Accumulation of 1,2-sn-Diradylglycerol with Increased Membrane-associated Protein Kinase C May Be the Mechanism for Spontaneous Hepatocarcinogenesis in Choline-deficient Rats*

Choline deficiency, via deprivation of labile methyl groups, is associated with a greatly increased incidence of hepatocarcinoma in experimental animals. This di- etary deficiency also causes fatty liver, because choline is needed for hepatic secretion of lipoproteins. We hy- pothesized that fatty liver might be associated with the accumulation of 1,2-sn-diradylglycerol and subse- quent activation of protein kinase C. Several lines of evidence indicate that cancers might develop second- ary to abnormalities in protein kinase C-mediated signal transduction. We observed that rats fed a choline- deficient diet for 1, 6, or 27 weeks had increased hepatic concentrations of 1,2-diradylglycerol. At 1 and 6 weeks, hepatic plasma membrane from choline-de-ficient rats had increased concentrations of 1,2-sn- diacylglycerol and 1-alkyl, 2-acylglycerol, with the latter accounting for 20-26% of membrane 1,2-sn-diradylglycerol (as compared with only 2-5% in con- trols). Protein kinase C activity was increased in hepatic plasma membrane at 1 week

Choline deficiency, via deprivation of labile methyl groups, is associated with a greatly increased incidence of hepatocarcinoma in experimental animals. This dietary deficiency also causes fatty liver, because choline is needed for hepatic secretion of lipoproteins. We hypothesized that fatty liver might be associated with the accumulation of 1,2-sn-diradylglycerol and subsequent activation of protein kinase C. Several lines of evidence indicate that cancers might develop secondary to abnormalities in protein kinase C-mediated signal transduction. We observed that rats fed a cholinedeficient diet for 1, 6, or 27 weeks had increased hepatic concentrations of 1,2-diradylglycerol. At 1 and 6 weeks, hepatic plasma membrane from choline-deficient rats had increased concentrations of 1,2-sndiacylglycerol and 1-alkyl, 2-acylglycerol, with the latter accounting for 20-26% of membrane 1,2-sndiradylglycerol (as compared with only 2-5% in controls). Protein kinase C activity was increased in hepatic plasma membrane at 1 week of choline deficiency. By Western blotting there was an increase in the amount of protein kinase C 5 and a decrease in the amount of protein kinase C 6 in liver at 1 week. By 6 weeks of choline deficiency, hepatic plasma membrane and cytosolic protein kinase C (PKC) activities were increased significantly, with increased amounts of hepatic plasma membrane protein kinase C a, and 6 detected by Western blotting. Glycogen synthase activity in liver was diminished after 1 week of choline deficiency; this enzyme is inhibited by PKC-mediated phosphorylation. We suggest that choline deficiency perturbed PKC-mediated transmembrane signaling within liver and that this contributed to the development of hepatic cancer in these animals. * This work was supported by a grant from the American Institute for Cancer Research and by the National Institutes of Health Grant HD16727. Some of the research described herein was submitted in partial fulfillment of the thesis requirement for the Doctor of Philosophy degree for K. D. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The authors contributed equally to the work described and the order for listing authorship does not indicate precedence. Choline is the major dietary source for labile methyl groups ) and its metabolism is inter-related with methionine and folate metabolism; choline deficiency depletes all of these methyl donors (Pomfret et al., 1990;Selhub et al., 1991;Zeisel et al., 1989). Choline is the only single nutrient for which dietary deficiency is associated with development of foci of premalignant hepatocytes (which express y-glutamyl transpeptidase and the placental form of glutathione S-transferase (Shinozuka and Lombardi, 1980;Takahashi et al., 1979)) and with the subsequent development of hepatocarcinomas in the absence of any known carcinogen (Chandar and Lombardi, 1988;Mikol et al., 1983;Newberne and Rogers, 1986).
