Proton magnetic resonance imaging and phosphorus-31 magnetic resonance spectroscopy studies of bromobenzene-induced liver damage in the rat

Magnerrc Resonance Imaging;-Vol. IO, pp. 257-261. 1992 0730-725X/92 $5.00 + .oO

Primed in the USA. All rights reserved. CopyrIght 0 1992 Pergamon Press Ltd.

l Original Contribution

PROTON MAGNETIC RESONANCE IMAGING AND PHOSPHORUS-31 MAGNETIC RESONANCE SPECTROSCOPY STUDIES OF

BROMOBENZENE-INDUCED LIVER DAMAGE IN THE RAT

MANFRED BRAUER AND STEVEN LOCKE

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

Respiratory-gated proton magnetic resonance imaging was used to study the response of the rat liver in situ to bromobenzene, a classic hepatotoxicant. A localized region of high proton signal intensity in the perihilar region of the liver was seen 24-48 hr after an intraperitoneal injection of bromobenzene. Localized proton magnetic resonance spectra from within this region indicated that the increased proton signal intensity was not due to accumulation of fat in the liver, but primarily due to a longer T2 for the proton resonance of water. This is consistent with acute edema in this localized region. In vivo 31P magnetic resonance spectroscopy studies of the same rat livers in situ were performed. Spectroscopic conditions were determined whereby localized, quantitative 3’P spectra could be obtained. Using these methods, 10 mmol/kg bromobenzene was found after 24 hr to cause a number of statistically significant (p < 0.05) effects: a decrease in adenosine S-triphosphate levels from 4.1 -t 0.5 to 3.0 f 0.5 mM, a decrease in phosphodiester levels from 11.3 f 0.9 to 9.3 f 0.7 mM and an increase in the phos- phomonoesters from 3.0 -t 0.4 to 5.5 + 1.2 mM (mean f standard deviation). High resolution in vitro 3’P spectra of perchloric acid extracts of these rat livers showed that the increased phosphomonoester resonance was due to a selective 4.3-fold increase in phosphocholine. Thus, our in vivo and in vitro 31P magnetic resonance spectra are consistent with the hypothesis that a phosphatidylcholine-specific phospholipase C (generating phosphocholine and diacylglycerol) is activated during tissue damage. Both the imaging and spectrocopy results obtained with bromobenzene closely resemble CC&induced liver changes previously reported, and may reflect a generalized response of the liver to any acutely acting toxic chemical.

Keywords: MRI; MRS; Hepatotoxicity; Bromobenzene; Edema.

INTRODUCTION

Magnetic resonance imaging (MRI) has gained univer- sal acceptance as the preferred clinical diagnostic method for imaging of CNS, spine, and many lower abdominal and musculoskeletal disorders. ’ It has also gained wide acceptance as a valuable imaging modal- ity for the liver.* The use of MRI as a research tool in the study of chemically induced liver toxicity has been very limited, and has centered mainly on the hepatotoxicity of carbon tetrachloride (CC&) as a model of acute liver necrosis. CC& is metabolized in the liver to a variety of free radical intermediates, which induce peroxidative damage to biomembranes. Stark and oth- ers studied the effects of ethionine, iron and CC& on

the livers of rats.3-6 They found that the intensities of the proton resonances increased with CC&, as did both T, and T2 values of the proton resonances within the liver. Brauer et al.’ found a similar increase in proton signal intensity after CC& treatment and, using localized proton magnetic resonance spectroscopy (MRS), found that this was largely due to an increase in the T2 relaxation time of the water protons rather than the accumulation of fat in the liver. Using 23Na MRI techniques, an increase in 23Na signal intensity in the liver was also found after CC& treatment, under identical experimental conditions as the ‘H MRI stud-

ies . * This increase in signal intensity was due to an increase in sodium ion concentration in the damaged liver region, rather than a significant change in 23Na

RECEIVED 4/11/91; ACCEPTED 7/11/91. Funding for this study was provided by the Natural Sci-

ences and Engineering Research Council and the University of Guelph MRI Facility.

Address correspondence to Manfred Brauer, Department of Chemistry and Biochemistry, University of Guelph, On- tario, Canada NlG 2Wl.

257

258 Magnetic Resonance Imaging 0 Volume 10, Number 2, 1992

T, or T2.* In vivo 31P MRS showed that CCL, induced a marked acidosis in the liver and an increase in phosphocholine.’ It is not known whether these CCL induced changes observed by MRI/MRS are generalized hepatic responses seen with any hepatotoxicant, or if the liver responds differently to different types of toxic compounds.

Bromobenzene (BB) is a classic indirect-acting hepatotoxicant which is metabolized in the liver to one or more chemically reactive, highly toxic metabolites. Unlike CC14, these metabolites are not free radical species. Other well-known hepatotoxicants, such as acetaminophen, furosemide, isoniazid, dimethylnitro- samine or benz(a)pyrene lo.1 ’ have similar mechanisms of toxic action to BB. The mechanism of BB’s toxic action involves metabolism by the cytochrome P450 system in the endoplasmic reticulum to the 3,4-epoxide, which is a potent electrophilic arylating agent. This chemically reactive BB intermediate binds with cysteine residues of glutathione, as well as enzymati- tally essential cysteines on Mg2+-dependent Na+-K+ ATPase Ca2+ ATPase and other crucial enzymes,” presumably inducing toxicity. An additional potentially important aspect of BB toxicity is mediated through oxidative stress and free radical damage. I2 This is due to the presence of highly reactive 0,. and H202 generated by: (a) the cytochrome P450 biotransformation step, (b) redox cycling of one or more of the secondary BB metabolites (ex. 4-bromocatechol) or (c) constitutive aerobic metabolism exacerbated by BB-induced depletion of glutathione. ‘I The targets of these highly reactive species include essential intracellular macromolecules and unsaturated fatty acyl moieties of the phospholipids in the membranes. Free radical attack of this latter group can initiate lipid peroxidation (as occurs with CCL,) leading to damaged and leaky membranes. Covalent attack of Na+-K+ ATPase or Ca2+ ATPase also affects membrane function by inhibiting these transport proteins which main- tain normal Na+, K+ and Ca2+ gradients. Toxicity due to both free-radical and covalent binding mechanisms of reactive intermediates has also been postulated for acetaminophen and furosemide. BB was selected as a model compound because its toxicity in the rat is reproducible, unlike the more idiosyncratic toxicity of acetaminophen, furosemide and isoniazid.

