V O L U M E 25, NO. 8, A U G U S T 1 9 5 3 (4) Cholak, J., Hubbard, D. M , and Burkey, R. E,, Ibid., 15, 754 (1943). (5) Fernandez, Obdulio, and Santaolalla, Manuel, Rev. real acad. cienc. ezact. its. 2/ nut. M a d r i d , 42, 421-39 (1949). (6) Greenleaf, C. A , , J. Aasoc. Ofic.A g r . Chemists, 24, 337 (1941). (7) Ibid., 25,385 (1942). (8) Ibid., 30,144 (1947). (9) Kimura, Kenjiro, and hlurakami, Yukio, Mikrochemie w r . Mikrochim. Acta, 36-7, 958-66 (1951).
1247 (IO) Michel, G., and Maron, N., A n a l . Chim. Acta, 4, 542-50 (1950). (11) Morrison, S. L., and Paige, H. L., IND.EXG.CHEM.,ANAL.ED., 18,211 (1946). (12) Murakami, Yukio, Bull. Chem. Soc. J a p a n , 23,99-102 (1950). (13) Reed, J. F.1 and Cummings, R. w., I N D . ENQ. CHEM.1 ANAL. E D . ,13, 124 (1941). RECEIVED for review July 9, 1952. Accepted May 2, 1953. Presented before the Division of Analytical Chemistry at the 121st Meeting of the AMERICAN CHEMICAL SOCIETY, Buffalo, N. Y.
Media Used for Microbiological Assay of Riboflavin Using Lactobacillus casei Method of Assay and Confidence Limits Attainable MIRIAnl F. CLARKE’ D e p a r t m e n t of C h e m i s t r y , Medical College of Virginia, R i c h m o n d , Va. Media commonly employed to estimate riboflavin potency by means of microbiological assays, using L. casei, did not seem sufficiently susceptible to tests of reliability or allow simple means of calculation (by accepted methods) of confidence limits of estimated potency. Therefore, changes in the media were made which yield these advantages. Three media were altered by increasing glucose content to 3970 and/or changing the amounts of yeast preparations, ammonium sulfate, and sodium acetate. These changes yielded curves with graded doses of riboflavin, nonlinear when the doses themselves were used, but linear when the logarithms of doses were plotted against the response, acid produced. No evidence of departure from linearity over an eightfold increase in dose was demonstrated. Curves of similar slope were obtained with graded quantities of vitamin concentrates. The value inherent in these results in planning 6-point, self-contained assays and calculating expected precision and confidence limits is discussed.
A
NUMBER of modifications of the original Snell and Strong
(17 ) medium for microbiological assay of riboflavin, using Lactobacillus casei, have been published ( 6 , 12, 16, 18) and are in more or less wide use for routine assays. Because three of the media described varied considerably in the amount of acid produced by the organsim in response to a given quantity of pure riboflavin, studies on these media were undertaken, directing attention to those ingredients considered most likely to influenee the performance of the organism. The results obtained could be used in deciding upon the most satisfactory modification to employ in assays. The early work in the field suggested a linear relationship between dose and response, titratable acidity, over a range approximately 0.02 to 0.2 microgram of riboflavin. A preliminary study of the modified media indicated that ( a ) an increase in acid output for any given dose, ( b ) a higher maximum value, and ( c ) the tendency toward a nonlinear dose-response curve on the usual arithmetic scalt. seemed to follow each change in the original medium. This nonlinear relationship between riboflavin dosage and acid output of L. casei made it obvious that some transformation of the data should be used, to obtain a straight line and to benefit from this in being able to test the validity of each assay and to calculate potency, and confidence limits thereof, by means of accepted procedures. The log-dose relationship to response, heretofore generally proved to be rectilinear with several vitamins tested on a variety of animals (Z),was found to be the most satisfactory method of plotting results in this study. Moreover, it had been reported previously (9) that a straight line described 1 Present address, Department of Physiological Chemistry, The Woman’s Medical College of Pennsylvania, Philadelphia, Pa.
