Chem. Res. Toxicol. 1992,5, 81-89
81
Characterization and Quantitation of Urinary Metabolites of [1,2,3-13C]Acrylamide in Rats and Mice Using 13C Nuclear Magnetic Resonance Spectroscopy Susan C. J. Sumner,* John P. MacNeela, and Timothy R. Fennel1 Chemical Industry Institute of Toxicology, P.O.Box 12137, Research Triangle Park, N o r t h Carolina 27709 Received July 15, 1991
Acrylamide, widely used for the production of polymers and as a grouting agent, causes neurotoxic effects in humans and neurotoxic, genotoxic, reproductive, and carcinogenic effects in laboratory animals. In this study, 13C NMR spectroscopy was used to detect metabolites of acrylamide directly in the urine of rats and mice following administration of [1,2,3-13C]acrylamide (50 mg/kg PO). Two-dimensional NMR experiments were used to correlate carbon signals for each metabolite in the urine samples and to determine the number of hydrogens attached t o each carbon. Metabolite structures were identified from the NMR data together with calculated values of shift for biochemically feasible metabolites and by comparison with standards. The metabolites assigned in rat and mouse urine are N-acetyl-S-(3-amino-3-oxopropyl)cysteine,
N-acetyl-S-(3-amino-2-hydroxy-3-oxopropyl)cysteine, N-acetyl-S-(l-carbamoyl-2-hydroxyethyl)cysteine, glycidamide, and 2,3-dihydroxypropionamide.These metabolites arise from direct conjugation of acrylamide with glutathione or from oxidation to the epoxide, glycidamide, and further metabolism. Acrylamide was also detected in the urine. Quantitation was carried out by integrating the metabolite carbon signals with respect to that of dioxane added at a known concentration. The major metabolite for both the rat (70% of total metabolites excreted) and the mouse (40%) was formed from direct conjugation of acrylamide with glutathione. The remaining metabolites for the rat (30%) and mouse (60%) are derived from glycidamide. The species differences in extent of metabolism through glycidamide may have important consequences for the toxic and carcinogenic effects of acrylamide.
Introduction Acrylamide (AM)' is widely used for the production of polymers and as a grouting agent. In humans, exposure to AM can cause neurotoxic effects ( 1 ) . Neurotoxic, genotoxic, reproductive, and carcinogenic effects have been reported for AM in laboratory animals (1,2). Disposition studies have shown that, in the rat, AM is rapidly removed from the bloodstream with a half-life of approximately 2 h (3,4). Within 24 h after administration of either [1-14C]or [2,3-14C]AM,approximately 60% of the radioactivity is excreted in the urine, with little further increase in excretion for up to 7 days ( 4 , 5 ) . With [1-14C]AM,approximately 6% of the administered radioactivity is exhaled as 14C02. Acrylamide is electrophilic and reacts readily with glutathione in vitro, and it is excreted in bile as a glutathione conjugate ( 3 , 5 ) . N-Acetyl-S-(g-amino3-oxopropy1)cysteine has been identified as the major urinary metabolite of AM in the rat (4),formed by further metabolism of the glutathione conjugate. In addition, a hemoglobin adduct formed from cysteine conjugation with AM has been detected in rats (6). Studies have suggested that acrylamide is oxidized to a reactive epoxide (7), and the in vivo detection of a hemoglobin adduct derived from cysteine conjugation with glycidamide has provided direct evidence for metabolism of AM to the epoxide glycidamide (8). In this study, we investigated the metabolism of AM by detection, identification, and quantitation of metabolites produced in the urine of rats and mice administered [1,2,3-13C]AMby gavage. These studies were carried out on the complex mixture of endogenous and exogenous compounds in urine, where two-dimensional NMR ex-
* To whom correspondence shouldbe addressed.
~
0 8 9 3 - 2 2 8 f~92 f 2705-O081$Q3,0Of 0
periments were used to correlate signals for the labeled carbons in each metabolite and to determine the number of hydrogens attached to each labeled carbon. Quantitation was accomplished by adding dioxane as a standard, and integrating metabolite carbon signals with respect to that of the dioxane. This method, as we have previously described (9-ll), enables metabolite structures to be determined and quantified without the need for authentic standards and without generating structural artifacts which can arise when using methods of extraction, chromatography, and mass spectrometry. The metabolites identified in both rat and mouse urine are N-acetylS(3-amino-3oxopropyl)cysteine, N-acetyl-S-(3-amino-2-hydroxy-3oxopropyl)cysteine, N-acetyl-S-(l-carbamoyl-2-hydroxyethyl)cysteine, glycidamide, and 2,3-dihydroxypropionamide. A c r y h i d e was also detected in the urine. Species differences in metabolism were determined for the conversion of AM to its metabolites.
Materials and Methods Chemicals. Carbon-13-enriched (99%) acrylamide ([ 1,2,313C]AM)was obtained from Cambridge Isotopea Ltd.(Cambridge, MA) and gave a 'H-decoupled 13C NMR spectrum in CDC13 consisting of two doublets at 127.7 ppm (69 Hz)and 167.1 ppm (59 Hz) and a doublet of doublets at 129.9 ppm (59, 69 Hz). Glycidamide was obtained from Polysciences, Inc. (Warrington, PA) and gave a 'H-decoupled I3C NMR spectrum in CDC1, consisting of three singlets at 47.3,49.2, and 171.4 ppm. D20and CDCl, were purchased from Merck & Co., Inc. (St. Louis, Mo.). S-(3-Amin~3-oxopropyl)cysteine was purified from the reaction of cysteine and acrylamide in the presence of triethylamine, as described by Calleman et al. (8),and identified by 'H and 13C Abbreviations: AM, acrylamide; INADEQUATE, incredible natural abundance double quantum transfer; HETZDJ, heteronuclear 2D J resolved; IR, infrared.
