Eur. J. Biochem. 71, 167- 170 (1976)
Further Studies on the Binding of Maltose to the Maltose-Binding Protein of Escherichia coli Maxime SCHWARTZ, Odile KELLERMANN, Sevec SZMELCMAN, and Gerald L. HAZELBAUER Unite de Genetique Moleculaire, Departement de Biologie Moleculaire, Institut Pasteur, Paris and the Wallenberg Laboratory, University of Uppsala (Received August 12/September 20, 1976)
Extensive binding studies, performed by equilibrium dialysis, suggest that the periplasmic maltosebinding protein of Escherichia coli is homogeneous with respect to its affinity towards maltose. This finding contradicts an earlier conclusion, which was based on insufficient data.
A protein binding maltose and maltodextrins can be released from cells of Escherichia coli [l],using the osmotic shock procedure of Neu and Heppel [2]. This protein, which is the product of gene malE has been purified to quasi-homogeneity and has been shown to be an essential element in maltose transport [3] as well as in maltose chemotaxis [4]. Studies performed in our laboratories [3- 51, suggested that the maltose-binding protein could exist in two conformations, differing in their affinity towards maltose. The results of equilibrium dialysis experiments yielded apparently non-linear, biphasic Scatchard plots with two slopes corresponding to dissociation constants of 1.5 and 10 pM, as if purified binding protein were a mixture of two functional types of molecules endowed with different affinities for maltose. Furthermore, the kinetics of maltose exit from dialysis bags containing binding protein was also biphasic and this was thought to reflect the dissociation of maltose from the two classes of sites. Only a single class of sites, characterized by a constant of 1 - 1.5 pM, is revealed when one follows the alteration of the intrinsic fluorescence of the protein upon addition of maltose. However this was interpreted to indicate that only binding to the highaffinity site induces the alteration [6]. The shock fluids from all malE mutants inactive in maltose transport and chemotaxis are devoid of maltose-binding activity [3,4] and some of them contain inactive binding protein which can be recognized immunologically [3]. This strongly suggested that the two classes of sites were carried by the same gene product. It was tempting to speculate that an interconversion between the two conformations of the protein occurred in vivo, played an important role in the transport process, and was perhaps catalyzed by
the products of genes other than malE (i.e. malF, malK and lamB) known to be involved in maltose transport [6,7]. The fact that another periplasmic binding protein had also been reported to display biphasic substrate binding [8,9] made the speculation particularly attractive. However, the results presented in this paper shed considerable doubt on the existence of two classes of sites on the maltose-binding protein. MATERIALS AND METHODS Materials Three different batches of [U-'4C]maltose (The Radiochemical Center, Amersham, England) have been used. The specific activities were 7, 10 and 7.9 Ci/mol. The concentration of maltose in the solutions was determined using the Somogyi/Nelson reaction [lo] or the method of Park and Johnson [ l l ] . At least 99 % of the radioactivity in the solution was verified to be in maltose using thin-layer chromatography (solvent: propanol/ethyl acetate/water, 7/1/2) and autoradiography. Bacterial strains and media were described previously [3,12]. The relevant genotype of the strains is given in the legend of Fig. 1. Preparation of Maltose-Binding Protein The shock fluids were obtained as described previously. Even though this later appeared unnecessary, an extensive dialysis step was introduced on crude or five-fold concentrated shock fluid, in order to remove any potential 'hidden ligand' [13]. A further 10-20fold concentration was then realized on PM 10 Amicon ultrafiltration membranes so that the final extract
168
Maltose Binding to a Periplasmic Protein of E. coli
