HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 280, Issue 24, 23073-23083, June 17, 2005

This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 280/24/23073    most recent M500992200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Bernal, A. J. Articles by Michelmore, R. W. Search for Related Content PubMed PubMed Citation Articles by Bernal, A. J. Articles by Michelmore, R. W. Social Bookmarking                      What's this?

Functional Analysis of the Plant Disease Resistance Gene Pto Using DNA Shuffling^*

Adriana J. Bernal, Qilin Pan, Jeff Pollack, Laura Rose, Alexander Kozik, Neil Willits, Yao Luo, Muriel Guittet, Elena Kochetkova, and Richard W. Michelmore

From the Department of Plant Pathology, Department of Statistics, The Genome Center and Department of Plant Sciences, University of California, Davis, California 95616

Received for publication, January 26, 2005 , and in revised form, March 23, 2005.

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Pto is a serine/threonine kinase that mediates resistance in tomato to strains of Pseudomonas syringae pv. tomato expressing the (a)virulence proteins AvrPto or AvrPtoB. DNA shuffling was used as a combinatorial in vitro genetic approach to dissect the functional regions of Pto. The Pto gene was shuffled with four of its paralogs from a resistant haplotype to create a library of recombinant products that was screened for interaction with AvrPto in yeast. All interacting clones and a representative sample of noninteracting clones were sequenced, and their ability to signal downstream was tested by the elicitation of a hypersensitive response in an AvrPto-dependent or -independent manner in planta. Eight candidate regions important for binding to AvrPto or for downstream signaling were identified by statistical correlations between individual amino acid positions and phenotype. A subset of the regions had previously been identified as important for recognition, confirming the validity of the shuffling approach. Three novel regions important for Pto function were validated by site-directed mutagenesis. Several chimeras and point mutants exhibited a differential interaction with (a)virulence proteins in the AvrPto and VirPphA family, demonstrating distinct binding requirements for different ligands. Additionally, the identification of chimeras that are both constitutively active as well as capable of binding AvrPto indicates that elicitation of downstream signaling does not involve a conformational change that precludes binding of AvrPto, as previously hypothesized. The correlations between phenotypes and variation generated by DNA shuffling paralleled natural variation observed between orthologs of Pto from Lycopersicon spp.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The outcome of challenges by potential pathogens on a plant is determined by multiple molecular interactions between host and pathogen. In bacterial diseases, some of these interactions involve effector proteins that are secreted directly into the host cell via the type III secretion system (1). When plants evolve to recognize an effector protein, a compatible interaction becomes incompatible, and the effector protein acquires an avirulence phenotype. Disease resistance genes have now been isolated from a wide variety of plants to diverse pathogens (2). The most abundant class of resistance genes encodes for proteins with nucleotide binding site and leucine-rich repeat domains (3). It is predicted that resistance proteins act in macromolecular complexes and that genes encoding different components can vary genetically (4, 5). Usually, variation occurs in nucleotide binding site-leucine-rich repeat encoding genes, the most common class of resistance genes. In some cases, however, other types of genes may determine the natural variation of resistance. This is the case of resistance in tomato to Pseudomonas syringae pv. tomato, where the observed intraspecific variation in specificity is determined by differences in the Pto gene that encodes a kinase, rather than in Prf that encodes a nucleotide binding site-leucine-rich repeat protein (6, 7).

The interaction between P. syringae pv. tomato (Pst) and its solanaceous hosts is currently one of the best understood plant-pathogen interactions at the molecular level (8). Resistance to Pst strains expressing the avirulence proteins AvrPto or Avr-PtoB is conferred by the Pto gene that encodes a functional serine/threonine protein kinase of 321 amino acids (9). Prf, a nucleotide binding site-leucine-rich repeat-encoding gene located within the Pto locus, is also required for this resistance (7). A direct interaction between the Avr proteins and Pto was observed in the yeast two-hybrid system (10–12). However, no interaction has been observed between Prf and Pto. Mutations in either Pto or AvrPto that disrupted the physical interaction in yeast resulted in the losses of disease resistance and the hypersensitive response (HR)¹ in planta (13, 14). AvrPto and AvrPtoB are introduced by the bacterial type III secretion system into the plant cell. The binding of AvrPto or AvrPtoB is thought to activate Pto, possibly by changes in protein phosphorylation and conformation (15–17). The activated Pto protein then induces a resistance response in a Prf-dependent manner, and pathogen growth is inhibited.

Pto belongs to a small multigene family that spans 65 kb (GenBank^TM AF220602 [GenBank] and AF220603 [GenBank] ).² The five tandem gene family members (paralogs) share 78–91% nucleotide identity with Pto (Supplementary Table S1). However, only Pto interacts with AvrPto in yeast and induces an HR in an AvrPto-dependent manner in planta (18). A Pto paralog, LescPth5, in the susceptible haplotype in Lycopersicon esculentum recognizes AvrPto in planta but does not confer complete resistance to bacteria expressing AvrPto. Fen, a paralog in the resistant haplotype, encodes sensitivity to the organophosphate insecticide Fenthion (19). Several paralogs can be activated by gain-of-function substitutions (18). All phenotypes of Pto, Fen, and gain-of-function variants are Prf-dependent (18, 20).

Several functional domains of Pto have been characterized. Ethyl methane sulfonate-induced mutagenesis identified three residues required for function in planta (10, 11, 21). Two phosphorylation sites are required for binding AvrPto in yeast and recognition in planta, respectively (22). Swaps between Pto and Fen together with site-directed mutagenesis identified a single amino acid (Thr²⁰⁴) in kinase subdomain VIII responsible for AvrPto binding specificity (10–12, 23). However, whereas these Fen-Pto hybrids did bind AvrPto in the yeast two-hybrid assay, wild type (Pto) levels of binding were not observed, indicating that other regions in the protein also participate in binding. Recently, a patch of residues on the surface of the protein was identified as a negative regulator for downstream signaling using three-dimensional structure prediction of sites located in the vicinity of the p + 1 loop of Pto (17). Residues critical for AvrPto and AvrPtoB recognition were also identified in a surface area that partially overlapped with the patch mediating negative regulation.

