- Email: [email protected]
Colorimetric and fluorometric carbohydrate determination with p-hydroxybenzoic acid hydrazide
BIOCHEMICAL
MEDICINE
7,
274-281
Calorimetric
(1973)
and
Fluorometric
Determination
with Acid
Carbohydrate
p-Hydroxybenzoic
Hydrazide M. LEVER
Pathology
Department,
Greenlane Received
Hospital, Auckland April
13,
3, New Zedund
1972
Aromatic acid hydrazides and reducing carbohydrates react in alkaline solution to give products suitable for the calorimetric and fluorometric assay of reducing sugars ( 1). Optimum results are obtained with p-hydroxybenzoic acid hydrazide (PAHBAH). A sensitive procedure has been described (1) which was found to be subject to interference from calcium and from high concentrations of bovine serum albumin. Described here are improved analytical methods that can be used over a wide range of glucose concentrations, from the calorimetric determination of serum glucose on less than a 1-J sample to the fluorometric determination of parts per billion glucose on a 0.5-ml sample. METHODS
instrumentation
Absorbance was read with a Gilford 300 N microsample spectrophotometer. Absorption spectra were scanned with a Unicam SP SOOA doublebeam spectrophotometer. Fluorescence measurements were made with an Aminco-Bowman spectrophotofluorometer fitted with a xenon lamp and an R 136 photomultiplier; mirrors and l-mm slits were placed in the cell housing. When recording spectra, O.l-mm slits were placed on the scanning monochromator side of the cell housing. Reagents Most reagents were obtained from BDH Chemicals Ltd., Poole, England; p-hydroxybenzoic acid hydrazide and diethanolamine were obtained from Fluka AG, Buchs, Switzerland. The bovine serum albumin was from Armour Pharmaceutical Co. Ltd., Eastbourne, England. Copyright All rights
@ 1973 by Academic Press, of reproduction in any form
274 Inc. reserved.
CARBOHYDRATE
DETERMINATION
275
Procedures 1. Calorimetric. The sample (10 ~1) is mixed with 5.0 ml color reagent (PAHBAH 0.5 M, sodium sulfite 0.05 M, calcium chloride 0.01 M, trisodium citrate 0.02 M and sodium hydroxide 0.3 M) and the mixture heated 5-10 min at lOO”, cooled, and the absorbance read at 420 nm. This procedure is suitable for glucose concentrations 05 mg/ml, such as occur in serum. ’ In a variant of this procedure, the sample (0.5 ml) is mixed with 1.0 ml of a more concentrated color reagent (PAHBAH 0.1 M, sodium sulfite 0.1 M, calcium chloride 0.02 M, trisodium citrate 0.05 M, sodium hydroxide 0.5 M), and the mixture heated and read as above. This procedure is suitable for glucose concentrations O-20 pg/ml. The color reagents used in these procedures are stable for over a week if air is excluded by a layer of paraffin. 2. Fluorometric. The sample (0.5 ml) is mixed with 0.2 ml PAHBAH reagent (0.25 M PAHBAH in 2 M sodium hydroxide) and the mixture heated 5 min at 100”. After cooling, 0.3 ml 1% polyvinylpyrrolidone (PVP) in 1 M sodium dihydrogen citrate is added and the mixture is extracted briefly (30-60 set) but vigorously with 2.5 ml 4-methylpentan2-one (iso-butyl methyl ketone). The PVP retains the PAHBAH in neutral solution, and assists extraction. When the phases have separated (by centrifugation if necessary) 2.0 ml organic phase is transferred to a Brown fluorometer cuvette (2) together with 0.6 ml fluorescence reagent, (0.2 M lanthanum chloride in diethanolamine, made by adding 0.5 M LaCl, in ethanol to warm diethanolamine slowly with mixing). All cuvettes are stoppered, inverted and shaken vigorously for 3-5 min on a vertical shaker (3). They are allowed to settle for 15 min or more and the fluorescence measured with excitation 470 nm and emission 545 nm. This procedure is suitable for glucose concentrations of 0.05-2 pug/ml. Variations of these procedures were tried with different cations instead of calcium or lanthanum, and with EDTA instead of citrate to hold calcium in alkaline solution. A range of carbohydrates and potential interfering substances were assayed, and the responses compared with glucose standards. Human serum samples were assayed calorimetrically and the results compared with those obtained by an o-toluidine procedure (4). RESULTS
Effects
of Cations
on Color Reaction
In the absence of calcium, has an absorption maximum
the colored product formed from glucose around 385390 nm (Fig. 1A). Calcium
276
M.
