Anal. Chem. 1993, 65, 321-326
521
Ascorbate and Dehydroascorbate Measurements in Aqueous Solutions and Plasma Determined by Gas Chromatography-Mass Spectrometry John C. Deutsch'st and J. Fred Kolhouset Divisions of Gastroenterology and Hematology, Department of Medicine, Department of Veterans Affairs Hospital and the University of Colorado Health Sciences Center, 4200 East Ninth Avenue, Denver, Colorado 80262
The terMutyidlmethylrllyi derivatives of ascorblc acld and dehydroascorbic acid were characterized by gas chromatography-mass rpectrometry, and an kotope dllutlon assay for ascorbate and dehydroascorbate was developed using [18C6)worblcackl and [I%,),and [8,&2H2]dehydroasco~ate. Thb assay was used to monltor ascorbic acld loss and the resulting r k e of dehydroascorblc acid in aqueous solutions and piaama. AIcorblc acld was shown to rapidly decompose In aqueous solutions containing transition metal Ions or when exposed to oxygen. Ethylenedlamlnaetraacetkadd chdatkn did not prevent ascorbic acld degradationIn aqueous solution, and ascorbate In ethylenediaminetetraacetic acld chelated piama was converted to dehydroascorbate on freezing. Gas chromatography-mars spectrometry appears to be a satisfactory method for determining the ascorbate and dehydroascorbate content of solutions includlng human blood plama, whether or not there is ongoing oxidation of ascorbate In those soiutlons.
INTRODUCTION Ascorbic acid (vitamin C) is an hexuronic acid with a pll of 4.17. In biologicalsolutions at neural pH, it exists primarily as the ascorbate anion. Ascorbic acidlascorbate is a vital component in the diet of humans. Ascorbate is the agent which prevents scurvy and is known to take part in several biological reactions including the formation of collagen, the formation of neurotransmitters,the degradationof tyrosine,',2 and use as a leukocyte stimulator.3 Ascorbate is well-known as an antioxidant in vitro, being used, for example, as a reducing agent in myeloperoxidase reactions3 and to prevent the oxidation of reduced folates to oxidized forms in red blood
* To whom reprint requests should be sent. +
Division of Gastroenterology.
t Division of Hematology.
(1)Jaffe, G. In Handbook of Vitamins;Machlin, L., Ed.; Dekker, Inc.: New York, 1984; p 199. (2) Levine, M. N . Engl. J. Med. 1986, 314, 892. (3) Marquez, L. A.; Dunford, H. B.; Van Wart, H. J. Biol. Chem. 1990, 265,5666. 0003-2700/93/0365-0321$04.00/0
cell folate assay^.^ Of interest, ascorbate is possibly the primary antioxidant in human blood plasma.5 Evidence suggeststhat other physiologicalantioxidants, such as vitamin E, are maintained in a reduced state at the expense of ascorbate.5.6 Ascorbate has also been shown to be more effective than glutathione at reducing the acetaminophen free radical.' The role of ascorbate as an antioxidant in vivo has important health implications since endogenously generated oxidantsare thought to be important in carcinogenesis? giving ascorbate a potential physiologic role in cancer prevention9 Assays for ascorbic acidlascorbate are important to better determine biochemical and physiological roles for ascorbate in health and disease. Unfortunately, ascorbate is unstable in solution with measurable degradation occurring within minutes to hours,lO presumably due to molecular oxygen and traces of contaminating metals.11 This problem may be corrected for using isotope dilution assay techniques. In addition, ascorbate is easily oxidized through a free radical intermediate (semi-dehydroaacorbate)to form dehydroascorbate, providing electrons to be used in various reactions. Transition metals, particularly Fe(II1) and Cu(II), are welldescribed catalysts for oxidizing ascorbate, producing hydrogen peroxide and hydroxyl radicals from molecular oxygen in the process.12J3 Understanding the role of ascorbate in vivo may well depend not only on the measurement of ascorbate but also on the relative quantity of dehydroascor(4) Rothenberg, S. P.; DaCosta, M.; Rosenberg, Z. N. Engl. J. Med. 1972,286, 1335. (5) Frei, B.; England, L.; Ames, B. N. R o c . Natl. Acad. Sci. U.S.A. 1989,86,6377. (6) Maguire, J. J.; Wilson, D. S.; Packer, L. J. Biol. Chem. 1989,264, 21462. (7) Ramakrishna-Rao,D. N.; Fischer, V.; Mason, R. P. J. Biol. Chem. 1990, 265, 844. (8) Weitzman, S. A.; Gordon, L. I. Blood 1990, 76, 655. (9) Stahelin, H. B.; Gey, K. F.; Brubacher, G. In Third Conference on Vitamin C; Burns, J., Rivera, J., Machlin, L., E&.; N.Y. Acad. Sci: New York, 1987; p 124. (10) Washko, P.; Rotrosen, D.; Levine, M. J. Biol. Chem. 1989, 264, 18996. (11) Buettner, G. R. J.Biochem. Biophys. Methods 1988,16,27-40. (12) Taqui, M. M.; Martell, A. E. J. Am. Chem. SOC.1967,89, 4176. (13) Miller, D. M.; Buettner, G. R.; Auat, S. D. Free Radic. Biol. Med. 1990, 8, 95.