DNA shuffling is a PCR-based combinatorial method for generating large numbers of variant progeny molecules from a set of sequence-related template molecules in vitro (24). Multiple versions of a gene (alleles, orthologs, or paralogs) from the same or different species are fragmented into small pieces (50–200 bp), allowed to reanneal, and reconstituted into chimeric genes from the different templates. DNA shuffling of naturally occurring genes or in vitro generated mutants has been used on a wide spectrum of proteins to enhance their performance rather than to dissect protein function (25). DNA shuffling has been used together with domain swaps in the tomato Cf4 and Cf9 genes, which confer resistance to Cladosporium fulvum, to identify regions responsible for recognition specificity (26). These Cf genes are more than 91% identical at the amino acid level. Consequently, direct visual comparisons between chimeras of these two templates were informative for the identification of regions defining specificity. However, resolution to a single amino acid level was not achieved with the analysis of chimeras obtained by DNA shuffling. Analysis of more divergent, multiple templates, such as the Pto gene family described in this paper, required statistical approaches to identify specificity determinants, since functional and nonfunctional chimeras not only differ in the specificity determinants but also in other regions. Additionally, constitutively active molecules were obtained by DNA shuffling of Cf9 homologs (27). In this case, regions responsible for the phenotype could not be identified, probably due to the low number of chimeras.

In this paper, we describe the use of DNA shuffling to dissect functional regions and individual amino acids of Pto that are important for the recognition of the cognate avirulence proteins and for signaling downstream. The importance of two regions previously identified by domain swaps was confirmed. Two regions identified by shuffling confirm the parallel site-directed analysis reported by Wu et al. (17). In addition, three regions important for recognition of the Avr proteins and signaling downstream were characterized that had not been identified previously. The importance of these regions was validated by site-directed mutagenesis. In two of the regions involved in Avr recognition, the amino acids required for interaction with AvrPto and AvrPtoB and homologs are similar but not identical. Furthermore, the phenotypes of multiple chimeras have implications for the role of conformational changes in the activation of Pto for downstream signaling. In addition, regions of Pto that can tolerate variation without abrogating function were also identified.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Shuffling of Pto and Homologs—The paralogs LpimPto, LpimFen, LpimPth2, LpimPth3, and LpimPth4 from the resistant haplotype (Supplementary Table S1 and GenBank^TM AF220602 [GenBank] ) were amplified from pCR Blunt (Invitrogen) clones derived from L. esculentum cv. Rio Grande 76R (18) using PCR primers with the attB recombination sites for the Gateway BP reaction (Supplemental Table S3) and cloned into the Gateway pDONR207 vector (Invitrogen). They were then amplified by PCR using Pfu DNA polymerase (Stratagene, La Jolla, CA) and vector-specific oligonucleotides (Supplemental Table S3). These PCR products were purified, and 2–5 µg of DNA were digested with 0.01 units of DNase I/µl (Sigma) in 50 µl of 50 mM Tris, pH 7.4, 1 mM MgCl₂ for 20 min at room temperature. Products were separated by electrophoresis; DNA fragments of 50–200 bp were excised, purified, and resuspended in 30 µl of 1x PCR buffer (Promega, Madison, WI) to a final concentration of 20–30 ng/µl. Five µl of each template treated with DNase were mixed together and amplified without primers using 2.5 units of Taq polymerase (Promega) per 100-µl reaction. A PCR program of 94 °C for 60 s, 94 °C for 30 s, 50 °C for 30 s, 72 °C for 30 s (40 cycles), and 72 °C for 5 min was used. The PCR products were diluted and amplified for 20 additional cycles (94 °C for 30 s, 50 °C for 30 s, 72 °C for 30 s) following the addition of primers Q43 and Q44, which contain the attB recombination sites for the BP reaction in the Gateway system (Supplemental Table S3). A single product of the correct size (1 kb) was obtained.

Plasmids and Cloning Procedures—Most cloning procedures utilized the Gateway cloning technology (Invitrogen). The yeast two-hybrid vectors pAS2.1 and pACT2 (Clontech, Palo Alto, CA) as well as the small binary vector for plant transformation pCB302-3 (20) were adapted as destination vectors using the protocol from Invitrogen. The resulting vectors had the Gateway attR cassette in the reading frame B and were designated pAS2.1GB, pACT2GB, and pCB302-3GB. The gene clones in pCB302-3GB were then under the control of the constitutive 35S promoter. The shuffled library was obtained by cloning the shuffled PCR products into pDONR207 using the BP clonase (Invitrogen). The resulting sequences were then transferred into pAS2.1GB using LR clonase (Invitrogen). A BP and subsequent LR ligation were used to transfer individual clones from pAS2.1GB to pCB302-3GB using pDONR207 as the intermediate. The shuffled clones selected for yeast and plant phenotyping were sequenced from both strands of pDONR207 using the primers PthF and PthR (Supplemental Table S3). The nonredundant set of clones in yeast and Agrobacterium tumefaciens were resequenced to confirm that they were identical to the corresponding clones in pDONR207.

The bacterial genes encoding effector proteins were cloned in pACT2 for yeast two-hybrid analyses. avrPto was amplified from pCR Blunt (Invitrogen) (13) and cloned into pACT2GB using pDONR207 as intermediate. avrPtoB was amplified from genomic DNA of P. syringae pv. tomato strain DC3000 (avrPtoB_PstDC3000) using gene-specific primers for AvrPtoB (12) (Supplemental Table S3) and a 10:6 mixture of Klentaq (DNA Polymerase Technology, St. Louis, MO) and Pfu (Stratagene) polymerases. The genes were then cloned into pCR2.1 using the TOPO TA cloning kit (Invitrogen), and their sequences were verified. avrPtoB_PstDC3000 was then cloned as an NcoI-XhoI fragment into the NcoI-XhoI sites of pACT2. avrPtoB_PstT1 was cloned using NcoI-SacI and SacI-XhoI fragments into the NcoI-XhoI sites of pACT2. The avrPtoB homologs virPphA_Pph, virPphA_Pgy, and virPphA_Psv (from Pseudomonas savastanoi pvs. phaseolicola, glycinea, and savastanoi, respectively) (29, 30) were amplified from clones provided by Dr. John Mansfield (Imperial College at Wye, Ashford, UK) and cloned into pCR2.1. The virPphA_Pph clone was digested with XhoI, filled-in using T4 DNA polymerase (New England Biolabs), and digested with BamHI; the resulting fragment was cloned into pACT2 that had been previously digested with EcoRI, filled in with T4 DNA polymerase, and cut with BamHI. virPphA_Pgy and virPphA_Psv were cloned as EcoRI-BamHI fragments into the EcoRI-BamHI sites of pACT2. The resulting clones were transformed into Y187 for yeast two-hybrid analyses.

Site-directed mutagenesis was carried out by overlapping extension PCR (31) using Klentaq and Pfu polymerases and oligonucleotides that introduced the desired mutations (Supplemental Table S3). PCR products were cloned into pDONR207 and sequenced. Clones with the desired sequence were then transferred to pCB302-3GB and pAS2.1GB.

