Identification of Ligand Binding Site of Phytosulfokine Receptor by On-column Photoaffinity Labeling*

Phytosulfokine (PSK), an endogenous 5-amino-acid-secreted peptide in plants, affects cellular potential for growth via binding to PSKR1, a member of the leucine-rich repeat receptor kinase (LRR-RK) family. PSK interacts with PSKR1 in a highly specific manner with a nanomolar dissociation constant. However, it is not known which residues in the PSKR1 extracellular domain constitute the ligand binding pocket. Here, we have identified the PSK binding domain of carrot PSKR1 (DcPSKR1) by photoaffinity labeling. We cross-linked the photoactivatable PSK analog [125I]-[Nϵ-(4-azidosalicyl)Lys5]PSK with DcPSKR1 using UV irradiation and mapped the cross-linked region using chemical and enzymatic fragmentation. We also established a novel “on-column photoaffinity labeling” methodology that allows repeated incorporation of the photoaffinity label to increase the efficiency of the photoaffinity cross-linking reactions. We purified a labeled DcPSKR1 tryptic fragment using anti-PSK antibodies and identified a peptide fragment that corresponds to the 15-amino-acid Glu503-Lys517 region of DcPSKR1 by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. Deletion of Glu503-Lys517 completely abolishes the ligand binding activity of DcPSKR1. This region is in the island domain flanked by extracellular LRRs, indicating that this domain forms a ligand binding pocket that directly interacts with PSK.

that has been identified in the medium of plant cell cultures, based on the results of assays of the growth-promoting activity of cultured cells (1). The addition of chemically synthesized PSK to culture medium, even at nanomolar concentrations, significantly promotes the proliferation of callus and suspension cells. PSK also promotes tracheary element differentiation (2), somatic embryogenesis (3,4), adventitious bud formation (5), adventitious root formation (6), and pollen germination in vitro (7). PSK is produced from Ϸ80-amino-acid precursor peptides via post-translational sulfation of tyrosine residues and proteolytic processing (8). Genes encoding PSK precursors are redundantly distributed in the genome and are expressed in cultured cells and in a variety of tissues, including leaves, stems, flowers, and roots (9,10).
PSK binds the membrane-localized PSK receptor PSKR1, which is a leucine-rich repeat receptor kinase (LRR-RK) that has been purified from solubilized carrot microsomes by ligand-based affinity chromatography (hereafter referred to as DcPSKR1) (11). The extracellular domain of DcPSKR1 contains 21 tandem copies of LRR interrupted by a 36-amino-acid island domain rich in hydrophilic and charged amino acid residues. Disruption or overexpression of the Arabidopsis ortholog of PSKR1 (AtPSKR1) significantly alters cellular longevity and potential for growth without interfering with primary morphogenesis of plants (10). PSK appears to activate the basic potential for cellular growth rather than directly determining cell fate and thereby exerts a pleiotropic effect on individual cells in response to environmental hormonal conditions.
Ligand binding generally causes a receptor protein to undergo a conformational change that directly activates the receptor so that it can interact with another cellular molecule and/or exert intrinsic enzyme activities such as kinase activity. PSK interacts with DcPSKR1 in a highly specific manner, with a high affinity dissociation constant of K d ϭ 4.2 nM (11). However, it is not known which amino acids in the DcPSKR1 extracellular domain constitute the ligand binding pocket.
Photoaffinity labeling is one of the most useful methods for analyzing ligand-receptor interactions. Identification of the labeled amino acid residues can yield valuable information about the ligand binding domain of the receptor. However, it is often quite difficult to identify the cross-linked residues by MS analysis due to the low efficiency of cross-linking reactions. These difficulties are further compounded by low concentrations of binding proteins, especially in the case of transmembrane receptors and channels that cannot be functionally overexpressed in bacteria. In fact, most known cross-linked regions of membrane-localized proteins have been identified by analyzing chemical and enzymatic fragmentation patterns of labeled proteins using radioactive photoaffinity ligands rather than direct MS analysis of purified peptide fragments cross-linked with photoaffinity ligands (12)(13)(14).