0 1993 American Chemical Society
322
ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993
bate that is present in the system being tested. This report describes a gas chromatography-mass spectrometry (GC-MS) method of quantitating and determining the relative changes in ascorbate and dehydroascorbata in solution using [13C&scorbic acid, [13C,Jdehydroascorbate, and [6,6-*H21dehydroascorbateinternal standards. This assay is rapid, simple, and sensitive. It has been used to study the degradation of ascorbate in aqueous solution and plasma under a variety of conditions.
METHODS [13C6]Ascorbicacid (96.1% [13C6])and [6,6-2H~]ascorbic acid (98.5% [~Hz])were purchased from MSD Isotopes, Montreal, Quebec,Canada. Dehydroascorbicacid was purchased from ICN Biochemicals, Cosa Mesa, CA. Trichloroacetic acid (TCA), L-ascorbate oxidase, and other chemical reagents of highest grade availablewere purchased from Sigma Chemicals, St. Louis, MO. N-Methyl-N-(tert-butyldimethylsily1)trifluoroacetamide (TBDMS) was purchased from Pierce, Rockford, IL. Diethylenetriaminepentaacetic acid (DTPA), 41% aqueous solution, was purchased from Kodak Chemicals, Rochester, NY. Gas chromatography was carried out on a Hewlett-Packard (Sunnyvale,CA) 5890Agas chromatographusing helium as carrier througha 10-mby0.25-mm (internaldiameter) Supelco (Belfont, PA) SPB-1 fused silica capillary column. The column head pressure was 50 kPa. The injector port temperature was 250 OC, and the initial column temperature was 80 "C. A temperature gradient of 30 OC/min was applied to the column until a final temperature of 300 OC was reached. Ionization was by electron impact. Mass spectrometry was carried out using a HewlettPackard 5971A mass detector. The electron multiplier was set at 1800-2000 V for high-quantity standards and at 2600-2800 V for plasma and serum samples. Spectra of the various standards were obtained in the scan mode, and quantitation was carried out by selected ion monitoring (SIM). Dehydroascorbate, [6,6-2H~ldehydroascorbate, and [l3C6ldehydroascorbate were made from 1mg/mL solutionsof ascorbic acid, [6,6-2H~lascorbic acid, and [13C6]ascorbicacid in 100 mM potaseium phosphate, 4 mM ethylenediaminetetraacetic acid (EDTA),pH 5.6. Ascorbate oxidase (AO)was added to a dialysis bag (with a molecular weight cut-off of 10 000) at a concentration of le100 units/mL and placed into the ascorbic acid solutions for a final enzyme concentration of 3 units of ascorbate oxidase/ mg of ascorbic acid. The reaction was carried out at 31 OC for 18 h, at which time the majority (95%)of the ascorbic acid was converted to dehydroascorbate or other products as judged by GC-MS. Derivatization was carried out by taking the sample (consisting of 1-10 pg of various ascorbic acid derived substances), drying under vacuum at room temperature (Savant Speed Vac Concentrator, Farmingdale, NY), adding 10 pL of TBDMS and 100 pL of acetonitrile, and incubating the mixture at 60 "C for 1-2 h and at room temperature for up to an additional 24 h. Borosilicate tubes were prepared for ascorbate assays by vacuum drying 100 pL of a 50 pg/mL solution of [13C~]ascorbic acid in water containing 0.01% (v/v) 2-mercaptoethanol. Stock solutions of [13C6]ascorbicacid (1 mg/mL) in water appeared stable over several months when kept at -70 OC. Quantitation of unknown ascorbate content was determined based on the ratio of [M - 571+ m/z 575 (ascorbic acid) to [M 57]+ m / z 581 ([13C6]ascorbicacid) using SIM. The quantity of [l3Ce]ascorbicacid was verified in triplicate for each experiment relative to a freshly prepared aqueous solution of known concentration of ascorbic acid. The variation of [13C&scorbic acid content was less than 5 % when determined in five consecutive samples. Dehydroascorbate assays were carried out by adding 1-2 pg of [13C6]dehydroascorbatein 100 mM potassium phosphate, 4 mM EDTA, pH 5.6, buffer directly to the sample being tested. Quantitation of the unknown dehydroascorbate content was determined based on the ratio of [M - 57]+ m/z 345 (dehydroascorbate) to [M - 57]+ m/z 351 ([13C6]dehydroascorbate). The [13C6]dehydroascorbatecontent was verified in triplicate