0 1992 American Chemical Society
82 Chem. Res. Toxicol., Vol. 5, No. 1, 1992 NMR. Products produced after reaction of glycidamide (1.1 "01) and N-acetylcysteine (1.1 "01) in the presence of excess triethylamine in HzO were identifed by 'H and 13CNMR without purification. Animals. Male Fischer 344 rats (216-228 g, 9-10 weeks) and male B6C3Fl mice (24-26 g, 6-7 weeks) were purchased from Charles River Laboratories (Raleigh, NC). They were supplied food (NIH 07 diet) and deionized water ad libitum and maintained on a 12-h light-dark cycle a t a temperature of 22 f 2 "C and relative humidity of 55; h 5%. Four rats and three mice were administered [1,2,3-13C]AMPO (50 mg/kg). The acrylamide was dissolved in distilled water and administered in volumes of 1.0 or 10.0 mL/kg body weight for rats and mice, respectively. The animals were placed in glass metabolism cages, and urine was collected for 24 h. Control urine samples for rats and mice were obtained in a parallel study (9). The urine samples were centrifuged at 600g for 20 min to remove particulate material and were analyzed immediately or stored a t -20 "C. NMR Spectroscopy. Samples were prepared for NMR studies by adding 100 pL of D 2 0 to 600 pL of the centrifuged urine. Concentrated samples were prepared for two-dimensional NMR studies by adding equal volumes (3-6 mL) of methanol to the urine, centrifuging, reducing the volume under a stream of nitrogen gas, and adding DzO to a final volume of 400-600 pL. NMR spectra were acquired with a 5-mm dual proton-multinuclear probe on a Varian VXR-300 spectrometer (Palo Alto, CA). Carbon-13 NMR spectra were acquired in the doubleprecision mode with an acquisition time of 0.9 s, 30K data points, a relaxation delay of 5 s, and a 45" pulse width. Spectra for urine samples prepared by addition of only DzO were acquired with approximately loo00 transients. All carbon spectra were referenced to urea a t 162.5 ppm. Two-dimensional incredible natural abundance double quantum transfer (INADEQUATE) spectra (12) were acquired using the INADQTprogram from the Varian pulse sequence library. Relaxation delays ranging between 5 and 10 s and 7 values corresponding to coupling constants of 10, 40, and 60 Hz were used to acquire data over the entire spectral window. Broad-band decoupling was employed throughout the pulse sequence, and spectra were acquired in the phase-sensitive mode with 1024 complex points in tz and 16-32 complex points in tl. Two-dimensional heteronuclear J-resolved (13)spectra were acquired using the HETZDJ program from the Varian pulse sequence library, with gated decoupling applied to the proton spin system. NMR spectra were acquired with 2048 complex points in t z and 64 complex points in tl. Calculated values of shift for carbons of biochemically feasible metabolites were obtained using incremental substituent effects for alkanes (14, 15), and using the Carbon-13 NMR Database (STN International, Columbus, OH). Samples were prepared for quantitative analysis by adding an aliquot of a dioxane solution in DzO. For rat urine, 100 pL of dioxane standard (116.2 mM) was added to 500 pL of urine (final concentration 19.36 mM). For mouse urine, 100 pL of dioxane standard (70.0 mM) was added to 400 pL of urine (final concentration 14.0 mM). The relaxation times of signals arising from metabolite and dioxane carbons were determined. Spectra were acquired with a delay time (24 s; a t least 4 times the TIfor metabolite or dioxane carbon signals) sufficient to relax carbons, and with proton decoupling only during acquisition. Concentrations were determined by comparing integrals for metabolite carbon signals to that of dioxane. A validation of this method is described on the quantitation of acrylonitrile metabolites (9).
Results Metabolites in Rat Urine. The 13CNMR spectra of control rat urine and of urine collected for 24 h following oral administration of 50 mg/kg [ 1,2,3-13C]acrylamideare shown in Figure 1, panels A and B, respectively. In the control spectrum (Figure lA), signals from carbons of endogenous compounds appear as single resonances. The most intense resonance in the control spectrum, at 162.5 ppm, arises from urea. Signals with weaker intensities appear for other endogenous compounds such as creati-
Sumner et al. Table I. Chemical Shifts (ppm), CarbonCarbon Coupling Constants (Hz), CarbonCarbon Correlations, and Proton Multiplicities for Resonances of Acrylamide and Metabolites in Rat and in Mouse Urine after Administration PO of Carbon-13-Enriched Acrylamide (CHz4HCONHz)" chemical carbon proton connectivity multiplicity Carbon-13 shift, coupling, Hzb (INADEQUATE)' (HETZDJ)~ Databasee DDm 27.22 36 la 28 34.83f 35,48 Ib 36 2a 35.96 37 34 36.14 38 2'a 34 25 4a 46.98 48 48.61 26,64 4b 49 49.81 38,50 3b, 3'b 47 37 61.60 3a 66 61.468 37 3'a 66 63.18 40 5a 65 70.22 38, 52 2b 79 70.34 38, 52 2'b 79 72.01 40, 52 5b 70 6a,b 126-130 127, 132 170.63 54 6c 171 4c 173.78 62 178 175.28 5c 173 3c 175.50 50 173 177.07 IC 49 173 177.55 53 2,2'c 175 Metabolite structures: 1, RSCH2CH2CONH2; 2, RSCH2CHOHCONH2;3, RSCH(CH2OH)CONHz;4, CHz(0)CHCONH2; 5, CHZOHCHOHCONH,; 6, CH24HCONH2 (R = Nacetylcysteine). Molecular structure 1-5 are proposed from the connectivitiy, multiplicity, and chemical shift data. Distinction between COOH and CONHz structures was not possible on the basis of chemical shift data. Amide assignments were made by analogy with the previously assigned urinary metabolite (4). Carbon-carbon coupling constants were measured from the 1D spectrum. Carbon connectivities were obtained using INADEQUATE spectroscopy. The carbon atom is assigned a number corresponding to the metabolite and a letter which designates its derivation from acrylamide (a, b, or c corresponding to the 3-, 2-, or 1-carbon of acrylamide). dThe number of hydrogens attached to each carbon was determined using HETzDJ spectroscopy. eValues of shift were obtained through the Carbon-13 Database. IAn additional set of peaks appear on the shoulder of these signals in mouse urine. gSignal present in spectra of mouse urines.
