Characterization of protein binding. to a nitrocellulose

Transcription

1 Vol. 33. No (293) Characterization of protein binding to a nitrocellulose membrane Kazuyuki Nakamura, Tatehiko Tanaka and Kazusuke Takeo SUMMARY Proteins bound to a nitrocellulose (NC) membrane tightly and quantitatively until the NC membrane was saturated with the proteins. The binding of proteins to the NC membrane was characterized by using twenty proteins having different molecular weight and isoelectric point as follows: 1) hydrophobic interactions between the protein and the NC membrane matrices may play a major role in the protein binding, and sugars, amino acids, DNA, nucleotides, neutral salts, or glycerol in the protein solution did not interfere with the protein binding. 2) The number of the protein molecule bound to the NC membrane was in the range from 1.13 to 1.98nmol/cm2 except for a few small and strongly charged proteins. 3) Non-ionic detergents such as Tween 20, Triton X-100, and Nonidet P-40 strongly interfered with the protein binding to the NC membrane in a concentration dependent manner. 4) Proteins could be extracted from the NC membrane with high yield by using a low concentration of the non-ionic detergent. These enables us to assay the proteins with high sensitivity and reproducibility. Furthermore, it may be possible to determine the amino acid composition and sequence with small amount of the proteins after the assay. Key words: protein binding, hydrophobic interaction, nitrocellulose membrane, non-ionic detergent, protein assay. INTRODUCTION A technique of immobilization of proteins to a nitrocellulose (NC) membrane has been developed to detect specific proteins with a high sensitivity.1-13) The technique has been also applied to the analyses of protein-ligand interactions.2 14,15) However, only a few papers concerning the principles of the binding of proteins to the NC membrane have been published ) Gershoni and Beisiegel have discussed the parameters which affected the interactions between the proteins and the NC membrane matrices in their recent reviews.19,20) They suggested that hydrophobic interactions may play a role in the protein binding to the NC membrane, but still the interactions between proteins and the NC membrane have not been clearly understood. Here, we describe the characterization of the binding of proteins to the NC membrane matrix Correspondence address: Kazuyuki Nakamura, Department of Biochemistry, Yamaguchi University, School of Medicine, 1144, Kogushi, Ube 755, Japan. (Received June 30, 1989, Accepted July 31, 1989)

2 from results obtained by using the technique for microassay of proteins on the NC membrane21) with a series of proteins having different molecular weights and isoelectric points. Furthermore, we examined effects of a series of different types of detergent on the protein binding to the NC membrane to optimize the conditions for extraction of proteins from the NC membrane. MATERIALS AND METHODS Materials Human gamma-globulin (Cohn fraction II), human serum albumin, human fibrinogen, histone (calf thymus, type II-S), cytochrome C (equine heart), concanavalin A, L-lactate dehydrogenases (rabbit muscle and rabbit heart), lysozyme (egg white), trypsin (porcine pancreas), ferritin (horse spleen), sodium cholate, ascorbic acid, adenosine 5'-monophosphate, adenosine 5'-diphosphate, adenosine 5'-triphosphate, and deoxyribonucleic acid (calf thymus, type I) were purchased from Sigma Chemical Company (St. Louis, MO., U.S.A.). Ponceau 3R (CI 16155), sodium dodecylsulfate, and Nonidet P-40 were obtained from Nacalai Tesque (Kyoto, Japan). Urea, Triton X-100, guanidine hydrochloride, polyethylene glycol 200, glycerol, maltose, 2-mercaptoethanol, thioglycolic acid, and amino acids were purchased from Wako Pure Chemicals Industries Ltd. (Osaka, Japan). Octyl-glucoside, CHAPS, and CHAPSO were obtained from Dojindo Laboratories (Kumamoto, Japan). Tween 20 was purchased from Bio Rad Laboratories (Richmond, CA., U.S.A.). Sodium chloride, ammonium sulfate, D-glucose, D-glucosamine, protamine sulfate (salmon sperm), and paraffin liquid werepurchased from Katayama Chemical (Osaka, Japan). Nitrocellulose membranes (HAWP: pore difluoride (PVDF) membrane were obtained from Millipore Corporation (Bedford, MA., U.S.A.). Phosphorylase a was purchased from Boehringer Mannheim (Darmstadt, F. R. G.). All other chemicals were of the highest grade commercially available. Methods Immobilization of proteins onto a nitrocellulose (NC) membrane using dot blotting procedure The protein binding assay was carried out as described in a previous paper. 21) Protein solutions were prepared with 10mM phosphate buffer, ph 7. 4 containing 150mM NaCl (P buffer). One hundred microliters of protein solutions diluted with P buffer to desired concentrations were pipetted into the wells of "Bio Dot" microfiltration apparatus (Bio Rad Laboratories, Richmond, CA., U.S.A.) and filtered under mildly reduced pressure by a water aspirator. Each well The proteins were spotted onto the NC membrane in uniform circles 3mm in diameter. The membrane was removed from the apparatus immediately and incubated in a freshly prepared dye solution (0.1% Ponceau 3R (w/w) in 7% (v/v) acetic acid) for 5min at room temperature. The excess dye was then removed from the membrane with continuous shaking in 7% acetic acid until the background of NC membrane became colorless. The NC membrane was dried on Whatman No. 1 filter paper and immersed into paraffin liquid to achieve translucency. Then the translucent membrane was set into a densitometer (Densitron PAN 802, Jookoo Co. Ltd., Tokyo, Japan) and the optical density (O. D.) of the stained protein spots was measured using a 1-mm slit with a 2-mm width at 500nm. Determination of maximum amount of protein bound to the NC membrane The proteins were spotted onto the double layered NC membrane in uniform circles 3mm in diameter using the microfiltration apparatus as described above. The protein spots on the double layered NC membrane were stained with 0.1 % Ponceau 3R, and their optical density was measured by the densitometer at 500nm. The optical density of the protein spots in the first

