Bile Acid Induction of Cytokine Expression by Macrophages Correlates with Repression of Hepatic Cholesterol 7α-Hydroxylase*

In the studies reported herein, we show that two complementary experimental models: inbred strains of mice (i.e. C57BL/6 and C3H/HeJ), and a differentiated line of rat hepatoma cells (i.e. L35 cells), require the activation of cytokines by monocyte/macrophages to display bile acid negative feedback repression of cholesterol 7α-hydroxylase (CYP7A1). Feeding a bile acid-containing atherogenic diet for 3 weeks to C57BL/6 mice led to a 70% reduction in the expression of hepatic CYP7A1 mRNA, whereas no reduction was observed in C3H/HeJ mice. The strain-specific response to repression of CYP7A1 paralleled the activation of hepatic cytokine expression. Studies using cultured THP-1 monocyte/macrophages showed that the hydrophobic bile acid chenodeoxycholate, a well established potent repressor of CYP7A1, induced the expression of mRNAs encoding interleukin 1 (IL-1) and tumor necrosis factor α (TNFα). In contrast, the hydrophilic bile acid ursodeoxycholate, which does not repress CYP7A1, did not induce cytokine mRNA expression by THP-1 cells. Chenodeoxycholate activation of cytokines by THP-1 cells was blocked by the peroxisome proliferator-activated receptor γ agonist rosiglitazone. The expression of cytokines (e.g. IL-1 and TNFα) by THP-1 cells paralleled with the ability of these cells to produce conditioned medium that when added to rat L35 hepatoma cells, repressed CYP7A1. Moreover, rosiglitazone, which blocks cytokine activation by macrophages, also blocked the repression of CYP7A1 normally exhibited by C57BL/6 mice fed the bile acid-containing atherogenic diet. The combined data indicate that the activation of cytokines may mediate CYP7A1 repression caused by feeding mice an atherogenic diet containing bile acids.

Bile acids, the major metabolites produced from cholesterol, are amphipathic steroid detergents necessary for the digestion and absorption of fat soluble nutrients from the intestine (1)(2)(3). The conversion of cholesterol to bile acids is regulated by the expression of cholesterol 7␣-hydroxylase (CYP7A1), 1 a cytochrome P450 enzyme unique to the liver parenchymal cell (4 -6). Bile acid synthesis exhibits negative feedback regulation (7,8) by decreasing the enzymatic activity of CYP7A1 (9). It is generally accepted that bile acids can inhibit the transcription of the CYP7A1 gene (1)(2)(3).
Many different experimental models have been used to examine bile acid negative feedback regulation of CYP7A1 and some have yielded conflicting results. Bile acid negative feedback repression of CYP7A1 has been experimentally demonstrated by infusing bile acids into the intestine of bile fistulae rats (10) and hamsters (11). The ability of different bile acids to repress CYP7A1 expression correlates with the hydrophobic index of the infused bile acid; CDCA is a potent repressor, whereas UDCA is not (12). The finding that infusing taurocholate into the portal vein of bile fistulae mice was unable to repress CYP7A1 led to the conclusion that a factor produced within the enterohepatic circulation may be required to repress CYP7A1 (10).
Bile acid repression of CYP7A1 has been demonstrated using primary cultured rat hepatocytes (13) and human hepatoma HepG2 cells (14 -16). Data from these cultured cell studies suggest that multiple mechanisms exist in regard to bile acid repression of CYP7A1 expression. These mechanisms include: "bile acid response" elements (BARE) (17), activation of protein kinase C (18), and activation of the farnesoid X receptor (FXR) (16,19).
L35 is a stable line of rat hepatoma cells that have been used for studies examining the expression of CYP7A1 (20 -22). L35 cells express CYP7A1 at levels equal to that of rat liver, which is 10-fold greater than the levels expressed by either HepG2 cells or primary rat hepatocytes (20). Moreover, with the one notable exception of resistance to repression by bile acids, the expression of CYP7A1 by L35 cells responded normally to essentially all the effectors established to alter CYP7A1 expression in vivo (20 -22). The inability of bile acids to repress CYP7A1 expression by L35 cells led to the proposal that they are missing factors necessary to mediate this repression (21).