the relationship between log-dose and acid output of L. ( m e when pantothenic acid was the limiting factor. Perhaps the most obvious limitation which could be made on acid production is the restriction of glucose, which is 100 ( 6 ) , 200 ( 1 5 ) , and 300 mg. (12) per test tube (10 ml.) in the references cited. In the experiments reported here, glucose was increased in the Greene-Black medium ( 6 ) from 100 to 300 mg. and in the RobertsSnell medium ( 1 5 ) from 200 t o 300 mg., the latter amount being that recommended in the widely used United States Pharmacopoeia assay (12). With the glucose thus increased, the curves tended to be nonlinear on an arithmetic scale, but the maximum titration was often above 20 ml. of 0.1 N sodium hydroxide instead of 9 to 10 ml. attainable when glucose was limited to 100 mg. Plotting responses against the logarithms of graded doses, using media modified only in glucose content, yielded curves of response which were sigmoid in nature, the straight central portion varying in slope and length with the various media. mith glucose virtually excluded as a limiting factor by incorporating it a t 395, two of the media previously described ( 6 , 12) were further modified in the study with regard to their “growth stimulating” and buffer content in order to analyze the effect of each of these on the performance of the organism and the character of the dose-response curves. The purpose of the study was to examine the dose-response curves obtained using three unmodified media and to compare the results with those obtained when the media were modified as to glucose content, or glucose plus growth stimulants and buffer. Conclusions were drawn regarding the merits of the modified media by comparison of the magnitude of the maximum response, the standard error of estimate, the range of linearity
1248
ANALYTICAL CHEMISTRY
Table I. Constituents of Basal Media for L . casei (Unmodified) in 10 M1. of Final Dilution GreeneBlack ( 6 ) Light treated peptone ( f 7 ) , ml
1 ( = 50 rug. peptone)
.
Enzymatic casein digest (16) ml Light-heated Bacto-yeast supplement (6) ml. Yeast supplemint froin baker'syeast (fb),ml. Salt solutionB (17), mI. Salt solution c ( 1 6 ) , m1. Glucose. me.
1 (a 50 mg. peptone)
4 ( = 200 mg.
casein)
0.1 ( e200 nrg.
fresh yeast) 0 05
0.05
0 2 200 1
300a
100 1
1
200 250 KHIPOI, nig. KnHYO,. mg. Adenine sulfate y Guanine hydroc'hloridr, Uracil, y Biotin, y Calcium pantothenate, y Folic acid, y Nicotinic acid, y p-Aminobenzoic acid y Pyridoxine hydrochldride, Thiamine hydrochloride,
RobertsSnell (fa)
U.S.P.(fa)
200 25 2: 100 100 100 0 04
5
0.02 b
1
10 5
y
a Amount of glucose is that required in U.S.P.XI1 supplement. 1943. and later specified in U.S.P.X I V . 1950, not the 10% higher amount prescribed for the U.S.P. XI11 medium.
perimental media. The dosage interval was 0.301 or the log of 2, each dose being twice the one next below. When concentrates were tested against riboflavin standard, the dosage interval was likewise 0.301. In certain cases, where original Greene-Black or U.S.P. media and corresponding procedures were employed, the recommended riboflavin doses, 0.05 to 0.5 microgram, and dosage intervals were used. All tubes were in duplicate. Points plotted in the graphs represent the average of two titrations. The stock culture was L. caset ATCC No. 7469. Stab transfers were prepared in glucose-agar-yeast extract tubes (11) a t biweekly intervals. For each group of assa s, inoculum was prepared by transferring organisms from a freszly incubated stab to tubes of complete liquid medium containing 3% gluo6se and 0.470 untreated Bacto yeast extract. Cultures incubated approximately 24 hours were centrifuged, wwhed by suspension in sterile 0.85% saline, centrifuged again, and resuspended to the original volume in saline. One tenth milliliter was introduced aseptically into each tube, previously autoclaved. The tubes were a t iirst autoclaved a t 15 pounds for 15 minutes. Later in the study the time was decreased to 10 minutes, which reduced the browning especially noticeable in the solutions containing considerable sodium acetate with 3y0 glucose. S o evidence of incomplete destruction of microorganisms was obtained in unseeded, incubated controls vr-hen the sterilization time was thus shortened. After incubation a t 37" C. for 70 to 72 hours, the assay tubes were removed and titrated to pH 6.8 to 7.0 with 0.1 N sodium hydroxide using bromthymol blue as indicator. The statistical analysis of the data for the standard dose-response curves followed conventional methods (3). An estimate of standard error, slope, linearity, curvature, etc., was made from an analysis of variance. Application of the information obtained to microbiological assays included tests for parallelism, curvature and confidence limits of potency estiniated.
