S cysteine‐rich (SCR) binding domain analysis of the Brassica self‐incompatibility S ‐locus receptor kinase
2007; Wiley; Volume: 175; Issue: 4 Linguagem: Inglês
10.1111/j.1469-8137.2007.02126.x
ISSN1469-8137
Autores Tópico(s)Plant Virus Research Studies
ResumoNew PhytologistVolume 175, Issue 4 p. 619-629 Free Access S cysteine-rich (SCR) binding domain analysis of the Brassica self-incompatibility S-locus receptor kinase Benjamin P. Kemp, Benjamin P. Kemp Department of Biology and Biochemistry, University of Bath, Claverton Down, Bath BA2 7AY, UKSearch for more papers by this authorJames Doughty, James Doughty Department of Biology and Biochemistry, University of Bath, Claverton Down, Bath BA2 7AY, UKSearch for more papers by this author Benjamin P. Kemp, Benjamin P. Kemp Department of Biology and Biochemistry, University of Bath, Claverton Down, Bath BA2 7AY, UKSearch for more papers by this authorJames Doughty, James Doughty Department of Biology and Biochemistry, University of Bath, Claverton Down, Bath BA2 7AY, UKSearch for more papers by this author First published: 11 June 2007 https://doi.org/10.1111/j.1469-8137.2007.02126.xCitations: 25 Author for correspondence: James Doughty Tel: +44 1225383485 Fax: +44 1225826779 Email: [email protected] AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Summary • Brassica self-incompatibility, a highly discriminating outbreeding mechanism, has become a paradigm for the study of plant cell–cell communications. When self-pollen lands on a stigma, the male ligand S cysteine-rich (SCR), which is present in the pollen coat, is transmitted to the female receptor, S-locus receptor kinase (SRK). SRK is a membrane-spanning serine/threonine receptor kinase present in the stigmatic papillar cell membrane. Haplotype-specific binding of SCR to SRK brings about pollen rejection. • The extracellular receptor domain of SRK (eSRK) is responsible for binding SCR. Based on sequence homology, eSRK can be divided into three subdomains: B lectin-like, hypervariable, and PAN. • Biochemical analysis of these subdomains showed that the hypervariable subdomain is responsible for most of the SCR binding capacity of eSRK, whereas the B lectin-like and PAN domains have little, if any, affinity for SCR. Fine mapping of the SCR binding region of SRK using a peptide array revealed a region of the hypervariable subdomain that plays a key role in binding the SCR molecule. • We show that residues within the hypervariable subdomain define SRK binding and are likely to be involved in defining haplotype specificity. Introduction The selective pressure on hermaphrodite flowering plants to prevent inbreeding has been a powerful evolutionary driving force, resulting in the evolution of numerous self-incompatibility (SI) systems (Whitehouse, 1950). SI is a prezygotic breeding barrier that allows a stigma to distinguish between self- and cross-pollen. Amongst members of the Brassicaceae, SI is characterized by the inhibition of self- or self-related pollen at the stigmatic surface (Kemp & Doughty, 2003). The molecular events of Brassica SI have been elucidated in some detail and depend on a receptor:ligand interaction that occurs at the stigmatic surface (Takayama & Isogai, 2005). The receptor is a membrane-spanning serine/threonine receptor kinase termed S-locus receptor kinase (SRK) (Takasaki et al., 2000). SRK is located in the plasma membrane of the papillar cells that cover the stigmatic surface (Stein et al., 1996). The extracellular domain (eSRK), which is responsible for ligand binding (Kachroo et al., 2001), is highly polymorphic between haplotypes, as one would expect for a specificity-determining molecule (Stein et al., 1991). The ligand for SRK is termed S cysteine-rich (SCR) (Schopfer et al., 1999) and has also been designated SP-11 (Takayama et al., 2000). SCR is a member of one family of small, cysteine-rich proteins which reside in the lipid-rich pollen coat (Doughty et al., 2000). SCR molecules are basic and hydrophilic and show extreme variability among haplotypes. Eight cysteine residues are conserved among haplotypes and these are involved in the formation of SCR tertiary structure. This disulphide framework means that SCR molecules from different