MAGNETIC RESONANCE I N MEDICINE 20, 100-1 12 (1991)

The Effects of Chronic Ethanol Consumption on the Intact Rat Liver Studied by in Vivo 3'P NMR Spectroscopy

MANFRED BRAUER* AND MINGFU LING

Guelph- Waterloo Center for Gruditate Work in Chemistry, University of Guelph, Guelph, Ontario, Canada NlG 2 Wl

Received April 2, 1990; revised June 13, 1990

In vivo "P nuclear magnetic resonance (NMR) spectroscopy provides unique oppor- tunities to study the biochemistry of an organ within the intact animal in a totally non- invasive way. We have used in vivo and in viiro "P NMR spectroscopy to study steady state changes in the major phosphorus-containing metabolites of the rat liver in control and chronically ethanol-treated rats. Chronic (4 month) ethanol treatment caused a sta- tistically significant increase in the inorganic phosphate and phosphodiester resonances of rat liver in in vivo 3'P NMR spectra relative to pair-fed control rats. Phosphomonoester and adenosine 5'-triphosphate resonances, as well as intracellular pH, were not appreciably altered. The effects of chronic ethanol treatment were particularly apparent in the response of the liver to a metabolic challenge ofglycerol. Glycerol is phosphorylated almost exclusively in the liver and metabolized predominately via glycolysis and gluconeogenesis. Our in vivo 3'P NMR results after administration of glycerol showed a significant increase in the phos- phomonoester resonance in the liver of chronic ethanol-treated rats but not for their pair- fed controls. In viiro "P NMR studies of perchloric acid extracts of liver showed that the increase was due to an accumulation of sn-glycerol 3-phosphate. This effect is due to the NAD+-dependent glycerol 3-phosphate dehydrogenase step being inhibited in the chronic ethanol-treated rats. This glycerol test may be useful in assessing the ability of the liver to rapidly regenerate NAD' in situ and may be a more sensitive indicator of redox imbalance than steady state ratios of redox pairs (e.g., lactate/pyruvate). 0 1991 Academic Press, Inc.

INTRODUCTION

The long-term use of high levels of ethanol in man can lead to a number of changes in the liver, including alcoholic steatosis (fatty liver), alcoholic hepatitis, fibrosis, and cirrhosis ( I ). Ethanol causes a wide range of related biochemical changes, including increased tnacylglycerol levels in the liver, increased ketosis, decreased glycolysis, de- creased citric cycle activity, etc. (2). Many of the acute metabolic effects of ethanol are due to its metabolism by alcohol dehydrogenase (ADH) and subsequently by acetaldehyde dehydrogenase, generating NADH from NAD' . The resulting ethanol- induced increase in NADH/NAD+ ratio shifts the equilibria between many key redox partners toward the more reduced form, for example, pyruvate --* lactate, and dihy- droxyacetone phosphate (DHAP) + glycerol-3-phosphate (G3P). After acute ethanol administration, lactate/ pyruvate ratios are greatly increased, while after chronic ethanol

* To whom correspondence should be addressed at: Department of Chemistry and Biochemistry, University of Guelph, Guelph, Ontario, Canada N IG 2W I .

0740-3194191 $3.00 Copyright 0 1991 by Academic Press. Inc. All nghts of reproduction in any form resewed.

100

CHRONIC ETHANOL CONSUMPTION 101

consumption, these levels return to normal (3 , 4 ) indicating apparent metabolic ad- aptation to the effects of ethanol. Chronic administration of ethanol induces an alternate pathway for ethanol degradation to acetaldehyde based upon the microsomal ethanol oxidizing system (MEOS). This pathway requires 0 2 and consumes (rather than gen- erates) reducing equivalents in the form of NADPH (5 , 6). The effect of ethanol on the redox balance of the hepatocyte is sensitive to the relative contribution of the ADH and MEOS pathways. Other adaptive processes, such as enhanced activity of the Mg2+ -sensitive Na-K-ATPase, increased hepatic oxygen consumption, and in- creased export of hepatic lactate ( 7) may also play a role in the long-term adjustment of hepatic redox balance. Ethanol-induced redox imbalance in the liver plays an im- portant, but incompletely understood, role in overall metabolic changes and in alcoholic liver disease.

