minutes of undisturbed standing is sufficient for full development of the barium sulfate turbidity, after which the absorbance remains constant for a t least 40 minutes. Sulfur-35 Tracer Study. A dilute solution of sulfur-35 as sulfate dissolved in dilute hydrochloric acid jerved as the tracer in t h e following series of experiments, in which a fixed amount was added to both samples and controls. The samples were digested, nThile controls ivere left undigested to determine sulfur loss caused by digestion. Known amounts of sulfate as ammonium sulfate \\-ere pipetted into a digestion vessel containing 1 ml. of 10% sodium nitrate and 4 ml. of 25% chloric acid and digested as outlined above. -4fter cooling, the residue was dissolved in 7 5 ml. of distilled water and transferred to 150-ml. beakers. One milliliter of 0.05M ammonium sulfate was added, which served as a carrier. The solution was made just acid to methyl red n-ith dilute hydrochloric acid and brought to boiling. One milliliter of 0.1M barium chloride dihydrate was then added dropm-ise n i t h stirring. The beakers 11ere covered with watch glasses and placed on a sand bath which maintained them lust below the boiling point for 2 hours. The samples were then filtered n i t h suction on a Hirsch funnel through Whatman S o . 42 filter paper. The precipitates were dried under an infrared lamp and counts Kere made for 5 minutes on each precipitate using a mica end window Geiger tube (Tracerlab TGC-2) and scaler. Corrections were made for the background count. The recovery of S35is dependent on the maintenance of a constant teniperature during the digestion process. This n a s accomplished n i t h a 3 X 3 X 12 inch aluminum heating block. Thirteen
holes 11/8 inches in diameter and 21/2 inches deep on the long surface of the block accommodate the test tubes used in the digestion. A thermometer well 21/2inches deep at one end of the block makes accurate temperature readings possible throughout the digestion. The block is heated by a Variac-controlled hot plate. Table I11 illustrates the recovery of sulfur-35 at 140' and 120' C., respectively. A temperature of 120' C. is most suitable, because losses are fairly consistent, ranging between 4 and 7%. A higher temperature gives erratic results. A temperature lower than 120' C. would prolong the digestion period considerably. Quantitative Determination of Sulfur. Compounds with sulfur in various oxidation states were used in this study. Table I V shows t h e per cent recovery of microgram quantities of sulfur after chloiic acid digestion in 25 x 150 mm. test tubes a t 120" C. measured against digested controls and illustrates the precision of this method. Chondroitin sulfate-Na, p-toluenesulfonic acid monohydrate, and p-toluenesulfonamide are representative of sulfur present in the +6 oxidation state. Recovery of the sulfur from 21 samples containing known amounts of sulfur averaged 99.2% with an average standard deviation of 1 3 . 0 y of sulfur when measured against digested standard ammonium sulfate. When compared to digested standard ammonium sulfate measured in the same way, sodium sulfite (S+4)showed an average of 907, recovery. The standard deviation in this instance was h5.0 y of sulfur for three samples. Cysteine hydrochloride monohydrate and l-cystine represent a sulfur valence state of -2. Recovery of sulfur from eight
samples averaged 787,, n ith an average standard deviation of +3.0 y of sulfur when compared t o digested standard ammonium sulfate. \Then sulfur is present in the lorrer valence states, there is an appreciable loss due to volatilization of sulfur dioxide during chloric acid digestion. The turbidity curve for sulfur over a range of 10 to 300 y describes a straight line. Digestion vessels other than the test tubes used in this study (beakers and Erlenmeyer flasks) permitted losses of sulfur-% ranging from 10% for Erlenmeyer flasks to 25% for 150-ml. beakers for S+6 when compared to undigested standard ammonium sulfate. The addition of sodium nitrate prevents losses arising from the vapor pressure of sulfuric acid a t 120' C. (1). After experimentation with 50- to 150-mg. amounts of sodium nitrate, 100 mg. was found the most suitable. The total digestion time required is between 3 and 4 hours. This may be reduced to 2 hours by first evaporating the contents of the digestion tube containing a boiling chip to one third of the initial volume over a microburner. S o additional losses are caused by this procedure, because only water is being evaporated. \There numerous samples are being analyzed, nothing is gained b y this individual manipulation. LITERATURE CITED
(1) Toennies, G., Bakay, B., - 4 s . i ~CHEM. . 25,160 (1953). ( 2 ) Zak, B., Willard, H. H., 31yers. G . B., Boyle, L4.J., Ibzd., 24, 1345-8 11952).