LEVER
0
WAVELENGTH
(nm 1
FIG. 1. Absorption spectra of PAHBAH-glucose derivative and of reagent blanks. Glucose ( 10 pl of 4 mg/ml) heated with 5 ml 0.05 it4 PAHBAH, 0.05 M sodium sulfite, 0.01 M EDTA and 0.3 M NaOH; (A) without added calcium and (B) with 10 mM calcium. Reagent blanks scanned against water, glucose derivative against reagent blank.
( 10 mM), however, shifts this to 413415 nm, with a shoulder about 445 nm (Fig. 1B). In the absence of a chelating agent to bind calcium and related cations, the spectrum described earlier ( 1) is obtained, where the shoulders at 415 nm and 440 nm are probably caused by traces of calcium in the reagents. Table 1 shows the effect of a number of cations on the colored derivative. Not all cations are soluble in alkaline solution at concentrations EFFECT
OF CATION
TABLE ON PRODUCT
1 OF COLORIMETRIC
(Increase
in absorbance)
ASSAY” X IOr
Cation
420 nm
470 nm
Mg2+ Cd+ SP+ Bar+ Zne+ Cd2+ Sn2+ Ala+ La3+ Bi3+
0 46 7 3 9 62 n 0 4 9
0 4 1 1 0 39 0 0 1 10
* Glucose (1~ ~1 of 5 mg/ml) heated 5 min at 100” with 5 ml of 0.05 M PAHBAH in 0.3 M NaOH and 0.1 M trisodium citrate. Cations (10 ~1 of 1 M solution) added after cooling. Spectra scanned against blanks (no glucose) and controls (no cation added) subtracted.
CARBOHYDRATE
277
DETERMINATION
of 10 mM, even with 0.1 M citrate added, so comparisons were made at 2 mM, which is not enough to produce the maximum shift. A comparison of the difference at 470 nm (where there is negligible absorbance without added cations) with that at 420 nm is used as a guide to the wavelength shift: it will be seen that cadmium and calcium produce the largest increases at 420 nm, and bismuth (III) and cadmium cause the most pronounced shifts. Magnesium and tin (II) give weak shifts at higher cation concentrations, but aluminum appears to have no effect. Some transition metal ions may interact, but these lead to unstable colored solutions. Calorimetric
Assay
Figure 2 shows that, in the calorimetric assay procedures described, absorbance is proportional to glucose concentration. The sensitivity can be readily adjusted to suit the levels to be measured by varying the dilution of sample with reagent. Serum glucose levels could be determined with samples of less than 1 ~1 (Fig. 2A). The color produced increases with longer heating times up to at least 15 min, but the change is small after 3 min. With a 5-min heating time, the absorbance of reagent blanks is less than 0.03 with the PAHBAH used in this work. Similar results can be obtained when EDTA is used instead of citrate to hold calcium in alkaline solution. A very small excess of EDTA significantly reduces the available calcium, whereas a 20-fold excess of citrate is needed to affect color yield. Marked differences between batches of disodium calcium EDTA were found, and for this reason citrate is the preferred additive. The colored products are slightly photolabile in aqueous solution. The tungsten lamp source of a spectrophotometer causes no loss in 30 min, lsl
(j/fok
I &/ml)
GLUCOSE
CONCENTRATION
FIG. 2. Response of calorimetric assay to different glucose concentrations sample ( 10 $1) with 5 ml 0.05 M PAHBAH color reagent; (B) sample (0.5 with 1.0 ml 0.1 M PAHBAH color reagent. Blanks subtracted.