In the present paper, we have reported identification of the PSK binding site of DcPSKR1 by SDS-PAGE mapping of the cross-linked fragments generated by chemical and enzymatic fragmentation of the photoaffinity-labeled ligand-receptor complex and by direct analysis of the purified cross-linked fragments by matrix-assisted laser desorption ionization time-offlight mass spectrometry (MALDI-TOF MS). We have also reported the usefulness of a novel solid-phase photoaffinity labeling technology, "on-column photoaffinity labeling," which allows repeated incorporation of a photoaffinity label to increase the efficiency of the photoaffinity cross-linking reactions.
Ligand Binding Assay-The ligand binding assay of plant microsomal fractions and affinity-purified DcPSKR1 was performed using [ 3 H]PSK and following a protocol described previously (11,17).
Affinity Purification of DcPSKR1-⌬KD-His 6 -Transformed BY-2 microsomal membranes (1,800 mg of protein prepared from 6-day-old culture) were solubilized in 320 ml of buffer containing 20 mM HEPES-KOH (pH 7.5), 50 mM KCl, and 1.0% Triton X-100. Solubilized materials were centrifuged at 100,000 ϫ g for 30 min at 4°C, and the supernatants were applied to a [Lys 5 ]PSK-Sepharose column (5.0-ml column) at a flow rate of 0.5 ml/min using the AKTA prime chromatography system (Amersham Biosciences) (11). After washing with 50 ml of buffer containing 20 mM HEPES-KOH (pH 7.5), 50 mM KCl, and 0.1% Triton X-100 (wash buffer), the column was eluted with 15 ml of buffer containing 20 mM HEPES-KOH (pH 7.5), 500 mM KCl, and 0.1% Triton X-100 (elution buffer). The eluates were added to a 1.0-ml Macro-Prep ceramic hydroxyapatite Type I column (Bio-Rad) at a flow rate of 0.5 ml/min at 4°C. The column was washed with 20 ml of wash buffer and eluted with an 18-ml gradient of KH 2 PO 4 (0 -400 mM) in wash buffer. Aliquots of active fractions (1.0 ml), as determined by [ 3 H]PSK binding assay, were concentrated by ultrafiltration (Ultrafree-MC MWCO 30,000; Millipore) and analyzed by SDS-PAGE using 7.5% gels for confirmation of purity. Protein bands in SDS-polyacrylamide gels were visualized using the fluorescent dye SYPRO Red (Molecular Probes). The immunoblot of the active fraction (20 l) was probed with anti-DcPSKR1 antibodies raised against the DcPSKR1 100-amino-acid N-terminal region (11) and was visualized using ECL Advance (Amersham Biosciences) according to the manufacturer's protocol.
PNGase F Treatment of Purified DcPSKR1-⌬KD-His 6 -Affinity-purified DcPSKR1-⌬KD-His 6 (10 l) was added to PNGase F buffer containing 100 mM Tris-HCl buffer (pH 8.6) and 0.1% SDS and was incubated at 95°C for 3 min. The denatured sample was then incubated with 2 milliunits of peptide N-glycosidase F (PNGase F; TakaraBio, Shiga, Japan) in the presence of Nonidet P-40 (1.0% final solution) at 37°C for 16 h at a total volume of 25 l. After deglycosylation, samples were mixed with SDS-PAGE sample buffer and were then analyzed by SDS-PAGE and immunoblotting, as described above.
In-solution Photoaffinity Labeling-Aliquots of purified DcPSKR1-⌬KD-His 6 (10 l) were incubated with 1 M [ 125 I]ASA-PSK for 10 min at 4°C. Thereafter, the samples were irradiated with a UV lamp (model ENF-260C/J, 365 nm; Spectronics Co. Ltd., NY) for 10 min on ice at a distance of Ͻ1 cm. SDS-PAGE sample buffer was added to each of the samples, which were then heated at 95°C for 5 min. The samples were loaded onto NuPage 12% BisTris gel (Invitrogen) and separated according to the manufacturer's protocol. The dried gels were exposed to a bio-imaging plate (Fujifilm, Tokyo, Japan) for 16 h at room temperature and were analyzed using a bio-imaging analyzer (BAS 2000, Fujifilm).
