I N STRONGLY ALKALINE SOLUTIONS ninhydrin (1,2,3 indantrione hydrate)

Transcription

1 Fluorimetric Determination of Creatine Rex B. Conn, Jr. I N STRONGLY ALKALINE SOLUTIONS ninhydrin (1,2,3 indantrione hydrate) combines with guanidine, monosubstituted guanidines, and N,N disubstituted guanidines to form highly fluorescent addition products (1). This reaction has been used as the basis for a simple, sensitive fluorimetric method for the determination of creatine in serum, blood, and urine. The most widely used methods for determining creatine are based upon the conversion of creatine to creatinine by heating in the presence of strong mineral acids and subsequent determination of the creatinine by the alkaline picrate reaction of Jaffe (2). The reaction is not specific for creatinine, and many naturally occurring substances give a similar color. The method has been criticized also because of the formation of noncreatinine chromogens and loss of creatine during heating. Many modifications of the alkaline picrate method have been proposed, but none has completely eliminated these objections (3). Creatine can be measured directly by the Barritt modification of the Voges-Proskauer reaction (4). This reaction is not specific for creatiiie since other naturally occurring guanidine compounds give a similar, although less intense, color with diacetyl. Sulfhydryl compounds interfere with color development and must be removed (5). Anderson et at. (5) have used the diacetyl reaction for the determina- From the Department of Laboratory Medicine, University of Minnesota, Minneapolis, Minn. This work was supported in part by Training Grant H.T.S given to the University of Minnesota by the National Heart Institute, U.S.P.H.S. The author wishes to express appreciation for advice given by Dr. 0. T. Evans and Miss Esther Freior of the Department of Laboratory Medicine and by Dr. Richard von Korif of the Department of Physiological Chemistry, University of Minnesota. Received for publication May 13,

2 538 CONN Clinical Chemistry tion of creatine after its separation from interfering materials with ion exchange resins. Another indirect method for the determination of creatine and other guanidinium compounds has been developed by Van Pilsum, Martin, Kito, and Hess (6). Tn this method creatine is converted to creatinine, the creatinine is degraded to methylguanidine, and the latter is determined by the nonspecific Sakaguchi reaction. Preformed creatinine is determined separately by degradation to methylguanidine and application of the Sakaguehi reaction. The difference between these values represents creatine. Van Pilsum (3) recently has reviewed current analytic methods for the determination of creatinine and related compounds. A recent development has been the application of a DPN-linked enzyme reaction to the determination of creatine by Tanzer and Gilvarg (7); however, their values for serum and erythrocyte creatine in animals cannot be compared to those in man. The fluorescent ninhydrin method for estimating creatine is simple and convenient, and it has a high degree of precision. The reaction itself is not specific for creatine; however, the fluorescence produced by creatine is greater than that produced by other guanidinium compounds, and specificity may be further enhanced by prior separation of arginine. METHODS REAGENTS Creatine Standard.* The standard should be recrystalized once from water before use. Standard solutions should give negligible color with the Sakaguchi and Jaffe reactions, and if necessary, creatine may be purified by the method of Hunter (8). Since there is a slow equilibrium between creatine and creatinine, new standard solutions should be made every day. Frozen aliquots of a stock creatine standard appear to be stable over a period of weeks, and their use eliminates daily weighing. Quinine Standard: 2 mg./l. in o.1n H2S04. It is convenient to use a stable fluorescent standard for calibration of the fluorimeter since the product of the alkaline ninhydrin reaction is unstable, and fluorimeter readings must be timed. The quinine solution is stable over long periods and may be used as frequently as necessary to check calibration of the instrument. Any convenient quinine concen- 8Distillation Products Industries, Rochester, N. Y.

