Photoaffinity labeling of the adenine binding sites of two Dolichos biflorus lectins.

Two differentially expressed lectins from the legume Dolichos biflorus, the seed lectin and a stem and leaf lectin (DB58), were photoaffinity-labeled at their adenine binding sites using the probe [2-3H]8-azidoadenine. Both heteromeric subunits I and II of the seed lectin and alpha and beta of DB58 were specifically labeled. This result, combined with the adenine binding site stoichiometries of two identical sites/seed lectin tetramer or one site/DB58 dimer, indicates that the adenine binding site resides at a heterologous subunit interface. Three radiolabeled peaks from seed lectin and one from DB58 were isolated from chymotryptic digests of the labeled lectins by reverse phase chromatography at pH 7.0. From these four peaks, six unique peptide sequences were determined. When aligned with the concanavalin A sequence, four of these peptides map to three loops in the metal binding domain of concanavalin A. The remaining two sequences represent carboxyl-terminal peptides unique to the D. biflorus lectins which may extend to the putative binding site from adjacent, heterologous subunits. It thus appears that the adenine binding sites of these D. biflorus lectins are within the metal binding domain and adjacent to the carbohydrate binding site.

The two best characterized lectins of D. biflorus, the seed lectin and DB58, derive from separate, differentially expressed genes (Harada et al., 1990), yet share 87.6% homology in their primary structures (Schnell and Etzler, 1988). The seed lectin is a heterotetrarner (Carter and Etzler, 1975) localized in the protein bodies of the seed cotyledons (Etzler et al., 1984), whereas DB58 is a heterodimer (Talbot and Etzler, 1978) localized in the vacuoles of young stems and leaves (Bunker and Etzler, 1993). The heteromeric subunits I and I1 of the seed lectin appear to differ only by a truncation a t their carboxyl termini,l wherein subunit I1 arises by the proteolytic cleavage of subunit I (Quinn and Etzler, 1989). Similarly, the only difference between the (Y and p subunits of DB58 is also at the carboxyl terminus.2 The seed lectin binds two molecules of GalNAc per tetramer . DB58 also binds GalNAc (Etzler and Borrebaeck, 19801, but does not agglutinate erythrocytes or precipitate glycoconjugates (Talbot and Etzler, 1978) and is presumed to be monovalent for carbohydrate binding. Recent studies have shown that adenine is also bound to two identical sitedseed lectin tetramer or one site/ DB58 dimer (Gegg et al., 1992), and that both these lectins bind adenine with similar affinity and specificity.
Because of the many uses for plant lectins as research tools (Lis and Sharon, 19861, the presence of high affinity hydrophobic binding sites should be a consideration in their application. Perhaps more intriguing is what the physiological significance of this activity might be in the plant. The finding that these sites bind the most active forms of cytokinin (Roberts and Goldstein, 1983b;Gegg et al., 1992) and, in a t least one case, appear to bind cytokinin cooperatively (Gegg et al., 1992), is suggestive that this class of phytohormone ligand may be associated with the function of the lectin in the plant. Localization of the adenine binding sites on the legume lectins may provide insights into the role of this activity, Photoaffinity labeling of the lima bean and kidney bean lectins with 8-azidoadenine has identified peptides which correspond to an asymmetric subunit interface of C o d 3 (Maliarik and Goldstein, 1988

5687
photoafinity labeling study was undertaken to better understand the structure/function relationship between the seed lectin and DB58 from D. biflorus with respect to their adenine binding sites. >98% radiochemical purity) and [2-3H18-azidoadenosine (14 CL'mmol, >98% radiochemical purity) were purchased from Moravek Biochemicals (Brea, CA). 8-Azidoadenosine and TLCK-treated chymotrypsin were obtained from Sigma. Ultrafree-MC filters with low binding cellulose membranes having a nominal molecular weight cutoff of 10,000 used for the competitive displacement assays were from Millipore. FPLC solvents were from Fisher. The D. biflorus lectins were isolated as previously described (Gegg et al., 1992).
Competitiue Displacement-Binding of unlabeled NRAde was measured using a centrifugal ultrafiltration assay to determine concentrations of unbound [l4C1adenine as described (Gegg et al., 1992). The determined concentrations of displaced [14Cladenine were used to calculate average relative affinity constants for N3Ade from the known affinity and binding stoichiometry of adenine with each lectin (Martin et al., 1991;Gegg et al., 1992).