In vivo 31P NMR spectroscopy provides biochemical information about the levels of various phosphomonoesters ( PME ), inorganic phosphate (Pi), phosphodiesters (PDE), adenosine 5'-tnphosphate (ATP)' , adenosine 5'-diphosphate (ADP)' , and intracellular pH (from the chemical shift of the Pi resonance) for an organ within the intact body in a totally noninvasive manner (8). 3iP NMR studies of the effects of ethanol on liver metabolism have until recently been limited to the acute effects of ethanol on isolated perfused rat liver or on surgically exposed rat livers. 31P NMR studies of perfused rat liver indicated that acute administration of ethanol exacerbated the acidosis and increase in Pi/ATP ratio induced by hypoxia ( 9 ) . An acute dose of ethanol to a rat with its liver surgically exposed raised the Pi/ATP ratio in a fed rat but lowered the ratio in a fasted rat ( 10) . NMR studies dealing with the chronic effects of ethanol found that administration of ethanol for 1 to 52 days caused a significant decrease in ATP and increase in ADP and Pi, relative to pair-fed controls, when an anoxic stress was administered to the rats ( 11 ). Chronic ethanol treatment also caused a decrease in 3-phosphoglycerate and phosphoethanolamine. However, this study was done on in vitro perchloric acid extracts of rat liver. In vivo 31P NMR studies of surgically exposed rat livers after 6 weeks of ethanol showed only a modest increase in the ADP/ATP ratio for the rat liver in vivo while an acute dose of ethanol had no effect on this ratio (12) . Using a spectroscopic localization method which did not necessitate surgical intervention, in vivo 31P NMR studies of chronic ethanol rats showed that the ratios of PME/ATP and Pi/ATP were increased with alcoholic rats (13, 14) . A "P NMR study of alcoholic liver disease in humans ( 1 5 ) also showed only minor differences in the ratios of ADP/ATP or Pi/ATP in the livers of alcoholic patients relative to normal controls, but showed a 13-509'6 decrease in absolute hepatic concentrations of all the NMR visible phosphorus-containing metabolites.

The major thrust of this study was to use in vivo 31P NMR spectroscopy to study the effects of chronic ethanol on both the resting liver in situ and the ability of the liver to handle a metabolic challenge. While steady state spectra indicate metabolic changes within the resting organ, the response of the liver to a metabolic challenge, the administration of glycerol, is a more sensitive indicator of tissue perturbation and damage. This metabolic challenge by glycerol requires the liver to rapidly phosphorylate

' Other purine and pyrimidine nucleotide triphosphates and diphosphates also contribute to the ATP and ADP resonances.

102 BRAUER AND LING

the glycerol to sn-glycerol-3-phosphate (G3P), which is mainly oxidized via an NAD+- requiring step to DHAP and enters glycolysis or gluconeogenesis (see Discussion):

ATP ADP NAD' NADH

Glycerol G3P DHAP - Glycolysis and Gluconeogenesis.

In a parallel situation, the steady state 3'P NMR spectra of skeletal muscle of many patients look normal, but obvious abnormalities are exposed during periods of muscle contraction ( 8 ) . In vivo 31P NMR spectra in this study were acquired such that absolute concentrations of phosphorus-containing metabolites could be determined, although by a different method than that of Meyerhoff et al. ( 1 5 ) .

METHODS AND MATERIALS

Male Wistar rats initially weighing 200-2 10 g (Charles River Lab, Inc.) were divided randomly into control or ethanol-treated groups and were pair-fed. Both groups were given a nutritionally adequate liquid diet, with ethanol substituted for carbohydrate to account for 36% of total calories for the treatment groups (16). The feeding protocol was generally introduced gradually, i.e., 9% ethanol for 3 days, then 18% for 3 days, then 27% for 3 days and finally 36%. All animals were on the 36% ethanol diet for at least 3 weeks before NMR examination. The animals were fasted for 3 h before NMR examination and were anesthetized with pentobarbital.

In vitro 3'P NMR spectra of perchloric acid extracts of rat livers were obtained after the rats had been anesthetized with pentobarbital, the livers freeze clamped and ex- tracted with 6% perchloric acid (PCA). Extracts were analyzed by conventional 31P NMR spectroscopy on a Bruker WH 400 NMR spectrometer, and were analyzed using our own previously published methods and resonance assignments (11). For quantitative analysis, 31P NMR spectra from rat liver PCA extracts were acquired with a 2 M methylenediphosphonic acid capillary concentrically mounted on a vortex plug in a 10 mm NMR tube. Spectral peaks were integrated using the integration program in the WH 400 spectrometer. The areas were then converted into concen- trations by comparison to the areas of standard solutions of ATP, AMP, and Pi. All chemical shifts are given relative to 85% phosphoric acid as an external standard.