In the studies reported herein, we show that these factors are cytokines produced by macrophages.

MATERIALS AND METHODS
Mouse Studies-Female C3H/HeJ and C57BL/6 mice 10 -12 weeks old were obtained from Jackson Laboratory, Bar Harbor, ME. The mice were housed in a room with a normal light cycle (lights on from 6 a.m. to 6 p.m.) were fed either normal Purina breeder chow or ground Purina breeder chow supplemented with 20% olive oil, 2% cholesterol, and 0.5% taurocholic acid (bile acid-containing atherogenic diet) and water ad libitum. Mice were maintained on the above diets for 3 weeks.
In the experiments examining the effect of rosiglitazone on CYP7A1 * This work was supported by National Institutes of Health Grants HL57974 and HL57974. 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.
Hepatic cytokine mRNAs were quantitated using RNase protection assays. In vitro transcribed [␣Ϫ 32 P]UTP-labeled antisense cytokine probes were generated using cytokine multiprobe template kits: mouse mCK-2 (catalog number 45002P) and mouse mCK-3 (catalog number 45003P; PharMingen International) and a MAXIscript in vitro transcription kit (catalog number 1314) using T7 RNA polymerase per the manufacturer's instructions. The content of human cytokines mRNAs was also quantitated by RNase protection assays except human template kits were used (human hCK-2, catalog number 45032P and human hCK-3, catalog number 45033P; PharMingen International).

RESULTS AND DISCUSSION
When fed the bile acid-containing atherogenic diet, C57BL/6 mice display repression of CYP7A1, whereas C3H/HeJ mice do not (23)(24)(25). Quantitative trait loci analysis of C3H/HeJ and C57BL/6 mice shows that marked phenotypic differences exist in regard to displaying inflammation in response to consuming the bile acid-containing atherogenic diet (26,27). C3H/HeJ display essentially a complete resistance to hepatic inflammation, whereas C57BL/6 display a remarkable susceptibility (26,27). We examined if strain-specific differences in cytokine ac-tivation might be the basis for the strain-specific differences in CYP7A1 repression. On the normal chow diet, the expression of CYP7A1 mRNA expression was similar in C57BL/6 and C3H/ HeJ mice (Fig. 1, A and B). In contrast, the bile acid-containing atherogenic diet caused marked differences in the expression of CYP7A1 by the two strains of mice. While C3H/HeJ mice displayed no significant change in CYP7A1 expression, C57BL/6 mice showed a marked 70% decrease, p Ͻ 0.01 (Fig. 1, A and B).
The individual strains also displayed distinct differences in the response of hepatic cytokine expression to the bile acidcontaining atherogenic diet (Fig. 1C). In C57BL/6 mice, the bile acid-containing atherogenic diet increased the hepatic expression of mRNAs encoding IL-1␣ (7-fold, p Ͻ 0.01), IL-1␤ (4-fold, p Ͻ 0.01), TNF␣ (3-fold, p Ͻ 0.01), IFN␤ (6-fold, p Ͻ 0.01), and TGF-␤1 (7-fold, p Ͻ 0.01) (Fig. 1C). In marked contrast, the expression of hepatic cytokines by C3H/HeJ mice was unaffected by the bile acid-containing atherogenic diet (Fig. 1C). The concordance between the ability of the bile acid-containing atherogenic diet to induce the expression of mRNAs encoding cytokines while repressing CYP7A1 mRNA expression suggested the possibility that cytokines might mediate the repression of CYP7A1 caused by the bile acid-containing atherogenic diet. Indeed, recent studies have shown that administering lipopolysaccharide as well as the cytokines TNF␣ or IL-1 to hamsters resulted in a marked suppression of CYP7A1 (28).
