Characterization of a tomato protein that inhibits a xyloglucan-specific endoglucanase


For correspondence (fax +1 706 542 4412; e-mail


A basic, 51 kDa protein was purified from suspension-cultured tomato and shown to inhibit the hydrolytic activity of a xyloglucan-specific endoglucanase (XEG) from the fungus Aspergillus aculeatus. The tomato (Lycopersicon esculentum) protein, termed XEG inhibitor protein (XEGIP), inhibits XEG activity by forming a 1 : 1 protein:protein complex with a Ki ≈ 0.5 nm. To our knowledge, XEGIP is the first reported proteinaceous inhibitor of any endo-β-1,4-glucanase, including the cellulases. The cDNA encoding XEGIP was cloned and sequenced. Database analysis revealed homology with carrot extracellular dermal glycoprotein (EDGP), which has a putative role in plant defense. XEGIP also has sequence similarity to ESTs from a broad range of plant species, suggesting that XEGIP-like genes are widely distributed in the plant kingdom. Although Southern analysis detected only a single XEGIP gene in tomato, at least five other XEGIP-like tomato sequences have been identified. Similar small families of XEGIP-like sequences are present in other plants, including Arabidopsis. XEGIP also has some sequence similarity to two previously characterized proteins, basic globulin 7S protein from soybean and conglutin γ from lupin. Several amino acids in the XEGIP sequence, notably 8 of the 12 cysteines, are generally conserved in all the XEGIP-like proteins we have encountered, suggesting a fundamental structural similarity. Northern analysis revealed that XEGIP is widely expressed in tomato vegetative tissues and is present in expanding and maturing fruit, but is downregulated during ripening.


Cell walls play an important role in the ability of plants to defend themselves against fungal pathogens (Côté and Hahn, 1994; Howard, 1997). The main components of primary cell walls are members of two polysaccharide networks, one consisting of cellulose and hemicellulose, and the other consisting of pectic polysaccharides. Most plant pathogens secrete a mixture of enzymes that can hydrolyze the polysaccharides in the primary cell wall (Walton, 1994). These enzymes include various polygalacturonases, pectin methyl esterases, pectin/pectate lyases, acetyl esterases, xylanases, and a variety of endoglucanases that cleave cellulose, xyloglucan, and other glucans.

Plants synthesize a variety of proteins that inhibit the cell-wall-degrading enzymes secreted by pathogens. The best characterized of these proteins are polygalacturonase-inhibitor proteins (PGIPs), which were first described more than 30 years ago (Albersheim and Anderson, 1971). PGIPs act by forming specific, reversible, and saturable complexes with microbial polygalacturonases (PGs), which cleave unesterified homogalacturonan (poly α-1,4-linked-d-galactosyluronic acid), the major pectic polysaccharide in primary cell walls (De Lorenzo et al., 2001; Stotz et al., 2000). The limitation of fungal colonization in transgenic tomato plants expressing pear PGIP is direct evidence of PGIP's role in defense (Powell et al., 2000). Sugar beet produces a protein pectin lyase inhibitor protein (PNLIP) that inhibits a pectin lyase from Rhizoctonia solani, which is a pathogen of sugar beet (Bugbee, 1993). Xylanase inhibitors have been detected in wheat flour (Debyser et al., 1999; Rouau and Surget, 1998), purified (McLauchlan et al., 1999), and characterized (Elliott et al., 2002; Gebruers et al., 2001). A kiwi fruit protein (Camardella et al., 2000) with homology to plant invertase inhibitors has been reported to inhibit tomato pectin methylesterase by forming a 1 : 1 complex (Balestrieri et al., 1990; Giovane et al., 1995).

Plants also produce enzymes that are capable of degrading the cell walls of pathogens. For example, the plant enzyme endo-β-1,3-glucanase (EGase) degrades β-1,3/1,6-glucans in fungal cell walls, releasing β-glucan oligosaccharide elicitors, which induce numerous plant defense responses (Côté and Hahn, 1994; Shibuya and Minami, 2001). Some plant pathogens have recently been shown to synthesize a family of proteins that inhibits plant EGases. For example, a glucanase inhibitor protein (GIP-1) has been purified from the oomycete pathogen of soybean, Phytophthora sojae (Psg; Ham et al., 1997), cloned, and characterized (Rose et al., 2002). GIP-1, which binds with high affinity to a soybean EGase, is homologous to serine proteases but lacks the catalytic triad required for protease activity.

Here, we describe the identification and molecular characterization of xyloglucan-specific endoglucanase inhibitor protein (XEGIP), the first plant protein that has been shown to inhibit an endoglucanase. Endoglucanase inhibition by XEGIP represents a new class of protein–protein interactions whose discovery may have important implications with regard to plant pathogenesis and plant–microbe co-evolution.


Identification and isolation of XEGIP from tomato cell culture

A protein (XEGIP) that inhibits the hydrolytic activity of a xyloglucan-specific β-1,4-endoglucanase (XEG) isolated from Aspergillus aculeatus (Pauly et al., 1999) was detected in the culture medium of suspension-cultured tomato cells. The filtered culture medium was mixed with 4 volumes of ethanol, and the resulting precipitate (containing xyloglucan) was re-suspended in water, dialyzed, and lyophilized. When the xyloglucan-enriched material was treated with XEG, none of the expected oligosaccharide products were detected by matrix assisted laser desorption – time of flight – mass spectrometry (MALDI-TOF-MS) analysis of the reaction mixture even after extensive treatment with an excess of XEG. The absence of XEG-generated oligosaccharides was confirmed by a colorimetric assay using p-hydroxybenzoic acid hydrazide (Lever, 1972) to detect the reducing ends of any oligosaccharides produced by the XEG. In contrast, the digestion proceeded to completion, and the reducing oligosaccharides were easily detected when the crude xyloglucan preparation was boiled for 5 min before adding XEG. The heat-labile nature of the inhibitory factor suggested that it was a protein.