Photoaffinity Labeling-25 pg of seed lectin (227 pmol) or DB58 (431 pmol) were equilibrated with 25 p~ NJ3H1Ade in 50 m~ Tris-HC1, pH 7.2 for 30 min a t room temperature then chilled 5 min in an ice bath. Samples (25-200 pl) in polypropylene microcentrifuge tubes were irradiated at 254 nm for 3 min in an ice bath from a height of 5 cm (1000 microwatts/cm2) using a Spectroline hand lamp (Westbury, N Y ) with the filter removed. Immediately after photolysis, the reactions were quenched with dithiothreitol added to a final concentration of 10 mM.
Lectin was removed from the reaction mixture by addition of 3 volumes of EtOH under conditions shown to precipitate >95% of the lectin. Photoincorporation of label was determined by scintillation counting EtOH precipitates of each reaction and subtracting the nonirradiated control. The degree of nonspecific photoincorporation was estimated by substituting 25 pg each of trypsin, alkaline phosphatase, or y-globulin for the lectin procedure described above. Alternatively, the quenched reaction mixtures were diluted with an equal volume of 87 m~ triethanolamine-HC1, pH 9.7, sample buffer containing 8 M urea, 0.2% SDS, 12.5% sucrose, and 0.0025% bromphenol blue and were immediately analyzed by SDS-urea-PAGE and fluorography (Carter and Etzler, 1975).
Identification of Labeled Peptides-200 pg of seed lectin or 100 pg of DB58 were labeled a t 1.0 and 0.75 mg/ml, respectively, with 25 p~ N,[,H]Ade a s described. The EtOH-precipitated lectins were resuspended in 8 M guanidine HCl for 20 min at room temperature then diluted 1:4 with 0.1 M Tris-HC1, pH 8.0. The diluted lectins were digested by overnight incubation a t room temperature with TLCKtreated chymotrypsin a t a ratio of 1:20 chym0trypsin:lectin (w/w).
The chymotryptic digests were initially separated by reverse phase FPLC on a 4.6 x 100 mm Aquapore RP-300 octyl column (Brownlee Laboratories) using a triethy1amine:phosphate (TEAP) buffer system at pH 7.0 and a 15-45% acetonitrile gradient. Peptide-containing peaks were detected by absorbance at 230 nm and plotted a s a function of elution time. Radioactivity profiles were determined by scintillation counting 400 p l of each fraction.
Peptide peaks containing radioactivity significantly above background were then dried and repurified on the same column using a trifluoroacetic acid buffer system a t pH 2.2 with a 0-50% acetonitrile gradient, Detection of peptides and radioactivity was the same as before. The isolated peptides were sequenced by automated Edman degradation on a model HPGIOOOA sequenator (Hewlett Packard) by the Protein Structure Laboratory at the University of California, Davis.

RESULTS
Lectin Afinity for 8-Azidoadenine-N3Ade is an effective competitor of [l4C1adenine binding to the seed lectin and DB58 ( Fig. 1). Displacement of [I4C]adenine by N d d e is concentration-dependent, and, at 250 p~ N3Ade, <20% of the [I4C]adenine remained bound by either lectin. Affinity constants calculated from the displacement of [14C]adenine by 0.5-250 PM N3Ade are based on adenine affinities of 7.0 x lo5 liters/mol of seed lectin with two binding sitedtetramer and 1.3 x lo6 liters/ mol of DB58 with 1 binding site/dimer (Gegg et al., 1992). The affinity constants determined for N d d e binding are 1.27 x lo5 liters/mol 0.27 for the seed lectin tetramer and 2.26 x lo5 liters/mol * 0.42 for the DB58 dimer.
Photoincorporation of [2-3H18-Azidoadenine-Protection by adenine of the seed lectin and DB58 from N,[3HlAde photoincorporation is dependent on adenine concentration (Fig. 2). Photoincorporation in the seed lectin is more responsive to protection at low adenine concentrations than DB58; however, both lectins respond similarly to 500-2000 p~ adenine. Background levels for nonspecific labeling by N3I3H1Ade were estimated by averaging photoincorporation of label into the proteins: trypsin, alkaline phosphatase, and y-globulin. Although nonspecific labeling of the non-lectin proteins averaged 0.35 nmol/mg 2 0.17, this represents only 13.2% of the seed lectin labeling and 19.6% of DB58.
