A study of the quaternary structure of glycinin

A study of the quaternary structure of glycinin

Biochimica et Biophysica Acta 827 (1985) 119-126 119 Elsevier BBA 32093 A study of the quaternary structure of glycinin M e r v y n J. Miles, Vict...

463KB Sizes 2 Downloads 37 Views

Biochimica et Biophysica Acta

827 (1985) 119-126

119

Elsevier BBA 32093

A study of the quaternary structure of glycinin M e r v y n J. Miles, Victor J. M o r r i s , D a v i d J. W r i g h t a n d J a m e s R. B a c o n AFRC, Food Research Institute, Colney Lane, Norwich, NR4 7UA (U.K.)

(ReceivedAugust24th, 1984)

Key words: Glycinin;11 S Globulin;Storageprotein; Quaternarystructure; Small-angleX-rayscattering;(Soybean seed)

A range of six- and twelve-subunit models for glycinin, in which the subunits are assumed to be identical and spherical, has been tested by comparing experimental sedimentation, translational diffusion and small-angle X-ray scattering data with theoretical predictions for each model. None of the models considered, including the stacked parallel hexagon model described by Badley et al. (Badley, R.A., Atkinson, D., Hauser, H., Oldani, D., Green, J.P. and Stubbs, J.M. (1975) Biochim. Biophys. Acta 412, 214-228), or the trigonal antiprism proposed by Plietz et al. (Pleitz, P., Damaschun, G., Mtiller, J.J. and Schwenke, K.-D. (1983) Eur. J. Biochem. 130, 315-320), provides acceptable interpretations of the sedimentation, diffusion and scattering data.

Introduction The 7 S and 11 S globulins are the major storage proteins of the seeds of many plant species. The 11 S globulins apparently consist of a range of oligomeric proteins with reported sedimentation coefficients between 10.8 S and 14.6 S. The quaternary structure of the 11 S globulins is stable but they can be induced to dissociate according to the following scheme [1,2]. 11 S (A6B6) ~ 7 S (2A3B3) -, 3 S (6AB) --, 2 S (6A+6B) Symbols contained within the brackets are intended to indicate the extent of 'subunit' interaction in terms of six acidic (A) and six basic (B) subunits. The conditions required to produce dissociation vary depending on the source of the protein. Dissociation of the 3 S protomer into the 2 S monomers generally requires the presence of a reducing agent to cleave disulphide bridges. Thus, it can be argued that the 3 S rather than the 2 S form of the 11 S globulin represents the funda0167-4838/85/$03.30 © 1985 ElsevierSciencePublishersB.V.

mental unit and, therefore, that the 11 S globulins should be modelled on the basis of six, not twelve, subunits. Knowledge of the quaternary structure of storage proteins is of importance to food scientists interested in understanding and exploiting the functional properties of seed proteins and to molecular biologists interested in understanding and controlling plant biosynthesis of such proteins. The 11 S globulin of soyabean is often referred to as glycinin. Two alternative quaternary structures have been proposed [3-5]. Catsimpoolas [3] and Badley et al. [4] have proposed a twelve-subunit model in which the acidic and basic subunits alternate in a two-layered parallel hexagonal model (Fig. 1A). Plietz et ~1. [5] have suggested a structural homology between the 11 S globulins from rape, soya bean and sunflower and proposed a six-subunit model in which the six (AB) subunits are arranged in a trigonal antiprism structure (Fig. 1B). In this paper, we have considered various plausible twelve- and six-subunit models for

120

glycinin. These models have been tested and compared by calculating theoretical sedimentation and translational diffusion coefficients and comparing these values with the experimental data available in the literature. In addition, the early small-angle X-ray scattering data of Badley et al. [4] have been extended and the experimental curves compared with theoretically calculated scattering curves for the various models. The merits and demerits of the various models are discussed.