(A) ml)
278
M.
LEVER
but sunlight causes substantial losses in 5 min. Calcium ( 10 mM), cadmium (2 mM) and zinc (10 mM) were used in modified methods ( 10 d of 5 mg/ml glucose with 5 ml reagent), and the cuvette was placed in the cell housing of the spectrophotofluorometer. The slit was removed and excitation monochromator set at 420 nm, and the loss in absorbance at 420 nm followed on exposure to the xenon lamp. In 30 min, the calcium sample decayed by 23%, the cadmium by 14%, and zinc by lO!%. Fluorescence
Assay
Figure 3B shows that the fluorometric assay is also linear up to 0.2 pg/ml, and linearity continues to at least 2 pg/ml, so that the range of reliability of this assay overlaps that of the calorimetric procedures. Figure 3A shows the excitation and emission spectra of the derivative in the fluorescence reagent. Attempts were made to use the strong fluorescence of the calcium chelate for analytical purposes, but this was found to be too unstable when exposed to the intense xenon lamp source. The main excitation maximum with calcium in diethanolamine is about 445 nm, with emission at 525 nm. Less fluorescent and photounstable chelates were also formed with cadmium, strontium and barium. No fluorescence was observed with tervalent or quadrivalent cerium.
LOO WAVELENGTH
500
600 (nm)
GLUCOSE
(ug/ml)
(A) excitation and emission spectra in FIG. 3. Fluorometric glucose assay: lanthanum-diethanolamine fluorescence reagent with excitation 470 nm and emission 545 nm; (B) response of assay to different glucose concentrations (0.5-ml sample). Reagent blank fluorescences subtracted in both cases.
Specificity Table 2 shows that both the calorimetric and fluorometric procedures detect reducing sugars that are capable of forming osazones. Glucose, fructose and mannose, which form the same osazone, give identical re-
CARBOHYDRATE
279
DETERMINATION
TABLE
2
RESPONSES OF CARBOHYDRATI~S IN ASSAY SYSTEMS Relative Carbohydrate Glucose Mannose Fructose Galactose Glucosamine Glucuronic acid Xylose Ribose 2-Deoxyribose Sucrose Maltose Lactose Sorbitol Ascorbic acid a Carbohydrate b Carbohydrate
solution solution
response
ColorimetrirY
= 100)
Fluorometricb 100 97 104 59 81 62 55 62 6 0 100 64
100 100 102 78 85 68 67 66 8 0 102 83
(10 ~1 of 10 mM) assayed (0.5 ml of 10 pi’@) assayed
(glucose
1
0
8
6 by calorimetric by fluorometric
procedure. procedure.
sponses. The relative responses of the other carbohydrates were found to vary slightly with reaction conditions, which possibly accounts for the differences of detail between the calorimetric and fluorometric procedures. Relative responses in the latter were affected also by efficiency of extraction into the fluorescence reagent, possibly because more than oncfluorogen was involved. The interferences from protein, calcium and bilirubin observed previously (1) were investigated with the modified calorimetric procedure. Calcium and magnesium ( 1 mg/ml) had no effect on the procedure proposed for serum. A colloidal solution of bilirubin (1 mg/ml) was prepared by adding 1% bilirubin in dimethylsulfoxide to phosphate buffer, pH 7.4, and a portion of this gave an apparent glucose of 0.40 mg/ml by the serum procedure, and this was additive when mixed with a glucose standard. In the same procedure, bovine serum albumin ( 10%w/v) gave an apparent glucose of 0.07 mg/ml, again additive when mixed with a glucose standard. The albumin solution responded strongly in the fluorometric procedure ( >lO pg/ml) and the fluorescence spectra closely resembled Fig. 3A, with slightly different relative intensities of shoulders and minor peaks in the excitation spectrum. Since the contaminating chromogen is not dialysable, it is probably high molecular weight, or protein-bound, carbohydrate. As observed before (l), ketoacids and non-carbohydrate reducing
280
M.