Pto and Pth3 variants for co-expression with AvrPtoB in tomato were PCR-amplified using oligonucleotides that introduced an NcoI and XbaI site at the 5' and 3' end of the genes, respectively. The PCR products were individually cloned into pCR2.1, and their sequence was verified. The resulting constructs were subcloned into the NcoI-XbaI sites of pSLJ4D4.1, between the CaMV 35S promoter and the octopine synthase terminator (32). The entire expression cassette was then inserted as a HindIII fragment into the HindIII site in the MCS1 of pCB302-3GB that had avrPtoB in the MCS2.

Yeast Strains and Yeast Two-hybrid Analysis—Yeast two-hybrid analysis was performed using the MATCHMAKER GAL4 system (Clontech). Yeast strains were grown and transformed using the media and protocols described in the yeast protocols handbook (Clontech). Both strain Y187 (MAT, ura3-52, his3-200, ade2-101, trp1-901, leu2-3, 112, gal4, met^–, gal80) and strain Y190 (MATa, ura3-52, his3-200, lys2-801, ade2-101, trp1-901, leu2-3, 112, gal4, gal80, cyh^r2) contain a chromosomal lacZ gene with the upstream regulatory elements replaced with GAL4 binding sites (GAL1_UAS-GAL1_TATA-lacZ).

To screen the library of shuffled molecules, bait protein fusions (GAL4AD:Avr protein) were expressed from plasmid pACT2GB, and prey protein fusions (GAL4 DNA BD:Pto chimeras) were expressed from plasmid pAS2.1GB. The library in pAS2.1GB was transformed into strain Y190 carrying GAL4AD:avrPto as bait. The library screen was performed following manufacturer's instructions (Clontech). Double transformants were selected for growth on SD (–leu, –trp), and positive interactions identified by -galactosidase assays were performed as described by Clontech.

Analysis of individual sequences utilized a yeast mating strategy. Haploid bait colonies of strain Y187 (MAT, transformed with pACT2GB:AvrPto or pACT2GB:AvrPtoB) were mated with haploid prey colonies of Y190 (MATa, transformed with pAS2.1GB:Pto chimeras) as described by Finley and Brent (33). Yeast colonies were then transferred to liquid media in 96-well ELISA plates, and their A₆₀₀ adjusted to 0.2. They were then replicated onto fresh SD (–leu, –trp) plates. Yeast were incubated at either 22 or 28 °C for 24 h, and colony lift -galactosidase assays were performed (Clontech). The standard Y2H temperature of 28 °C is higher than temperatures reflecting normal in planta conditions. Therefore, assays were conducted at two temperatures: at 28 °C, the standard yeast two-hybrid assay conditions, and 22 °C, reflecting more physiological conditions under which these proteins interact naturally. The lower temperature would also accommodate possible temperature-sensitive molecules generated by the shuffling process. Color intensities were recorded after 6 h of incubation. For the site-directed mutants, diploid yeast colonies were streaked onto SD medium (–leu, –trp) and incubated for 36 h at 28 °C. Colony lift assays were performed as recommended (Clontech). Color intensities were recorded after 6 h of incubation.

For Western blot analysis, yeast colonies were grown in SD (–trp) to an A₆₀₀ of 0.5–0.6. Proteins were extracted by resuspension of the yeast pellet using glass beads and MURB buffer (50 mM sodium phosphate, pH 7.0, 25 mM MES, pH 7.0, 3 M urea, 1% SDS, 10% -mercaptoethanol, 0.01% bromphenol blue). Proteins were separated in a 15% SDS-polyacrylamide gel and detected using GAL4 DNA-BD monoclonal antibody (Clontech).

Agrobacterium-mediated Transient Assays in Planta—Agrobacterium strains C58C1(pCH32) and 1D1249 (35) were transformed with variants of pCB302-3 using electroporation. Transient assays of Nicotiana benthamiana were performed as previously described (15), except for the infiltration media composition (10 mM MES, pH 5.6, 10 mM MgCl₂, 150 µM acetosyringone). For all procedures, A. tumefaciens C58C1(pCH32) containing pCB302-3 clones was grown in LB containing kanamycin at 50 µg/ml and tetracycline at 5 µg/ml. Wild type N. benthamiana and plants expressing AvrPto under the dexamethasone-inducible promoter (34) were assayed as described by Chang et al. (18). The presence or absence of an HR was scored 2 and 4 days after dexamethasone induction.

For transient assays in tomato, colonies of A. tumefaciens strain 1D1249, which does not elicit an HR when infiltrated into leaves of tomato (35), were grown for 24 h in LB containing 66 µg/ml kanamycin at 250 rpm. Bacteria were washed as for the experiments with N. benthamiana and infiltrated at an A₆₀₀ of 1.0. The presence or absence of HR was recorded at 2 and 4 days after inoculation.

Data Management and Statistical Analysis—An alignment of the predicted amino acid sequences was generated using ClustalW (available on the World Wide Web at clustalw.genome.ad.jp/) and refined manually using GeneDoc (available on the World Wide Web at www.psc.edu/biomed/genedoc/). Duplicated sequences were identified using GeneDoc. A MySQL database was generated to assist in data management and analysis (available on the World Wide Web at michelmorelab. ucdavis.edu/shuffledb/data). This database contains all of the protein and nucleotide sequences of the chimeras as well as the phenotypes associated with them. The protein alignment is displayed with the location of the 11 kinase subdomains; amino acids in the alignment can be color-coded. The origin of each fragment among the templates can be graphically displayed from the Web site. An SQL-based program was generated to analyze the distribution of recombination sites within the protein with user-specified sliding window sizes.

Correlations between amino acid and phenotypes were analyzed using SAS (SAS Institute, Cary, NC). For the analysis of yeast two-hybrid interactions with Avr proteins, the alignment was converted into a matrix of binary data as follows: "1" was assigned to each amino acid residue in the shuffled molecules if it matched the residue present in Pto, and "0" was assigned if it did not. For the analysis of the ability to signal downstream, the alignment was converted into binary data as follows: "1" was assigned to each amino acid residue if it matched either Pto, Pth3, or Fen, and "0" was assigned if it did not. These groupings were used, because previous studies (18) had shown that these molecules were capable of inducing an HR (i.e. they were competent to signal downstream). Univariate logistic regression analysis (36) at each position was carried out for the transformed data. This resulted in a ² value that represented the correlation between the variation at each amino acid position and the phenotype (binding in yeast or induction of HR in planta).