Chemical Cleavage and Protease Digestion-[ 125 I]ASA-PSKlabeled DcPSKR1-⌬KD-His 6 , with or without PNGase F treatment, was used for chemical cleavage and protease digestion. Cyanogen bromide (CNBr) cleavage of cross-linked proteins (100 l) was performed in 75% formic acid containing 0.5 mg/ml CNBr under a nitrogen atmosphere. After this solution was incubated for 16 h at room temperature, it was diluted 10-fold with distilled water and lyophilized. For trypsin digestion, 10 l of labeled DcPSKR1-⌬KD-His 6 was treated with 100 pmol of TPCK-treated trypsin (Sigma) in buffer containing 20 mM Tris-HCl (pH 7.6), 10 mM CaCl 2 and 10% CH 3 CN (v/v) at 37°C for 16 h at a total volume of 20 l. For Asp-N digestion, 10 l of labeled DcPSKR1-⌬KD-His 6 was treated with 100 pmol of endoproteinase Asp-N (TakaraBio, Shiga, Japan) in 50 mM sodium phosphate buffer (pH 8.0) at 37°C for 16 h at a total volume of 20 l. All of these samples were analyzed by SDS-PAGE using the NuPage 12% BisTris gel and autoradiography.
On-column Photoaffinity Labeling and Trypsin Digestion-Affinity-purified DcPSKR1-⌬KD-His 6 (9 ml) was immobilized on Ni 2ϩ -loaded HiTrap chelating HP-Sepharose (bed volume 200 l; Amersham Biosciences) by recycling three times. After the column was washed with 1 ml of the above-described wash buffer, 1 ml of 10 nM [ 125 I]ASA-PSK or 1 M cold ASA-PSK (dissolved in the wash buffer) was loaded onto the column, which was then left to stand for 5 min for ligand binding. After the column was washed with 600 l of the wash buffer to remove unbound ligand, the bound ligand was cross-linked to DcPSKR1-⌬KD-His 6 by directly irradiating the column with UV light (365 nm) for 10 min. Uncross-linked ligand was removed by washing the column with 1 ml of the wash buffer, and the labeling cycle was then repeated up to 3 times for [ 125 I]ASA-PSK and 10 times for ASA-PSK. For trypsin digestion, labeled DcPSKR1-⌬KD-His 6 on Sepharose was suspended in 200 l of digestion buffer containing 20 mM Tris-HCl (pH 7.6), 10 mM CaCl 2 , and 10% CH 3 CN in a total volume and was then digested by 100 pmol of TPCK-treated trypsin (Sigma) at 37°C for 16 h. Sepharose was then removed by filtration, and liberated peptides were used for the following analysis.
Gel Filtration of the Labeled Tryptic Peptides-Gel filtration analysis was performed using Sephadex G-25 Superfine (1.5 cm inner diameter ϫ 35 cm; Amersham Biosciences) at a flow rate of 0.2 ml/min. Column equilibration and chromatography were performed using buffer containing 20 mM HEPES-KOH (pH 7.0) and 150 mM NaCl. On-column tryptic digests of [ 125 I]ASA-PSK-labeled DcPSKR1-⌬KD-His 6 (200,000 counts/min eq.) were applied to the column, and 1.0-ml fractions were analyzed for radioactivity using an autowell ␥ system (ALOKA). Blue dextran (Amersham Biosciences) and cyanocobalamin were used for molecular mass calibration.