The purpose of this present study was to determine whether acute sublethal exposure to BB produces MRI/ MRS-observable hepatic changes similar to those observed with CC&. Since MRI/MRS is a noninvasive probe, we could determine which toxic processes were important and readily observable in situ, as opposed to those processes which are only significant in tissue culture or subcellular systems.

METHODS AND MATERIAL

BB and Myrj 45 were obtained from Sigma Chem- ical Co. Male Wistar rats (250-350 gm) were starved for 12 hr, and given 10 mmol BB/kg body weight, in- traperitoneally (i.p.) as a 40% aqueous emulsion in 2% Myrj 45. The LDsO for BB is 25 mmol BB/kg i.p. and 17 mmol BB/kg orally (P.o.).“*‘~ Before MRI examination the rats were anesthetized with sodium pentobarbital (Somnotol 1.0-I .5 ml/kg body weight, i.p.). For most of the animals MRI and 31P MRS data were collected sequentially on the same animal. Generally MRS was done first; then the rat was supplemented with pentobarbital and MRI data was acquired. No MRI/MRS data was obtained after 4 hr of anesthesia to avoid possible metabolic effects of long-term

anesthesia. All the MRI/MRS experiments described in this

study were performed on a Spectroscopy Imaging Sys- tem (SIS) 85/310 spectrometer at a magnetic field strength of 2.0 T and a 31-cm horizontal-bore magnet. The rats were imaged using respiratory gating such that each inspiration was used to trigger acquisition of phase-encoding steps in the spin-warp imaging sequence.’ Multiple slices were taken in the transverse plane using an echo time (TE) of 20 msec. The image slices were 3 mm thick with a field-of-view of 5 x 7 cm. 256 phase-encoding steps were taken with two ac- quisitions per step, and 512 frequency-encoding points were acquired. This yielded a 5 12 x 5 12 pixel image after zero-filling in the phase-encoding direction. Vi- als of 2 mM CuS04 in water, and corn oil were used as absolute intensity standards, so that proton signal intensities could be compared from one experiment to the next. Gaussian filtering of the spin echo signal was used to improve the signal-to-noise ratio prior to quantitation of proton signal intensities, and 0.6 x 0.6 cm areas were used for quantitation.

Localized proton spectra were taken from (4-mm)3 voxels within the liver based on spatial coordinates determined from the images. The VOSY (volume selective spectroscopy) method, based on stimulated echo methods, 13*14 was used for spatial localization of the proton spectra. A 12-msec delay was used after the first and third 90” pulses, and a 65-msec delay was used after the second 90” pulse. As with the imaging protocol, the pulse sequence was gated with the respiration of the animal. Localized T2 determinations were obtained for proton resonances within the voxels by a modification of the VOSY method. The delay after the first and third 90” pulses was incremented from 12 to 40 msec in 2-msec increments, while the middle delay was kept constant at 65 msec. T2 determinations were calculated via nonlinear regression analysis.

Studies of bromobenzene-induced liver damage 0 M. BRAUER AND S. LOCKE 259

In vivo 31P MRS studies of the livers within the intact rats were obtained without surgical intervention. In addition to the localization achieved by the simple two-turn 2.0 cm (o.d.) surface coil itself, two localization strategies were implemented to improve the volume selection. First DEPTH pulsesi [e/S, 813, 8,281 were used to narrow the zone of excitation and eliminate high flux regions. Secondly, the MRS signals from skin and muscle covering the liver were eliminated using an immobilized ferrite screen which intro- duces B. field inhomogeneity in the region of the skin.i6 7-i effects were minimized by acquiring spectra with a repetition time of 5.6 sec. Saturation factors of 0.95 to 1.0 were determined for all in vivo 3’P MR spectra, comparing spectra taken every 5.6 set with spectra taken every 10 set, and saturation factors of 0.80 to 0.95 for all standard solutions were similarly determined. The resonance for the a-phosphate of ATP was used as an internal chemical shift reference for intracellular pH determination, as reported previously.’ An external capillary of 0.1 M phosphonitrilic chloride plus 16 mM Cr(II1) acetylacetonate in CC4 (PNC) mounted 1.5 cm above the center of the surface coil was used as a signal intensity and radiofrequency pulse strength standard. All spectra were acquired with the PNC capillary subjected to 90” DEPTH pulses to obtain reproducible radiofrequency penetration into the sample to a depth of - 1.5 cm. This correlates with the approximate depth of the region of high signal intensity in the ‘H MR images. All spectra were normalized relative to the area of the PNC resonance to control for variations in surface coil sensitivity between experiments. Signal intensities were calibrated against standard solutions of 40 mM NaCl, and 3-20 mM inorganic phosphate (Pi) and ATP. Spectra from the standard solutions were acquired with the same arrangement of surface coil, ferrite screen and PNC capillary and with the same DEPTH pulses and repetition times (5.6 set) as the in vivo spectra from the rats. Thus, in vivo spectra from one animal to the next could be obtained reproducibly and quantitatively.