LEVER
agents do not interfere. Closely similar results, and a correlation coefficient better than 6.99, were obtained when 24 serum specimens were assayed by the calorimetric procedure and by an o-toluidine method (4). DISCUSSION
The reaction of glucose and benzoic acid hydrazide in dilute alkali was shown by Pinkus (5) to yield glyoxal bis (benzoylhydrazone) and methylglyoxal bis ( benzoylhydrazone ) by a still ill-understood mechanism (6). The colored products obtained in the present assaysare probably related hydrazones, which, like the similar products obtained from ,8-diketones and aromatic acid hydrazides, reversibly form yellow anions in alkali. In support of this hypothesis, the product obtained from dimethylglyoxal and excess p-hydroxybenzoic acid hydrazide in acid has very similar properties. Its yellow anion gives an intensely colored and fluorescent chelate with calcium, and the strong fluorescence in the lanthanum-diethanolamine reagent has spectra closely similar to Fig. 3A. This is a potential reagent for calcium. A mechanism based on the intermediate formation of aroylhydrazones, and possibly osazones, is consistent with the specificity of the analytical procedures (Table 2). Addition of calcium eliminates interference from calcium in the sample, and increases the sensitivity of the calorimetric method. Cadmium may be preferred for some purposes because of increased wavelength shift and improved stability compared with calcium, but cadmium is difficult to retain in alkaline solution, and if large numbers of glucose analyses are to be undertaken its toxicity and greater cost must be considered. Cadmium and calcium have similar ionic radii, and smalle_ror larger cations do not interfere at levels likely to be encountered in biological material. The calorimetric procedure provides a simple and inexpensive method for the determination of reducing sugars over a wide range of concentrations. Low reagent blanks are associated with high color yields, and the reagents are relatively non-toxic and non-corrosive compared to other procedures currently in use. Studies of the application of these methods in clinical chemistry are being undertaken. The fluorometric method has not been thoroughly evaluated since the calorimetric procedure provides sufficient sensitivity for the purposes of this laboratory. High blanks limit its sensitivity at present, but the spectra of these suggest that they are 66-76% carbohydrate contamination, which is ubiquitous in a clinical laboratory. As with similar chelates (7, 8) fluorescence yields are higher in nonaqueous hydrogen bonding solvents than in water. Ethanediol (but not formamide) may be used as the solvent, but unstable blanks
CARBOHYDRATE
are obtained, possibly react with PAHBAH.
281
DETERMINATION
because
oxidation
products
in the solvent
can
SUMMARY
The reaction of p-hydroxybenzoic acid hydrazide and reducing carbohydrates in alkaline solution can be used in calorimetric and fluorometric assay procedures. These are capable of determining glucose at concentrations from less then 0.1 pg/ml up to levels found in human serum. A calorimetric procedure is based on the formation of an intensely yellow calcium chelate of the derivative, and a fluorometric procedure on the fluorescence of a lanthanum chelate in diethanolamine solution. The methods are specific for reducing carbohydrates and are subject to few noncarbohydrate interferences. ACKNOWLEDGMENTS The author thanks Mr. J. C. Powell and Dr. C. W. Small for discussions. REFERENCES 1. LEVER, 2. BROWN,
M,, Anal.
Biochem. J. B., MACNAUGHTAN,
( 1972). C., SMITH, M. A., AND SMYTH,
47, 273
40, 175 (1968). 3. POWELL, J. C., N.Z.J. Med. Lab. Tech. 26, 69 (1972). 4. FETEFUS, W. A., Amer. J. Med. Tech. 31, 17 (1965). 5. PINKUS, G., Chem. Ber. 31, 31 (1898). 6. RUSSEL, C. S., AND LYONS, R., Carbohyd. Res. 9, 347 7. LEVER, M., Biochem. Med. 6, 65 (1972). 8. LEVER, M., Biochem. Med. 6, 216 (1972).
