Direct interaction of ligand–receptor pairs specifying stomatal patterning

EPF1 and EPF2 associate with ERECTA family in planta

We first tested whether EPF1 and EPF2 bind to TMM, ERECTA, and/or ERL1. ERL2 was not included in the analysis due to its minor role in stomatal patterning (Shpak et al. 2005). Induction of EPF1 or EPF2 in Arabidopsis inhibits stomatal differentiation, resulting in seedling lethality (Hara et al. 2007, 2009; Hunt and Gray 2009). This makes it difficult to obtain sufficient amounts of EPF peptides for biochemical analysis. We therefore used a Nicotiana benthamiana transient expression system (Voinnet et al. 2003) to combinatorially coexpress epitope-tagged ligand–receptor pairs and performed co-IP assays (Fig. 1; Supplemental Fig. 1). To ensure that proper ligand–receptor interactions would be reconstituted in N. benthamiana, we used ligand and receptor constructs that are functional when expressed in Arabidopsis (for functionality, see Figs. 2, 3).

EPF1 and EPF2 possess an N-terminal signal peptide, a propeptide region with low sequence similarity, and a C-terminal predicted mature EPF peptide (MEPF) domain with eight conserved cysteines (Supplemental Fig. 1; Rychel et al. 2010; Shimada et al. 2011). Expression of EPF1-Flag and EPF2-Flag in N. benthamiana resulted in major products corresponding to epitope-tagged MEPF1 and MEPF2 as well as faint signals of higher molecular mass, which are likely partially processed propeptide (Fig. 1). Immunopurified EPF1-Flag and EPF2-Flag conferred severe reductions in stomatal differentiation similar to EPF1 and EPF2 overexpression, respectively (Supplemental Fig. 2), indicating that tobacco-expressed peptides are functional and bioactive.

The epitope-tagged full-length ERECTA and ERL1 were expressed under the control of their respective native promoters. To increase yield, we also expressed their kinase-deleted versions (ΔK) under the control of a strong CaMV35S promoter, since removal of the kinase domain of ERECTA increases protein stability (Shpak et al. 2003). Biochemical fractionation experiments show that ERECTAΔK, ERL1ΔK, and TMM expressed by 35S promoters were predominantly localized in the membrane (microsomal) fraction (Supplemental Fig. 3), confirming that these fusion proteins are localized properly.

Both full-length ERECTA-YFP and ERL1-YFP as well as their kinase-deleted versions (ERECTAΔK-GFP and ERL1ΔK-GFP) were able to coimmunoprecipitate EPF1-Flag and EPF2-Flag (Fig. 1A–D). Thus, as with other LRR-RKs, such as CLAVATA1 (CLV1) and PSKR (Shinohara et al. 2007; Ogawa et al. 2008), the kinase domain of ERECTA family proteins is not required for ligand binding. In contrast, TMM-GFP failed to coimmunoprecipitate EPF1-Flag (Fig. 1E), although EPF2-Flag signal was obtained by TMM-GFP co-IP (Fig. 1E). To examine the specificity of interactions, we performed co-IP assays using epitope-tagged LURE2, an unrelated CRP controlling pollen tube guidance in Torenia (Okuda et al. 2009). LURE2-Flag did not coimmunoprecipitate with ERECTAΔK-GFP, ERL1ΔK-GFP, or TMM-GFP (Supplemental Fig. 4). Likewise, no co-IP signals were detected when GFP was coexpressed with EPF1-Flag and EPF2-Flag (Fig. 1F). The results indicate that EPF1 and EPF2 associate with the ERECTA family in planta, while TMM may show differential binding to each EPF.