haplotypes have similar predicted three-dimensional structures despite having variable primary structures (Mishima et al., 2003). The haplotype-specific binding of SCR to SRK resulting in rejection of self-pollen was demonstrated in two seminal papers of 2001 (Kachroo et al., 2001; Takayama et al., 2001). SRK bound self-SCR but not cross-SCR in experiments with both native protein from stigmatic plasma membranes (Takayama et al., 2001) and recombinant protein produced in Nicotiana benthamiana (Kachroo et al., 2001). Although both papers agreed that the haplotype-specific interaction between SRK and SCR was of critical importance in the control of pollen rejection, there were some differences in the fine details of the interactions reported. Recombinant eSRK6, produced in N. benthamiana, was able to bind SCR6 but not SCR13 (Kachroo et al., 2001), whereas recombinant eSRK8, produced in silkworm larvae (Bombyx mori), did not bind SCR8 (Takayama et al., 2001). Analysis of the proteins responsible for SCR8 binding in S8S8 stigmatic microsomes suggested that an S-locus glycoprotein (SLG)-like molecule forms a high-affinity binding site for SCR8 (SLG is similar in sequence to eSRK). A second, low-affinity, binding site was also present in the receptor complex (Takayama et al., 2001). Takayama's group have recently further characterized the high-affinity SCR8 binding site in stigmatic microsomal membranes and determined that it is composed of SRK and a truncated membrane-anchored version of the receptor (tSRK) (Shimosato et al., 2007). They also demonstrated that high-affinity binding between SCR and SRK could only be achieved when the receptor was membrane-anchored. No requirement for membrane localization, second binding site, or role of other molecules was found in S6S6 stigmas (Kachroo et al., 2001). These differences could be explained by variation among haplotypes either in the structure of the receptor complex or in the binding sites found in specific SRK molecules. Among-haplotype variation was uncovered when the specificity determinants of SCR were investigated. Only a few amino acids need be exchanged to swap the specificities of some SCR haplotypes, whereas others are more recalcitrant to change (Chookajorn et al., 2004; Sato et al., 2004). SCR6 was converted to S13 specificity by replacing a stretch of five amino acids in region V (VPTGR) with four amino acids from the corresponding region in SCR13 (TDTQ); however, the specificity of SCR13 could only be changed to S6 by altering the majority of the protein (Chookajorn et al., 2004). Using a similar rationale, Sato and coworkers analysed the amino acids governing specificity in two very similar SCR molecules with different recognition specificities and found that two regions of SCR (regions III and V) were the most important in determining specificity (Sato et al., 2004). Chookajorn et al. (2004) managed to create SCR chimeras that bound eSRK strongly but were unable to elicit an SI response, indicating that binding and activation may be distinct for some haplotypes (Chookajorn et al., 2004). To date, the specificity determinants of SRK have not been examined experimentally; however, bioinformatic speculation has resulted in a number of hypotheses as to which region(s) of the molecule is involved. It is known that eSRK is responsible for ligand binding (Kachroo et al., 2001) and eSRK can be split into three subdomains based on sequence similarity (Shiu & Bleecker, 2001, 2003). The N-terminal region is similar to that of mannose-binding lectins, the middle hypervariable region contains most of the variability seen among haplotypes, and the final C-terminal region is most similar to a PAN or apple domain involved in protein–protein or protein–carbohydrate interactions. Of most interest to researchers has been the hypervariable region, where three distinct regions of variability (HVI, HVII and HVIII) have been identified and are thought to be under balancing selection (Nishio & Kusaba, 2000; Sato et al., 2002). In addition, a number of specific amino acids are speculated to be under selection to change physiochemical properties, which may explain the evolution of new S haplotypes (Sainudiin et al., 2005). Here, data