A protein whose presence is correlated with the inhibitory activity was purified from the ethanol-precipitated material by ion-exchange and size-exclusion chromatography (Figure 1, lanes 1–4, described in Experimental procedures). The molecular weight of the purified protein was approximately 50 kDa as determined by SDS–PAGE (Figure 1, lane 4). MALDI-TOF-MS established that the protein had a molecular weight of 50 853 Da (data not shown). The amount of this protein in the tomato cell culture medium appears to depend on the culture conditions, constituting approximately 2% of the ethanol-precipitated protein from one culture, but up to 10% of the protein in other cultures, as estimated by SDS–PAGE.

Figure 1.

SDS–PAGE of fractions obtained during purification of XEGIP (see Experimental procedures for details).

Lane M, molecular weight standards (208, 115, 79, 49.5, 35, 28, 20, and 7 kDa). Lane 1, ethanol-precipitated material from the medium of suspension-cultured tomato cells. Lane 2, non-bound fraction from the Q-Sepharose anion-exchange column. Lane 3, XEGIP-active fractions that were eluted from the HiTrap-S cation-exchange column during a pH and salt gradient. Lane 4, XEGIP-active fraction from the Superdex-75 column.

Various amounts of the purified 51 kDa protein were mixed with XEG, and the resulting mixtures were added to xyloglucan solutions. Aliquots of each reaction mixture were taken after 10 min, 30 min, and 24 h, and the extent of xyloglucan digestion was determined using the reducing sugar assay. The results (Figure 2) showed that the XEG-catalyzed reaction was completely inhibited in the presence of an excess of the purified 51 kDa protein, even when the incubation time was extended to 24 h. Accordingly, the 51 kDa protein was named XEGIP.

Figure 2.

Inhibition of XEG activity by increasing amounts of XEGIP.

The response of the PAHBAH assay for reducing sugars (see Experimental procedures), which corresponds to the amount of the hydrolysis product generated by XEG, is plotted versus the amount of XEGIP. The linear range of the assay extends to an absorption of about 0.75. Data points with higher absorption values underestimate the amount of reaction product. The amount of XEGIP required to completely inhibit XEG activity is independent of the reaction time. When slightly less than one equivalent of XEGIP is added, significant XEG-catalyzed hydrolysis is detected when the reaction time is extended, consistent with a non-proteolytic mechanism of inhibition.

Stoichiometry of the XEGIP–XEG interaction

Xyloglucan-specific endoglucanase inhibitor protein inhibits the activity of XEG by binding to the enzyme. Binding was first detected as a shift in the retention time of XEG and XEGIP during size-exclusion chromatography (SEC) on Superdex-75 (Figure 3). A constant amount of XEG was mixed with increasing amounts of XEGIP, and the resulting mixtures were subjected to SEC. Increasing the XEGIP concentration resulted in the appearance and growth of a higher molecular mass peak and the disappearance of the XEG peak. The newly formed peak was collected and analyzed by SDS–PAGE (data not shown). The resulting gels consistently contained two bands, corresponding to XEG and XEGIP, irrespective of the length of time for which the enzyme and inhibitor were pre-incubated before they were subjected to SEC. The molar extinction coefficients (ɛ) for XEG and XEGIP were determined (see Experimental procedures) allowing the XEGIP:XEG ratio to be measured for each mixture applied to the column. Ideally, if a 1 : 1 complex forms with a dissociation constant (Kd) that is significantly smaller than the concentration of XEG, the XEG peak should disappear just when the XEGIP:XEG ratio in the mixture is 1 : 1. When the XEGIP:XEG ratio is greater than 1, a (free) XEGIP peak should be visible as a shoulder on the XEG–XEGIP complex peak. Within the accuracy of our estimations of ɛ for XEG and XEGIP, these are the results obtained (Figure 3). Taken together, the appearance of a high-molecular weight complex peak, the disappearance of the XEG peaks when the XEGIP:XEG ratio is greater than or equal to 1, and the absence of detectable proteolytic fragments support the hypothesis that XEGIP inhibits XEG by binding to XEG to form a 1 : 1 complex and that XEGIP does not proteolytically degrade XEG.

Figure 3.

Interaction of XEG and XEGIP as determined by SEC on Superdex-75.

Pure XEG (A), pure XEGIP (B) and mixtures of the two with varying ratios (C–F) were chromatographed. Increasing the XEGIP:XEG ratio resulted in a decrease in the concentration of free XEG and in the appearance of an XEG-XEGIP complex (C–F). When the XEGIP:XEG ratio reached 1 : 1, the XEG peak completely disappeared (E). When sufficient XEGIP was added, the XEGIP:XEG ratio became greater than 1 (F) and a free XEGIP peak appeared as a shoulder on the XEG–XEGIP complex peak. Profile G is a computer-generated simulation (*) obtained by linear combination of profiles B (free XEGIP) and E (the saturated complex), taking into account the absolute amount of protein used to obtain each of these profiles (0.75 × B + 1.0 × E). This simulated profile should be identical to profile F if XEG and XEGIP combine to form a 1 : 1 complex. The agreement of profiles F and G confirms the identification of free XEGIP as a shoulder on the complex peak and supports the hypothesis of a tightly bound 1 : 1 complex.

