Regulation of Hepatocytic Glycoprotein Sialylation and Sialyltransferases by Peroxisome Proliferators*

Short-term dietary exposure of rats to a representa- tive member of each of the three classes of peroxisome proliferators was found to elicit: (i) 71-80 and 6675% reductions in the specific activities of the hepatic p-ga-lactoside a2,6- and a2,3-sialyltransferaes, respectively; (ii) a 67439% reduction in the level of hepatic B-galacto-side a2,6-sialyltransferase protein; and (iii) 41-46 and 628% reductions in the levels of the hepatic B-galacto- side &,6- and a2,3-sialyltransferase &As, respectively. These changes were found to correlate with a re- duction in the sialylation of the N-linked glycans of a prototypical hepatocytic sialoglycocoqjugate, the inte-gral plasma membrane glycoprotein CE9, as was evident through (i) a decrease in apparent molecular mass, (ii) a conversion to a more basic distribution of isoelectric points, and (iii) 56-72 and 3344% decreases in the ability to bind lectins specific for sialic acid in a2,3- and &,6- linkage, respectively. When assessed by labeling semithin frozen sections of liver tissue with a fluorescent lectin specific for a2,6-linked sialic acid, the reduced sialylation observed for CE9 was

Sialylation is a posttranslational modification of glycoproteins implicated in the regulation of processes as diverse as receptor-mediated endocytosis, protein targeting, cell adhesion, virus-host cell recognition, and hormone signal transduction (e.g. see Ashwell and Harford, 1982;Rutishauser et al., 1988;Lasky, 1992;Weiss et al., 1988;Stockell Hartree and Renwick, 1992). Sialic acids are added to membrane and secretory glycoproteins during their posttranslational processing in the Golgi complex to become the terminal sugars on N-and 0linked oligosaccharides (Kornfeld and Kornfeld, 1985). Two sialyltransferases are responsible for adding sialic acids in a linkage-specific manner to the galactose residues of nascent complex-type N-linked oligosaccharides: the P-galactoside a2,3-sialyltransferase (2,3-ST)l and the P-galactoside a2,6-sialyltransferase (2,6-ST) (Weinstein et al., 1982b(Weinstein et al., , 1987Wen et al., 1992). We have investigated the effects of the peroxisome proliferators (PPs) on the activities and/or levels of these sialyltransferases.
*This work was supported by Grant CA53997 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The PPs are a structurally diverse group of relatively low molecular weight xenobiotic compounds that includes certain hypolipidemic drugs and phthalate-ester plasticizers (Reddy and Rao, 1986). Dietary administration of PPs to rodents elicits a complex pleiotropic response that is remarkably specific to hepatocytes. The best characterized aspects of the response are the proliferation of hepatocytic peroxisomes and smooth endoplasmic reticulum and the induction of many of their enzymatic constituents, particularly those involved in the oxidation of fatty acids (Reddy and Rao, 1986;Rao and Reddy, 1987;Hawkins et al., 1987). Despite the observation that PPs are nonmutagenic, chronic dietary exposure to these compounds causes hepatocellular carcinoma in rodents with an extremely high efficiency (Rao and Reddy, 1987).
Previously we observed that short-term dietary exposure of rats t o PPs caused the hepatocytic plasma membrane glycoprotein CE9 to migrate slightly faster in SDS-gels (Bartles et al., 1990). Encoded by a single gene and mRNA in the rat, CE9 is a widely distributed Type-Ia transmembrane protein and a member of the immunoglobulin superfamily (Nehme et al., 1993). When expressed by the rat hepatocyte, CE9 exhibits an apparent molecular mass of 48 kDa, contains three N-linked glycans, and is concentrated within the basolateral plasma membrane domain (Hubbard et al., 1985;Bartles et al., 1985;Nehme et al., 1993). We determined that the PP-induced difference in the electrophoretic mobility of CE9 could be eliminated by prior chemical deglycosylation, but was not yet apparent when comparing pulse-radiolabeled high-mannose precursors (Bartles et al., 1990). Thus, we tentatively concluded that CE9 experienced an altered pattern of posttranslational glycosylation in the hepatocytes of PP-treated rats. In this article, we demonstrate that dietary exposure to the PPs brings about a reduction in the sialylation of CE9 and other hepatocytic glycoconjugates and that this reduction in sialylation mirrors decreases in the specific activities and/or levels of the relevant hepatic sialyltransferases.