Based on these combined findings, we formulated the following experimentally testable model (Fig. 1D). Following their active absorption in the distal intestine, bile acids return to the liver via the portal vein entering the hepatic parenchymal cell by crossing through the sinusoids. As bile acids move across the sinusoids, they may interact with resident macrophages (i.e. Female C57BL/6J and C3H/HeJ mice were housed in a room with a normal light cycle (lights on from 6 a.m. to 6 p.m.) and fed water ad libitum with either a chow diet or a bile acid-containing atherogenic diet. After 3 weeks, mice were sacrificed at 9 a.m. A, livers were extracted for RNA, and poly(A) RNA was isolated, blotted onto nitrocellulose, and probed with 32 P-labeled cDNA encoding rat CYP7A1 and GAPDH as a loading control. B, the abundance of CYP7A1 mRNA relative to GAPDH is shown as the mean Ϯ S.D. of three individual mice in each group. The asterisk denotes a significant difference between the C57BL/6 mice fed the chow diet and C57BL/6 mice fed the bile acid-containing atherogenic diet (p Ͻ 0.01). C, the hepatic content of the indicated cytokine mRNAs was determined using an RNase protection assay. L32 and GAPDH are RNA loading controls. D, based on the findings showing that dietary bile acids alter the expression of CYP7A1 in a manner that is inversely related to their activation of hepatic cytokines, we propose the model schematically represented in this figure. This model predicts that bile acids returning to the liver via the portal blood interact with hepatic macrophages (Kupffer cells) as they traverse the sinusoidal surface to enter the parenchymal cell. The type of bile acid and its concentration in portal blood determines whether cytokine expression by macrophages is increased. The ability of bile acids to induce the expression of regulatory cytokines subsequently determines whether CYP7A1 will be repressed by this mechanism.
Kupffer cells) which reside along the sinusoidal surface. At sufficient concentration, bile acids may cause the activation of cytokines by Kupffer cells. These regulatory cytokines may subsequently act on hepatic parenchymal cells, leading to the repression of CYP7A1 (28).
We attempted to reconstruct this model using the cultured rat hepatoma cell line (L35 cells) and human monocyte/macrophages (THP-1 cells). To approximate the intercellular relationships that may exist between hepatic macrophages and parenchymal cells (Fig. 1D), THP-1 cells were exposed to bile acids and the effects of the conditioned medium was examined on the expression of CYP7A1 by L35 cells. CYP7A1 expression by L35 cells was unaffected by changing the culture medium to serum-free DMEM, without dexamethasone but containing either 0.1% BSA or 0.1% BSA containing the hydrophobic bile acid CDCA (100 M) or conditioned medium obtained from THP-1 cells (Fig. 2A). However, changing the cultured medium to serum-free DMEM, without dexamethasone but containing conditioned medium obtained from THP-1 cells exposed to CDCA, repressed CYP7A1 expression by Ͼ70% (Fig. 2A). These data indicate that: 1) CDCA requires THP-1 cells to repress CYP7A1 expression by L35 cells, and 2) CDCA caused THP-1 cells to secrete a factor that repressed CYP7A1.
According to our hypothesis (Fig. 1D) cytokines mediate the repression of CYP7A1 caused by dietary bile acids. Thus, our model predicts that blocking the activation of cytokines by bile acids should block the repression of CYP7A1. PPAR␥ agonism inhibits the production of inflammatory cytokines by monocyte/ macrophages (29). Therefore, if our model is valid, the PPAR␥ agonist rosiglitazone should prevent repression of CYP7A1. Treating THP-1 cells with rosiglitazone completely blocked the ability of CDCA to induce the expression of cytokine mRNAs (Fig. 2B). Rosiglitazone also blocked the ability of THP-1 cells exposed to CDCA to produce conditioned medium that could repress CYP7A1 expression by L35 cells (Fig. 2C). The additional finding that TNF␣ caused a dose-dependent decrease in the expression of CYP7A1 mRNA by L35 cells (Fig. 2D) further indicates that cytokines produced by THP-1 cells in response to CDCA are responsible for repression of CYP7A1.