Immunoaffinity Purification of the Labeled Tryptic Peptides-For the purification of the anti-PSK antibodies, anti-PSK antiserum (5 ml) (19) was loaded onto a [Lys 5 ]PSK-Sepharose column (1.0-ml column). After the antiserum was recycled three times, the column was washed with phosphate-buffered saline, and the bound antibodies were then eluted with 0.1% formic acid (pH 2.6) and immediately neutralized with NaHCO 3 . Purified antibodies were coupled with HiTrap NHSactivated HP-Sepharose (Amersham Biosciences) for 16 h at 4°C. The immobilized anti-PSK antibodies were stored in phosphate-buffered saline until use. On-column tryptic digests of ASA-PSK-labeled DcPSKR1-⌬KD-His 6 were incubated at 95°C for 10 min to inactivate trypsin and were then loaded onto the immunoaffinity column (20-l column). After the column was washed with 200 l of the trypsin digestion buffer and 200 l of distilled water, bound fragments were eluted with 1% trifluoroacetic acid (v/v). The eluate was used for MALDI-TOF MS analysis.
MALDI-TOF MS Analysis-The eluate from the immunoaffinity column was concentrated by evaporation, desalted using Zip Tip C18 TM pipette tips (Millipore), and mass-profiled with a 4700 proteomics analyzer (Applied Biosystems) using ␣-cyano-4-hydroxycinnamic acid as the matrix. A pulsed nitrogen laser (337 nm) was used to induce desorption/ionization, and mass spectra were obtained using the reflector mode. Each representative mass spectrum shown in the present figures was the smoothed average of 100,000 laser shots.

Overexpression of DcPSKR1 in Tobacco BY-2 Cells-To
obtain the relatively large amount of DcPSKR1 protein required for the direct identification of the PSK binding domain by photoaffinity labeling, we first overexpressed DcPSKR1 in BY-2 suspension cells, which grow rapidly and can be harvested weekly. For DcPSKR1 expression, BY-2 cells were transformed using two constructs, 35S::DcPSKR1-His 6 , as a positive control for functional expression of full-length DcPSKR1 in heterologous cells, and the 35S::DcPSKR1-⌬KD-His 6 construct, in which the coding region for the kinase domain of DcPSKR1 had been removed (Fig. 1A). The latter construct was used to test whether the extracellular domain of DcPSKR1 is sufficient for interaction with PSK. The His 6 tag was introduced to immobilize these proteins on the nickel column to perform on-column photoaffinity labeling (described below).
Northern blot and immunoblot analysis of the membrane fractions of each transformed BY-2 cell clone revealed that both DcPSKR1-His 6 and DcPSKR1-⌬KD-His 6 proteins were successfully overexpressed in BY-2 cells (Fig. 1B). Both proteins migrated, with apparent molecular sizes of 150 and 120 kDa, respectively. The theoretical molecular masses of these proteins, without the signal sequence, are 110.2 and 78.8 kDa, respectively, indicating that both proteins are post-translationally modified by the addition of an ϳ40-kDa moiety, most likely by glycosylation.
The ligand binding assay using [ 3 H]PSK confirmed a significant increase in PSK binding activity in the membrane fractions derived from both transformants, compared with fractions derived from untransformed BY-2 cells (Fig. 1C). These results indicate that DcPSKR1 is functionally expressed in BY-2 cells and that the intracellular kinase domain of DcPSKR1 is not essential for PSK binding. Scatchard analysis of the binding of [ 3 H]PSK to membranes expressing DcPSKR1-⌬KD-His 6 showed that the dissociation constant (K d ) of DcPSKR1-⌬KD-His 6 is 2.1 Ϯ 0.3 nM, which is comparable with that of wild-type DcPSKR1 (4.2 Ϯ 0.4 nM) (Fig. 1D) (11). Because this truncated DcPSKR1 lacks the kinase domain, and therefore theoretically yields fewer peptide fragments after chemical and/or enzymatic digestion, we used DcPSKR1-⌬KD-His 6 in the following experiments.