RECEIVED for review October 22, 1959. iiccepted January 2, 1960. IT-ork supported by grants-in-aid from the Michigan Heart Association and Sational Institutes of Health.
Fluorometric Microdetermination of Human Serum Albumin JOSEPH J. BETHEIL Department of Biochemistry, Albert Einstein College of Medicine, Yeshiva University, New York 61,
A new fluorometric procedure for the determination of human serum albumin is based upon the interaction in alkaline solution of albumin with a commercially available dye, Vasoflavine. The method is specific, highly sensitive, and rapid. The albumin content of a 5-pI. sample of serum can b e determined wiih precision. A comparison with procedures which are commonly employed for the deter-
560
ANALYTICAL CHEMISTRY
mination of serum albumin indicates that the new procedure yields comparable results and is suitable for routine use.
T
HE quantitative determination of serum albumin is useful in the diagnosis and treatment of many clinical entities. Several methods for this determination are based upon the separa-
N. Y.
tion of the protein components of serum by salt (8)or solvent fractionation (6, 6 ) . Most procedures are difficult to carry out and are timeconsuming in routine clinical use. The procedure described 1' based upon the interaction in alkaline solution of human serum albumin nith R commercially available fluorescent d!.e, T7asoflavine, in the presence of an amine buffer system (3). The ability of
albumin to interact 11ith certain dyes has been the basis of a number of methods for measuring the quantity of this protein in serum (4, 7 , 9 ) . The present method is exceedingly sensitive, is specific for serum albumin, and can be performed rapidly. I n alkaline solution, T'asoflavine (a sulfonated, methylated, benzothiazole derirative), upon interaction n-ith human serum allnmiin, exhibits an enhancement or increase in fluorescence ( 2 , 3 ) . This is related linearly to the albumin concentration of the solution ( 3 ) . Thus, n-ith :I given T'asoflavine concentration the fluorescelice increases linearly 11 ith increasing albumin concentration until all of the T'asoflavine has interacted n ith albumin. The fluorescence level reaches a plateau upon further addition of albumin. JJ-hen a comparibon nas made of the absorption spectra of Vasoflarine and of T'asoflavine-albuniiii mixtures in alkaline solution, a shift in the absorption maximum of the d j e toward higher wave lengths n i t h an increase in the absorptivity a t the absorption maximum was observed. Subsequent to the development of the method described, the interaction of T'asoflavine and human serum albumin n as studied spectrofluorometrically employing the Farrand spectrofluorometer. I n alkaline solution the dye has an excitation maximum a t 360 nip which is qhifted to 390 mp upon interaction n ith human serum albumin. The fluorescent peak for the d j e when activated a t its excitation ma.;imum (360 nip) is at 440 nip, while that of the dye-human serum albumin coniplex activated a t its excitation maximum (390 mp) is a t 420 nip. The spectrofluorometric behavior is consistent n ith the changes observed in absorption spectra. I n evaluating the ne17 procedure for the determination of human serum albumin based upon albumin-Vasoflavine interaction, the conditions dewibeci below have been employed. Vnder these conditions the interaction of human serum albumin with the dye is highly specific. S o n e of the albuminfree fractions obtained from human plasma by alcohol fractionation ( 5 ) interact n-ith Vasoflavine. There is also no interference with albuminVasoflnvine interaction by any of the other serum proteins. The interaction exhibits a stoichiometric relationship between reactants and results in the emission of a fixed quantity of fluorescent energy under specified conditions. If the dye concentration is varied, the range of albumin concentrations ivhich can be determined will also vary. A linear relationship between fluorescence and albumin concentration can be obtained with Vasoflarine levels as high as 0.5 y per ml. and as low as 0.125 y per ml. At the higher dye concen-
tration the sensitivity of the method is decreased, as the increment in fluorescence which is obtained for each microgram of albumin is relatively lower than that obtained a t the lower dye concentration. This is due to the fact that n ith higher dye concentrations, the fluorometer must be standardized with a solution n hich has greater fluorescence and the fixed quantity of fluorescence obtained by the addition of albumin will be compared with a solution of greater fluorescence. METHOD
ETH.I~LCNEDI.4~llNE-clTRATEBUFFER, PH 9.0. T o 21 grams of citric acid monohydrate dissolved in 500 ml. of water is