The in vivo spectra for both BB-treated and control rat livers were analyzed by computer-fit spectral simulation of the resonances as the sum of Lorentzian line-shapes. A broad resonance with a line-width of 500-1200 Hz was due to resonances from immobilized membrane phospholipids and peripheral muscle tissue broadened by the ferrite screen and was eliminated by convolution difference methods. The areas of the resonances were normalized relative to that of the PNC resonance at -24 ppm.

In vitro 3’P MR spectra of perchloric acid (PCA) extracts of whole rat livers were obtained after the rats had been anesthetized with pentobarbital, the livers

freeze-clamped and extracted with 6% PCA.17 Ex- tracts were analyzed by conventional 31P MRS on a Bruker WH 400 NMR spectrometer, and were analyzed using our own previously published methods and resonance assignments. I7 All chemical shifts are given relative to 85% phosphoric acid as an external standard. Ten separate extractions and 31P MR spectra were obtained from five control and five BB-treated rats.

RESULTS

All rats were imaged in a supine position after they had been anesthetized with pentobarbital. Coronal images were initially taken to establish the coordinates for multiple transverse slices, using the dark outline of the lung as an anatomical reference point. Since each acquisition was gated to one expiration, the repetition time (TR) was controlled by the rat. Typically, the rat breathed once every 1 .O-2.0 sec. The TE was kept fairly short at 20 msec, as the signal intensity of the liver decreased substantially at longer TE times, indicative of a short liver T2 value.

Control rats were starved 12 hr and given an aqueous emulsion of 2% Myrj 45. Control transverse slices taken 1.8 cm catrdal to the apex of the diaphragm (Fig. 1A) show a large hypointense region to the left of the image corresponding to the empty stomach. The spinal column and skeletal muscle of the back are clearly visible at the bottom of the image. The liver is seen to the right of the stomach and above the muscu- lature of the back. The two bright circles flanking the rat are vials of 2 mM CuS04 in water (on the left) and corn oil (on the right) which serve as proton signal intensity standards.

BB-treated rats were starved 12 hr, given 10 mmol BB/kg body weight i.p. as an aqueous emulsion and anesthetized with pentobarbital prior to imaging. Trans- verse images taken 12 hr after BB injection did not show any significant change in the MRI examination. However, transverse images taken 24 hr after BB injection showed a region of the liver with significantly increased proton signal intensity (Fig. 1B) compared to the same region in control rats (Fig. 1A). This region of the liver corresponds to the perihilar region where the portal vein initially enters the liver. Other regions of the liver did not show significantly altered signal intensity at 24 hr. Five days after the BB treatment the images had returned to normal signal intensity and appearance. Qualitatively similar MRI results were observed following BB treatment in over 20 different rats. For quantitative experiments, in which images were acquired with vials of CuS04 in Hz0 and corn oil, the overall signal intensity for a 0.6 x 0.6 cm perihilar region increased by 42 + 24% (mean + stan-


Fig. 1. (A) Normal rat liver. A respiratory-gated MRI was obtained as a transverse slice through the liver of a normal rat. The stomach is clearly visible as a region of lower intensity in the lower left part of the image and the spinal cord is near the bottom of the image. The liver is the large, relatively homogeneous region below and to the right of the stomach. (B) BB-treated rat liver. An analogous image was obtained 24 hr after BB treatment. Note the area of high signal intensity in the liver to the right of the stomach.

dard deviation, SD, n = 5) relative to the same region of uninjured liver in control rats.

Pentobarbital pretreatment (75 mg/kg i.p. every day for 3 days) induces the cytochrome P450 system which metabolizes BB to its toxic form. l”,ll Pen- tobarbital pretreatment prior to BB injection resulted in a very intense increase in proton MRI signal intensity in the perihilar region after only 12 hr and death of the animal after 24 hr (data not shown).

In order to establish whether the increase in signal intensity in the rat liver caused by BB was due to an increase in the water resonance, that is, edema, or an increase in the fat (methylene protons of fatty acyl moieties) signal, that is, fatty infiltration, localized proton spectroscopy was done. The VOSY method based on the generation of stimulated echoes was used, with the spatial coordinates determined from the

transverse images. The resonance for water is at 4.8 ppm (from tetramethylsilane) and that for fat is at 1.1

ppm. Spectra taken from regions of apparent tissue damage showed a very large water proton resonance (Fig. 2B) compared to the regions of the liver appar- ently unaffected by the hepatotoxicant (Fig. 2A). As with the imaging experiments, data acquisition was triggered by the respiration of the rat. (The shoulders on the water proton resonance are due to residual eddy current effects following the last selective 90” pulse.)

The VOSY method was modified so as to obtain relative T2 values within selected voxels of the rat livers. The T2 decays obtained were not corrected for diffusion of water within a magnetic field gradient, but this effect was assumed to be constant for control and BB-treated rats. Thus, changes seen between the


10 5 0 PPM

Fig. 2. (A) In vivo proton MR spectrum of an unaffected region of liver. Localization was achieved by the VOSY method (see Methods and Materials). The predominant resonance (6 = 4.8 ppm) is that of water with a smaller peak for the methylene protons of fat (6 = 1.1 pm). (B) In vivo proton MRS spectrum of an injury region of liver. Localization was achieved by the VOSY method. The region from which this spectrum was taken corresponds to a (4-mm)’ voxel in the center of the region of high signal intensity such as that in Fig. l(B).

T2 decays of control and BB-treated rats do reflect real changes in T2-relaxation rates due to tissue injury. The regions of apparent liver damage showed significantly longer water proton T2 values than regions of unaffected liver (Fig. 3A, B). Apparent proton T2 values in regions of increased signal intensity were 34 -+ 9 msec (mean f SD, n = 3), while apparent T2 values from the same region of control rats was 16 f 5 msec (mean f SD, n = 6). No significant deviation of the data from a monoexponential decay was observed during the time measured for either the af- fected or the unaffected regions of the liver.