Purification and Photoaffinity Labeling of DcPSKR1-⌬KD-His 6 -We purified DcPSKR1-⌬KD-His 6 protein from solubilized microsomal fractions derived from transgenic BY-2 cells using [Lys 5 ]PSK-Sepharose and hydroxyapatite column chromatography. The relative expression level of DcPSKR1-⌬KD-His 6 protein was estimated to be 190 fmol/mg microsomal proteins based on Scatchard analysis. SDS-PAGE analysis of the purified fractions confirmed that DcPSKR1-⌬KD-His 6 protein had been recovered at relatively high purity (Fig. 1E, left panel). After the PNGase F treatment, the apparent molecular mass of DcPSKR1-⌬KD-His 6 changed from 120 to ϳ110 kDa, indicating that at least 10 kDa of the presumed 40-kDa post-translationally added moiety is composed of N-linked glycans (Fig. 1E,  right panel). It has been reported that ␣1,3-fucosylated glycans, which are often found in N-linked glycans in plants, are resistant to PNGase F treatment (20). Incubation of the purified DcPSKR1-⌬KD-His 6 with 10 nM [ 125 I]ASA-PSK (photoaffinity ligand) ( Fig. 2A), followed by cross-linking by UV irradiation, resulted in specific labeling of the 120-kDa band that corresponds to DcPSKR1-⌬KD-His 6 . The incorporation of [ 125 I]ASA-PSK into the 120-kDa band was completely abolished by the presence of excess unlabeled PSK, indicating specific cross-linking between [ 125 I]ASA-PSK and DcPSKR1-⌬KD-His 6 (Fig. 2B). Treatment of labeled DcPSKR1-⌬KD-His 6 with CNBr, which cleaves polypeptides at the N-terminal of Met residues (21), and subsequent analysis by SDS-PAGE revealed a labeled 45-kDa polypeptide (Fig. 2B). After the PNGase F treatment, the apparent molecular mass of this band changed from 45 to ϳ35 kDa, indicating that a 10-kDa N-linked glycan moiety is attached to this fragment. After tryptic digestion of this band, MALDI-TOF MS analysis showed a signal at m/z 1752.8, which corresponds to Glu 503 -Lys 517 , indicating that this fragment contains the island domain (data not shown). Because the presence of the Met-Thr sequence causes conversion of Met to homoserine without cleavage, we speculated that the 45-kDa polypeptide (35 kDa after PNGase treatment) corresponds to the Pro 295 -Met 537 fragment (Fig. 2C). We confirmed that both of the two Met-Thr sequences within DcPSKR1-⌬KD-His 6 are resistant to CNBr treatment, using bacterially expressed recombinant DcPSKR1-⌬KD-His 6 (data not shown). The theoretical molecular size of this fragment is 26.6 kDa, indicating that it contains several post-translationally added moieties that are resistant to PNGase F treatment.

Chemical Fragmentation and Enzymatic
Digestion of labeled DcPSKR1-⌬KD-His 6 by endoproteinase Asp-N, which cleaves polypeptides at the N-terminal of Asp residues (five sites within the Pro 295 -Met 537 fragment) yielded fast migrating labeled peptides with an approximate size of 5 kDa, with or without PNGase F treatment (Fig. 2B), suggesting that the cross-linking site is in the C-terminal region of the CNBr fragment (Fig. 2C). In contrast, the labeled fragment obtained by trypsin digestion (18 sites within Pro 295 -Met 537 fragment) was no longer detectable on the gel, indicating that the molecular size of the labeled fragment was Ͻ3.5 kDa (Fig.  2B). These results indicate that the cross-linking site is confined to one tryptic fragment within the island domain. The most likely location of the cross-linking site in DcPSKR1-⌬KD-His 6 is the region Glu 503 -Lys 517 (Fig. 2C).