Immunoprecipitation, Glycosidase Deatments, and Tho-dimensional ElectrophoresisCES was quantitatively immunoprecipitated from Triton X-100/n-octyl-P-o-glucopyranoside extracts of rat liver homogenates in the presence of protease inhibitors using either mouse monoclonal or rabbit polyclonal anti-CE9 IgG-Sepharose with equivalent results (Bartles et al., 1987). Complete desialylation of the immunoprecipitated CE9 required sequential incubation with neuraminidases isolated from Clostridium perfringens and Athrobacter ureafaciens. CE9 was treated first with 2 unitdml of C. perfringens neuraminidase in 50 m~ sodium acetate, 10 m~ calcium acetate, 3 m M sodium azide, pH 5.5, for 4 h at 37 "C while still attached to the immunoadsorbent beads. After recollecting the beads by microcentrifugation, the partially desialylated CE9 was quantitatively removed from the beads by heating in 0.5% (w/v) of SDS at 100 "C for 3 min. The eluted protein was then treated with 2 unitsfml ofA. ureafaciens neuraminidase in 20 m M sodium acetate, 2.5% (v/v) of Nonidet P-40,0.16% (wh) of SDS, 3 m M sodium azide, pH 6.0, for 18 h at 37 "C before running in two-dimensional gels. For enzymatic deglycosylation using N-glycanase, the immunoprecipitated CE9 was eluted from the beads by heating in 0.5% (w/v) of SDS at 100 "C for 3 min and then treated with 10 unitsfml of N-glycanase in 0.2 M sodium phosphate, 0.03 M 2-mercaptoethanol, 1 m~ 1,lO-phenanthroline, 1.25% (v/v) of Nonidet P-40,0.16% (wh) of SDS, 3 m M sodium azide, pH 8.6, for treated CE9 were heated at 100 "C in 1% (w/v) of SDS and 5% (w/v) of 24 h at 37 "C. Samples of liver homogenates or immunoprecipitated and 2-mercaptoethanol. Nonidet P-40 and urea were added to final concentrations of 7% (vh) and 0.25 g/ml, respectively, and the samples were the method of O'Farrell (1975) in the pH ranges of 4-6 (untreated) or resolved in two-dimensional isoelectric focusing SDS-gels according to 5-7 (treated). Blotting-Western blotting, lectin blotting, and Northern blotting were performed under conditions that were established empirically to give a linear or near-linear response as a function of input over the concentration range of interest (e.g. see Bartles et al., 1991). Protein samples resolved in one-or two-dimensional SDS-gels were transferred electrophoretically to nitrocellulose. Western blots were labeled sequentially with affinity-purified rabbit polyclonal antibodies directed against either rat CE9 or rat liver 2,6-ST and lZ5I-protein A, and the relative levels of lZ5I-protein A binding were determined using an Pharmacia LKB Ultroscan XL laser densitometer to scan autoradiograms (Bartles et al., 1991). When quantifying the 2,6-ST on Western blots, gels were loaded with equal amounts of total homogenate protein as determined using a modified Lowry assay (Markwell et al., 1978). For lectin blotting, samples containing nearly equal amounts of immunoprecipitated CE9, as determined by Western blotting, were electrophoresed in SDS-gels. The corresponding blots were labeled sequentially with digoxigenin-labeled SNA or MAA and anti-digoxigenin-alkaline phosphatase conjugate, and the color reaction was developed by incubation with 4-nitro blue tetrazolium chloride and 5-bromo-4-chloro-3indolylphosphate according to the manufacturer's recommendations (Glycan Differentiation Kit; Boehringer Mannheim). The relative levels of lectin binding were determined using a Bio-Rad model GS-670 imaging densitometer to scan the blots in the reflective mode and were normalized on the basis of CE9 content as determined by Western blotting. Total RNA was isolated from frozen chunks of liver (Chomczynski and Sacchi, 1987). Samples containing 35-pg of RNA were electrophoresed in formaldehyde-denaturing 1% agarose gels, transferred to nitrocellulose by capillary action, and labeled with randomly primed [32Pl~DNAs (Sambrook et al., 1989) encoding the 2,3-ST or 2,6-ST. The relative levels of bound [32PlcDNA were determined using a Pharmacia LKB Ultroscan XL laser densitometer to scan the corresponding autoradiograms and were normalized to the level of 28 S rRNA as determined subsequently on the same blots using an end-labeled 32P-labeled oligonucleotide probe (Sambrook et al., 1989).