3 Vol. No. 6, 1960 DETERMINATION OF CREATINE 539 tration may be used since this figure is not used in any calculations; furthermore, if the same quinine solution is used, the day-to-day readings of the creatine standards may be compared. 5.0% (w/v) Zinc Sulfate. One hundred grams of ZnSO4. 7H20 (Reagent Grade) are dissolved in double distilled water and diluted to 2000 ml. 0.3 N Barium Hydroxide. Ninety grams of Ba(0H2. 8H20 (Reagent Grade) are dissolved in double distilled water and diluted to 2000 ml. The solution is filtered if cloudy, and it should be protected from the atmosphere to prevent formation of barium carbonate. The exact concentrations of Reagents 3 and 4 are not critical, but they should neutralize each other exactly. The stronger solution should be diluted so that 10 ml. of ZnSO4 requires 10.0 ± 0.05 ml. of Ba(OH)2 to give a pink color with phenolphthalein. The Ba(OH)2 should be added slowly to prevent a false end point, and the pink color should persist for 1 mm. 1.0% (w/v) Alcoholic Ninhydrin Solution. One gram of Ninhydrin* is dissolved in 100 ml. of 95% ethyl alcohol. 10% (w/v) Alcoholic Potassium Hydroxide. The required amount of KOH (Reagent Grade) is dissolved in 95% ethyl alcohol. The solution should be filtered if cloudy, and evaporation should be avoided. Amberlite IRA -401 Ion Exchange Resin.t The resin is converted to the hydroxide cycle by immersion in 0.1N NaOH followed by thorough washing. The washings should be near neutrality and should show no appreciable fluorescence before the resin is used. A high blank reading indicates that the washing has been insufficient. After use the resin may be recycled with NaOH and reused 8-10 times before discarding. As in any fluorimetric procedure, it is desirable to reduce the background fluorescence of the blank to as low a value as possible, to enhance the sensitivity of the method. All reagents should be checked for fluorescence at the wavelengths used, and any showing appreciable fluorescence compared to distilled water should be discarded. FLUORIMETRY All fluorimetry was done with a Photovolt Fluorimeter consisting *Distilation Products Industries, Rochester, N. Y. trohm and Haas Co., Philadelphia, Pa. iphotovolt Corporation, New York, N. Y.

4 540 CONN Clinical Chemistry of a line-operated Multiplier Photometer Model 520-M and the Fluorescence Unit Model 54. An excitation beam was isolated using Corning glass filters No and No together, giving peak transmission at 410 m with a band pass of 70 m at half-maximal transmission. The emission beam was isolated using Corning glass filters No and No together, giving peak transmission at 525 mjh with a band pass of 60 rn/h at half-maximal transmission. DETERMINATION OF CREATINE IN SERUM AND BLOOD A barium hydroxide-zinc sulfate ifitrate is prepared using 1.0 ml. of serum or blood, 15.0 ml. of water, 2.0 ml. of 0.3N Ba(OH)2, and 2.0 ml. of 5% ZnSO4. The solution is mixed by inversion and filtered through fine ifiter paper. A filtrate blank is prepared by substituting water for serum. The creatine concentration in the filtrate must not exceed 2.5 X 105M (0.3 mg.%). Standard solutions containing 0.05, 0.10, and 0.20 mg.% are run with each set of determinations. A distilled water blank is included for the standard solutions since the fluorescence of the filtrate blank is slightly higher than that of one using water. For development of the fluorescent reaction, cuvettes are set up containing 2.0 ml. of filtrate, blank or standard, and 1.0 ml. of 1% alcoholic ninhydrin solution. One milliliter of 10% alcoholic KOH is added to each cuvette and mixed immediately. The fluorimeter is set to a convenient sensitivity using the quinine standard, and fluorimeter readings are taken 5 mm. after addition of the KOH. It is convenient to set up a series of cuvettes containing the unknowns and standards with the ninhydrin solution and then to add the KOH at timed intervals. Five minutes later the readings may be taken at intervals in the same order. Although the decay of fluorescence is accelerated slightly by light, it is not necessary to shield the cuvettes from ordinary lighting. Direct sunlight should be avoided, and the cuvette should not be placed in the excitation beam until fluorimeter readings are to be taken. After subtraction of the water blank, the fluorimeter readings for the standards are plotted on ordinary graph paper. The ifitrate blank is subtracted from the unknowns, and the concentration of creatine in the unknowns may be read directly from the standard curve. Since the relationship of fluorescence to creatine concentration is linear