added, with rapid mixing, 10 ml. of 98% ethylenediamine (Eastman Grade, white label). This is diluted to about 900 ml. and then adjusted to p H 9.0 with sodium hydroxide. The volume is made up to 1 liter after final p H adjustment. VSSOFLAVINESTOCKSOLUTIOK contains 40 y per ml. of Vasoflavine in distilled water. Vasoflavine can be obtained from the National Aniline Division, Allied Chemical Corp., S e w Pork. N. Y. KORKING
The foregoing procedure should be repeated daily to check reagents and must be carried out \\-hen nerr reagents are prepared. Reproducible solutions and results are obtained ~ ~ i t h o utliffit cult- if this procedure is followed. T'BSOFLAVISE
Apparatus. T h e procedure has been developed for use n-ith several readily available fluorometers. Because t h e level of fluorescence evoked by t h e interaction of dye a n d albumin is high, i t can be measured with instruments which do not have photomultiplier detection units. T h e procedure a s described has been conducted routinely with t h e Lumetron Model 402 EF and with the Coleman Model 12C fluorometers. The Farrand photofluorometer, which employs a photomultiplier, can also be used if provision is made for reducing the intensity of incident energy and also for filtering out a portion of the fluorescent energy. This can be accomplished readily by insertion of a special aperture stop for the incident beam and filtering of the fluorescent beam through a neutral density filter. The follo\+ing filters should he used. The primary filter for the incident beam is Corning No. 5874 and the secondary filter consists of a combination of Corning S o s . 3389 and 4308. These filters are the same as those commonly used for the determination of thiamine by fluorescence methods. Reagents. QUININE STAXDARD SOLUTIOXcontains 0.845 y of quinine sulfate per ml. in O.1N sulfuric acid.
VA'SOFLAVIXE
should read 78. If it does not, the concentration must be adjusted t o give that reading. Different samples of dye vary somcnhat and this comparison permits the preparation of Vasoflavine solutions which exhibit the same initial fluorescence as well as the same fluorescence increment JTith albumin.
SOLUTION
contains approximately 0.4 y per nil. of dye in ethylenediamine-citrate buffer. This reagent is prepared by a 100-fold dilution of the Vasoflavine stock solution with the buffer. The fluorescence of this solution should be compared with that of the quinine sulfate standard as follows. The fluorometer is adjusted to give a reading of 80 with quinine standard solution and a reading of zero with n-ater. K i t h the fluorometer so adjusted, the Vasoflavine working solution
DILCEXT SOLUTION FOR
MICROPROCEDGRE contains 0.16 y per ml. in etliylenetlianiine-citrate buffer. The Vasoflavine working solution is diluted 2.5-fold with bhe buffer. The reagents are stable for a t ltaast one month u-hen kept a t room teniperature. For masimum stability, the reagents should he kept in amber bottles under refrigeration. These conditions result in no change for a t least 6 months. PROCEDURE
Preparation of Standard Curve for H u m a n Serum Albumin. Conditions for t h e determination of serum albumin by this method a r e satisfactory with a final dye concentrabioii of 0 16 y per nil. in ethylenediamine-citrate buffer, p H 9.0. The procedure is standardized using a sample of pooled normal human serum. This pooled sample should be analyzed for allwniin content by the Tiselius moving-boundary electrophoresis procedure, the methanol fractionation procedure ( 6 ) : or the sodium sulfate-sodium sulfite fract'ionation procedure (8j. The use of pooled serum is necessary, as preparations of purified albumin n-hich are available commei,ci:illy are not reliable as standards for this procedure. Human serum must he used because species differences exist n-ith regard to the fluorescence incremmt which is observed. The stantlard curve for the T':isoflavine procedure covers a range of 0 to 5 y of albumin per nil. (final conwntration) and is obtained as follows. Dilute 0.1 ml. of serum of known albumin content with the buffer solution to a volume of 200 ml. to establish the standard curve. Use the same curve for both the semimicro- and micromodifications of the method. Add graded amounts of this diluted swum to 25-ml. volumetric flasks containing 10 ml. of Vasoflavine working solution. Then make the flasks up to volume with the buffer and mix the contents. Transfer the contents of each flask t o fluorometer tubes and read in a fluorometer. Use Vasoflavine working solution to adjust the fluorometer to give a reading of 100. Use the blank tube containing no albumin to adjust the lo&-errange of the fluorometer scale. It should give a reading of 40 or lower. VOL. 32, NO. 4, APRIL 1960
561
Table
1.