In vivo 31P NMR spectra of the steady state livers within the intact rat were obtained using a Spectroscopy Imaging System 85 MHz NMR spectrometer with 3 1-cm hori- zontal bore magnet. In addition to the localization achieved by the simple two-turn 2.0 cm (0.d.) surface coil itself, two localization strategies were implemented to improve the volume selection. First DEPTH pulses [ 8 / 5 , 8 / 3 , 8, 281 ( 1 7 ) were used to narrow the zone of excitation and eliminate high flux regions. Secondly, the NMR signals from skin and muscle covering the liver were eliminated using an immobilized ferrite screen which introduces Bo field inhomogeneity in the region of the skin (14, 18). An external capillary of methylene diphosphonate mounted 1.5 cm above the center of the surface coil was used as a signal intensity, and radio frequency pulse strength standard. T I effects were minimized by acquiring spectra with a repetition time of 5 s. Control experiments, showed that all resonances were over 90% of their fully relaxed intensities at a repetition time of 5 s relative to a repetition time of 10 s. 256 scans were taken for each spectrum, requiring 2 1.3 min. Signal intensities were

CHRONIC ETHANOL CONSUMPTION 103

calibrated against standard solutions of 30 m M NaCl, pH 7.4, and 10 m M Pi and ATP. All spectra were acquired with the same BI field strength by calibrating the methylene diphosphonate resonance for maximal intensity (90" pulse) to obtain re- producible spectra from the same selected volume within the sample. All spectra were normalized relative to the area of the methylene diphosphonate resonance to control for variations in magnetic field homogeneity and surface coil sensitivity between ex- periments. Thus, in vivo spectra from one animal to the next could be obtained re- producibly and quantitatively.

Fifty percent glycerol (0.8 ml/ kg) was given intrapentoneally to anesthetized rats after a control spectrum had been taken. Spectra were taken over the next 3 h after glycerol administration. In order to maximize the time resolution of the glycerol effect, signal to noise had to be maximized at the cost of spacial selectivity and quantitation. Hence, glycerol time-course studies were done without the femte screen and DEPTH pulses and with a faster (2.5 s ) repetition time.

Statistical analysis was done using a two-way analysis of variance to partition variance due to ethanol treatment from variance between pairs. A software program (Canadian Academic Technology, Inc.) was used. The experimental protocol followed the guide- line of the Canadian Council on Animal Care and was approved by the local Animal Care Committee.

RESULTS

I n vitro 31P NMR spectra of PCA extracts for a pair-fed control and chronic ethanol- treated rat liver after euthanization with pentobarbital are shown in Figs. 1A and 1 B,

10 1-10 16

D i

5 O -5 -10 -15 -20 5 4 3 2 6 (PPM)

FIG. I . In vitro 3'P NMR spectra of PCA extracts of ( A ) pair-fed control and ( B ) chronic ethanol-treated rat liver. The downfield region is expanded for (C) control and (D) ethanol-treated rats. Resonances of particular interest are G3P (peak 2), +4.21 ppm; 3-phosphoglycerate (peak 3), +3.99 ppm; adenosine monophosphate (peak 5 ) , +3.70 ppm; phosphocholine (peak 8). +3.34 ppm; Pi (peak lo), +2.38 ppm; ADP, doublets 14 and 15 at -6.02 and -10.33 ppm; and ATP, doublets 13 and 16 at -5.72 and -10.79 QQW apparent triplet 19 at -21.13 ppm

104 BRAUER AND LING

respectively. The results for three pairs of rats (mean i standard error) are summarized in Table 1. The ratios of Pi/ATP and ADP/ATP were 1.37 and 0.41 for control rats and 2.1 1 and 0.56 for ethanol-treated rats, respectively. Thus the Pi/ATP and ADP/ ATP ratios were 1.6 fold and 1.4 fold higher for the ethanol-treated rats relative to the controls. Substantial changes were also seen in the PME region (Figs. 1C and ID), with ethanol-treated PCA extracts exhibiting increased G3P (peak 2), adenosine monophosphate (peak 5) and phosphocholine (peak 8 ) and decreased 3-phospho- glycerate (peak 3). The assignment of 31P resonances in PCA extracts was described previously ( f I ).

