Determination of the carbohydrate-binding properties of peanut agglutinin by ultraviolet difference spectroscopy.

The anti-T lectin from peanuts was purified on a new affinity matrix, and the number of carbohydrate binding sites was determined by equilibrium dialysis with [14C]lactose to be four per tetramer. Methyl-alpha- and methyl-beta-D-galactopyranoside and lactose were found to perturb the UV spectrum of the lectin in the aromatic region and their association constants were determined by UV difference spectroscopy to be 1.8, 1.0, and 1.3 X 10(3) M-1, respectively, at 25 degrees C. Thermodynamic parameters were also obtained for the two galactosides from measurements at several temperatures. For the alpha anomer, delta H degrees = -42 kJ mol-1 and delta S degrees = -78 J K-1 mol-1; for the beta anomer, delta H degrees = -43 kJ mol-1, and delta S degrees = -86 J K-1 mol-1. Analysis of the lectin for metal atoms disclosed 0.98 mol of Ca2+ and 0.78 mol of Mg2+ subunit, while manganese was present in trace amounts only. The results of the present study indicate that recent improvements in instrumentation should make UV difference spectroscopy more widely applicable to studies of protein-ligand interactions.

. v The anti-T lectin from peanuts was purified on a new affinity matrix, and the number of carbohydrate binding sites was determined by equilibrium dialysis with ['4C]lactose to be four per tetramer.
Methyl-a-and methyl-/?-D-galactopyranoside and lactose were found to perturb the UV spectrum of the lectin in the aromatic region and their association constants were determined by UV difference spectroscopy to be 1.8, 1.0, and 1.3 x lo3 M-~, respectively, at 25°C. Thermodynamic parameters were also obtained for the two galactosides from measurements at several temperatures.
Analysis of the lectin for metal atoms disclosed 0.98 mol of Ca*+ and 0.78 mol of M&+/subunit, while manganese was present in trace amounts only. The results of the present study indicate that recent improvements in instrumentation should make UV difference spectroscopy more widely applicable to studies of protein-ligand interactions.
tinin (4), is being used widely for monitoring the distribution of the T-antigen on erythrocyte surfaces (5.6) and as a probe for studying T lymphocyte subpopulations (7-11). Moreover, this lectin is a useful tool for the separation of mature and immature murine and human thymocytes (12,13), permitting the study of in vitro maturation of immature thymocytes into immunocompetent cells and providing better characterization of lymphocyte subpopulations with regard to their surface properties (14). PNA receptors have also been established as a new cell surface marker in studies of in vitro differentiation of teratocarcinomas (15,16 After equilibration for 2 days at 4OC, 8O-pl aliquots were taken from the chambers and counted in 10 ml of scintillation fluid (Aquasol, New England Nuclear, Montreal, Quebec) on a Beckman LS-7OOO scintillation counter. UV Difference Spectroscopy-UV difference spectra were recorded on a Cary 219 spectrophotometer in masked semimicro cells, of 1-cm pathlength. This instrument is capable of reading absorbance to 0.O001 unit. Aliquots of 1 ml of PNA solution (approximately 2 mg/ ml in PBS) were added to both sample and reference cells and the base-line was recorded into the instrument's memory unit to be subtracted automatically from subsequent spectra. Small aliquots (1 p l ) of a solution of the sugar under study was added to the sample cuvette, while the reference cuvette received the same amount of PBS buffer. The difference spectrum was recorded after thermal equilibration had been re-established, as indicated by a thermocouple in the cell holder. The total volume of ligand solution added was 10 to 15 pI, rendering concentration corrections for diiution unnecessary. Control experiments were conducted with methyl-a-D-glucoside, which is not an inhibitor of PNA (3).

