paper chromatogram, a linear curve is obtained (Figure 3). DISCUSSION
The use of a factor based on the height, H , of the radioactive zone would be restricted to a single steroid in a rigidly controlled system, because steroids near the origin are in narrow and compact bands, whereas those migrating near the front are more broad and diffuse. This holds true for all the radioactive steroids employed in biological experiments (4) (~ortisone-4-C~~, 11P-hydroxyandro~tenedione-4-C~~, and 4-pregnene17,20,21 - triol - 3,11 - dione - 4 - CY4)). The results have all fallen within i8% of the factor E / A , which is 6.12 (Table I). Therefore, if the area recorded by the strip counter (Figure 2) is multiplied by this factor, the amount of radioactivity present in the paper can be determined with this accuracy. This eliminates the necessity of elution steps @), FT-hich often result in loss of appreciable quantities of steroid.
E / P is a useful factor, which has been employed in previous studies (9). When a known quantity of steroid is chromatographed, the recovery from the zone of migration of the steroid in the Zaffaroni chromatographic systems varies from 80 t o 92%. The percentage of radioactivity not accounted for in the chromatographed zone can be demonstrated a t the origin and front and in the streaking zones. The distribution of radioactivity outside the chromatograph zone is usually greater in the Bush system ( 5 ) of chromatography, because of the low capacity of this system in the presence of fats in biological extracts. When the area method is applied to the determination of radioactive steroids extracted from biological fluids and tissues, the percentage loss can be determined by adding a known quantity of nonradioactive steroid to the extract before chromatography (1). The loss of nonradioactive steroid in the chromatographed zone is proportional to the loss of the radioactive steroids, when they are identical and in mixture.
A quantitative ultraviolet strip recording spectrophotometer in combination with the strip counter employed in the work should be of great value in the simultaneous determination of radioactive and nonradioactive steroids directly on paper chromatograms. LITERATURE CITED
(1) Berliner, D. L., Proc. SOC.Esptl. Biol. & M e d . 94, 126 (1957). (2) Berliner, D. L., Salhanick, H. A,, ANAL.CHEX 28, 1608 (1956). (3) Berliner, D. L., Wiest, TV. G., J . Biol. Chem. 221, 449 (1956). (4) Berliner, M. L., Berliner, D. L., Dougherty, T. F., Proc. Am. Assoc. Cancer Research 2, 94 (1956).
(5) Bush, I. E., (‘Recent Progress in Hormone Research,” 1‘01. IX, p. 321, Bcademic Press, Xew York, 1954. (6) Zaffaroni, A., Ibid., Vol. VIII, p. 51, 1953. RECEIVED for review Sovember 15, 1956. Accepted July 27, 1957. Work supported in part by research grants from the S a tional Cancer Institute, National Institutes of Health, U. S.Public Health Service, and the American Cancer Society upon recommendation of the Committee on Growth, National Research Council.
Accuracy of Quantitative Paper Chromatography in Amino Acid Determination Using Direct Photometry HENRY R. ROBERTS and MICHAEL G. KOLOR Research laboratories Division, National Dairy Products Corp., Oakdale, 1. I., N. Y.