3 Vol. 33. No (295) layer was increased proportionally to the increase of the amount of the proteins applied to the wells, unless the amount of the proteins exceeded the capacity of the NC membrane binding to the proteins. When the amount of the proteins exceeded the capacity of the NC membrane, the optical density of the protein spots in the first layer became constant and the protein spots appeared on the second layer. Hence, the maximum amount of proteins bound to the NC membrane corresponds to the amount of the proteins which gives the maximum optical density of the protein spots in the first layer that will not to make a spot on the second layer of the NC membrane. Examination of the effects of chemicals on the protein binding to the NC membrane The effects of chemicals on the protein binding to the NC membrane were examined by using protein solutions (0.1mg/ml of human gammaglobulin in P buffer) containing a series of the concentrations of the chemicals. The protein so- spotted on the NC membrane in uniform circles 3mm in diameter followed by washing the spots sity of the protein spots stained with 0.1 Ponceau 3R was measured by the densitometer at 500nm. The effects of the chemicals were estimated from the magnitude of the changes in the optical density of the protein spots. Examination of the effects of detergents on the protein binding to the NC membrane The protein solutions (0.1mg/ml of human gamma-globulin) were prepared with P buffer containing a series of the different concentrations were blotted onto the NC membrane in uniform circles 3mm in diameter. After washing the stained with 0.1% of Ponceau 3R in 7% acetic acid. Then the optical density of the stained spots was determined by the densitometer at 500 nm. The magnitude of the decrease in the optical density of the spots indicated the strength of the interference with the protein binding to the NC membrane by the detergent in the protein solution. Extraction of proteins from the NC membrane The protein solutions (0.1mg/ml of human gamma-globulin) were prepared with P buffer. the NC membrane in uniform circles 3mm in diameter. After washing the spots twice with bated with P buffer containing a series of different concentrations of a detergent with gentle shaking at room temperature. One hour later the NC membrane was transfered to the solution of 0.1% Ponceau 3R in 7% acetic acid to stain the protein spots. The optical density of the stained protein spots was determined by the densitometer at 500nm. The magnitude of the decrease in the optical density of the protein spots indicated the extractability of the protein from the NC membrane by incubating with the detergent. RESULTS Immobilization of proteins to a nitrocellulose (NC) membrane As shown in Fig. 1, the amount of human gamma-globulin bound to the NC membrane increased proportionally to the increase of the amount of protein in the samples, unless the protein amount exceeded the binding capacity of the NC membrane at both ph 3.7 and ph 7.4. The maximum amount of human gamma-globulin bound to the NC membrane was estimated to be The protein bound to the NC membrane tightly and could not to be removed from the membrane by extensive washing with the buffer (Fig. 2). Slight decrease of the optical density of the spots the washing out of the protein which weakly attached to the surface of the protein tightly bound to the NC membrane. As summarized in Table 1,

