Biochemistry 1981, 20, 3076-3083
3076
Thermodynamics of Concanavalin A Dimer-Tetramer Self-Association: Sedimentation Equilibrium Studies? Donald F. Seneart and David C. Teller*
ABSTRACT: The effects of temperature and pH on the reversible dimer-tetramer association of concanavalin A were studied by the high-speed sedimentationequilibrium technique. Both commercial and highly purified preparations of concanavalin A were used. Equilibrium constants were analyzed as a Wyman linked function by using truncated van't Hoff temperature dependence. In the concentration range 0.1-3.0 mg/mL, from 5 to 35 "C and between pH 5.5 and pH 7.5 at 0.5 M ionic strength, only dimer and tetramer species were present in both preparations. For purified concanavalin A, association constants ranged from 1.5 X lo3 to 8.0 X lo7 M-I. Constants for our commercial preparation were 10-fold lower due to the decreased competency of some subunits to self-associate. From the fit of the Wyman model to the experimental data, A G O , AIP,AS",and ACpowere calculated
for the association and association-linked ionization reactions. From the valucs of the ionization thermodynamic parameters, the association is governed by the ionization of a histidine side chain on each subunit, either histidine51 or histidme121. The association is characterized by large entropy (66.3 cab mol-ldeg-l at 25 "C) and heat capacity (-821 cal*mol-'.deg-') changes in accordance with the large hydrophobic association surface observed in crystallographic studies [Reeke, G. N., Jr., Becker, J. W., & Edelman, G. M. (1975) J. Biol. Chem. 250,1525-15471, In addition, there is a large enthalpy change (10.4 kcal.mo1-' at 25 "C). We propose a model for the interaction based on a more detailed thermodynamic description than was obtained in an earlier, incomplete study [Huet, M., & Clavarie, J. M. (1978) Biochemistry 17, 236-24 11.
Interest in the unusual biological properties of the jack bean lectin concanavalin A (Con A)' has made it the subject of intense biochemical study. The primary and three-dimensional structures are known (Cunningham et al., 1975; Becker et al., 1975; Hardman & Ainsworth, 1972), and the interactions between the subunits that form the crystalline quaternary structure have been well described (Reeke et al., 1975). Although it crystallizes as a tetramer of identical 25 500 molecular weight protomers, some of which are fragmented (Kalb & Lustig, 1968; Wang et al., 1971), the solution quaternary structure is pH and temperature dependent. Con A has been reported to be a dimer at pH 5.5 and a tetramer at pH >7 (McKenzie et al., 1972; Huet, 1975). An analysis of this self-polymerization has shown that a dimer-tetramer equilibrium exists (Huet & Clavarie, 1978). This equilibrium is of interest because some biological properties of Con A have been shown to depend on the valency of Con A for saccharide ligands [e.g., Gunther et al. (1973) and Yasaka & Kambara (1979)l. It is also of interest because it provides a model for the energetics involved in maintaining the folded structures of globular proteins. Chothia (Chothia & Janin, 1976) has shown that the packing densities of amino acid residues in subunit interfaces and in subunit interiors are the same, both resembling amino acid crystals. This suggests that the energetics for the folding of polypeptide chains into globular subunits and for the assembly of subunits into oligomeric structures are similar. Thus, self-associating systems make simple models for folding energetics which feature well-defined initial and final states and which involve relatively few interactions. The conformational entropy of folding transitions is avoided. Con A, which exhibits only two molecular weight species related by a single equilibrium constant,
provides the simplist possible model. It is necessary to measure equilibrium constants over a wide temperature range to obtain accurate thermodynamic parameters. For typical, pH-dependent self-association reactions it is also necessary to analyze the pH dependence in order to subtract out the thermodynamics of association-linked ionization reactions. Because this approach has not often been followed for proteins whose three-dimensional structure is known, there is very little available data that can be applied to deduce the roles of the various types of interactions involved in maintaining oligomeric structures. An earlier study of the Con A self-polymerization ignored the effect of associationlinked ionization reactions (Huet & Clavarie, 1978). In the present work, we have studied the Con A self-association from 5 to 35 "C and between pH 5.5 and pH 7.5 by using the high-speed sedimentation equilibrium technique and computer programs developed in this laboratory (Teller, 1973). The pH dependence was analyzed by the Wyman linked function theory to separate the self-association and association-linked ionization reactions. Truncated van't Hoff equations were used to express the temperature dependence of the reactions. The group responsible for the pH dependence was identified from the thermodynamic parameters deduced for the association-linked ionization reaction.