Fig. 1. Scatchardplots ofmaltose binding by shockfiuids from various strains. Unless otherwise stated all dialyses were performed in buffer A at 4 "C, as described in Materials and Methods, and no correction was made for variations in the protein concentration in the different bags. (A) Maltose-grown HfrG6 (wild type). Protein concentration is 3 mg/ml. (B) Maltose-grown HfrG6 (wild type). Protein concentration is 4.5 mg/ml. Dialyses was performed in buffer C. A correction was made for observed variations in protein concentration in the different bags, as determined by Azso. (C) Maltose-grown pop 1080 (ochre lamB mutant). Protein concentration is 4.2 mg/ml. (D) Glycerol-grown pop 1749 (amber malF mutant). Protein concentration is 7.2 mg/ml. (E) Glycerol-grown pop 1761 (amber malK mutant). Protein concentration is 5 mg/ml. (F) Glycerol-grown pop 1730 (deleted for part of malK and IamB). Protein concentration is 5.6 mg/ml. The experimental points obtained from the two bags placed in the same flask are shown joined by a bar (-0). The symbol 0 indicates that the two points fall in the same place. When a single point is shown (0)one of the duplicate has been lost
M. Schwartz, 0. Kellermann, S. Szmelcman, and G. L. Hazelbauer
169
Fig. 2. Scatchardplots ofmaltose binding by purified binding protein from HfrGd andpop 1730. ( A )Protein from HfrG6 (22 pM). (B) Protein from pop 1730 (36 pM). Same conditions and symbols as in Fig. 1 . The concentration of protein inside the bags was calculated from the Azso assuming that the protein is pure, that its specific absorbance is A$Gg""'= 1.47, and that its molecular weight is 40000 [3]
contained between 6 mg and 12 mg of protein per ml, as measured by the technique of Lowry [14]. In some cases this extract was diluted 2- 3-fold with buffer before use. The purification of binding protein was done as previously described [3] except that 2-mercaptoethanol was omitted from all buffers. The additional dialysis step described above was not performed during the preparation of the extracts from which binding protein was purified. Electrophoresis on dodecylsulfate-polyacrylamide gels revealed the existence in the preparation of two contaminating proteins which together represented at most 10% of the total.
Binding Assay Unless stated otherwise the dialyses were in the same buffer A as used by Hazelbauer [4,5] (10 mM Tris pH 7.3, 2 mM MgCl2, 3 mM NaN3). In one experiment we used buffer C (50 mM sodium phosphate pH 7.0, 0.2 M NaCl, 3 mM NaN3). 18 50-ml conical flasks were each filled with 10 ml of buffer containing [14C]maltose at concentrations ranging from 0.25 pM to 75 pM. Two identical dialysis bags (SVIP Paris), prepared as previously described [3], and from which liquid was drawn as much as possible using Kleenex paper, were filled with 100 p1 of binding protein solution (previously dialyzed against the same buffer as used for the experiment) and put into each flask. Dialysis was allowed to occur for 15- 18 h at 4 "C or at 22-24 "C with gentle shaking. Equilibrium was attained under these conditions, even at the highest protein concentrations used in this work. Duplicate 25-p1 samples were removed (using Corning
micropipettes) from the outside medium, for determination of free maltose, and from each of the two bags, for determination of bound maltose plus free maltose. The samples were transferred into 10 ml of Bray's solution, and counted for 10 min in an Intertechnique scintillation counter. The duplicate values from the same bag, or from the outside medium, almost never differed by more than 5 % , and were averaged. The values from the two bags from the same flask, on the other hand, often differed more appreciably, and are both shown on the plot. In several experiments an extra 25-pl sample was removed from the bags, and diluted in 1 ml of buffer, for determination of the A2so. The values obtained were generally constant f 6 %. Since the variation was not much greater than could be excepted from the error in pipetting, no correction was applied (except in one experiment, Fig. 1 B) which would correspond to variations in the concentration of protein in the bags.