Additional evidence for the binding of XEGIP and XEG was obtained using an affinity chromatography matrix consisting of XEG immobilized on CNBr-activated Sepharose. Neutral and anionic polysaccharides and proteins were removed from the ethanol-precipitated materials from tomato cell culture media by cation-exchange chromatography. XEGIP and other cationic molecules that had bound to the ion-exchange matrix (Figure 4, lane 1) were eluted with a high-salt buffer and applied to the XEG column. Material that bound to the XEG column was eluted with 2 m imidazole/HCl (pH 7) and analyzed by SDS–PAGE (Figure 4, lane 4), revealing a single protein, which was identified as XEGIP by its electrophoretic mobility and whose inhibitory activity was recovered upon removal of the imidazole by dialysis. The immobilized XEG in the regenerated affinity column retains its ability to digest xyloglucan and can be reused to bind XEGIP (data not shown), thus confirming that the inhibition is not due to proteolysis.

Figure 4.

SDS–PAGE of affinity chromatography fractions.

Lane 1, cationic proteins prepared by applying the material obtained by ethanol precipitation of tomato cell culture medium (see Experimental procedures) to a Hi-Trap S cation-exchange column, washing the column with the starting buffer, and then eluting it with the starting buffer containing 2 m NaCl. Lane 2, non-bound fraction obtained by applying cationic proteins (shown in lane 1) to an XEG column and washing with the starting buffer. Lane 3, proteins that were eluted from the loaded XEG column by the starting buffer containing 2 m NaCl. Lane 4, XEG-bound proteins that were eluted from the XEG column with 2 m imidazole (pH 7).

Kinetic studies of XEGIP–XEG binding

An attempt was made to determine the dissociation constant (Kd) for the interaction of XEGIP with XEG by surface plasmon resonance (SPR) using a BiaCore® apparatus. Toward this goal, purified XEGIP was covalently immobilized on a BiaCore® sensor chip. The sensogram obtained by passing soluble XEG over the XEGIP-coated chip indicated that the interaction of XEG with XEGIP was essentially irreversible under the experimental conditions used. Thus, the value of Kd could not be obtained by this method.

Most standard methods for determining the inhibition constant Ki are based on the assumption that formation of the enzyme–inhibitor complex does not significantly decrease the concentration [I] of the free inhibitor relative to the total concentration [I]t of added inhibitor (Segel, 1975). That is, [I] is assumed to be equal to [I]t. However, this is not the case when XEGIP is mixed with XEG at a concentration normally used to determine enzyme activity. As illustrated in Figure 3, when the XEGIP:XEG ratio is less than 1, essentially all the added XEGIP forms a complex with the XEG, and [I] is much less than [I]t. Free XEGIP cannot be detected until the XEGIP:XEG ratio is greater than 1, at which time the XEG is completely saturated and the rate of the XEG-catalyzed reaction is zero. A graphical method specifically designed to determine the association constants of such tightly bound inhibitors has been proposed by Dixon (1972) and described by Segel (1975). This analysis is carried out at low protein concentrations such that the enzyme–inhibitor interactions are minimized and the equilibrium between complex formation and disassociation can be characterized, that is, at concentrations where more than half of the inhibitor is free and the amount of free enzyme is sufficient to detect its activity. In the case of the XEGIP–XEG interaction, this corresponds to an XEG concentration of approximately 1 nm. As illustrated in Figure 5, Ki for the interaction of XEG and XEGIP was determined to be 0.5 nm by this method.

Figure 5.

Determination of Ki for the interaction of XEG with XEGIP (Segel, 1975).

The initial rate of the XEG-catalyzed depolymerization of tamarind xyloglucan (4 mg ml−1) was determined by the reducing sugar assay (A = 410 nm) and plotted versus the concentration of XEGIP. Straight lines are drawn connecting the intersection of the resulting curve with the y-axis (i.e. at V0) to points along the curve whose y-value equals V0/n, where n is an integer (2, 3, 4, 5, etc.). These diagonal lines are extended so as to cross the x-axis, and the distance between the x-intercepts corresponds to the value of Ki,app = Ki(1 + [S]/Km). The data shown indicate that Ki,app for the XEGIP:XEG interaction is 1.0 nm. Solving for Ki = Ki,app/(1 + [S]/Km) and using the previously determined value of 3.6 mg ml−1 for Km of tamarind xyloglucan as a substrate for XEG (Pauly et al., 1999) resulted in a Ki value of 0.5 nm.

Specificity of the XEGIP–XEG interaction

Xyloglucan-specific endoglucanase inhibitor protein did not show any capacity to inhibit other plant cell-wall-degrading enzymes tested. For example, fungal polygalacturonases (Cook et al., 1999), an endoglucanase (Megazyme, Wicklow, Ireland) from Trichoderma longibrachiatum that cleaves xyloglucan and other β-1,4-linked glucans, and a xyloglucan endotransglycosylase (LeXET2) that specifically uses xyloglucan as a substrate (Cataláet al., 2001) are not inhibited by XEGIP.