(1969).
B., J. Endocrinol.
MEDICINE
7,
274-281
Calorimetric
(1973)
and
Fluorometric
Determination
with Acid
Carbohydrate
p-Hydroxybenzoic
Hydrazide M. LEVER
Pathology
Department,
Greenlane Received
Hospital, Auckland April
13,
3, New Zedund
1972
Aromatic acid hydrazides and reducing carbohydrates react in alkaline solution to give products suitable for the calorimetric and fluorometric assay of reducing sugars ( 1). Optimum results are obtained with p-hydroxybenzoic acid hydrazide (PAHBAH). A sensitive procedure has been described (1) which was found to be subject to interference from calcium and from high concentrations of bovine serum albumin. Described here are improved analytical methods that can be used over a wide range of glucose concentrations, from the calorimetric determination of serum glucose on less than a 1-J sample to the fluorometric determination of parts per billion glucose on a 0.5-ml sample. METHODS
instrumentation
Absorbance was read with a Gilford 300 N microsample spectrophotometer. Absorption spectra were scanned with a Unicam SP SOOA doublebeam spectrophotometer. Fluorescence measurements were made with an Aminco-Bowman spectrophotofluorometer fitted with a xenon lamp and an R 136 photomultiplier; mirrors and l-mm slits were placed in the cell housing. When recording spectra, O.l-mm slits were placed on the scanning monochromator side of the cell housing. Reagents Most reagents were obtained from BDH Chemicals Ltd., Poole, England; p-hydroxybenzoic acid hydrazide and diethanolamine were obtained from Fluka AG, Buchs, Switzerland. The bovine serum albumin was from Armour Pharmaceutical Co. Ltd., Eastbourne, England. Copyright All rights
@ 1973 by Academic Press, of reproduction in any form
274 Inc. reserved.
CARBOHYDRATE
DETERMINATION
275
Procedures 1. Calorimetric. The sample (10 ~1) is mixed with 5.0 ml color reagent (PAHBAH 0.5 M, sodium sulfite 0.05 M, calcium chloride 0.01 M, trisodium citrate 0.02 M and sodium hydroxide 0.3 M) and the mixture heated 5-10 min at lOO”, cooled, and the absorbance read at 420 nm. This procedure is suitable for glucose concentrations 05 mg/ml, such as occur in serum. ’ In a variant of this procedure, the sample (0.5 ml) is mixed with 1.0 ml of a more concentrated color reagent (PAHBAH 0.1 M, sodium sulfite 0.1 M, calcium chloride 0.02 M, trisodium citrate 0.05 M, sodium hydroxide 0.5 M), and the mixture heated and read as above. This procedure is suitable for glucose concentrations O-20 pg/ml. The color reagents used in these procedures are stable for over a week if air is excluded by a layer of paraffin. 2. Fluorometric. The sample (0.5 ml) is mixed with 0.2 ml PAHBAH reagent (0.25 M PAHBAH in 2 M sodium hydroxide) and the mixture heated 5 min at 100”. After cooling, 0.3 ml 1% polyvinylpyrrolidone (PVP) in 1 M sodium dihydrogen citrate is added and the mixture is extracted briefly (30-60 set) but vigorously with 2.5 ml 4-methylpentan2-one (iso-butyl methyl ketone). The PVP retains the PAHBAH in neutral solution, and assists extraction. When the phases have separated (by centrifugation if necessary) 2.0 ml organic phase is transferred to a Brown fluorometer cuvette (2) together with 0.6 ml fluorescence reagent, (0.2 M lanthanum chloride in diethanolamine, made by adding 0.5 M LaCl, in ethanol to warm diethanolamine slowly with mixing). All cuvettes are stoppered, inverted and shaken vigorously for 3-5 min on a vertical shaker (3). They are allowed to settle for 15 min or more and the fluorescence measured with excitation 470 nm and emission 545 nm. This procedure is suitable for glucose concentrations of 0.05-2 pug/ml. Variations of these procedures were tried with different cations instead of calcium or lanthanum, and with EDTA instead of citrate to hold calcium in alkaline solution. A range of carbohydrates and potential interfering substances were assayed, and the responses compared with glucose standards. Human serum samples were assayed calorimetrically and the results compared with those obtained by an o-toluidine procedure (4). RESULTS
Effects
of Cations
on Color Reaction
In the absence of calcium, has an absorption maximum
the colored product formed from glucose around 385390 nm (Fig. 1A). Calcium
276
M.