RESULTS AND DISCUSSION The binding of maltose to binding protein was studied by equilibrium dialysis, using shock fluids from HfrG6 (wild-type) as well as from malF, malK and lumB mutants (Fig.1). Some experiments were performed using purified binding protein from either HfrG6 or a strain deleted for malK and lamB (Fig. 2). The precision of the determination is illustrated by showing on the Scatchard plot [15] the two values obtained from identical bags which had been placed in the same flask. Given the variation between duplicate values, there is no compelling reason to draw a
170
M. Schwartz, 0. Kellermann, S. Szmelcman, and G. L. Hazelbauer: Maltose Binding to a Periplasmic Protein of E . coli
complicated curve through the points. The data is most simply represented by a single line, indicating a single binding site. The same result was obtained in further binding experiments not shown'here, ten performed at 4 "C, and nine at 21 -24 "C. The apparent dissociation constant of the bindingprotein . sugar complex, as determined from the slope of the Scatchard plot, averaged 3.5 pM for binding at 4 "C (range 1.9- 5 pM) and 2.2 pM for binding at 21 -24 "C (range 1.9- 2.9 pM). These values are in reasonable agreement with the value of 1-1.5 pM determined by spectrofluorometry at 21 "C. The concentration of binding sites, determined from the plots in Fig. 2, is lower than the concentration of protein molecules (80% in Fig.2A and 75% in Fig. 2B). This difference probably simply reflects the use of purified protein which is less than 100 % active protein and the error generated by the use of approximate values for molecular weight and specific absorbance [3]. We are then left with the question of what happened to 'biphasic binding'. Two lines with different slopes can be drawn through the data points of previously published binding experiments as well as some of the experiments shown here. However, the line which would correspond to the low-affinity binding is determined by data which is subject to significant error along the horizontal axis. We thus believe that the deviation from linearity, apparent in some experiments, results from variation inherent in the binding experiments and the graphical analysis. In addition the biphasic kinetics found for the exit of maltose from dialysis bags containing binding protein are in fact compatible with the existence of a single class of binding sites [16]. Therefore the simplest interpretation of all available data, whether obtained by dialysis
or by spectrofluorometry, is that the maltose-binding protein of E. coli is homogeneous with respect to the binding of maltose. This protein appears to have a single binding site characterized by a dissociation constant of 3-4 pM at 4"C, and 1.5-2.5 pM at 21 - 24 "C. The excellent technical assistance of Madeleine Jolit and Celestine Derval is gratefully acknowledged. This investigation was supported by Grant 927 from the North Atlantic Treaty Organization, by Grant 75.7.0039 from the Delegation GPntrale a la Recherche Scientifique et Technique, as well as Grants from the Centre National de la Recherche Scientifique and the Commissariat a I'Energie Atomique, and the Swedish Naturul Sciences Research Council.
REFERENCES 1. Hazelbauer, G. L. & Adler, M. S . (1971) Nut. New Biol. 230, 101- 104. 2. Neu, H. C. & Heppel, L. A. (1965) J . Biol. Chem. 240, 36853692. 3. Kellerman, 0. & Szmelcman, S. (1974) Eur. J.Biochem. 47,, 139-349. 4. Hazelbauer, G. L. (1975) J . Bacteriol. 122, 206-214. 5. Hazelbauer. G. L. (1975) Eur. J . Biochem. 60, 445-449. 6. Szmelcman, S., Schwartz, M., Silhavy, T. J. & Boos, W. (1976) Eur. J. Biochem. 65, 13- 19. 7. Hofnung, M. (1974) Genetics, 76, 169-184. 8. Boos, W., Gordon, A . S., Hall, R. E. & Price, H. D. (1972). J. Biol. Chem 247, 917-924. 9. Silhavy, T. J. & Boos, W. (1975) Eur. J . Biochem. 54, 163- 167. 10. Somogyi, M. (1945) J . Biol. Chem. 160, 61. 11. Park, J. T. &Johnson, M. S. (1969) J . Mol. B i d . 181,149- 151. 12. Hofnung, M., Jezierska, A. & Braun-Breton, C. (1976) Mol. Gen. Genet. 145, 207-213. 13. Richarme, G. & Kepes, A. (1974) Eur. J . Biochem. 45,127- 133. 14. Lowry, 0. H., Rosebrough, M. J., Farr, A. J. & Randall, R. J. (1951) J . Biol. Chem. 193, 265-275. 15. Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51, 660-672. 16. Silhavy, T. T J., Szmelcman, S., Boos, W. & Schwartz, M. (1975) Proc. Natl Acad. Sci. U.S.A. 72, 2120-2124.
M. Schwartz, 0. Kellermann, and S. Szmelcman, Unite Ghetique Moleculaire, Departement de Biologie Moltculaire, Institut Pasteur, 25 Rue du Docteur-Roux, F-75224 Paris-Cedex-15, France G . L.Hazelbauer, Wallenberg Laboratoriet, Uppsala Universitet, Box 562, S-751 22 Uppsala, Sweden
Note Added in Proof (November 26, 1976). It should be noted that binding of galactose to galactose-binding protein, which had appeared biphasic, has been recently shown to be single site-single affinity binding. [Strange, P. G. & Koshland (1976) Proc. Natl Acad. Sci. U.S.A. 73, 762-766; Zukin, R. S., Strange, P. G., Heavy, L. R. & Koshland, D. E., Jr, Biochem. in press.]