A study has been made of the accuracy and precision of the quantitative paper chromatographic procedure developed by McFarren and Mills for determination of amino acids in protein hydrolyzates. Known solutions of 20 common amino acids were prepared and analyzed. Each amino acid was assayed nine times, the average value obtained, and the per cent deviation of this average value from the known value calculated. Eleven acids gave average values which differed from the theoretical value by less than 2%, 13 less than 370, 17 less than 5%. Only lysine, norleucine, and tyrosine had values more than 5% in excess of the true values. Statistical evaluation of precision showed that the assays can b e repeated with good reproducibility. ARIOUS quantitative methods have been developed for the determination of amino acids by paper chromatography (3, 4, 18). A number of investigators (9-5, 7, 10, 11, 13-15, l Y ,
1800
ANALYTICAL CHEMISTRY
13, 20) have shown that the direct photometric or maximum color density procedure yields results which agree with other more involved methods and is convenient, less tedious, and more rapid. This paper presents data on the accuracy and precision of McFarren and Mills’ ( I S ) maximum color density procedure for amino acid determination. I n determination of the amino acids content of p-lactoglobulin, McFarren and hlills’ results were comparable to those obtained by other methods. Their appraisal of the procedure was based only on this comparison. The present study is based on analysis of a known standard solution of amino acids, for without a comparison iTith a standard, accuracy has no meaning. I n addition to ascertaining the accuracy and precision of McFarren and Mills’ technique, this study also ivas attempted:
To point out that the procedure gives excellent separation of the amino acids as round or elliptical spots, the lack of which is often offered as an argument
against the use of the maximum density procedure. To offer additional proof that the maximum density technique, as a quantitative procedure, can be used with confidence. EXPERIMENTAL
hIcFarren and Mills’ one-dimensional descending paper chromatographic procedure ( I S ) uses seven solvent systems a t a selected p H to separate each amino acid from all others in a mixture containing 20 common amino acids. The paper is buffered at the same p H as the solvent. Strips of Khatman No. 1filter paper are dipped into the appropriate buffer solution ( I d ) , suspended by one end, and dried in air. Each solvent system is a two-phase system. The waterrich layer is placed in the bottom of the chamber. To achieve the desired separations, the bottoms of the chambers must be kept wet with the water-rich phase. The developing solvents, equilibrated a t 22’ C. in a temperaturecontrolled room m-hich houses the chromatographic chambers, are prepared as follows: SOLVENT 1. Equal volumes of phenol
(Merck, reagent grade, melted in a 60" C. water bath) and pH 12.0 buffer (0.067M). SOLVENT 2. Equal volumes of redistilled m-cresol (Matheson Co., practical grade) and pH 8.4 buffer (0.067M). The buffer solution must be within &O.l pH to obtain the reported separations. SOLVENT 3. Equal volumes of redistilled 2,4-lutidine (Koppers Co., Inc., refined grade) and pH 6.2 buffer (0.022.V). SOLVENT 4. Equal volumes of phenol and pH 1.0 buffer (0.2M). SOLVENT 5. Equal volumes of redistilled o-cresol (Fisher Scientific Co., melted in a 60" C. water bath) and pH 6.2 buffer (0.067M). SOLVENT 6. Benzyl alcohol, butyl alcohol, pH 8.4 buffer (0.067Jf) (1-1-2). SOLVEKT i . Equal volumes of redistilled 2,4,6-collidine (Koppers Co., Inc., refined grade) and pH 9.0 buffer (0.067Mj. The quantitative procedure has been published in detail ( I S ) . The follon-ing modifications, which do not alter the separations, are incorporated in this present study.
Chromatographic Chambers. Stainless steel chambers 30 inches long, 25 inches high, a n d 8 inches n-ide are lined with filter paper t h a t dips into t h e water-rich solvent layer in the bottom of t h e chamber. T h e filter paper linings become saturated with t h e water-rich phase a n d t h e atmosphere of t h e chamber becomes saturated with respect t o the solvent phases. Stainless steel troughs made from 1inch sanitary tubing are held in place 22 inches from the bottom by stainless steel brackets fastened to the sides of the chamber. Normally each chamber houses one solvent trough, but the size of the chamber easily permits the housing of two troughs, thereby doubling the number of chromatograms per chamber. A sponge rubber gasket cemented to the flanged surface of the chamber is coated with stopcock grease, to make a n airtight seal with the glass-plate cover. Table I.