Precision of Method
Serum 1
2 3
9 10
./
Vasoflavine
Figure 1 . Comparison of Vasoflavine interaction and sa It fractionation procedures
Av. Value and Dev., Albumin, Grams % 2.81 i 0.06" 4.35 i 0.05 3.78 0.07 4.90 f 0 10 4.40i 0.09 3.80 f 0.05 4.96 =k 0.07 4 . 0 7 i 0.10 4.27 dz 0.11 4.28 i 0.05
*
0 Identical results obtained with more than one sample. Results from 73 samples a r e summarized here
Average of four determinations on each sample.
The scale can be expanded or contracted by adjusting the resistor knob which is used to suppress the fluorescence of the blank solution. The range of readings should be expanded as much as possible with a particular instrument, because this nil1 result in the greatest sensitivity for each increment in fluorescence reading. The fluorometer readings are plotted us. albumin concentrations. A straight-line curve is obtained M hich can be used for the analysis of sera of unknown albumin content. This standard curve is reproducible. However, the practice has been to repeat it daily. This serves as a check on the procedure and is Jvorth the additional small effort. Analysis of Unknown Solutions. T h e same standard albumin curve is used for both the semimicro- and microprocedures. The final concentration of all components is the same in both procedures; t h e difference between the t n o is in the size of the serum aliquot 11hich is required and the dilution factor.
SEMIMICROPROCEDURE. Pipet 0.1 mi. of serum into a 200-mi. volumetric flask and dilute to volume with distilled water. Accurately transfer 3 ml. of this diluted serum to a 25-ml. volumetric flask containing 10 ml. of Vasoflavine working solution. Then adjust the volume to 25nil. with the buffer solution and mix the contents of the flask. Transfer the solution to a fluorometer tube and read in the fluorometer. Adjust the fluorometer with Vasoflavine working solution and with the reagent blank as for the standard curve. ~IICROPROCEDURE. To a 100-ml. volumetric flask containing Vasoflavine diluent solution add 5 pl. of serum, using a n accurately calibrated micropipet. Make up to volume with Vasoflavine diluent solution, transfer the mixed solution to a fluorometer tube, and determine the fluorescence reading as for the semimicroprocedure. Calculation of Results. If the albumin standard curye is drawn directly in terms of grams of albumin per 100 ml. of serum, t h e results can be
562
ANALYTICAL CHEMISTRY
/ , I
I
2
I
3
I
4
I
5
Albumin, grams per 100ml. Salt Fractionation Procedure.