nine, sugars, and hippurate (16). The spectrum of urine obtained after AM administration to rats (Figure 1B) contains resonances from the endogenous compounds and also resonances not present in the control spectrum. The signals which are not present in the control spectrum and that possess coupling patterns associated with enriched carbon-13 nuclei are assigned to metabolites of AM. The coupling patterns are produced by spin-spin interactions between carbons in the portion of each metabolite derived from the enriched carbons of AM. Here, a 13C nucleus with one adjacent 13C nucleus gives rise to two equally intense resonances, while a 13C nucleus with two adjacent 13C nuclei gives rise to four equally intense resonances. 13Ccarbons in normal urinary compounds (1.1 0'5 natural abundance) give rise to single resonances because of the low incidence of adjacent 13C nuclei. The chemical shifts (for the center of the multiplet patterns) and the carbon-carbon one-bond coupling constants for metabolites of AM are listed in Table I. The carbon-carbon coupling constants measured for resonances in the 1D spectrum can be used to determine carbon connectivity. However, in the complex urine mixture, several multiplet patterns have s i m i i intensities and couplings, resulting in ambiguous assignment. INADEQUATE spectroscopy enables the determination of carbon-carbon connectivity by providing a correlation
Urinary Metabolites of Acrylamide
Chem. Res. Toxicol., Vol. 5, No. 1, 1992 83
urea
A
d
diagram for the signals which result from each metabolite. An INADEQUATE spectrum obtained on a concentrated
sample of rat urine (see Materials and Methods) is shown in Figure 2. The 1D 13C NMR spectrum of the concentrated urine, which is plotted on the F1 axis, does not contain some signals (at 46.98 and 48.61 ppm) which are in the spectrum obtain$ on the urine before concentration (Figure 1B). In addition, the spectrum of concentrated urine has additional signals which arise from carbon-13labeled carbons, indicating that a metabolite undergoes structural changes on concentration. Signals corresponding to all other metabolites of AM are identical in the spectra obtained before and after concentration. In the INADEQUATE spectrum, signals arising for adjacent carbons in each metabolite are connected by contour peaks. For example, the two contours a t the chemical shift of carbon 3b (50 ppm) are horizontally aligned with one contour at 60 ppm (carbon 3a) and one a t 176 ppm (carbon 34. Contours for carbons 3a and 3c are only aligned with 3b. These connectivities establish a C,-Cb-C, carbon skeleton for the structural portion of metabolite 3 which is derived from AM. The INADEQUATE spectrum was plotted to specifically show connectivities for carbon signals having low intensities, resulting in l i e s of noise for the intense signals (27,35, and 177 ppm) arising from carbons of metabolite 1. The INADEQUATE spectrum enables determination of connectivities (listed in Table I) for seven compounds, where two
of these compounds (see below) are produced after concentrating the urine. H E ~spectroscopy J was used to determine the number of hydrogens attached to each carbon. An expanded region of the HETODJ spectrum is shown in Figure 3. The 13C NMR spectrum is plotted on the F2 axis. The number of hydrogens attached to a carbon is equal to the number of contours located on the Fl axis at the carbon shift position minus one. Each signal for carbons 3a and 5a has three contours located a t ita chemical shift position on the F1 axis and is assigned to methylene (CH,) carbons. The signals for carbon 2,2'b have two contours located at their chemical shift position on the F1 axis and are assigned to methine (CHI carbons. The number of hydrogens attached to other metabolite carbons are listed in Table I. In the HETPDJ spectrum, the signals from carbon-13-enriched carbons have contours which are not symmetrical about 0 Hz (Fl axis), while signals from natural abundance carbons have contours which do have symmetry about 0 Hz (data not shown). This difference is due to spin-pin interactions which only occur for the enriched carbons and enables further distinction of exogenous (100% W) and endogenous (1.1% 13C) carbon signals. Assignment of Metabolites. The assignments of metabolite structure were carried out using the data obtained on carbon-carbon correlation and proton multiplicity together with calculated values of shift for biochemically feasible metabolites. The assigned metabolites and
Sumner et al.
84 Chem. Res. Toxicol., Vol. 5, No. 1, 1992
0 2000 I
4000
I ( '
Y
I
6000
8000 10000 12000 14000
200
ieo
160
140
loo
120
F2
60
80
20
40
0
(PPMI
Figure 2. INADEQUATE spectrum on rat urine collected for 24 h following administration of [1,2,3-13C]AM. The urine was concentrated as described under Materials and Methods. The spectrum was acquired with 2048 points in tz, 32 increments in tl, a relaxation delay of 10 s, and a T value corresponding to a 4 0 - H ~coupling. The spectrum was plotted to show carbon-carbon connectivities for compounds with low signal intensities and therefore contains three lines of noise for the intense signals from carbons in metabolite 1. Correlation of signals for each metabolite is determined by tracing the connectivitiesfor signals at different chemical shifts (along the x-axis) which have the same double quantum frequency along the y-axis. Three signals (*) are not present in the I3C spectrum obtained directly on the urine (Figure 1B).
a
250
200
150
100
50
0
-50
-100
-150
-200
-250
-3 0
F 1 ihZ1
Figure 3. An expanded region of the HETPDJ spectrum of rat urine collected for 24 h following administration of 50 mg/kg [1,2,3-13C]AM. The number of hydrogens attached to a carbon is equal to the number of contours minus one at the chemical shift of the carbon.
the proposed pathways of formation are shown in Scheme I. The INADEQUATE data for metabolite 1 shows correlation for the carbon signal at 35 ppm (lb) with signals at 27 (la) and 177 (IC)ppm. The HETPDJ spectrum shows that the peaks at 27 and 35 ppm are CH2 carbons, and the chemical shift at 177 ppm indicates a carbonyl (CO) carbon. Together, these data indicate an XCHzCH2COY structure for the portion of the metabolite derived from AM. Using incremental shift values for substitution of functional groups on alkanes, the groups X and Y can be determined. A CONH2group has a and j3 effects of 22 and
3,respectively. To fit the experimental values of 27 and 35 ppm, and assuming the functional group Y is NH2, the functional group X on an ethane carbon must have a and j3 effects of 18 and 7, respectively. An RS group (a= 20, j3 = 9) is the only feasible substituent which satisfies these conditions, resulting in a structure of RSCH2CH2CONH2. Given the previous description of N-acetyl-S-(3-amino-3oxopropy1)cysteine as the major urinary metabolite of AM in the rat (4), RS is most probably N-acetylcysteine. Chemical shift values for a synthetic standard of S43amino-3-oxopropy1)cysteine (Table 11) and calculated values for the N-acetyl derivative (Table I) are consistent
Urinary Metabolites of Acrylamide
Chem. Res. Toxicol., Vol. 5, No. 1, 1992 85
-
-
Scheme I. Proposed Metabolism Scheme for AM in the Rat and Mouseo H~C-CH-CONH,
0 H~C‘&-CONH, Metabollte 4
HOCH,CHOH-CONH,
Metabolite 5
CHpOH I
GS-CHp-CHp-CONHp I
GS-CHp-CHOH-CONHp
GS-CH-CONHp
I I I
I
I
i
.;.