RESULTS
PNA was purified on an affinity chromatography matrix prepared by reacting p-aminophenyl-P-D-lactoside with an active ester derivative of agarose beads. The capacity of the gel for PNA was 10 mg of PNA/ml of packed gel, which is higher than the best previously reported matrix (21) and the column had good flow properties (flow rate approximately 100 ml/h). Although we used p-nitrophenyl-b-D-lactoside for the preparation, essentially the same results should be obtainable with the p-nitrophenyl-D-galactosides, which are commercially available.
Examination of the product obtained by this procedure for metals, using atomic absorption spectroscopy, showed calcium and magnesium were present in significant amounts, i.e. approaching 1 mol/mol of subunit assuming a molecular weight of 27,50O/subunit. The contents were: Ca", 0.98 mol/mol subunit; Mg2+, 0.78 mol/mol; Zn, 0.11 mol/mol; other metals, including Nn2+, were present in trace amounts only (t0.05 mol/mol). These contents were not changed by prior dialysis against 0.01 M EDTA, pH 7.2. The total of magnesium and zinc is close to 1 mol/mol of subunit, suggesting these may be in the same site. Unlike many lectins, PNA did not contain significant amounts of manganese.
Equilibrium Dialysis-PNA has been shown to be a tetramer of identical subunits (3, 22), but its valency was reported as only 2 (Ref. 23, no experimental details given). Since valency cannot be determined directly by UV difference spectroscopy when the protein is the absorbing species, the number of carbohydrate binding sites on the PNA tetramer was determined by equilibrium dialysis, using [1-'4C]lactose. The data are shown in Fig. 1, plotted according to the Scatchard equation r/c = K(n-r), where r is the number of moles of lactose bound per mol of PNA, c is the free sugar concentration, n is the number of binding sites per mol of PNA, and I( is the association constant. The molecular weight of PNA was taken to be 110,000 (3). The number of binding sites, obtained from the x intercept in Fig. 1, was found to be 3.7 * 0.3. AS PNA is a tetramer at neutral pH (3,22), our data indicate one binding site per subunit of 27,500 molecular weight. The apparent association constant for lactose at 4°C was found to be 1.1 x lo4 M" (see Table I), Since mutarotation of the lactose will be occurring under the conditions of this experi-

Properties of Peanut Agglutinin
Me-P-D-Gd UV Difference Spectroscopy-The three sugars examined, methyl-a-and methyl-P-D-galactoside and lactose, all perturbed the UV spectrum of PNA (Fig. 2). Fig. 3 shows titration curves for the binding of the three sugars to PNA, obtained from the difference maximum at 285 nm or 286 nm for various sugar concentrations. The association constants (Table I) (Fig. 2 A ) . The binding of methyl-a-D-glucopyranoside to concanavalin A also gave a positive maximum at 285 nm (24), but no feature at 278 nm in the difference spectrum. When the effect of temperature on the binding of methyla-D-galactoside to PNA was studied, no change in the overall shape of the difference spectrum was found, but its magnitude decreased with increasing temperature. Thus, the 285 nm magnitude was 0.03 at 7.6"C, 0.028 a t 16"C, 0.025 a t 25°C and 0.021 a t 35"C, respectively (corresponding to Ac values, based on PNA subunit concentration, ranging from 430 to 300 M" cm"). Table I shows the association constants obtained; from these, thermodynamic parameters for this interaction were determined using a Van't Hoff plot, as displayed in Fig. 4. The enthalpy and entropy changes were found to be AH" = -42 kJ mol" and AS" = -78 J K" mol-'. Methyl-P-D-gdactopy- ranoside gave a difference spectrum (Fig. 2B) very similar to that observed with the a anomer, exhibiting positive maxima a t 286 nm and 278 nm, an isosbestic point at 274 nm and a slightly more pronounced negative maximum at 266 nm, but the magnitude of the difference spectrum was smaller. The magnitude of the absorbance change at 286 nm a t sugar saturation was 0.023 a t 8.6"C, 0.021 at 16.8"C, 0.017 a t 25"C, and 0.015 at 35"C, respectively (corresponding to AE values, based on PNA subunit concentration, ranging from 332 to 216 M" cm"). Association constants were obtained again a t each temperature from titration curves, and thermodynamic parameters were calculated from these constants by constructing a Van't Hoff plot (Fig. 4); AH" was -43 kJ mol" and AS" -86 J K" mol". In the case of lactose (Fig. 2C), again positive maxima a t 286 nm and at 278 nm were observed. However, the shape of the difference spectrum is different from that obtained with the monosaccharides, there being a third, smaller positive maximum a t 268 nm.
Pea Lectin-To test the UV difference spectroscopy method on a system with a smaller absorbance change on binding, the pea lectin was chosen. On binding methyl-a-Dglucoside, the absorbance of a 1.3 mg/ml solution of the lectin changed by 0.01 at 7°C. The difference spectrum had a maximum at 292 nm, which may be due to tryptophan involvement in sugar-binding as chemical modification suggested (25), and a smaller maximum a t 285 nm. The association constant for methyl-a-D-glucoside was 7.8 X 10' M", in agreement with the value of 8 X 10' M" obtained by equilibrium dialysis at 4°C (20).