To test if protection by high adenine concentrations was due to chemical quenching of the photoactivated label, labeling reactions prepared with seed lectin as described in the identification of labeled peptides were analyzed by TLC as described for the preparation of [2-3H]8-azidoadenine. The addition of a 1000 p~ concentration of the scavenger p-aminobenzoate (RF = 0.75) prevents photoincorporation and causes the appearance of a new radiolabeled species (RF = 0.12) which is presumably the scavenger adduct. In contrast, the addition of 1000 PM adenine ( R F = 0.66) does not yield a new radiolabeled species but instead causes an increase in the level of unincorporated radiolabel (RF = 0.29).
The dependence of photoafinity labeling on irradiation time is demonstrated in Fig. 3. Maximum photoincorporation of N3[3H]Ade in both lectins is achieved after a 1-min irradiation with 15-18% of the seed lectin and DB58 binding sites becoming labeled. In the presence of adenine, maximum photoincorporation of N3r3H]Ade is achieved after 2 min for both lectins and at a lower level of labeling. A comparison of these curves indicates that the kinetics of photoincorporation in the presence of adenine is slower than in its absence. There was no radiolabel incorporation in the absence of ultraviolet irradiation or when the label was prephotolyzed prior to adding protein. Photoincorporation into each of the lectin subunits was observed by SDS-urea-PAGE and fluorography (Fig. 4). Subunits I and I1 of the seed lectin tetramer and subunits C Y and p of DB58 were each labeled with N3t3H]Ade; however, adenine effectively protects these subunits from photoincorporation of label. There was no evidence for ultraviolet light-induced crosslinking of subunits or photodegradation of either the seed lectin or DB58.
Isolation of Photoaffinity-labeled Peptides-Preliminary efforts to separate photoafinity-labeled lectins from the unbound label by trichloroacetic acid precipitation indicated that the photoincorporated radiolabel was acid-labile. Isolation of radiolabeled peptides from tryptic or chymotryptic digests by reverse phase FPLC using a 0.1% trifluoroacetic acid:acetonitrile solvent system at pH 2.2 was also unsuccessful. These observations are consistent with those of other laboratories (Brinegar et al., 1988;Maliarik and Goldstein, 1988). Alternatively, a TEAP:acetonitrile solvent system at pH 7.0 was found to preserve -15-20% of the photoincorporated radioactivity in discrete peptide-containing peaks.
Chymotryptic digests of the seed lectin and DB58, both photoafinity-labeled with N3t3H]Ade, were separated at pH 7.0 with the TEAP:acetonitrile solvent system and their respective chromatograms, and radioactivity profiles are shown in Fig. 5. The seed lectin digest produced two distinct peaks of radioactivity after the flow-through. The first radioactive peak in- seed lectin (A) or 100 pg of DB58 ( E ) were initially separated by reverse phase chromatography a t pH 7.0 using a TEeacetonitrile solvent system. Solvent A was 0.3% triethylamine, 0.1% phosphoric acid, pH 7.0, and solvent B was 40% TEAP, 60% acetonitrile. Samples were loaded during 15 min a t 85% solvent A15% solvent B followed by a 60-min linear gradient to 55% solvent A:45% solvent B. The radioactivity profiles for each chromatogram are plotted for lectins labeled in the absence (-----) or presence (---1 of 1000 p~ adenine. cluded two small, poorly resolved peptide peaks with elution times of 32.63 min (peptide 1) and 33.53 min (peptide 2). The second radioactive peak contained a single small peptide peak at 37.16 min (peptide 3). Radioactivity in all three peaks was eliminated when N3[3H]Ade labeling was performed in the presence of adenine. The DB58 digest gave only one distinct peak of radioactivity with a corresponding peptide peak at 45.88 min (peptide 4).
The isolated peaks (1-4) were further purified by a second chromatographic step utilizing reverse phase FPLC a t pH 2.2 using a 0.1% trifluoroacetic acid:acetonitrile solvent system. Peaks 1 and 2, from the seed lectin, were resolved with elution times of 27.10 and 25.07 min, respectively (Fig. 6A). A small amount of peak 3, also from the seed lectin, was eluted a t 25.33 min and contained some trace contaminants eluting at 27.17  Fig. 5, A and B , were rechromatographed at pH 2.2 using a trifluoroacetic acid:acetonitrile solvent system. Solvent A was 0.1% trifluoroacetic acid, pH 2.2, and solvent B was 0.1% trifluoroacetic acid in acetonitrile. Samples were loaded in 200 pl of solvent A during 5 min at 100% solvent A followed by a 30-min linear gradient to 50% solvent A50% solvent B. and 27.86 min (Fig. 63). These contaminants as well as peptide 3 may have been derived from the large adjacent peak eluting a t 40.18 min in Fig. 5A, as this peak yields three peaks at 25.36, 27.15, and 28.97 min in the 0.1% trifluoroacetic acid: acetonitrile solvent system (data not shown). Peak 4, from DB58, gave a large single peak at 26.84 min in the 0.1% trifluoroacetic acid: acetonitrile solvent system (Fig. 6C). The large trailing peak eluting at 8 min in all the chromatograms is the TEAP solvent carried over from the initial chromatographic separation. The small peaks at 21 and 38 min are also solventrelated. None of the peaks contained detectable levels of radioactivity after pH 2.2 chromatography.