Further Studies on the Binding of Maltose to the Maltose-Binding Protein of Escherichia coli Maxime SCHWARTZ, Odile KELLERMANN, Sevec SZMELCMAN, and Gerald L. HAZELBAUER Unite de Genetique Moleculaire, Departement de Biologie Moleculaire, Institut Pasteur, Paris and the Wallenberg Laboratory, University of Uppsala (Received August 12/September 20, 1976)
Extensive binding studies, performed by equilibrium dialysis, suggest that the periplasmic maltosebinding protein of Escherichia coli is homogeneous with respect to its affinity towards maltose. This finding contradicts an earlier conclusion, which was based on insufficient data.
A protein binding maltose and maltodextrins can be released from cells of Escherichia coli [l],using the osmotic shock procedure of Neu and Heppel [2]. This protein, which is the product of gene malE has been purified to quasi-homogeneity and has been shown to be an essential element in maltose transport [3] as well as in maltose chemotaxis [4]. Studies performed in our laboratories [3- 51, suggested that the maltose-binding protein could exist in two conformations, differing in their affinity towards maltose. The results of equilibrium dialysis experiments yielded apparently non-linear, biphasic Scatchard plots with two slopes corresponding to dissociation constants of 1.5 and 10 pM, as if purified binding protein were a mixture of two functional types of molecules endowed with different affinities for maltose. Furthermore, the kinetics of maltose exit from dialysis bags containing binding protein was also biphasic and this was thought to reflect the dissociation of maltose from the two classes of sites. Only a single class of sites, characterized by a constant of 1 - 1.5 pM, is revealed when one follows the alteration of the intrinsic fluorescence of the protein upon addition of maltose. However this was interpreted to indicate that only binding to the highaffinity site induces the alteration [6]. The shock fluids from all malE mutants inactive in maltose transport and chemotaxis are devoid of maltose-binding activity [3,4] and some of them contain inactive binding protein which can be recognized immunologically [3]. This strongly suggested that the two classes of sites were carried by the same gene product. It was tempting to speculate that an interconversion between the two conformations of the protein occurred in vivo, played an important role in the transport process, and was perhaps catalyzed by
the products of genes other than malE (i.e. malF, malK and lamB) known to be involved in maltose transport [6,7]. The fact that another periplasmic binding protein had also been reported to display biphasic substrate binding [8,9] made the speculation particularly attractive. However, the results presented in this paper shed considerable doubt on the existence of two classes of sites on the maltose-binding protein. MATERIALS AND METHODS Materials Three different batches of [U-'4C]maltose (The Radiochemical Center, Amersham, England) have been used. The specific activities were 7, 10 and 7.9 Ci/mol. The concentration of maltose in the solutions was determined using the Somogyi/Nelson reaction [lo] or the method of Park and Johnson [ l l ] . At least 99 % of the radioactivity in the solution was verified to be in maltose using thin-layer chromatography (solvent: propanol/ethyl acetate/water, 7/1/2) and autoradiography. Bacterial strains and media were described previously [3,12]. The relevant genotype of the strains is given in the legend of Fig. 1. Preparation of Maltose-Binding Protein The shock fluids were obtained as described previously. Even though this later appeared unnecessary, an extensive dialysis step was introduced on crude or five-fold concentrated shock fluid, in order to remove any potential 'hidden ligand' [13]. A further 10-20fold concentration was then realized on PM 10 Amicon ultrafiltration membranes so that the final extract
168
Maltose Binding to a Periplasmic Protein of E. coli
