Purification and properties of a novel xyloglucan-specific endo-(1----4)-beta-D-glucanase from germinated nasturtium seeds (Tropaeolum majus L.).

Endo-(1----4)-beta-D-glucanase activity has previously been detected in the cotyledons of germinated nasturtium (Tropaeolum majus L.) seeds, and has been linked to the hydrolysis in vivo of storage xyloglucan (amyloid) (Edwards, M., Dea, I. C. M., Bulpin, P. V., and Reid, J. S. G. (1985) Planta (Berl.) 163, 133-140). Extracts from the cotyledons of 14-day seedlings are now shown to contain a single endo-glucanase activity, and it has been purified to apparent homogeneity by sequential anion-exchange chromatography, cation-exchange chromatography, and gel filtration. The purified enzyme gave a single protein band on polyacrylamide gel electrophoresis, dodecyl sulfate gel electrophoresis, and isoelectric focusing. The isoelectric point was 5.0, the pH optimum 4.5-5.0, and the temperature optimum 40 degrees C. Dodecyl sulfate electrophoresis gave a molecular weight of 29,000; permeation methods gave lower values. The enzyme contained about 5% carbohydrate. An endo mode of action on tamarind seed xyloglucan was deduced by monitoring solution viscosity and reducing power, and by analyzing the end products of the reaction. The specificity of the enzyme toward a wide range of substrates, including celluloses, carboxymethyl and hydroxyethyl celluloses, glucomannan, galactoglucomannans, mixed-linkage (1----3 and 1----4)-beta-D-glucan, laminarin, and xyloglucans from seeds and primary cell walls, was tested. Only the xyloglucans were hydrolyzed. It is concluded that the enzyme is a pure, endo-acting (1----4)-beta-D-glucanase which is novel in its apparently complete specificity toward xyloglucans. It is speculated that enzymes of this type may have escaped earlier detection because it is normal practice to screen for endo-(1----4)-beta-D-glucanase using artificial cellulose derivatives and that they may be widely involved in xyloglucan metabolism in plant cell walls.

Extracts from the cotyledons of 14-day seedlings are now shown to contain a single endo-glucanase activity, and it has been purified to apparent homogeneity by sequential anion-exchange chromatography, cationexchange chromatography, and gel filtration. The purified enzyme gave a single protein band on polyacrylamide gel electrophoresis, dodecyl sulfate gel electrophoresis, and isoelectric focusing. The isoelectric point was 5.0, the pH optimum 4.5-5.0, and the temperature optimum 40 "C. Dodecyl sulfate electrophoresis gave a molecular weight of 29,000; permeation methods gave lower values. The enzyme contained about 5% carbohydrate. An endo mode of action on tamarind seed xyloglucan was deduced by monitoring solution viscosity and reducing power, and by analyzing the end products of the reaction. The specificity of the enzyme toward a wide range of substrates, including celluloses, carboxymethyl and hydroxyethyl celluloses, glucomannan, galactoglucomannans, mixed-linkage (143 and 144)-&D-glUCan, laminarin, and xyloglucans from seeds and primary cell walls, was tested. Only the xyloglucans were hydrolyzed. It is concluded that the enzyme is a pure, endo-acting (1+4)-@-D-glucanase which is novel in its apparently complete specificity toward xyloglucans. It is speculated (a) that enzymes of this type may have escaped earlier detection because it is normal practice to screen for endo-(l+4)-B-~glucanase using artificial cellulose derivatives and (b) that they may be widely involved in xyloglucan metabolism in plant cell walls.
In the early 19th century it was observed that the cell walls in the cotyledonary or endosperm tissues of some seeds could be stained blue with iodine. The material responsible for the staining reaction was assumed to be starch-like and it was named "amyloid (Vogel and Schleiden, 1839). Amyloid-positive cell walls are present in the seeds of at least 230 dicotyledonous species (Kooiman, 1960), and the amyloids from four species-Tropaeolum majus L. (Le Dizet, 1972), Tamarindus indica (Kooiman, 1961), Impatiens balsamina (Courtois * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and Le Dizet, 1974) and Annonu muricata (Kooiman, 1967)have been isolated and subjected to full structural analysis. All four amyloids are xyloglucans (or more correctly, galactoxyloglucans), based on a (14)-@-linked D-glucan (cellulosic) backbone which carries a-D-xylopyranosyl and @-Dgalactopyranosyl-(1+2) -a -D -xylopyranosyl substituents. Both types of substituent are linked ( 1 4 ) to the D-glucan backbone. Galactose/xylose/glucose ratios range from 1:2:3 in Tropaeolum and Tamarindus amyloids to 1:1:4 in that of Annona (Reid, 1985). Polysaccharides closely similar in structure to the seed amyloids are present in the primary cell walls of dicotyledonous plants (Bauer et al., 1973;Darvill et al., 1980). The primary wall xyloglucans differ from the amyloids in that they contain an additional structural feature; some 60-80% of their side chain D-galactosyl residues are further substituted by (1*2)-linked D-fUCOSY1 residues. The fucosyl residues are probably in the a-anomeric form (Darvill et al., 1980).