On-column Photoaffinity Labeling of DcPSKR1-⌬KD-His 6 -To confirm the location of the cross-linking site of ASA-PSK by mass spectrometry, we performed large scale photoaffinity labeling followed by trypsin digestion. The main difficulty in identifying photoaffinity-labeled peptide fragments contained in the complex enzymatic digests of the labeled protein is that the relative abundance of the labeled fragment is extremely low due to the low efficiency of the photoaffinity cross-linking reaction. To overcome this limitation, we established a novel oncolumn photoaffinity labeling methodology that allows repeated incorporation of the photoaffinity label. We immobilized DcPSKR1-⌬KD-His 6 on nickel-chelating HiTrap TM HP-Sepharose beads using a His 6 tag and performed solid-phase photoaffinity labeling in a transparent narrow column by directly irradiating the column with UV light (365 nm). The crucial advantage of the on-column photoaffinity labeling is that the cross-linking reaction can be repeated after washing out the uncross-linked ligands that act as potential competitors of the newly added ligands in the next round of photoaffinity reaction. In addition, this system allows direct buffer exchange without dialysis upon enzymatic digestion.
We repeated sequential on-column photoaffinity labeling three times and confirmed the significant increase in cross-linking of [ 125 I]ASA-PSK to immobilized DcPSKR1-⌬KD-His 6 (Fig. 3A). We also confirmed that labeled DcPSKR1-⌬KD-His 6 can be successfully digested by trypsin even if it is immobilized on nickel-chelating beads (Fig. 3A). Released labeled fragment was no longer detectable on SDS-PAGE after trypsin digestion, confirming that its molecular size was Ͻ3.5 kDa.
To gain more information about the molecular size of the labeled tryptic fragment, we chromatographed the tryptic digest of labeled DcPSKR1-⌬KD-His 6 on a gel filtration column. Autowell ␥ counting of each fraction revealed the presence of two clear peaks (Fig. 3B). High pressure liquid chromatography analysis of the larger peak fraction showed that radioactivity was not retained on the reverse-phase column, suggesting that the radioactive molecule contained in this fraction was free 125 I rather than labeled fragments (data not shown). Based on the separation range of Sephadex G-25 (M r 5000 -800) and the elution profiles of blue dextran (marker for V 0 ) and cyanocobalamin (marker for molecular mass separation, M r 1355), the apparent molecular size of the labeled peptide contained in the smaller peak was estimated to be between 1.5 and 5 kDa. Together with the data from the SDS-PAGE of the labeled tryptic peptide, we estimated the apparent molecular size of the labeled fragment to be between 1.5 and 3.5 kDa.
Immunoprecipitation and MALDI-TOF MS Analysis of the Labeled Peptide-To directly detect the peptide fragment crosslinked with ASA-PSK using MALDI-TOF MS, we purified the labeled fragment by immunoprecipitation using anti-PSK antibodies. We performed large scale on-column photoaffinity labeling of Ϸ200 pmol of purified DcPSKR1-⌬KD-His 6 using non-radioactive ASA-PSK (repeated 10 times) and purified labeled fragments derived from the tryptic digest of the labeled protein using the immunoaffinity column. Using positive mode MALDI-TOF MS analysis, we identified two specific molecular ion peaks at m/z 1752.83 and 1880.87 in the eluate of the immunoaffinity column; these peaks were not detected in the control experiments using the tryptic digest of the unlabeled protein (Fig. 4A). Upon assignment of these peaks, we considered two possibilities, (i) that cross-linked peptide fragments were detected as an adducted form and (ii) that cross-linked peptide fragments were detected in their free form due to the loss of the photo-incorporated moiety. MALDI-TOF MS analysis often causes cleavage of unstable cross-linked sites. For example, MALDI-TOF MS analysis of the purified labeled fragment of porcine guanylyl cyclase C yielded a considerable amount of the free form due to cleavage of the unstable nitrene-mediated cross-linked site (22). Similar results have been reported in studies in which ligand binding domains were examined using photoaffinity labeling (23,24).
