The Interaction Between Endopolygalacturonase From Fusarium Moniliforme and PGIP from Phaseolus Vulgaris Studied by Surface Plasmon Resonance and Mass Spectrometry

A combination of surface plasmon resonance (SPR) and matrix-assisted laser-desorptionionization- time-of-flight mass spectrometry (MALDI-TOF-MS) was used to study the interaction between endopolygalacturonase (PG) from Fusarium moniliforme and a polygalacturonase-inhibiting protein (PGIP) from Phaseolus vulgaris. PG hydrolyses the homogalacturonan of the plant cell wall and is considered an important pathogenicity factor of many fungi. PGIP is a specific inhibitor of fungal PGs and is thought to be involved in plant defence against phytopathogenic fungi. SPR was used either to study the effect of the PG glycosylation on the formation of the complex with PGIP, and as a sensitive affinity capture of an interacting peptide from a mixture of PG fragments obtained by limited proteolysis. Mass spectrometry allowed to characterise the interacting peptide eluted from the sensor surface.


Introduction
Surface plasmon resonance (SPR) biosensors are important and versatile tools for studying proteinprotein interactions; they allow to measure interactions in real time and require very little material, which usually does not need any chemical modification. SPR technology coupled with mass spectrometry can be also used to identify and characterise proteins eluted from sensor surfaces (Sö nksen et al., 2000;Williams, 2000). Once a sample containing a mixture of possible ligands is passed over the sensor surface, and binding is detected by the SPR signal, the identification of interacting proteins at the femtomole level is made possible by the use of sensitive mass spectrometers and advanced database searching algorithms (Nelson et al., 2000).
We are studying the interaction between the endopolygalacturonase (PG) from the phytopathogenic fungus Fusarium moniliforme with PGIP-2 of Phaseolus vulgaris. PGs catalyse the fragmentation and solubilisation of homogalacturonan in the plant cell wall and play an important role during pathogenesis. Polygalacturonase-inhibiting proteins (PGIPs), present in the cell wall of many plants, form specific complexes with fungal PGs and favour the accumulation of oligogalacturonides able to elicit plant defence responses (Cervone et al., 1997). PGIPs belong to a super-family of leucine-rich repeat (LRR) proteins. In a previous study, we showed that F. moniliforme polygalacturonase and PGIP-2 specifically interact with high affinity (K D = 47 nM), and we demonstrated that the residues in the predicted b-strand/b-turn motif of PGIP are critical for its affinity and specificity for the PG ligands (Leckie et al., 1999). The residues of PG involved in the interaction with PGIP are still unknown.
One of the best characterised members of the LRR superfamily is ribonuclease inhibitor (RI), that utilises a large set of interactions to achieve tight binding to different members of the RNase family (Kobe et al., 1996;Papageorgiou et al., 1997).
In analogy with what is known for RI complexes, it is likely that a network of contacts occur between PG and PGIPs. The aim of this study was to gain information on which domain of PG contains the PGIP interacting residues and whether the glycosylation of the PG molecule has any effect on the interaction.

Materials and methods
Protein purification PGIP-2 was purified from Nicotiana benthamiana plants infected with PVX as previously described .
FmPG expressed in yeast was prepared and purified as previously described .

Surface plasmon resonance
The interaction between FmPG and immobilised PGIP-2 was measured in real time by surface plasmon resonance using BIACORE X2 equipment (BIACORE AB, Uppsala, Sweden). Protein interaction analyses were performed on research grade BIACORE CM5 sensor chips. For the immobilisation of PGIP-2 the sensor chip was activated by injection of 35 ml of 1 : 1 mixture of N-ethyl-Nk-(3-diethylaminopropyl)carbodiimide and N-hydroxysuccinimide at 5 ml/min flow rate. Running buffer used during the immobilisation procedure was HBS (10 mM Hepes, pH 7.4, 150 mM NaCl, 0.005% [v/v] surfactant P20 from BIA-CORE, in distilled water). 40 ml of bean PGIP-2 at 100 ng/ml in 10 mM sodium acetate, pH 4.5, were injected over the sensor chip, followed by 35 ml of 1 M ethanolamine hydrochloride to block the remaining ester groups. A total of 4000-5000 RU (Resonance Units, which are proportional to the mass of protein bound on the surface of the chip) of PGIP-2 were immobilised onto the sensor chip surfaces, corresponding to a density of 4-5 ng/mm 2 . A second flow-cell of the sensor chip was treated in the same way, except that buffer was injected instead of PGIP-2, and this surface was used as the reference flow-cell. In the interaction assays, PG solutions in 25 mM ammonium acetate buffer pH 5.0 were injected onto the PGIP-2 sensor chip at a flow rate of 5 ml/min. Limited proteolysis of FmPG was performed by incubating the native enzyme with Endoproteinase Lys C (Boehringer Mannheim GmbH, Frankfurt, Germany) using an enzyme/substrate ratio of 1 : 50 (w/w) in ammonium bicarbonate buffer pH 7.5 at 37uC; aliquotes taken at different times were analysed.