Table 11. Constituents of Four Media Found to Be Best Adapted to Microbiological Assays of Rihoflavin, Using
L.
casei Amounts in 10 MI. Final Dilution of: Ian Iba I1 b IIIC MillilitPm
DI
0 ,
0.9
0 4
0 5
Micrograms Riboflavin per TestTube Figure 1.
20
-,I
4.5
0
*Ob
Lo$ Micrograms Riboflavin
Light-treated peptone (1 7) Enzymatic casein digest (16) Light-treated Bacto-yeast supplement (6) Yeast supplement from baker's east ( f b ) s a 6 solution B (17) Salt solution C (15)
1
1
1
0.3
0 05
0 0:
4
plotted on log scale
of the dose-response curve and the slope of this curve. Because the curves of response with some of the modified media were proved rectilinear on a log-dose basis, it was possible, further, to estimate the precision of assays using conventional methods, and to compare these value,c with any claimed for previously described media and procedures.
0.5 0.05 0.2
AIilligrains
Curves of Response, Greene-Black Medium
Left, A . Unmodified medium B. Unmodified medium with glucose increased to 3 5% Right. Glucose increased to 3%. Data obtained simultaneously but
1
~
~
~
Glucose 300 300 3 00 300 1 1 1 1 Cystine 200 100 100 ("4)zSO4 Sodium acetate (approx.) 250 150 150 200 5 5 5 25 KHzPO4 5 5 5 2.5 IGHPO4 a Derived from Greene-Black (61,modified t o contain 3 % glucose. hledium Ib has decreased yeast supplement, ainmoniurn sulfate, and sodium acetate. b Modified U.S.P. medium, with 5 X yeast supplement, added animoniuin sulfate and sodium acetate. C Mbdified Roberts-Snell mediiim with increased glucose. The uracil, two purines, and seven vitamins are the saine as indicated in Table I.
RESULTS
PROCEDURE
The composition of the media used as a basis for this study is found in Table I. The principal modified media, found most satisfactory for reasons discussed below, are in Table 11. The alterations were confined to the quantity of glucose, yeast preparations, ammonium sulfate, or sodium acetate, or a combination of these. Stock riboflavin solution, 50 micrograms per milliliter, prepared by diluting pure riboflavin with 0.02 N acetic acid, was kept under toluene in the refrigerator. Standard solutions were made by suitable dilution on the day of the assay. Care was exercised to shield all solutions from light: transfers were made in a room lighted only with red lamps. A series of doses of riboflavin, usually from 0.0625 to 2 micrograms, was used to establish the standard curve with all the ex-
Experiments on Greene-Black Medium. Figure 1 shows the riboflavin response curve using unmodified Greene-Black medium, 1% glucose, original doses ( A ) compared with the same medium with 3% glucose ( B ) . Using the medium with increased glucose and logarithmically spaced doses, the dose-response curve was obtained, Figure 1 (right), showing the straight line relationship of log-dose to milliliters of acid produced through the range 0.0625 to 2 micrograms-Le., log - 1.204 to +0.301. Further changes in the Greene-Black medium with 3% glucose involved the sodium acetate, ammonium sulfate, and yeast preparation, Figure 2. Since these were relatively high in the medium as described, and probably contributed to the dark color,
~
_
_
1249
V O L U M E 25, N O . 8, A U G U S T 1 9 5 3
t
I
Log Figure 2.
Micrograms Riboflavin
:
shows satisfactory changes in the position, slope, and length of the linear portion of the curves of response. As with the GreeneBlack medium, the lower quantities of ammonium sulfate and sodium acetate (100 and 150 mg., respectively) seem to provide the modification most favorable for bioassay purposes, medium I1 in Table 11, and curve E in Figure 4. Experiments on Roberts-Snell Medium. Since a satisfactory straight line and a high maximum acid output were obtained (Figure 5 ) by merely increasing the glucose content to 375, n o further modification of this medium was made.