relative to a freshly prepared aqueous solution of known concentration (based on mass) of commercially obtained dehydroascorbic acid. Permission to obtain blood for the measurement of ascorbate was granted by the human subjects committee at the University of ColoradoHealth ScienceaCenter. Human blood was collected in 10-mL syringes and added to vacuum phlebotomy tubes (Becton-Dickinson,Rutherford, NJ) containing either KaEDTA, heparin, or no anticoagulant prior to the preparation of either plasma or serum. One hundred microliters of plasma or serum was removed in triplicate, being careful not to disturb the interface between the supernatant fluid and the red cell pellet. For ascorbate assays, the plasma and serum were placed in borosilicate t u b containing dried [13C6]ascorbicacid. [13C6]Dehydroascorbateused as internal standard was stored at -70 OC at a concentration of 125250 pg/mL in 100 mM potassium phosphate, 4 mM EDTA, pH 5.6. For dehydroascorbate assays, 10 pL of the PC~ldehydroascorbate solution was added to the sample. Ten microliters of a 100 g/100 mL solution of TCA was then added to 100 pL of plasma or serum while vortexing,followed by 890 pL of Hz0. The solutions were centrifuged at 3000g for 10min, and 500 pL of the supernatant was removed and washed three times with 1.5 mL of hydrated diethyl ether. The aqueous phase was dried by vacuum centrifugation, and 10 pL of TBDMS and 100 pL of acetonitrile were added followed by incubation at 60 "C for 2 h. The solutions were then centrifuged at 3oooOg for 10 min, and 40 p L of supernatant fluid was removed for analysis. Plasma and serum samples were also stored at -70 and -20 OC for 1-20 days, either with or without [1q6]ascorbic acid. Cell-free plasma samples were made by passing the plasma through a 0.22-pm filter. One hundred micromolar solutions of ascorbic acid and [l3C6]ascorbic acid were oxidized by adding either 40 nM CuSOd or 1% (v/v) HzOz or by bubbling oxygen through the solutions 20 cm from a 50-Wlamp. Ten-microliter aliquots of these solutions were removed at timed intervals, dried, derivatized,and analyzed by GC-MS. Ascorbate degradation in the presence of chelating agents was measured after adding 4 mL of a 100 pM solution of ascorbate to tubes containing either 18pmol of KsEDTA, 3 pmol of DTPA, or no chelator and which, additionally,contained either 1.2 pmol of FeC13, CuClz, or an equal volume of water. These samples were immediately frozen for 48 h at -20 OC prior to examination by GC-MS. Metal salts were added to plasma by adding equal volumes of EDTA chelated plasma (having a final concentration of approximately 7 mM KaEDTA)to either 5 or 10mM solutions of FeClz, FeCl3, CuC12, MgClZ, CaClZ, or NaCl and then freezing the solutions at -20 OC for 96 h. On thawing, the protein was precipitated from 200 pL of sample solutions with 20 pL of 100 g/100 mL of TCA; the solutions were brought to a final volume of 1mL with HzO and processed as described above. A calibration curve for ascorbic acid was constructed using fresh heparinized plasma and [13C~]ascorbic acid as the internal standard. Onemilliliteraliquotsof plasma containing [l~~]ascorbic acid diluted with either 50 pL of HzO (final volume 1.05 mL), 50 pL of 40.0 pg/mL ascorbate in HzO, or 50 pL of 80.0 pg/mL ascorbate in H20 (net increase of 1.90 and 3.80 pg/mL) were processed as described above. A separate calibration curve was also constructed by using [13C6]dehydroaecorbateas the internalstandard and adding increasing amounts of unlabeled dehydroascorbate (made from ascorbicacid using ascorbate oxidase). The absolute quantity of unlabeled dehydroascorbate added to plasma was determined relative to a known quantity of freshly prepared dehydroascorbic acid from ICN Biochemicals. Samples were processed in the standard fashion. All samples were run in triplicate or quadruplicate, and the mean and standard deviation were calculated. Significancedefiied at p I0.05 was determined using a two-tailed t-test.