Elution from the sensor chip
Elution from the sensor chip was performed essentially as described by Sö nksen et al. (Sö nksen et al., 1998)with the following modifications: 10 ml of NH 4 HCO 3 buffer pH 8 were used to elute most of the bound PG. The elution was performed at the beginning of the dissociation phase, without rinsing the system. The eluted sample was subjected to purification from salts and preparation for MALDI analysis by using reversed-phase nano-columns prepared as described by Gobom et al. (Gobom et al., 1999).

Mass spectrometry
MALDI mass spectra were acquired on a Voyager-Elite MALDI-TOF (Applied Biosystems, MA) mass spectrometer, operated in positive ion linear mode.

Results and discussion
PG from F. moniliforme (FmPG) contains four potential N-linked glycosylation sites (Caprari et al., 1993). The enzyme used for this study is FmPG expressed in Saccharomyces cerevisiae. When analysed by SDS-PAGE, FmPG showed three protein bands with molecular masses of 43, 46 and 50 kDa, respectively, corresponding to different glycoforms of the same polypeptide chain . The calculated molecular mass of the polypeptide is 36.2 kDa, indicating that the enzyme purified from yeast is heavily glycosylated with a carbohydrate content ranging from 16% in the lightest glycoform to 28% in the heaviest one. As shown in the MALDI spectrum of a glycopeptide arising from the tryptic digest of PG, carbohydrate microheterogeneity gives rise to multiple peaks with mass differences corresponding to the carbohydrate residues ( Figure 1). Up to 14 Man residues are present in the high-mannose type glycan chain typical of the yeast-secreted proteins. To investigate whether the glycosylation of the enzyme is involved in the binding of FmPG to PGIP-2, PG deglycosylation was performed using endo-H, which leaves a single GlcNAc residue present at each of the occupied Asn residues. The enzymatic deglycosylation does not affect the proper folding of the enzyme which maintains activity and produces an oligogalacturonide profile comparable to that produced by the wild type enzyme (Bergmann et al., 1996). Native PG and PG treated with endo-H were passed over a sensor chip with immobilised PGIP-2 ( Figure 2). The sensorgram in Figure 2A shows that the deglycosylated enzyme binds to the inhibitor. Interestingly, the affinity for the deglycosylated enzyme is higher than that of the glycosylated form, with a difference in the equilibrium response of ca. 1000 RU. We concluded that not only the glycosylation of FmPG is not required for binding to PGIP, but also that the presence of the glycans might sterically reduce the number of contacts between the two proteins.
A modification of the method developed by Sö nksen et al., (1998) was used for the recovery of the affinity-bound enzyme from the sensor surface for mass spectrometric analysis ( Figure 2B). The amount of protein eluted from the sensor chip was calculated to be ca. 40 fmol, as based on the molecular weight of PG. The protein eluted from the sensor surface showed an average molecular mass of 37110 Da, in good agreement with the mass calculated for the polypeptide (36198 Da) plus four GlcNAc residues attached at the four potential N-glycosylation sites.
In order to locate the domain of PG recognised by the inhibitor, a peptide mixture was prepared by limited proteolysis of the native enzyme with Endoproteinase Lys-C (Figure 3). The peptide mixture derived by treatment with Endoproteinase Lys-C was passed in flow over the sensor surface with immobilised PGIP-2, and over a reference flow-cell as a control. One peptide with m/z 6479.9 was specifically recovered from the flow-cell with immobilised PGIP-2 ( Figure 3B). In addition to the  peak at m/z 6479.9, a non-specific peak at m/z 4818.6, also present in the fraction eluted from the reference flow-cell, was detected ( Figure 3B and C).
The last peak is not present in the original mixture of peptides ( Figure 3A) and is probably a contaminant that comes from the elution procedure. The peak at m/z 6479.9 corresponds to the PG fragment spanning from amino acid 181 to amino acid 244 comprising several residues of FmPG which are conserved in all the known fungal PGs and form the active site of this class of enzymes (Armand et al., 2000). Among them Asp191, Asp212, Asp213 correspond to residues that in PGII from Aspergillus niger (AnPGII) have been shown to be involved in catalysis, and His 234 corresponds to a residue that in AnPGII is thought to play an indirect role in catalysis (Armand et al., 2000). Recently we reported the crystal structure of FmPG, that allows us to locate the peptide 181-244 within the putative active site cleft of the enzyme (Figure 4). By site-directed mutagenesis and SPR analysis, we have demonstrated that several amino acids of the active site and His188, at the edge of the active site cleft, are critical for the formation of the complex with PGIP . Our isolation of the peptide 181-244 which has a strong capacity of interaction with PGIP-2, confirms that most of the residues critical for the interaction are located within this region.