In vivo 31P NMR spectra of pair-fed control and chronic ethanol-treated rat liver in situ are shown in Figs. 2A and 2B, respectively. The two spectra are overlayed in Fig. 2C for easy comparison. An ethanol-induced increase in the PDE and Pi resonances and a slight decrease in the ATP is apparent in this pair. The presence of a very small PCr resonance at +2.3 ppm indicates that the in vivo 31P resonances came almost exclusively from the liver. Less than 8% of the spectrum was determined to originate from extrahepatic skeletal or smooth muscle tissue assuming an in vivo concentration for PCr of 16.9 m M in skeletal muscle ( 1 9 ) . The localization of these liver spectra, using the DEPTH pulses in conjunction with the ferrite strips, is particularly good as

TABLE 1

Results from in Vitro 3'P NMR Spectra of Long-Term Liquid Diet Rats

Control" Ethanol" (pnol/mg wet wt) (pmol/mg wet wt) P b

Glycerol 3-phosphate 0.151 f 0.021 0.949 f 0.16 0.0072 3-Phosphogl ycerate 0.141 f 0.015 0.0353 f 0.0046 0.0024 Phosphoethanolamine 0.427 f 0.089 0.306 ?c 0.023 0.259 (N.S.)e Adenosine monophosphate 0.276 f 0.0 15 0.501 f 0.040 0.0064 Phosphocholine 0.348 f 0.063 1.08 f 0 . 1 2 0.0054 Pi 2.875 f 0.233 3.76 f 0.15 0.033 Glycerophosphoethanolamine 1.029 f 0.083 0.487 f 0.088 0.0057 Glycerophosphocholine 1.163 f 0.127 0.658 f 0.13 0.047 ATP(7-P) 2.103 f 0.189 1.778 f 0.019 0.161 (N.S.)' ADP(&P) 0.859 f 0.055 0.996 f 0.025 0.086 (N.S.)' ADP(a-P) 0.716 f 0.056 0.905 f 0.015 0.080 (N.S.)' ATP(a-P) 2.224 f 0.092 1.963 f 0.045 0.070 (N.S.)e Diphosphosphodiesters ( I ) 3.126 f 0.155 3.237 f 0.043 0.528 (N.S.)e Diphosphosphodiesters (2) 1.238 f 0.064 1.004 f 0.112 0.144 (N.S.)e ATP(P) 2.162f 0.160 1.681 f 0.0326 0.040 Total adenylate pool 3.234 f 0. I26 3.276 f 0.046 0.803 (N.S.)' Energy charge' 0.781 f 0.019 0.695 f 0.01 1 0.019 Phosphorylation potentiald 0.900 f 0.210 0.521 f 0.0123 0.146 (N.S.)'

a Mean f standard error ( n = 3). ' Probability-2 way analysis of variance.

dmM-'. ' N.S., not significant (P > 0.05).

No unit.

CHRONIC ETHANOL CONSUMPTION 105

10 0 -10 -20 6 (PPM)

FIG. 2. In vivo 3'P NMR spectra of (A) pair-fed control, ( B ) chronic ethanol-treated rat liver, and (C) the two spectra overlayed. See Methods and Materials for details.

the spectra were acquired with a 5-s repetition time, which does not diminish the signal of the PCr (long T I ) relative to other resonances with shorter T , values.

The in vivo spectra for both ethanol-treated and control rat livers were analyzed by spectral simulation of the resonances as the sum of Lorentzian line-shapes (Fig. 3) . The areas of the resonances were normalized relative to that of the methylene di- phosphonate resonance (at N +2 1 ppm, not shown). The results for four pairs of rats are shown in Table 2. The mean values f standard error for each resonance are indicated. Chronic ethanol treatment resulted in a statistically significant increase in the Pi and PDE resonances ( P < 0.05 ). No statistically significant change in the PME or ATP resonances was obtained. The a resonance of ATP also contains resonances for the CY phosphate of ADP, and the phosphates of NAD', NADH, NADP+ and NADPH and is hence more intense than the y or 6 resonances of ATP. Concentrations of ADP can be determined by the difference between the intensity of the ATP, y resonance (y phosphate of ATP plus 6 phosphate of ADP) and the intensity of the ATP resonance ( p phosphate of ATP only). Ratios of ADP/ATP were less than 0.15 for both control and ethanol-treated rats and were not significantly different. No

106 BRAUER AND LING

-20 10 0 -10 -20 10 0 -1 0

6 (PPM)

FIG. 3. I n vivo3'PNMR spectraof(A)pair-fedcontrol. and (B)chronicethanol-treated rat liver. Simulated spectra (middle) correspond closely with the experimental spectra ( t o p ) . The individual Lorentzian line-shapes contributing t o the final simulated spectra are indicated in the bottom spectra. In vivo "P NMR spectra were assumed to be due to a PME, Pi, PDE, residual PCr, and ATP 7, a , and p resonances.

significant change in pH between control and ethanol-treated rat liver was observed (P > 0.05).