Clamps can be easily attached to ensure this seal, as the equilibrium maintained within the chamber during the chromatographic development is a critical factor in obtaining good resolutioii and round compact spots. Preparation of Test Solutions. Three known amino acid solutions n-ere prepared and run as unknowns. T h e first solution contained cystine (40 mg. per 100 ml.) adjusted t o p H 1.0. T h e second contained tyrosine (40 nig. per 100 ml.) adjusted t o p H 2.0. The third contained t h e remaining 18 common amino acids (Table I) a t a concentration of 40 mg. of each amino acid per 100-nil. total volume, p H 6.5. The need for preparing special solutions for cystine and tyrosine was dictated by their solubility limits. Each solution contained lOyc isopropyl alcohol as a preservative. The standard solutions used to derive the standard curves and the solutions to be analyzed were prepared from the same amino acid source. Thus, no error due to amino acid impurity was introduced and the results give a true picture of the accuracy of the method. Preparation of Quantitative Paper The quantitative Chromatograms. procedure uses strips of Whatman KO. 1 filter paper, 23 X 57 cm., buffered a t t h e appropriate p H . Five known standard solutions containing, respectiyely, 1.25, 1.00, 0.75, 0.50, and 0.25 y of amino nitrogen for each amino acid, and three dilutions of the solution, are applied to the origin in 5-pl. quantities a t a point 7.5 cm. from one end, using a Gilmont ultramicroburet. The test solution is spotted to fall within the 0.25- to 1.25-y range of the standard solutions. It was necessary to double the concentration of the standards and test solutions applied to the paper n-hen assaying for hydroxyproline, in order to obtain significant density readings. All solutions applied to the paper, standards, and unknowns are adjusted to p H
6.5, except cystine (pH 1.0) and tyrosine (pH 2.0). Quantitative Procedure for Hydroxyproline and Norleucine. AhFarren and Mills ( I S ) did not give quantitative data on hydroxyproline and norleucine. Hydroxyproline is separated by using m-cresol buffered a t p H 8.4, as are tyrosine, histidine, valine, and methionine. Because of the faint color produced with ninhydrin, hydroxyproline is made to react with 0.47, isatin in water-saturated butanol containing 47, acetic acid. The color is developed by heating a t 100" C. for 10 minutes and then storing a t room temperature for 24 hours. I n addition to the oven heat treatment (13), histidine and glycine required 24 hours a t room temperature for full color development. Korleucine was separated with phenylalanine by using o-cresol buffered a t p H 6.2. The quantitative procedure is the same as for phenylalanine (13). RESULTS AND DISCUSSION
Each amino acid was assayed nine times. The results obtained from one quantitative paper chroniatogram containing three dilutions of the amino acid test solution represented one determination. The average value of the nine determinations was obtained and the per cent deviation of this average value, from the theoretical value of 40 mg. per 100 ml., was calculated for each amino acid (Table I). The average values of threonine, serine, and cystine differed from theoretical values by less than 1%. Values for aspartic acid, histidine, proline, isoleucine, leucine, alanine, tryptophan, and glutamic acid were in error by 1 to 2%. The error was betlTeen 2 and 37, for arginine and valine, 3 and 47, for hydroxyproline and phenylalanine, 4 and 5% for glycine and methionine,
Accuracy and Precision of Paper Chromatographic Determination of Amino Acids Calculated Concentration, hIg./100 111. Av. Concn., Dev." from Standardb Chromntogram Iig./lOO Theory, Dev., 111, 2 3 4 5 6 7 8 9 % hIg./100 MI.
Amino Acid 1 Threonine 40.8 42.5 39.1 39.1 Serine 42.8 42.8 38.3 39.1 Cystine 39.4 39.4 41.2 39.4 Aspartic acid 40.9 40.9 40.9 40.9 Histidine 45.4 44.3 42.i 38.i Proline 35.3 41.9 41.9 38.6 Isoleucine 40.2 38.4 43.1 37.4 Leucine 42.1 43.1 38.4 44.0 Alanine 44.5 36.3 3i.5 38.2 Tryptophan 36.5 39.4 43.7 37.9 Glutamic acid 3 8 . 9 44.1 44.1 37.8 -4rginine 3i.3 44.7 42.3 41.1 Valine 36.8 38.5 36.8 41.8 Hydroxyproline 3 7 . 4 37.4 39.3 40.2 Phenylalanine 3 8 . 9 42.4 40.1 36.5 G1y cin e 39.7 42.3 41.8 40.7 hiethionine 38.3 39.4 37.3 37.3 Lysine 39.6 42.8 42.8 41.7 Norleucine 43.1 36.5 35.6 40.2 Tyrosine 44.0 38.8 42.7 40.1 ' Theoretical value = 40 mn . / l o 0 ml. yzxz- ( z x ) z , Standard deviation = &(Ai