4 Fig. 1. Quantitative binding of human gammaglobulin to a nitrocellulose membrane. The experiments were carried out by using double layered nitrocellulose membranes as described in MATERIALS AND the protein in sample solutions prepared with 100mM acetate buffer, ph 3.7 con- the protein spots in the second layer solutions prepared with P buffer. Fig. 2. Tight binding of human gamma-globulin to a nitrocellulose membrane. The experiments were carried out by the following procedures: Protein solu- of the microfiltation apparatus and filtered through the NC membrane under mildly reduced pressure by a water aspirator as described in MATERIALS was pipetted into the wells and filtered through the NC membrane to wash the protein spots. The washing time varied Fig. 3. Effect of ph on the protein binding to a nitrocellulose membrane. the maximum amounts of the proteins bound to the NC membrane, which were detemined as described in MATERIALS AND METHODS. the maximum amount of proteins bound to the NC membrane was well correlated to their molecular weight (M. W.), with a few exceptions of ceruloplasmin and histone. Therefore, the number of the bound protein molecules which is a value of the maximum amount of the proteins bound to NC membrane divided by their molecular weight was in the range from 0.08 to 0.14 nmol in a uniform circle 3mm in diameter. The number of the bound protein molecules seemed to increase as the hydrophobic indices of proteins calculated from the content of hydrophobic side chains of amino acids composed of the proteins by Levitt's method22 and Tanford's method23) increased. However, good correlation between the number of the bound protein molecules and the hydrophobic indices of the proteins was not documented. The isoelectric point (pi) of proteins was not correlated to the number of bound protein molecules at all. The amount of the protein bound to the NC membrane was not largely influenced by changing the ph of the loading buffer, as shown in Fig. 3. Effects of contaminating chemicals on the protein binding to the NC membrane The effects of the contaminating chemicals on

5 Vol. 33. No (297) Table 1. Protein binding to a nitrocellulose membrane and factors affecting the protein binding. the protein binding to the NC membrane are summarized in Table 2. Non-ionic detergents, Triton X-100, Nonidet P-40, and Tween 20 strongly interfered with the protein binding to the NC membrane at their concentration of 0.01%. SDS moderately interfered with the protein binding at a low concentration of 01%. Other detergents, CHAPS, CHAPSO, sodium cholate, and octyl-glucoside also interfered with the protein binding to the NC membrane at a concentration of 1%. On the other hand, urea and guanidine hydrochloride interfered with the protein binding only at their concentration of 6M significantly. Ammonium sulfate, glycerol, and polyethylene glycol 200 weakly interefered with the protein binding, however, sugars, amino acids, reducers, or nucleotides did not significantly interfere with the protein binding to the NC membrane. But interestingly, a supercoiled double stranded DNA from calf thymus strongly interfered with the binding of histone to the NC membrane, although the DNA did not interfere with the binding of human gamma-globulin or serum albumin to the NC membrane at all. Effect of detergents on protein binding to the NC membrane The interference of protein binding to the NC membrane by detergents contaminating the pro-

6 Table 2. Effects of contaminating chemicals on the binding of human gamma-globulin to nitrocellulose membrane. ** The optical density (O. D.) of protein spots was determined as described in MATERIALS AND METHODS, and the values for relative O. D, were shown as a percentage of the ratio to the O. D. obtained by using the protein sample in the absence of the contaminating chemicals. gamma-globulin in 10mM phosphate buffer, ph 7.4 containing 150mM NaCI (P buffer). *2 The DNA Was a supercoiled double stranded deoxyribonucleic acid from calf thymus. albumin prepared with P buffer. The values for the reative O. D. were shown as a percentage of the ratio to the O. D. obtained by using the protein sample in the absence of DNA. tone type II-s from calf thymus prepared with P buffer. The values for the relative O. D. was shown as a percentage of the ratio to the O. D. obtained by using the protein sample in the absence of DNA. a The concentrations of the chemicals were shown as a percentage of weight per weight. tein samples was examined with human gammaglobulin in more detail by using a series of different concentrations of the detergents, as shown in Fig. 4. Non-ionic detergents, Triton X-100, Nonidet P-40, and Tween 20 strongly interfered with the protein binding to the NC membrane their concentrations were increased, and only 32 of protein remained on the membrane with 0.01% Tween 20. An anionic detergent, SDS interfered as with the protein binding at a low concentration of 0.1%, however, a significant amount of the protein (40%) remained on the NC membrane even at a high concentration of 1% (data not shown). Zwitterionic detergents, CHAPS and CHAPSO, also interfered with the protein binding to the NC membrane strongly as their concentrations were increased, and completely interfered with the protein binding at their concentration of 1%.