0.3 Dose Intervals
Curves of Response, Modified Greene-Black Medium
Points on each curve represent responses to 0.03125,0.0625, 0.115, 0.25,
0.5, and 1 y of riboflavin per test tube
A. B. C. D.
Glucose increased to 3% Same but sodium acetate decreased from 2.5 to 1.570 Same but ammonium sulfate decreased from 2 to 1% Same but Greene-Black yeast preparation decreased from 0.04 to
0.02%
E. Same but decrease of sodium acetate, ammonium sulfate, and yeast preparation as in B, C,and D
20
Log Micrograms Riboflavin : 0.3 Dose Intervals Figure 4. Curves of Response, U.S.P. M e d i u m
-
Points represent responses to doses a5 in Figure 3 A. U.S.P. medium unmodified B. Same with fivefold inerewe in U.S.P. yeast supplement C. Same yith 2 % each d u m acetate and ammonium sulfate added D. Like B, with 2 each sodium acetate and ammonium sulfate added E. Like B, with 1 % each sodium acetate and ammonium sulfate added
Log Micrograms Riboflavin:0,3DoseInten/aIs Figure 3.
Curves of Response,
U.S.P. M e d i u m Points represent responses to 0.0625,0.125, 0.25, 0.5, 1, and 2 y of riboflavin
A. U.S.P. medium unmodified B. Same with tenfold increase in U.S.P. yeast supplement
a disadvantage, they were reduced in quantity, one a t a time or all together. It will be noted that although decrease to approximately half quantities of each of the above constituenh lowered the acid output for a given dose or caused the curve to depart somewhat from linearity, the final curve ( E ) where all three substances were decreased, has a long rectilinear portion with a satisfactory slope. Repetition of the test, using the medium for curve E, yielded consistent results. The pertinent data regarding media are summarized in Table 11, where the media Ia and I b correspond to those used for curves A and E in Figure 2, respectively. Experiments on the U.S.P. Medium. A study of the U.S.P. medium involved an increase of the yeast preparation and sodium acetate and the addition of ammonium sulfate. Figure 3 shows the greater acid output per unit dose, increase in slope, and lengthening of the linear portion of the curve following a tenfold increase in the amount of U.S.P. yeast supplement. A fivefold increase, Figure 4, curve B , resulted in a less marked change. However, the latter amount of yeast supplement, with added ammonium sulfate and sodium acetate (Figure 4, curves D and E ) ,
J
E -lib
-1.L
-06
0
t06
Log
Micrograms Riboflavin Figure 5 . Curves of Response A.
Roberts-Snell medium with glucose increased to 3 70;points represent responses to 0.0625 0.125, 0.25, 0.5, 1, and 27 of riboflavi)n per test tube B . Plotted from data of Roberts and Snell ?5, p. 505); glucose content of medium 2 %
Comparison of Results with Certain Modiiied Media. Four modified media were selected (Table 11) which seemed to offer most possibilities for bioassay because of their long, straight doseresponse curves with steep slopm Figure 6 illustrates four typical standard riboflavin curves. Calculation of the statistics s2, b, and X (s/b) was limited to the pooled results of replicate tests on these four media, Table 111. The range over which the dose-response curve was a straight line did not vary markedly amongst the four media. In each
1250
ANALYTICAL CHEMISTRY
affected alike. For reasons mentioned, medium 11 is favored for assays, with I11 a close second. With a more objective method of estimating acid output, s2 would probably be much smaller and Ia Ib I1 I11 Approximate range of linearity medium I11 would then be best because of the of dose-response curve, y 0.0625-0.6 0.125-1, 0 0.0626-0.6 0.0625-0.5 steep slope and high sensitivity. Ib and Ia are Maximum acid produced, ml. 26 23 22 26 ?/lo less satisfactory. Variances for difference between duplicates, sampling error, s z 0.094 0.056 0.030 0.063 hIicrobiologica1 assays of concentrates (Difco Slopem of dose-response curve. b 13.13 14.26 13.63 16.34 Bacto-Yeast Extract and Sirup of Pantabee: the Expected precision5 of bioassays employing linear portion of latter an aqueous liver concentrate with added 0.0234 0.0166 curve, 8/b = A 0.0128 0.0164 Number of tests contributing to B complex vitamins, prepared by Charles C. 5 4 2 2 data Haskell and Co.) using the four media discussed a Statistirs sp, b, and X calculated from data lying on rectilinear portion of dose-response above were carried out. An example of such an curves. assay is presented in Figure 7 where the same sample was assayed by identical procedures four times. Only portions of the dose-response curves are shown. Because of the symmetrical arrangement of the assay it was possible to make use of polynomial coefficients in the analysis of variance (3). This analysis indicated that there was no signifi- u . cant curvature or departure from parallelism in the two dose0 3 20response curves of each pair. The potency and its standard error U L could therefore be computed from the variances for the difference n between samples, D2, and for slope BZ. 2 In estimating the log ratio of potency of the unknown, the equation o 'I- kID Table 111. Data Calculated to Evaluate Standard Riboflavin DoseResponse Curves, with Four Best Media, Using Results of Replicates Run on Different Days
/
fl
P -
s -
-.