Experimental

Preparation of glycinin. Soya bean meal (Hopkins & Williams, Chadwell Heath, Essex, U.K.) was defatted with hexane and then extracted with 5 vol. water for 1 h at room temperature. After centrifugation, the supernatant was stored at + I°C overnight and the resultant cryoprecipitate collected by centrifugation. The latter was redissolved in starting buffer (0.1 M KH2PO4 (pH 7.6)/1 mM dithiothreitol/0.02% NAN3) and applied to a column (4.4 × 55 cm) of HA Ultragel (LKB Instruments, S. Croydon, Surrey). Elution was performed at a flow-rate of 80 cm3. h-1 and after 1800 cm3, the column was developed with a 24-h linear gradient to 0.3 M KH2PO4 (pH 7.6)/1 mM dithiothreitol/0.02% NaN 3. Fractions corresponding to glycinin were pooled, concentrated by ultrafiltration and stored frozen at -20°C. Small-angle X-ray scattering (SAXS). Small-angle scattering curves were measured using an automatic Kratky camera (Anton Paar, Austria) fitted with a proportional counter. A Philips PW 1730 generator provided a highly-stabilized X-ray source at a wavelength (~,) of 0.154 nm. Glycinin solutions were prepared in 0.05 M NaH2PO 4 (pH 7.0) buffer/1% NaC1/1 mM dithiothreitol/0.02% NaN 3. Six concentrations were studied in the range 1.625-50.0 mg. m1-1. Scattered intensities were measured for the globulin solutions contained in quartz capillaries and for buffer solutions in the capillary tubes over the range h = 0.2448 nm-1 to 3.6725 nm -1 (h = (4~r/X)sin 0 where 20 is the scattering angle). Background scattering was subtracted prior to analysis of data. Experimental data was smoothed and desmeared by the indirect-transform method described by Glatter [61 using the criteria set out by Mi~ller and Glatter [7].

Additional small-angle scattering studies were made on a 50 mg. ml-1 glycinin solution using the SERC Daresbury synchrotron radiation source (X = 0.1608 nm) and a one-dimensional position-sensitive detector. Data were collected for 10 min. Background scattering from the buffer and the capillary tube were recorded and subtracted from the scattering curve for the protein solution contained in the capillary tube to obtain the scattering curve for the protein solution. Scattering curves were normalised using the total flux entering the specimen during the measurement of each scattering curve, as determined by an ionisation counter. A collagen fibre (64 nm repeat) was used to calibrate the camera length and the angular spacing of the position-sensitive detector channels.

Theoretical

Calculation of sedimentation and translational diffusion coefficients The sedimentation coefficient S and the translational diffusion coefficient D are given by the equations [8]:

s

MO- ~o)

Nf

kT

O = -f

(1) (2)

The quantities ~, M, p, N, k, T and f are the partial specific volume, the molecular weight of the protein, the solvent density, Avogadro number, Boltzman constant, the absolute temperature and the translational frictional coefficient, respectively. For the spherical protein of radius a: 4rra 3 3

M

N

(3)

and from Stokes law [8]: f= 6~0a

(4)

where ~0 is the solvent viscosity. Thus, from Eqns. 1, 3 and 4: S~, (1-~0)

N -2/3 2.3[ 4~r ,~1/3 ~ M / ~T) =KM2/3

(5)

121

where K = 0.012 S. cm. g-a • mo1-2/3. Experimentally, it has been observed [9,10] that: S~1/3

(a-~p)

K ' M 2/3

(6)

Thus, from Eqns. 2 and 10:

D.

kT 7.2~ona I

r,;1

1+-n

i-1 j=l

kTX

(11)

7.2~r'~ona 1

i4. j

where K' = 0.010 S. cm. g-1. mo12/3. This discrepancy has been attributed [10] to surface roughness (rugosity). It has been suggested [10] that surface roughness may be approximated empirically by using a frictional coefficient f = 1.2fth~or. Bloomfield and co-workers [11] have adapted the formulations of Kirkwood [12] to cover solid 'rigid' objects composed of subunits. It will be assumed that the subunits are spherical and identical. For such a model, the sedimentation coefficient of the oligomer (S~) is related to the sedimentation coefficient ($1) and the frictional coefficient ( f l ) of the monomer or subunit: S.= (S~nfl)/f.