7 Vol. 33. No (299) Fig. 4. Effects of contaminating detergents on the protein binding to a nitrocellulose membrane. X-100. The ordinate indicates the values for the relative optical density (O. D.) of the protein spots as a percentage of the ratio to the O. D. of the protein spots in the absence of the detergents. Fig. 5. The extractability of the bound protein from a nitrocellulose membrane with detergents. The striated bar indicates the amount of the protein remained on the NC membrane after the incubation with 1% of the detergents except for 1M of guanidine hydrochloride and urea for 1h. The white bar indicates the amount of the protein remained on the NC membrane after incubation with 10% of the detergents except for 6M of guanidine hydrochloride and urea for 1h. The ordinate indicates the percentage of the amount of proteins remained on the NC membrane to the total amount of the proteins in the applied samples. The letters appearing under the abscissa indicate the kinds of detergent: a, Tween 20; b, Triton X- 100; c, Nonidet P-40; d, octyl-glucoside; e, SDS; f, CHAPS; g, CHAPSO; h, sodium cholate; i, urea; and j, guanidine hydrochloride. Fig. 6. Time course of the extraction of bound protein from a nitrocellulose membrane with detergents. All the experiments Extraction of protein from the NC membrane with a series of detergents were carried out by The extractability of bound proteins from the NC membrane was examined with human gamma-globulin by using a series of detergents as shown in Fig. 5. Non-ionic detergents, Triton X-100, Nonidet P-40, and Tween 20 extracted the bound protein from the NC membrane efficiently. The results with other types of detergent indicated that the efficiency of extracting the bound protein was well correlated with the

8 strength of interference with the protein binding to the NC membrane matrices by those detergents, as shown in Table 2. The amount of the protein bound to the NC membrane in the presence of 0.1% Tween 20 in the sample was only 3% of that in the absence of the detergent. However, 8% of the protein still remained on the NC membrane after the extraction of the protein from the membrane by incubating with 10% Tween 20. After incubating with 10% SDS, 40% of the protein still remained on the NC membrane. Urea and guanidine hydrochloride did not extract the protein from the NC membrane at 1M. But 6M guanidine hydrochloride did extract 70% of the protein. As shown in Fig. 6, the extraction of the bound protein was completed by 1h incubation with 1% Tween 20, and only a trace of the bound protein could be extracted by further incubation for 24h. The extraction of the bound protein was almost completed by 6h incubation with 1% SDS, however, still 20% of the protein remained on the membrane. DISCUSSION Factors involved in the immobilization of proteins to a nitrocellulose membrane It is apparent that proteins bind to the NC membrane tightly and quantitatively until the NC membrane is saturated with the proteins (Figs. 1 and 2). The number of protein molecules bound to the NC membrane was in the range from to 1.98nmol/cm2 even though the proteins had considerable variations in their molecular weights (from 448kd to 18kd) and isoelectric points (from pi 4.6 to 8.9) (Table 1). Furthermore, the protein binding to the NC membrane was not influenced much by changing ph of the buffer (Fig.3). Small size proteins having strong negative charges (e.g. trypsin inhibitor, 23kd and pi 4.3) or positive charges (e. g. lysozyme, 14kd and pi 11.0), however, bound very weakly to the NC membrane, as reported by Lin and Kasamatsu.24) Protamine, a small basic protein (4kd and pi>10), did not bind to the NC membrane at all, as shown in Fig. 7. In the case of histone, the number of protein molecules bound to the NC membrane became much larger than that of other proteins. This may be due to the formation of protein aggregates which remained on the membrane because only the surface of the spot of histone was stained densely (data not shown). On the other hand, histone did not bind to the NC membrane at all in the presence of a supercoiled double stranded DNA from calf thymus, although the DNA did not interfere with the binding of human gamma-globulin or serum albumin at all (Table 2). This indicates that histone formed a stable complex with the DNA to have strong negative charges on their molecular surface, so that histone could not bind to the membrane. The interference of the binding of proteins to the NC membrane with non-ionic detergents indicates that hydrophobic interactions between the proteins and the NC membrane matrices play a major role in the protein binding to the NC membrane. However, there was no interference by urea, neutral salts, glycerol, sugars, nucleotides, or charged amino acids (Table 2). A series of detergents for extraction of protein from the NC membrane Non-ionic detergents, Triton X-100, Nonidet P- 40, and Tween 20, an anionic detergent, SDS, and zwitterionic detergents CHAPS and CHAPSO strongly interfered with the protein binding to the NC membrane in a concentration dependent manner (Fig. 4). As was expected, those detergents efficiently extracted the bound protein from the NC membrane (Fig. 5). In contrast, sodium cholate did not interfere with the protein binding to the NC membrane strongly, and it did not extract the bound protein from the NC membrane efficiently. In these experiments, the amount of proteins bound to the NC membrane was estimated from the optical density of the protein spots, which were stained with the dye, Ponceau 3R. So, the effect of those detergents on the