The most dramatic effects of chronic ethanol treatment were apparent when glycerol was added as an intraperitoneal injection. The in vivo 3'P NMR spectra of pair-fed control rats did not show significant time-dependent changes after glycerol adminis- tration (Fig. 4B), while the chronic ethanol-treated rats showed a dramatic increase

TABLE 2

Results from Spectral Simulation of in I,.ivo 3 'P NMR Spectra of Long-Term Liquid Diet Rats

Control" Ethanol" (pmol/mg wet wt) (pmol/mg wet wt) P b

PME Pi PDE PCr ATP, Y ATP, tc ATP, P

PME/ATP,O Pi/ATP, 6

PH

1.09 f 0.21 0.84 f 0.17 5.40 rt_ 1.65 1.27 t 0.42 3.24 * 0.43 5.73 f 0.72 2.88 k 0.39 7.26 f 0.15 0.390 f 0.106 0.325 f 0.075

1.43 f 0.19 2.09 f 0.13 8.64 f 1.09 1.31 * 0.40 2.60 f 0.45 4.71 ? 0.71 2.34 -1-0.19 7.15 f 0.21 1.143 k0.312 0.903 -C 0.087

0.318 (N.S.)' 0.032 0.036 0.910 (N.S.)' 0.383 (N.S.)' 0.527 (N.S.)' 0.333 (N.S.)' 0.687 (N.S.)' 0.139 (N.S.)' 0.008

a Mean -+ standard error ( n = 4). * Probability-2 way analysis of variance.

N.S.. not significant ( P > 0.05).

CHRONIC ETHANOL CONSUMPTION 107

dA TIME (MIN.)

26 'k 104p 208

6 4 2 0

B

P P 6 4 2 0

6(PPM)

FIG. 4. I n vivo "P NMR spectra (downfield region) of ( A ) chronic ethanol-treated and (B) pair fed control rat liver immediately before and at various times after an intraperitoneal injection of 50% glycerol (0.8 ml/kg).

in the PME resonance starting 20 min after injection of glycerol (Fig. 4A). The PME resonance returned to normal after about 3 h. The increase in the in vivu PME resonance is consistent with the accumulation of G3P, since it is the downfield portion of the PME resonance which increased. The pH was constant throughout the time course, as judged by the fixed position of the Pi resonance. The results for four pairs of rats, including pair-fed control and ethanol-treated rats, are shown in Fig. 5. The means * standard error for each time period are indicated. The resonances for Pi and ATP did not change significantly with time for either ethanol-treated or control rats, in- dicating that the amount of glycerol given did not exceed the ability of the livers in either group to regenerate ATP following the glycerol kinase reaction.

To confirm the assignment of the PME as being predominately G3P, the chronic ethanol-treated rats were given the same dose of glycerol and were sacrificed after 140 min, and the livers were quickly freeze-clamped, as before. In vitru 31P NMR spectra of the perchloric acid extracts showed a >20-fold increase in the G3P resonance for the ethanol-treated rats (Fig. 6B) relative to control rats (Fig. 6 A ) . In vitru 13C NMR spectra of the same extracts showed a threefold increase in the glycogen C1 resonance (103 ppm from TMS) in control rat livers relative to ethanol-treated rat livers (data not shown).

108 BRAUER AND LING

- 175.0

0

e 150.0

0 125.0

100.0

5 75.0

L 0

v

5 0 . 0 5

a, 2 50.0 0 40 80 120 160 200

t ime (min)

FIG. 5 . Changes in the in vivo PME, Pi, and ATPp resonances of pair-fed control (0) and chronic ethanol- treated (e) rat livers with time after glycero1 injection. Each data point represents the mean & standard error for four rats. Changes are expressed as percentages relative to the spectra immediately before glycerol injection.

DISCUSSION

In vitro 31P NMR spectra of chronic ethanol-treated versus control rats indicated an increase in several PME resonances, the most predominant being G3P. This may be due to a redox-balance-induced shift in the equilibrium between G3P and DHAP. Other bioenergetic parameters, such as ADP/ATP and Pi/ATP ratios were increased 1.4- and 1.6-fold, respectively, by chronic ethanol administration. Our in vitro bio- energetics results are in general agreement with the in vitro results of Helzberg et al. (12) who found a 2.1- to 2.4-fold and 1.8-fold increase in ADP/ATP and Pi/ATP, respectively, with chronic ethanol consumption in rats.