Due to the inability to form phosphatidylcholine, a constituent of the lipoprotein envelope, secretion of triglyceride (TG)' is inhibited in choline deficiency Vance, 1988, 1989), causing lipid to accumulate (Lombardi, 1971;Lombardi et al., 1968;Yao and Vance, 1988). In addition to massive accumulation of TG, we have observed the accumulation of 1,2-sn-diradylglycerol (DRG) within liver during choline deficiency (Blusztajn and Zeisel, 1989). DRG is an intermediate for the biosynthesis of TG and membrane phospholipids. DRG is also a second messenger, formed when plasma membrane receptors for certain hormones, neurotransmitters, or growth factors are coupled to phospholipases (Blackshear et al., 1988;Exton, 1988). The DRG molecule remains within the membrane after hydrolysis of phospholipids and can activate a regulatory enzyme, protein kinase C (PKC), by facilitating the enzyme's translocation from cytosol to plasma membrane (Nishizuka, 1986). There are at least 10 isoforms of PKC encoded by at least six distinct genes; only a, 6, and { isoforms have been identified in liver (Wetsel et al., 1992). The predominant components of DRG in tissues are the esterlinked 1,2-diacyl (DAG) species, and the 1-alkyl-2-acyl glycerols (AAG; ether-linked 1-0-alkyl-2-acyl and vinyl-linked 1-0-alk-l'-enyl-Z-acyl species). These subclasses of DRG may differ in their ability to activate PKC (Bass et al., 1989;Cabot and Jaken, 1984;Daniel et al., 1988;Dawson and Cook, 1987;Ford et al., 1989;Heymans et al., 1987). The targets for phosphorylation by PKC have not been completely described, but they include receptors for insulin, epidermal growth factor, and many proteins involved in control of gene expression, The abbreviations used are: TG, triglyceride; DRG, 1,2-sn-diradylglycerol; PKC, protein kinase C; DAG, 1,2-diacylglycerol; AAG, 1alkyl-2-acyl glycerols; TPCK, n-tosyl-L-phenylalanine chloromethyl ketone; CHAPSO, 3-[3-cholamidopropyl)dimethylammonio]-2-hydroxy-1-propanesulfonic acid BSA, bovine serum albumin; PBS, phosphate-buffered saline. This is an Open Access article under the CC BY license. cell division, and differentiation (Nishizuka, 1986;Weinstein, 1990). The appearance of DRG in membranes is usually transient, and therefore PKC is activated for only a short time after a receptor has been stimulated (Price et al., 1989). In choline-deficient liver, chronic accumulation of DRG could result in prolonged activation of PKC. We now report upon the subcellular distribution and subclasses of DRG in liver during choline deficiency and upon the increase in PKC protein and activity within these livers.

MATERIALS AND METHODS
Animal Care-Animal care was in accordance with institutional guidelines of the University of North Carolina at Chapel Hill. Male Fischer 344 rats (90 g; Charles River Breeders, Wilmington, MA) were housed in an isolated room in stainless steel hanging cages and were exposed to light for 12 h/day. They were fed a semipurified choline sufficient or choline-deficient. diet (contained 0.2% methionine, 0.005 g/kg folate, 0.003 g/kg vitamin B12; sufficient diet contained 0.46% choline, deficient diet contained 0.006% choline by our assay; Lombardi Choline Diets, Dyets, Bethlehem, PA) for 1, 6, or 27 weeks. Water was offered ad libitum. Animals were anesthetized with ether, and a sample of liver was collected and chilled on ice for immediate use in subcellular fractionation and Western blotting. A lobe of liver was also collected by freeze-clamping with tongs cooled in liquid nitrogen and was stored at -80 "C until used for phosphocholine and glycogen synthase assays. Subcellular Fractionation-Weighed aliquots (2 g) of the fresh livers were homogenized in a cold sucrose buffer (0.25 M; ultracentrifugation grade, Fisher) containing 20 mM Tris, 0.5 mM EGTA, 2 mM EDTA, 2 mM MgCI,, 0.1% aprotinin, 20 p M leupeptin, 0.1 mM n-tosyl-t-phenylalanine chloromethyl ketone (TPCK), 5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride (Sigma), pH 8.0. The lipid layer on top was aspirated off and saved, and subcellular fractions were prepared (Aronson and Touster, 1974). Marker enzymes were assayed in all fractions and were used to check purity of the following fractions: delipidated cytosol (lactate dehydrogenase (Cabaud and Wroblewski, 1958)), mitochondria + lysosomal pellet (succinate dehydrogenase (Pennington, 1961), and acid phosphatase (Andersch and Szczypinski, 1947)), plasma membrane from microsomal pellet (5'-nucleotidase (Dixon and Purdom, 1954)), endoplasmic reticulum (glucose 6-phosphatase (Aronson and Touster, 1974)). Purity of nuclei and of lipid droplets was determined by measuring DNA (Labarca and Paigen, 1980), and TG (Tacconi and Wurtman, 1985), respectively. The cytosol, plasma membrane, nuclei and lipid fractions had < 2% contamination with any of the other fractions based upon measurements of the marker enzymes.