are presented demonstrating that the majority of SCR binding is focused in the hypervariable subdomain. Further, the binding of the hypervariable subdomain is mapped using a peptide array and we speculate that pollen rejection is dependent on a primary initial binding followed by haplotype-specific reinforcement. Materials and Methods Production of SCR SCR cDNA (minus the predicted signal sequence) from haplotypes S5, S29 and S63 was cloned into pET32a and transformed into the Origami cell line (Merck Biosciences, Nottingham, UK). Protein was expressed by growing cells to OD600 (optical density at 600 nanometres) 0.5 at 37°C and then induced with 1 mm isopropyl-beta-D-thiogalactopyranoside at 6000g for 10 min, resuspended in phosphate-buffered saline (PBS) with 1% Triton X-100 and lysed by three cycles of freeze–thawing, and chromosomal DNA was sheared using a probe sonicator. Insoluble material was pelleted by centrifugation at 10 000g for 15 min and the SCR:TRX (thioredoxin) fusion protein purified using Talon Co2+ affinity resin (Clontech, Saint-Germain-en-Laye, France). The TRX tag was cleaved from the SCR by adding thrombin and dialysing in a 3000-kDa molecular weight cut-off dialysis bag against 1 × thrombin buffer (Merck Biosciences). Several different isoforms of SCR were present, presumably the result of incorrect disulphide bond formation. SCR was purified to a single, active, isoform by reverse-phase high-performance liquid chromatography (HPLC), proteins being eluted in a linear gradient of 15–45% (volume/volume (v/v)) acetonitrile 0.1% (v/v) trifluoroacetic acid. SCR bioassay Bioassays were performed essentially as described in Takayama et al. (2001), except that a Hamilton syringe was found to give more reproducible results in applying the SCR than a micropipette. Iodination of SCR Seven milligrams of purified SCR was iodinated with 10 mCi Na125I (GE Healthcare, Pollard Wood, UK) using Iodogen reagent (Perbio Science, Northumberland, UK) and purified from unreacted Na125I using a column packed with Sephadex G-10 (GE Healthcare). Isolation of crude stigmatic membranes Stigmas (100 mg) were homogenized in 200 µl of membrane buffer (30 mm Tris-Cl, pH 7.5, 75 mm NaCl, 0.1 mm ethylenediaminetetraacetic acid (EDTA) and 20% glycerol) supplemented with 5 mm dithiothreitol and phenylmethylsulphonyl fluoride. The homogenates were then centrifuged twice for 10 min at 4000g. Membranes were pelleted by centrifugation at 35 000g for 1 h, washed in 1 ml of membrane buffer and pelleted at 35 000g for 45 min. The pellet was resuspended in 100 µl of membrane buffer and the protein concentration was determined using a micro Bradford assay (Sigma-Aldrich, Gillingham, UK). SCR binding assays using dot blots Between 10 and 100 µg of protein was blotted onto Immobilon P membrane (Millipore, Watford, UK) and blocked for 1 h in 1 × Tris-buffered saline (TBS), 0.01% (v/v) Tween 20 and 0.3% (weight/volume (w/v)) casein. Blots were then incubated overnight with iodinated SCR in blocker and washed for 3 × 10 min in blocker before exposure to X-ray film. Expression of eSRK in N. benthamiana Transient expression of 6xHis and c-myc tagged eSRK29 and eSRK63 was carried out according to Voinnet et al. (2003). Expression levels were checked by excising a leaf disc with a 1.5-ml microfuge tube lid, homogenizing it in sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer and heating it to 95°C for 5 min. Aliquots were run on 12% acrylamide SDS-PAGE gels before blotting onto Immobilon P membranes (Millipore). Equal loading was determined using Ponceau S stain (Sigma-Aldrich). Protein expression was analysed using c-myc primary antibody (Invitrogen, Paisley, UK) and a goat anti-mouse-AP conjugate secondary antibody (Sigma-Aldrich) using nitro-blue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate colorimetric detection. To purify proteins, leaves (100 g) expressing eSRK were ground in 100 ml of 0.1 m NaH2PO4, 0.25 m NaCl, 0.3% (w/v) Sarksoyl and 5 mmβ-mercaptoethanol, pH 7.0, filtered through miracloth and centrifuged at 10 000g for 10 min to remove insoluble debris. A volume of 1 ml of Talon resin (Clontech-Takara Bio Europe) was added to the filtrate, stirred for 1 h at room