Cloning of XEGIP cDNA

Peptide sequences of tryptic fragments of XEGIP were determined by the Edman procedure (see Experimental procedures) and used to search the publicly available databases. Several homologous gene sequences were identified. Degenerate primers designed on the basis of well-conserved regions of these homologous sequences were used to generate several PCR products, which were cloned and sequenced (see Experimental procedures). Short nucleotide sequences within one of the PCR products corresponded to data obtained by Edman sequencing. Searches of the TIGR Gene Indices database ( with this cloned fragment revealed that it overlapped one Tentative Consensus sequence (TC85129) and one EST singleton (EST 258116 Accession number AI 777151). A new set of gene-specific primers were designed accordingly and used to clone the full-length cDNA of XEGIP. The cDNA sequence was deposited in GenBank (Accession number AY 155579). Its deduced amino acid sequence is shown in Figure 6.

Figure 6.

Alignment of the amino acid sequence of XEGIP with those of structurally related proteins.

Bg7S, basic globulin 7S protein from soybean; EDGP, extracellular dermal glycoprotein from carrot; conglutin, conglutin γ protein from Lupinus albus. Protein sequences deduced from tomato and Arabidopsis genes are also included. ClustalW ( and GeneDoc ( were used, respectively, to align and present the sequences. Identical residues are highlighted in black. Conserved cysteines are marked with C. The predicted signal peptide of XEGIP is underlined, and the cleavage site is marked by an arrow. Potential N-glycosylation sites are marked with X. Lines above XEGIP sequence indicate the tryptic fragments whose sequences were verified by Q-TOF mass spectroscopy.

The deduced amino acid sequence of the XEGIP gene product agrees closely with the results of quantitative amino acid analyses of purified XEGIP (data not shown) and with the results of mass spectral analysis of tryptic fragments of XEGIP. That is, the measured molecular weights of all 18 major peptide fragments detected by MALDI-TOF analysis of trypsin-treated XEGIP matched (within 0.1%) the theoretical values predicted for tryptic peptides generated in silico (data not shown). Eight of the trypsin-generated XEGIP peptides were also isolated by RP-HPLC and sequenced by MS/MS. All of the resulting peptide sequences (marked in Figure 6) matched the deduced amino acid sequence of XEGIP.

The deduced protein sequence has a putative signal peptide (22 aa) predicted by SignalP program ( The putative cleavage site is marked by an arrowhead in Figure 6. The predicted mature XEGIP protein has 415 amino acids and a molecular weight of 44229.74 Da. The mature protein also has five potential N-glycosylation sites, which are likely to account for some or all of the 6625 Da difference between the predicted molecular weight and the molecular weight measured by MALDI-MS (50 853 Da). Indeed, XEGIP binds to a Con-A lectin column and is eluted by methyl-α-d-mannoside (data not shown), confirming that XEGIP is a glycoprotein. Extracellular dermal glycoprotein (EDGP), the carrot homolog of XEGIP (see below), is also reported to be glycosylated (Satoh and Fujii, 1988).

Xyloglucan-specific endoglucanase inhibitor protein has 62% amino acid sequence identity to EDGP from carrot (Satoh et al., 1992) and homology to multiple EST sequences from tomato and diverse plant species including soybean, lotus, carrot, cotton, maize, rice, sorghum, Medicago, and Arabidopsis. (For each of these genes, the accession number and homology to the XEGIP sequence is listed at the end of this article.) In addition, XEGIP shares some conserved residues with basic globulin 7S protein (Bg7S) from soybean (35% identity) (Kagawa and Hirano, 1989), conglutin γ (Cγ) from Lupinus albus (35% identity), and some less related sequences from tomato and Arabidopsis. Eight of the 12 cysteines in XEGIP are generally conserved in all the XEGIP-like proteins illustrated in Figure 6, and all 12 of these cysteines are conserved in most of these proteins, suggesting similarities in their three-dimensional structures.

Gel blot analyses

Genomic DNA analysis identified a single XEGIP gene in tomato, regardless of whether a full- or partial-length probe was used (Figure 7), indicating that XEGIP is present as a single copy in tomato. However, five other XEGIP-related genes from tomato were identified in the databases. These genes encode proteins whose amino acid sequences are from 24% (partial sequence for tomato-2) to 51% (XEGIP-5) identical to those of XEGIP (Figure 6), which presumably are sufficiently divergent to preclude cross-hybridization in the Southern analysis. The two most closely related genes in Arabidopsis (Arabidopsis-1 and -2) share 58 and 63% amino acid sequence identity with XEGIP, respectively, and share 89% identity with each other (Figure 6).

Figure 7.

Genomic DNA analysis of XEGIP.

Genomic DNA (10 µg per lane) was digested with the indicated restriction enzymes, and the DNA gel blot was hybridized with the full-length XEGIP cDNA probe and washed with 0.2× SSC at 65°C. Molecular weight markers are indicated in kilobases.

Northern analysis revealed that XEGIP mRNA was expressed in all the vegetative tissues examined, with lower expression levels in young healthy leaves (Figure 8a). XEGIP mRNA abundance increased during fruit expansion, peaked immediately prior to the onset of ripening at the mature green stage, and declined as ripening progressed (Figure 8b).

Figure 8.

Analysis of XEGIP mRNA abundance in tomato tissues.

Total RNA gel blot analysis (15 µg RNA per lane) of XEGIP expression in (a) vegetative tissues and flowers, and (b) in tomato fruit spanning a development series. (Ov, ovary; I–III, expansion stages I–III; IG, immature green; MG, mature green; Br, breaker; Tu, turning; Pi, pink; LR, light red; RR, red ripe).