Fig. 1. Scatchardplots ofmaltose binding by shockfiuids from various strains. Unless otherwise stated all dialyses were performed in buffer A at 4 "C, as described in Materials and Methods, and no correction was made for variations in the protein concentration in the different bags. (A) Maltose-grown HfrG6 (wild type). Protein concentration is 3 mg/ml. (B) Maltose-grown HfrG6 (wild type). Protein concentration is 4.5 mg/ml. Dialyses was performed in buffer C. A correction was made for observed variations in protein concentration in the different bags, as determined by Azso. (C) Maltose-grown pop 1080 (ochre lamB mutant). Protein concentration is 4.2 mg/ml. (D) Glycerol-grown pop 1749 (amber malF mutant). Protein concentration is 7.2 mg/ml. (E) Glycerol-grown pop 1761 (amber malK mutant). Protein concentration is 5 mg/ml. (F) Glycerol-grown pop 1730 (deleted for part of malK and IamB). Protein concentration is 5.6 mg/ml. The experimental points obtained from the two bags placed in the same flask are shown joined by a bar (-0). The symbol 0 indicates that the two points fall in the same place. When a single point is shown (0)one of the duplicate has been lost
M. Schwartz, 0. Kellermann, S. Szmelcman, and G. L. Hazelbauer
169
Fig. 2. Scatchardplots ofmaltose binding by purified binding protein from HfrGd andpop 1730. ( A )Protein from HfrG6 (22 pM). (B) Protein from pop 1730 (36 pM). Same conditions and symbols as in Fig. 1 . The concentration of protein inside the bags was calculated from the Azso assuming that the protein is pure, that its specific absorbance is A$Gg""'= 1.47, and that its molecular weight is 40000 [3]
contained between 6 mg and 12 mg of protein per ml, as measured by the technique of Lowry [14]. In some cases this extract was diluted 2- 3-fold with buffer before use. The purification of binding protein was done as previously described [3] except that 2-mercaptoethanol was omitted from all buffers. The additional dialysis step described above was not performed during the preparation of the extracts from which binding protein was purified. Electrophoresis on dodecylsulfate-polyacrylamide gels revealed the existence in the preparation of two contaminating proteins which together represented at most 10% of the total.
Binding Assay Unless stated otherwise the dialyses were in the same buffer A as used by Hazelbauer [4,5] (10 mM Tris pH 7.3, 2 mM MgCl2, 3 mM NaN3). In one experiment we used buffer C (50 mM sodium phosphate pH 7.0, 0.2 M NaCl, 3 mM NaN3). 18 50-ml conical flasks were each filled with 10 ml of buffer containing [14C]maltose at concentrations ranging from 0.25 pM to 75 pM. Two identical dialysis bags (SVIP Paris), prepared as previously described [3], and from which liquid was drawn as much as possible using Kleenex paper, were filled with 100 p1 of binding protein solution (previously dialyzed against the same buffer as used for the experiment) and put into each flask. Dialysis was allowed to occur for 15- 18 h at 4 "C or at 22-24 "C with gentle shaking. Equilibrium was attained under these conditions, even at the highest protein concentrations used in this work. Duplicate 25-p1 samples were removed (using Corning
micropipettes) from the outside medium, for determination of free maltose, and from each of the two bags, for determination of bound maltose plus free maltose. The samples were transferred into 10 ml of Bray's solution, and counted for 10 min in an Intertechnique scintillation counter. The duplicate values from the same bag, or from the outside medium, almost never differed by more than 5 % , and were averaged. The values from the two bags from the same flask, on the other hand, often differed more appreciably, and are both shown on the plot. In several experiments an extra 25-pl sample was removed from the bags, and diluted in 1 ml of buffer, for determination of the A2so. The values obtained were generally constant f 6 %. Since the variation was not much greater than could be excepted from the error in pipetting, no correction was applied (except in one experiment, Fig. 1 B) which would correspond to variations in the concentration of protein in the bags.