Lipid Analyses-Lipids were extracted from liver and subcellular fractions of liver using the method of Bligh and Dyer (1959). Hepatic T G was isolated by thin layer chromatography, hydrolyzed, and measured as methyl esters using capillary gas chromatography (Tacconi and Wurtman, 1985). Triheptadecanoin (C17:O; Nu Chek Prep, Elysian, MN) was used as an internal standard to correct for variations in recovery. DRG was determined using diacylglycerol kinase (Lipidex, Westfield, NJ) and [32P]ATP (Amersham Corp.) to form radiolabeled phosphatidic acid (Preiss et al., 1986), which was isolated by thin layer chromatography (CHC13/MeOH/acetic acid, 65:15:5), and quantitated using an imaging radiometric scanner (Bioscan, Washington, DC). This assay is specific for 1,2-sn-diradylglycerol and will not detect other isomers nor TG. AAG was determined by converting the total diradylglycerol to [3ZP]phosphatidic acid using diacylglycerol kinase according to Preiss et al. (1986) and then adding 500 units of Rhizopus lipase (Sigma) for 30 min at 37 "C to destroy the DAG present (Tyagi et al., 1989). The radiolabeled phosphatidic acid was then quantitated as described earlier. Amounts of AAG were calculated by comparison with analyses of authentic standard synthesized from l-O-alkyl-2-oleoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids, Alabaster, AL) using phospholipase C (Blank et al., 1984;Mavis et al., 1972) and purified with a C18 cartridge (Seppak; Waters Associates, Milford, MA). DAG was determined as the difference between total DRG and AAG measurements. Diacylglycerol kinase and [32P]ATP will react with ceramide (Turinsky et al., 1990) to form a phosphorylatedproduct that was completely separated from [32P]diradylphosphatidate during thin layer chromatography. We believe that the addition of protease inhibitors (0.1% aprotinin, 20 p M leupeptin, 0.1 mM TPCK, 1 mM phenylmethylsulfonyl fluoride) a t the start of membrane preparation prevented artifactual formation of DRG during subcellular fractionation. DRG concentrations were lower in plasma membrane preparations (6-week animals) that were prepared in the presence of the protease inhibitors as compared with membrane prepared from liver without the inhibitors (36% lower in controls, 49% lower in deficients; p < 0.01 by t test). The addition of protease inhibitors at the end of plasma membrane preparation did not prevent increased DRG concentrations. Thus, we concluded that one or more of these inhibitors prevented phospholipid breakdown and that protease inhibitors should be added when DRG is to be measured in subcellular fractions.
The lipid droplets in the choline-deficient livers contained more than 1000 times more TG than DRG. If lipid droplets contaminated subcellular fractions during preparation, then the TG content of fractions would provide an upper bound to the amount of such contamination. Using the DRG/TG ratio in the droplets of deficient liver (0.001 at 1 week, 0.0005 at 6 weeks, 0.0006 at 27 weeks), and the TG concentrations measured in the plasma membrane fraction, we found that an insignificant amount of the DRG in plasma membrane could have come from contamination by lipid droplets (Table I).
Thus, the hepatic DRG concentrations in plasma membrane were not an artifact of TG contamination from the lipid droplets. Cytosol, from which the fat layer had been removed after low speed centrifugation, was examined in a similar manner (Table I). In this fraction, most of the DRG could have come from lipid droplets. In another experiment, we determined whether DRG that was associated with lipid droplets might have been shifted to plasma membrane during subcellular fractionation, we took the lipid layer from 6 week cholinedeficient liver and mixed it with the homogenate of liver from control rats. There was no difference between DRG concentrations in plasma membrane prepared from control liver or from control liver with added fat layer (10.4 f 0.7 and 10.6 It 0.3 nmol/mg of protein, respectively). Protein in samples used for DRG measurements was determined using a colorimetric assay (Bradford, 1976).