0
I
I
I
CHpOH
I
N-AcCys-S-CHp-CHp-CONHp Metabolite 1
N-AcCys-S-CHp-CHOH-CONHp Metabolite 2
N-AcCyS-S-CH-CONHp
Metabolite 3 GS represents a glutathionyl residue, and N-AcCys-S represents an N-acetylcysteine residue, Table 11. Carbon-13 Chemical Shifts (ppm) of Acrylamide (AM), Glycidamide (GA), and Their Cysteine (Cys) or N-Acetylcysteine (NAC) Conjugates carbon derived from AM or GA carbons derived from N-Ac or Cys C CRHz C,H COCHq CO a b 127.65 (CHJ 129.89 (CH) 167.07 (CO) acrylamide 47.26 (CHJ 49.15 (CH) 171.37 (CO) glycidamide S-(3-amino-3-oxopropyl)-Cys 26.95 (SCHZ) 34.57 (CHZ) 176.89 (CO) 31.99 53.42 172.45 N-Ac-S-(3-amino-2-hydroxy-3-oxopropyl)-Cys 36.01 (CH& 70.23 (CH) 177.26 (CO) 34.87 54.46 176.08 172.91 36.21 70.37 34.63 54.58 21.97 N-Ac-S-(l-carbamoyl-2-hydroxyethyl)-Cys 61.58 (CH,) 49.73 (CH) ndb 33.77 54.28 nd nd 49.98 33.84 54.38 2,3-dihydroxypropionamide 63.12 (CH,) 71.96 (CH) nd I
_
-
The 13Cspectra of AM and GA were referenced to CDC13at 77 ppm. The products produced on reaction of AM with Cys and GA with NAC were run in DzO. These spectra were referenced by using the same reference values (RFL and RFP values on the Varian spectrometer) which positioned the urea peak at 162.5 ppm in the spectra of rat and mouse urines. GA was also run in DzO and referenced using these RFP and RFL values. This spectrum contained peaks at 46.96,48.52,and 173.76 ppm. * nd: Peaks not assigned due to overlapping signals or insufficient signal to noise (not determined).
with the experimental shift values for this metabolite in the rat urine. The NMR data obtained for the identification of metabolites in this study do not readily distinguish between amide and carboxyl groups for the carbonyl carbon derived from acrylamide. The assignments made in this study assume that the amide group is retained. Metabolite 2 (designated 2,2’) has two sets of similar multiplet pattems which have INADEQUATE correlations between signals at 70 ppm (2,2’b) and signals at 36 (2,2’a) and 178 (2,2’c) ppm. The HETSDJ spectrum results in assignment of CH2and CH carbons for 2,2’a and 2,2’b, respectively, while the shift of 178 ppm indicates a carbonyl (CO) carbon for 2,2’c. These data suggest an XCH2CHZCOY structure, where diastereomers can account for the similar multiplet patterns. Calculated values of shift (Table I) are consistent with the structure RSCH2CH(OH)CONH2.Assuming R is N-acetylcysteine, this metabolite is assigned N-acetyl-S-(3-amino-2hydroxy-3-oxopropy1)cysteine. The INADEQUATE data for metabolite 3 show correlation for the carbon signal at 50 ppm (3b) to signals at 61 (3a) and 176 (3c) ppm. The HETSDJ spectrum results in assignment of CH2 and CH carbons for 3a and 3b, respectively, while the shift at 176 ppm indicates a CO carbon for 3c. The resulting structure is XCH(CH2Z)COY. Calculated values of shift (Table I) are consistent with RSCH(CH20H)CONH2,indicating N-acetyl-S-(l-carbamoyl-2-hydroxyethy1)cysteineas a metabolite of AM. Metabolite 4 has shifts centered at 47 (J = 25 Hz), 49 (J= 25,64 Hz),and 174 (J = 64 Hz) ppm. The correlation of these signals was based on the distinct carbon-carbon coupling constantsmeasured in the 1D 13CNMR spectrum
obtained directly on the urine. The coupling constant of 25 Hz between two of the carbons is consistent with a cyclic structure (15). Chemical shift values for a standard of glycidamide (Table 11)and calculated values of shift for glycidamide (Table I) are nearly identical to those recorded in the urine, indicating that the epoxide of AM is excreted in the urine. Further evidence for the excretion of the epoxide is found from the addition to the urine sample of sodium thiosulfate, which would be expected to react with the epoxide. The 13C NMR spectrum obtained on the urine after addition of sodium thiosulfate does not contain the signals (47,49, and 174 ppm) assigned to glycidamide. However, the spectrum does have additional signals from carbon-13-enriched nuclei at 38 and 69 ppm. Signals with these shift positions are also present in the 13Cspectrum obtained after the direct reaction of glycidamide and sodium thiosulfate in DzO. The signals assigned to glycidamide (at 47,49, and 174 ppm) are not present in the spectra obtained on some of the concentrated urine samples. However, additional signals from carbon-13-enriched nuclei are present and have connectivity in the INADEQUATE spectrum between 28 and 61 ppm and between 46 and 70 ppm. These compounds may be products formed from reaction of glycidamidewith endogenous compounds in the urine upon concentration. A fifth metabolite is present which has INADEQUATE connectivities between signals at 72 ppm (5b) and signals at 63 (5a) and 175 (512)ppm. HETSDJ shows a CH2 carbon for the signals at 63 ppm, and the shift at 175 ppm indicates a CO carbon. These data indicate an XCH2CHYCONH2structure for metabolite 5. Calculated
Sumner et al.