Sialyltransferase Assays-All enzyme assays were performed under conditions of substrate excess and were tested to ensure that the generation of product was linear as a function of time and enzyme input. Assays were performed on nonionic detergent extracts of rat liver according t~ a modification of the procedure described by Weinstein et al. (1982a). To assay for 2,3-ST activity, a frozen chunk of liver was thawed and homogenized in 3.3 mug of 50 m~ sodium cacodylate, 20 rn MnCIZ, 0.5% (v/v) Triton CF-54, pH 6.0. Following extraction for 1 h at 4 "C, the mixture was centrifuged for 16 min at 13,000 x g in an Eppendorf microcentrifuge at 4 "C. Reaction mixtures of 60 pl total volume were prepared using 40 pl of supernatant, 30 pg of lacto-N-tetraose, and 9 nmol of CMP-[14C1NeuAc (6200 cpdnmol) and were incubated for 15 min at 37 "C. Reactions were stopped by dilution into 1 ml of ice-cold 5 rn sodium phosphate buffer, pH 6.8, followed by immediate application to a 3-cm-high column of Dowex 1-X8 in a Pasteur pipette. The column was rinsed with an additional 1 ml of this same buffer. The flow-through was collected directly into scintillation vials, and radioactivity was measured in a liquid scintillation counter. Enzyme activities were normalized to the amount of total protein present in the liver homogenates as determined by a modified Lowry protein assay (Markwell et al., 1978). To our knowledge, there is no substrate specific for the 2,6-ST. Therefore to determine the specific activity of the 2,6-ST, it was necessary to first assay for the 2,6-ST and 2,3-ST combined using a procedure identical to that described above, with the exception that 480 pg of N-acetyllactosamine, was substituted for lacto-N-tetraose, and reaction mixtures were incubated for 10 min at 37 "C. At the relatively high concentration utilized (21 mM), N-acetyllactosamine is sialylated at approximately the same rate by these two enzymes (Weinstein et al., 1982b). The specific activity of the 2,6-ST was then estimated by taking into account both the activity measured for the 2,3-ST using 1acto-Ntetraose as a substrate and the observation that the 2,6-ST is 3.5 times more prevalent in the livers of normal rats . sections were obtained from livers fixed with 2% paraformaldehyde/ Lectin Fluorescence and Zmmunofluorescence-1.5-pm-thick frozen lysine/periodate by perfusion and labeled according to the procedure outlined by Bartles et al. (1990), substituting 2% (w/v) polyvinylpyrrolidone (average M,, 40,000) for gelatin as the blocking agent when labeling with the lectins. For lectin fluorescence, the sections were labeled with 10 pg/ml of fluoresceinated SNAor MAA. For immunofluorescence, the sections were labeled with affinity-purified rabbit polyclonal antibody to the rat liver 2,6-ST or nonimmune rabbit IgG followed by rhodaminated goat anti-rabbit IgG. The specimens were examined and photographed using a Leitz Diaplan fluorescence microscope.