LEVER
0
WAVELENGTH
(nm 1
FIG. 1. Absorption spectra of PAHBAH-glucose derivative and of reagent blanks. Glucose ( 10 pl of 4 mg/ml) heated with 5 ml 0.05 it4 PAHBAH, 0.05 M sodium sulfite, 0.01 M EDTA and 0.3 M NaOH; (A) without added calcium and (B) with 10 mM calcium. Reagent blanks scanned against water, glucose derivative against reagent blank.
( 10 mM), however, shifts this to 413415 nm, with a shoulder about 445 nm (Fig. 1B). In the absence of a chelating agent to bind calcium and related cations, the spectrum described earlier ( 1) is obtained, where the shoulders at 415 nm and 440 nm are probably caused by traces of calcium in the reagents. Table 1 shows the effect of a number of cations on the colored derivative. Not all cations are soluble in alkaline solution at concentrations EFFECT
OF CATION
TABLE ON PRODUCT
1 OF COLORIMETRIC
(Increase
in absorbance)
ASSAY” X IOr
Cation
420 nm
470 nm
Mg2+ Cd+ SP+ Bar+ Zne+ Cd2+ Sn2+ Ala+ La3+ Bi3+
0 46 7 3 9 62 n 0 4 9
0 4 1 1 0 39 0 0 1 10
* Glucose (1~ ~1 of 5 mg/ml) heated 5 min at 100” with 5 ml of 0.05 M PAHBAH in 0.3 M NaOH and 0.1 M trisodium citrate. Cations (10 ~1 of 1 M solution) added after cooling. Spectra scanned against blanks (no glucose) and controls (no cation added) subtracted.
CARBOHYDRATE
277
DETERMINATION
of 10 mM, even with 0.1 M citrate added, so comparisons were made at 2 mM, which is not enough to produce the maximum shift. A comparison of the difference at 470 nm (where there is negligible absorbance without added cations) with that at 420 nm is used as a guide to the wavelength shift: it will be seen that cadmium and calcium produce the largest increases at 420 nm, and bismuth (III) and cadmium cause the most pronounced shifts. Magnesium and tin (II) give weak shifts at higher cation concentrations, but aluminum appears to have no effect. Some transition metal ions may interact, but these lead to unstable colored solutions. Calorimetric
Assay
Figure 2 shows that, in the calorimetric assay procedures described, absorbance is proportional to glucose concentration. The sensitivity can be readily adjusted to suit the levels to be measured by varying the dilution of sample with reagent. Serum glucose levels could be determined with samples of less than 1 ~1 (Fig. 2A). The color produced increases with longer heating times up to at least 15 min, but the change is small after 3 min. With a 5-min heating time, the absorbance of reagent blanks is less than 0.03 with the PAHBAH used in this work. Similar results can be obtained when EDTA is used instead of citrate to hold calcium in alkaline solution. A very small excess of EDTA significantly reduces the available calcium, whereas a 20-fold excess of citrate is needed to affect color yield. Marked differences between batches of disodium calcium EDTA were found, and for this reason citrate is the preferred additive. The colored products are slightly photolabile in aqueous solution. The tungsten lamp source of a spectrophotometer causes no loss in 30 min, lsl
(j/fok
I &/ml)
GLUCOSE
CONCENTRATION
FIG. 2. Response of calorimetric assay to different glucose concentrations sample ( 10 $1) with 5 ml 0.05 M PAHBAH color reagent; (B) sample (0.5 with 1.0 ml 0.1 M PAHBAH color reagent. Blanks subtracted.