86 Chem. Res. Toxicol., Vol. 5, No. 1, 1992
Table 111. Quantitative Determination of Metabolites of [ 1,2,3-1SC]AMExcreted in Rat Urine" chemical concentration, % of total metabolite shift, ppm mM metabolitesb 1 27.22 5.46 f 0.22 67.4 f 3.6 2,2' 36.05 1.27 f 0.16 15.7 f 1.3 3 61.60 0.73 f 0.12 9.0 f 1.1 4 46.98 0.44 f 0.09 5.5 f 1.0 5 63.18 0.19 f 0.06 2.4 f 0.7
I
'''I"-
d Figure 4. 13C Nh4R spectrum obtained on a mixture of products produced from reaction of glycidamide with N-acetylcysteine in the presence of excess triethylamine. Signals from the three reaction products are labelled 1-3, with the letters a, b, and c representing the 3-, 2-, or 1-carbon,respectively, derived from glycidamide. The in vitro products have shift values for the carbons derived from glycidamide (Table 11) which are consistent with signals for in vivo metabolites. 00
80
SC
4C
20
'
8
N
8
8
r
DPY
values of shift (Table I) are consistent with hydroxyl groups on both the CH2and CH carbons. This compound (2,3-dihydroxypropionamide)is a hydrolysis product of glycidamide. Metabolite 6 has signals at 126130 ppm which overlap signals from the endogenous hippurate. The INADEQUATE spectrum shows connectivity between signals at 130 and 170 ppm. The signals for metabolite 6 are consistent with those obtained for a standard of [1,2,3J3C]AM (Table 11),indicating that a small amount of the parent compound is excreted in the urine. In Vitro Reaction of N-Acetylcysteine and Glycidamide. The reaction of N-acetylcysteine with glycidamide has been investigated to compare reaction products with the metabolites found in the urine. The 13C NMR spectrum obtained on a 1:l mixture of glycidamide and Nacetylcysteine reacted in the presence of triethylamine is shown in Figure 4. Three reaction products give rise to carbon signals in the spectrum. Two of these products (labeled 1 and 2) have shift positions consistent with compounds formed after conjugation of N-acetylcysteine at the 3- and 2-carbon positions of glycidamide. These two products, formed in a 1O:l ratio, are assigned to N acetyl-S-(3-amino-2-hydroxy-3-oxopropyl)cysteine and N-acetyl-S-(l-carbamoyl-2-hydroxyethyl)cysteine.Calleman et al. have previously characterized S-(3-amino-2hydroxy-3-oxopropy1)cysteine as the product produced on the in vitro reaction of glycidamide with cysteine (8). The third in vitro reaction product (labeled 3) has chemical shifts consistent with 2,3-dihydroxypropionamide,a hydrolysis product of glycidamide. All three in vitro products have chemical shifts for the carbons derived from glycidamide (Table 11) which are nearly identical to those detected for the in vivo metabolites 2,2', 3, and 5. The relative intensities of carbon resonances produced in the in vitro reaction (101 for products 1:2) differ significantly from the ratios of the in vivo metabolites (107 for 2,2':3). This suggests that enzymatic processes are involved in the in vivo formation of these metabolites. Quantitation of Rat Urinary Metabolites. Quantitation of metabolites was carried out after adding a known concentration of dioxane to the urine samples (see Materials and Methods). After measuring Tl values for dioxane and the metabolite carbons,quantitation experiments were run with a delay time (25 s) sufficient to relax carbons in compounds to be quantitated, and with proton decoupling only during acquisition. The concentrations of metabolites were determined by comparing the integrals for the me-
"Rata were administered a dose of 150.89 f 3.18 pmol of [1,2,3I3C]AM, and metabolites were measured in 24-h urine samples using dioxane as a standard. The total amount excreted in 24 h was 76.1 f 2.7 pmol. Values listed represent mean & SD (n = 3). bThe amount of each metabolite is expressed as a percentage of total metabolites excreted in the urine in 24 h.
tabolite and the dioxane carbon signals. For some metabolites, two carbon signals derived from AM were quantitated, resulting in similar concentrations, A validation of these quantitation experiments is given for the analysis of acrylonitrile metabolites (9). The amount of AM excreted in the urine was not quantitated due to the long relaxation times for alkene carbons and due to the overlap of AM signals with those of an endogenous compound. The amounts of metabolites detected in rat urine, collected for 24 h following administration of AM, are shown in Table 111. Metabolites 1-5 amounted to 50% of the administered dose. The major metabolite (67% of the total metabolites excreted) for the rat was N-acetyl-S-(3amino-3-oxopropy1)cysteine(metabolite l),formed after direct conjugation of AM with glutathione. Metabolite 2,2', formed after glutathione conjugation with the 3-carbon position of glycidamide, accounted for 16% of the excreted metabolites. Conjugation of glutathione with the 2-carbon position of glycidamide (metabolite 3) accounted for 9% of the excreted metabolites. Glycidamide (metabolite 4) accounted for 6% of the total metabolites excreted, while the hydrolysis product (2,3-dihydroxypropionamide,metabolite 5) accounted for 2% of the amount excreted. The total amount of metabolites derived via glycidamide accounted for 33% of excreted metabolites. Metabolites in Mouse Urine. The 13CNMR spectra of control mouse urine and of urine collected for 24 h following oral administration of 50 mg/kg [1,2,3-13C]AM are shown in Figure 5, panels A and B, respectively. The control spectrum (Figure 5A) is very similar to that obtained for rat urine, where the most intense resonance at 162.5 ppm is assigned to urea. The spectrum of urine obtained after AM administration (Figure 5B) contains signals from metabolites with chemical shifts and coupling patterns like those detected in rat urine. The similarities in chemical shift and coupling show that AM is converted to the same metabolites in the mouse as in the rat. An additional set of signals (at 61.5 and 49.8 ppm) from carbon-13-enriched nuclei in the spectrum of mouse urine nearly overlap those assigned to carbons 3a and 3b (61.60 and 49.8 ppm). This suggests the presence of diastereomers for N-acetyl-S-(l-carbamoyl-2-hydroxyethyl)cysteine (metabolite 3) in the mouse, whereas only one set of signals was present in the spectrum of rat urine. This may be due to differences in urine composition (e.g., pH or concentration) between rats and mice, or the absence of a specific diastereomer in the rat. A second additional set of signals from carbon-13-enriched nuclei appear as shoulders on the doublet of doublets near 35 ppm. The multiplet at 35 ppm is assigned to carbon l b of N-acetyl-S-(3-amin0-3-0~0propy1)cysteine (metabolite 1). These shoulder signals suggest the presence of a compound similar to metabolite 1where the RS portion of the molecule is slightly altered,
Chem. Res. Toxicol., Vol. 5, No. 1, 1992 87
Urinary Metabolites of Acrylamide
A
/1.