Sequence Determination of Photoafinity-labeled Peptides-
The results of automated sequence analysis of the rechromatographed peaks 1-4 are given in Table I. The sequence determined for peak 1 corresponds to residues 228-241 of the seed lectin (Schnell and Etzler, 1987) which is located near the carboxyl termini of both subunits I and 11. Two sequences from the seed lectin were identified for peak 2; the more abundant sequence 2a corresponds to residues 90-104, while the 10-fold less abundant sequence 2b is a near match for residues 138-146. Peak 3 also gave two seed lectin sequences; however, both were in low yield. The first sequence 3a from peak 3 corresponds to residues 125-137 and is contiguous with the minor sequence 2b. Sequence 3a also contains 2 mismatched residues and a tryptophan. The second sequence 3b, obtained from peak 3, is a partial match to sequence 2a, and the residue corresponding to cycle 7 was indeterminable. Peak 4 from DB58 also gave two sequences in nearly equivalent amounts. The first sequence 4a aligns with residues 145-154 of DB58 (Schnell and Etzler, 1988) and overlaps the two mismatched residues at the carboxyl terminus of sequence 2b from the seed lectin. The other sequence 4b, obtained from peak 4, matches residues 248-251 which is at the extreme carboxyl terminus of the DB58 a subunit. The amino-terminal residues of all peptides with the exception of peptides 2b and 4a are consistent with chymotryptic cleavage sites. These discrepancies in peptide sequence and chymotryptic cleavage may represent sites of modification by the photoaffinity probe.

DISCUSSION
The 8-azidopurine nucleotides have been used successfully to label the nucleotide binding sites of a variety of different proteins (Potter and Haley, 1983;King, et al., 1991); their popularity stems from the relative nonspecific reactivity of the photoactivated nitrene which reacts by bond insertion (Bayley and Knowles, 1977). The yield of photoincorporation can be highly variable, 0.1-50%, and the new bond is frequently labile. In these experiments, it is essential to demonstrate the affinity and specificity of the target protein for binding the label and that photoincorporation is specific and dependent on irradiation. (Roberts, et al., 1986) have shown that this binding site is most tolerant of modifications at C-2 and C-8 of adenine. Maliarik and Goldstein (1988) have determined that the single adenine binding sites on each of the tetrameric lectins from the lima bean and kidney bean have reduced affinities for 8-azidoadenine; however, N d d e binding could be significantly enhanced in the presence ofANS. Similarly, the D. biflorus lectins bind NaAde with affinity constants that are approximately 80% of those found for adenine binding. In contrast to the lima bean and kidney bean lectins, the D. biflorus lectins show a higher overall affinity for N d d e with a binding stoichiometry of 2 siteslseed lectin tetramer and 1 sitelDB58 dimer. Also, unlike the lima bean and kidney bean lectins, ANS has no effect on N3Ade binding to either D. biflorus lectin. From the affinity constants of NaAde for the D. biflorus lectins, we calculate that 63% of the seed lectin binding sites and 77% of the DB58 binding sites are occupied by N3[3HIAde at equilibrium under the conditions used for photolabeling.