RESULTS
When examined in two-dimensional isoelectric focusing SDSgels, the hepatocytic plasma membrane glycoprotein CE9 present in the liver homogenate of control rats was observed to focus as a constellation of seven to nine partially resolved spots in the pH range of-4-5 (Fig. li). Ten-day dietary exposure to a PP caused the constellation to become more basic, yet comparable numbers of spots were retained. Fig. 1, j-Z, illustrates the typical results obtained for three different PPs: the class 1 hypolipidemic drug ciprofibrate, the class 2 hypolipidemic drug Wy-14,643 and the phthalate-ester plasticizer DEHP. At the concentrations tested, the effect appeared slightly more pronounced for ciprofibrate and Wy-14,643 than for DEHP, thus mirroring the relative potencies of these compounds a t eliciting other aspects of the pleiotropic response (Reddy and Lalwani, 1983). Identical results were obtained when comparing the behavior of CE9 isolated by immunoprecipitation from nonionic detergent extracts of the liver homogenates (data not shown). The constellations were converted to three common more basic spots focusing near a pH of -6 following sequential treatment  N-glycanase  (N-GLY). Rats were fed a control diet (CON, a, e, and i) or a diet containing ciprofibrate (CIO, b, f, a n d j ) , DEHP (DIO, c, g, and k), or Wy-14,643 (W10, d, h, and 1 ) for 10 days. Samples containing CE9 were resolved in two-dimensional isoelectric focusing SDS-gels, and the CE9 was revealed by Western blotting. The direction of isoelectric focusing was from left to right (-, (Fig. 1, e-h), suggesting that the shift to more basic isoelectric points was due to a reduction in the level of sialylation of CE9 in the livers of the PP-treated rats. Patterns consisting of three common spots were also observed when the CE9 immunoprecipitated from the livers of control and PP-treated rats were stripped of their N-linked glycans by enzymatic deglycosylation using N-glycanase ( Fig. 1, a d ). The persistence of multiple spots upon enzymatic desialylation or deglycosylation most likely reflects the differential phosphorylation of the CE9 protein.2 Lectin blotting was used to quantify the levels of sialic acid on CE9 isolated by immunoprecipitation from the livers of control and PP-treated rats. The 02,3-linked and the a2,6-linked sialic acids most commonly found as part of complex-type Nlinked oligosaccharides (Kornfeld and Kornfeld, 1985) were quantified using the linkage-specific lectins MAA (Wang and Cummings, 1988) and SNA (Shibuya et al., 1987), respectively. As reported previously (Bartles et al., 19901, the CE9 obtained from the livers of PP-treated rats was observed to migrate slightly faster in SDS-gels (Fig. 2). When normalized on the basis of CE9 protein, the binding of MAA was decreased to 28 2 6%, 44 f 8%, and 37 4% of controls, and the binding of SNA was decreased to 56 2 lo%, 67 2 ll%, and 67 2 12% of controls in the livers of rats fed the ciprofibrate, DEHP, or Wy-14,643 diets, respectively (Fig. 2). In each case, the binding of the lectin was shown to be specific for sialic acid, because neither lectin bound to CE9 following enzymatic desialylation with C. perfringens and A. ureafaciens neuraminidases (data not shown).
To help determine whether the PP-mediated decrease in sialylation observed for CE9 might extend to other hepatocytic glycoconjugates, fluorescently tagged versions of these same two sialic acid-binding lectins were used to label semithin frozen sections of liver. When sections obtained from the livers of control rats were labeled with fluoresceinated SNA, specific fluorescence signals were found at the surfaces of both hepatocytes and sinusoidal-lining cells (Fig. 3, a and b). With the resolution afforded by immunofluorescence, it was not possible to distinguish the relative labeling contributions of plasma membrane glycoconjugates per se from those present within K. S. Hospodar and J. R. Bartles, unpublished data. the surrounding extracellular matrix or subplasmalemmal cytoplasm. Both the basal (sinusoidal) and apical (bile canalicular) surfaces ofhepatocytes were labeled brightly relative to the lateral surfaces between adjacent hepatocytes, perhaps as a result of limited access. In addition, there was bright specific labeling a t certain intracellular sites within hepatocytes that generally exhibited the size, shape, and localization expected for elements of the Golgi complex (Roth et al., 1985). When fixation, processing, labeling, and photography were carried out in parallel under identical conditions, the binding of fluoresceinated SNA to sections obtained from the livers of ciprofibrate-treated rats was found to be decreased, both a t the hepatocyte surface and at intracellular sites within hepatocytes (Fig. 3, c and d ). Such a decrease was not observed for the surfaces of the sinusoidal-lining cells (Fig. 3, arrowheads), which are generally thought not to be affected by the PPs (Reddy and Rao, 1986). Similar overall results were obtained when examining sections obtained from the livers of rats treated with DEHP or Wy-14,643 (data not shown). These observations suggested that decreased derivatization with a2,6linked sialic acid may extend to a variety of cell surface and intracellular glycoconjugates in the hepatocytes of PP-treated rats. Unfortunately, the levels of labeling observed using fluoresceinated MAA were too low to allow such a comparison to be made for a2,3-linked sialic acids (data not shown).