DISCUSSION
Amino acid sequence work on PNA has shown it has some homology to Con A (12 out of 40 residues being identical) in the region so far sequenced (26). Although the two sugar specificities are quite different, the results obtained here bear out a general similarity of PNA to Con A. One major exception is that the metal analysis points to the PNA subunits having 1 mol each of calcium and magnesium rather than the calcium and manganese of Con A. There have been two reports of other lectins with calcium and magnesium, namely, Bandeiraea simplicifolia I which had 0.5 mol of Ca" and 0.31 mol of Mg''/subunit (27) and a lectin from Phaseolus vulgaris, 0.98 mol of CaZ+ and 0.63 mol of Mg"/subunit (28). The B. simplicifolia lectin is also galactose-specific and it, PNA, and a Ph. vulgaris lectin also show similarities in their CD spectra, both in the far UV and aromatic regions.' There is some similarity to the behavior of Con A (24) in the form of the UV difference spectra obtained here for PNA, but of more interest are the subtle differences between the spectra given by the three sugars. The a-and @galactosides differ by a nanometer in their maxima and in the intensity of the difference spectra they generate, while lactose, although of intermediate association constant (Table I), gives rise to a difference spectrum with an additional peak. These variations may be taken to support the view of ligand binding by the site as a mutual fit process, where the actual conformation of the site as well as the major ligand site contacts vary subtly with changes in the ligand structure (29).
In Table 11, the binding constants and thermodynamic parameters obtained for PNA are listed with values from the literature for three comparable systems. The PNA values are very close to those measured by microcalorimetry for the binding of D-galactose to ricin (30). The AHo values for the various systems are essentially the same within experimental error, suggesting similar site interactions. The variations in the binding constants hence arise mainly from the differences in the entropy contribution.
At 25"C, the ratio of PNA binding constants for methyl-aand methyl-@-D-galactoside is 1.8:1, and for methyl-a-D-galactoside and lactose it is 1.4:l. These ratios for the binding constants agree well with the hemagglutination inhibition data of Lotan et al. (3). From their data for the ratio of the inhibitory activity of the T-antigen to that of methyl-a-Dgalactoside, and the association constant obtained here for methyl-a-D-galactoside, the association constant for the interaction of PNA with the T-antigen can be estimated to be about 1 X lo7 M" at 25"C, a value similar to those found for the interaction of glycopeptides with Con A, up to 2.3 X 10' M" (33). These values exceed those found for anti-lactose antibodies, K = 1.6 X IO5 M" (32) or carbohydrolytic enzymes such as lysozyme, which binds chitotriose with a K of lo6 (34). The binding data also suggest that lactose binds more strongly than methyl-a-D-galactoside at 4"C, the reverse of the situation at 25°C.
The observed absorption changes (1 to 2%) reported here are quite small, particularly in the case of methyl-P-D-galactopyranoside, but were found to be accurately measurable with very good reproducibility. With the instrumentation used in the present study, even smaller total absorption changes (around 0.01 A ) were sufficient for spectrophotometric titrations to be performed to determine an association constant for the pea lectin and methyl-a-D-glucoside. The value found was in good agreement with the literature value (20). Provided they have little or no UV absorbance, the interaction of more complex sugars such as glycopeptides with PNA may be a N. M. Young and R. E. Williams, unpublished observations. studied by this technique, and only milligram amounts would be needed. UV difference spectroscopy may also be useful with lectins such as Ph. vulgaris agglutinin for which there is no simple sugar inhibitor. The results indicate that the current generation of spectrophotometers should make UV difference spectroscopy more widely applicable to studies of proteinligand interactions.