Afinity, Specificity, and Photoincorporation of [2-3H18hidoadenine-Previous studies on the adenine binding site of the lima bean lectin
The specificity of NJ3H]Ade for the adenine binding site is further evidenced by the protection from NaL3H1Ade occupation and photoincorporation afforded both lectins by adenine. Although some nonspecific photoincorporation is indicated by the labeling observed with trypsin, alkaline phosphatase, and y-globulin, the labeling of both the seed lectin and DB58 was a t least 5-fold higher than the non-lectin proteins. Because protection from photoincorporation requires a large excess of adenine, despite the higher affinity of the lectins for adenine, it was considered possible that adenine was chemically quenching the photoactivated label. The TLC analysis of the labeling reactions indicates that, unlike the scavenger p-aminobenzoate, adenine does not react directly with the photoactivated label to generate a chemical adduct with a unique mobility. Instead, adenine appears to protect the lectin SO that the portion that would have become photoincorporated simply returns to the pool of unincorporated label. If 80% of the lectin labeling is specific and adenine is not chemically quenching the photoactivated label, then it is difficult to understand the high levels of adenine necessary for protection. An explanation for this may be derived from the comparatively long half-life of the activated affinity label, on the order of a millisecond, which could be very much larger than the rates of dissociation for both the label and adenine from the active site. In that context, several bindingldissociation events may occur during the life-is no residue determined. time of the activated label, providing multiple opportunities for photoincorporation (Ruoho et al., 1973). Therefore, adenine becomes less effective as a protectant because it would eventually be displaced by the activated label. Although the dissociation of activated label from the binding site may lead to higher nonspecific labeling, the highest levels of photoincorporation shouid occur at multiple sites in and around the binding site, decreasing as a function of distance away from the true binding site orientation.
The dependence of N3[3H]Ade photolabeling on irradiation time indicates that label incorporation occurs through a photoactivated intermediate, most probably a nitrene. The absence of labeling from prephotolyzed N3L3H]Ade shows that the intermediate nitrene is not so long-lived that the observed labeling continues after the irradiation is discontinued. Photoincorporation is complete for both lectins after 1 min of irradiation; however, when 1000 VM adenine is included, labeling saturation is not complete until 2 min of irradiation. The change in labeling kinetics may be due to quenching by the ultravioletabsorbing adenine. The yield of photoincorporation in the absence of adenine is .-15-17% of the total binding sites present during labeling or -22-24% of the binding sites estimated to be occupied by N3[3HlAde.
Photoaffinity labeling of the adenine binding sites of seed lectin and DB58 with N3L3H1Ade yields specific photoincorporation into both subunits I and I1 of the seed lectin and a and p of DB58. Taken in the context of adenine binding stoichiometry, which is two identical sitedseed lectin tetramer and one sitelDB58 dimer, this observation indicates that the binding site must occur at a heterologous subunit interface. Similarly, photoaffinity labeling of the lima bean and kidney bean lectins with N3L3H1Ade has identified a tryptic peptide that aligns with a sequence in the ConA crystal structure which traverses a homologous subunit interface (Maliarik and Goldstein, 1988).
Localization of Photolabeled Peptide-The acid lability of the photoprobe was partially overcome by performing the initial separations of photoaffinity-labeled peptides by reverse phase FPLC at pH 7.0. Even at pH 7.0, only -5 2 0 % of the photoincorporated radioactivity was recovered in discrete peptide peaks, the rest eluting in the flow-through or as a n elevation in background throughout the gradient. The background radioactivity observed in the chromatograms of Fig. 5 may derive from nonspecific labeling of the lectins. Alternatively, the background radioactivity may represent small populations of specifically labeled peptides and peptide fragments from around the binding site which are labile in response to the acetonitrile gradient. Rechromatography at pH 2.2 allowed the further isolation of peaks 1-4, but resulted in a complete loss of label.
Sequence analysis of the photoafinity-labeled peaks 1-4 gave six different peptide sequences, each of which corresponds to unique seed lectin or DB58 peptides. Two of the seed lectin sequences are contiguous, and one DB58 sequence overlaps the carboxyl terminus of these contiguous peptides on alignment with the corresponding seed lectin sequence. One sequence each from the seed lectin and DB58 correspond to carboxylterminal peptides. In two cases, peptides were identified which did not correspond to chymotryptic peptides as judged by their amino-terminal residues. In a third peptide, a n aromatic residue was detected within the sequence. These three inconsistencies may represent sites of N3[3HlAde modification.