B
I
6b
Figure 5. I3C NMR spectrum of control mouse urine (A) and urine collected for 24 h after oral administration of 50 mg/kg [ 1,2,3-13C]AM(B). Chemical shift positions for metabolites of AM are nearly identical to those obtained for rat urine. Table IV. Quantitative Determination of Metabolites of r1.2.3-WlAM Excreted in Mouse Urine" ______~ chemical concentration, % of total metabolite shift, ppm mM metabolites* 41.2 f 2.2 1 27.22 2.67 f 0.10 1.37 f 0.06 21.3 f 0.6 2,2' 36.05 11.7 f 0.6 3,3' 61.53 0.76 f 0.06 1.09 f 0.11 16.8 2.1 4 46.98 0.34 f 0.08 5.3 f 1.2 5 63.18
*
Mice were administered a dose of 16.66 f 0.32 pmol of [1,2,313C]AM,and metabolites were measured in 24-h urine samples using dioxane as a standard. The total amount excreted in 24 h was 8.45 f 0.74 pmol. Values listed represent mean f SD (n = 3). bThe amount of each metabolite is expressed as a percentage of total metabolites excreted in the urine in 24 h. An epoxide degradation product is also detected in 2 of the 3 mouse urine samples (see text) and accounts for approximately 4% of the total metabolites excreted.
while the carbons derived from acrylamide are in the same type of structural environment. This compound may be S-(3-amino-3-oxopropyl)cysteine, as previously described by Dixit et al. (17). S-(3-Aminc-3-oxopropyl)thioaceticacid is also consistent with the NMR data and with the assignment of S-(2-~yanoethyl)thioaceticacid2 as a metabolite of acrylonitrile in mouse urine (9). Quantitation of Metabolites in Mouse Urine. Quantitation of metabolites (Table IV) in mouse urine collected for 24 h following administration of AM was carried out as described for rat urine. Metabolites 1, 2,2', 3,3', 4, and 5 accounted for 51% of the administered dose. N-Acetyl-S-(3-amino-3-oxopropyl)cysteineaccounted for 41% of the total metabolites excreted and was the major metabolite in mouse urine. Metabolites 2,2' and 3,3' (derived from glutathione conjugation with glycidamide) accounted for 21% and 1270,respectively, of the excreted metabolites. Glycidamide (metabolite 4) was excreted at
* G . L. Kedderis, S. C. J. Sumner, S. D. Held, R. Batra, M. J. Turner, Jr., A. E. Roberts, and T. R. Fennell, unpublished results.
17% while the hydrolysis product (metabolite 5) accounted for 5% of the total metabolites excreted. For two of the three mice, a degradation product of the epoxide (with chemical shift positions similar to those observed after concentration of rat urine) accounted for 4% of the excreted metabolites. The amount of metabolites formed via glycidamide accounted for 59% of the excreted metabolites.
Discussion Methods of NMR spectroscopy have been used to detect, assign, and quantitate metabolites directly in the urine of rats and mice administered [ 1,2,3-13C]acrylamide PO. Metabolites were assigned from structural information obtained for the carbon-13-labeled portion of each metabolite, using one- and two-dimensional NMR spectroscopy and calculated values of shift for biochemically feasible metabolites. General advantages of this approach for characterizing the metabolism of small polar molecules include the simultaneous detection directly in the urine of all excreted metabolites, the fact that structural information can be derived from the spectral data without the need for synthetic standards, the lack of a requirement for radiolabeled compounds, and the avoidance of chromatographic techniques that may alter metabolite structure. In this study, NMR was also used to distinguish metabolites that are structural isomers and diastereomers. Although the sensitivity of NMR is much lower than that of HPLC and GC/MS, carbon-13 NMR provides resolution of individual metabolites at both low and high concentrations. This enables structural information to be directly obtained for unknown metabolites, even when they represent a small fraction of the total metabolites. Metabolites of acrylamide were identified that account for as little as 2 % of the total dose excreted (76 pmol) in rat urine and 4% of the total dose excreted (8pmol) in mouse urine. The major metabolite detected in both rat and mouse urine results from direct reaction of AM with glutathione and is assigned to N-acetyl-S-(3-amino-3-oxopropyl)cysteine. This metabolite, which has been previously described in rat urine (4,17), accounts for 67% of the total metabolites excreted in rat urine and 41% of the total metabolites excreted in mouse urine. The epoxide of AM, glycidamide, was detected in rat and mouse urine and accounts for 6% and 17% of the total metabolih excreted, respectively. The remainder of the products identified were derived from glycidamide. Two metabolites in rat and mouse urine are assigned to products produced after glutathione conjugation with the 3- and 2-carbon positions of glycidamide and are assigned to N-acetyl-S-(&amino2-hydroxy-3-oxopropy1)cysteineand N-acetyl-S-( l-carbamoyl-2-hydroxyethyl)cysteine,respectively. These metabolites account for 16% and 9% of the total metabolites excreted in rat urine and 21% and 12% of the total metabolites excreted in mouse urine. A hydrolysis product of glycidamide, 2,3-dihydroxypropionamide,accounts for 2% of the total metabolites excreted in rat urine and 5 % of the metabolites in mouse urine. A n additional product produced after degradation of the epoxide accounted for approximately 4 % of the metabolites excreted in mouse urine. Previous studies have reported that approximately 62 % of the administered dose of [14C]AMPO was recovered in urine in 24 h (with an additional 7% recovered by 7 days) and that approximately 12%, 6%, 5%, and 3% of the administered label of [I4C]AMiv was retained in blood, muscle, feces, and skin, respectively ( 4 , 5 ) . The data reported in this study are in reasonable agreement, with
88 Chem. Res. Toxicol., Vol. 5, No. 1, 1992