Early microscopic studies suggested that amyloids disappear from seeds after germination (Heinricher, 1888;Reiss, 1889) and it has been assumed that they are reserve polysaccharides. Surprisingly, this was confirmed only very recently in the course of an investigation of xyloglucan metabolism in the cotyledons of the nasturtium seed (T. majus L.) after germination (Edwards et al., 1985). The amyloid of the nasturtium seed was shown to be mobilized completely following germination. It constituted 30% of the total substrate reserves utilized by the seed, and must therefore be classified as a major cell-wall storage polysaccharide (Meier and Reid, 1982).

EXPERIMENTAL PROCEDURES
Materials-Tamarind seed xyloglucan was prepared from commercial tamarind flour as described previously (Edwards et al., 1985).
Assays-In crude or partially purified enzyme preparations endo-P-D-glucanase activity was assayed by viscometry, using tamarind seed xyloglucan as substrate (Edwards et al., 1985). Once purified free of other activities which were capable of releasing reducing sugars from xyloglucan or its breakdown products, the endo-glucanase was assayed, more accurately, by monitoring the release of reducing power by the direct ferricyanide method (Halliwell and Riaz, 1970) or by the more sensitive p-hydroxybenzoic acid hydrazide method (Lever, 1972). D-Glucose was used as standard. In both assays the substrate was at 10 mg ml" and the buffer was McIlvaine phosphate citrate, pH 5.0 (Dawson et al., 1982). Viscometric units were calculated as before (Edwards et al., 1985). Activities quoted in nanokatals were calculated on the basis of glucose equivalents released.
p-D-Galactosidase activity was measured by determining the release of p-nitrophenol from p-nitrophenyl-P-D-galactopyranoside and by assaying the release of D-galactose from tamarind seed xyloglucan specifically using D-galactose dehydrogenase (Edwards et al., 1985).
a-D-Xylosidase activity was assayed by determining the release of pentose from tamarind seed xyloglucan as before (Edwards et al., 1985). D-Glucose was determined using the hexokinase/~-glucose-6-phosphate dehydrogenase procedure , D-galactose by using D-galactose dehydrogenase (Kurz and Wallenfels, 1974), and pentose by a modification of the p-bromoaniline procedure of Roe andRice (1948) (Edwards et al., 1985).
Neutral sugars in polysaccharide hydrolysates were determined quantitatively by gas-liquid chromatography of their alditol acetate derivatives (Albersheim et al., 1976).
Protein was assayed by the dye-binding procedure of Sedmak and Grossberg (1977) using bovine serum albumin as standard.
Standard carbohydrate and protein solutions were prepared daily using the appropriate buffer solution.
Zsotation of the Enzyme-Seeds, germination, and growth conditions were as before (Edwards et al., 1985). Cotyledons from 14-day seedlings (typically about 400 g, fresh weight) were harvested and homogenized for 30 s in a blender (M.S.E. Ltd. "Atomix") with 0.2 M sodium phosphate buffer, pH 7.2 (600 ml), and insoluble polyvinylpolypyrrolidone (Sigma) (6 g). This was carried out at 4 "C, as were all subsequent operations during isolation and purification. The suspension was stirred for 1 h to allow diffusion of enzyme from cell walls, centrifuged (19,000 X g for 30 min), and the supernatant brought to 90% saturation with ammonium sulfate by addition of the crystalline salt. The precipitate was collected by centrifugation (19,000 X g for 30 min), dissolved in the minimum volume of unbuffered 0.1 M sodium chloride, and dialyzed against the same solution. The precipitate which formed in the dialysis sac was removed by centrifugation (25,000 X g for 20 min), and the supernatant dialyzed The abbreviations used are: DP, degree of polymerization; DS, degree of substitution; SDS, sodium dodecyl sulfate; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
Anion-exchange Chromatography-The dialyzed supernatant was applied to a column (2.2 X 20 cm) of DEAE-cellulose (Whatman DE52) and the column was washed with the above buffer. When the absorbance at 280 nm of the eluate had fallen to a constant value, the column was eluted with a sodium chloride gradient (0-0.5 M in the same buffer, over 7 bed volumes). The column flow-rate was 80 ml h-'. Fractions containing high endo-b-D-glucanase activity were pooled, dialyzed against 20 mM sodium acetate buffer, pH 5.0, and subjected to cation-exchange chromatography.