(7)

Following Teller and De Hahn [9,10]: S . = ( 1 - r P ) O'OlOMff/3nl/3( ~

(8a)

_ ( 1 - ~p) 0.010M,2/3F ~ ~1/3

(8b)

where M, is the molecular weight of the oligomer

(= nM1) and Fn is a geometric factor dependent on the number and spatial distribution of subunits. In the Kirkwood formulation [12,14]: Fn=n -2/3 1 + __ ai

riJ 1 = 71- 2/3X

n i ~1 j = l

(9)

]

where a~ is the spherical radius of the subunit ( = (3M~/4~rN) 1/3) and r,j is the separation of the ith and j t h monomers. This is equivalent to replacing f = 6 ~ 0 a n by: -1

r,71 i=1 j =1

i¢j

(10)

Calculations of /9, and Sn for various subunit arrangements involved the following steps. The number of subunits was chosen. The molecular weight of the oligomer was calculated from experimental values of D and S by eliminating f from Eqns. 1 and 2. The subunits were considered to be identical and M 1 was calculated. For spherical subunits, a 1 was determined and X computed for a known subunit distribution. Values of S~ and D~ were then found using Eqns. 8b, 9 and 11.

Small-angle X-ray scattering Theoretical scattering curves were calculated for different distributions of subunits using the Debye method [15]. The subunits were taken to be spherical and identical. For a given arrangement of subunits, the radius of the subunit was adjusted until the radius of gyration of the oligomer was equal to the experimentally determined value. This effectively normalised the theoretical and experimental curves at low scattering angles.

Results The subunit arrangements investigated are shown in Fig. 1. The model proposed by Catsimpoolas [3] and Badley et al. [4] is shown in Fig. 1A. Fig. 1B represents the trigonal antiprism model proposed by Plietz et al. [5]. Fig. 1D and E are natural variations of models 1A and lB. Fig. 1C and F represent twelve-subunit models intended as a first approximation to trigonal antiprism (1B) and trigonal prism (1E) models with asymmetric subunits extended radially. Provided the acidic and basic subunits in the twelve-subunit models are chemically distinct, then all the models possess the symmetry 32. The mirror plane is intended to reflect the dissociation behaviour 11 S --, 7 S. Badley et al. [4] quote Sn = (12.3 5: 0.1)S and Dn = (3.44 + 0.1). 10 -11 m E. s -1. Assuming v = 0.73 and p = 1.018 g. cm -3 and using Eqns. 1 and 2, this yields M = (3.22 + 0.15)- 105. Table I shows

122

i

Fig. 1. A, Two-layered parallel hexagonal model; B, trigonal antiprism model; C, two-layered parallel star model; D, twolayered rotated hexagonal model; E, trigonal prism; F, twolayered rotated star model.

TABLE I CALCULATED SEDIMENTATION (S) AND DIFFUSION (D) COEFFICIENTS FOR THE MODELS SHOWN IN FIG. 1 Model (Fig. 1)

S (S)

D (m2. s- 1)( × 1011)

A B C D E F

12.8±0.4 14.3±0.4 12.2±0.3 13.2±0.4 13.7±0.4 12.3±0.3

3.57±0.06 4.~±0.07 3.41±0.06 3.68±0.06 3.82±0.06 3.~±0.06

the calculated values of S~ and Do for the various models. In addition, Table II shows the values of So and D, predicted for the two fragments obtained after dissociation along the mirror plane. The models C and F (Fig. 1) yield calculated So and D~ values which agree to within the estimated accuracy with the experimental values. Judged by this criterion alone, these twelve-subunit models provide the best approximation to the quaternary structure. The parallel stacked hexagon model (1A) is just acceptable, but the remaining models yield unacceptably high values for both Do and So. The trigonal prism (1E) and the antiprism (1D) would also yield a vary large value for So for the two dissociated fragments (Table II). The small-angle X-ray scattering curve extrapolated to zero concentration resulted in the Guiner plot shown in Fig. 2. The z-average radius of gyration (Rg) calculated from the slope of the plot was 4.42 + 0.06 nm and the weight average molecular weight (Mw) was (3.18 _+ 0.23)- 105. The calculated molecular weight is in good agreement with the values reported by Badley et al. [4]. The value of Rg is similar to that reported by Badley et al. [4] (4.4 nm). Fig. 3 shows data obtained over the full range of angles measured for a glycinin concentration of 25 mg. cm -3. Also shown are the smoothed and desmeared curves calculated using the indirect transform algorithm. The desmeared curve shows a submaximum at h = 1.0 nm -1 with a shoulder on the higher angle side of the maximum and two separate submaxima at higher angles, indications of which are apparent in the non-desmeared data also shown in Fig. 3. Badley et al. also observed a shoulder on the first submaximum in their desmeared data, but they do not report measure-