A plant protein that inhibits a xyloglucan-specific fungal endoglucanase (XEG) was purified from suspension-cultured tomato cells and termed XEG inhibitor protein (XEGIP). XEGIP, the first protein from plants or any other source that has been shown to inhibit an endo-β-1,4-glucanase, shows no detectable proteolytic activity and inhibits XEG via a strong 1 : 1 interaction with a Ki of approximately 5 × 10−10 m. The XEG–XEGIP complex has no detectable endoglucanase activity, but active XEG and XEGIP can be regenerated upon dissociation of the immobilized complex with 2 m imidazole buffer. The molecular basis of its interaction with XEG is yet to be determined, as XEGIP does not contain any recognizable protein–protein interaction motifs, such as leucine-rich repeats.

Xyloglucan-specific endoglucanase inhibitor protein appears to represent the newest class of plant-derived proteins that inhibit microbial enzymes that, in turn, degrade plant cell walls. Endoglucanases, which constitute a major class of these microbial enzymes, hydrolyze cellulose and xyloglucan and are implicated in pathogenicity. For example, endoglucanases of Ralstonia solanacearum (Denny et al., 1990; Kang et al., 1994; Roberts et al., 1988; Saile et al., 1997) and Clavibacter michiganensis (Jahr et al., 2000) are believed to be involved in the pathogenesis of tomato wilting disease. Cellulase (endoglucanase) CelV mutants of Erwinia carotovora exhibited delayed maceration of plant tissues, suggesting reduced virulence (Mae et al., 1995; Walker et al., 1994) and further indicating the involvement of endoglucanases in pathogenesis. In A. aculeatus, seven endoglucanases, including cellulase and XEG, have been identified (de Vries and Visser, 2001). Although A. aculeatus is a saprophyte and not considered to be a tomato pathogen, the spectrum of plant cell wall hydrolases that it secretes is likely to be similar to those of pathogenic fungi and bacteria. For example, the maize pathogen Cochliobolus carbonum secretes MLG2, a mixed-linked (β-1,3-β-1,4) glucanase into its culture medium. MLG2 and XEG both belong to Family 12 of the glycosyl hydrolases (Coutinho and Henrissat, 1999; Goedegebuur et al., 2002; Kim et al., 2001; Yuan et al., 2001), which include cellulases, endo-β-1,4-glucanases, and β-1,3-β-1,4-glucanase from fungi and bacteria. It remains to be seen whether the ability to interact with Family 12 enzymes is a general characteristic of XEGIP-related proteins. It is possible that XEGIP-related proteins have evolved to counteract cellulolytic and hemicellulolytic enzymes produced by plant pathogens, analogous to the inhibition of pathogenic polygalacturonases by plant PGIPs.

Alternatively, XEGIP may play a role in regulating endogenous plant enzymes, thereby affecting the modification and reorganization of cell walls during growth and development. However, we have demonstrated that the tomato xyloglucan endotransglucosylase LeXET2 (Cataláet al., 2001) is not inhibited by XEGIP. Except those that are closely related to the xyloglucan endotransglucosylases, all known plant endo-β-1,4-glucanases belong to Family 9 (Henrissat and Bairoch, 1996), and have an inverting rather than a retaining mechanism of action. None of the Family 9 enzymes have been tested for inhibition by XEGIP.

Extracellular dermal glycoprotein, the XEGIP homolog in carrot, is a 57 kDa protein that is expressed at a high level in the dermal tissues of roots, petioles, and leaves, as well as in developing seeds. EDGP has been proposed to be involved in pathogen resistance because of its localization in dermal tissues and expression in response to wounding (Satoh et al., 1992). However, its function has not been unambiguously established in vivo or in vitro. The sequence homology between EDGP and XEGIP (62% identity) suggests that they have similar functions; indeed, the ethanol precipitate of carrot cell culture medium contains a heat-sensitive component that inhibits XEG. A protein was purified (95% homogeneity) from the precipitated material and identified as EDGP by its apparent molecular weight (SDS–PAGE) and tryptic peptide-mass fingerprinting (data not shown). Peak shift experiments such as those shown in Figure 3 confirmed that EDGP binds to XEG (data not shown).

Xyloglucan-specific endoglucanase inhibitor protein and EDGP are both secreted into the medium of suspension-cultured cells that were derived from callus, which normally grows in response to wounding a plant tissue. As pointed out by Satoh et al. (1992), plant proteins that are typically expressed in response to wounding are also detected in suspension-cultured cells. These include invertase (Lauriere et al., 1988), hydroxyproline-rich glycoproteins (Brownleader and Dey, 1993; Esaka et al., 1992; Hirsinger et al., 1997; Lamport, 2001; Lamport and Northcote, 1960), chitinase (Arie et al., 2000; Kunze et al., 1998; Wojtaszek et al., 1998), β-1,3-glucanase (Kunze et al., 1998), and peroxidase (Breda et al., 1993; Schnabelrauch et al., 1996). Thus, the high expression of XEGIP-related proteins in cultured cells is consistent with a possible role in plants' response to stress.

An Ageratum conyzoides gene similar to EDGP (the carrot homolog of XEGIP), which is strongly upregulated when the plant is infected by Agrobacterium tumefaciens, has been detected by cDNA-amplified fragment length polymorphism (AFLP) studies (Ditt et al., 2001). The 73-amino acid sequence coded by the gene fragment is homologous to both XEGIP and EDGP. Furthermore, the N-terminal sequence (28 amino acids) reported for a 46 kDa cotton seed protein with antifungal activity (Chung et al., 1997) is similar to a sequence within XEGIP (50% identity). Taken together, these observations suggest that XEGIP-related proteins may be involved in plants' defense against pathogenic attack.