Lir = xa - xu
s-
E
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Log Micro$rams Riboflavin : 0.3Dose Intervals Figure 6. Sample Dose-Response Curves Using Four of the Most Favorable Modified Media
was used, where and &&are the mean log doses for standard and unknown, k is d 8 / 3 for a 3-dose assay, I is the log-interval between successive doses, here 0.301, D and B are the square roots of the first two treatment varimces described by Bliss and Marks -Le., for difference between samples and for slope. The standard error of ill is
Points represent responses to 0.0625, 0.125, 0.25, 0.5, 1, and 2 y of riboflavin per test tube For description of media see Table I1
case, an analysis indicated no significant departure from linearity within this range. The maximum acid output was obtained with I a and 111,but the latter is preferred because of a smaller sampling error and a greater slope, both of which contributed to a greater precision in an assay, shown by a considerably smaller A. I n comparing Ia with Ib, which differ only with respect to yeast extract, ammonium sulfate, and sodium acetate content, one finds that the medium with the smaller proportions of these ingredients has a steeper slope, a smaller sampling error, and a greater precision as indicated by A. All these point to Ib as a better medium for riboflavin assays than Ia, although the sensitivity of Ib t o a given dose of riboflavin is somewhat less than that of Ia. The larger sampling error of Ia may be due to its darker color which may influence accuracy of titration. The modified U.S.P. medium I1 has a relatively low maximum acid output and slope. However, owing to a very small sampling error, s2, and a small A, this medium could be expected to yield results with precision greater than medium I11 (modified Roberts-Snell), although the latter has the steepest slope. The low sampling error of I1 is probably related to its pale color which assists in determining the end point of titrations. lMedium I11 is considerably more sensitive than medium I1 in that half the dose of riboflavin allows about the same acid output compared with that of the latter medium. The zero dose with medium I11 and the inoculum used by the author results in a titration of about 3 ml., 50% higher than with medium 11. This is not a serious drawback when one considers that the lowest response one needs to use is a t least three times this value, a larger gap than with the other media. Also both unknown and standard are
+
SM =
s k ~ d B 2 + ~ 2
where s is the standard deviation of the sampling error. The calculated data are in Table IV. The four media yield values for potency differing by only 8% and with overlapping confidence limits which are consistent with the values for A calculated from the pooled data for the standard curves (Table 111).
Table IV. Comparison of Four Assays Run on Same Sample of Concentrate Using Four Different Media Media Ia
Potency, y/gram Range, corresponding t o M f t s M yb
Confihence limits, % of estimated potency
Ib
I1
12 12 12 0.174 0.072 0.021 13.71 13.45 12.44 2.656 2.660 2.628 0.0115 0.0176 0.0062 452 457 424
I11 12 0.070 16.91 2.660 0.0096 447
414-495
431-485
411-438
427-468
92-110
94-106
97-103
96-105
Actual ratio + 6 , to convert t o terms of ?/gram. b t corresponds to a probability of 0.05 for 11 degrees of freedom.
Q
An assay of the same sample using original Greene-Black medium and U.S.P. doses of standard, yielded a curve of response with only two of the dose levels, 0.05 and 0.1 microgram, below the maximum level. Following the single-curve, graphic method, the riboflavin content of the concentrate was estimated from two dose levels of test material, by interpolation from this standard curve, as 430 and 460 micrograms per gram. Only 8
V O L U M E 25, NO. 8, A U G U S T 1 9 5 3
1251
78 .