Significance thresholds for ² values were calculated for each test using permutation analysis (37). The phenotypes of the chimeras were randomized against their sequences 500 times, and the analysis was carried out in the same manner as with the actual data for each of the 500 permutations. The highest ² value was recorded for each set of permuted data. The p = 0.05 significance threshold was determined by the 5% highest ² values among the 500 analyses.

Three-dimensional Modeling of Pto—Sequences of protein kinases for which the three-dimensional structure was known were selected based upon similarity to Pto using sequence and structural similarity methods. Sequence similarity methods used included PSI-BLAST (available on the World Wide Web at www.ncbi.nlm.nih.gov/blast) and MODELLER (38) (available on the World Wide Web at www.salilab.org/modeler/modeler.html). Structural similarity methods included 123D+ (39) (available on the World Wide Web at 123d.ncifcrf.gov/123D+.html) or 3DPSSM (40) (available on the World Wide Web at www.sbg.bio.ic.ac.uk/~3dpssm/). Two families gave good overall results: the MuSK tyrosine kinase (Protein Data Bank accession number 1LUF_A) and the cytoplasmic domain of the type I transforming growth factor- receptor (Protein Data Bank accession number 1B6C_B). Pto was individually aligned with the two kinases using ClustalW with manual refining (available on the World Wide Web at clustalw.genome.ad.jp/). MODELLER was then used to build the three-dimensional model of Pto. PROCHECK (41) (available on the World Wide Web at www.biochem.ucl.ac.uk/~roman/procheck/procheck.html) and PROSAII (42) (available on the World Wide Web at www.came.sbg.ac.at/) were used to assess the quality of the model. Multiple cycles of refining the alignment and validation were carried out to optimize the model. The best overall final validation results were obtained with MuSK kinase. According to PROCHECK, 88.2% of the residues were in the core regions of the Ramachandran plot, none were in the disallowed regions, and the overall G-factor was –0.08. In PROSAII, the energy profile showed no significant areas with positive energy, except in the regions where the template also had positive energy, and the combined Z-score was –6.71.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
DNA Shuffling, Library Screening, and Phenotypes—A library of shuffled molecules was generated using Pto and its four paralogs from the resistant haplotype, LpimFen, LpimPth2, LpimPth3, LpimPth4 (hereafter designated Fen, Pth2, Pth3, Pth4 for brevity) as templates. With the exception of Pto, these proteins did not interact with AvrPto in the yeast two-hybrid system (18). The library was screened for interaction with AvrPto in yeast. Approximately 40,000 clones were screened, and 150 that were positive for -galactosidase activity were selected. All of these clones were selected for subsequent analyses. Additionally, 100 negative colonies were randomly chosen for subsequent analyses. The 250 shuffled molecules selected from the primary screen were isolated from yeast and retested individually by yeast mating for their ability to interact with AvrPto. This provided 148 interacting and 92 noninteracting clones. These were subcloned into pDONR207 (Invitrogen), and their sequence was determined. Sequence analysis of the 240 shuffled clones showed that they were all chimeras generated by shuffling; 70 (29%) were duplicated sequences, indicating that we had sampled the majority of the variation present in the library. 96% of the 170 nonredundant sequences encoded full-length proteins with the correct subdomain order of functional serine/threonine kinases. All of the interacting molecules belong to this class. The remaining 4% (six clones) were predicted to encode proteins truncated at various points. On average, there was 1.09 mutations in each clone predicted to be generated by PCR. None of the nonredundant, full-length clones had the same sequence as any of the templates used.

Recombination events were distributed throughout the molecule (Fig. 1). DNA shuffling produced diverse, predominantly full-length, chimeras useful for functional analysis. With the exception of Pth2, each of the original templates were represented in different regions of the shuffled molecules. Pth2 is not the most divergent member of the templates used in this study and was expected to recombine well in the mixture. The lower representation of Pth2 in the chimeras may have been due to the inadvertent use of smaller amount of DNA of Pth2 compared with the other templates. The distribution of crossovers in the nucleotide sequence was manually determined for a random sample of 50 nonredundant chimeras. These chimeras contained 9–18 fragments of the parental templates. There was an average of 12.2 detectable crossovers per molecule. There were therefore a total of 2,000 crossovers in the nonredundant data set (164 molecules) distributed over the 321 amino acids of Pto.

Analysis of Candidate Regions Required for Interaction with AvrPto and AvrPtoB in Yeast—The 164 nonredundant, full-length clones were tested individually for autoactivation as well as their interaction with AvrPto and AvrPtoB. They were also resequenced to confirm correspondence of the yeast clones with pDONR-derived sequences. The complete nonredundant data set is presented in Supplemental Table S2. 26 sequences exhibited autoactivation. In addition, in 23 cases, there was a lack of correspondence between the pAS2.1GB- and pDONR207-derived sequences. We ascribed this to the initial yeast colonies harboring multiple plasmids. This resulted in 115 nonredundant, full-length, nonautoactivating, sequence-validated clones available for further analysis.

Only 56 nonredundant clones from the entire data set were confirmed as interacting with AvrPto. This fraction of positive colonies from the original 40,000 clones screened suggested that four or five independently recombining variable regions of the Pto molecule were necessary for the interaction with AvrPto ([frequency of Pto in the template]^4.5 x total number of clones screened x frequency of nonredundant clones = [0.25]^4.5 x 40,000 x 0.7 = 55). The mutation rate was not included in this calculation, because it was approximately the same for interactors and noninteractors (data not shown). Therefore, the frequency of interacting clones was consistent with the number of peaks subsequently detected by ² analysis.

² analysis was carried out independently for each of the 179 polymorphic residues to detect correlations between the variation at each residue and the ability to interact with AvrPto (Fig. 2) (see "Materials and Methods"). The plot for interaction with AvrPtoB is almost identical (data not shown). Seven regions exceeded a ² value threshold of 3.84 for a single test and were therefore potentially important for the ability to interact with AvrPto and AvrPtoB. Peaks 1 and 2 are located in the predicted kinase small lobe that spans the N terminus through subdomain IV. Peak 3 is located in the cleft that links the small and large lobe (kinase subdomain V), and peaks 4–7 are located in the large lobe, at kinase subdomains VIII, IX, and XI. Amino acids Asn¹²¹ and Thr²⁰⁴, which were identified in this shuffling experiment (Fig. 2, peaks 3 and 4, respectively), had been previously identified as important for AvrPto binding (10, 23). These positions therefore served as positive controls and confirmed the validity of our screening system and the sensitivity of our analysis.