RESULTS AND DISCUSSION The binding of maltose to binding protein was studied by equilibrium dialysis, using shock fluids from HfrG6 (wild-type) as well as from malF, malK and lumB mutants (Fig.1). Some experiments were performed using purified binding protein from either HfrG6 or a strain deleted for malK and lamB (Fig. 2). The precision of the determination is illustrated by showing on the Scatchard plot [15] the two values obtained from identical bags which had been placed in the same flask. Given the variation between duplicate values, there is no compelling reason to draw a
170
M. Schwartz, 0. Kellermann, S. Szmelcman, and G. L. Hazelbauer: Maltose Binding to a Periplasmic Protein of E . coli
complicated curve through the points. The data is most simply represented by a single line, indicating a single binding site. The same result was obtained in further binding experiments not shown'here, ten performed at 4 "C, and nine at 21 -24 "C. The apparent dissociation constant of the bindingprotein . sugar complex, as determined from the slope of the Scatchard plot, averaged 3.5 pM for binding at 4 "C (range 1.9- 5 pM) and 2.2 pM for binding at 21 -24 "C (range 1.9- 2.9 pM). These values are in reasonable agreement with the value of 1-1.5 pM determined by spectrofluorometry at 21 "C. The concentration of binding sites, determined from the plots in Fig. 2, is lower than the concentration of protein molecules (80% in Fig.2A and 75% in Fig. 2B). This difference probably simply reflects the use of purified protein which is less than 100 % active protein and the error generated by the use of approximate values for molecular weight and specific absorbance [3]. We are then left with the question of what happened to 'biphasic binding'. Two lines with different slopes can be drawn through the data points of previously published binding experiments as well as some of the experiments shown here. However, the line which would correspond to the low-affinity binding is determined by data which is subject to significant error along the horizontal axis. We thus believe that the deviation from linearity, apparent in some experiments, results from variation inherent in the binding experiments and the graphical analysis. In addition the biphasic kinetics found for the exit of maltose from dialysis bags containing binding protein are in fact compatible with the existence of a single class of binding sites [16]. Therefore the simplest interpretation of all available data, whether obtained by dialysis
or by spectrofluorometry, is that the maltose-binding protein of E. coli is homogeneous with respect to the binding of maltose. This protein appears to have a single binding site characterized by a dissociation constant of 3-4 pM at 4"C, and 1.5-2.5 pM at 21 - 24 "C. The excellent technical assistance of Madeleine Jolit and Celestine Derval is gratefully acknowledged. This investigation was supported by Grant 927 from the North Atlantic Treaty Organization, by Grant 75.7.0039 from the Delegation GPntrale a la Recherche Scientifique et Technique, as well as Grants from the Centre National de la Recherche Scientifique and the Commissariat a I'Energie Atomique, and the Swedish Naturul Sciences Research Council.
REFERENCES 1. Hazelbauer, G. L. & Adler, M. S . (1971) Nut. New Biol. 230, 101- 104. 2. Neu, H. C. & Heppel, L. A. (1965) J . Biol. Chem. 240, 36853692. 3. Kellerman, 0. & Szmelcman, S. (1974) Eur. J.Biochem. 47,, 139-349. 4. Hazelbauer, G. L. (1975) J . Bacteriol. 122, 206-214. 5. Hazelbauer. G. L. (1975) Eur. J . Biochem. 60, 445-449. 6. Szmelcman, S., Schwartz, M., Silhavy, T. J. & Boos, W. (1976) Eur. J. Biochem. 65, 13- 19. 7. Hofnung, M. (1974) Genetics, 76, 169-184. 8. Boos, W., Gordon, A . S., Hall, R. E. & Price, H. D. (1972). J. Biol. Chem 247, 917-924. 9. Silhavy, T. J. & Boos, W. (1975) Eur. J . Biochem. 54, 163- 167. 10. Somogyi, M. (1945) J . Biol. Chem. 160, 61. 11. Park, J. T. &Johnson, M. S. (1969) J . Mol. B i d . 181,149- 151. 12. Hofnung, M., Jezierska, A. & Braun-Breton, C. (1976) Mol. Gen. Genet. 145, 207-213. 13. Richarme, G. & Kepes, A. (1974) Eur. J . Biochem. 45,127- 133. 14. Lowry, 0. H., Rosebrough, M. J., Farr, A. J. & Randall, R. J. (1951) J . Biol. Chem. 193, 265-275. 15. Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51, 660-672. 16. Silhavy, T. T J., Szmelcman, S., Boos, W. & Schwartz, M. (1975) Proc. Natl Acad. Sci. U.S.A. 72, 2120-2124.
M. Schwartz, 0. Kellermann, and S. Szmelcman, Unite Ghetique Moleculaire, Departement de Biologie Moltculaire, Institut Pasteur, 25 Rue du Docteur-Roux, F-75224 Paris-Cedex-15, France G . L.Hazelbauer, Wallenberg Laboratoriet, Uppsala Universitet, Box 562, S-751 22 Uppsala, Sweden
Note Added in Proof (November 26, 1976). It should be noted that binding of galactose to galactose-binding protein, which had appeared biphasic, has been recently shown to be single site-single affinity binding. [Strange, P. G. & Koshland (1976) Proc. Natl Acad. Sci. U.S.A. 73, 762-766; Zukin, R. S., Strange, P. G., Heavy, L. R. & Koshland, D. E., Jr, Biochem. in press.]