To examine the basis of the effects of PPs on the sialylation of CE9 and other hepatocellular glycoconjugates, rats were fed diets containing one of the three PPs for 10 days, and the specific activities of the two major hepatic glycoprotein sialyltransferases were assayed. The specific activity of the hepatic 2,3-ST was found to be reduced to 25-34% of control levels in the livers of rats fed the PPs (Table I). Likewise, the specific activity of the hepatic 2,6-ST was found to be reduced to 20-29% of control levels in the livers of PP-treated rats (Table I). When assayed as a function of time after initiating dietary treatment with ciprofibrate, 2-5 days were required for the decreases in the specific activities of the sialyltransferases to reach one-half those observed after 10 days of treatment (data not shown).
To determine whether the decreases in the specific activities of the sialyltransferases observed upon PP treatment reflected decreased levels of enzyme, Western blotting was used to quantify the levels of 47-kDa 2,6-ST protein (Weinstein et al., 1987)  in the livers of control and PP-treated rats. When normalized on the basis of total homogenate protein, the level of 2,6-ST protein was found to be decreased to 31-33% of control levels in the livers of PP-treated rats (Table I). To our knowledge, there are no antibodies available that would allow similar quantification of the levels of the rat liver 2,3-ST.

CON C10 Dl0
Immunofluorescence was used to compare the localization of the 2,6-ST in the livers of the control and PP-treated rats. Semithin frozen sections of liver were prepared, immunolabeled, and photographed in parallel under identical conditions. Consistent with the localization of the 2,6-ST to the trans-Golgi network in normal hepatocytes (Roth et al., 19851, specific immunostaining was observed over multiple foci within the hepatocyte cytoplasm, often near bile canaliculi (Fig. 4, a and b).
A qualitatively similar pattern of labeling was observed for sections obtained from the livers of rats fed ciprofibrate (Fig. 4,  c and d ) and the other two PPs (data not shown). But, as expected from the results of Western blotting (Table I), the

s by Peroxisome Proliferators
intensity of labeling was found to be reduced for those sections obtained from the livers of the PP-treated rats, thus causing the foci to also appear somewhat smaller (cf. Fig. 4, a and c ) .
There was, however, no evidence of a gross redistribution of the 2,6-ST protein in the hepatocytes of rats fed the PPs. By Northern blotting, the decrease in the specific activity of the 2,6-ST and the decrease in the level of the 2,6-ST protein were found to reflect consistent decreases in the level of hepatic 2,6-ST mRNA. Following normalization to the level of 28 S rRNA, the level of 4.3/4.7-kilobase 2,6-ST mRNA  was found to be decreased to 54-59% of control levels in total RNA preparations isolated from the livers of rats fed the PPs (Table I). In contrast to the consistent decrease observed for the 2,6-ST mRNA, the effect of the PPs on the level of the 2,3-ST mRNA was smaller and considerably more variable. Following normalization to the level of 28 S rRNA, the level of 2.5-kilobase 2,3-ST mRNA (Wen et al., 1992) was found to be decreased to 72-94% of control levels in total RNA preparations isolated from the livers of PP-treated rats ( Table I). DISCUSSION Three possible explanations for the observed PP-mediated decrease in the sialylation of CEYs N-linked glycans are: (i) a decrease in the number of glycans, (ii) a failure to complete the addition of terminal sugar residues, or (iii) a decrease in the branching of the glycans. On the basis of cDNA sequence, there are three Asn-X-SerfThr consensus sites for N-linked glycosylation predicted to reside within the extracellular domain of CE9 (Nehme et al., 1993). The observation of three products upon partial deglycosylation with N-glycanase substantiates the existence of three N-linked glycans on rat hepatocytic CE9 (Nehme et al., 1993). It appears as though three N-linked glycans must also be present on CE9 in livers of PP-treated rats, because pulse-radiolabeled high-mannose precursors of CE9 from livers of control and PP-treated rats were observed to comigrate in SDS-gels (Bartles et al., 1990). Edlund et al. (1986) observed a 31-34% decrease in the rate of glycosylation of endogenous proteins by dolichol