5 Vol. 6, No. 6, 1960 DETERMINATION OF CREATINE 541 over the concentration limits given, an alternative method is to omit construction of the standard curve and to calculate creatine concentrations directly. DETERMINATION OF URINE CREATINE CONCENTRATION Urine contains substances that interfere with the direct determination of creatine by producing a fluorescence at the wavelengths used for creatine determination. These naturally fluorescent materials may be effectively removed by treating the urine with a strong anion exchange resin prior to the determination of creatine and by using a highly diluted urine sample. The excretion of these materials appears to be rather constant and was found in every case in a large group of individuals. Fifteen to twenty milliliters of a 1:50 dilution of urine are placed in a 50-mi. flask with about 3 gm. of IRA-401 resin, and the flask is agitated for 30 min. The resin is allowed to settle, and 2.0 ml. of the supernatant urine dilution are pipetted into a cuvette. One milliliter of alcoholic ninhydrin and 1.0 ml. of 10% alcoholic KOH are added. The procedure described above for the determination of creatine in serum is then followed. A blank consisting of distilled water treated with the ion exchange resin is included. It is essential that the urine be sufficiently diluted to make the final concentration of creatine less than 2.5 X 105M (3.3 mg./l.). Thus a 1:50 dilution may be used for creatine concentrations of up to 150 mg./l., and dilutions in pathologic states with creatinuria as high as 1:500 or higher may be required. RESULTS With the concentrations described above, the fluorescence due to creatine reaches a maximum at about 41/2 min. after adding KOH and then declines very slowly. The loss of fluorescence from 5-10 mm. after adding KOH is negligible under most circumstances, and the time of taking fluorimeter readings may vary a minute or so without loss of precision (Fig. 1). Under the conditions described, the relationship of fluorescence to creatine concentration (in filtrates and urine dilutions) is linear from 1.0 X 107M to 2.5 X 105M. At higher concentrations quenching becomes evident, and below 1.0 X 107M the fluorescence of the blank approaches that due to creatine. Recoveries of creatine added to serum prior to making protein-free

6 542 CONN Clinical Chemistry filtrates averaged 99.5 per cent (S.D., 1.9 in 10 experiments) over a range of added creatine from 5.0 X 105M to 2.5 X 10-4M ( mg.%). l0 >- i-8 Fig. 1. Development of z 6 fluorescence with time in fluorescent ninliydrin reaction. Equimolar solutions of W 4 creatine (A), dimethylguani. dine (B), guanidoacetic acid (C), arginine (D), methylo guanidine (B), and guani- 2 dine (F). U MINUTES Recoveries of creatine added to whole blood were similar, averaging per cent (S.D., 1.5 in a series of four experiments) over a range of added creatine from 1.0 X 10-4M to 4.0 X 10M ( mg.%). Recoveries of creatine added to urine prior to resin treatment averaged per cent (S.D., 2.3 in a series of five experiments) over a range of added creatine from 2.0 X 10-4M to 2.0 X 10-3M ( mg./l.). The serum creatine concentration from 19 normal adult male blood donors averaged 0.41 mg.% (S.D., 0.17), with a range of mg.%. Whole-blood creatine concentrations ranged between 3 and 4 mg.%. Normal males were found to excrete only small amounts of creatine. Urine concentrations ranged from 18.6 to 58.5 mg./l. on urine specimens from 6 normal adult males. Replicate samples correlate well when run in the same batch or in different batches. Average differences between replicate samples approximated 1% in a large series including urine, blood, and serum determinations.