Protein Kinase C Assay and Western Blotting-Total homogenate, delipidated cytosol, and plasma membrane from microsomal pellet were prepared and PKC activity was assayed by measuring the transfer of [32P]phosphate from [T-~'P]ATP to histone H1 by a slight modification of the method of Kikkawa et al. (1983). The samples were solubilized with 1% Nonidet P-40 and kept overnight at -80 "C to increase solubilization and maximize activity. They were thawed and incubated on ice with 1% CHAPSO (Sigma). After sonicating with 10 1-s pulses (setting 7; model W225R, Heat Systems Ultrasonics, Plainview, NJ), the samples were applied to a DEAE-cellulose column (Sigma) equilibrated with Buffer A (20 mM Tris-HCI, 2 mM sodium EDTA, 0.5 mM EGTA at pH 7.5). Columns were washed twice with 0.5 ml of Buffer A. The PKC was eluted with 0.45 ml of 0.15 M NaCl in Buffer A. Assay tubes contained the following in a volume of 100 pl: 20 mM Tris-HCI, 1 mM CaC12, 10 mM MgCl,, 500 gg/ml histone Type 111-S (Sigma), 0.2 mM ATP, and IO6 dpm [y-"PI ATP (Amersham Corp.). The reaction was started with the addition of 5-15 pg of protein of the PKC preparation. Samples were assayed in triplicate in the presence and absence of 90 pg of phosphatidylserine (Sigma) and 1 p~ phorbol 12-myristate 13-acetate (Sigma). Reactions were incubated for 5 min at 32 "C and stopped by addition of 0.4 ml of 20% trichloroacetic acid and 10 p1 of 20 pg/pl bovine serum albumin (BSA; Sigma). Trichloroacetic acid-precipitated protein was collected on filter paper with a Skatron Cell Harvester (Sterling, VA) and unincorporated label was removed by washing with 7.5% trichloroacetic acid. Radioactivity of 32P on the filters was determined by liquid scintillation counting, and PKC activity was expressed as activity in the presence of activators less the activity measured in the absence of the phospholipid and phorbol ester activators. Protein was determined by the bicinchoninic acid method of Smith et al. (1985).
In order to determine whether PKC might be associated with lipid droplets, we also determined activity in this fraction. Lipids were aspirated from the top of the tube after the first centrifugation of the subcellular fractionation procedure, and this fraction was washed three times with Tris buffer (20 mM, pH 7.5). Nonidet P-40 (1%) was added and sample was frozen overnight. Prior to assay, CHAPSO (1%) was added and sample sonicated. No PKC activity was associated with the lipid droplets once they had been washed.
An aliquot of liver homogenate (25 pg total protein/lane) was subjected to 10% SDS-polyacrylamide gel electrophoresis according to the method of Laemmli (Laemmli, 1970  I Triacylgl.vcero1 and 1.2-sn-diradylRlycero1 in p h m a membrane and cytosol Using the DRG/TG concentration ratio present the lipid droplets in the liver (0.001 at 1 week; 0.0005 at 6 weeks, 0.0006 at 27 weeks), and the TG concentrations measured in the plasma membrane and cytosolic fractions, the possible contamination of DRG from the lipid droplets was calculated in these two fractions from livers of rats fed a choline-deficient diet for 1, 6, and 27 weeks. Percent maximum contamination was calculated by comparing these values to the DRG concentration in the fractions (from Tahles 11-IV). Data are expressed as mean k S.D. were transferred by electroblotting (50 min at 400 mA using a semidry blotting apparatus; Owl Scientific Plastics, Cambridge, MA) on nitrocellulose paper (pore size, 0.45 pm). Nonspecific binding sites were blocked by incubation with 5% nonfat dry milk in 1 X phosphatebuffered saline (PRS) for 1 h at room temperature. The nitrocellulose sheets were incubated with the respective PKC antisera (a, 6, or j-at 1:1000 dilution in 1% RSA in 1 X PRS; antibodies were a generous gift of Dr. Yusuf A. Hannun, Duke University) for 4 h at room temperature. These three isozymes were studied because they have previously been shown to be the only isozymes of PKC present in liver (Wetsel et al., 1992). After five washes with 1 X PRS at 10-min intervals, secondary antiserum (sheep anti-rabbit IgG conjugated with alkaline phosphatase, diluted 1:1000 in 1% RSA in 1 X PRS; Sigma) was added and incubated for 2 h at room temperature. The above washing procedure was repeated, and color was developed using bromochloroindoyl phosphate/nitro blue tetrazolium. Density of the 80-kDa bands were determined by scanning densitometry. Glycozen Synthase-Aliquots of frozen liver were homogenized in cold 0.25 M sucrose, 50 mM Tris, 10 mM EDTA, pH 7.8 (6 volumes/g of tissue) and centrifuged for 30 rnin at 16,000 X g and 4 "C. The supernatants were collected, and glycogen synthase activity was assayed by measuring the incorporation of radioactivity from uridine 5'-diphospho["C]gIucose (Amersham Corp.) into glycogen (Thomas et nl., 1968). Samples were added to an assay mixture containing 50 mM Tris, 10 mM EDTA, pH 7.8, 6.7 mM UDP-["C]glucose (0.0s pCi/ pmol) and 10 mg/ml rabbit liver glycogen (Sigma) in the presence and absence of 6.7 mM glucose 6-phosphate (Sigma). After incubating for 15 min at 30 'C, an aliquot was spotted on filter paper (2 X 2 cm; Whatman 31ET, Hillshoro, OR), and the unincorporated label was removed with 66% ethanol. Radioactivity of "C on the filters was determined by liquid scintillation counting and glycogen synthase activity was expressed as the ratio of the activity measured without/ with glucose &phosphate. I'hosphocholine-Phosphocholine was measured in samples of frozen liver using the method of Pomfret et al. (1989).

