TIBS 1 6 - SEPTEMBER 1991

LETTERS Affinity of guanine nucleotide binding proteins for their ligands: facts and artefacts Guanine nucleotide binding proteins play an impo~.ant role in biology. They act as molecular switches in such seemingly diverse cellular processes as growth.receptor-mediated signal transduction, polypeptide elongation and protein transport. The switch between the GTP bound 'on' state and the GDP bound 'off' state is mediated by an intrinsic GTPase activity. As our understanding of the biological processes controlled by these proteins increases, one error is made repeatedly in their characterization. This error concerns the affinity between the proteins and their controlling cofactors GDP and GTP. in the cases that we have investigated, the reported affinities are wrong by factors of up to 10 000 fold! Guanine nucleot~de binding proteins (for recent reviews see Refs 1,2) have, in most cases, a high affinity for GDP/GTP. Thus, G proteins such as elongation factor Tu .(EF-Tu), Gsa or the H-ras protein (o21) are usually isolated with one mole of guanine nucleotide bound per mole of protein 3-6. Because of this, the determination of the binding affinity is obviously not straightforward and has resulted in many errors. It arises because a labeled Iigand is titrated against a protein that is effectively 100% saturated with another ligand at the beginning of the titration and because the degree of saturation does not change significantly throughout the experiment. Most investigators have incubated the guanine nucleotide binding protein-GDP complex, at some arbitrary concentratiolt, with varying

concentrations of a radioactive nucleotide untl~!equilibrium is r~ached and then measured the |,abeled nucleotide bound to the protein. Obviously, this procedure only measures th~ exchange of unlabeled with labeled ~i~g~md.Figure la shows a titration curve obtaiaed on addi,ng [3H]GDP to p21-GDP at a protein concentration of approximately 8 nM. The data points follow an approximately hyperbolic course, and fitting equation 1 (see Fig. 1 legend) to the data gives a value of 7.3 nM for the p21-GDP dissociation constant (solid line). Fitting the correct equation (2) gives the dotted line in Fig. la, which, with the quality of

data obtained by this technique, is not obviously a better fit. The same experiment with 26 nM (not shown) and 248 nM p21 (Fig. Ib) leads to significant shifts of the binding curves to higher concentrations. Thus, increasing the p21 concentration increases the value of the apparent dissociation constant. The reason for this is easy to understand. The value of the dissociation constant obtained from such curve fitting corresponds to the concentration of [3H]GDP nucleotide needed to produce 50% of the maximal amount of p21-[3H]GDP that can be achieved at saturation, which is 50% of the total p21 concentration. (There are easier and more appropriate methods to determine protein concentrations,r) What is an appropriate method for determining the binding constant between guanine nucleotide binding proteins and nucleotides? One way would be to prepare nucleotide-free protein (using methods developed by us and others; Ref. 7), incubate it with increasing concentrations of radioactive nucleotide and determine the concentration of free and bound nucleotide by equilibrium dialysis. However, since nucleotide-free guanine nucleotide binding proteins are Rpm

usually unstable for extended periods of time and equilibrium dialysis with nucleotides takes 20-24 h, this is not feasible. Equilibrium methods using spectroscopic signals would have to be done at very low concentrations and would also need long incubation times. A procedure that avoids these problems involves measurements of the rate constants of the association and dissociation reactions. This allows calculation of the affinity constant, since it is simply the ratio of the rate constants for the forward and reverse reactions. This approach has been used by US 7-9 and others .° to determine the affinity between p21 and GDP/GTP. This is found to be in the region of I0 H M-I at 4°C, rather than 107-109 M-l, as reported by numerous authors using the experimental design presented above. While it may seem that the absolute values of these affinity constants are of esoteric interest only, incorrect estimates can lead to incorrect interpretations of the situation in a biological environment. For example, it has been reported for various mutants of p21 (e.g. SI7N) that their affinity for GDP/GTP is reduced by a factor of I000 or more compared to wild-type p21, as