TABLE II CALCULATED SEDIMENTATION (S) AND DIFFUSION (D) COEFFICIENTS FOR THE FIRST DISSOCIATION PRODUCTS OF THE MODELS SHOWN IN FIG. 1 Model (Fig. l)

S (Sv~bergs)

D (m2.s-1)(×1011)

A D E

7,6±0.2 7,3±0.2 8,5±0.2

4.25±0.07 4.06±0.07 4.76±0.08

123

3o 1

'~ 2"5

>, 2.0 o

2

I

1'5

100

I

(20) 2 m r a d s 2

I

200

300

Fig. 2. Guinier plot of K r a t k y data extrapolated to zero concentration.

ments of the scattered intensity at higher angles. There must always be some uncertainty in the scattering curves produced by any desmearing procedure. Inappropriate use of desmearing routines may produce submaxima from noise in regions of

the scattering curve where no submaxima are apparent in the initial data. Interpretations which rely heavily on these features of the curves must be treated with caution. Use of synchrotron radiation can, however, obviate the need for desmearing the data as the system is a good approximation to the ideal pinhole geometry and, for most purposes, instrumental smearing may be neglected [16]. In Fig. 4, preliminary results of synchrotron radiation experiments on glycinin (50 mg. cm -3) are compared with the desmeared data obtained using the conventional generator and the Kratky camera. The general features of the curves are, of course, similar. However, the synchrotron data shows a clear submaximum in the same position as the shoulder on the first submaximum observed in the desmeared Kratky data. A suggestion of a third submaximum in the synchrotron data is also seen in the Kratky data. Besides eliminating the requirement to desmear the data, synchrotron radiation has the advantage that measurements may be made very rapidly owing to the very high intensity of the beam. The synchrotron data reported here was collected in about 1/800 of the time required for the Kratky measurements. Clearly, any aggregation and denaturation occurring with time would be more significant in the Kratky studies than for

5.0¸~...~ 3 " 0 : ~ ~n

4"C %

3"0 1"0i -o:4

o:o IOg~oh

0.4 (:n m -~ )

Fig. 3. Plot of 1o810 (intensity) against log]0 h from K r a t k y data ( × ) smoothed data ( . . . . . . ) and desmeared data ( ).

-0"6

-0"3 IOglO h

0"0

0"3

(nm "1)

Fig. 4. C o m p a r i s o n of the desmeared K r a t k y scattering data ( . . . . . . ) and the synchrotron scattering data ( × - × - × ).

124

the synchrotron radiation work. This could be a factor in explaining the observed improvement in the resolution of the submaxima in the synchrotron experiment. Theoretically calculated scattering curves for the models shown in Fig. 1 are compared with the experimental scattering curves in Fig. 5. The starlike models (C, F) yield such poor agreement with

5"0

*~"

the experimental scattering curves that they can be dismissed as implausible, despite the good agreement between calculated and measured sedimentation and diffusion coefficients. All the remaining models (Fig. 5) show good agreement with the experimental data at low scattering angles. They all give a first submaximum in approximately the correct angular position but none of the models

~ -~.

'\ \

,<

4"0

4-0

%

~.

% a~

o

\\

o

)-,

3"0

-0-6

-0"3

IOgloh

0'0

0'3

-0.6

1Oglo h

( n m -1) 5.0

4"0

\,?\

\



o~ 0

3"O

-0"6

-0"3 IoglOh

0"0 ( n m "1)

0"3

(nm "1)

\~ \!~

%

3"0

0"3

d



\ \\ '\

0"0

~'~

4-0

\\

¢m 0

-0"3

\'\

~i

-0"6

-0-3 Iog~oh

0"0

0"3

( n m "1)

Fig. 5. Comparison of the experimental scattering curves with scattering curves calculated for: (a) parallel ( . . . . ) and rotated ( ..... ) hexagonal models; (b) trigonal prism ( - - - - - - ) and antiprism ( . . . . . ) m o d e l s ; (c) parallel ( . . . . ) and rotated ( . . . . . ) star models and (d) spherical model.