RESULTS AND DISCUSSION
When rats were fed a semisynthetic diet devoid of choline for 1, 6, or 27 weeks, fat accumulated in liver (Fig.  1). T h e livers of the control animals were smooth, shiny, and deep red in color at all the time intervals. However, the livers of t h e choline-deficient animals a t 1 and 6 weeks were yellowbrown in color with a micronodular greasy appearance. They were also larger (9.3 k 0.5 g a t 1 week, 13.2 f 0.2 g a t 6 weeks) than the control livers (4.9 f 0.2 g at 1 week, 7.5 f 0.3 g at 6 weeks). The choline-deficient livers at 27 weeks also looked micronodular, swollen, and pale brown, but their weights (

pm for A-C.
Choline deficiency was verified by showing that hepatic concentrations of phosphocholine were markedly diminished in deficient animals as compared with controls. Phosphocholine content of liver has been shown previously to be a sensitive indicator of choline deficiency (Pomfret et al., 1990;Zeisel et al., 1989). In our study, at 1 week phosphocholine content of liver was 617 nmol/g (+.40) in control rats' liver and 37 nmol/g ( k 3 ) in deficient rats' liver. At 6 weeks, phosphocholine content of liver was 906 nmol/g (5.54) in control rats' liver and 50 nmol/g (58) in deficient rats' liver. At 27 weeks, phosphocholine content of liver was 2.273 nmol/ g (+.285) in control rats' liver and 225 nmol/g (224) in deficient rats' liver.
As expected, the massive accumulation of lipid in liver was largely due to accumulation of TG (Lombardi et al., 1968). Hepatic TG concentrations in control animals remained relatively constant during 27 weeks (approximately 200 nmol/ mg protein). In deficient animals TG in liver was 941 f 81 a t 1 week, 1,339 +. 161 a t 6 weeks, and 646 +. 39 a t 27 weeks (a11 values p < 0.01 different from control at same time point).