View larger version (56K): [in this window] [in a new window]   FIG. 1. Pattern of recombination in chimeras resulting from DNA shuffling of Pto and four of its paralogs. Top panel, diagrammatic representation of 164 nonredundant sequences (shown in rows) with the N terminus to the left of the panel and the C terminus to the right. Each chimera was divided into segments of 20 amino acids, and each segment was compared with the templates and colored to indicate its origin. Bottom panel, the genotypic frequencies calculated for each segment in the top panel, showing an even distribution of recombination across the molecule. Indistinguishable, segments that have the same sequence in more than one template, and therefore their origin cannot be determined; Non-Pto, segments that are indistinguishable between individual templates but did not originate from Pto. An interactive version that allows variable window sizes is available on the World Wide Web at michelmorelab.ucdavis.edu/shuffledb/data.

View larger version (23K): [in this window] [in a new window]   FIG. 2. Candidate regions important for binding to AvrPto. Diagrammatic representation of 2 values identifying correlations between variation at individual amino acids and the ability of 115 chimeras to bind AvrPto in the yeast two-hybrid system at 28 °C (see "Materials and Methods" for analysis procedures). A 2 value of 3.84 is the 95% significance threshold for a single independent test with one degree of freedom, indicated by the lower red line. The experiment-wide significance threshold of 9.24 was determined using permutation analysis, indicated by the upper red line. The significant peaks are numbered according to the nomenclature used under "Results." The red asterisks show the peaks containing residues that were previously demonstrated as significant for Pto function (10, 11, 17). The green arrow indicates the location of the catalytic aspartate (Asp164 in Pto).

Validation by Site-directed Mutagenesis—In order to test the inferences from the statistical analyses, site-directed mutagenesis was carried out in Pto for each of the seven regions that were correlated with AvrPto binding with ² values above 3.84, except for the region corresponding to Thr²⁰⁴. As detailed below, the alteration of at least one residue in most of these regions resulted in a change in phenotype in the yeast two-hybrid system (Fig. 3). All site-directed mutants were tested with AvrPto, AvrPtoB, and three AvrPtoB homologs that interact with wild type Pto (VirPphA homologs from Pseudomonas savastanoi pvs. phaseolicola, glycinea, and savastanoi) (29, 30). To test whether the inability to interact with AvrPto in the yeast two-hybrid system was also seen in planta, we carried out Agrobacterium-mediated transient expression of these molecules in N. benthamiana plants expressing AvrPto under the glucocorticoid-inducible promoter (18, 34). All of the site-directed mutant molecules that were unable to interact with AvrPto in the yeast two-hybrid system were also unable to elicit an AvrPto-dependent HR in N. benthamiana (Fig. 3). Therefore, the lack of interaction in yeast was reflected in a lack of recognition in planta.

Peak 1 (maximum ² = 5.11) was located N-terminal to the kinase subdomain I of Pto. The peak exceeded the significance threshold for single comparisons but was well below that for multiple comparisons. The amino acid residues with highest value in this peak were Asp⁴² and Phe⁴⁵. Pto molecules with substitutions D42N and F45S interacted with AvrPto and AvrPtoB, resulting in -galactosidase activity similar to that of wild type Pto. Therefore, these amino acids were not solely critical for binding of Pto to these avirulence proteins. However, other sites in this region with lower ² values that were still above the threshold for single comparisons were not investigated, and they could be important for interaction with the avirulence proteins.

Peak 2 (maximum ² = 4.56) was located between predicted kinase subdomains II and III of Pto. Four polymorphic residues in this region, Pro⁷³, Glu⁷⁴, Ser⁷⁶, and Gly⁷⁸ exceeded the single comparison significance threshold for AvrPto binding; however, they were well below the threshold for multiple comparisons. Each of these residues was individually changed to amino acids present in the other paralogs. With the exception of PtoG78S, all of the Pto site-directed mutants in this region (Pto P73L, P73C, E74D, S76Q, S76A, G78S, and Sw71–76, a substitution of amino acids 71–76 with the corresponding amino acids from Pth3) had unaltered phenotypes in yeast (Fig. 4). Pto G78S was affected in its ability to bind AvrPtoB but not AvrPto or any of the VirPphA homologs tested. The VirPphA proteins are 50–52% identical to AvrPtoB (12, 29, 30). Conversely, any of the mutations introduced in this region in Pth3, a noninteracting paralog, conferred to this molecule the ability to interact with AvrPtoB and its homologs, but not with AvrPto (Pth3 Q76S, S78G, Sw71–76, and Sw73–78) (Fig. 4). We then tested if these phenotypes reflected the interactions in planta by transient co-expression of AvrPtoB and the Pth3 mutants in tomato. When co-expressed with AvrPtoB in tomato, Pth3 Q76S did not elicit a detectable hypersensitive response (data not shown). However, co-expression of AvrPtoB and Pth3 Sw71–76, containing these amino acids from Pto, resulted in confluent HR in 11 of 25 cases, and nonconfluent HR was obtained in 12 of 25 cases tested (Fig. 5). Taken together, these results suggest that this region is necessary and sufficient to confer the ability to bind AvrPtoB but is not critical for binding AvrPto. It is not clear why there was a significant peak for binding AvrPto. It is possible that this region is required for binding to AvrPto in combination with a second region in the protein.

View larger version (69K): [in this window] [in a new window]   FIG. 3. Phenotypes of site-directed mutants of Pto in five of the candidate regions important for interaction with AvrPto and AvrPtoB. The first three columns show yeast two-hybrid interactions at 28 °C with AvrPto, AvrPtoB, and VirPphA, respectively. The phenotypes with VirPphAPsv and VirPphAPgy (29, 30) were the same as for VirPphA (not shown). The next three columns show the ability to elicit HR in N. benthamiana: Agrobacterium-mediated transient expression of the site-directed mutants alone, co-expression of the site-directed mutants with AvrPto using dexamethasone-inducible transgenic N. benthamiana, and expression of the site-directed mutants with an additional mutation at L205D. The L205D mutation makes the Pto protein constitutively active for downstream elicitation of an HR. The last column shows the reference to peak numbering used under "Results" and in Fig. 2. Sw280–283, substitution of amino acids 280–283 in the Pto molecule with the corresponding amino acids from Pth4; I, interaction; NI, no interaction; HR, hypersensitive response; O, no HR; WT, wild type.