In vivo 31P NMR spectra of steady state livers of chronic ethanol-treated versus control rats showed some statistically significant changes (Table 2). The in vivo 31P resonance for PME was not significantly higher for the ethanol-treated rats, despite increased G3P seen in the in vitro results. The Pi levels were significantly higher for the ethanol-treated rats, indicating that deleterious bioenergetic changes were in fact occurring. The decrease in ATP for ethanol-treated rats was not statistically significant, however. A number of mechanisms for bioenergetic deterioration have been proposed including increased rates of ATP hydrolysis (the "hypermetabolic state"), decreased

CHRONIC ETHANOL CONSUMPTION 109

A

l,a, FIG. 6. In vitro 31P NMR spectra of PCA extracts of (A) pair-fed control and ( B ) chronic ethanol-treated

rat livers 140 min after glycerol injection. A > 20-fold increase in G3P for ethanol-treated rats is readily apparent.

rates of ATP synthesis (due to inhibited adenine translocase or ATP synthetase activity, or partial uncoupling of oxidative phosphorylation ), and increased liver oxygen uti- lization with functional centrilobular hepatic hypoxia ( 1, 2, 11, 14).

The increase in the PDE resonance in the ethanol-treated rats could be due to increased phospholipid breakdown, since the two major metabolites comprising this resonance, glycerophosphocholine and glycerophosphoethanolamine, are major phos- pholipid breakdown products. However, a decrease in glycerophosphocholine and glycerophosphoethanolamine with ethanol treatment was observed in vitro (see Table 1 ). The nature of the PDE resonance has been brought into question by a recent study indicating that the line width of the PDE resonance increased substantially with mag- netic field strength (20). The PDE resonance was hence postulated to be due largely to immobilized phospholipids in the biomembranes. Hence, our findings of a chronic ethanol-induced increase in PDE may be interpreted as an alteration in the fluidity, structure, or amount of membrane in the liver. It is known that ethanol affects the structure/dynamics of biological membranes, increasing their fluidity, for example. Chronic ethanol also induces increased endoplasmic reticular membrane formation ( I , 2). A third interpretation of our findings is that the increase in the PDE resonance is due to an accumulation of lipoprotein within the ethanol-treated liver. The partially immobilized phospholipids within, for example, the very low density lipoproteins (VLDL) (30-70 nm diameter), would have a 3'P chemical-shift and resonance line-

110 BRAUER AND LING

width compatible with the observed PDE resonance in our studies. Our linewidths for the PDE resonance, after correcting for line-broadening and the chemical-shift differ- ence between glycerophosphocholine and glycerophosphoethanolamine, were about 40 Hz broader than the linewidths of the Pi or PCr resonances. The accumulation of VLDL and other lipoproteins within the ethanol-treated liver is a well-known obser- vation ( 1 , 2) and is an important causal factor in fatty infiltration. It should be noted that our studies were done at a magnetic field strength of 2 T and with a long repetition time where the PDE resonance is particularly narrow and readily observable.

Our in vivo results generally agree with those of other groups with some notable exceptions. Our ratios of Pi/ATP and PME/ATP, p were 0.325 and 0.390, respectively, for control rat livers, in good agreement with the results of Iles et al. (21 ) (Pi/ATP, p 0.33 to 0.36) and the results of Meyerhoff et al. ( 1 5 ) (PME/ATP, p of 0.4 5 0.2). Chronic ethanol caused an increase in Pi/ATP, p (P = 0.008) and a slight increase in PME/ATP, p (P = 0.139) in general agreement with Takahashi, et al. (14). However our results were normalized relative to an external intensity and excitation pulse width standard so that intensities of individual resonances could be assessed rather than simply ratios of intensities. Thus, we found a significant increase in Pi but only a minor increase in PME and decrease in ATP, p. Our observed increase in PDE was not detected by Takahashi et al. ( 14). This is probably because at their magnetic field strength of 4.7 T, the PDE is much broader than at 2 T. Also their repetition time of 2 s is not optimal for this long T,-relaxing PDE resonance. Meyerhoff et aI. (IS) also obtained quantitative 31P spectra of ethanol-treated livers from human patients. They found no statistically significantly changes in Pi/ATP, p, PME/ATP, p, or PDE/ATP, p ratios but a uniform decrease in the concentrations of all the 31P resonances. Their results are most likely due to a “dilution effect” of actively metabolizing liver tissue with fibrotic tissue or extensive fatty infiltration. In addition, it is presumed that al- coholic patients in hospital would be abstaining (at least temporarily) from ethanol, allowing the metabolic effect of ethanol to revert back to normal.