(A) ml)
278
M.
LEVER
but sunlight causes substantial losses in 5 min. Calcium ( 10 mM), cadmium (2 mM) and zinc (10 mM) were used in modified methods ( 10 d of 5 mg/ml glucose with 5 ml reagent), and the cuvette was placed in the cell housing of the spectrophotofluorometer. The slit was removed and excitation monochromator set at 420 nm, and the loss in absorbance at 420 nm followed on exposure to the xenon lamp. In 30 min, the calcium sample decayed by 23%, the cadmium by 14%, and zinc by lO!%. Fluorescence
Assay
Figure 3B shows that the fluorometric assay is also linear up to 0.2 pg/ml, and linearity continues to at least 2 pg/ml, so that the range of reliability of this assay overlaps that of the calorimetric procedures. Figure 3A shows the excitation and emission spectra of the derivative in the fluorescence reagent. Attempts were made to use the strong fluorescence of the calcium chelate for analytical purposes, but this was found to be too unstable when exposed to the intense xenon lamp source. The main excitation maximum with calcium in diethanolamine is about 445 nm, with emission at 525 nm. Less fluorescent and photounstable chelates were also formed with cadmium, strontium and barium. No fluorescence was observed with tervalent or quadrivalent cerium.
LOO WAVELENGTH
500
600 (nm)
GLUCOSE
(ug/ml)
(A) excitation and emission spectra in FIG. 3. Fluorometric glucose assay: lanthanum-diethanolamine fluorescence reagent with excitation 470 nm and emission 545 nm; (B) response of assay to different glucose concentrations (0.5-ml sample). Reagent blank fluorescences subtracted in both cases.
Specificity Table 2 shows that both the calorimetric and fluorometric procedures detect reducing sugars that are capable of forming osazones. Glucose, fructose and mannose, which form the same osazone, give identical re-
CARBOHYDRATE
279
DETERMINATION
TABLE
2
RESPONSES OF CARBOHYDRATI~S IN ASSAY SYSTEMS Relative Carbohydrate Glucose Mannose Fructose Galactose Glucosamine Glucuronic acid Xylose Ribose 2-Deoxyribose Sucrose Maltose Lactose Sorbitol Ascorbic acid a Carbohydrate b Carbohydrate
solution solution
response
ColorimetrirY
= 100)
Fluorometricb 100 97 104 59 81 62 55 62 6 0 100 64
100 100 102 78 85 68 67 66 8 0 102 83
(10 ~1 of 10 mM) assayed (0.5 ml of 10 pi’@) assayed
(glucose
1
0
8
6 by calorimetric by fluorometric
procedure. procedure.
sponses. The relative responses of the other carbohydrates were found to vary slightly with reaction conditions, which possibly accounts for the differences of detail between the calorimetric and fluorometric procedures. Relative responses in the latter were affected also by efficiency of extraction into the fluorescence reagent, possibly because more than oncfluorogen was involved. The interferences from protein, calcium and bilirubin observed previously (1) were investigated with the modified calorimetric procedure. Calcium and magnesium ( 1 mg/ml) had no effect on the procedure proposed for serum. A colloidal solution of bilirubin (1 mg/ml) was prepared by adding 1% bilirubin in dimethylsulfoxide to phosphate buffer, pH 7.4, and a portion of this gave an apparent glucose of 0.40 mg/ml by the serum procedure, and this was additive when mixed with a glucose standard. In the same procedure, bovine serum albumin ( 10%w/v) gave an apparent glucose of 0.07 mg/ml, again additive when mixed with a glucose standard. The albumin solution responded strongly in the fluorometric procedure ( >lO pg/ml) and the fluorescence spectra closely resembled Fig. 3A, with slightly different relative intensities of shoulders and minor peaks in the excitation spectrum. Since the contaminating chromogen is not dialysable, it is probably high molecular weight, or protein-bound, carbohydrate. As observed before (l), ketoacids and non-carbohydrate reducing
280
M.