7 Vol. 6, No. 6, 1960 DETERMINATION OF CREATINE 543 DISCUSSION FLUORESCENT REACTION The addition product of creatine with ninhydrin in alkaline solution is a highly fluorescent material with two sharp excitation peaks at 305 and 390 rn/h, the fluorescence at 390 rn/h being approximately twice as intense as that at 305 rn/h. The single emission peak has a maximum at 495 rn/h (1). Guanidine, monosubstituted guanidines, and N,N disubstituted guanidines form addition products with similar spectra, with the fluorescence intensity increasing in the order of listing (Fig. 1). In alkaline solution the five-mernbered ring of ninhydrin is opened to produce o-carboxyphenylglyoxal (9), which is yellow and interferes by strongly absorbing the excitation wavelengths. The degradation product of o-carboxyphenylglyoxal is o-carboxyrnandelic acid (9), which has negligible absorbancy at the wavelengths used. The fluorescent addition product apparently is formed by o-carboxyphenylglyoxal, but since this compound interferes with fluorimetry, it is necessary to delay fluorimeter readings until any excess is degraded to o-carboxymandelic acid. The intensity of the yellow color and its rate of disappearance are functions of the concentrations of ninhydrin and alkali, respectively. For this reason the concentrations of ninhydrin and alkali must be such that there is sufficient o-carboxyphenyiglyoxal to combine with all of the creatine, but not so much that degradation of the excess requires too long a time, thus delaying development of rnaximal fluorescence. The fluorescent compound probably is formed by the combination of the arnidine group with one or both of the adjacent ketone groups on o-carboxyphenylglyoxal (1). This is of interest because the fluorescent ninhydrin reaction has certain similarities to the diacetyl reaction in which adjacent ketone groups are involved. The specificity of both reactions is similar since free guanidines, monosubstituted guanidines, and N,N disubstituted guanidines give a color with diacetyl or a fluorescent product with ninhydrin, the color or fluorescence being greatest with N,N disubstituted guanidines. Neither the diacetyl reaction nor the fluorescent ninhydrin reaction takes place with N,N disubstituted guanidines, and thus creatinine, creatine phosphate, and guanine produce no color or fluorescent product. Although the fluorescent ninhydrin reaction may be used with most of the usual protein precipitants, the Ba(OH)2-ZnSO4 filtrate was

8 544 CONN Clinical Ch.misfry selected because more than 95 per cent of any arginine present is removed, and because the filtrate is neutral, thus eliminating any changes caused by neutralization of part of the KOll in the relationship of development of fluorescence to time. Concentrations of alcohol higher than the final 50% solution in the cuvette can not be used with the Ba(OH)2-ZnSO4 ifitrate because of the formation of a slight cloudiness that interferes with fluorimetry. Ninhydrin is widely used for the detection and estimation of alpha amino nitrogen and is most effective for this purpose in nonaqueous media in a p11 range of 5-7 (10). In contrast, the fluorescence reaction takes place at a ph above 13 and works equally well in aqueous and alcoholic media. At this ph the action of ninhydrin on alpha amino nitrogen is nil; however, it will give a color with the ammonium ion in high concentrations. The use of ninhydrin in a fluorescence reaction for determining y-aminobutyric acid and other amino acids has been described by Lowe, Robins, and Eyerman (11). This is evidently a different fluorescence reaction from the one described here, since it has a p11 optimum between 9.0 and 10.6 and gives little fluorescence above ph It forms fluorescent products with amino acids that are nonreactive in the method described here; also, the fluorescent spectra have different maxima. SPECIFICITi OF THE METHOD Since the fluorescent ninhydrin reaction is not specific for creatine, results are slightly high because of the presence of noncreatine compounds that give fluorescent products. In vertebrate tissue these compounds are arginine, guanidoacetic acid, methylguanidine, and guanidine. The latter two, if they are actually present and are not artifacts due to chemical manipulation (6), are present in such small amounts that any fluorescence that they might produce can be disregarded. In the method described for serum and blood more than 95 per cent of the arginine is removed by the Ba(OH)2-ZnSO4 filtrate. Since arginine produces less than J/5 the fluorescence of creatine (Fig. 1), the presence of an equimolar quantity of arginine would cause an error of less than 1 per cent in the determination of creatine. The removal of arginine from urine prior to estimation of creatine does not appear to be necessary except in states accompanied by aminoaciduria. Hoberman (12), using an enzyrnatic method, found that normal male urine contained less than 15 rng./l. of arginine. Arginine