RESULTS AND DISCUSSION The Characterization of Ascorbic Acid and Dehydroascorbic Acid by GC-MS. Based on the structure of ascorbic acid (Figure lA), the TBDMS derivative was computed to have a mass of 632 Da and [13C~]ascorbicacid
ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993
A
B
Table I. Proposed Structure of Ions Shown in Figure 3 Generated from TBDMS-Derivetized Ascorbic Acid.
1-10- CH:!
HO-CH:! I HC-OH
Ascorbic Acid/ [13C~]hcorbic Acid M+ 632/638 m/z 575/581 loss of tert-butyl group from one of four TBDMS-
I
0
\ /
0
HC-OH
\ /
\HO
HC
HC
C
\
I
/c=c
HO
\
\HO C
rnlz 5311536
I
o//c -cNo
\
OH
mlz 443/449
MW 176
MW 174
Figure 1. The structure of (A) ascorbic acid and (B) dehydroascorblc acid. Each TBDMS derkatlzation adds a net mass of 114 Da.
rnlz 4151420
(m/z 150-650)
rnlz 3431347
TIC
323
derivatized hydroxyl groups (generating the ion); no loss from parent molecule loss of tert-butyl group from one of four TBDMSderivatized hydroxyl groups and decarboxylation at CI loss of tert-butyl group from one of four TBDMSderivatized hydroxyl groups and loss of entire TBDMS-OH; no loss of carbon loss of tert-butyl group from one of four TBDMSderivatized hydroxyl groups, loss of entire TBDMS-OH, and lose of CO at C1 loss of entire derivatized CSand CSportion of the molecule
M+,the totalmolecular maas of the derivatizedproduct in daltons. The mass of the l3C6-labeledcompound appears on the right side of the slash mark. m / z , mass to charge.
TIC (m/z
150-630)
Time (min)
4.0
7.0
I "
TIME (mfn)
7.0
TIC (m/z 1563-650)
Figurr 2. The TIC (mlz 150-650) of TBDMS derkatked (A) ascorbic acid. acid and (B) [1%8]as~orbi~
.h
s7s
3
w
E
2m C
4
58 I I
9
$1
34;
E < p '
\
'y
i r z 351
4.0
Y
(14)Stabler, S. P.; Marcell, P. D.; Podell, E. R.; Allen, R. H. Anal. Biochern. 1987,162. (15)Marcell, P. D.; Stabler, S. P.; Podell, E. R.; Allen, R. H. Anal. Biochem. 1985, 150.
T i m e (min)
6.0
Flgwe 4. (A) The TIC (mlz 150-650) of commercially prepared dehydroascorbicacid. X marks the peak where rnlz 345 occurs. The asterisk (*) marks a peak having a spectra identical to ascorbk add. (B) The TIC (mlz 150-650) of the product of ascorbate oxidase and [1%6]ascorbic acid. (C) The chromatogram of mlz 351 (['%e]dehydroascorbic acid) taken from the TIC shown in B.