approximately 50% of the administered dose identified and quantitated as metabolites of AM. The excretion of AM unchanged in the urine was observed in this study, but not quantitated due to the long relaxation times for alkene carbons and overlap of AM signals with an endogenous compound. However, previous studies have shown that less than 2% of the dose of AM was excreted in the urine of F-344 rats (4). Although a complete characterization of the distribution of AM and its metabolites to tissues and excreta has not been conducted, the quantitative results from this study suggest that there are species differences in the metabolism of AM to glycidamide. The quantitation of metabolites in urine suggests that AM is converted to glycidamide to a greater extent in the mouse (59%) than in the rat (33%). Similar findings were obtained for acrylonitrile (9),where conversion to cyanoethylene oxide was greater in the mouse (80%)than in the rat (60%). However, the percent of AM conversion to glycidamide (at 50 mg/kg PO) is considerably less than found for acrylonitrile conversion to cyanoethylene oxide in mice (10 mg/kg PO) and in rats (30 mg/kg PO). These may be dose-related differences or may reflect differences in the substrate specificity of the enzymes involved. Three non-sulfur-containing metabolites of AM have been previously described but not characterized (4). These metabolites may correspond to the glycidamide, 2,3-dihydroxypropionamide, or the epoxide degradation products assigned in this study. Analysis by IR spectroscopy of a biliary metabolite of AM in rainbow trout indicated the presence of a primary alcohol function in addition to carboxylic acid and amide carbonyl functional groups (18). This metabolite may correspond to the N-acetyl-S-(3amino-2-hydroxy-3-oxopropy1)cysteine or N-acetyl-S-(1carbamoyl-2-hydroxyethy1)cysteinedetected in this study. The exhalation of 14C02from [ 1J4CJAM administered to rats by iv injection comprised 6% of the dose. While no two-carbon metabolites, which might arise after loss of COz, were detected in urine in this study, the further metabolism of 2,3-dihydroxypropionamideto glycerate and hydroxypyruvate could result in the release of COz and production of glycoaldehyde. The excretion of the product in the urine identified as glycidamide was unexpected, given that epoxides are usually highly reactive. However, the stability of this epoxide and its hydrophilicity appear to enable its excretion in urine. In some cases, concentration of the urines resulted in degradation of glycidamide. The analysis of metabolites in the urine by the techniques used in this study afforded the opportunity to monitor readily the effects of manipulations on labile metabolites, which is not usually feasible with other methods of analysis. The metabolites detected in this study appear to be unique to acrylamide, unlike metabolites of acrylonitrile, which can be produced from a number of compounds (9). This suggests that the metabolites observed in this study have the necessary specificity to serve as indicators of exposure to acrylamide. The relative contribution of acrylamide and its epoxide to the various health effects observed for AM is not well understood. Few studies have been carried out on the direct effects of glycidamide. In Salmonella typhimurium (191, mouse lymphoma cells (20),and human lymphob l a s t ~glycidamide ,~ is directly mutagenic, suggesting a possible role in the carcinogenic effects of AM. Modulation of the metabolism of acrylamide by administration of inL. Recio, unpublished results.
Sumner et al.
ducers and inhibitors of cytochrome P-450, and examination of the subsequent effects on neuropathy, has produced disparate results (2). Preliminary studies have suggested that glycidamide is capable of inducing unscheduled DNA synthesis in mouse spermatids (7),and in primary hepatocyte culture." If the toxic and carcinogenic effecb of amylamide are mediated by glycidamide, the data from this study suggest that mice have a greater susceptibility to AM than rats. Evaluation of the potential role of glycidamide in the toxic and carcinogenic effects of AM and examination of the dose, route, and species dependence of metabolism should aid in the assessment of risk to humans on exposure to AM.
Acknowledgment. We thank P. W. Ross, C. U. Parkinson, and T. Shepard for their assistance with animal experiments and C. J. Calleman (University of Washington), G. Csanady, and G. L. Kedderis for helpful discussions. Registry No. AM, 79-06-1; N-AcCys, 616-91-1; metabolite 1, 81690-92-8;metabolite 2,2', 137698-08-9;metabolite 3,3', 137698-09-0; metabolite 4,5694-00-8;metabolite 5,54393-33-8.
References (1) Miller, M. S.,and Spencer, P. S. (1985)The mechanisms of acrylamide axonopathy. Annu. Reo. Pharmacol. Toxicol. 25, 643-666. (2) Dearfield, K. L., Abernathy, C. O., Ottley, M. S., Brantner, J. H., and Hayes, P. F. (1988)Acrylamide, ita metabolism, developmental and reproductive effects, genotoxicity, and carcinogenicity. Mutat. Res. 195,45-77. (3) Edwards, P.(1975)The distribution and metabolism of acrylamide and its neurotoxic analogues in rats. Biochem. Pharmucol. 24, 1277-1282. (4) Miller, M. J., Carter, D. E., and Sipes, I. G. (1982)Pharmacokinetics of acrylamide in Fischer 344 rats. Toxicol. Appl. Pharmacol. 63,36-64. (5) Hashmoto, K., and Aldridge, W. N. (1970)Biochemical studies on acrylamide, a neurotoxic agent. Biochem. Pharmacol. 19, 2591-2604. (6) Bailey, E., Farmer, P. B., Bird, I., Lamb, J. H., and Peal, J. A. (1986)Monitoring exposure to acrylamide by the determination of S-(2-carboxyethyl)cysteinein hydrolyzed hemoglobin by gas chromatography-mass spectrometry. Anal. Biochem. 157, 241-248. (7) Sega, G.A,, Generoso, E., and Brimer, P. A. (1990)Acrylamide exposure induces a delayed unscheduled DNA synthesis in germ cells of male mice that is correlated with the temporal pattern of adduct formation on DNA. Environ. Mol. Mutagen. 16,137-142. (8) Calleman, C. J., Bergmark, E., and Costa, L. G. (1990)Acrylamide is metabolized to glycidamide in the rat: evidence from hemoglobin adduct formation. Chem. Res. Toxicol. 3,406-412. (9) Fennell, T. R., Kedderis, G. L., and Sumner, S. C. J. (1991) Urinary metabolites of [1,2,3-13C]acrylonitrilein the rat and mouse detected by 13Cnuclear magnetic resonance spectroscopy. Chem. Res. Toxicol. 4,678-687. (10) Sumner, S. C. J., MacNeela, J. P., and Fennell, T. R. (1989) Urinary metabolites of [1,2,3-13C]acrylamidedetected by 13Cnuclear magnetic resonance spectroscopy. Toxicologist 10,332. (11) Sumner, S.C. J., Clarke, D. O., Welsch, F., and Fennell, T. R., (1991)Urinary metabolites of 2-methoxyethanol: determined by NMR spectroscopy. Toxicologist 11, 50. (12) Bax, A., Freeman, R., and Frenkiel, T. A. (1981)An NMR technique for tracing out the carbon skeleton of an organic molecule. J. Am. Chem. SOC.103,2102-2104. (13) Muller, L., Kumar, A., and Emst, R. R. (1975)Two-dimensional carbon-13 NMR spectroscopy. J. Chem. Phys. 63, 549Ck5491. (14) Wehrli, F. W., and Wirthlin, T. (1976)Interpretation of carbon-13 NMR spectra, Heyden & Son, New York. (15) Breitmaier, E.,and Voelter, W. (1987)Carbon-13 NMR spectroscopy: high resolution methods and applications in organic chemistry and biochemistry, VCH, New York. (16) Nicholson, J. K.,and Wilson, I. H. (1987)High resolution nuclear magnetic resonance spectroscopy of biological samples as B. E. Butterworth, unpublished results.