Bioactive, recombinant MEPF1 and MEPF2 trigger unique responses in stomatal development

To detect direct ligand–receptor binding, we moved into a synthetic system using receptor biosensor chips. For this purpose, we first produced bioactive, predicted mature epitope-tagged EPF1 and EPF2 peptides (MEPF1-His and MEPF2-His) from Escherichia coli (Materials and Methods; Fig. 2; Supplemental Figs. 1, 5). The C-terminal MEPF domain of EPF1/2 possesses eight cysteines that are predicted to form intramolecular disulfide bonds (Supplemental Fig. 1; Higashiyama 2010; Kondo et al. 2010; Matsubayashi 2010; Sugano et al. 2010). Hence, unlike small post-translationally modified peptides such as MCLV3 and phytosulfokine, proper refolding is critical for the biological activities of CRPs (Higashiyama 2010). We developed protocols for refolding both MEPF1 and MEPF2 expressed and purified from E. coli (see the Materials and Methods).

Application of refolded MEPF1-His (2.5 μM) to Arabidopsis wild-type seedlings conferred an epidermis devoid of stomata with arrested stomatal precursors called meristemoids (Fig. 2A,B), a phenotype identical to induced EPF1 overexpression (Fig. 2C,D). Likewise, application of MEPF2-His (2.5 μM) peptide solutions led to severe inhibition of asymmetric divisions initiating stomatal lineages (Fig. 2E,F), a phenotype identical to induced EPF2 overexpression (Fig. 2G,H). These observations demonstrate the unique biological activities of MEPF1 and MEPF2 in different steps of stomatal development, restricting initial asymmetric division and enforcing proper spacing. Furthermore, the results support the notion that, like EPFL9/Stomagen (Kondo et al. 2010; Sugano et al. 2010), MEPF1 and MEPF2 are mobile signals controlling stomatal development in a non-cell-autonomous manner.

Direct binding of bioactive MEPF1 and MEPF2 to plant-produced receptors on synthetic biosensor platform

Purified, bioactive MEPF1 and MEPF2 peptides were used to perform ligand–receptor-binding assays using biosensor platforms of QCM (quartz crystal microbalance) and SPR (surface plasmon resonance) (Tamerler et al. 2006; Baltus et al. 2007). For both techniques, we first immobilized purified ERECTAΔK-GFP, ERL1ΔK-GFP TMM-GFP, or control GFP from N. benthamiana on gold surfaces via anti-GFP antibody linkages and then introduced the purified MEPF-His peptide solutions (Materials and Methods).

To accurately measure the amounts of bioactive peptide, we performed reverse-phase high-pressure liquid chromatography (HPLC) followed by mass spectrometry and bioassays. Properly folded (and bioactive) peptides were separated from misfolded (and inactive) forms, and their amounts were determined (Supplemental Figs. 6, 7).

Robust binding of MEPF1 and MEPF2 to the ectodomain of ERECTA and ERL1 was detected as a function of frequency change (for QCM) or the shift in SPR wavelength (for SPR) (Fig. 2I–N; Supplemental Fig. 8). Based on the SPR sensorgrams, the binding of MEPF1 to both ERECTA and ERL1 was rapid, saturable, and in similar kinetics (Fig. 2M). The QCM data support this observation (Fig. 2I). In contrast, no obvious binding of MEPF1 to TMM-GFP or control GFP was detected by either SPR or QCM (Fig. 2I,J,M). MEPF2, unlike MEPF1, exhibited binding to TMM in addition to ERECTA and ERL1 (Fig. 2K,L,N). The apparent affinity difference for TMM between MEPF1 and MEPF2 was consistently detected in three different methods (Figs. 1, 2; Supplemental Fig. 8). Additionally, we detected a difference in the binding kinetics of MEPF2 to ERECTA and ERL1, with ERECTA having higher saturation (Fig. 2N).

To address the specificity of ligand binding, we further performed SPR analysis using LURE2 (Okuda et al. 2009). LURE2 showed no binding to ERECTA, ERL1, or TMM under the same conditions (300 nM total refolded peptide solution) (Fig. 2O), and its sensorgram was identical to the buffer-only control (Fig. 2P). Furthermore, heat denaturation of MEPF1 and MEPF2 severely compromised their binding to receptors (Supplemental Fig. 9). Together, the results demonstrate specific, direct binding of bioactive MEPF peptides to the ERECTA family RKs as well as their differential binding to TMM.