Based on the structure of dehydroascorbic acid (Figure lB),the predicted derivatizedmasses of dehydroascorbicacid, [2H2ldehydroascorbic acid, and [13C61dehydroascorbicacid were 402,404, and 408 Da with [M - 571+ ions of rnlz 345, 347, and 351. The TIC of commercially prepared derivatized dehydroascorbic acid was examined and found to contain several peaks (Figure 4A). A major peak contained m/z 345 at 4.8 min. Dehydroascorbate, [W61dehydroascorbate,and [6,6-2H2]dehydroascorbatewere then prepared from their respective ascorbic acids and ascorbate oxidase in 100 mM potassium phosphate, 4 mM EDTA, pH 5.6. An aliquot from each of these reaction mixes was derivatized with TBDMS and scanned by GC-MS. When these derivatized reaction mixes were scanned, the TIC revealed several peaks (Figure 4B). However, the respective predicted [M - 57]+ ions occurred in only one of these peaks in each TIC (Figure 4C). Full spectra of derivatized dehydroascorbate and [lsC~]dehydroascorbate are shown in Figure 5A,B,respectively,and the proposed structures of the ions are described in Table 11.
924
ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993
x
35 1
I 2a0
250 M a s s/C hara a
3 00
iSP
i
I I
i
I
V .
0
Figurr 5. The spectra obtained at 4.8 min for (C) dehydroascorbate and (D) [13C8]dehydroascorbate.
Table 11. Proposed Structure of Ions Shown in Figure 5 Generated from TBDMS-Derivatized Dehydroascorbic Acids Dehydroascorbic Acid/ [13C~lDehydroascorbic Acid M+ 402/408 m/z 3871393 loss of a methyl group from one of two TBDMSderivatized hydroxyl groups (generating the ion) m/z 345/351 loss of tert-butyl group from one of two TBDMSderivatized hydroxyl groups (generating the ion); no loss from parent molecule m/z 3011306 loss of tert-butyl group from one of two TBDMSderivatized hydroxyl groups and decarboxylation at C1 a M+,the total molecular mass of the derivatized product in daltons. The mass of the Ws-labeled compound appears on the right side of the slash mark. miz, mass to charge.
To derive a calibration curve for ascorbic acid quantitation in aqueous solution, the ratio of ion abundance5 at rnlz 575 and 581 was plotted against the quantity of added ascorbic acid. The ratio of rnlz 575-581 at 6.7 min was linear in the tested range with a correlation (r) between the increase in added ascorbic acid to the increase in the ratio of rnlz 575:581 of greater than 0.997 (Figure 6, top). Similar experiments were performed with commerciallyprepared dehydroascorbic acid and [6,6-2H2]dehydroascorbate.As shown in Figure 6, bottom, a linear relationship existed between the addition of dehydroascorbic acid to fixed amounts of [6,6-2H21dehydroascorbate and ratio of rnlz 345:347. The correlation ( r )was greater than 0.999. This linear relationship of the calibration curve did not hold when large quantities (approximately 500 ngor more) (peakwidths greaterthan0.15 min) of derivatized ascorbic acids were applied to the column, giving an overload pattern (Figure 7). Degradation of Ascorbic Acid. One hundred micromolar solutions of ascorbic acid and [13C61ascorbic acid were subjected to direct light (50-W lamp at 20 cm) or dark (aluminum foil wrap) with either argon or oxygen bubbling through the solutions at 2 Llmin. The solutions were kept in a common metal block at 22 OC. Aliquota were taken at the start and at intervals through 4 h. These aliquota were examined by GC-MS with SIM at rnlz 345,351 (group 1)and mlz 443,449,575, and 581 (group 2). Essentially no rnlz 345 was present in the starting material. The ratio of rnlz 345: 575 (dehydroascorbicacid to ascorbic acid) increased 25 OOOfold in the solution exposed to light and oxygen compared to the starting material. A 6OOO-fold increase in the ratio of rnlz 345575 occurred in an ascorbic acid solution exposed to oxygen without light. Smaller increases in the ratio of rnlz 345:575 (about 100-fold compared to starting material), occurred in ascorbic acid containing solutions kept in argon whether or not exposed to light. The reversibility of ascorbic acid oxidation was then examined. Previous reports suggest that 10 mM 2,3-dimer-
i
s
i
ug of Addd AmorbIc Add
2.0
+
0
/I
2
4
ug of Added D&ydroa.corblc Acid
Flgure 6. (Top)A callbratlon curve In water for the ion abundance of mlz 575 to 581 (varying quantltles of ascorbic acld to 2 pg of ['%e]ascorblc acM] plotted against the known quantltles of added ascorbic acM. (Bottom) A calibration curve in water for the Ion abundance of mlz 345-347 (varylng quantlties of dehydroascorbate to 2.5 pg of [6,&*H2]dehydroascorbate) plotted against the known quantlties of
added dehydroascorbic acM.