Cation-exchange Chromatography-The pooled fractions from the preceding step, in 20 mM sodium acetate buffer, were applied to a column (2.2 X 20 cm) of CM-cellulose (Whatman CM52). The column was washed with the above buffer until the absorbance at 280 nm of the washings was constant. The column was then eluted with a sodium chloride gradient (0-0.5 M in the same buffer, over 7 bed volumes). Fractions containing high endo-s-glucanase activity were pooled and concentrated by rotary evaporation at 25 "C prior to gel filtration.
Gel Filtration-The concentrated solution from the previous step was applied to a gel filtration column. The column was eluted with McIlvaine phosphate citrate buffer (Dawson et al., 1982) diluted to %o strength. In earlier preparations a Sephadex G-200 column (2.5 X 33 cm) was used, and in later preparations a Bio-Gel P-60 column (3.0 X 73 cm) was used. Although both columns gave pure endo-8-Dglucanase, the Bio-Gel column was more satisfactory because of its greater capacity, and because the contaminating proteins were eluted at or near the void volume. The Sephadex column was run at 10 ml h" and the Bio-Gel column at 1.6 ml h". Fractions with high endob-D-glucanase activity were pooled and frozen in aliquots.
Isoelectric Focusing-Analytical isoelectric focusing was carried out using Ampholine PAGplates (LKB Ltd). They were run and stained with Coomassie Blue exactly according to the manufacturer's instructions.
Estimation of the Molecular Weight of the Native Enzyme-Both gel filtration columns mentioned earlier were calibrated and used to estimate the molecular weight of the nondenatured enzyme. The molecular weight was also determined using high performance liquid chromatography (LKB System) in conjunction with permeation chromatography columns (TSK 2000 SW and TSK 3000 SW). Molecular weight standards were cytochrome c (12,4001, ribonuclease A (13,700), myoglobin (17,0001, chymotrypsinogen (24,000), ovalbumin (45,000), and bovine serum albumin (66,000). pH Optimum and Stability; Temperature Optimum-These parameters were determined using tamarind seed xyloglucan as substrate, and assaying for reducing power by the p-hydroxybenzoic acid hydrazide method. The pH optimum was obtained using McIlvaine phosphate citrate buffers covering the range pH 3.0-9.0. To determine the pH stability, preincubations were carried out at different pH values and assays performed at pH 5.0. The temperature optimum was obtained by assaying at different temperatures.
Carbohydrate Content of the Purified Enzyme-The phenol-sulfuric acid method (Dubois et al., 1956) was used with D-mannOSe as standard.
Mode of Action of the Enzyme-Enzyme and tamarind seed xyloglucan solutions were mixed in the same proportions as in the standard assay (final substrate concentration 10 mg m1-I). Reaction progress was monitored by measuring viscometric flow time as described previously (Edwards et al., 1985) and by determining reducing power on boiled aliquots by the direct ferricyanide method. When the reducing power of the reaction mixture had reached a constant value, the enzyme was inactivated by heating on a boiling-water bath (5 min), and the end products were analyzed by TLC (McCleary et al., 1982).
Substrate Specificity of the Enzyme-Soluble and insoluble sub-strates were usually dissolved or suspended in McIlvaine phosphate citrate buffer (%o strength), pH 5.0, and mixed with enzyme (0.06 nkat/ml, final volume, determined using tamarind xyloglucan as substrate) to give a final concentration in the assay of 10 mg ml-I. Over a period of 24 h viscometric flow times were measured if appropriate and reducing power was determined on aliquots using the p-hydroxybenzoic acid hydrazide method. A control incubation was carried out for each substrate. Those substrates which gave inconveniently high viscosities at 10 mg ml" were used at lower concentrations. The action of the enzyme on cello-oligosaccharides was determined by examining the incubation mixture by TLC.