37.4 39.8 44.6 41.8 39.9 39.5 40.2 38.4 38.2 42.3 36.8 39.8 38.5 41.2 36.5 41.8 38.3 42.8 34.6 45.3
39.1 42.1 37.i 38.0 42.1 40.3 41.2 37.4 3i.5 40.8 41.0 44.i 39.3 37.4 37.7 44.5 35.1 43.8 37.4 46.5
42.5 38.3 42.9 39.9 36.5 34.5 38.4 38.4 40.1 40.8 43.1 43.5 41.0 37.4 42.4 42.9 37.3 41.7 37.4 44.0
41.i 38.3 37.7 40.9 38.i 40.3 37.4 43.1 39.4 43.7 37.8 38.5 39.3 39.3 35.4 39.1 39.4 41.7 37.4 44.0
38.3 40.6 36.0 39.9 36.5 43.6 39.3 39.3 43.2 40.8 43.1 37.3 38.5 37.4 36.5 42.3 39.4 41.7 35.6 41.4
40.1 40.2 39.8 40.5 40.5 39.5 39.5 40.5 39.4 40.i 40.7 41 .O 38.9 38.6 38.5 41.7 38.0 42.1 37.5 43.0
0.25 0.50 0.50 1.25 1.25 1.25 1.25 1.25 1.50 1.i5 1.i5 2.50 2.i5 3.50 3.75 4.25 5.00 5.25 6.25 7.50
1 8i ~. 1.92 2.i0 1.09 3.21 3.02 1.87 2.57 2.i5 2.44 2.96 2.97 1.67 1.48 2.63 1.65 1.41 1.19 2.63 2.43
- 1)
VOL. 29, NO. 12, DECEMBER 1957
1801
and 5 and S70 for lysine, norleucine, and tyrosine. Average values of 11 amino acids differed from the theoretical value by less than 2%, 13 less than 395, and 17 less than 5%. Only lysine, norleucine, and tyrosine had values more than 5% greater than the true value. Even then, the value for tyrosine (the most in error) differed from the true value by only 3 mg. or 7.57, error. Considering the values of the 20 amino acids as a whole, the average deviation (from the theoretical value of 40 mg.) calculated by the maximum color density method on paper chromatograms is 2.66Y,, or less than 3%. The precision of the method was evaluated by calculating the standard deviation, using the formula:
where S equals the standard deviation,
S equals the number of determinations, and X equals the individual values. This equation (21) is algebrfiically iden-
dz, I
tical to S
=
but is preferred
because it allow computation without rounding off data. The data (Table I) show that the assays can be repeated with good precision. I n only two cases, histidine and proline, x i s the standard deviation in excess of 3 mg. per 100 ml. The standard deviation of a determination is betmen 1 and 2 mg. per 100 ml. for threonine, serine, aspartic acid, isoleucine, valine, hydroxyproline, glycine, methionine, and lysine; and between 2 and 3 mg. per 100 ml. for glutamic acid, arginine, phenylalanine, norleucine, tyrosine, cystine, leucine, tryptophan, and alanine. This study has demonstrated that
quantitative determination of amino acids on paper chromatograms, by the maximum density method developed by NcFarren and llills, yields values possessing low errors and good reproducibility. Khile the method employs seaen solvent systems and a maximum solvent development time of 40 hours, the resolutions so obtained yield amino acid spots which readily lend themselves to quantitation. The method employs a heat treatment to dry the chromatograms following solvent derelopment and also to develop the color of the amino acid spots after the chromatogram has been dipped in the appropriate color reagent solution. Following color development, the maximum densitips of the spots are read. This eliminates the 24-hour development in the dark normally employed. The heat treatments do not introduce significant errors. Although losses of amino acids may occur when chromatograms are heated, as demonstrated by a number of inwstigators ( 1 , 6 , S ,9 , 1 6 ) , chromatographing the standards and unknova solution on the same sheet does much to cancel out these losses, as they are both subjected to thp same heat treatment. ACKNOWLEDGMENT
The authors wish to thank F. E. Hawkins for valuable suggestions pertaining to statistical treatment of the data. LITERATURE CITED
Berry, H . K., Cain, T., Arch. Biochem. 24, 179 (1949). Block, R. J., ANAL. CHEJI.22, 1327 (1950).
(3) Block, R. J., Durrum, E. L., Zweig, G., “Manual of Paper Chromatography and Paper Electrophoresis,” Chap. 4-5, Academic Press, Kew York, 1955. (4) Block, R. J., Weiss, K. FV., “Amino Acid Handbook. Rlethods and Results of Protein Analysis,” Chap. 5, Charles C Thomas, Springfield, Ill., 1956. (5) Block, R. J., Weise, K. W.,Arch. Biochem. and Bzophys. 5 5 , 315 (1955). (6) Brush, RI. K., Boutwell, R. I