View larger version (57K): [in this window] [in a new window]   FIG. 4. Phenotypes of site-directed mutants of Pto and Pth3 in the region spanning amino acids 71–78 (Fig. 2, peak 2). The first three columns show yeast two-hybrid interactions at 28 °C with AvrPto, AvrPtoB, and VirPphA, respectively. The phenotypes with VirPphAPsv and VirPphAPgy (29, 30) were the same as for VirPphA (not shown). The last three columns show the ability to elicit HR in N. benthamiana: alone, co-expression of the site-directed mutants with AvrPto using dexamethasone-inducible transgenic N. benthamiana, and expression of the site-directed mutants with an additional mutation at L205D. This mutation renders Pto constitutively active for signaling to HR. Sw71–76, substitution of amino acids 71–76 in the Pto molecule with the corresponding amino acids from Pth3 or vice versa; Sw73–78, substitution of amino acids 73–78 in the Pth3 molecule with the corresponding amino acids from Pto; I, interaction; NI, no interaction; HR, hypersensitive response; O, no HR; WT, wild type. Note that substitutions into Pth3 could not be tested in N. benthamiana, because wild type Pth3 alone elicits an HR in this species.

View larger version (83K): [in this window] [in a new window]   FIG. 5. Co-expression of AvrPtoB and Pth3 with a substitution of amino acids 71–76 from Pto in the susceptible tomato variety Rio Grande 76S. Agrobacterium-mediated transient expression of Pto and Pth3 variants with AvrPtoB expressed from the same T-DNA. Note that the molecules could only be assayed in tomato because Pth3 causes an HR in N. benthamiana. WT, wild type; Sw71–76, substitution of amino acids 71–76 in the Pth3 molecule with the corresponding amino acids from Pto.

Peak 6 (maximum ² = 11.83) spanned amino acid residues Glu²⁸⁰–Arg²⁸³ and clearly exceeded the significance threshold for multiple comparisons. These are located in subdomain XI of Pto. A substitution of amino acids 280–283 from Pth4 into Pto disrupted the ability to bind to both Avr proteins, suggesting that this region is necessary for binding or for protein stability. However, a Pto R283G mutant was recently reported to bind AvrPto to wild type levels (44). Additionally, one of the shuffled molecules with an asparagine at this position is able to bind both Avr proteins; therefore, Asn²⁸³ was at least not an absolute destabilizing factor. It is therefore possible that Lys²⁸⁰ or Tyr²⁸¹ is responsible for the phenotypes observed in the mutant molecule.

Peak 7 (maximum ² = 22.07) includes a large segment of the C-terminal amino acid residues from Val²⁹⁰ to Ile³²¹ in Pto. This was the largest of all the peaks and exceeded the significance threshold for multiple comparisons by 2.4-fold. Pth4 was the major source of sequence polymorphism that was correlated with the inability to bind AvrPto in this region. The predicted Pth4 protein is shorter than the rest of the templates, including Pto. Therefore, we generated C-terminal deletions of Pto of 5 and 10 amino acid residues. Both deletions disrupted the ability of Pto to bind AvrPto and AvrPtoB proteins (Fig. 3). In addition, residue Ala³¹³ in this region had been identified in a study of natural Pto variants as a unique residue present in a nonfunctional Pto ortholog in a wild species of tomato (45). A PtoA313D site-directed mutant was generated; this mutant was unable to bind AvrPto and AvrPtoB. However, both the Pto 5-amino acid C-terminal deletion and the A313D mutants were able to bind three VirPphA homologs (VirPphA, VirPphA_Psv and VirPphA_Pgy) (Fig. 3 and data not shown), indicating that these Pto mutants were folding correctly and were stable. Consequently, the C terminus of Pto seems to be specifically required for interaction with AvrPto and AvrPtoB proteins in the yeast two-hybrid system.

View larger version (38K): [in this window] [in a new window]   FIG. 6. Western analysis of yeast strains expressing site-directed mutant variants or Pto. The upper panel shows the Western blot analysis of protein fusions of site-directed mutant molecules of Pto using the anti-GAL4 monoclonal antibody. The bottom panel shows the corresponding polyacrylamide gel stained with Coomassie Blue. The phenotypes of interaction in yeast are shown below. +, interacts with all Avr proteins tested; –, does not interact with any of the Avr proteins tested; (+), interacts with some but not all of the Avr proteins tested. Sw71–76, substitution of amino acids 71–76 in the Pto molecule with the corresponding amino acids from Pth3; Sw280–283, substitution of amino acids 280–283 in the Pto molecule with the corresponding amino acids from Pth4; Sw243–258, substitution of amino acids 243–258 in the Pto molecule with the corresponding amino acids from Pth4.

Analysis of Regions Potentially Important for Downstream Signaling—To test for the ability to signal downstream, all 164 nonredundant shuffled molecules were expressed in N. benthamiana plants that expressed AvrPto under the glucocorticoid inducible promoter using Agrobacterium-mediated transient expression. As expected, all of the molecules that were unable to interact with AvrPto in yeast were also unable to signal downstream in planta in an AvrPto-dependent manner. However, only seven of the 56 molecules that interacted with AvrPto in yeast were able to produce an HR in an AvrPto-dependent manner. All functional molecules had a complete rather than truncated C terminus. Nine molecules induced HR in an AvrPto-independent manner and were therefore constitutively active, similar to Pto L205D, Fen, and Pth3. Eight of these interacted with AvrPto, and one was autoactive in yeast. All of the molecules that could signal downstream for an HR had the six residues identified as being important for binding AvrPto in yeast. The low number of chimeras that could signal downstream may reflect the differences in subcellular localization of the chimeras in yeast and plants or that several additional regions were necessary for downstream signaling in addition to those required for Avr binding.

A similar analysis to the one employed for interactions in yeast was carried out for the downstream signaling to an HR in planta. All 16 molecules that induced HR in an AvrPto-dependent or -independent manner were classified as capable of signaling downstream. The only peak obtained in this analysis that was validated by site-directed mutagenesis was the region between amino acids Gln²⁴³–Glu²⁵⁸ that are located in the kinase subdomain X. The sensitivity and resolution of this analysis may have been limited by the low number of molecules present in the functional class. It is therefore possible that additional regions important for signaling downstream were not detected. This peak exceeded the single comparison value but was below the experiment-wide significance threshold (² = 8.53).