In vivo 3’P MR spectra of the livers of control or BB-treated rats in situ were obtained under reproducible and quantitative conditions. Reproducibility from day to day was ensured by using an external capillary of PNC mounted in one fixed location within the surface coil, as both a pulse-strength and sensitivity standard. To minimize the possible confounding effects of T, changes in the 31P resonances, the repetition time (5.6 set) was kept long relative to the in vivo 31P T,

values. Quantitation of the integrated intensities of the in vivo 31P resonances of rat liver (relative to PNC) was done by obtaining a calibration curve with known concentrations of Pi and ATP (Fig. 4). The surface coil plus external PNC were placed over sealed con-

! B ti

Fig. 3. (A) T, relaxation decay in region of normal liver. The T, relaxation time was determined using a modification of the VOSY method (see Methods and Materials). (B) T, relaxation decay in region of injured liver.

tainers of 3-20 mM Pi, 3-20 mM ATP and 40 mM NaCl in an identical arrangement to the rat studies. 40 mM NaCl was used to adjust the conductivity of a standard sample of 10 mM ATP, 10 mM Pi with the conductivity of a rat (i.e., identical Q values for the loaded surface coil in each case). Figure 4 shows a lin- ear caIibration curve (R = 0.978) with its ordinate in- tercepting almost at zero (0.01).

In vivo 31P MRS results for control and BB- treated (24 hr) rat liver in situ are shown in Fig. 5(A) and (B), respectively. Localization of the 31P resonances to the liver could be assessed from the magni- tude of the phosphocreatine (PCr) resonance which should be negligible for the liver. The presence of a very small PCr resonance at f2.3 ppm indicates that the in vivo 31P resonances came almost exclusively from the liver, with less than 8% of the spectrum coming from extra-hepatic skeletal muscle tissue (assuming that skeletal muscle has 18 mM PCr).‘,‘* The localization of these liver spectra, using the DEPTH pulses in conjunction with the ferrite shield, is partic- ularly good as the spectra were acquired with a 5.6-set TR, which does not diminish the signal of the PCr (long Tl) relative to other resonances with shorter Tl

values. The in vivo spectra for both BB-treated and control

rat livers at 24 hr were analyzed by computer-fit spectral simulation of the resonances as the sum of Lor-


1.20

0.20

0.00 0 5 10 15 20 25

[ATP] mM

Fig. 4. Calibration curve for quantitation of in vivo 31P MR spectra. The integrated intensities of the resonance of the /3- phosphate of ATP relative to that for the external capillary of 0.1 M phosphonitrilic chloride (PNC) as a function of ATP concentration in the sealed phantom. Error bars for three determinations of each phantom are of the order of the data points.

r..-.--_.Cl-rl.~ m ___-_,... 1 - ., ,...,..,.,

30 20 10 0 -10 -20 30 20 10 0 -10 -20

PPM

Fig 5 . . In vivo 31P MR spectra of (A) control, and (B) BB-treated rat liver, 24 hr. Simulated spectra (C) and (D) correspond closely with the experimental spectra (A) and (B) respectively. The individual Lorentzian line-shapes contributing to the fi- nal simulated spectra (C) and (D) are indicated in (E) and (F), respectively. In vivo 3*P spectra were assumed to be due to the PME, Pi, PDE, ATP y, (Y, and /3 resonances, as well as the PNC standard (-24 ppm).

entzian line-shapes (Fig. 5). The areas of the resonances

were normalized relative to that of the PNC resonance (at = +24 ppm). The results for control and BB- treated rats are shown in Table 1. The mean values

*SD for each resonance are indicated. BB-treatment resulted in a statistically significant (p < 0.05) decrease in the ATP and phosphodiester (PDE) resonances and a statistically significant increase in the


Table 1. Results from spectral simulation of in vivo “P MR spectra of control and BB-treated rats

Control* (mmol/liter tissue HzO)

BBt (mmol/liter tissue H20) PS

PME 2.97 f 0.41 5.46 + 1.16 Pi 2.92 k 0.53 3.49 +I 1.06 PDE 11.31 + 0.94 9.27 + 0.70 ATP,y 4.65 f 0.72 3.55 + 0.76 ATP,ar 6.87 + 0.96 5.82 + 1.26 ATP,P 4.09 + 0.51 2.98 + 0.52 PH 7.39 f 0.13 7.39 k 0.06 PME/ATP, @ 0.73 IO.13 1.83 i 0.38 Pi/ATP$ 0.71 Ik 0.17 1.17 + 0.28

<O.OOl N.S.8

<O.OOl 0.001 0.011

<O.OOl N.S.9

<O.OOl <O.OOl

*Mean + standard deviation (n = 10). tMean k standard deviation (n = 5). SProbability, Student’s t-test (2-tailed). §Not significant (p > 0.05).

phosphomonoester (PME) resonance at 24 hr. The Pi resonance tended to increase with BB treatment, but this was not statistically significant (p > 0.05). No significant change in pH between control and BB-treated rat liver was observed (p > 0.1).

In vitro 3’P MR spectra of PCA extracts for a typ-

ical control and BB-treated rat liver after euthaniza- tion with pentobarbital are shown in Fig. 6(A) and (B), respectively. No significant changes in the bioenergetic metabolites ATP, ADP or Pi were observed in vitro, although ATP was decreased in vivo by BB treatment. This apparent discrepancy may be due to the fact that in vitro spectra measure total ATP, while in vivo spectra do not detect protein-bound or intramito- chondrial ATP. 19,20 As well, there is likely to be significant contribution from surrounding normal tissue to the extract, obscuring the changes occurring in the damaged tissue. Substantial BB-induced changes were seen in the PME region (2.5-4.5 ppm), with BB-treated PCA extracts exhibiting a dramatic increase in the resonance for phosphocholine at 3.3 ppm. Phosphocho- line was increased 4.29 +_ 1.24 fold (mean f SD, n = 5) 24 hr after BB treatment relative to controls. The as- signment of 3’P resonances in PCA extracts was described previously.”