Glycerol is an ideal test compound for liver function because it can be administered enterally or parenterally, and it is rapidly absorbed and transported to the liver. Glycerol is phosphorylated primarily in the liver to G3P catalyzed by glycerol kinase. This reaction requires ATP. The resulting ADP and Pi are used to regenerate ATP via oxidative phosphorylation or substrate-level phosphorylation. G3P levels within the liver will increase if the glycerol kinase reaction is much faster than those pathways which utilize G3P, i.e., triglyceride and phospholipid biosynthesis, via phosphatidate, and glycolysis and gluconeogenesis, via DHAP. The glycerol kinase reaction is known to be fast and not the rate-limiting step in glycerol-utilizing pathways (22, 23). While triglyceride biosynthesis is important in the etiology of fatty infiltration, the oxidation of G3P to DHAP and subsequent entry into gluconeogenesis or glycolysis is the primary route by which most of the G3P is metabolized (24, 2 5 ) . NAD’ is consumed in the first step of this sequence, and it must be regenerated, mainly via the electron trans- port chain.

In vivo 31P NMR spectroscopy allows us to monitor changes in the concentrations of a number of these phosphorus-containing metabolites without cutting the animal open. 3’P resonances for ATP and Pi were easily observable in the NMR spectrum of the rat liver (see Fig. 2). The fall in the hepatic ATP pool and a rise in the Pi pool

CHRONIC ETHANOL CONSUMPTION 111

would have been expected if the glycerokinase reaction had exceeded the liver's ability to rapidly regenerate ATP. This was not observed in either the ethanol-treated or the control rats. G3P is the major phosphorylated compound which makes up the down- field part of the PME resonance. The presumption that the increase in the downfield part of the PME resonance for ethanol-treated rats after administration of glycerol was due to the accumulation of G3P was confirmed by parallel in vitro 31P NMR experiments (Fig. 6). It is readily apparent that the conversion of G3P to DHAP is rate-limiting for the chronic ethanol-treated rats.

The oxidation of G3P to DHAP is catalyzed by glycerol-3-phosphate dehydrogenase (G3PDeH), and, like ethanol, this oxidation occurs in the cytoplasm and requires NAD' as the electron acceptor. (A mitochondria1 G3PDeH also exists, but its activity in the liver is low.) In the ethanol-treated rat, the G3PDeH enzyme must compete with ADH for available NAD'. A shortage of available NAD+ may inhibit the G3PDeH reaction and make it rate-limiting in G3P utilization. An ethanol-induced inhibition of the G3PDeH reaction has in fact been reported (26 ) . Thus, it is reasonable to propose that the rate of disappearance of G3P in ethanol-treated rats may be limited by the rate of the G3PDeH reaction and the liver's ability to regenerate NAD'. It should be noted that this presumably NAD+-limiting glycerol effect is observed after a period of metabolic adaptation to chronic ethanol treatment where lactate / pyruvate ratios have returned to normal ( 5 , 6 ) . The administration of a large amount of glycerol requires the liver to generate an unusually large amount of NAD' quickly and may reveal chronic ethanol-induced deterioration in hepatic oxygen availability as the final electron acceptor for the electron transport chain. This may be due to competition for oxygen by the MEOS system, decreased flow of blood through the hepatic sinusoids, decreased ability of the liver to rapidly export reduced metabolites such as lactate (from pyruvate) , or possibly inhibited electron transport activity due to mitochondrial damage. We have found less available lactate in ethanol-treated rat livers than in controls (data not shown). This glycerol test may hence be a more sensitive indicator of redox imbalance than lactate to pyruvate ratios. Glycerol is a virtually nontoxic compound and a normal constituent of many foods. Fructose, at a comparable dosage to our studies, has been given to human subjects and its phosphorylation to fructose- 6-phosphate has been studied as a clinical tool by in vivo 31P NMR ( 2 7 ) . This glycerol test in conjunction with noninvasive in vivo 31P NMR spectroscopy may hence be useful as a safe clinical diagnostic tool to assess the ability of the liver to regenerate NAD' in situ, in a clinical setting for alcoholic patients.