N
+ From
the Department of Biochemistry, School of Medicine, University of Washington, Seattle, Washington 98 195. Received September 24, 1980. This work was supported by National Institutes of Health Grant GM 13401. t Present address: Institute of Molecular Biology, University of Oregon, Eugene, OR 97403.
0006-296018 110420-3076$01.25/0
Experimental Procedures Chemicals. Concanavalin A, prepared from jack bean meal by using acetic acid as per Olson & Liener (1967) and lyophilized, was purchased from Sigma Chemical Co. (Type IV, lot no. 16C-7090). The Con A supplied by Sigma contains N 50% fragmented chains.2 These were removed by precipI Abbreviations used: Con A, concanavalin A rms, root mean square; NaDodSO,, sodium dodecyl sulfate. About 50% of concanavalin A subunits in commercial preparations are hydrolyzed between residues 1 1 8 and 119. We term such preparations as "fragmented". However, the hydrolyzed subunits maintain their normal folded structure. In no case are materials smaller than dimer observed in sedimentation equilibrium of this Con A (see Figure 2).
0 1981 American Chemical Society
CON A S E L F - A S S O C I A T I O N T H E R M O D Y N A M I C S
itation in NH,HCO, as described by Cunningham et al. (1972). Their procedure was modified slightly by first preparing the Con A at 10 mg/mL in 0.005 M sodium acetate, pH 2.3, and then dialyzing this solution against 1% NI-bHCO,, pH 7.90, a t 37 'C for 12 h. After removal of the pellet, the supernatant was filtered through a 22-pm millipore filter. When dialyzed exhaustively into the phosphate buffer described below (pH 6.5). these solutions were exceptionally stable and could be stored for up to 12 weeks at 4 OC without any deterioration detectable by ultracentrifuge analysis. Solutions of Con A were prepared by dialysis at 20 "C in 0.05 M sodium phosphate buffer containing 0.2 mM CaCI, and MnCI,, 0.1 mM NaN,, and enough NaCl to bring the total ionic strength to 0.5 M at the desired pH. The final protein concentration was estimated spectrophotometrically at 280 nm by using an extinction coefficient, E," = 11.4 (Agrawal & Goldstein, 1968). All buffer components were reagent grade products of J. Baker Chemical Co. or Mallinckrodt Chemical Works. Analytical Ultracentrifugation: High-Speed Sedimentation Equilibrium Technique. Sedimentation equilibrium experiments were performed in a Beckman Spinco Model E ultracentrifuge equipped with electronic speed control. Data were collected as Rayleigh interference patterns, photographed on Kodak 11-G spectroscopic plates. A mercury arc light source, a Baird-Atomic B-9 interference filter with a 546-nm transmittance maximum, and an interference mask with the slit width enlarged to 0.014 in. were used to produce the fringes. Temperature was controlled by an RTIC temperature control unit. At least 12 fringes were routinely resolved. Thus the effective concentration range measured was 0.1-3 mg/mL. For runs at temperatures >25 'C, the chamber was lined with mirrored stainless steel to avoid excessive use of the heater (Aune et al., 1971). Centrifuge experiments were performed at 18 000-22000 rpm in Yphantis-style six-channel charcoal-filled epoxy centerpieces with matched sapphire windows. Solution channels were filled to a 3-mm column height with 0.13 mL of Con A solution, which had been dialyzed against the appropriate buffer. A 3:l dilution series (0.5, 1.0, and 1.5 mg/mL) of Con A was used with dialyzate placed in the solvent channels. Photographs were taken at least 4 h after the time required to reach equilibrium (12-14 h), calculated according to Teller (1973). Sedimentation equilibrium tracked the temperature closely in runs in which the temperature was changed after equilibrium was attained, implying that chemical equilibrium is rapid relative to the time scale of the experiment. Photographs were taken at least 1 h after the new temperature was reached. For these runs, the pH of the buffer was calculated at each temperature other than 20 OC by using the data from Bates & Acree (1943). Base line photographs were taken at 3200 rpm after shaking the assembled cell in a test-tube vortex mixer. Method of Analysis. Equilibrium and base h e photographs were read on a modified automated Nikon microdensitometer described by De Rosier et al. (1972) and interfaced to a digital equipment PDP-12 computer. The system uses Fourier transform analysis to determine the displacement from the position of the zeroth-order fringe. Crude base line and equilibrium data were smoothed, and point-by-point number-, weight-, and z-average molecular weight moments were calculated as described by Teller (1973). Point-by-point association constants for the reaction 2D T were calculated from each molecular weight moment according to expressions given by Hoagland & Teller (1969). and a weighted-average value
VOL. 20, N O . I I , 1981 A
B
3077
.c
L L NATIVE FRAGMENTS
FIGURE I: NaDod.30.-polyacrylamide gel electrophoresis analysis of Con A. These are 15% gels, electrophoresed for 4 h at 4 mA/gel and 20 OC. The gels are (a) commercial Sigma Type IV Con A, (b) the pellet after precipitation of Con A in 1% ammonium bicarbonate. and (c) the ammonium bicarbonate supernatant fraction.