DISCUSSION

The increase in proton signal intensity observed in the perihilar region of the liver (Fig. 1B) is consistent

with the mechanism of BB hepatotoxicity. BB given i.p. is brought directly from the mesenteric veins

r-~--- ---‘-7-----T-v I...,,,-..,.,

5 0 -5 -10 -15 -20 5 4 3 -PPM

Fig. 6. In vitro 3’P MR spectra of PCA extracts of (A) control and (B) BB-treated rat liver. The PME region is expanded and shown at the right of the full spectra. Resonances of particular interest are: G3P, + 4.21 ppm; 3-phosphoglycerate, +3.99 ppm; adenosine monophosphate, +3.70 ppm; phosphocholine; +3.34 ppm; Pi, +2.38 ppm; ADP, doublets at -6.02 and -10.33 ppm; and ATP, doublets at -5.72 and -10.79 ppm, apparent triplet at -21.13 ppm.


through the portal vein to the liver. Thus, the hepatic regions closest to the initial point of entry of the portal vein will get the highest concentration of BB. A similar localized increase in proton signal intensity in the perihilar region was observed after i.p. administration of CCb. ‘,* The toxic effects of BB require it to be metabolized in the liver to the 3,4-epoxide and other toxic intermediates. The lag time needed to see changes by MRI (> 12 hr after BB administration) are consistent with the lag time needed to see histological, serum transaminase and other changes (12-24 hr).2’,22 This lag time almost certainly reflects the time needed for absorption of BB, biotransformation of the BB within the liver to toxic intermediates and the time needed for these intermediates to act.

images to those based on every breath (data not shown). Hence the images are not appreciably T,-

weighted. Tissue damage and edema generally increase the T, of tissue water,1,24s25 which would decrease rather than increase proton signal intensity in the MRI scans. A significant increase in the amount of water present in the liver as a result of BB-induced tissue damage is very likely. Since the normal liver is about 68% water24 and the water content typically rises by lo-15% for edematous tissue,23,24 an increase in water content (d) per se would not be sufficient in itself to explain the 42 f 24% increase in proton signal intensity observed.

The increase in observed proton signal due to BB- induced hepatotoxicity could be attributed to a variety of possible causes. One possibility is the induction of local fatty infiltration. Steatosis is a commonly observed effect of hepatotoxicants, because of the liver’s major role in processing and distribution of triglycer- ides. BB does induce steatosis, although not as mark- edly or selectively as other common hepatotoxicants such as CCL or ethionine.23 However, our localized proton spectroscopic results (see Fig. 2) indicate that this is not a major mechanism in the acute phase of BB-induced toxicity, because no major increase in the methylene region (1,l ppm) corresponding to fatty acyl protons could be detected. The increased intensity of the water resonance is, however, quite apparent.

The increase in proton signal intensity in the MRI images is due to a change in the water resonance, as determined by localized proton spectroscopy. For nonflowing water within the liver, the observed signal intensity (I) in a normal spin warp MRI experiment is determined by:

I= d x exp(-TE/T,) x [l - exp(-TR/7’,)]

where d is the density of protons, T, and T2 the lon- gitudinal and transverse relaxation times for water and TE and TR are the echo and repetition times for the spin warp pulse sequence.

The major mechanism for increased proton signal intensity was a dramatic increase in the T2 values of the water resonance in the damaged areas of the liver. By modification of the VOSY method, the apparent T2 for the proton resonance of water within the region of increased signal intensity in the BB-treated liver was shown to be substantially longer than the apparent T2 in the same perihilar regions of control rats (see Fig. 3A and B). An increase in apparent T2 from 16 + 5 msec (n = 6) to 34 + 9 msec (n =3) was observed. This increase in apparent T2 would increase the observed signal intensity by 94070, easily account- ing for the effects seen in our MRI studies. Literature values for in vivo normal liver are reported as 37 msec at 15 MHz6 and 30 msec at 30 MHz.~ These values are for both water and fat protons and employ the two-point method of T2 determination. No literature values for the water resonance alone have been reported for in vivo rat liver at 85 MHz, although in vitro proton T2 values for tissue water range from 24 to 50 msec1,24*25 and are independent of magnetic field strength. While our apparent T2 values may not reflect the absolute T2’s of the liver water, due to the effects of water diffusion in a magnetic field gradient, imperfect radiofrequency pulses, field and phase insta- bilities, and uncompensated eddy currents, these effects should not change between BB-treated and control rats. Thus, the increased MRI signal intensity we observed subsequent to BB treatment is due predomi- nantly to a real increase in the T2 of the tissue water.