In vivo association of ERECTA family RKs and TMM

Unlike ERECTA family RKs, TMM lacks a cytoplasmic effector domain (Nadeau and Sack 2002). To clarify the role of TMM in ERECTA family-mediated signaling, we next investigated the physical associations of these receptors in vivo. For this purpose, we generated transgenic Arabidopsis plants expressing a series of epitope-tagged TMM, ERECTA, and ERL1 driven by their own native promoters. Because ERECTA requires introns for functionality (Karve et al. 2011), genomic DNA was used for the constructs. Epitope-tagged ERECTA, ERL1, and TMM rescued their respective mutant phenotypes, indicating that they are functional (Fig. 3). Since ERL1 is functionally redundant, we introduced the epitope-tagged ERL1 into the erecta erl1 erl2 triple mutant to confirm its functionality (Fig. 3A–E). The epitope-tagged TMM, ERECTA, and ERL1 were detected in membrane (microsomal) fractions, consistent with their predicted subcellular localizations (Supplemental Fig. 10; Nadeau and Sack 2002; Shpak et al. 2004).

We subsequently generated double-transgenic Arabidopsis plants expressing pairwise combinations of functional epitope-tagged receptors and performed co-IP assays (Fig. 4). ERECTA-Flag protein was coimmunoprecipitated with ERECTA-YFP using anti-GFP antibody, indicating that ERECTA associates with itself in vivo (Fig. 4A). Likewise, co-IP experiments demonstrated in vivo association of ERL1 with itself (Fig. 4B). Thus, ERECTA family RKs form receptor homomers in vivo. In contrast, reciprocal co-IP assays using double-transgenic tmm seedlings expressing both proTMM::TMM-YFP and proTMM::TMM-HA failed to precipitate TMM-HA or TMM-YFP by anti-GFP or anti-HA antibody, respectively (Fig. 4C). This indicates that TMM does not form homomers in vivo. However, TMM-YFP and ERECTA-Flag as well as TMM-YFP and ERL1-Flag associate with each other in vivo (Fig. 4D,E).

ERECTA and ERL1 exhibit unique yet overlapping expression patterns in developing epidermis, with ERECTA predominantly in the protoderm and ERL1 in meristemoids and SLGCs (Shpak et al. 2005). We thus tested whether they could form heterodimers in vivo. Indeed, co-IP experiments show that ERECTA-YFP associates with ERL1-Flag (Fig. 4F). Taken together, the results suggest that the ERECTA family function as receptor homodimers as well as heteromers among them, while TMM modulates the signal transduction via association with the ERECTA family.

In vivo specificity of ligand–receptor pairs controlling two critical steps of stomatal development

Genetic studies have shown that EPF1 and EPF2 possess unique functions: EPF1 orients asymmetric spacing divisions to prevent stomatal pairing, while EPF2 inhibits asymmetric entry divisions initiating stomatal cell lineages (Hara et al. 2007, 2009; Hunt and Gray 2009). Consistently, the phenotypes of EPF1 and EPF2 overexpression (or MEPF1 and MEPF2 application) are different (Fig. 2; Hara et al. 2007, 2009). Genetically, however, both EPF1 and EPF2 require TMM and ERECTA family genes to function (Hara et al. 2007, 2009). How can different EPFs elicit distinct responses during stomatal development? The difference in binding affinities of receptors to each MEPF peptide, as seen for TMM (Figs. 1, 2), may determine specificity. In the case of ERECTA and ERL1, however, both receptors show robust ligand binding.

To dissect the in vivo developmental specificities between EPFs and the ERECTA family, we examined the phenotypic consequences of blocking specific ERECTA family signaling (Fig. 5; Supplemental Fig. 11). Simple loss-of-function experiments do not suffice due to partial genetic redundancy, and severe stomatal clustering results only when the entire ERECTA family of genes is missing (Supplemental Fig. 11; Shpak et al. 2004, 2005). We introduced kinase-deleted versions of ERECTA and ERL1 driven by their native promoters into their respective knockout backgrounds. A removal of the ERECTA kinase domain confers severe dominant-negative effects (Shpak et al. 2003). Thus, replacing ERECTA and ERL1 with their respective dominant-negative versions will likely block all EPF signals normally perceived by ERECTA or ERL1 from further signal transduction.