of the association constant (k,) was calculated for each experiment. The root-mean-square deviation of the predicted vs. the observed weight-average molecular weights was used to assess how well the model described the data. For these calculations, the dimer molecular weight was always taken to be 51 000,the analytical dimer molecular weight (Becker et al., 1975). The data were always best described by a dimer molecular weight, M , = 51 OOO 1500. The partial specific volume was calculated from data given in Cohn & Edsall (1943) and the amino acid composition (Cunningham et al., 1975). This gave B = 0.731 mL.g-l at 20 OC in agreement with Sumner's measurement (Sumner et al., 1938). The temperature dependence of D was taken to be dB/dT = 3.7 X lo4 mL.g-ldeg? (Bull, 1976). A refractive index increment of 4 fringesmg-l-mL was assumed.
*
Results The result of the purification of native intact Con A from the Sigma product is shown in Figure 1. In the presence of 1% NaDodSO, on polyacrylamide gels, the Sigma Type IV Con A shows the pattern described by Wang et al. (1971) as being typical for Con A purified by crystallization or as per Olson & Liener (1967). The Sigma product appears to contain about equal amounts of native polypeptide chains and hydrolyzed fragments. The NH,HCO,-purified fraction is substantially free of fragments. The reversibility and stoichiometry of the self-association reaction were tested by high-speed sedimentation equilibrium of a single Con A sample a t pH 6.70 which was run to equilibrium at 25 OC and again at 20 OC. The pH was adjusted by dialysis to 7.30, and the experiment was repeated. Typical molecular weight average distributions (Figure 2) are independent of the initial protein concentration as shown by the superimposability of the curves in the figure, indicating a chemically reversible equilibrium (Teller, 1973). There is a smooth transition in the molecular weight moments between the analytical dimer (51 OOO) and tetramer (lOZOOO) molecular weights as a function of concentration. At each experimental condition, the two-component plot of molecular weight aver-
3078
BIOCHEMISTRY
SENEAR A N D TELLER
c
5
0
5
m
-
IO Concentrotion (fringes)
15
20
FIGURE 2: Molecular weight distribution of commercial Con A obtained by sedimentation equilibrium. Data presented are computer-generated plots of molecular weight vs. concentration (fringes) from analysis of a Rayleigh plate. Initial loading concentrations of approximately 0.5, 1.0, and 1.5 mg/mL were centrifuged for 17 h at 22000 rpm and 20 OC in 0.14 M sodium phosphate, pH 6.70, and 0.2 mM CaC12and ZnC12. The solid curves represent the best fit to the data with a dimer molecular weight, M2 = 51 000, calculated according to Teller (1973). This gave In k2 = 10.06 f 0.05rms. M,, M,, and M, represent the number-average molecular weight, the weight-average molecular weight, and the z-average molecular weight.
ages developed simultaneously by Horbett (Teller et al., 1969) and Roark (Roark & Yphantis, 1969) did not indicate the presence of monomer, trimer, or polymer species larger than the tetramer. Dyson analysis of the data (Van Holde et al., 1969; Teller, 1973) where number, weight, and z averages were used simultaneously to determine equilibrium constants found zero or negative constants for the formation of polymers larger than tetramers. The data were best described by a model that included only dimer and tetramer species in a single, rapid, reversible equilibrium, confirming the hypothesis that such an equilibrium exists under these conditions (Huet, 1975). Attempts were made to include the effect of nonideality in the Dyson calculations. These invariably produced negligibly small or negative second virial coefficients. For the concentration range of the sedimentation equilibrium experiments, Le.,