Of the two above-described molecular ion peaks, the peak at m/z 1752.83 perfectly matched the calculated mass value for the free form of the tryptic peptide fragment (Glu 503 -Lys 517 ) of DcPSKR1-⌬KD-His 6 (calculated exact mass of ENAVEEPSPD-FPFFK (in protonated form), 1752.80) (Fig. 4B). This domain corresponds to the N-terminal side of the 36-amino-acid island domain of DcPSKR1. Similarly, the peak at m/z 1880.87 matched the miscleaved tryptic peptide fragment (Glu 503 -Lys 518 ) derived from the same region of DcPSKR1-⌬KD-His 6 (calculated exact mass of ENAVEEPSPDFPFFKK (in protonated form), 1880.89) (Fig. 4B). Because the synthetic peptide ENAVEEPSPDFPFFK alone exhibited no interaction with the immunoaffinity column (data not shown), it is likely that these fragments were immunoprecipitated with the cross-linked ASA-PSK and were generated by the cleavage of cross-linked sites presumably due to the high energy from the laser beam during MALDI-TOF MS analysis. This possibility is supported by the following three observations. (i) Negative mode MALDI-TOF MS of the same sample detected an ion peak at m/z 900.36, which corresponds to the [M-H-SO 3 ] Ϫ ion of the cleaved ligand (Fig. 4A). (ii) The appearance of the peak at m/z 1880.87, which corresponds to the miscleaved tryptic peptide fragment (Glu 503 -Lys 518 ), is consistent with Lys 517 being a cross-linking site that is resistant to trypsin digestion due to modification of its side chain. (iii) The calculated mass of the adduct form (2741.10) is within the estimated range of the molecular size obtained from gel filtration experiments. Because negatively charged peptides, such as phosphorylated and sulfated peptides, are often resistant to ionization in positive mode MALDI-TOF MS analysis, it is possible that the level of ionization of the adduct form was below the limit of detection.
Deletion of Glu 503 -Lys 517 of DcPSKR1 Abolishes Ligand Binding Activity-To confirm that the 15-amino-acid Glu 503 -Lys 517 region within the island domain of DcPSKR1 is involved in ligand binding, we generated a deletion mutant of DcPSKR1 that lacks Glu 503 -Lys 517 (DcPSKR1-⌬ID[Glu 503 -Lys 517 ]) (Fig.  5A). The ligand binding assay using [ 3 H]PSK showed that PSK binding activity in DcPSKR1-⌬ID[Glu 503 -Lys 517 ] membranes decreased to background levels despite successful expression of truncated proteins detected by immunoblot analysis, indicating that the region Glu 503 -Lys 517 is necessary for ligand binding (Fig. 5, B and C). We also prepared a deletion mutant of DcPSKR1 lacking Lys 518 -Ile 538 , which is a region adjacent to Glu 503 -Lys 517 within the island domain. The ligand binding assay showed that DcPSKR1-⌬ID[Lys 518 -Ile 538 ] also completely lacked PSK binding activity, suggesting that the entire island domain is necessary for ligand recognition.
We determined that the ligand contact domain of DcPSKR1 is located within the 15-amino-acid Glu 503 -Lys 517 region of the island domain and that Glu 503 -Lys 517 and several adjacent residues together form a functional ligand binding pocket.

DISCUSSION
Based on the present results of on-column photoaffinity labeling, MALDI-TOF MS analysis, and fragmentation of labeled DcPSKR1-⌬KD-His 6 by CNBr and endoproteinase Asp-N, we conclude that the 15-amino-acid Glu 503 -Lys 517 region located on the N-terminal side of the 36-amino-acid island domain is the ligand contact domain of DcPSKR1. This region is also highly conserved in Arabidopsis PSK receptor AtPSKR1 (10). On-column photoaffinity labeling allows repeated cross-linking, thus increasing labeling efficiency without interference by uncross-linked photolyzed ligands, which act as potential competitors of the newly added photoaffinity ligands in the next round of the photoaffinity reaction. This system is also compatible with enzymatic digestion after changing to the appropriate buffer and enables us to use MALDI-TOF MS to directly analyze labeled fragments derived from small quantities of natural receptor preparations.