9 Vol. 33. No (301) Table 3. Effect of detergents on the staining of the proteins bound to a nitrocellulose membrane with Ponceau 3R. * All the experiments were carried out by us - human gamma-globulin prepared with 10mM phosphate buffer, ph7.4 containing 150mM NaCl. After the spotting of the protein in uniform circles 3mm in diameter, the protein spots were stained with 0.1% Ponceau 3R in 7% accetic acid solution in the presence or absence of the detergents. The optical density (O. D.) of the protein spots was determined as described in MATERIALS AND METHODS. The relative O. D. was shown as a percentage of the ratio to the O. D. obtained by using the staining solution in the absence of the detergents. a The concentration of the detergents was 1% in the staining solution of 0.1% Ponceau 3R prepared with 7% acetic acid. bthe concentration of detergents was 6M in the staining solution (0.1% Ponceau 3R (w/w) in 7% (w/w) acetic acid). Ponceau 3R aggregated at high concentration of guanidine hydrochloride in the staining solution. Thus, the protein spots were stained with 0.1% Ponceau 3R in 7% acetic acid immediately after the protein spots were dipped for 10s into 6M guanidine hydrochloride prepared with 10mM phosphate buffer, ph 7.4 containing 150mM NaCl. protein-dye binding must be accounted for. The effect of the detergents on the protein-dye binding, however, was negligible, as shown in Table 3. Non-ionic detergents such as Triton X-100, Nonidet P-40, and Tween 20 strongly interfered with the protein binding to the NC membrane, Fig. 7. Binding of protamine to a nitrocellulose membrane in the presence of sodium dodecylsulfate. The experiments were carried out by ing a series of the concentrations of protamine sulfate from salmon sperm. The optical density (O. D.) of the protein spots was determined as described in teln solutions containing 1% SDS spotted onto untreated NC membrane. and only a trace of the protein could be detected on the NC membrane at 1% (data not shown), as reported by others ) An anionic detergent, SDS also interfered with the protein binding to the NC membrane. However, a significant amount of the protein still remained on the NC membrane even at 1% (data not shown). This suggests that SDS tightly bound to NC membrane through its hydrophobic moiety of hydrocarbon chain, and another moiety of charged group, sulfate group, strongly interacted with positive charges on the proteins to form a stable complex.27) This may be supported by the results from the examination of the binding of a small basic protein, protamine, to the NC membrane as shown in Fig. 7. Although the protein itself did not bind to the NC membrane at all, the protein tightly bound to the NC membrane in both cases, with the protein samples containing SDS and with the NC membrane treated with SDS prior to spotting the protein.