n I\ 4
JI I I I I 1 1 1 ! 6.7 6.8 6.9 Figure 7. (A) An example of an ascorbate TIC where a linear relationship exists between rnlz 575 and mlz 581 and the content of ascorblc acM and [ 1%8]ascorblcacid. (B) An overbad pattern where there Is loss of the llnear relatbnshlp of mlz 575 and mlz 581 and the content of ascorbic acid and [i3Ce]ascorblcacid.
captopropanol completely reduces dehydroascorbic acid to ascorbic acid in 10 rnin.loJ6J7 A 100 pM solution of ascorbic acid was exposed to light and oxygen for 5 h. The ascorbic (16)Washko, P. W.; Hartzell, W. 0.; Levine, M. Anal. Biochern. 1989; 181, 216.
525
ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993
Table 111. Effect of Chelation on Metal Ion Degradation of Ascorbic Acid ratio of ascorbic relative abundancee of internal standard acid to intemal sample standard mlz 575/581 m/z 581 controla 1.00 f 0.08 1.00f 0.12 DTPA 1.00 f 0.02 0.33 f 0.02 EDTA 0.82 i O M b co.01 CUClZ 0.44 f 0.02b co.01 DTPA/CuClZ 0.62f 0.05 0.32 f 0.01 EDTA/CuClZ 0.28i 0.02 0.39 f 0.10 FeCl3 0.73 .+ 0.08 0.13 f 0.07 DTPA/FeC13 0.40 f 0.10 0.28 i 0.04 EDTA/FeC13 0.04 f 0.01 0.25 f 0.04 Control = 100 1M solution of ascorbic acid frozen at -20 O C for 48 h. The mean of three samples was set at 1. Due to low abundance of rnlz 575/581,this value is an estimate based on mlz 4431449.c The internal standard was added after the freezethaw and was present only during processing of the samples. The relative loss in internal standard represents both oxidation of the internal standard by the test solution during proceeaing and interference with derivatization by test solutions. This relative abundance is only semiquantitative since it is based on a single ion response. acid concentration decreased over this time from 100 to 48 pM,with a large increase noted in rnlz 345 at 4.8 min. This oxidized ascorbic acid solution was then reduced by 50 mM 2,3-dimercaptopropanol for 15 min. Following reduction, the rnlz 345 at 4.8 min was not detectable, even though a small amount of rnlz 345 (dehydroascorbate) was found in the original 100 pM ascorbic acid solution. However, this reduction increased the ascorbic acid concentration from 48 pM to only 91 pM. It appeared that a small, but significant, quantity of the ascorbic acid (approximately 10%) was therefore irreversibly oxidized. The effect of chelators on the degradation of ascorbic acid solutions was examined by freezing (-20 "C) a 100pM solution of ascorbic acid for 48 h in the presence of either EDTA, DTPA, FeC13, or CuClz with an excess of EDTA or DTPA, or metal ions without chelators. As shown in Table 111, ascorbic acid degradation occurred to a greater extent in EDTA solutions compared to either DTPA solutions or solutions withno chelator or addedmetals. DTPA was better at protecting ascorbic acid than EDTA when either copper or iron was added. However, neither DTPA or EDTA was efficient at protecting ascorbic acid from degradation by FeCl3. The Measurement of Ascorbate and Dehydroascorbate in Plasma. The ascorbate content in human plasma from fasting subjects was determined by drawing blood into KBEDTAor heparinized tubes, separating the cells from the plasma by low-speed centrifugation and adding 100 pL of plasma to known amounts of [13C61ascorbicacid. Plasma proteins were precipitated by the addition of TCA. The sample was then ether extracted to remove TCA, derivatized, and analyzed by GC-MS. Plasma ascorbate levels ranged from 7 to 12 pg/mL (40-70 pM). The coefficient of variation of the determination of ascorbate concentrations in five aliquots of plasma from a single blood draw was less than 10%. Heparinized plasma was then examined after the addition of known quantities of unlabeled ascorbic acid and dehydroascorbate where either [13C61ascorbic acid or [l3CSldehydroascorbate was added as an internal standard. As shown in Table IV, an excellent quantitative correlation existed between the predicted and measured increments of ascorbic acid and dehydroascorbate in these plasma samples. The limit of detection of ascorbate in plasma was estimated (17)Dhariwal, K. R.; Washko, P. W.; Levine,M.A ~ d B i o c h e m1990, . 189, 18.