24
-
2 -
0 /) ///// JI
Ib
jle: 3 =
z Q
0
12-
\
z -2
E
8
-
Standard
+ Unknown 1
1
1
,
1
,
,
1
1
1
1
1
1
1
1
1
1
1
1
Wood ( 2 1 ) has also found that the data of Roberts and Snell for riboflavin fit a curve in which the log-dose is linearly related to log-response for the range 0.04 to 0.16 microgram of riboflavin. This observation puts this assay in the same category as two other vitamins and 12 or more amino acids tested by a variety of microorganisms. Vood points out that the slope of the curve of response over a three- or fourfold increase in dose is 0.56. Replotting the loner portion of the curves obtained by the author indicated that a similar relationship with a slope about 0.5 may exist. However, the range of doses over which this seems to apply is only for about a twofold increase in dose and thus the use of this device is impracticable. Moreover, the mathematical procedures involve extra steps which do not seem to be justified if a simpler transformation holds. Price (13) found a straight line relating log-dose and acid production of L. casei over the range 0.04 to 0.125 microgram using the Greene-Black medium enriched with factors provided by yeast. Shils et al. ( 1 6 ) , using seeded agar instead of liquid media, fouiid a straight line relating log-dose of riboflavin and
,
1
,
1
~
confidence limits of riboflavin assays, but details are lacking. Wood ( 2 0 ) , in describing his procedure for slope-ratio calculation, presented one 20-tube assay which yielded confidence limits of 94 to 106% ( P = 0.05). The most direct comparison can be made with the results of Price and Graves (14) who, as the author did, used semilogarithmic dose-response curves. For an assay set up with 45 tubes, five levels for standard and four for unknown, they reported fiducial limits of 95 to 106% ( P = 0.05), account having been taken of temperature differences within the incubator. -4larger range was calculated for assays with fewer tubes. These
fluence of any temperature differences M hich may have existed. The effects, if any, appeared in the statistic, s. Evidence presented indicates that a high degree of accuracy may be expected of riboflavin microbiological assays, using L. casei. Anything which increases the slope, b, of the curve of response or decreases the sampling error, s, will enhance the confidence of an assay. Since s includes errors inherent in pipetting and titration as aell as temperature differences in the incubator, increasing uniformity or objectivity of these operations will enhance the accuracy of the assay. But the linear relationship discussed also makes it a simple matter to increase confidence in estimated potency by merely adding to the number of dose levels, decreasing the dosage interval, or increasing the number of tubes for each dose. Following rather simple mathematical steps, the contribution of each of these factors may be evaluated. ACKNOWLEDGMEKT
The author wishes to express her thanks to D. B. DeLury, director of the Department of LIathematical Statistics Ontario
1252
ANALYTICAL CHEMISTRY
Research Foundation, for valuable suggestions for the preparation of this paper and the statistical analysis. LITERATURE C I T E D
(1) Bliss, C. I., Ann. Math. Statistics, 17, 232-7 (1946). (2) Bliss, C . I., and Cattell, McK., Ann. Rev. Physiol., 5 , 479-539 (1943). (3) Bliss, C. I., and Marks, H. P., Quart. J . Pharm. and Pharmacol., 12,82-110,182-205 (1939). (4) Crandall, W. A,, and Burr, Muriel &I., Federation Proc., 6, 246-7 (1947). (5) Finney, D. J., Quart. J . Pharm. and Pharmacol., 18, 77-82 (1945). (6) Greene, R. D., and Black, A , J . Am. P h a r m . Assoc., Sci. Ed., 32,217-20 (1943). (7) Kent-Jones, D. W., and Meiklejohn, M., Analyst, 69, 330-6 (1944). (8) Kerly, Margaret, Biochem. J., 38, 423-5 (1944). (9) Light, A. E., and Clarke, &I. F., J. Biol. C h a . , 147, 739-47 (1943).