temperature and then collected by centrifugation (500g for 5 min). The resin was washed with 20 ml of buffer before bound proteins were eluted with buffer supplemented with 150 mm imidazole. Protein was refolded by dialysis for 4 h at 4°C against 10 mm Tris-Cl, pH 7.5, and 5 mmβ-mercaptoethanol to remove the sarkosyl, then for 4 h at 4°C against 10 mm Tris-Cl, pH 7.5, to remove β-mercaptoethanol, and then overnight at 4°C against 10 mm Tris-Cl, 1 mm reduced glutathione and 0.2 mm oxidized gluthathione to encourage disulphide bond formation. Precipitated protein was removed by centrifugation. Production of eSRK subdomains Subdomains of eSRK were amplified by polymerase chain reaction (PCR), cloned into pBADgIII (Invitrogen) and transformed into the TOP10 cell line (Invitrogen). Protein was expressed by growing cells to OD600 0.5 at 37°C and then induced with 0.2% (w/v) l-arabinose for 3 h at 37°C. Cells were harvested by centrifugation at 6000g for 10 min, resuspended in PBS with 1% Triton X-100 and lysed by three cycles of freeze–thawing. Chromosomal DNA was sheared using a probe sonicator. Inclusion bodies were pelleted by centrifugation at 10 000g for 15 min and solubilized in 8 m urea, 500 mm NaCl and 50 mm sodium phosphate, pH 7.0, before purification with Talon Co2+ affinity resin (Clontech). Pull-down assays Recombinant SCR was bound to S-protein agarose (Merck Biosciences) utilizing the S-tag. SCR (1 µg) was bound to S-protein agarose (50 µl) and the mixture was washed three times in 1 × bind buffer (20 mm Tris-Cl, pH 7.5, 150 mm NaCl and 0.1% Triton X-100) to remove unbound protein. Recombinant eSRK29::myc (5 µg) in bind buffer was incubated with the SCR resin for 3 h at room temperature with gentle agitation. The resin was washed twice in bind buffer to remove unbound proteins. The S-tag/S-protein interaction was disrupted by incubation in 0.2 m citrate, pH 2, for 10 min at room temperature, releasing the tagged SCR and any bound eSRK29::myc. Equivalent volumes from each step were analysed by western blot using a mouse anti-c-myc antibody (Invitrogen). Microtitre plate analysis of binding Enzyme-linked immunosorbent assay (ELISA) plates were coated with 1 µg of eSRK::myc in TBS Tween-20 (TBST) overnight at 4°C and blocked with 0.3% (w/v) casein in TBST for 1 h at room temperature. Increasing amounts of recombinant SCR29 in TBS were then added to the coated plates which were left for 1 h at room temperature. Plates were washed three times with TBST before detection of bound SCR29 either with a rabbit anti-SCR29 polyclonal antibody (raised against the peptide LRKRGPEHYSLPGVC fused to keyhole limpet haemocyanin and tested for SCR29 specificity by comparing binding to S29 pollen coat proteins and S25 pollen coat proteins) or by the use of the FRETworks S-tag assay kit (Merck Biosciences). Peptide array production The hypervariable subdomains of SRK5, SRK29, SRK63 and SLG29 were aligned using ClustalW and 20mer peptides spanning 160 amino acid residues of the hypervariable domain were designed, each peptide overlapping its neighbour by 10 amino acids (see Supplementary Material, Table S1, for peptide sequences). Peptides were synthesized as a PEPscreen custom peptide library (Sigma-Genosys, Haverhill, UK). Four peptides unsuccessfully synthesized using this procedure (peptides 6 and 7 from both SRK29 and SRK63) were synthesized by Cambridge Peptides (Haverhill, UK). Peptides were dissolved according to the scheme developed by Sigma-Genosys for dissolving PEPscreen libraries to 1 mg ml−1. Using a vacuum manifold, 100 µg of each peptide was spotted onto Immobilon P polyvinyl fluoride (PVDF) membrane and allowed to dry. The membrane was then blocked in TBST with 0.3% casein (w/v) for 1 h before the addition of iodinated SCR and overnight incubation at room temperature. Membranes were washed three times for 10 min each in TBST with 0.3% casein (w/v) and exposed to X-ray film. Results The extracellular domain of SRK consists of three subdomains To uncover the ligand binding region of SRK, we first reviewed the possible domain architecture of eSRK. Previous work had suggested that eSRK is composed of three separate regions: a B lectin-like, a hypervariable and a PAN-like domain (Shiu & Bleecker, 2001, 2003). We analysed the sequence of eSRK from Brassica oleracea haplotypes S5, S29 and S63 using the Pfam database (http://www.sanger.ac.uk/software/pfam/) (Fig. 1a). Figure 1Open in figure viewerPowerPoint Structural architecture of the extracellular receptor domain of S-locus receptor kinase (eSRK). (a) Scale diagram of the extracellular S-domain of SRK29 highlighting the distinct subdomains. Hypervariable regions (HVI–HVIII) are indicated above the diagram. Black vertical lines indicate cysteine residues, and the predicted signal peptide is shown by an arrow. (b) ClustalW alignment of the lectin-like domains from SRK5, SRK29 and SRK63 with the Pfam domain for B lectin. Residues that are conserved or similar between the Pfam domain and SRK sequences are shown in uppercase and residues that are different are shown in lowercase. *, conserved residue; :, similar residue; -, gap introduced to the alignment. The QxDxNxVxY motif essential for mannose binding (Ramachandraiah & Chandra, 2000) is highlighted in grey. The initial 170 amino acids of SRK, after the predicted signal peptide, show significant similarity to the Pfam B lectin domain (Fig. 1b). Lectins are involved in sugar binding, and the B lectin type is specific for mannose. This raised the intriguing possibility that SRK was involved in binding mannose, or possibly mannose containing glycoproteins. Both SRK and SLG are glycoproteins, and it has long been speculated that carbohydrate interactions may be important in Brassica SI (e.g. Sarker et al., 1988). We analysed all SRK sequences present in the GenBank database but none contained the QxDxNxVxY motif essential for mannose binding (Ramachandraiah & Chandra, 2000), indicating that the lectin-like domain does not bind mannose (Fig. 1b). Around the mannose binding motif the sequence is highly conserved, indicating that the B lectin-like domain may function in a structural capacity. After the B lectin-like domain there is a region that shows much less identity among haplotypes. The Pfam domain that corresponds to this region is called the S-locus glycoprotein family domain. As most of the variability observed among haplotypes is located in this region, we refer to it as the hypervariable subdomain. Three adjacent regions of hypervariability in the subdomain have been identified (Nishio & Kusaba, 2000) and have been shown to be under balancing selection (Sato et al., 2002; Sainudiin et al., 2005). Finally, a PAN or apple domain is present in the last 90 amino acids before the transmembrane region. PAN domains are predicted to be involved in protein–protein or protein–carbohydrate interactions. This domain contains six cysteine residues assumed to be linked via disulphide bonds; in human plasma prekallikrein the three disulphide bonds link the first and sixth, second and fifth and third and fourth cysteines (McMullen et al., 1991). We thus split eSRK into three subdomains: the lectin-like, hypervariable and PAN domains. Using ClustalW we aligned 26 full-length eSRK amino acid sequences (14 from Brassica oleracea, eight from Brassica rapa, two from Brassica napus and two from Arabidopsis lyrata) and compared the percentage identity across the three subdomains. The lectin-like subdomain had an identity of 82%, the hypervariable subdomain an identity of 70% and the PAN subdomain an identity of 80%. Although the variability between S haplotypes is concentrated in the hypervariable subdomain, other regions also contain haplotype-specific polymorphisms that may have functional significance. Recombinant SCR can initiate pollen rejection and binds stigmatic plasma membrane protein in a haplotype-dependent manner To identify which region(s) of SRK was responsible for ligand binding, a source of functional SCR was required. Recombinant mature SCR29 (without a signal peptide) with 6xHis and S-tags was produced in Escherichia coli. After purification, biological activity was tested using the bioassay initially developed by Stephenson and colleagues (Stephenson et al., 1997) but without using pollen coat as a carrier (Takayama et al., 2001). Recombinant SCR29 applied to S29S29 stigmas before cross-pollination with S63 pollen was able to block this normally