1

Apparent dissociation constants of H-ras p21 for GDP.p21 was incubated at two different concentrations with increasing amounts of radiolabeled GDP in the presence of 1 mM EDTA. After equilibrium was obtained (45 min, 25°C) 10 mM Mg2. was added and the mixture was filtered through nitrocellulosefilters. Fromthe retained bound radioactivity the concentrationof protein-boundlabeled GDP was measured and free labeled GDP was calculated. The r;ata were fitted using a hyperbolic bindingcurve (solid line) [EL2] = [E]o/(1 + K/[L2]) (Eqn 1) where L2 is the radioactive ligand that is titrated, EL2 is its complex with protein E and [Eo] is the total protein concentration. However, this equationneglectsthe presenceof the unlabelednucleotide boundat the start of the titration. If a stoichiometric amount of unlabeled nucleotide (Lt) is present the equation becomes [EL2]= --C[L2]/2 + {(c[L2])2 + 4c[E]0[L2]}t/2/2(Eqn2) where c is the ratio of affinities of L2 and Lt. This describes a curve (dotted line) which appears approximatelyhyperbolic.The indicated apparent dissociation constants were obtained from equation (1). These are seen to vary with the concentrationof p21 usedfor the titration, and are of the same order as the protein concentration used. The true KD is 22 pM underthe conditions used7.

(a)

8

• . - - "I"

. . . . . . . . . . . . . . . . . .

6 10

"

4

0 JQ g. a

(3

p21 = 8 nM I Kd = 7.3 nM I

2 I

I

I

I

200

i

I

I

800

6O0

4OO

GDP free (nM)

(b)220 20O 180

.o •

~'160



°

"

" ""



..r"

~'140

~, 120 n ~ 100

~ eo 60 4O 20 0

'

[Kd-- 143nM j I

I

200

=

t

I

I

400 600 GDP free (nM)

=

J

800

327

TIBS 16 - S E P T E M B E R 1 9 9 1 judged from the increase in their dissociation rate constants. This has led to the hypothesis that the biological activity of these particular mutants is due to the proteins being nucleotide-free in the cell. Apart from the fact that nucleotide-free p21 (Ref. 8) would be extremely unstable at 37°C, the argument is wrong for another reason: with a binding constant of 10m-10 n M-~ the protein would still be saturated with guanine nucleotide even if the binding constant dropped 10 000 fold, since the concentration of nucleotide in the cell is of the order of I mM. Thus, to understand the biological role of such proteins, it seems to be important to have reasonably

accurate estimates in each case (within an order of magnitude) of their affinities for guanosine nucleotides. Finally, if we consider it necessary to measure these values in the first place, is it not preferable to do it properly?

Periplasmic space and the concept of the periplasm

spaces; some bacterial species may require more periplasmic activity and consequently, more space than others. Still, recognizing that size may not be absolute, general measurements of periplasmic thickness should not be so extreme as to render them meaningless. Having observed numerous bacterial species by various electron microscopic techniques, we feel capable of discussing the size of the periplasmic space. Our measurements consistently suggest that this space is greater than 7 nm but less than 50-71 nm in thickness (see previous discussions in Refs 1,2). Thin-section electron microscopy of chemically fixed (usually glutaraldehyde and/or osmium tetroxide) E. coil cells has provided most microscopic measurements of periplasmic thickness. However, these measurements are only approximate, since this fixation frequently produces a wavy outer membrane• Our experience with conventionally fixed E. coli cells suggests a relatively empty space, c. 11-15 nm in thickness, containing a thin