Table IV. The Measured Increase in Ascorbate and Dehydroascorbate after the Addition of Known Quantities of Ascorbic Acid or Dehydroascorbate to Plasma ascorbic acid added, m/mL measured value. udmL 0
8.0 f 0.2 9.9 f 0.3 11.8 f 0.2 measured value. udmL 1.6 f 0.1 3.2 f 0.1 4.9 f 0.2
1.90 3.80
dehydroascorbate added, rrdmL 0 1.70 3.40
40
-
h
T
2 v 3 0 r W
ap 4
2
1
I
10
Y-
01
0
1
I
I
I
60
120
180
240
Time After Oral Ingestion ( m i n ) Flgure 8. The standarddevietlonof plasma ascorbatevalues measured In a single humansubject fotbwlngthe oral ingestkmof 0.1 g of ascorblc ackllkg of body weight.
by taking fresh heparinized plasma and using a serial 1:lO dilutions of an aliquot of that plasma to which [13C61ascorbic acid had been added. The standard was detectable when present in quantities between 50 and 500 nglmL. However, at concentrations less than 50 ngImL, the natural isotope spillover from the endogenous plasma ascorbate at rnlz 581 masked the further decrements in rnlz 581that occurred from diluting out the added [13C61ascorbicacid. A human subject then had plasma ascorbate levels determined prior to and at intervals after the oral administration of 100mglkg of ascorbicacid. The plasma ascorbate increased 3-foldwithin 1h of the ingestion of ascorbic acid and remained elevated for the next 3 h (Figure 8). Plasma dehydroascorbate levels did not appreciably change during ascorbate loading. After ascorbate content in fresh plasma was shown to be accurately quantitated based on the ratio of known amounts of exogenous [13C6]asCOrbiC acid to endogenous ascorbate, studies of the stability of endogenous plasma ascorbate upon freezing were performed. The ascorbate content of plasma collected a t KSEDTA was found to decrease by greater than 80% after 24 h at -20 "C, from 8.99 f 0.34 pg/mL when fresh to 1.66 f 0.16 pglmL after freezing. Degradation of ascorbate in K3EDTAalso occurred in aliquots frozen at -70 "C but to a lesser extent (ascorbate value decreased to 6.90 f 0.15 pg/mL from 8.99 f 0.34 pgl mL). In all instances, as the ascorbate levels decreased in the K3EDTA plasma samples, a corresponding increase was noted in rnlz 345 eluting a t 4.8 min (dehydroascorbate). Heparinized plasma and serum frozen at -20 "C both exhibited approximately 20 % degradation of endogenous ascorbate after 24 h, in contrast to the 80 % degradation found in the K3EDTA plasma samples. However, at -70 "C no
326
ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993
significant degradation of ascorbate was found in either heparinized plasma or serum stored up to 2 weeks. When DTPA, EDTA, and heparin were compared in their ability to protect plasma ascorbate at -20 "C, there was no significantdifferencebetween heparin or DTPA. Heparinized or DTPA-chelatedplasma had greater than 7-fold higher level of ascorbate than an identical sample of EDTA-chelated plasma. Accurate quantitation of ascorbate and dehydroascorbate in biological solutions is important to define the role of ascorbate in vivo redox reactions. This paper outlines a GCMS isotope dilution assay for both of these species that appears to be sensitive, accurate, and relatively simple to perform. Although ascorbic acid has been previously characterized by electron ionization-mass spectrometry,181aserdesorptionmass spectrometry,lQand GC-MS,2021 the mass spectroscopic characterization of TBDMS-derivatized dehydroascorbate, and isotope dilution assays for these compounds have not been reported. Isotope dilution assays for unstable compounds, such as ascorbate, are useful due to the sensitivity and specificity of these techniques and also because