obtained directly from t h e curve. If the curve is set u p in terms of micrograms of albumin per milliliter (final concentration), a proper dilution factor must be calculated and the value obtained from the curve multiplied b y this factor to obtain t h e final result. When both the semimicroand microprocedures are used in the same laboratory, the use of a factor may be more convenient, as the same curve is used for both procedures. EVALUATION
OF
VASOFLAVINE METHOD
The results using the Vasoflavine procedure for the determination of human serum albumin have been compared with those obtained by procedures which yield values for albumin in good agreement with moving boundary electrophoresis measurements. For this comparative study, the methanol fractionation procedure of Pillemer and Hutchinson (6) and salt fractionation with sodium sulfate and sodium sulfite (8) have been employed. For the methanol fractionation, protein concentration was determined by a semimicro-Kjeldahl nitrogen volumetric procedure. For the salt fractionation, both the semimicro-Kjeldahl and the biuret procedure were used to determine protein concentration. Good agreement \vas obtained between the tn-o procedures for protein determination. The results presented for the Vasoflavine-albumin interaction procedure were obtained using the semimicroprocedure described above, unless otherwise noted. The reproducibility of the Vasoflavine procedure is illustrated in Table I. Samples analyzed in quadruplicate yielded values the average deviation of which varied between 1 0 . 0 5 and 0.1 gram yo of albumin. When human serum albumin in known quantities (within physiological
limits) was added to normal serum, the amount recovered as determined by the Vasoflavine interaction method was between 97 and 105% of the quantity added. This is within the experimental error for the procedure. The methanol fractionation and the Vasoflavine procedures are compared in Table 11. These samples were obtained from normal individuals. Table I11 compares the two methods with serum samples obtained from patients with various pathological entities. All of these sera exhibited low albumin values and abnorinal albumin-globulin ratios. There is reasonably good agreement betreen the results obtained by the two methods for sera exhibiting both normal and pathologically low albumin levels. Because the salt fractionation procedure ( 8 ) is most commonly employed for the determination of albumin in clinical chemistry laboratories, it \vas extensively compared with the Vasoflavine procedure (Figure 1). The samples consisted of both normal and pathological sera. The data obtained illustrate that agreement between the two methods is good. I n another large series of routine serum samples obtained from the Clinical Chemistry Laboratory of the Bronx Municipal Hospital Center, there \\as no statistically significant difference between the results obtained by the salt fractionation and Vasoflavine interaction procedures. The semimicro- and micro modifications of the Vasoflavine interaction procedure are compared in Table IV. The results obtained n-ith the microprocedure are in good agreement with those obtained by the semimicro- and the salt fractionation procedures. It can be concluded from these studies that the Vasoflavine procedure yields results which are reliable for clinical use.
DISCUSSION
A procedure for the determination of human serum albumin based upon its interaction with Vasoflavine has been critically evaluated. The new method yields results which are in agreement with those obtained by standard, accepted procedures for albumin determination. The Vasoflavine procedure has been employed in the Clinical Chemistry Laboratory of the Bronx Municipal Hospital Center for an 18-month period, during which 12,500 determinations have been performed. .4n analysis of the data obtained indicates that good agreement also exists with the results obtained by paper electrophoresis employing the Spinco Paper Electrophoresis Apparatus Model R. No difficulty has been encountered in the routine use of the T’asoflavine procedure. When only the albumin content of the serum is desired, the Vasoflavine interaction procedure eliminates the necessity for the fractionation of serum proteins. Similarly, this method, when combined with a suitable procedure for the determination of the total protein content of serum, provides a convenient and simple technique for the determination of the albumin and globulin concentration of serum without a fractionation procedure. The usefulness of the procedure is enhanced by the rapidity with which it can be conducted, and by its sensitivity. Albumin determinations can be performed on serum obtained from fingertip samples of blood. This is of special importance in pediatric studies where only small quantities of blood are available. Serum obtained from blood samples in which hemolysis has occurred can be accurately analyzed for albumin by this method. The diagnostic dyes, bromsulphthalein (BSP) and phenolsulfonphthalein (PSP), even at high concentrations do not interfere with the determination. Similarly, roentgenographic contrast media which are administered orally or parenterally do not interfere. I n sera obtained from some patients with cirrhosis of the liver, albumin values obtained by the Vasoflavine method are lower than those found by the salt fractionation procedure (1). This discrepancy has been encountered in a small number of sera in which the bilirubin concentration is 7 mg. per 100 ml. or higher. It is possible that the dye and bilirubin may occupy a common binding site on the albumin molecule. Further studies are in progress to determine in detail the extent and nature of the effect of these high bilirubin levels.
Table II.
Comparison of Vasoflavine and Methanol Fractionation Methods as Applied to Normal Human Sera
Albumin, Grams yo Vasoflavine Methanol Difference
KO. 1 2 3 4
3.7 3.7 4.6 3.5 3.9 4.1 3.7 4.2 3.9 4.9
5 6
-
i
9 10
3.7 3.6 4.2 3.4 3.8 3.8 3.4 3.6 3.7 4.5
Albumin, Grams Vasoflavine Methanol Difference
So.