The increase in signal intensity was not due to T, The increase in tissue water T2 is likely due to local

effects, because of the slow TR for data acquisition edema as a result of the BB-induced tissue damage. compared to the known T, relaxation times for liver The T2 values of water resonances are strongly depen- proton signals. Typical literature values for TI of dent upon the amount of “bound” water present in tis- liver in vitro and in vivo range from 240 to 500 msec sue.26 The overall observed water T2 value ( T20bS) is a for 15 to 100 MHz frequency.3,24 If we assume a TI fast-exchange averaged value determined from the T2 of 500 msec at 85 MHz proton frequency and a typi- of “free” water (T[), the very short T2 of water cal TR (based on the respiration rate of the rat) of bound to proteins and membranes (Tt), and the 1000 msec, the steady-state signal amplitude observed fraction of total water in the free (Ff) and bound would already be 0.86 of the maximum amplitude at (1 - Ff) pools: 1/T20bs = Ff/T,F+ (1 - Ff)/Tf. Lo- an infinite TR. Images taken based on respiratory gat- cal tissue damage caused ty the metabolism of BB ing to every second breath gave essentially identical could induce edema. This localized edematous re-


sponse increases the molar fraction of “free” water, and hence decreases the molar fraction of “bound” water. Since the observed TZ is dominated by the (1 - Ff)/Tf term, a small decrease in the fraction of “bound” water will result in a large increase in observed water T2.

The absolute molar concentrations determined from in vivo 31P MRS for ATP and Pi in rat liver in this study (Table 1) are in excellent agreement with values calulated from the literature for perfused rat liver. l9 Assuming that the normal liver contains 68% tissue water by weight24 (i.e., 1 mmol metabolite per kg wet weight = 1.47 mmol metaboiite per liter tissue water), perfused rat liver contains 4.12 mmol ATP/liter tissue water and 2.94 mmol Pi/liter tissue water, compared to our values of 4.09 and 2.92 mmol/liter tissue water for ATP and Pi, respectively. In vivo 31P studies of liver in human subjects have reported absolute molar concentrations’8,27 in general agreement with our own values for rat liver. Again converting all concentrations to mmol/liter tissue water, concentrations of PME, Pi, PDE and ATP are 4.70, 4.85, 14.85 and 4.26, respectively, from Buchli and Boesiger,r8 1.18, 3.27, 7.79 and 2.94, respectively, from Meyerhoff et aL2’ and 2.97, 2.92, 11.31 and 4.09, respectively, from our study (Table 1). It should be noted that the in vivo studies involving human patients18,27-29 both used the ISIS method of spectral localization, while we used DEPTH radiofrequency pulses plus a ferrite screen. One distinct advantage of our spectral localization procedure is that components with rapid T2 decay, such as the PDE and the /3 resonance of ATP are not attenuated relative to other resonances. Potential in- strumental problems which arise in ISIS and other pulsed gradient localization methods, such as eddy current effects, off-resonance effects and imperfect cancellation of signals outside the volume of interest, are not important in our localization method. How- ever our 3’P localization methods are not image- guided and are optimized to acquire spectra from the perihilar region of the rat liver.

BB induced a significant 27% decrease in in vivo levels of ATP (Table I), which indicates that BB has had substantial deleterious effects on the bioenergetic status of the liver. This may be due to the covalent binding of BB metabolites to one or more enzymes needed for production of ATP. Alternatively, BB- induced peroxidative damage to membranes may cause increased permeability to Na+, K+, or Ca2+, and hence higher rates of Na+-K+ ATPase or Ca2+ ATPase. No significant decrease in ATP was seen in in vitro 31P spectra 0 f e xtracts taken from the whole liver. This is likely due to significant contribution from surrounding normal tissue to the extract spectra.

The intracellular pH was not substantially altered,

indicating that the liver continues to rely mainly on aerobic metabolism rather than anaerobic metabolism with the generation of lactic acid. However, limita- tions in spectral resolution and sensitivity could ob- scure a possible region of localized acidosis within the BB-damaged liver. These results are in marked contrast to the pronounced acidosis observed following CCkinduced liver damage; however, this latter acidosis was postulated to be due to the metabolism of some of the CC4 to phosgene and eventually to HCl.9

BB also induced a significant 11% decrease in the in vivo 3’P resonance for PDE (Table 1) although in vitro 31P MRS results showed an increase in the major water-soluble PDE metabolites, glycerophosphoryl- choline and glycerophosphorylethanolamine (Fig. 6). A major contribution to the in vivo 31P resonance of PDE has been shown to be due to the relatively mo- bile phospholipids of the endoplasmic reticulum.30 BB is metabolized to toxic intermediates at the endoplasmic reticulum. Hence, a significant degree of peroxidative damage to these particular membranes would result in the decrease in the in vivo PDE resonance that we observed.

BB induced an 84% increase in intensity for the overall PME resonance relative to controls in in vivo 3’P MRS studies (Fig. 5 and Table 1). Analogous in vitro 31P MR spectra of PCA extracts of BB-treated rat livers showed a dramatic 4.3-fold increase in the resonance for phosphocholine (Fig. 6), which would account for the observed in vivo results. (BB did also induce a small increase in 3-phosphoglycerate levels, at +3.99 ppm.) In contrast to the large increase in phosphocholine, no significant increase in phospho- ethanolamine (+3.87 ppm), phosphoserine (+4.65 ppm) or sn-glycerol3-phosphate (+4.21 ppm) was observed. This indicates a selective alteration of phosphatidylcholine metabolism, but not a change in other major phospholipids. Recent studies have shown the importance of a phosphatidylcholine (PhCh)-specific phospholipase C (PLC), which degrades PhCh to phosphocholine and diacylglycerol.3’-35 Interleukin- 1, which is known to be released from a variety of cells in response to injury, infection or immunologic chal- lenge, stimulates PhCh-specific PLC31 Other extracellular messengers, such as platelet derived growth factor, vasopressin and c+adrenergic stimulation also stimulates this specific PLC.32,33 Energy depleted or structurally damaged myocytes again demonstrate this specific PLC stimulation. 34 Cultured rat hepatocytes exposed to BrCC13, a more potent Ccl4 analog, ex- hibited increased PhCh degradation by a rapidly activated PLC.35 It should be noted that all these studies were done on isolated cellular or subcellular systems. PhCh-specific PLC cleaves PhCh into phosphocholine and the second messenger, diacylglycerol, which in-


duces translocation and stimulation of protein kinase C. This key enzyme in turn activates and inhibits a

wide variety of major metabolic processes.36 A major regulatory role for PhCh-specific PLC in cell prolif- eration, complementary to but distinct from the phosphatidylinositol/phosphoinositol regulatory scheme, has hence been proposed. 3’-33 Thus, our noninvasive observation of an increased PME resonance in in vivo 31P MR spectra after BB-induced tissue damage may well reflect an important regulatory process occurring within the rat liver in situ. The present study is consistent with the hypothesis that PhCh-specific PLC activ- ity has a significant role in the mechanism of tissue damage at the intact whole-animal level.