monophosphate-mediated UDP-glucosaminyl-and GDP-mannosyltransferases in microsomal fractions prepared from the livers of rats fed a diet containing DEHP for 2 weeks. Our data suggest that such a change does not translate into a reduction in the number of N-linked glycans, at least in the case of the plasma membrane protein CE9. As for the possibility of incomplete processing, surplus terminal galactose residues were not detected on CE9 isolated from the livers of PP-treated rats using a sensitive digoxigenin-Ricinus communis agglutinin-I-binding assay." Furthermore, CE9 isolated from the livers of control and PP-treated rats proved to be resistant to digestion by endoglycosidase H, suggesting that the N-linked glycans of both forms of the protein have been processed significantly beyond their high-mannose precursors.3 On the basis of these additional observations, the most likely explanations are that either: (i) the terminal processing of some of the branches of CE9's N-linked glycans is aborted in the livers of PP-treated rats, but prior to the addition of galactose, or (ii) there is less branching of CEYs N-linked glycans in the livers of PP-treated rats. An interruption in the terminal processing prior to the addition of galactose would be expected to yield glycans with terminal N-acetylglucosamine residues (Kornfeld and Kornfeld, 1985). The existence of such truncated glycans may explain the observation that CE9 isolated from the livers of PP-treated rats binds disproportionately larger amounts of wheat germ agglutinin, a lectin specific for both sialic acid and N-acetylglucosamine (Bartles et al., 1990).
B. E. Fayos and J. R. Bartles, unpublished data. To determine the specific activities of the 2,3-ST or 2,6-ST, nonionic detergent extracts o f liver homogenates were assayed using the exogenous substrates lacto-Ntetraose or N-acetyllactosamine as described under "Experimental Procedures." The data were normalized on the basis of total homogenate protein and are reported as mean 2 S.D. (triplicate determinations on three rats of each type) relative to values of 100% for the corresponding controls. To determine the level of 2,6-ST protein, samples o f liver homogenates were resolved in SDS-gels, and the resultant blots were labeled sequentially with affinity-purified anti-2,B-ST antibody and I2"I-protein A. The levels of antibody binding were determined by densitometry, were normalized on the basis of total homogenate protein, and are reported as mean t S.D. (duplicate determinations on three rats of each type) relative to a value of 100% for the control. To determine levels of mRNA encoding the 2,3-ST and 2,6-ST, samples of total liver were labeled with randomly primed 2,3-ST or 2,6-ST fT2PlcDNA. The RNA were resolved in 1% agarose gels, and the resultant Northern blots levels of cDNA binding were determined by densitometry, were normalized on the basis of 28 S rRNA, and are reported as mean * S.D. (single or duplicate (C10) determinations on three rats of each type) relative to values of 100% for the corresponding controls. 3 4 t 8 2 4 2 4 ND" 3 3 2 2 9 4 k 3 55 k 3 W10 2 5 2 7 2 9 2 5 ND" 3 1 k 2 81 k 7 5 9 k 4 " Protein levels were not determined because anti-2,3-ST antibodies were not available.
CE9 is a basolateral plasma membrane protein of hepatocytes, both in control and in PP-treated rats (Bartles et al., 1990). Yet the decrease in labeling by fluoresceinated SNA (Fig.  3c) was found to apply to both the basal and apical surfaces of hepatocytes as well as to their intracellular compartments. Therefore, the decrement in sialylation noted for CE9 appears to extend to other hepatocellular glycoconjugates in the hepatocytes of PP-treated rats. Even though CE9 is synthesized a t 1.7 times the normal rate3 and is induced 1.8-fold in the livers of ciprofibrate-treated rats (Bartles et al., 19901, there is no evidence of an intracellular accumulation of CE9 by immunofluorescence (Bartles et al., 1990). This suggests that transport through the hepatocytic secretory pathway is neither blocked nor slowed to a great extent by treatment with PPs. Additional support comes from the observations that six other hepatocytic plasma membrane proteins continue to be sent to their correct surface domains in PP-treated rats (Bartles et al., 1990) and that there is no drastic change in the localization of the hepatocytic 2,6-ST (Fig. 4, c and d ) .