It has been reported that the brassinosteroid (BR) receptor BRI1, which also belongs to the LRR-RK family, directly interacts with photoactivatable BRs and that the 90-amino-acid region containing the island domain and an adjacent single LRR can recognize BRs even when they are bacterially expressed as a glutathione S-transferase fusion (25). These findings indicate that this region can, on its own, form a functional binding pocket for recognition of BRs. However, in the present study, recombinant glutathione S-transferase fusion proteins containing the island domain of DcPSKR1 and several adjacent LRRs did not bind to PSK (data not shown), suggesting that this domain is not sufficient for the formation of a stable ligand binding pocket. The 70-amino-acid island domain of BRI1 contains two cysteine residues, which may form a disulfide bond that stabilizes the ligand binding pocket. In contrast, there are no cysteine residues within the 36-aminoacid island domain of DcPSKR1 or the several LRRs adjacent to it. Although further experiments are required to demonstrate how extracellular 21 tandem LRRs of DcPSKR1 contribute to the formation of a stable ligand binding pocket, we speculate that the global ternary structure of the DcPSKR1 extracellular domain, rather than a local sequence motif within the island domain, defines the specific conformation of the island domain by which DcPSKR1 recognizes PSK with high affinity and specificity.
The theoretical structural model of Cf-9, a membrane-localized LRR-receptor-like protein of tomato cells (Lycopersicon pimpinellifolium) involved in the resistance response to the fungal pathogen Cladosporium fulvum, shows that its 38-aminoacid island domain forms a unique "loop-out" structure that is isolated from the adjacent LRR loops (26). Secondary structure prediction of DcPSKR1 using the multivariate linear regression combination software (27) suggests that its 36-amino-acid island domain has no characteristic structural motif, such as ␣-helix and ␤-strand, which are highly conserved in the LRR domain (data not shown). We propose that the island domain acts as a flexible hinge-like region that modulates the relative conformation of the two surrounding LRR regions upon ligand binding. An island domain has been found in several LRR-RKs, including BRI1 (28) and tBRI1/SR160 (29), and in some LRR-receptor-like proteins including CLV2 (30), Cf-9 (31), and LeEIX (32); each of these island domains has a unique and specific amino acid sequence that is distinct from that of the highly conserved LRR motif (33).
There are also many LRR-RKs and receptor-like proteins that have no island domain, such as FLS2 (34), Xa21 (35), and TMM (36). In a recent report, photoaffinity labeling indicates that flagellin, an elicitor-active structural component of bacterial flagella, FIGURE 4. MALDI-TOF MS analysis of tryptic-digested on-column-labeled DcPSKR1-⌬KD-His 6 . A, labeled DcPSKR1-⌬KD-His 6 fragments purified using the immunoaffinity column were analyzed by MALDI-TOF MS. The control experiment was performed using the tryptic digest of the unlabeled proteins. Specific ion peaks were detected in both the positive and negative mode. B, photolabeled peptides that were detected by MALDI-TOF MS are summarized. directly binds and cross-links to FLS2 (37). Although the ligand binding site of FLS2 has not been identified, two lines of evidence suggest that a LRR domain is involved in its ligand binding. First, the Arabidopsis mutant fls2-24, which carries a point mutation in one of its 28 LRRs, completely lacks flagellin binding activity (37). Second, studies indicate that in mammals, flagellin binds in a highly specific manner to the extracellular LRR domain of TLR5 (38,39). The existence of two distinct ligand perception systems among plant LRR-RKs may reflect plant strategies for adaptation to dynamic environmental conditions using limited molecular components. Indeed, the tomato LRR-RK tBRI1/SR160 recognizes both BRs and systemin, and the lack of competition between these two ligands suggests the presence of two distinct binding sites in tBRI1/SR160 (29).
A fundamental question in the study of transmembrane surface receptors is how ligand binding switches the receptor signaling state between active and inactive states. Identification of the ligand binding site of DcPSKR1 is an important step toward clarifying the regulatory mechanisms of receptor-mediated signal transduction. The biochemically detectable interaction between PSK and DcPSKR1 can serve as a model system for studying the molecular basis of interaction between yet-uncharacterized ligands and putative receptors.