in recent letters to TIBS, the size of the periplasmic space in Gram-negative eubacteria has been questioned ~,2.in Gram-negative cell envelopes, the periplasmic space is considered to be the region between the outer face of the plasma membrane and the inner face of the outer membrane 3. One of the main problems in the debate so far is the conceptual view of the 'periplasmic space' and the 'periplasm'. These terms are often used synonymously with one another when they can be quite diflerent3, Periplasm implies substance and functionality and is dynamic; perlplasmic space is the physical region that contains most, but not necessarily all, of the periplasm, In Gram-negative eubacteria some enzymes and molecules may be so intimately associated with plasma or outer membranes that they may be partially embedded within them. By this reasoning, Gram-positive eubacteria and archaebacteria have a periplasm (much of their periplasmic material is associated with either the plasma membrane or the wall fabric), but maybe no periplasmic space (Gram-positive bacteria do not have an outer membrane to clearly delineate the outermost boundary of this space). The design of walled prokaryotic cells implies, in a functional sense, that all must have a periplasm. The contours of an E. co~~cell must be quite malleable; depending on the growth conditions, the mean cell volume can differ by a factor of five4. Clearly, if the periplasm is of a dynamic nature, reflecting at least to some extent the metabolic activity of the cell, it is possible that the dimensions of this space are alterable too. Furthermore, it is probable that bacteria in taxons other than E. co/i could have differently sized periplasmic

328

References I Bourne, H. R., Sanders, D. A. and McCormick,F. (1990) Nature 348, 125-132 2 Bourne, H. R., Sanders, D. A. and McCormick,F. (1991) Nature 349, 117-127 3 Ferguson,K. M., Higashijima,T., Smigel, M. D. and Gilman,A. G. (1986) ./. BioL Chem. 261, 7393-7399 4 Miller, D. L. and Weissbach, H. (1970) Arch. Biochem. Biophys. 141, 26-37 5 Poe, M., Scolnick,E. M. and Steitz, R. B. (1985)



~

'

/

,

:

J. BioL Chem. 260, 3906-3909 6 Tucker, J. et al. (1986) EMBOJ. 5, 1351-1358 7 John, J. et al. (1990) Biochemistry 29, 6058--6065 8 Feuerstein, J., Goody, R. S. and Wittinghofer, A. (1987) J. BioL Chem. 262, 8455-8458 9 Feuerstein, J., Kalbitzer, H. R., John, J., Goody, R.S. and Wittinghofer, A. (1987) Eur. J. Biochem. 162, 49-55 10 Neal, S. E., Eccleston,J. F., Hall, A. and Webb, M. R. (1988) J. BioL Chem. 263,19718-19722

ROGERS. GOODY,MAI"rHIASFRECH AND ALFREDWITrlNGHOFER Abteilung Biophysik, Max-PlanckInstitutfor MedizinischeForschung,Jahnstrasse29, D-6900 Heidelberg, FRG.

peptidoglycan layer c. 2.5 nm wide 3. Cryofixation is a better method of accurately preserving the spatial limits of the periplasmic space. Early attempts using freeze-etch replicas were difficult to interpret, often used cryoprotectants (which can perturb lipid packing order in membranes) and did not recognize the importance of vitreous ice (which is an index of accurate preservation). Still, measurements of cross fractures through cell envelopes s-7 suggest a greater width (18-22 rim) for the periplasmic space than that observed by conventional embedding methods. Freeze-substitution, which combines the advantages of cryofixation with those of thin sectioning has altered our perception of the periplasmic space and the periplasm in E. co//. These Images show the periplasmic space to span 13 nm (Ref. 8) and to be entirely packed with visible substance, so much so that clear differentiation betweenthe peptidoglycan and periplasmic substance or 'periplasmic gel' is difficult. We believe that freeze-substitution provides a more

/

:

~

:, •

~:

.

.

.

Rgum 1 Thin section of cryo-fixed, freeze-substituted E. co/i K-12 cell. The envelope profile shows the thick, electron-dense periplasmicgel (arrow)typicallyobserved in this species. Bar represents 100 nm.

.

Affinity of guanine nucleotide binding proteins for their ligands: facts and artefacts.

TIBS 1 6 - SEPTEMBER 1991 LETTERS Affinity of guanine nucleotide binding proteins for their ligands: facts and artefacts Guanine nucleotide binding p...
711KB Sizes 0 Downloads 0 Views