A likely explanation for these defects in the sialylation of CE9 and other hepatocytic glycoconjugates lies in the observation that the PPs, irrespective of class, were found to elicit substantial reductions in the specific activities of the two major sialyltransferases involved in the terminal processing of hepatocytic N-linked oligosaccharides (Table I). In the case of the 2,6-ST, these reductions in specific activity could be completely accounted for by comparable decreases in the level of 2,6-ST protein (Table I). Given the precedent for the regulation of 2,6-ST activity through changes in the level of its corresponding mRNA (Wang et al., 1989(Wang et al., ,1990Svensson et al., 1990;Shah et al., 1992;Grollman et al., 1993), the observed reduction in the level of 2,6-ST mRNA (Table I) is most likely responsible for the decreases in the level and specific activity of the 2,6-ST in the livers of PP-treated rats. It is presently unclear how the PPs might act to alter the level of 2,6-ST mRNA. Regrettably, the magnitude of the effect at the level of the mRNA is sufficiently small so as to make it difficult to distinguish between the options of decreased transcription and decreased mRNA stability by experimental means, especially without the benefit of a model cell culture system. While the information available concerning the so-called PP-activated receptor suggests that PPs may directly increase the rate of transcription of certain genes (Isseman and Green, 1990;Kliewer et al., 1992), no instances of the PPs eliciting a decrease in the rate of transcription have yet been documented. A subset of hepatocytic plasma membrane proteins is also known to be expressed at a lower level in the livers of PP-treated rats (Bartles et al., 1990). Our observations reinforce the notion that many profound physi- ological and biochemical changes are occurring in the livers of PP-treated rats. These changes may reflect a shift in cellular emphasis toward the biosynthesis of peroxisomal constituents at the expense of the biosynthesis and maintenance of the secretory pathway and its organelles.
Although the PPs may exert their effects on the specific activity and level of the 2,6-ST by affecting the rate of transcription or stability of the hepatocytic 2.6-ST mRNA, it would seem to be considerably more difficult to invoke such a n explanation in the case of the 2,3-ST. The reductions observed in the levels of 2,3-ST mRNA were not only more modest, but were found to vary considerably among the PPs (Table I), despite the uniformity and magnitude of the effects of these compounds on the specific activity of the 2,3-ST (Table I) and the binding of MAA to CE9 (Fig. 2). Thus, although the net effects of the PPs on the specific activities of the two sialyltransferases were similar, there is a distinct possibility that these agents will prove to affect the 2.3-ST and 2,6-ST, and hence the sialylation of hepatocytic glycoproteins, by alternate pathways. The elucidation of the basis for the PP-mediated reduction in the specific activity of the hepatic 2,3-ST awaits further experimentation. In preliminary experiments, we have failed to detect a direct inhibitory effect of ciprofibrate on the activity of the 2,3-ST when assayed at final concentrations as high as 0.3 mM in nonionic detergent extracts of rat liver homogenate.3 But this by no means rules out the possibility that some metabolite of the PPs or some cellular change brought about by exposure to the PPs might somehow affect the activity or stability of the 2,3-ST protein.
Regardless of their mechanism of action, the data reported here indicate that short-term dietary exposure to the PPs can bring about significant changes in the sialylation of hepatocytic glycoconjugates and that these changes reflect decreases in the specific activities and/or levels of expression of the hepatic glycoprotein sialyltransferases. Given the pivotal roles identified for the sialic acid residues of glycoproteins, this newly described aspect of the pleiotropic response to dietary PPs may prove to have a profound influence on the activities, localizations, and/or stabilities of the affected hepatocytic membrane and secretory glycoproteins.