0 +0. 1 +0.4 +o. 1 $0.1 +0.3 $0.3 +0.6 +0.2 f0.4
11 12
4.3 3.3 4.0 4.6 3.5 3.9 5.0 3.8 4.5 3.6
l14 3
16 17 20 l9
3.6 3.7 3.8 4.0 3.5 3.7 4.5 3.7 4.3 3.4
+0.7 -0.4 $0.2 $0.6
0
+0.2 +0.5 $0.1 +0.2 +0.2
Values are averages from duplicate analyses. Table 111.
Comparison of Vasoflavine and Methanol Fractionation Methods as Applied to Pathological Human Sera
AlbuminGlobulin Ratio
KO.
1
T‘asoflavine
Albumin, Grams yo Methanol
2.4 2.6 2.8 2.7 2.8 2.1 2.7 2.4 2.0 2.7
2.2 2.4 2.8 2.8 2.5 1.7 2.9 2.5 1.7 2.5
0.58 0.60 0.80 1.03 0.63 0.27
2 3 4 5 6 7 8 9 10
...
...
0.30 0.72
Difference $0.2 +0.2
0
-0.1 $0.3 +0.4 -0.2 -0.1
$0.3 +0.2
Values are averages from duplicate analyses.
ACKNOWLEDGMENT
The author thanks J. U. Schlegel, director, Busn ell Urological Research Laboratory, University of Rochester School of Medicine and Dentistry, where this study was initiated, for his cooperation and encouragement. He also thanks Albert Hanok, clinical chemist of the Bronx Municipal Hospital Center, S e w York, for providing some of the samples used in this study, Sam Sorof, Institute for Cancer Research, Philadelphia, Pa., for performing the moving boundary electrophoresis measurements on the serum used for the standardization of the method, and Myron hlehlman for technical assistance. The Vasoflavine used in this study was provided b y the National hniline Division, Allied Chemical Corp. LITERATURE CITED
( 1 ) Betheil, J. J., Hanok, A., Federation Proc. 17, 190 (1958). (2) Betheil, J. J., Johnston, H. W., Schlegel, J. U., Ibid., 13, 13 (1954). (3) BFtheil, J. J., Moukous, R., Schlegel, .T. Ti.. Ihid..> 12. 15 11953). - ~ - , ~
Table IV. Comparison of Micro and Semimicro Vasoflavine Procedures
Albumin, Grams $& Vasoflavine %lt Semi- fractionaMicro micro tion V U l Y
Serum 1 2 3 4 5 6 7 8 9 10
4.7 4.3 4.5 3.6 4.4 4.0 4.3 3.9 4.3 3.9
4.9 4.3 4.5 3.8 4.5 4.1 4.3 4.0 4.3 4.0
4.9 4.3 4.6 3.7 4.4 4.1 4.4 4.0 4.2 4.1
Values are averages from duplicate analyses.
(7) Rees, V. H., Fildes, J. E., Lawrence, D. J. R., J. Clin. Pathol. 7, 336 (1954). (8) Reinhold, J. G., “Standard Methods of Clinical Chemistry,’’ Vol. 1, p. 88, Academic Press, New York, 1953. (9) Rutstein, D. D., Ingenito, E. F., Reynolds, W. E., J. Clin. Invest. 33, 211 (1954).
\
(4j-Bracken, J. ’S., Klotz, I. M., Am. J.
Clin. Patho!. 23,1055 (1953). (5) Cohn, E. J., Strong, L. E., Hughes, W. L., Jr., Mulford, D. J., Ashworth, J. K., hlelin, M., Txylor, H. L., J . Am. Chem. SOC.68, 459 (1946). it?’, Pillemer. L.. Hiitchir Eon, M. C.,
RECEIVEDfor review August 12, 1959. Accepted December 21, 1959. Study supported in part by a grant (No. A1415C) from the National Institutes of Health, United States Public Health Service.
VOL. 32, NO. 4, APRIL 1960
563