9 Vol. 6, No. 6, (960 DETERMINATION OF CREATINE 545 can be removed prior to the application of the fluorescence reaction by treatment with an ion exchange resin, as described by Anderson etal. (5). Guanidoacetic acid is the compound that interferes most with the specificity of the method for creatine. In a series of experiments using various precipitants, methods of extraction, and ion exchange resins, no practical method was found for the separation of guanidocetic acid from creatine. This interference is not great in most situations, however, since equimolar concentrations of creatine give about four times the fluorescence of guanidoacetic acid (Fig. 1), and creatine occurs in higher concentrations than guanidoacetic acid. In whole blood the concentration of creatine is approximately 10 times that of guanidoacetic acid (6), and inaccuracies are small. The average excretion of urinary guanidoacetic acid is about 40 mg. per day (6, 13), and thus measured creatine concentrations are about 10 mg. per day too high. Values for serum guanidoacetic acid have been reported as rng.% (12), mg.% (13), and rng.% (14). Thus measurements of creatine in serum by the fluorescence method are probably rng.% too high because of the presence of guanidoacetic acid. It perhaps should be pointed out that available methods for determining guanidoacetic acid require the separation of much larger quantities of arginine, either on Permutit columns (13, 14) or by enzymatic destruction (12), before application of the nonspecific Sakaguchi reaction. These reported values may, therefore, be higher or lower than those actually present. INTERFERING MATERIALS Many naturally occurring substances can interfere with the development of the fluorescence reaction; however, few of these are present in sufficiently high concentration to impair the usefulness of the method. These interfering substances are of three types: (1) those whose inherent fluorescence causes fallaciously high readings, (2) those that inhibit the development of fluorescence due to creatine, and (3) those that react with ninhydrin to produce fluorescent products. The latter have already been discussed. Compounds of the first type interfere with the determination of creatine in urine but can be removed with ion exchange resins, as described. The batch procedure appears entirely satisfactory, removing more than 93 per cent of these materials. For even more complete removal, running the urine dilution through a column of the resin

10 546 CONN Clinical Chemistry (8 X 200 mm.) is highly effective. If urine samples are not treated with the resin before application of the fluorescence reaction, results are about rng./l. too high, due to background fluorescence. Resin treatment lowers this interference to about 1 mg./l. The variation in background fluorescence of urine from specimen to specimen is small enough that errors caused by using the resin-treated water blank for all samples average less than 1 mg./l. Background fluorescence of serum filtrates is approximately the same as that of the filtrate blanks, and use of resins does not appear to be necessary. Although there is a slight variation in the background fluorescence from ifitrate to filtrate, it is small compared to the fluorescence developed by the ninhydrin reaction, and it can be disregarded. Numerous compounds have been found to inhibit the fluorescence produced by the ninhydrin reaction. These conditions were found, however, only with extremely high concentrations, which are very unlikely even under pathologic conditions. For example, a urea concentration of 1000 mg.% decreases the fluorescence developed by creatine by about 5 per cent. Sulfhydryl compounds, which interfere with the diacetyl reaction, produce negligible interference with the fluorescence reaction until the concentration reaches about 100 rng.%. The recoveries of creatine added to blood, serum, or urine approximate 100 per cent and indicate that interference with the fluorescence reaction by the constituents of body fluids must be small. COMPARISON WITH OTHER METHODS The fluorescent ninhydrin method for the determination of creatine possesses considerable advantage over currently available methods in degree of sensitivity and simplicity. After preparation of the protein-free filtrate, the fluorescence reaction is carried out in the cuvette, and thus the actual mechanics of the method approach the simplicity of common colorimetric procedures. The theoretic disadvantages, as well as the laboratory manipulations involved in the indirect methods, which require conversion to creatinine, are avoided. The higher sensitivity of the fluorescent method is an added advantage, since the normal serum creatine concentration is at the lower limits of sensitivity for available colorimetric methods. Normal values of serum and urine creatine concentrations agree well with published values obtained with the best colorimetric methods (Table 1). The values obtained for urine creatine concentrations are in the lower range of reported values (Table 2), partly because