Replacement of the endogenous ERECTA by its dominant-negative version (proERECTA::ERECTAΔK-YFP in erecta) conferred a phenotype nearly identical to that of epf2: a vast increase in asymmetric entry divisions resulting in increased nonstomatal cell density (Fig. 5B,C; Supplemental Fig. 11C,D,K). In contrast, proERL1::ERL1ΔK-CFP in erl1 led to modest stomatal pairing, a phenotype highly resembling that of epf1 (Fig. 5D,E; Supplemental Fig. 11G,H,J). None of the genotypes showed significant changes in stomatal density (Supplemental Fig. 11I). The results place ERECTA and ERL1 as major receptors for EPF2- and EPF1-mediated signaling during stomatal development, respectively.

To ultimately address the in vivo specificity of EPF2–ERECTA and EPF1–ERL1 ligand–receptor pairs, we treated seedlings with bioactive MEPF1 and MEPF2 (Fig. 5F–O). MEPF1 application to wild type, epf1, epf2, and ERECTAΔK in erecta conferred arrested meristemoids (Fig. 5F–H,J), indicating that MEPF1 is triggering a cellular response. Consistent with the separate roles of EPF1 and EPF2, upon MEPF1 application, seedlings of epf2 and ERECTAΔK in erecta exhibited additive phenotypes: excessive asymmetric divisions, none of which differentiate into stomata (Fig. 5G,H). In contrast, seedlings of ERL1ΔK in erl1 developed paired stomata even in the presence of MEPF1 (Fig. 5I), indicating that ERL1ΔK confers insensitivity to MEPF1 application. Similarly, seedlings of ERECTAΔK in erecta are insensitive to MEPF2 application and differentiate stomata (Fig. 5L), while seedlings of all other genotypes (wild type, epf1, epf2, and ERL1ΔK in erl1) showed severe inhibition of asymmetric division and stomatal development (Fig. 5K,M–O). Based on these findings, we conclude that EPF1 and EPF2 act as specific, upstream signaling ligands for ERL1- and ERECTA-mediated signal transduction, respectively.

Plant materials

Arabidopsis thaliana Columbia (Col) accession was used as wild type. Mutants and transgenic lines used in this study are in the Col background unless otherwise specified. The following mutants and transgenic lines have been described previously: er-105, erl1-2, and er-105 erl1-2 erl2-1 (Shpak et al. 2004, 2005); epf1 and tmm-KO (Hara et al. 2007); epf2 (Hara et al. 2009); and pNLB115 (proERECTA::ERECTA-3xFlag) in er-105 (Uchida et al. 2011). Each transgene was introduced into its respective mutant background by either Agrobacterium-mediated transformation or genetic crosses. Arabidopsis seeds were surface-sterilized with bleach solution (33% bleach solution, 0.01% Triton X-100) and grown on Murashige-Skoog (MS) medium. Two-week-old seedlings were transplanted onto soil and grown as described previously (Shpak et al. 2003). N. benthamiana was grown in a temperature-controlled growth room maintained with a 16-h light/8-h dark cycle at 25°C. Three-week-old to 4-wk-old plants were used for experiments.