It is noteworthy that our in vivo 3’P MRS studies of Ccl* and BrCCl,-induced liver damage also showed a significant increase in the PME resonance.’ This increase in PME occurred over roughly the same time course as the development of tissue damage reflected in a localized increase in proton signal intensity seen via MRI.’ Again, in vitro 31P MRS analysis of PCA liver extracts showed the increase in PME to be due to a selective increase in phosphocholine. The stimulation of PhCh-specific PLC, by interleukin-1 or other extracellular messengers, may thus be a general response of damaged tissue to initiate tissue regeneration.

The pattern of MRI/MRS observable changes seen in this study after acute exposure to BB involves an increased proton signal intensity in the perihilar region of the liver (due to an increased T2 of tissue water) seen via MRI, and an increase in PME (specifically phosphocholine) and decrease in ATP seen via in vivo 3’P MRS. A remarkably similar pattern of MRUMRS changes was seen after CC& treatment.7m9 As with the BB results, a localized increase in MRI signal intensity was seen due to an increase in tissue water T2, and an increase in the in vivo 31P resonance of PME was seen due to a selective elevation of phosphocholine levels in the damaged livers. The MRI changes are consistent with hepatic edema and swelling of the hepatocytes secondary to peroxidative damage to the membranes. The decrease in ATP would then reflect an increased rate of ATP hydrolysis by the Na+-Kf ATPase and Ca2+ ATPase in attempting to re-establish the normal gradient of Na+ (-150 mM extracellular, - 10 mM intracellular), K+ (-10 M extracellular, 140 mM intracellular) and Ca’+ (-2 mM extracellular, 0. l-l PM intracellular). 37 The same MRI changes were seen in CC&-induced liver damage,‘,’ as well as the increase in phosphocholine observed by both in vivo and in vitro 3’P MRS.9 The increase in the in vivo PME resonance due to a selective increase in phosphocholine may indicate a phosphatidylinositol-independent pathway for the generation of the second messenger,

diacylglycerol, via activation of PhCh-specific PLC. The fatty acyl composition of phosphatidylcholine dif- fers significantly from that of phosphatidylinositol,32,37 potentially leading to chemically distinct diacylglycerols generated from each pathway, each with their own spec- ificity and sites of control.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

REFERENCES

Andrew, E.A.; Bydder, G.B.; Griffiths, J.; Iles, R.; Styles, P. Clinical Magnetic Resonance Imaging and Spectroscopy. New York: Wiley-Liss; 1990. Stark, DD.; Bradley, W.G. Magnetic Resonance Imag- ing. Langham, MD: Mosby Press; 1989. Stark, DD.; Goldberg, H.I.; Moss, A.A.; Bass, N.M. Chronic liver damage. Evaluation by magnetic resonance. Radiology 150: 149; 1984. Stark, D.D.; Bass, H.M.; Moss, A.A.; Bacon, B.R.; McKerrow, J.H.; Cann, C.E.; Brito, A.; Goldberg, H.I. NMR imaging of experimentally induced disease. Radiology 148:743-751; 1983. Davolio-Marani, R.; Federico, M.; Iannone, A.; Magin, R.L.; Schwartz, H.M.; Vannini, V. CC&-induced liver lesions in rat liver studied by MRI: Contrast enhanced by dextran magnetite. Proc. Sot. Magn. Reson. Med. 8: 793; 1989. Ratner, A.V.; Carter, E.A.; Pohost, G.M.; Wands, J.R. NMR spectroscopy and imaging in the study of experimental liver diseases. AIcoholism: C/in. Exp. Res. 10: 241-245; 1986. Brauer, M.; Towner, R.A.; Renaud, I.; Janzen, E.G.; Foxall, D.L. In vivo proton NMR imaging and spectroscopy studies of halocarbon-induced liver damage. Magn. Reson. Med. 9:229-239; 1989. Brauer, M.; Towner, R.A.; Foxall, D.L. Sodium-23 and proton NMR imaging studies of CC&-induced liver damage in the rat. Magn. Reson. Imaging 8:459-463; 1990. Towner, R.A.; Brauer, M.; Janzen, E.G.; Ling, M. In vivo and in vitro 3’P-NMR spectroscopy of rat liver treated with halocarbons. Biochim. Biophys. Acta 993: 92-99; 1989. Gillette, J.R. Formation of reactive metabolites of for- eign compounds and their covalent binding to cellular constituents. Handbook Physiol. 9:577-588; 1980. Monks, T.J.; Lau, S.S. Reactive intermediates and their toxicological significance. Toxicology 52:1-53; 1988. Casini, A.F.; Pompella, A.; Comporti, M. Liver glutathione depletion induced by bromobenzene and its rela- tion to lipid peroxidation and necrosis. Am. J. Pathol. 118:225-237; 1985. Frahm, J.; Merboldt, K.D.; Hanicki, W.; Haase, A. Stimulated echo imaging. J. Magn. Reson. 64:81-93; 1985. Granot, J. Selected volume excitation using stimulated echoes. J. Magn. Reson. 70:488-492; 1986. Bendall, M.R. Surface coils and depth resolution using the spatial variation of radio frequency field. In: T.L.