compatible interaction, with stigmas showing few pollen tubes (Fig. 2a). When SCR29 was applied to S63S63 stigmas before the application of S29 pollen, pollen tubes developed normally, indicating that pollen rejection was haplotype specific and not attributable to a general inhibitory effect of recombinant SCR (Fig. 2a). Figure 2Open in figure viewerPowerPoint Recombinant S cysteine-rich (SCR) is biologically active. (a) Purified recombinant SCR29 was applied to the stigmas homozygous for either the S63 (1) or S29 (2) self-incompatibility (SI) haplotypes before pollination. Recombinant SCR29 caused the normally compatible cross S63 pollen to fail on S29 stigmas (2), whereas S29 pollen produced large numbers of tubes (arrow) on S63 stigmas (1). (b) Recombinant SCR binds stigmatic membrane proteins in a haplotype-specific manner. Crude stigmatic microsomal membrane preparations from either S63S63 (right) or S29S29 stigmas (left) were probed with iodinated recombinant SCR29 before exposure to X-ray film. The S29S29 membranes bound SCR29 much more tightly than those derived from S63S63 stigmas. Recombinant SCR29 was further characterized by binding assays with crude microsomal membrane proteins isolated from S29S29 and S63S63 stigmas. Membrane proteins were incubated with increasing amounts of S-tagged SCR29 and the amount of bound protein determined utilizing a FRET-based S-tag detection assay. Scatchard analysis showed that membranes from S29S29 (disassociation constant Kd = 6.6 µm) stigmas have fivefold higher affinity for SCR29 than S63S63 stigmas (Kd = 29.5 µm). To confirm this result, recombinant SCR29 was radiolabelled with 125I and bound to stigmatic microsomal membrane proteins on dot blots. SCR29 was found to bind self-membranes more than cross-membranes (Fig. 2b), but a small amount of cross-binding was still observed. Taken together, these results indicate that E. coli expressed SCR29 is biologically active and binds to self-S29 stigmatic microsomal membrane proteins with a greater affinity than cross-membranes. eSRK produced in N. benthamiana binds SCR in a haplotype-independent manner Before subdomain analysis, it was of interest to determine if the extracellular portion of SRK alone retained S specificity. A number of heterologous expression systems have been used to express eSRK previously: E. coli was found to degrade eSRK (Stein et al., 1996); silkworm larvae expressed eSRK which did not bind SCR (Takayama et al., 2001), and eSRK expressed transiently in N. benthamiana was found to bind SCR in a haplotype-specific manner (Kachroo et al., 2001). With this in mind, we utilized an Agrobacterium-mediated transient expression system using N. benthamiana and the p19 suppressor of gene silencing (Voinnet et al., 2003). We expressed eSRK29 with c-myc and 6xHis tags in N. benthamiana leaves, and western blot analysis identified a number of tightly clustered Mr versions of eSRK. These polypeptides are all above the predicted size for tagged eSRK (48 kDa) and represent different glycoforms of the receptor (Fig. 3a). Figure 3Open in figure viewerPowerPoint S cysteine-rich (SCR) binding of the extracellular receptor domain of S-locus receptor kinase (eSRK) expressed in Nicotiana benthamiana. (a) Expression of eSRK29 in N. benthamiana. A blot of total protein from leaf discs of N. benthamiana previously infiltrated with Agrobacterium carrying pBIN61 with the p19 suppressor of gene silencing either alone (p19) or coinfiltrated with Agrobacterium carrying pBIN61 eSRK29::myc::6xHis. The blot was probed with an anti-myc antibody. Several molecular weight variants of eSRK were detected which correspond to the multiple glycoforms typical for this receptor. (b) SCR29 binding of eSRK29 and eSRK63. Microtitre plates were coated with eSRK29, eSRK63 or bovine serum albumin (BSA) and incubated with increasing amounts of SCR29. Bound SCR29 was detected with an SCR29-specific polyclonal antibody. The two eSRK molecules bound SCR29 with similar efficiencies. (c) Pull-down assays of eSRK29::myc. eSK29::myc was incubated with either S-tagged SCR29 or S-tagged SCR63. eSRK–SCR complexes were then pulled down with S protein agarose, washed twice and finally eluted with sodium dodecyl