Plasmid construction and generation of transgenic plants

The following constructs were generated and used in this study: pMK111 (proERL1::3xFlag), pMK110 and pMK210 (proERL1::ERL1-YFP), pMK112 and pMK212 (proERL1::ERL1-3xFlag), pJG102 (proTMM::TMM-3xHA), pJM125 (proTMM::TMM), pJM251 (proTMM::TMM-YFP), pJM122 (TMM cDNA), pJM140 (ERL1ΔK), pJM154 and pJM254 (proERECTA::ERECTAΔK-YFP), pJM159 and pJM259 (proERL1::ERL1ΔK-CFP), pJM184 and pJM284 (proERECTA::ERECTA-YFP), pTK102 (proEst::EPF1), pTK103 (proEst::EPF2), pTK127 (EPF1 cDNA), pTK128 (EPF2 cDNA), pMK334 (LURE2 cDNA), pTK146 (ERECTAΔK), pJSL55 (CaMV35S::ERECTAΔK-GFP), pJSL56 (CaMV35S::ERL1ΔK-GFP), pJSL57 (CaMV35S::TMM-GFP), pJSL51 (EPF1-3xFlag), pJSL52 (EPF2-3xFlag), pJSL77 (LURE2-3xFlag), pJSL59 (CaMV35S::EPF1-3xFlag), pJSL60 (CaMV35S::EPF2-3xFlag), pJSL81 (CaMV35S::LURE2-3xFlag), pJSL11 (pBAD::MEPF1-6xHis), and pJSL68 (pBAD::MEPF2-6xHis). Genomic clones of ERECTA and ERL1 were used for all epitope-tagged constructs. Stable transgenic plants were generated using Agrobacterium-mediated transformation by the floral-dip method (Clough and Bent 1998). The epitope-tagged receptors were transformed into their respective knockout (null) mutant backgrounds to confirm their functionality via complementation. For all transgenic Arabidopsis lines, >20 T1 plants per construct were subjected to phenotypic characterization. Two to three lines were selected based on single insertion status inferred by the segregation of resistance genes and stability of the phenotype in subsequent generations. For information on the plasmid construction and primers used in this study, see Supplemental Tables 1 and 2.

Histochemical analysis and microscopy

For confocal microscopy, seedlings were treated with propidium iodide (PI) (Sigma) to highlight cell periphery (excitation, 555 nm; emission, 570–610 nm) using a Zeiss LSM700 with a 20× objective lens. Histochemical staining of epidermis using toluidine blue O (TBO) (Sigma) was performed as described in Hara et al. (2009). Quantitative analysis of stomatal phenotypes was performed using TBO-stained epidermal samples as described previously (Guseman et al. 2010).

Preparation of membrane proteins, co-IP, and protein gel immunoblot analysis

For co-IP assays, Arabidopsis plants expressing functional, epitope-tagged receptor proteins that complement the respective mutant phenotypes were used as starting materials. Twelve-day-old Arabidopsis seedlings were ground to fine powder in liquid nitrogen and solubilized with 3× (w/v) extraction buffer (100 mM Tris-HCl at pH 8.8, 150 mM NaCl, 1 mM EDTA, 20% glycerol, 20 mM NaF, 1 mM PMSF, 1:1000 Complete protease inhibitor cocktail [Sigma]). The extracts were sonicated on ice and ultracentrifuged at 100,000g for 30 min at 4°C to obtain the membrane fraction as precipitate. The pellet was resuspended in membrane solubilization buffer (100 mM Tris-HCl at pH 7.3, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 1% Triton X-100, 20 mM NaF, 1 mM PMSF, 1:1000 Complete protease inhibitor cocktail) to release membrane proteins. The solution was sonicated on ice and ultracentrifuged again at 100,000g for 30 min at 4°C, and the supernatant was incubated with anti-GFP (ab290, Abcam) or anti-HA (Sigma) overnight at 4°C. Immune complexes were bound to protein G-agarose beads (Pierce) for 1.5 h with slow rotation at 4°C, followed by washing in IP buffer (25 mM Tris-HCl at pH 7.2, 150 mM NaCl) four times. The immunoprecipitates were eluted with 2× SDS sample buffer (200 mM Tris-HCl at pH 6.8, 8% SDS, 0.4% Bromophenol Blue, 40% glycerol, 10% β-mercaptoethanol) by boiling for 5 min. Either total membrane or immunoprecipitated proteins were separated on a SDS-PAGE gel and transferred to PDVF membrane (Millipore) for immunoblot analysis using monoclonal anti-Flag M2 (Sigma), mouse anti-GFP (Zymed Laboratory), or mouse anti-HA (12CA5, Roche) antibodies as primary antibodies. As secondary antibodies, goat anti-mouse IgG horseradish peroxidase-linked antibodies (GE Healthcare) were used at a dilution of 1:50,000. The protein blots were visualized using Chemiluminescence assay kit (Thermo Scientific).