Studies of bromobenzene-induced liver damage 0 M. BRAUER AND S. LOCKE 26-t

James, A.R. Margulis (Eds). Biomedical Magnetic Res- onance, San Francisco: Radiol. Research and Education

Foundation; 1987: pp 99-134. 16. Geoffrion, J.; Rydzy, M.; Saunders, J.K.; Smith,

1.C.P.; Jarrell, H.D. Application of immobilized ferrite particles to in vivo NMR surface coil spectroscopy. Sot. Magn. Reson. Med. 6:968; 1987.

17. Ling, M.; Brauer, M. In vitro 3’P NMR spectroscopic

studies of rat liver subjected to chronic ethanol administration. Biochim. Biophys. Acfa 1051:151-158; 1990.

18. Buchli, R.; Boesiger, P. Absolute 3’P metabolite concentrations in human brain, liver and muscle. Sot. Magn. Reson. Med. 8:598; 1989.

19. Desmoulin, F.; Cozzone, P.J.; Canioni, P. Phosphorus- 3 1 nuclear magnetic resonance study of phosphorylated metabolites. Compartmentation, intracellular pH and phosphorylation state during normoxia, hypoxia and ethanol perfusion in the perfused rat liver. Eur. .I. Bio- them. 162:151-159; 1987.

20. Iles, R.A.; Stevens, A.N.; Griffiths, J.R.; Morris, P.G. Phosphorylation status of liver by 3’P-NMR spectroscopy and its implications for metabolic control. Bio- them. J. 229:141-151; 1985.

21. Casini, A.F.; Maellaro, E.; Pampella, A.; Ferrali, M.; Comporti, M. Lipid peroxidation, protein thiols and calcium homeostasis in bromobenzene-induced liver damage. Biochem. Pharmacol. 36:3689-3695; 1987.

22. Brodie, B.B.; Reid, W.D.; Cho, A.K.; Sipes, G.; Krishma, G.; Gillette, J.R. Possible mechanisms of liver necrosis caused by aromatic compounds. Proc. Natl. Acad. Sci. USA 68:160-164; 1971.

23. Zimmerman, H.J. Chemical hepatic injury and its de- tection. In: G.L. Plaa, W.R. Hewitt (Eds.) Toxicology of the Liver. New York: Raven Press; 1982:~~ l-45.

24. Beall, P.T.; Amtey, S.R.; Kasturi, S.R. NMR Data Handbook for Biomedical Applications. New York: Pergamon Press; 1984.

25. Holzmueller, P.; Gamiscek, G.; Moser, E.; Markis, E.M.; Echsel, H. Proton Relaxation time alterations in rat liver treated with different hepatotoxic model agents. Sot. Magn. Reson. Med. 8:780; 1989.

26. Lynch, L.J. Water relaxation in heterogeneous and bio- logical systems. Magn. Reson. Biol. 2:248-304; 1980.

27. Meyerhoff, D. J.; Boska, M.D.; Thomas, A.M. ; Weiner, M.W. Alcoholic liver disease: Quantitative image-guided P-31 MR spectroscopy. Radiology 173:393-400; 1989.

28. Roth, K.; Hubesch, B.; Meyerhoff, D.J.; Naruse, S.; Gober, J.R.; Lawry, T.J. Boska, M.D.; Matson, G.B.; Weiner, M.W. Noninvasive quantitation of phos- phorous metabolites in human tissue by NMR spectroscopy. J. Magn. Reson. 81:299-311; 1989.

29. Bottomley, P.A.; Hardy, C.J. Rapid, reliable in vivo as- says of human phosphate metabolites by nuclear magnetic resonance. Clin. Chem. 35:392-395; 1989.

30. Williams, S.R.; Murphy, E. J.; Bates, T.E.; Watson, T.; Brindle, K.M.; Rajagopalan, B.; Radda, G.K. Endo- plasmic reticulum contributes to PDE signal in 3’P MRS. Proc. Sot. Magn. Reson. Med. 9:284; 1990.

31. Rosoff, P.M.; Savage, N.; Dinarello, C.A. Interleukin- 1 stimulated diacylglycerol production in T lymphocytes by a novel mechanism. Cell 54:73-81; 1988.

32. Moscat, J.; Cornet, M.E.; Lopez-Barahona, M. Activa- tion of phosphatidylcholine-specific phospholipase C in cell growth and oncogene transformation. Biochem. Sot. Trans. 17:988-991; 1989.

33. Besterman, J.M.; Duronio, V.; Cuatrecasas, P. Rapid formation of diacylglycerol from phosphatidylcholine: A pathway for generation of a second messenger. Proc. Natl. Acad. Sci. USA 83:6785-6789; 1986.

34. Myrmel, T.; Larsen, T.S.; Skulber, A.; Forsdahl, K.; Little, C. Phospholipase C-evoked glycerol release in energy depleted rat myocardial cells. Mol. Cell. Biochem. 88:107-111; 1989.

35. Lamb, R.G.; Jarrett, M.; Gregg, P.; Bush, R. Bromo- trichloromethane-dependent alternations in liver cell phosphatidyl choline metabolism. Fed. Am. Sot. Exp. Biol. 49:2797; 1990.

36. Vance, D.E.; Vance, J.E. Biochemistry of Lipids and Membranes. Menlo Park, CA: Benjamin/Cummings Publishing; 1985:~. 319.

37. Staub, N.C.; Taylor, A.E. Edema. New York: Raven Press; 1984.

Documents

Proton magnetic resonance imaging and phosphorus-31 magnetic resonance spectroscopy studies of bromobenzene-induced liver damage in the rat