Basic 7S globulin (Bg7S) from soybean seed and conglutin γ (Cγ) from L. albus seed (Blagrove and Gillespie, 1975; Blagrove et al., 1980; Elleman, 1977) also have amino acid residues that are conserved with XEGIP (Figure 6), even though amino acid sequences of these proteins are only 35% identical to those of XEGIP and EDGP, suggesting a remote evolutionary relationship. The cysteine residues are especially well conserved, suggesting that the three-dimensional structures of these proteins may have features in common with XEGIP.

Proteins that are closely related to Bg7S and Cγ are widely present in the seeds of leguminous plants (Kagawa et al., 1987) and have long been considered as storage proteins (Blagrove et al., 1980; Kolivas and Gayler, 1993). However, recent research suggests that these proteins may have additional physiologic roles. When soybean or L. albus seeds are immersed in water at 50–60°C, large amounts of Bg7S and Cγ, respectively, are released along with other proteins (Kagawa et al., 1987). The Cγ was found to be newly synthesized rather than constitutive (Duranti et al., 1994), suggesting a potential role in stress response. Bg7S binds insulin and insulin-like growth factors (Komatsu and Hirano, 1991), while Cγ was observed to have lectin-like activity (Duranti et al., 1995). Thus, it appears that the two common characteristics of XEGIP-related proteins are their capacity to bind to specific proteins or peptides and their enhanced expression when the plant is exposed to a stressful stimulus such as heat shock, wounding, or infection.

In summary, we suggest that XEGIP may be a plant defense protein that functions as an inhibitor of a microbial Family 12 glycanase, and as such, XEGIP may play an important protective role by preventing such glycanases from degrading the plant cell wall. Alternatively, XEGIPs and related proteins may play a general role in protecting plants against biotic or abiotic stresses, as suggested by their localization and expression upon wounding, pathogen infection, and heat shock. Further evidence is required to establish the biologic function of XEGIP and XEGIP-related proteins in vivo.

Experimental procedures

Chemicals, reagents, substrates and enzymes

Buffer salts, acids, and bases were obtained from J.T. Baker (Philipsburgs, NJ, USA). Organic solvents were from Fisher Scientific (Pittsburgh, PA, USA) and other reagents were purchased from Sigma Chemical Co. (St Louis, MO, USA). All chromatography columns and matrices used for protein purification were purchased from Amersham Biosciences (Piscataway, NJ, USA).

Xyloglucan-specific endoglucanase (XEG) from A. aculeatus was generously provided by Novozymes (Copenhagen, Denmark) and was purified as described (Pauly et al., 1999).

Tomato cell culture

Tomato (Lycopersicon esculentum‘Bonnie Best’) cell suspension cultures were generated and maintained as described (Smith et al., 1984).

Enzyme inhibition assay

Varying amounts of column fractions (2–50 µl depending on the XEGIP content) were mixed with a defined amount of XEG before adding 0.5–1.0 ml of 4 mg ml−1 of purified tamarind xyloglucan in 20 mm NaOAc buffer (pH 5.3). The reaction was incubated at room temperature, and aliquots were analyzed by a colorimetric assay (Lever, 1972) at different time points to determine the amount of reducing sugar generated by the reaction. The tamarind xyloglucan was reduced with NaBH4 in advance to minimize the background in the reducing sugar assay.

The inhibition constant (Ki) was obtained by the graphical method for determining binding constants for tightly bound inhibitors (Dixon, 1972; Segel, 1975), as illustrated in Figure 5. The initial rates of catalysis were determined as described above except that purified XEG and XEGIP were used, and the concentrations of the two proteins were kept sufficiently low so that both XEG and XEGIP remained partially free in the assay solution (that is, both [Complex] < [XEGIP]tot and [Complex] < [XEG]tot).

Purification of XEGIP

A culture medium from 4 l of 7-day-old cells was filtered through two layers of cheesecloth. Ethanol (95%, 4 volumes) was added to the filtrate to precipitate proteins and polysaccharides. After keeping the solution at 4°C overnight, the ethanol-precipitated material was collected by decantation and centrifugation. The precipitate was re-suspended in water (50 ml) and stirred for several hours at 4°C, any remaining insoluble material was removed by a second centrifugation, and the resulting supernatant was analyzed by SDS–PAGE (Figure 1, lane 1). The supernatant was concentrated at 30°C and dialyzed overnight at 4°C against 20 mm NaOAc, pH 5.2–5.4. The pH of the retentate was adjusted to approximately 7.2 by adding 1 m imidazole (free base), and it was applied to a Q-Sepharose anion-exchange column to remove pectins and anionic proteins. An aliquot of the flow-through fraction was analyzed by SDS–PAGE (Figure 1, lane 2), and the remainder was dialyzed against 20 mm NaOAc (pH 5.2) buffer and then loaded on a HiTrap-S cation-exchange column equilibrated in the same buffer. The cation-exchange column was washed with loading buffer to elute neutral polysaccharides, which were collected for structural analysis. The HiTrap-S column was then eluted with a pH and salt gradient (eluent A, 20 mm NaOAc (pH 5.2); eluent B, 20 mm HEPES (pH 7.4) containing 2 m NaCl; 0–100% B in 60 min at 0.5 ml min−1). The fractions (0.5 ml) were collected and assayed for XEG inhibition. Active fractions (Figure 1, lane 3) were pooled, concentrated, applied to a Superdex-75 SEC column, and then eluted with 20 mm NaOAc (pH 5.2) containing 0.3 m NaCl. The fractions with inhibitory activity against XEG were pooled. SDS–PAGE of the purified inhibitor protein resulted in a single band (Figure 1, lane 4).