degradation of the endogenous compound should also cause degradation of the internal standard without changing the ratio between internal standard and the endogenous substance. GC-MS was shown to be a useful method to study the oxidation of ascorbic acid/ascorbate to the dehydro species because direct changes in specific mlz (345:575) could be monitored under a variety of circumstances. This reduces the potential error that may occur using a high-performance liquid chromatography with electrochemical detection, the currently accepted beat method for ascorbate and dehy-
droascorbate quantitation, where dehydroascorbate is measured indirectlylO,16,1',21-23 and apparently negative values for dehydroascorbatecontent sometimesoccur.1o Furthermore, the GC-MS assay appears able to increase the Sensitivity of analysis by lW-fold, based on the limit of detection of the mass detector compared to the electrochemical detector.loJ6 In addition, dehydroascorbatehas been shown to be unstable under conditions typically used for biologic and dehydroascorbateloss does not always lead to an increase of ascorbate content.24 A major problem with the GC-MS analysis of plasma for dehydroascorbate is the chemical impurity of the available standards (Figure 4A). Although the sensitivity of GC-MS was adequate to examine plasma based on Table IV, potential error in absolute quantitations could occur when content is related to an added impure standard. The data in this manuscript shows that the ratio of dehydroascorbate to ascorbate reflecta oxidative stress in vitro and potentially could reflect oxidative stress in vivo. If so, the ratio of plasma dehydroascorbate to ascorbate may be useful to epidemiologic studies linking oxidative stress and injury to common human illnessessuch as cancer and ischemic heart disease.9~~~ This GC-MS assay appearsto offer an advance in the ability to measure ascorbate and dehydroascorbate. The steps required for sample processing are similar to the sample preparation used for HPLC with electrochemical detection.lOJ6J7~21-23 The sensitivity appears to be improved,lOJ6 and the specificity should be superior since GC-MS involves direct analysis based on retention time and mass. GC-MS analysis provides another excellent mechanism to examine ascorbate metabolism and degradation.
(18) Ng, Y. C.; Akera, T.; Han, C. S.; Braselton, W. E.; Kennedy, R. H.; Temma, K.; Brody, T. M.; Sato, P. H.Biochem. Pharmacol. 1985,34, 2525. (19) McMahon, J. M. A w l . Biochem. 1985,147, 535. (20) Knaack, D.; Podleski, T. h o c . Natl. Acad. Sci. U.S.A. 1985,82, 575. (21)Honegger, C. G.; Krenger, W.; Langemann, H.; Kempf, A. J. Chromatogr. 1986, 381, 249. (22) Margolis, S. A,; Paule, R. C.; Ziegler, R. G. Clin. Chem. 1990,36, 1750. (23) VanderJagt, D. J.; Garry, P. J.; Bhagavan, H. N. Am. J. Clin. Nutr. 1989, 49, 511. (24) Bode, A. M.; Cunningham, L.; Rose, R. C. Clin. Chem. 1990,36, 1807. (25) Gey, K. F.; Stahelin, H. B.; Puska, P.; Evans, A. In Third Conference on Vitamin C; Burns, J., Rivers, J., Machlin, L., Eds.; N.Y. Acad. Sci.: New York, 1987; p 110.
We wish to thank Valarie Allen-Ruebush for her assistance in the preparation of this manuscript. This work was supported in part by a Senior Fellow Research Award from the American Gastroenterological Association, a Veterans Administration Research Advisory Group Award OOO1 (J.C.D.), and a Department of Health and Human Services Research Grant No. GM26486 awarded by the National Institute of General Medical Sciences (J.F.K.), National Institute of Health.
~~
ACKNOWLEDGMENT
RECEIVED for review August November 17, 1992.
17, 1992.
Accepted