Pth4 was the source of polymorphism in the Gln²⁴³–Glu²⁵⁸ peak; none of the functional molecules had these amino acids from Pth4. We compared the chemical nature of each amino acid in Pto versus Pth4 in this region and created single site-directed mutants on the two most chemically variable sites (Fig. 7). Neither of the E248G and E254A single substitutions impaired the ability to bind to AvrPto or AvrPtoB or to signal downstream in an AvrPto-dependent or independent manner. However, a substitution of amino acids 243–258 in Pto with the corresponding ones from Pth4 resulted in the loss of the ability to bind to AvrPto and AvrPtoB in yeast and to elicit HR in an AvrPto dependent or independent manner in planta. This chimeric protein was detected in yeast protein extracts using Western blots with the anti-GAL4 binding domain antibody. Therefore, there is no evidence that this chimeric protein was unstable. This suggests that this region is important either for correct protein folding or for binding to the Avr proteins and downstream components. A Pto L245D mutant was recently identified that was impaired in downstream signaling but still bound AvrPto (17). Although Leu²⁴⁵ is one of the polymorphic residues between Pto and Pth4, the change is conservative (Leu to Ile), and it is unlikely that it is responsible for the phenotype observed in the chimera. Therefore, more than one residue in this region may be critical for downstream Pto function.

Three-dimensional Distribution of Functionally Important Sites—We modeled the three-dimensional structure of Pto by threading its sequence onto the crystal structure of the kinase domain of the MuSK tyrosine kinase (see "Materials and Methods"). The positions Ser⁷⁶, Gly⁷⁸, and Asn¹²¹ that were critical for binding the avirulence proteins were located on the same side of the protein as the activation domain (Fig. 8). Glu²³³ and the Gln²⁴³–Glu²⁵⁸ region were predicted to be behind and below the activation domain, respectively. However, amino acids Glu²⁸⁰–Arg²⁸³ and the major part of the C terminus that were also identified as critical for interaction with the avirulence proteins form -helices that fold in the opposite face from the activation domain. Therefore, the C terminus and Glu²⁸⁰–Arg²⁸³ may not be important for direct binding of the Avr proteins but rather for the correct three-dimensional configuration of the protein required for function.

View larger version (55K): [in this window] [in a new window]   FIG. 7. Phenotypes of a domain swap of Pto in the region around subdomain X. The first two columns show yeast two-hybrid interactions with AvrPto and AvrPtoB, respectively. The last three columns show the ability to elicit HR in N. benthamiana: Agrobacterium-mediated transient expression of the site-directed mutants alone, co-expression of the site-directed mutants with AvrPto using dexamethasone-inducible transgenic N. benthamiana, and expression of the site-directed mutants with an additional mutation at L205D. I, interaction; NI, no interaction; HR, hypersensitive response; O, no HR; WT, wild type; Sw243–258, substitution of amino acids 243–258 in the Pto molecule with the corresponding amino acids from Pth4.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
We used DNA shuffling as an in vitro genetic approach to identify seven regions important for the function of Pto. Previous studies using domain swaps between Pto and Fen, as well as site-directed mutagenesis, identified a single amino acid responsible for AvrPto binding (10, 11, 23). However, the decreased strength of the interaction of the chimeric Pto-Fen molecules in the yeast two-hybrid system indicated that additional regions of the protein are required for this recognition. Also, only a limited number of residues differed between Pto and Fen and were therefore informative. Our identification of additional regions involved in interaction with AvrPto and/or AvrPtoB confirmed the premise that multiple regions are involved in ligand recognition. This was facilitated by the use of DNA shuffling as a combinatorial approach with diverse templates. DNA shuffling detected two regions that were also recently identified by site-directed mutagenesis (17). In addition, we discovered three previously unidentified regions that are responsible for the ability of Pto to recognize AvrPto and AvrPtoB. Of these, Ser⁷⁶ and Gly⁷⁸ are critical for recognition of AvrPtoB but not AvrPto, and the C terminus as well as the region between Glu²⁸⁰ and Arg²⁸³ are required for both recognition and signaling downstream.

View larger version (97K): [in this window] [in a new window]   FIG. 8. The predicted three-dimensional structure of Pto showing the location of the seven regions detected as being functionally important. The three-dimensional model was obtained by threading the sequence of Pto onto the muscle-specific kinase. The five regions important for binding both AvrPto and AvrPtoB are shown in red. The region that was critical for interacting with AvrPtoB but not AvrPto is blue. The region in subdomain X that was required for interaction with the avirulence proteins and downstream signaling is colored yellow. For reference, the catalytic aspartate 164 is shown in green. The activation domain is indicated by a white arrow. The blue arrow indicates Leu215, a residue identified by Wu et al. (17) as critical for recognition of AvrPtoB but not AvrPto.

As with classical genetic studies, the number of recombination events analyzed determines the resolution at which loci that are important for a specific characteristic can be identified (46). The number of recombination events is determined by the frequency of recombination, the progeny size, and the number of rounds of mating. In our analysis, resolution to a single amino acid residue was obtained in some cases (e.g. Glu²³³). 12 amino acids upstream and 5 amino acids downstream of Glu²³³ are nonvariable between the templates. However, linkage drag was observed in several regions for amino acids that were close (linked) to amino acids important for the phenotype. For example, linkage to Asn¹²¹ elevated Arg¹⁰⁷ through Lys¹¹⁵ above the significance threshold for single comparisons; however, mutagenesis indicated that only Asn¹²¹ was critical for interaction (10) (this paper). To increase resolution, more recombination events would have to be studied. This could be achieved by a second generation of shuffling using sequences selected from this experiment or by selecting smaller fragments prior to PCR.

View larger version (46K): [in this window] [in a new window]   FIG. 9. Amino acid positions that tolerate variation in Pto. The three-dimensional structure was modeled as for Fig. 8. A, predicted three-dimensional location of amino acids that are polymorphic in chimeras that can interact with AvrPto (green). B, three-dimensional location of amino acids that are polymorphic in fully functional chimeras (green) and amino acids that are polymorphic both in fully functional chimeras generated in vitro and in natural orthologs of Pto (yellow) (45).

Nine chimeras showed a constitutive gain-of-function phenotype; however, eight of these were still able to interact with AvrPto in the yeast two-hybrid system, in contrast with previously identified constitutively active variants of Pto, which were not able to interact with AvrPto (15). The kinase activity of Pto is not necessary for activation of downstream signaling, and the current model for elicitation of downstream signaling is a conformational change in Pto induced by interaction with AvrPto (17). The existence of multiple chimeras, which both bind AvrPto and are constitutively active, indicates that elicitation of downstream signaling does not involve a conformational change that precludes binding of AvrPto, as previously hypothesized. A single constitutively active mutant (I214D) that was able to interact with AvrPto (17) is also consistent with our data.