Internal peptide sequencing by Edman degradation

Internal peptide sequencing by Edman degradation of tryptic fragments of XEGIP was performed by Michigan State University Macromolecular Structure, Sequencing and Synthesis Facility.

Amino acid assay and protein quantification

Quantitative amino acid analyses (AAA) were conducted by Harvard Microchemistry Facility. The amount of alanine or phenylalanine in standard XEG and XEGIP samples, along with the deduced amino acid sequences of the proteins, was used to accurately determine the protein concentration in these samples. This allowed extinction coefficients (ɛ) of XEG and XEGIP (82 123 and 35 867 m−1 cm−1, respectively) to be obtained by measuring the absorbance of these samples at 280 nm in 50 mm NH4COOH (pH 6.3). Absorption at 280 nm was used to quantify the purified proteins used in kinetic studies. Otherwise, protein concentrations were determined by the Bradford assay (Bradford, 1976).

Trypsin digestion

Purified XEGIP (approximately 400 µg) was dissolved in 50 µl of the reducing buffer (8 m urea, 50 mm Tris–HCl, pH 7.5), and 5 µl of 0.5 m DTT was added. The mixture was incubated at 50°C for 15 min. Iodoacetamide (5 µl of 0.1 m in the reducing buffer, freshly made) was then added, and the reaction was incubated in dark for 15 min. A Microcon 3K filter (Millipore, Bedford, MA, USA) was used to replace the reducing buffer with 50 mm NH4HCO3 (pH of approximately 8.0), the final volume being approximately 150 µl. After adding 5 µl of trypsin (Sigma sequencing grade 1.0 µg µl−1), the reaction was incubated at 37°C in a water bath. The progress of the reaction was monitored by MALDI-MS until no further change was observed. The resulting peptides were vacuum-dried and dissolved in 200 µl of 0.1% trifluroacetic acid (TFA) for HPLC.

HPLC separation of trypsin-generated peptides

Peptides were separated on a Phenomenex Prodigy octadecylsilica (ODS) (5 µm particles, 250 mm × 2.0 mm, 100 Å) or a Supelco Discovery C-18 (5 µm particles, 150 mm × 2.1 mm) column. Peptides were eluted from the columns at a flow rate of 0.2 ml min−1 in the following gradient: (0–10% buffer B over 10 min, 10–60% buffer B over 90 min, and 60–100% buffer B over 20 min; buffer A: 0.1% TFA (Sequanal Grade, Pierce, Rockford, IL, USA) in water; buffer B: 0.085% TFA in 80% acetonitrile). Peptides were detected by monitoring the absorbance at 214 nm. Fractions were manually collected and assayed by MALDI-MS. Selected pure peptides were subjected to MS/MS sequencing.

Mass spectrometry

MALDI-TOF mass spectrometry of xyloglucan oligosaccharides was performed as described previously (Pauly et al., 2001). MALDI-TOF mass spectrometry of peptides and proteins was performed according to the standard procedures (Jiménez, 1998). Mass spectra of peptides were also recorded with a Q-TOF hybrid mass spectrometer (Q-TOF2, Micromass, UK) equipped with an electrospray source (Z-spray) operated in the positive mode. The HPLC-separated tryptic peptides were reconstituted in 50% methanol with 1% formic acid and infused into the Q-TOF mass spectrometer with a syringe pump (Harvard Apparatus Cambridge, MA, USA) at a flow rate of 5 µl min−1. A potential of 3 kV was applied to the capillary, and nitrogen was employed as both the drying and the nebulization gas. [Glu]-Fibrinopeptide B was used as the calibration standard in the positive mode. For MS analysis, quadrupole Q1 was operated in the RF-only mode with all the ions transmitted into the pusher region of the TOF analyzer and the MS spectrum was recorded from m/z 400–2000 with a 1 sec integration time. For MS/MS spectra, the transmission window of quadrupole Q1 was set up to about 3 mass units and the selected precursor ions were allowed to fragment in the hexapole collision cell. The collision energies (40–55 eV) were optimized for maximized product ion yield, and argon was used as the collision gas. The mass spectrometry (MS/MS) data were integrated over a period of 1–2 min for each precursor ion.

Cloning the XEGIP cDNA

XEGIP amino acid sequence fragments, derived from Edman sequencing of the native protein, were used to search public databases, and several hits were aligned using dnasis (Hitachi Software, San Francisco, CA, USA). Conserved regions near 5′ and 3′ ends were used to design degenerate primers (Sense: 5′-TGG GTI GAY TGY GAY MAR RRI TA-3′; antisense: 5′-TCI ADY TGR WRI CCI CCD ATI ACI A-3′). First-strand cDNAs were obtained by reverse transcription (M-MMLV reverse transcriptase; Invitrogen, Carlsbad, CA, USA) of mRNA extracted from suspension-cultured tomato cells (Plant RNeasy Kit, Qiagen, Valencia, CA, USA) and were used as PCR templates. PCR was performed using Taq polymerase (Fisher, Pittsburgh, PA, USA) on a Bio-Rad iCycler thermal cycler under the following conditions: 95°C for 4 min, 40 cycles (95°C, 30 sec; 43–50°C, 50 sec; 72°C, 1 min), and 72°C for 7 min. PCR products were analyzed by electrophoresis on 1% agarose gels. Bands with the expected size were recovered with the Qiagen gel extraction kit. The purified fragments were cloned into pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA, USA), and competent TOP 10 cells provided with the kit were transformed with the vector according to the manufacturer's instructions. The cell colonies that grew on ampicillin plates were selected and amplified. The plasmids were purified using the Qiagen Miniprep kit, digested with EcoRI, and analyzed by electrophoresis. Those with an insert of the predicted size were sequenced at the Molecular Genetics Instrumentation Facilities of the University of Georgia.