Transient protein expression in N. benthamiana by agroinfiltration

Agrobacterium tumefaciens strain GV3101 was transformed with the indicated expression clones and grown in YEB medium supplemented with kanamycin and hygromycin. Bacterial cultures were precipitated and resuspended in infiltration medium containing 10 mM MgCl₂, 10 mM MES (pH 5.6), and 150 μM acetosyringone. To enhance the transient expression in tobacco, cultures were mixed with an Agrobacterium culture that allows expression of the silencing suppressor p19 (a gift from Professor Sir David Baulcomb) (Voinnet et al. 2003), giving an optical density at 600 nm of 0.3 for each strain. The mixed bacterial suspensions were then infiltrated into young, but fully expanded, leaves of N. benthamiana plants using a needleless syringe. After infiltration, plants were grown further for 48–72 h at 25°C and collected for further biochemical assays.

Immunopurification of EPFs from N. benthamiana

Tobacco leaves expressing MEPF1-Flag or MEPF2-Flag were ground in liquid nitrogen and homogenized in extraction buffer (100 mM HEPES at pH 7.5, 5 mM EDTA, 5 mM EGTA, 150 mM NaCl, 2 mM Na₃VO₄, 2 mM NaF, 1 mM PMSF, 1 tablet/50 mL extraction buffer of protease inhibitor mixture [Roche Applied Science], 10% glycerol, 7.5% [w/v] PVPP). After centrifugation at 10,000g for 15 min at 4°C, the supernatant was filtered through a 0.45-μm filter. The filtrate was then incubated with prewashed anti-Flag M2 affinity gel (Sigma-Aldrich) overnight at 4°C, washed extensively, and eluted by either competition with Flag peptide (Sigma-Aldrich) or acid elution with glycine. The immunopurified EPF1 and EPF2 were subjected to bioassays.

Chemical induction of transgenes

Transgenic Arabidopsis seedlings carrying estradiol-inducible EPF1 and EPF2 were germinated on MS medium supplemented with 5 μM estradiol (Sigma). Induction of EPF1 and EPF2 gene expression was confirmed by RT–PCR. The induction was confirmed by observing the epidermal phenotypes of cotyledons and rosette leaves using a confocal microscope.

Expression, purification, refolding, and bioassays of recombinant MEPF peptides

Recombinant LURE2 peptide was prepared according to Okuda et al. (2009). For preparation of recombinant MEPF1 and MEPF2, the following modifications were employed. Coding sequences of predicted MEPF1 and MEPF2 were amplified from the cDNA of Arabidopsis and cloned into the vector pBAD/gIII A (Invitrogen) to fuse a poly-histidine (His) tag to the C terminus. Peptides were expressed in E. coli overnight at 37°C and purified by a Ni column (His-Trap FF, GE Healthcare). The recombinant peptides were dialyzed (Mini dialysis kit, MWCO:1,000, GE Healthcare) and refolded for 4 d at 4°C using glutathione (reduced and oxidized forms; Wako) and L-arginine ethyl ester dihydrochloride (Sigma). For heat denaturation experiments, refolded peptide solution was incubated for 30 min at 70°C. For bioassays, either buffer alone (50 mM Tris-HCl at pH 8.0) or refolded recombinant MEPF peptides in buffer were applied to 1-d-old Arabidopsis plants that had germinated on MS medium. After 5 d of further incubation in MS liquid medium containing each peptide, stomatal phenotypes of abaxial cotyledons were determined by inspection with a confocal microscope.

HPLC purification and MALDI-TOF mass spectrometry analysis

Affinity column-purified MEPF1 and MEPF2 were further purified on reverse-phase HPLC (Waters Delta Prep 3000 HPLC) to assess MEPF1 and MEPF2 conformational isomers. The solvents used were solvent A, 0.1% TFA in DI water; and solvent B, 75% acetonitrile, 1 mL/min flow rate using the following gradient program: 0–2 min in an isocratic flow (0% B), 2–102 min in a linear gradient of 1% per minute, and 102–117 min in an isocratic flow (0% A). The separated peaks were collected by autothreshold collection, and the peaks' contents were identified by MALDI-TOF mass spectrometry on an Autoflex II mass spectrometer in positive ion mode (Bruker Daltonics) using 2:1 α-cyano-4-hydroxycinnamic acid and 2,5-dihydroxy-benzoic acid matrix. The collected HPLC-purified MEPF1 and MEPF2 peptide peaks were freeze-dried, then redissolved to appropriate concentration in MS-0 medium, and used for bioassays.