Some of the sites that we identified may be important for basic kinase structure and function. The predicted Pto structure resembles a typical kinase with a small and a large lobe (Fig. 8) (48). The small lobe is composed of subdomains I–IV and has been shown to anchor the ATP molecule in other kinases. The large lobe spans subdomains VI–XI and functions in recognition of the substrate and catalysis. There are 12 invariant residues that are conserved in all kinases and have been identified as critical for kinase function (48). The majority of these were conserved in all of the templates used in this study; only Gly⁴⁸ and Arg²⁸³ were polymorphic among the templates. Not all functional molecules had a glycine at position 48, suggesting that this residue is not critical for interaction with the avirulence proteins or for signaling downstream. Arg²⁸³ was identified as highly significant in our analysis. However, a single chimera present in our analysis and a point mutant (44) were both able to bind AvrPto. Therefore, the two conserved residues Gly⁴⁸ and Arg²⁸³ are not absolutely required for interaction with AvrPto.

Previous molecular and biochemical studies have identified other residues in Pto that are important for interaction with AvrPto and AvrPtoB and downstream signaling. Biochemical studies to identify phosphorylation sites in the Pto kinase identified Thr³⁸ as important for binding AvrPto and downstream signaling; Ser¹⁹⁸ was shown to be important for downstream signaling but not for binding AvrPto (22). Ethyl methane sulfonate mutagenesis identified Val⁵⁵ and His⁹⁴ as important for resistance; mutations in these residues disrupted Pto-mediated resistance and interaction in yeast with AvrPto (10, 21). None of these four residues were polymorphic among the templates involved in this analysis and were consequently not detected. Our templates were naturally occurring genes, and therefore variation at sites critical for basic kinase function may be less likely than for other sites.

The majority of the sites that we identified are conserved in 53 naturally occurring alleles and orthologs of Pto from six Lycopersicon spp (45); however, they are polymorphic relative to other kinases that have different ligands and substrates (data not shown). With the exception of Ser⁷⁶, all the sites identified as important for recognition of AvrPto are conserved in LescPth5, a paralog that confers minor recognition of AvrPto in the susceptible haplotype (18). A substitution in the C terminus, A313E, in a single ortholog was the only change correlated with a loss of recognition of AvrPto (45). Site-directed mutagenesis of this residue as part of this study confirmed the importance of this residue in AvrPto binding.

A patch for negative regulation of Pto was recently identified using a combination of three-dimensional structure prediction and site-directed mutagenesis (17). Additionally, a surface area that overlaps with this patch was partially delimited as being required for interaction with AvrPto and AvrPtoB. Of the 17 residues in this surface area, seven were not polymorphic between our templates used in our shuffling experiments and therefore have been conserved in the Pto paralogs. Ten were polymorphic between our templates; eight of these are located in three of the regions detected in our study. Phe²¹³ is included in the activation loop (peak 4). Glu²³³ was identified in both studies (peak 5). Residues Leu²⁴⁵, Asn²⁵¹, Ala²⁵³, Glu²⁵⁴, and Glu²⁵⁸ are included in the Gln²⁴³–Glu²⁵⁸ region identified in the downstream signaling analysis. Our shuffling approach identified three regions (the region around Ser⁷⁶, Glu²⁸⁰–Arg²⁸³, and the C terminus) that had not been previously identified by directed mutagenesis approaches. The importance of these regions could not be inferred from a priori data. This underscores the utility of a nondirected, combinatorial approach as complementary to site-directed approaches.

The low number of chimeras that were functional in planta precluded a comprehensive analysis of regions important for signaling downstream. Kinase subdomain X was identified as important for signaling downstream in an AvrPto-dependent and -independent manner. Subdomain X is highly variable among kinases and seems to be more conserved in subfamilies that share similar functions. Interestingly, this region has been shown to be important for the signaling in the c-Jun N-terminal kinase/stress-activated protein kinase cascade leading to apoptosis in response to stress in mammalian cells (49). A swap of the Pth4 sequence in this region of Pto disrupted all phenotypes but was stable in yeast, suggesting the importance of this region for all Pto functions tested in this study or for correct protein folding. The C terminus of Pto was also important for signaling downstream. The A313D mutant and the 5-amino acid C-terminal deletion of Pto were capable of binding VirP-phA homologs in yeast and therefore were presumably folded correctly; however, the A313D substitution to the constitutively active Pto L205D was incapable of signaling in planta. Therefore, this region is required for both recognition of AvrPto and AvrPtoB as well as signaling downstream. Avr proteins and downstream signaling components may interact with the C terminus directly, or this region might have an indirect conformational effect. The separation of the C terminus and the activation domain in the three-dimensional model indicates that the latter is more probable.

A high proportion of the Pto protein can tolerate variation without affecting recognition of AvrPto and downstream signaling. More than half of the residues that were variable among all chimeras, representing 28% of the total protein, were polymorphic among the seven fully functional chimeras. Sites that are variable both among the orthologous Pto sequences (45) and among the paralogous sequences (our templates) may not be selectively constrained. Comparisons between eleven functional orthologs identified 30 sites and one indel that were polymorphic (45). 90 sites were polymorphic among our functional chimeras. 23 sites were polymorphic in both the functional chimeras and orthologs. None of the sites that were identified as being important for function by DNA shuffling were variable among the natural functional orthologs. In both cases, these variable sites are distributed along the length of the molecule (Fig. 9). A ² test was used to determine if there was significant overlap in the number of residues that were variable between the functional orthologs and the chimeras. The null hypothesis was rejected, indicating that there was significant overlap among residues identified in these two data sets. This degree of significant overlap (p < 0.001) strongly supports the validity of the shuffling approach to efficiently distinguish important from unimportant sites in Pto.

Future experiments will analyze additional molecules to dissect the regions important for downstream signaling. This is being pursued by an additional round of shuffling using the seven fully functional chimeras identified in this experiment with seven chimeras that are capable of recognizing AvrPto in yeast but not of signaling downstream. It would also be interesting to use DNA shuffling for in vitro evolution of Pto. It may be possible to select chimeras that interact with different virulence effector proteins. If such interaction results in activation of Pto for downstream signaling, it might be possible to increase the spectrum of resistance available in tomato and other solanaceous species. In addition, shuffling Pto from tomato with orthologs from nonsolanaceous species may provide molecules that confer the ability on new species to recognize and respond to AvrPto, AvrPtoB, and homologs of VirPphA as well as lead to an understanding of the domains involved in the restricted taxonomic functionality of this resistance protein.

	FOOTNOTES

 
^* This work was supported by National Science Foundation (NSF) Cooperative Agreement BIR-8920216 (to Center for Engineering Plants for Resistance against Pathogens) and the NSF Integrative Plant Biology Program award 0133993 and a Fulbright-Colciencias-IIE scholar