Hepatic plasma membrane TG was similar in controls and deficient animals at all time points (70-250 nmol/mg protein). In addition to accumulation of TG, we observed that, in livers from choline-deficient animals, there was a remarkable increase in DRG concentrations (Tables 11,111,and IV). DRG only increased in plasma membrane and nuclear membrane and not in endoplasmic reticular membrane. In plasma membrane, DRG reached values higher than those occurring after stimulation of a receptor linked to phospholipase C activation (e.g vasopressin receptor) and of the order of magnitude   needed to activate PKC in platelets and to modify responses t o &,-adrenergic receptors in the liver (Bocckino et al., 1985;Cooper et al., 1985;Kaibuchi et al., 1983;Lapetina et al., 1985;Preiss et al., 1986). At all time points, most of total hepatic DRG was present within the lipid droplets (see Fig. 1; Tables 11-IV that accumulated in choline deficiency. AAG in hepatic lipid droplets from deficient animals accounted for 14,20, and 18% of total DRG at 1, 6, and 27 weeks, respectively. The increase in plasma membrane DRG observed at 1 and 6 weeks was the result of increases in both AAG and DAG content (Table V; note that these data are from a replication of the experiment reported in Tables 11-IV and differ slightly in total DRG present). In control animals, AAG accounted for 2-5% of DRG in liver plasma membrane. In deficient animals, AAG accounted for 20-26% at times when DRG was elevated (1 and 6 weeks). Previously, such high concentrations of AAG have been mainly observed in tumor cells (Snyder and Snyder, 1975) or in stimulated neutrophils (Bauldry et al., 1988). These subclasses of DRG may differ in their ability to activate PKC. Although AAG has been considered to be incapable of activating PKC (Cabot and Jaken, 1984), recent studies indicate that naturally occurring species of AAG activate PKC (Dawson and Cook, 1987;Ford et al., 1989). This activation may require free calcium concentrations found only in stimulated cells (Ford et al., 1989). The alkyl group in the 1-position diminishes the diglyceride's ability to activate PKC relative to the corresponding DAG (Heymans et al., 1987), and AAGs with short chain fatty acid substituents appear to be better activators than are those with longer chain fatty acids (Bass et al., 1989). 1-0-Alkyl-2-acetyl analogs of AAG inhibit PKC activation by DAG (Daniel et al., 1988). This acetyl analog may differ from other AAGs in biologic activity; Rider et al. (1988) have speculated that the biologic activity of AAG may be dependent on conversion to alkylacetylglycerophosphocholine (platelet-activating factor).
We observed significant increases in PKC activity associated with hepatic plasma membranes (Tables VI-VIII). That PKC increased at 6 weeks in both plasma membranes and cytosol (Table VII) suggests that there was a stable activation of PKC and/or an increase in the total PKC pool in the cell.

TABLE V
Subclasses of 1,2-sn-diradylglycerol i n liver plasma membrane during choline deficiency Rats were treated as described in Table 11. Total diradylglycerol was determined using a radioenzymatic assay. AAG was determined after hydrolysis of diacylglycerol species using Rhizopus lipase. These data are from a replication of the experiments reported in Tables II-IV. DAG was calculated as the difference between total diradylgly-cero1 and AAG. Data are expressed as mean f S.E. for n = 6 samples per point. Statistical differences were determined using Student's unpaired t test.    Rats were fed defined diets containing choline ( C ) or devoid ofcholine ( I ) ) for 1 or 6 weeks. liver was collected and homogenate prepared as descrihed under "Materials and Methods." Proteins (25 p g total protein/lane) were analyzed by SDS-polyacrylamide gel electrophoresis, electrohlotted, and immunostainetl with anti-protein kinase C ( Y , 6, or {antihody as described "Materials and Methods." Representnt ive hands are shown for each isozyme and hoth diets. The amounts o f I'KC isozymes in deficients expressed as percent mean hand density as compared with control are also presented for all animals studied ( n = :)/point). * = p < 0.05 different from control.
The changes in membrane-associated PKC (Tables VI-VIII) t h a t we observed might he due to changes in one or more of these isoforms. Using Western blotting (Fig. 2), in 1-week liver homogenate, we ohserved that amounts of PKC {were increased (3-fold; p < 0.05), whereas PKC d decreased (to 23% of control; p < 0.05). At 6 weeks (Fig. 2), amounts of P K C (Y a n d d were increased (2and IO-fold, respectively; p < 0.05).
W e observed changes in plasma memhrane but not in endoplasmic reticulum. This may explain why other investigators did not find that PKC in total particulate fraction changed during choline deficiency (Singh et al., 1990). Our ohservations of simultaneous accumulation of AAG, DAG, and increased amount. and activation of PKC suggests that, in oiuo, accumulation of these DRGs perturhed PKC signal transduction.
In order t,o assess whether functional changes consistent with activation of P K C were present, we measured glycogen synthase activity in livers from animals fed a choline-deficient diet. A t 1 week, glycogen synthase activity was lower in the deficient (0.29 f 0 . 0 3 activity ratio) than in the control group (0.49 k 0.04 activity ratio). Glycogen synthase contains multiple sites which when phosphorylated, inactivate the enzyme (Camici et a/., 1984a;Camici et al., 1984h: Iioach, 1990. PKC phosphorylates and inactivates hepatic glycogen synthase (Ahmad et a/., 1984: Roach, 1990). Thus, we have identified what we helieve is the first dietary manipulation which influences hoth the amount and activity of I'KC isozymes in any organ.