LEVER
agents do not interfere. Closely similar results, and a correlation coefficient better than 6.99, were obtained when 24 serum specimens were assayed by the calorimetric procedure and by an o-toluidine method (4). DISCUSSION
The reaction of glucose and benzoic acid hydrazide in dilute alkali was shown by Pinkus (5) to yield glyoxal bis (benzoylhydrazone) and methylglyoxal bis ( benzoylhydrazone ) by a still ill-understood mechanism (6). The colored products obtained in the present assaysare probably related hydrazones, which, like the similar products obtained from ,8-diketones and aromatic acid hydrazides, reversibly form yellow anions in alkali. In support of this hypothesis, the product obtained from dimethylglyoxal and excess p-hydroxybenzoic acid hydrazide in acid has very similar properties. Its yellow anion gives an intensely colored and fluorescent chelate with calcium, and the strong fluorescence in the lanthanum-diethanolamine reagent has spectra closely similar to Fig. 3A. This is a potential reagent for calcium. A mechanism based on the intermediate formation of aroylhydrazones, and possibly osazones, is consistent with the specificity of the analytical procedures (Table 2). Addition of calcium eliminates interference from calcium in the sample, and increases the sensitivity of the calorimetric method. Cadmium may be preferred for some purposes because of increased wavelength shift and improved stability compared with calcium, but cadmium is difficult to retain in alkaline solution, and if large numbers of glucose analyses are to be undertaken its toxicity and greater cost must be considered. Cadmium and calcium have similar ionic radii, and smalle_ror larger cations do not interfere at levels likely to be encountered in biological material. The calorimetric procedure provides a simple and inexpensive method for the determination of reducing sugars over a wide range of concentrations. Low reagent blanks are associated with high color yields, and the reagents are relatively non-toxic and non-corrosive compared to other procedures currently in use. Studies of the application of these methods in clinical chemistry are being undertaken. The fluorometric method has not been thoroughly evaluated since the calorimetric procedure provides sufficient sensitivity for the purposes of this laboratory. High blanks limit its sensitivity at present, but the spectra of these suggest that they are 66-76% carbohydrate contamination, which is ubiquitous in a clinical laboratory. As with similar chelates (7, 8) fluorescence yields are higher in nonaqueous hydrogen bonding solvents than in water. Ethanediol (but not formamide) may be used as the solvent, but unstable blanks
CARBOHYDRATE
are obtained, possibly react with PAHBAH.
281
DETERMINATION
because
oxidation
products
in the solvent
can
SUMMARY
The reaction of p-hydroxybenzoic acid hydrazide and reducing carbohydrates in alkaline solution can be used in calorimetric and fluorometric assay procedures. These are capable of determining glucose at concentrations from less then 0.1 pg/ml up to levels found in human serum. A calorimetric procedure is based on the formation of an intensely yellow calcium chelate of the derivative, and a fluorometric procedure on the fluorescence of a lanthanum chelate in diethanolamine solution. The methods are specific for reducing carbohydrates and are subject to few noncarbohydrate interferences. ACKNOWLEDGMENTS The author thanks Mr. J. C. Powell and Dr. C. W. Small for discussions. REFERENCES 1. LEVER, 2. BROWN,
M,, Anal.
Biochem. J. B., MACNAUGHTAN,
( 1972). C., SMITH, M. A., AND SMYTH,
47, 273
40, 175 (1968). 3. POWELL, J. C., N.Z.J. Med. Lab. Tech. 26, 69 (1972). 4. FETEFUS, W. A., Amer. J. Med. Tech. 31, 17 (1965). 5. PINKUS, G., Chem. Ber. 31, 31 (1898). 6. RUSSEL, C. S., AND LYONS, R., Carbohyd. Res. 9, 347 7. LEVER, M., Biochem. Med. 6, 65 (1972). 8. LEVER, M., Biochem. Med. 6, 216 (1972).
(1969).
B., J. Endocrinol.