Considerable sequence homology has been observed among the legume lectins , despite in some cases dramatic differences in quaternary structure (Goldstein and Poretz, 1986). Based on these differences in structure, three subclasses have been proposed for the legume lectins (Becker, et al., 1983). The circularly permuted ConA is a member of the first subclass, the "two-chain" lectins favin and pea belong to the second subclass, and, among the third, are the G. simplicifolia isolectins and the GalNAc binding Erythrina corallodendron and D. biflorus lectins. The three-dimensional crystal structures have been solved for ConA (Hardman and Ainsworth, 1972;Reeke et al., 19751, the broad bean lectin, favin (Reeke and Becker, 1986), pea lectin (Einspahr et al., 19861, and the lectins from E. corallodendron (Shaanan et al., 1991) and G. simplicifolia (Delbaere et al., 1990). The protomers from each lectin subclass are nearly identical in secondary and tertiary structure and are quite similar to the predicted secondary structure of the D. biflorus seed lectin and DB58 (Schnell and Etzler, 1988).
The striking similarity in tertiary structure among the legume lectins supports using the three-dimensional structure of ConA as a model for an approximation of the topographical location of the adenine binding sites of the D. biflorus lectins.
When mapped to the a-carbon backbone of ConA we find that peptides 2a and b, 3a a n t b, and 4a comprise three loops which approach to within 10 A of each other in the metal binding domain of ConA. Peptide 3a aligns with the metal binding loop of ConA (residues 10-23) and the residues determined in sequencing cycles 6 and ll which are inconsistent with the seed of l b o Dolichos biflorus Latins lectin sequence are positioned in proximity to one another and the other two loops. Peptides 2b from the seed lectin and 4a from DB58 correspond to the second loop of ConA (residues 24-32 and 31-40, respectively). The point where peptides 2b and 4a overlap represents the point of nearest approach for this loop with the other two loops of C o d . This domain of peptide 2b and 4a overlap also contains residues inconsistent with the predicted seed lectin sequence and represents a nonchymotryptic cleavage site for peptide 4a. The third loop is described by peptides 2a and 3b and corresponds to residues 213-229 of C o d . The final few carboxyl-terminal residues of this peptide are the most proximal to the other two loops and in peptide 2a includes an aspartate in sequencing cycle 12 which is inconsistent with the seed lectin sequence. Peptide 4b represents the carboxyl-terminal peptide of the DB58 OL subunit and is not present on the p subunit. It may be possible for this peptide to extend to the three loops described above from either the same or adjacent subunits arranged as in the ConA tetramer. Finally, peptide 1 is something of a n enigma because it clearly represents a carboxyl-terminal portion of the seed lectin, but it is unclear if it derives from subunit I or 11. It is therefore difficult to suggest what interaction this peptide may have with the three-loop domain defined by the other peptides. Although the legume lectins are largely devoid of a helix, the proposed threeloop adenine binding domain is reminiscent of the highly conserved mononucleotide binding domains of the dehydrogenases (Rao and Rossman, 1973;Rossman et al., 1974).
The close proximity of the adenine binding sites of the D. biflorus lectins to their metal binding domains is inconsistent with the location of the adenine binding sites proposed for the lima bean and kidney bean lectins (Maliarik and Goldstein, 1988). Superimposed on the ConA model, the photoaffinitylabeled peptides isolated from the lima bean and kidney bean lectins comprise a p-strand in the middle of a six-member sheet with one side making contact with an asymmetric subunit interface and the other side facing the monomer core and central hydrophobic pocket. The disparity in placement of the adenine binding sites between these lectins and the D. biflorus lectins may have been influenced by their differences in adenine binding affinity, stoichiometry, or the effects ofANS used to enhance photoincorporation into the lima bean and kidney bean lectins.
The consequences of localizing the adenine binding site in such close proximity to the metal and carbohydrate binding sites is unclear. It has been shown that GalNAc binding is metal-dependent for the D. biflorus seed lectin , but that adenine binding is independent of GalNAc for both the seed lectin and DB58 (Gegg, et al., 1992). It is interesting that the active forms of the cytokinin class of phytohormone are high affinity ligands for the adenine binding sites of the D. biflorus lectins, because, in addition to their role in plant cell growth and differentiation, they have been shown to complex divalent metals (Hideshi and Hirobe, 19861, scavenge superoxide radicals (Frimer, et al., 1983), and retard plant senescence (Leshem, et al., 1981;Leshem, 1988). The oxidation of cytokinin by superoxide converts the N6 amine to an amide. In the case of benzyladenine, the oxidation product is benzoyladenine, which is also a ligand for the adenine binding site of seed lectin albeit with a somewhat lower affinity than ben-zyladenine (Gegg, et al., 1992). The localization of a binding site for cytokinin near the metal ions and adjacent to the carbohydrate site may be consistent with a catalytic role for this domain in protecting the plant and its seeds from oxidative damage and senescence.