RESULTS
Purification of the Enzyme-Nasturtium (T. majus L.) seedlings were harvested 14 days after sowing, when endo-P-Dglucanase, a-D-xylosidase, and @-D-galactosidase activities were all at or near their maximum levels in the cotyledons (Edwards et al., 1985). The endo-~-D-glucanase was purified from cotyledon extracts by a combination of DEAE-cellulose chromatography, CM-cellulose chromatography, and gel filtration ( Table I and Fig. l). Fig. l shows that only one form of endo-0-D-glucanase was present, whereas both 8-galactosidase and a-xylosidase were multiple activities. The purified enzyme migrated as a single band on polyacrylamide gel electrophoresis (PAGE), SDS-PAGE (Fig. 2), and isoelectric focusing (Fig. 3). When concentrated, purified enzyme preparations were used to completely depolymerize tamarind seed xyloglucan, no galactose, pentose, or glucose could be detected in the final mixture of products, confirming the absence of even trace amounts of @-galactosidase, a-xylosidase, or Pglucosidase.
The purification data in Table I suggest that an approximately 9.5-fold purification results in a homogeneous enzyme. This implies that the enzyme constituted about 10% of the protein in the cotyledon extract after ammonium sulfate precipitation and dialysis. However it is clear from the electrophoretic separation pattern (Fig. 2) that it represented a very much smaller percentage of the protein in the extract. Substances interfering with the protein determination and/ or synergistic interactions between xyloglucan-degrading enzymes may have led to an overestimation of specific activity during the earlier stages of purification. Molecular Properties, pH, and Temperature Optima-On SDS-PAGE the purified enzyme had an apparent molecular weight of 29,000 (Fig. 2). The apparent molecular weight of the nondenatured enzyme was somewhat lower. Gel chromatography on a calibrated Bio-Gel P-60 column and by high performance liquid chromatography gave values of 13,000 and 22,000, respectively. A small amount of carbohydrate (-5%) was associated with the enzyme, but the amount is unlikely to be sufficient to interfere significantly with the estimation of the molecular weight of the molecule by SDS-PAGE. We therefore conclude that the enzyme is a single polypeptide  chain, of molecular weight 29,000. The lower values obtained for the nondenatured enzyme suggest that it has a compact tertiary structure.
With 1.0% tamarind seed xyloglucan as substrate, the pH optimum was at pH 4.5-5.0 (stable from pH 4-8). The temperature optimum was 40 "C. The isoelectric point of the molecule was pH 5.0 (Fig. 3).
Mode of Action and Substrate Specificity-The enzyme brought about a very rapid decrease in the specific viscosity of xyloglucan solutions in the early stages of the reaction, whereas the reducing power of the incubation mixture increased slowly in a linear fashion over a very much longer period (Fig. 4). This type of relationship between specific viscosity decrease and reducing power release is characteristic of endo-as opposed to exo-hydrolytic cleavage. When the hydrolysis of tamarind seed xyloglucan by the enzyme was complete, i.e. when the reducing power of the reaction mixture had reached a constant value, the products of the reaction were examined by TLC. TLC confirmed that no monosaccharides were present, and did not reveal any oligosaccharides  The gel was stained with Coomassie Blue. Activity and pH were obtained by eluting strips of unstained gel. For activity determination elution was done overnight with McIlvaine phosphate citrate buffer, pH 5.0. For pH determination elution was done for 1 h with distilled water. Points are plotted to correspond with centers of eluted strips. with mobilities greater than that of cellopentaose. The method separated three mobile components all migrating more slowly than cellopentaose from an unresolved mixture of even higher molecular weight material (Fig. 5).
The enzyme is clearly an endo-(1+4)-/&D-glucanase which hyrolyzes tamarind seed xyloglucan rapidly, but only to a limited extent. The final value of the reducing power released in Fig. 4 suggests a degree of hydrolysis of 9% and indicates that some 18% of the D-glucosidic links in the D-glucan backbone are hydrolyzed.