Chem. Res. Toxicol. 1992,5,89-94 an aid to drug development. B o g . Drug Res. 31,427-479. (17) Dixit, R.,Seth, P. K., and Mukhtar, H. (1982) Mercapturates from acrylamide. Drug Metab. Dispos. 10,196-197. (18) Petersen, D. W.,and Lech, J. J. (1987) Hepatic effects of acrylamide in rainbow trout. Toxicol. Appl. Phormacol. 89, 249-255.
89
(19) Hashimoto, K.,and Tanij, H. (1985) Mutagenicity of acrylamide and ita analogues in Salmonella typhimurium. Mutat. Res. 158,129-133. (20) Barfknecht, T. R.,Mecca, D. J., and Naismith, R. W. (1988) The genotoxic activity of acrylamide. Enuiron. Mol. Mutagen. 11, Suppl. 11.
Synthesis and Neurotoxicological Evaluation of Putative Metabolites of the Serotonergic Neurotoxin 2-( Methylamino)-1-[3,4-( methylenedioxy)phenyllpropane [(Methylenedioxy)methamphetamine] Zhiyang Zhao and Neal Castagnoli, Jr.* Department of Chemistry, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061
George A. Ricaurte, Thomas Steele, and Mary Martello Department of Neurology, Johns Hopkins University School of Medicine, Francis Scott K e y Medical Center, Baltimore, Maryland 21224 Received August 9,1991
Theoretical considerations and recent experimental data have prompted an investigation of the neurotoxicologikal prooerties of the 6-hydroxydopamine analogue 2-(methylamino)-1(2,4,5trihydroxyphenyl)propane( 5 ) and its possible precursor 1-[2-hydroxy-4,5-(methylenedioxy)phenyl]-2-(methylamino)propane(4), potential metabolites of the serotonergic neurotoxin MDMA. Systemic, intracerebroventricular,and intraparenchymal (intrastriatal and intracortical) administration of 4 led to no detectable alterations of hippocampal or cortical serotonin or striatal dopamine levels in the rat under conditions that caused significant biogenic amine depletions by established neurotoxins. By contrast, intraparenchymal administration of 5 caused profound depletions of dopamine and serotonin, with the former being more severely depleted than the latter. Although not conclusive, these data suggest a possible role for 5 in the mediation of MDMA’s neurotoxic actions.
Introduction The methamphetamine derivative 2-(methylamino)-1[3,4-(methy1enedioxy)phenyllpropane [(methylenedioxy)methamphetamine,MDMAl (1);see Chart I], a commonly abused substance (l),has been promoted by some practitioners as a useful adjuvant psychotherapeutic agent (2). These human exposures are of some concern since studies in rats (3-5)and primates (6)have documented that MDMA is a neurotoxin with selectivity for serotonergic neurons. The molecular events responsible for the neurotoxic effects of MDMA and structurally related amphetaminetype neurotoxins remain unknown (7). Serotonin (5-HT) uptake inhibitors protect against the neurotoxicity of MDMA (8). However, since MDMA itself does not appear to be a substrate of the 5-HT transporter (9),it has been postulated that one or more metabolites of MDMA may mediate the observed neurotoxic effects of the parent drug (10).Consistent with this metabolic theory is the report Abbreviations: MDMA, (methy1enedioxy)methamphetamine [2(methylamino)-l-[3,4-(methylenedioxy)phenyl]propane];5-HT, 5-
hydroxytryptamine (serotonin);6-OHDA, 6-hydroxydopamine [ 2-(2,4,5trihydroxypheny1)ethylaminel;DA, dopamine; ppm, parts per million; TMS, tetramethylsilane; HP, Hewlett Packard; DIPEI, direct insertion probe electron ionization; IR, infrared; UV, ultraviolet; TLC, thin-layer chromatography;THF, tetrahydrofuran; GC/EIMS, gas chromatography/electron ionization mass spectra; ip, intraperitoneal; 5,7-DHT,5,7dihydroxytryptamine; EC, electrochemical;DIPCI, direct insertion probe chemical ionization; EDTA, ethylenediaminetetraaceticacid.
that intraventricular administration of MDMA does not lead to alterations in brain 5-HT levels (11). These considerations have led to a series of metabolic studies in rats (12)and humans (13)designed to evaluate the possible conversion of MDMA to neurotoxic metabolites. Of particular interest is the ability of rat liver and brain enzymes to catalyze the oxidative cleavage of the 3,4-(methylenedioxy)group present in MDMA to generate the corresponding catechol 2 (12).This dopamine analogue is reported to be oxidized rapidly to the corresponding electrophilic o-quinone (3), a potential neurotoxic alkylating agent (14). An alternative bioactivation pathway would involve initial C-2 oxidation of the aromatic ring, a type of biotransformation known to be catalyzed by liver (15)and brain (16)cytochrome P-450 monooxygenases, to yield the species 4. corresponding 2-hydroxy-4,5-(methylenedioxy) Subsequent oxidative cleavage of the methylenedioxy group in the brain would generate the 2,4,5-trihydroxy compound 5, a close structural analogue of the potent neurotoxin “&hydroxydopamine” (6-OHDA, 6) (17). Since the a-methyl analogue 7 of 6-OHDA (18)and the structurally related hydroquinone 8 (19) display neurotoxic propertiea similar to those of 6-OHDA, the a-methyl group present in 6 is unlikely to affect the toxic potential of this system. The neurotoxic properties of the corresponding N-methylated analogue of 6-OHDA have not been reported. In an effort to evaluate the possible toxicological significance of this putative metabolic sequence, we have
0893-228~/92/2705-0089$03.00/00 1992 American Chemical Society