One fragment cloned by the above method showed sequence overlap with one Tentative Consensus sequence (TC85129) and one EST singleton (EST258116) in the TIGR Gene Indices database (Quackenbush et al., 2000, 2001). Assembly of these three sequences in silico resulted in a sequence that contained a full-length open reading frame. New sets of primers were synthesized accordingly to clone the full-length cDNA (Sense: 5′-CGT GCC GAT TAA ATC ATG GCT TCT-3′; antisense: 5′-ATT CAA TAC ATG AAT TAA AAC AAC-3′). The same templates were used as described above. PCR was performed with the high-fidelity Taq polymerase AccuTaq (Sigma, St Louis, MO, USA) under the following conditions: 95°C for 4 min, 40 cycles (95°C, 30 sec; 48°C, 50 sec; 72°C, 1 min), and 72°C for 7 min. The resulting PCR fragments were cloned and sequenced as before.

DNA gel blot analysis

Genomic DNA was extracted from young tomato leaves (cv. Ailsa Craig) as described (Murray and Thompson, 1980), and 10 µg aliquots were digested with the appropriate restriction enzymes, fractionated on agarose gels, and transferred to nylon membranes as previously described (Rose et al., 1997). The membranes were hybridized at 42°C in 50% formamide, 6× SSPE, 0.5% SDS, 5× Denhardt's solution, and 100 mg ml−1 sonicated salmon sperm DNA, with radiolabeled DNA probes corresponding to either the full-length XEGIP cDNA, or an 819 bp DNA fragment obtained by digestion of the XEGIP cDNA with PvuII. The probes were synthesized with the Ready-To-Go DNA Labeling Beads (–dCTP) Kit (Amersham Biosciences, Piscataway, NJ, USA) using 32P (dCTP), and purified with ProbeQant G-50 Micro Columns (Amersham Biosciences). Following hybridization, the membranes were washed thrice in 5× SSC, 1% (w/v) SDS at 42°C for 15 min followed by three washes in 0.2× SSC, 0.5% SDS at 65°C for 20 min, and then exposed to the film.

RNA gel blot analysis

RNA was extracted from tomato vegetative tissues and from a series of fruit developmental stages (ovaries, expanding stages I–III, immature green, mature green, breaker, turning, pink, light red, and red ripe) as outlined by Rose et al. (1996) and Cataláet al. (2000). Total RNA (15 µg per lane) samples were subjected to electrophoresis on 1.2% agarose, 10% formaldehyde gels and transferred to Hybond-N membrane (Amersham Biosciences) as previously described (Rose et al., 1997). Blots were hybridized and washed as described for the DNA blot analysis, but with the high-stringency washes at 65°C in 0.5× SSC, 0.5% SDS.


The authors would like to thank Dr Carmen Catalá for determining inhibitory activity of XEGIP on XET, and Novozymes A/S for generously supplying the XEG. The Harvard Microchemistry Facility performed the amino acid analysis, the University of Georgia Molecular Genetics Instrumentation Facility performed the DNA sequencing, and the Michigan State University Macromolecular Structure Facility performed the protein sequencing. This research was supported by US Department of Energy (DOE) grant DE-FG05-93ER20220 and the DOE-funded (DE-FG02-93ER20097) Center for Plant and Microbial Complex Carbohydrates.

The accession number and homology (per cent amino acid sequence identity to XEGIP) for each gene and protein described in this article is given below. AY155579 (tomato XEGIP, 100%); D14550 (carrot EDGP, 61%); AJ297490 (conglutin γ from Lupinus albus, 32%); P13917 (soybean Bg7S, 32%); AAK59531 (Arabidopsis-1, 62%); AAC72120 (Arabidopsis-2, 58%); BAB89707 (rice-1, 27%), BAB89708 (rice-2, 25%); BAB89703 (rice-3, 25%); BAB89705 (rice-4, 27%); BAB89709 (rice-5, 29%). The TIGR Gene Indices Tentative Consensus numbers and homologies are TC1987 (lotus-1, 46%); TC2190 (lotus-2, 37%); TC43357 (Medicago-1, 63%), TC43356 (Medicago-2, 58%); TC43854 (Medicago-3, 34%), TC40118 (Medicago-4, 29%); TC44724 (Medicago-5, 40%); TC51626 (Medicago-6, 33%); TC51629 (Medicago-7, 33%); TC24971 (potato-1, 90%); TC80165 (soybean-1, 33%); TC80166 (soybean-2, 62%); TC110649 (soybean-3, 63%); TC100743 (soybean-4, 33%), TC102507 (tomato-2, 22%); TC99653 (tomato-3, 39%); TC108379 (tomato-4, 36%); TC103252 (tomato-5, 51%); TC102514 (tomato-6, 35%).