ACKNOWLEDGMENTS

Funding for this study was provided by the Natural Science and Engineering Research Council, the University of Guelph Research Board, and the University of Guelph MRI Facility. This is paper number 14 from the U. of G. MRI Facility.

REFERENCES

1. V. D. D E S M E T , A ~ ~ ~ M ~ ~ . Scand. Suppl. 703, 1 1 1 (1987). 2. C. S. LIEBER, Acta Med. Scand. Suppl. 703, 1 1 ( 1987). 3. J . ALDERMAN, T. TAKAGI, AND C. S. LIEBER, J. Biol. Chem. 262,7497 ( 1987). 4. H. NUTENIN, K. 0. LINDROS, AND M. SALASPURO, Alcoholism: Clincial Exp. Rex 1, 163 ( 1983). 5. s. DOMSCHKE, w. DOMSCHKE, AND c. s. LIEBER, Life SCi. 15, 1327 (1974).

112 BRAUER AND LING

6. M. J . SAVOLAINEN, E. BARAONA, P. PIKKARAINEN, AND C. s. LIEBER, Hepatology 1, 33 ( 1981 ). 7. Y. ISREAL AND H. ORREGO, Ann. N.Y. Acad. Sci. 492, 303 (1987). 8. S. M. COHEN. Ann. N. Y. Acad. Sci. 508, 1 ( 1987). 9. F. DESMO LIN, P. CANIONI, C. CROTTE, A. GEROLAMI, AND P. J. COZZONE, Hepatology 7 , 3 15 ( 1987 ).

10. C. CUNNINGHAM, C. MALLOY, AND G. RADDA, Biochem. Biophys. Actu 885, 12 ( 1986). 11. MINGFU LING AND M. BRAUER, Biochim. Biophys. Aria 1051, 151 (1990). 12. J. H. HELZBERG, M. S. BROWN, D. J. SMITH, J. C. GORE, AND E. R. GORDON, Hepatology 7, 83

13. M. BRAUER AND M. LING, “Proceedings, 32nd Annual Meeting of the Canadian Federation of Biological

14. H. TAKAHASHI, Y. GEOFFRION, K. W. BUTLER, A N D S. W. FRENCH, Hepatology 11,65 ( 1990). IS. D. J. MEYERHOFF, M. D. BOSKA, A. M. THOMAS, AND M. W. WEINER, Radiology 173, 393 (1989). 16. C. S. LIEBER AND L. M. DECARLI, Alcoholism: Clin. Exp. Rex 6, 523 (1982). 17. M. R. BENDALL, in “Biomedical Magnetic Resonance” (T. L. James and A. R. Margulis, Eds.), pp.

99- 134, Radiol. Research and Education Foundation, San Francisco, 1987. 18. J. GEOFFRION, M. RYDZY, J. K. SAUNDERS, I. C. P. SMITH, AND H. D. JARRELL, in “Soc. Magn.

Reson. in Medicine, Sixth Annual Conference, New York,” p. 958, 1987. 19. R. BUCHLI AND P. BOESIGER, in “SOC. Magn. Reson. in Medicine, Eighth Annual Conference, Am-

sterdam.” p. 598, 1989. 20. E. J . MURPHY, B. RAJAGOPALAN, K. M. BRINDLE, AND G. K. RADDA, Magn. Reson. Med. 12, 282

(1989). 21. R. A. ILES, A. N. STEVENS, J. R. GRIFFITHS, AND P. G. MORRIS, Biocliem. J. 229, 141 (1985). 22. L. SESTOFT AND P. FLERON, Biochim. Biophys. Actu 375,462 ( 1975). 23. C. VIND. A. HUNDING, AND N. GRUNNET, Biochim. Biophys. Actu 375, 462 ( 1975). 24. C . VIND, A. HUNDING, AND N. GRUNNET, Biochem. J. 243,625 (1987). 25. H. F. WOODS AND H. A. KREBS, Biochem. J. 132,55 (1973).

27. R. D. OBERHAENSLI, G. J. GALLOWAY, D. J. TAYLOR, P. J. BORE, B. RAJAGOPALAN, AND G. K.

(1987).

Societies. Calgary, p. 97, June 1989.”

26. T. CRONHOLM A N D T. CURSTEDT, Biochem. J. 224,731 ( 1984).

RADDA, A4a~n. Re.von. Imaging 4 , 4 I3 ( 1986).