK (41), but a generalization is not possible because a preference for Ser was described for other RLK of the S-domain type (42).
The results with the Thr-720 3 Ala mutant suggest that this Thr is the autophosphorylated residue. It should be noted, however, that it is not possible at this point to rigorously exclude the possibility that Thr-720 is necessary for these activities, but is not phosphorylated itself. The exchanges of the other Thr residues had no such drastic effects on autophospho- rylation. However, complex roles in the regulation of enzyme activity are possible, as shown with the Thr-767 mutant that revealed a drastic stimulation of ␤-casein phosphorylation (Fig.  7C). The Ser autophosphorylation remains to be investigated, but this will be difficult because it is very low and because of the high abundance of Ser residues in the protein. The complete loss of ␤-casein phosphorylation with the Thr-720 mutant suggests that Ser autophosphorylation plays a minor role, if any, in the phosphorylation of other proteins.
The Thr Important for Auto-and Substrate Phosphorylation Is Conserved in Other Plant Protein Kinases-This is the first identification of a residue in a plant kinase that is important for auto-and substrate phosphorylation. The other plant RLK do not contain a Thr in the position corresponding to Thr-720 in CrRLK1, but the most closely related kinases (Pto and Fen from tomato) do have a Thr (Fig. 8). These do not belong to the RLK group because they contain not much more than the catalytic kinase domain, but they may be membrane-anchored via a myristyl residue (43). Interestingly, these proteins also predominantly autophosphorylated on Thr residues (44), but a further characterization is not available. The EST (expressed sequence tag) data base contained three partial putative plant kinase sequences with about 50% identity to the C-terminal of Pto, Fen, and CrRLK1, and these also had a Thr at the same position (Fig. 8). It would be interesting to see whether the proteins belong to one of the known RLK types. Remarkably, all of the related proteins contain a Ser at the position of Thr-734 (end of subdomain XI) in CrRLK1 in a region that is otherwise highly conserved (Fig. 8). A possible functional significance of these similarities remains to be elucidated. FIG. 8. Threonines in the C-terminal 14.1 kDa of CrRLK1 and alignment with related sequences. The numbers refer to the AA position in the complete kinase. Dots, identical AA; underlined, sequence similarity. The peptides OSR07581A (from rice, Oryza sativa) and ATTS4134 and H36409 (both from Arabidopsis thaliana) were translated from the partial cDNA sequences in the EST data base; ?, unknown residue (undefined base in sequence).