The substrate specificity of the enzyme was tested initially using soluble cellulose derivatives. Under conditions where tamarind and nasturtium seed xyloglucans were depolymerized rapidly, the enzyme brought about no change in either the viscosity or the reducing power of sodium carboxymethyl celluloses (DP 400,1100, 3600;DS 0.4,0.7, 12), hydroxyethyl celluloses, hydroxypropylmethyl celluloses, and methyl celluloses. Table I1 summarizes the results of further experiments in which the action of the enzyme on a wider variety of polymeric substrates containing the (lh)-P-D-glucosidic link (plus the (1+8)-b-D-glUCan laminarin) was tested. Clearly, the enzyme hydrolyzes only xyloglucans. It has no effect on cellulose (microcrystalline or phosphoric acid repre-

TABLE I1
Substrate specificity of endo-@D-glucanase from T. majus cotyledons Substrates were at 10 mg ml-'. Viscosity was determined at frequent intervals (see Fig. 4). Controls showed no change in viscosity or reducing power. cipitated), cello-oligosaccharides, carboxymethyl cellulose, hydroxyethyl cellulose, konjac glucomannan, galactoglucomannans, barley mixed-linkage (1+3) and (l-A)-P-glucan or laminarin. At a substrate concentration of 10 mg ml" the xyloglucans of tamarind and nasturtium seeds were hydrolyzed more extensively than the xyloglucan from the primary cell walls of mung bean hypocotyls.

DISCUSSION
The endo-( 1+4)-P-D-glucanase of the germinated nasturtium seed is novel in its apparently total specificity toward xyloglucans, and in future work we will investigate the molecular basis of this specificity. On the basis of our present data one may propose two general hypotheses to explain the specificity of the enzyme. It may be a true xyloglucanase in the sense that it recognizes structural features in the side chains of xyloglucans. Alternatively, it may be an endo-(l-*4)-P-Dglucanase with a subsite binding requirement of a type which has not hitherto been observed.
Our observation (Table 11) that the enzyme hydrolyzes the xyloglucans of T. indica and T. majus seeds more extensively than that from mung bean primary cell walls tends to support the first hypothesis. The galactose/xylose/glucose ratios are similar in all three xyloglucans. The primary cell-wall xyloglucan, however, differs from the other two in that some 60% of the galactose-containing side chains are terminated by Dfucosyl residues (Kato and Matsuda, 1976). Our observation that only xyloglucans were hydrolyzed is, however, also explicable if the enzyme recognizes only the (1+4)-P-D-glucan backbone, but requires, for example, a soluble substrate with a relatively long sequence of residues which are unsubstituted at any position other than C-6. Celluloses and cello-oligosaccharides of DP > 6 are insoluble; artificial cellulose derivatives have a relatively high frequency of substitution at C-2 and C-3; soluble cello-oligosaccharides, konjac glucomannan, galactoglucomannans, and barley P-glucan have relatively short regions of contiguous (1+4)-@-linked D-glucose residues. Clearly, the natural substrate of the present enzyme is the storage xyloglucan in the cotyledonary cell walls of the nasturtium seed. The enzyme does, however, hydrolyze other xyloglucans including the fucose-containing primary cell-wall type. Primary cell-wall xyloglucans are now known to be major non-cellulosic polysaccharides in dicotyledonous plants (Darvill et al., 1980), and there is good evidence that they are turned over during elongation growth (Labavitch and Ray, 1974). Two cellulases (more correctly, end0-(1+4)-P-D-glUcanases) have been isolated from elongating pea epicotyls (Byrne et al., 1975;Wong et al., 1977). Both hydrolyze cellulose, sodium carboxymethyl cellulose, and xyloglucan (Wong et al., 1977). The levels of both enzymes increase after auxin treatment and they may participate in the cell-wall changes associated with cell elongation. Although their natural substrate is not known, it has been suggested that it may be xyloglucan (Hayashi and Maclachlan, 1984). This is possible, but the participation of xyloglucan-specific endo-P-D-glucanases in the turnover of primary cell-wall xyloglucans must now be considered. Recently, York et al. (1984) have reported that a specific nonasaccharide product obtained by the hydrolysis of a primary cell-wall type xyloglucan by Trichoderma uiride endo-P-D-glucanase inhibits the auxin-stimulated elongation of pea-stem segments. This intriguing observation may indicate a wider regulatory role for those enzymes which catalyze the endo-hydrolysis of xyloglucan in vivo.
Although this is the first report of a xyloglucan-specific endO-(l~)-B-D-glUCanaSe, enzymes with a similar substrate specificity need not necessarily be uncommon in nature. The virtually universal practice of screening for endo-(1-+4)-@-Dglucanase using sodium carboxymethyl cellulose or some other soluble cellulose derivative may simply have obscured their presence.