There may he differences between receptor-mediated and choline deficiency-mediated activation of I'KC. Normally PKC that has heen activated during signal transrluction is rapidly degraded, leading to down-regulation. For example. this occurs after treatment with phorbol esters (Castagna rt al., 1982). However, prolonged activation of PKC by Dl<(; is not always associated with such down-regulation (1)iaz-Laviada et Price et al., 1989). It is unclear whether the initial activation of PKC (Megidish and Mazurek, 1989) or the suhsequent down-regulation (Brooks v t al., 1991 1 is the critical event in carcinogenesis. Choline deficiency could influence PKC activity via mechanisms that do not involve accumulation of DRG. Metaholites generated from some choline-cont,aining phospholipids (sphingosine nnd Ivsosphingolipids from sphingomyelin and lvsophosphatid~lcholine from phosphatidylcholine) act as negative effectors motlulat ing PKC activation (Hannun, 1989: Hannun andHell, 198% 1,nvie et ai., 1990;Merrill and Stevens, 1989). \Ye a r e currently examining changes in sphingosine, ceramide, sphingomyelin. and phosphatidylserine during choline deficiency.2 We suggest that the accumulation of DRG and suhsequent elevation of hepatic PKC protein and activity during choline deficiency is a critical ahnormality which eventually contrih-Utes to the development of hepatic cancer in these animals. Several lines of evidence indicate that cancers might develop secondary to abnormalities in I'KC-mediated signal transduction. The phorbol esters, potent mitogens, and tumor promoters are analogs of DRC (Nishizuka, 1986). Many other mitogens also activate PKC (Pessin et al., 1990: liozengurt, 1986). Abnormalities in the expression of genes which are often associated with tumors can also he associated with alterations in DRG-and PKC-mediated pathways (Diaz-Laviada et a/., 1990; Kato ef al., 1089: Price rt al.. 1989: Weinstein, 1990Wilkison et a/., 1989;b'olfman and Macara. 1987;. Activated I'KC, in turn, may induce expression of the c-my oncogene (Kaihuchi et al.. 1986;Rozengurt. 1986). Indeed, c -m y was over expressed in liver and tumor tissue from choline deficient rat liver (Chandar et ai., 1989;Hsieh et a/., 1989). Transfection of fihrohlasts with a gene for constantly active mutant forms of PKC ( ( I or p) causes the cells to hecome transformed and tumorigenic (Krauss et al., 1989;Megidish and Mazurek, 1989: Persons rt ai., 1988). Rat liver epithelial cells which constitutively over express PKCI3I were not transformed hut did overexpress crnyc (Hsieh et al., 1989). Thus, many ohservntions suggest that increased DRG concentration with subsequent activat ion of PKC may be a key event in carcinogenesis. Although PKCmediated signal transduction has heen intensely investigated, we do not know the precise link between activation o f t h i s pathway and carcinogenesis. It is noteworthy that the isozyme which we observed to increase most is I'KC 6. Nuclear PKC' activit,y (probably the i , isozyme) in liver has heen recently characterized (Rogue et al., 1990) and it phosphorylates several nuclear proteins (including DNA met hvlt ransferase nnd ADP-ribosyltransferase). Activation of PKC also modulates the binding of proteins to cis-regulatory elements on DNA which control gene expression (Hata et al., 1989). There are several other mechanisms which have been suggested for the cancer-promoting effect of a diet devoid of choline. In the choline-deficient liver there is a progressive increase in cell death and cell proliferation (Chandar et al., 1987;Chandar and Lombardi, 1988), consistent with the effects of prolonged activation of PKC. The associated increased rate of DNA synthesis could be the cause of greater sensitivity to chemical carcinogens (Ghoshal et al., 1983). Choline deficiency also affects the genome. Hypomethylation of DNA (Dizik et al., 1991;Locker et al., 1986), abnormalities in DNA repair (Li and Randerath, 1990), and DNA damage caused by lipid peroxides (Banni et al., 1990) could contribute to the initiating events leading to carcinogenesis.
It is interesting that choline-deficient rats not only have a higher incidence of spontaneous hepatocarcinoma but that they are markedly sensitized to the effects of administered carcinogens (Newberne and Rogers, 1986). We suggest that perturbed PKC signal transduction may lower the threshold dose of carcinogen needed to initiate the development of cancers. The choline-deficient rat model could be used to elucidate the relationship between PKC signaling and chemical carcinogenesis.