125 predicts the correct intensity for this submaximum. None of the models predicts the presence of the second and third submaxima observed in the synchrotron data. The agreement between the calculated and experimentally observed first submaximum improves as the model adopts a more spherical shape. Indeed, the theoretical scattering curve for a simple sphere yields a very close approximation to the position and intensity of the first submaximum (Fig. 5). The spherical model also predicts the second better than any of the subunit models, indicating that the molecule shows only slight departures from a spherical shape. If the molecular models are restricted to six- or twelve-subunit structures in which the subunits are identical and spherical, then none of the models or their variants proposed for glycinin shown in Fig. 1 provide a satisfactory fit to the combined sedimentation, diffusion and scattering data. Of the available models, the stacked parallel hexagons (1A) provide the best fit to the sedimentation and diffusion data and a reasonable fit to the X-ray data. Plietz et al. [5] have refined the trigonal antiprism model for rape and sunflower 11 S globulins by building the molecular structure from tiny spheres arranged on a cubic lattice. They have demonstrated that such refinements can improve the agreement between theoretical and calculated scattering curves. It would be possible to calculate sedimentation and diffusion coefficients for such models using the methods outlined in the present text. Refinements of the present six- and twelvesubunit models by varying the shape of the subunits could be used to improve fits to the scattering, sedimentation and diffusion data. However, such procedures are tedious and time-consuming and there appears to be no obvious systematic approach to such modelling procedures. Therefore, it is intended to collect detailed diffusion, sedimentation and scattering data on a number of 11 S globulins and their dissociation products and to ascertain common features of the globulins before embarking on detailed modelling of the structures. The synchrotron radiation studies reported here justify the belief that scattering data of sufficient quality may be obtained at sufficiently high scattering angles to permit such modelling studies to be carried out with confidence.

Conclusions A variety of six- and twelve-subunits models for glycinin have been tested and compared. None of the models considered provides a satisfactory description of experimental sedimentation, translational diffusion and small-angle X-ray scattering data. The modelling is restricted by the assumption that the subunits are identical and spherical. Procedures are available for refining such models by varying the shape of the subunits. It has been shown that synchrotron radiation studies can be used to obtain unambiguous scattering data showing several submaxima. Such data, together with sedimentation and diffusion data obtained on a number of 11 S globulins and their dissociation products, provide the means of refining the structural model for the quaternary structure of these proteins.

Acknowledgements The authors wish to express their thanks to Dr. C. Nave, SERC Daresbury Laboratory, for his technical assistance with the synchrotron measurements, and to Professor E.D.T. Atkins, Bristol University, for making it possible for these experiments to be performed.

References 1 Derbyshire, E., Wright, D.J. and Boulter, D. (1976) Phytochemistry 19, 3-24 2 Wolf, W.J. in (1977) Food Proteins (Whitaker, J.R. and Tannenbaum, S.R., eds.), Chapter 10, pp. 291-314, AVI Publishing, Wesport 3 Catsimpoolas,N. (1969) FEBS Lett. 4, 259-261 4 Badley, R.A., Atkinson, D., Hauser, H., Oldani, D., Green, J.P. and Stubbs, J.M. (1975) Biochim. Biophys. Acta 412, 214-228 5 Plietz, P., Damaschun, G., Mi~ller,J.J. and Schwenke, K.-D. (1983) Eur. J. Biochem. 130, 315-320 6 Glatter, O. (1977) J. Appl. Cryst. 10, 415-421 7 Mailer, K. and Glatter, O. (1982) Makromol. Chem. 183, 465-479 8 Tanford, C. (1961) Physical Chemistry of Macromolecules, Chapter 6, p. 381, John Wiley & Sons, New York 9 Teller, D.C. and De Hahn, C. (1975) Fed. Proc. 34, 598, abstr. 2143 10 Teller, D.C., Swanson, E. and De Harn, C. (1979) Methods Enzymol. 61, 103-124

126 11 Bloomfield, V.A., Dalton, W.O. and Van Holde, K.E. (1967) Biopolymers 6, 135-148 12 Kirkwood, J.G. (1949) Rec. Trav. Chim. Pays-Bas 68, 649-660 13 Van Holde, K.E. (1975) in The Proteins, 3rd Edn. (Neurath, H., Hill, R.L. and Boeder, C.-L., eds.), Academic Press, New York

14 Kirkwood, J.G. (1954) J. Polymer Sci. 12, 1-14 15 Debye, P. (1915) Ann. Phys. 46, 809-823 16 Miles, M.J., Morris, V.J., Carroll, V., Wright, D.J. and Bacon, J.R. (1984) Int. J. Biol. Macromol. 6, 291-292