sulphate (SDS) loading buffer. The pull-down was analysed by western blot using anti-c-myc antibody. eSRK29 is the input protein, F is the flow-through, W1 and W2 are the two washes and E is the final elution. No specificity was observed in the abilities of SCR63 and SCR29 to pull down eSRK29. eSRK29 did not bind S protein agarose in the absence of SCR (data not shown). Despite good levels of expression in this system we were unable to purify eSRK in significant quantities without the use of ionic detergents and reducing agents. This suggests that the high levels of expression overwhelmed the protein secretion machinery, probably resulting in incorrect folding and inappropriate disulphide bond formation. Co-expression with SLG29 did not improve the yield or solubility of eSRK29, despite previous reports that SLG may function to stabilize SRK (Dixit et al., 2000). Utilizing in vitro refolding we were able to purify enough soluble eSRK29 and eSRK63 to analyse SCR29 binding. To test the binding specificities of recombinant eSRK, microtitre plates were coated with eSRK29 and eSRK63, and then incubated with increasing amounts of SCR29. Bound SCR29 was detected in an ELISA assay using an SCR29-specific polyclonal antibody. Both receptors bound SCR29 in a similar manner, showing no obvious haplotype specificity (Fig. 3b). We conducted pull-down assays of eSRK29 using both SCR29 and SCR63 to confirm this result. SCR29 and SCR63 were able to pull down eSRK29 with equal efficiency (Fig. 3c). These data indicate that, for these haplotypes at least, eSRK in isolation retains the ability to bind SCR but does not display S specificity in vitro. A functional S-specific complex is likely to require either a native membranous environment or specific accessory molecules to assemble. The hypervariable subdomain of eSRK is responsible for the majority of SCR binding To analyse which subdomains of eSRK were responsible for SCR binding, an efficient method for expressing these regions of the molecule was required. Subdomains were expressed in E. coli because of the aforementioned problems with Agrobacterium-mediated transient expression. We were able to express all three subdomains from eSRK29 in E. coli, without the problems of degradation previously reported when the whole of eSRK was expressed (Stein et al., 1996). All three subdomains were present as inclusion bodies, which could be solubilized with 8 m urea. We attempted to improve solubility by varying the expression vector (Sheffield et al., 1999), the level of induction and the bacterial growth conditions, and by using in vitro refolding, but the subdomains remained insoluble (data not shown). We used 6xHis and c-myc-tagged protein for binding studies following purification, via the 6xHis tag, under denaturing conditions. Subdomains were immobilized before the removal of the denaturant. Limited refolding could be expected to occur under these conditions (analogous to matrix-assisted refolding) (Machold et al., 2005), where spatial separation of polypeptides reduces protein aggregation. Even if these proteins did not form the correct tertiary structure it is likely that a number of linear epitopes would be presented, enabling us to distinguish which subdomains were responsible for ligand binding. Purified domains from SRK29 were blotted onto Immobilon PVDF membranes under denaturing conditions and probed with radiolabelled SCR29. The majority of SCR29 binding capacity was present in the hypervariable subdomain with little or no binding detected for the B lectin-like and PAN subdomains (Fig. 4a). Figure 4Open in figure viewerPowerPoint Binding of isolated subdomains of the extracellular receptor domain of S-locus receptor kinase (eSRK) to S cysteine-rich (SCR). (a) Subdomains from SRK29 (B lectin, hypervariable and PAN) were expressed in Escherichia coli and purified under denaturing conditions. Immobilized subdomains were then probed with radioiodinated SCR29. The majority of binding was to the hypervariable subdomain, although limited binding to the PAN domain was also detected. BSA, bovine serum albumin negative control. (b) As for (a), but subdomains from both SRK29 and SRK63 were probed with radioiodinated SCR63. The m
Referência(s)