(10) Oser, B. L., Proc. I n s t . Food Technol., [ 6 ] ,169-84 (1946). (11) Pennington, D., Snell, E. E., and Williams, R. J., J . Biol. Chem., 135,213-22 (1940). (12) Pharmacopoeia of the C.S,.4., XII, First bound supplement, pp. 79-81,1943. (13) Price, S. A,, ’Irature, 156, 171-2 (1945). (14) Price, S. A , . and Graves. H. C. H.. Ibid.. 153. 461 (1944) (15) Roberts, E. C., and Snell, E. E., J . Biol. Chem., 163, 499-509 (1946). (16) S h h , hi. E., Horowitz, 31., Gelman, h.C., and Sass, hI., Science, 115,517-19 (1952). (17) Snell, E. E., and Strong, F. AI., IND.EXG.CHEM.,AXAL.ED., 11,346-60 (1939). (18) Society of Public Analysts, Analytical Methods Committee, Sub-committee on Vitamin Estimations, A n a l y s t , 71, 397406 (1946). (19) Toennies, G., and Frank, H. G., Growth, 14, 341-51 (1950). (20) Wood, E. C., Analyst, 71,l-14 (1946). (21) Ibid., 72,84-90 (1947). (22) Wood, E. C., and Finney, D. J., Quart. J . Pharm. and PhamnaCOZ., 19, 112-27 (1946). RECEIVEDfor review January 5, 1953.
Accepted April 14, 1953.
Accuracy of Quantitative Paper Chromatography in Amino Acid Analysis-Addendum R. C. SALANDERI, MARCUS PIANO, AND A. R. PATTON Chemistry Department, Colorado A & M College, Fort Collins, Colo. and Chism recently published a “quantitative” P method for paper chromatography of amino acids. In response t o inquiries regarding the accuracy of the method as well as ATTON
(6)
certain unpublished details of procedure, the following is presented. METHOD
One-microliter spots of 5 to 100 m M acid solutions are applied to Whatman No. 1 paper with a self-filling transfer-type pipet capillary, which is wiped dry before each application. Ascending chromatography is conducted for 18 to 24 hours in a suitable solvent. The solvents most often used are 80% phenol (Merck reagent grade) and water-saturated 2,4-lutidine (Matheson Co.). After development the solvent is allowed to evaporate a t room temperature for 18 to 24 hours. The papers are then sprayed with 0.3% ninhydrin in 95% ethyl alcohol and color development is allowed to take place in the dark a t room temperature for 18 hours. Maximal absorbances are read through the vertical axis of the colored spots in the Welch Densichron. Concentrations are derived from comparison with standard series run on the same chromatogram. The standard series curves are made by plotting peak absorbances against molar concentrations in a geometric progression. Increment curves which represent the distribution of color density in a spot are obtained by recording the absorbance every 3 mm. along the vertical axis of the spot.
Table I. Added, m M 6.25 5.00 2.50 1.25 5.00 5.00 5.00 5.00
densest area of the color spot as the concentration increases. Figure 1, B, is an example of a standard series curve for alanine. Alanine increment curves a t the higher concentrations are shown in Figure I, C. Using this technique, with water-saturated lutidine as solvent, 35 replicate spots of a casein hydrolyzate showed the sample to contain 8.3% lysine with a standard deviation of 1.5%. Block ( 3 ) gives 8.5% as an approximate value for lysine in casein. To test recoveries, various concentrations of lysine and alanine were added to the casein hydrolyzate as s h o m in Table I. The student method described by Patton ( 5 ) was tested by giving an “unknown” alanine solution to each of 11 students, none of whom had attempted quantitative paper chromatography before. Using the Welch Densichron and uncalibrated capillary .OS M
Recovery of Alanine a n d Lysine Recovered, mA4 Recovery of Alanine 5.60 5.37 2.50 1.22 Recovery of Lysine 5 00 5.00 4.75 5.12
Recovered, % ,0125
89.5 107.4 100.0 97.5
Rf
(mm)
,025
9 5
MOLAR C O N C E N T R A T I O N
1,
100.0 100.0 95.0 102.4
DATA AND R E S U L T S
D
Figure 1, A , shows increment curves of 1-pl. spots of alanine in the molar concentrations listed. Spot diameter increases with concentration and there is a slight increase in the R, of the 1 Present address, Department of Agricultural Biochemistry, Rutgers University, New Brunswick, N. J.
\\ \I Rf
(mmh
I
.os
0.1
0.E
MOLAR CONCENTRATION
Figure 1