10 Table 4. Effect of pore size of nitrocellulose membrane on the protein binding. * The maximum amount of bound protein to the nitrocellulose membrane was deter - mined as described in MATERIALS AND METHODS. N. A.:. Not available for determination of precise amount of the bound protein because of difficulty in yielding uniform protein spot of 3mm diameter. The values in parenthesis indicate approximate amount of the bound protein to the NC membrane. N. D.: Not detectable. Originally; the NC membrane was prepared in as a surface filter.19) However, it is apparent that proteins bind to the NC membrane tightly and quantitatively by chemical interactions between the proteins and the NC membrane matrices in which hydrophobic interactions may a major role, but not by the mechanical sieving effect of the membrane filter. Therefore, the NC membrane became a powerful tool for the detection and quantitation of a very small amount of the proteins having considerable variations in their molecular weights and isoelectric points. However, it should be noted that some proteins having a low molecular weight and strong charges on their surface tightly. do not bind to the NC membrane The effect of pore size of the NC membrane on the protein binding was also examined by using other type of NC membranes whose Table 4. The amount of bound proteins was increased by decreasing the pore size of the NC membrane. This increase may be due to the increase of the capacity of the small pore membrane to bind the proteins, but not due to the sieving effect of the small pore membrane because the proteins could passed through the first layer of the membrane to give the protein spot on the second layer of the membrane even with the small pore NC membrane. Thus, NC mem- also be useful for the protein assay. However, was not useful because of its difficulty of protein application onto the membrane to yield the uniform circles of protein spot under mildly reduced pressure. Other type of membrane, polyvinylidene difluoride membrane was useful for the protein assay because its high binding capacity and high affinity for the proteins provided a high reproducibility and the translucency of this membrane could be attained by immersing into silicone liquid to determine the amount of proteins on the membrane densitometrically (data not shown). Non-ionic detergents, Tween 20 and Triton X- 100 strongly interfered with the protein binding to the NC membrane in a concentration dependent manner, and they extracted the proteins from the NC membrane efficiently at their concentrations lower than 0.1%. Thus, these detergents are much more useful than SDS for extracting proteins from the NC membrane to determine their amino acid composition and sequence in micro scale after blotting the proteins. 28)

11 Vol. 33. No (303) REFERENCES 1) Towbin, H. et al.: Proc. Natl. Acad. Sci. U.S.A., 76: 4350, ) Bowen, B. et al.: Nucl. Acids Res., 8: 1, ) Shen, V. et al.: Nucl. Acids Res., 8: 4639, ) Wang, R. et al.: Olin. Chem. Chem. Acta, 102: 169, ) Glass, W. F., Jr. et al.: Anal. Biochem., 115: 219, ) Burnette, W. N.: Anal. Biochem., 112: 195, ) Hawkes, R. et al.: Anal. Biochem., 119: 142, ) Herbrink, P. et al.: J. Immunol. Methods, 48: 293, ) Reinhart, M. P. and Malamud, D.: Anal. Biochem., 123: 229, ) Palfree, R. G. E. and Elliot, B. E.: J. Immunol. Methods, 52: 395, ) Kranz, R. G. and Gennis, R. B.: Anal. Biochem., 127: 247, ) Kakita, K. et al.: Diabetes, 31: 648, ) Domin, B. A. et al.: Anal. Biochem., 136: 390, ) Kawasaki, H. et al.: Anal. Biochem., 148: 297, ) Jones, B. B, et al.: Proc. Natl. Acad. Sci. U.S.A., 74: 4914, ) Wallis, C. et al.: Ann. Rev. Microbiol., 33: 413, ) Presswood, W. G.: Membrane Filtration, Applications, Techniques, and Problems, Dutka, B. J. (ed), Dekker, New York and Basel, ) Farrah, S. R. et al.: Proc. Natl. Acad. Sci. U.S.A., 78: 1229, ) Gershoni, J. M. and Palade, G. E.: Anal. Biochem., 131: 1, ) Beisiegel, U.: Electrophoresis, 7: 1, ) Nakamura, K. et al.: Anal. Biochem., 148: 311, ) Levitt, M.: J. Mol. Biol., 104: 59, ) Tanford, C.: J. Am. Chem. Soc., 84: 4240, ) Lin W. and Kasamatsu, H.: Anal. Biochem., 128: 302, ) Batteiger, B. et al.: J. Immunol. Methods, 55: 297, ) Bjerrum, O. J. et al.: Electrophoresis, 8: 388, ) Reynolds, J. A. and Tanford, C.: Proc. Natl. Acad. Sci. U.S.A., 66: 1002, ) Brandt, W. F. and von Holt, C.: Advanced Methods in Protein Microsequence Analysis, Wittman-Liebold, B., Salnikow, J. and Erdmann, V. A. (eds.), Springer Veriag, Berlin, Heidelberg New York, London Paris and Tokyo, 1986, p. 161.