11 Vol. 6, No. 6, 1960 DETERMINATION OF CREATINE 547 Table 1. REPORTED NORMAL SERUM CREATINE CONCENTRATION Value,,ng.% Method Reference Fluorescent ninhydrin... O Picric acid Picric acid 14 O Destruction by bacterial enzyme; Picric acid (S.D. = 0.21)* Picric acid (S.D. = 0.15)* o-nitrobenzaldehyde ± 0.15 Ion exchange separation; 5 Diacetyl *Plasma I Arterial, as creatinine. As creatinine. methods that require conversion of creatine to creatinine are inaccurate in the presence of the large amounts of creatinine normally found in urine (6). The present values, however, are comparable to those obtained with the direct diacetyl method (17) and the DPNcoupled enzymatic methods (7). The normal range of creatine concentrations in whole blood was not determined since creatine within the erythrocyte does not follow water distribution and the physiologic significance of this intracellular creatine is not known. SUMMARY A new fluorirnetric method for determining creatine in serum, blood, and urine is described. The method is based upon the coupling of guanidine compounds with ninhydrin in alkaline solution to produce highly fluorescent products. In degree of sensitivity and simplicity the method possesses several advantages over currently used Table 2. REPORTED NORMAL URINE CEZAnNE VALUES Volue Method Reference mg./l. Fluorescent ninhydrin mg./day Picric acid mg./day Picric acid ef mg./day Pieric acid cf. 6 <150 mg./day onitrobenzaldehyde mg./l. Bacterial enzyme destruction; 17 Diacetyl 48.4 ± 26.2 mg./day Ion exchange separation; 5 Diacetyl nig./day DPN-linked enzyme system 7

12 543 CONN Clinical Chemistry colorimetric methods, and creatine can be determined in concentrations as low as 1.0 X 10-7M. The serum creatine concentration was found to be 0.41 mg.% in normal adult males, who excrete only small amounts of creatine in the urine. REFERENCES 1. Conn, B. B., and Davis, B. B., Nature (London) 183, 1053 (1959). 2. Polin, 0., J. Biol. Chem. 17, 475 (1914). 3. Van Pilsum, J. F., in Wick, D., Ed., Methods of Biochemical Analysis. New York, Iiiterscience Publishers, (1959), vol 7, p Eggleton, P., Elsden, 8. B., and Gough, N., Biochesa , 526 (1943). 5. Anderson, D. R., Williams, C. M., Knee, G. M., and Dowben, R. M., Biochem. J. 67, 258 (1957). 6. Van Pilsum, J. F., Martin, B. P., Kito, E., and Hess, J., J. Bioi. Chem. 222, 225 (1956). 7. Tanzer, M. L., and Gilvarg, C., J. Biol. Chem. 234, 3201 (1959). 8. Hunter, A., Creatine and Creatinine. New York, Longmans (1928). 9. Buhemann, S., J. Cliem. Soc. 97, 2025 (1910). 10. Meyer, H., Bwohem. J. 67, 333 (1957). 11. Lowe, I. P., Robins, E., and Eyerman, G. S., J. Neurochem. 3, 8 (1958). 12. Hoberman, H. D., J. Bwl. Cheon. 167, 721 (1947). 13. Levedahl, B. H., and Samuels, L. T., J. Biol. Cliem. 176, 327 (1948). 14. Sandberg, A. A., Hecht, H. H., and Tyler, F. H., Metabolism, GUn, and Exptl. 2, 22 (1953). 15. Allinson, M. J. C., J. Biol. Chem. 157, 169 (1945). 16. Van Pilsuin, J. P., and Bovis, M., Clin. Chem. 3, 90 (1957). 17. Ennor, A. H., and Stocken, L. A., Biochern , 310 (1953).