Preparation of biosensor surfaces

Gold-coated QCM and SPR slides of the specifications described below were used after cleaning by sequential sonication in acetone, ethanol, and water (three times per solvent, 5 min each) followed by UV ozone cleaning for 20 min. The GFP-tagged receptors were immobilized on clean chip surfaces via anti-GFP antibody linkage. First, the surfaces were drop-incubated with 1:50 dilution of anti-GFP solution (ab290, Abcam) in phosphate buffer (pH 7.4) in a humidity-controlled environment overnight at 4°C and rinsed 10 times by dilution rinse with phosphate buffer. The surfaces were then blocked in a 2% bovine serum albumin (BSA) solution for 2 h and washed by dilution as before. The resulting samples were incubated with a membrane fraction of either control (GFP only) or receptor solution for either 4 h (QCM) or 2 h (SPR), followed by the dilution rinse and an additional 1 h of BSA blocking to yield receptor-immobilized biosensor chips. All chips were used on the day they were produced.

QCM

QCM measurements were performed using QCM-Z500 (KSV Instruments) and commercially available AT-cut polished QCM crystals with a fundamental resonant frequency of 4.95 MHz (International Crystal Manufacturing Co.). The sensor surface was precoated by evaporation with a 100 Å titanium adhesion layer, followed by 1000 Å of gold. The QCM crystal dimensions used were 1.397 cm (0.550 in) blank diameter and 0.5 cm and 1.176 cm (0.440 in) wraparound electrode diameters. The receptor-modified QCM crystal (described above) was placed in the holder and sealed with two O-rings. In order to establish a stable baseline, a sufficient amount of phosphate buffer solution was injected to fill the QCM chamber, and the frequency shifts were monitored overnight. Then, sequential serial concentrations of recombinant MEPF1 or MEPF2 peptides in phosphate buffer solution were introduced to the QCM chamber, and the frequency change was recorded continuously until no further change was observed, indicating equilibrium. The QCM resonant frequency shifts recorded at the seventh harmonic were used to analyze ligand–receptor interactions, since the observed resonant QCM frequency decreases proportionally to the mass of adsorbed molecules following the Saurbrey equation

where f₀ is the frequency of the fundamental mode of the crystal, Δm is the mass change per unit area, A is the piezoelectrically active area, ρ_q is the density of quartz, and μ_q is the shear modulus of quartz. All experiments were performed at 4°C in stop-flow mode in three independent experiments.

SPR

SPR measurements were made on a four-channel instrument (Kretschmann configuration: Radio Engineering Institute, Czech Republic) equipped with a polychromatic light source (Ocean Optics LS1). The slides were prepared via thermal evaporation by coating a BK-7 glass slide with 20 Å of chromium as the adhesion layer and 480 Å of gold. The instrument can detect changes at a level of 0.0001 refractive index unit and is temperature controlled (10°C–55°C). Buffer and recombinant MEPF solutions were degassed to avoid bubble formation in the flow cell. First, phosphate buffer solution was flowed over the surface until a stable baseline signal was established (<0.05 nm change over 5 min). Then, recombinant MEPF solutions in phosphate buffer at concentrations indicated were flowed over the surface, and the adsorption was monitored. The temperature within the flow cell of the SPR was kept at a constant 25°C via a heating element and a cooling fan controlled by a temperature controller. All of the solutions used were introduced to the flow cell at a rate of 140 μL/sec. The Langmuir isotherm model was used to calculate the kinetics of the adsorption process using the experimental data obtained from SPR experiments as described previously (Tamerler et al. 2006), by fitting the following equation to the adsorption data:

where Δn is the plasmon resonance peak shift in nanometers, n_max is the expected sift ath equilibrium, k_obs is the apparent rate of adsorption, and t is time in seconds.