Further analysis showed that L35 cells did not express detectable levels of mRNAs encoding the regulatory cytokines TNF␣, TGF-␤1, or IL-1␤ (data not shown). Treatment of L35

FIG. 2. The expression of cytokine mRNAs by THP-1 cells correlates with the ability of conditioned medium to repress CYP7A1 mRNA expression by rat hepatoma L35 cells.
A, CDCA requires THP-1 cells to repress CYP7A1 expression by L35 cells. Rat hepatoma L35 cells were cultured in serum-free medium containing 100 M dexamethasone for 24 h. The cultured medium was then changed to serum-free DMEM medium containing the following additions: 0.1% BSA (Control), 0.1% BSA containing 100 M CDCA (CDCA), the addition of 50% by volume of medium from THP-1 cells, which was incubated for 48 h with either 0.1% BSA (THP-1) or 0.1% BSA containing 100 M CDCA (THP-1 ϩ CDCA). After 24 h, cells were harvested, and poly(A) RNA was isolated, blotted onto nitrocellulose, and probed sequentially with 32 P-labeled cDNAs encoding rat CYP7A1 and ␤-actin. B, CDCA, but not UDCA, induces the expression of cytokine mRNA by THP-1 cells via a process that is blocked by the PPAR␥ agonist rosiglitazone. THP-1 cells were cultured in RPMI 1640 medium containing 10% FBS and then treated with either 0.1% BSA (lanes 1 and 5), 0.1% BSA containing 100 M UDCA (lanes 2 and 6), 0.1% BSA containing 100 M CDCA (lanes 3 and 7), or 0.1% BSA containing 100 M CDCA and 500 nM rosiglitazone (BRL) (lanes 4 and 8). RNA was isolated and subjected to RNase protection assay for the indicated cytokines. L32 and GAPDH are RNA loading controls. C, the PPAR␥ agonist rosiglitazone blocks the CDCA-induced repression of CYP7A1 in L35 cells by conditioned medium from THP-1 cells. Rat hepatoma L35 cells were cultured in serum-free medium containing 100 M dexamethasone for 24 h. The cultured medium was then changed to serum-free DMEM containing the following additions (50% by volume): conditioned medium obtained from THP-1 cells incubated with 0.1% BSA (THP-1) or 0.1% BSA containing 100 M CDCA or 0.1% BSA containing 100 M CDCA and 500 nM rosiglitazone (BRL). After 24 h, cells were harvested, and the relative level of CYP7A mRNA to ␤-actin mRNA was quantitated. Each value represents the mean of duplicate plates of cells. D, TNF-␣ represses the expression of CYP7A1 mRNA by L35 cells. L35 cells cultured in serum-free DMEM medium containing 100 M dexamethasone were treated with the indicated concentrations of human TNF-␣ for 24 h. Each value represents the level of CYP7A mRNA to ␤-actin mRNA as the mean Ϯ S.D of three replicate plates of cells. cells with culture medium containing CDCA or rosiglitazone did not induce the expression of these cytokines to detectable levels (data not shown). In marked contrast to L35 cells, HepG2 cells display the ability to express most inflammatory cytokines (30,31). The inability of L35 cells to express TNF␣, TGF-␤1, or IL-1␤ may explain their requirement for THP-1 cells to observe a CDCA repression of CYP7A1 (Fig. 2, A and B).
Our model also predicts that blocking the activation of cytokines by dietary bile acids in C57BL/6 mice should result in the C3H/HeJ phenotype (i.e. resistance to both the activation of hepatic cytokines and the repression of CYP7A1; Fig. 1). We, therefore, determined if treating C57BL/6 mice with rosiglitazone would prevent repression of CYP7A1 by the bile acidcontaining atherogenic diet. Rosiglitazone treatment of chowfed C57BL/6 mice did not alter the expression of CYP7A1. However, in C57BL/6 mice fed the bile acid-containing atherogenic diet, administration of rosiglitazone blocked the repression of CYP7A1 (Fig. 3). Our combined data suggest that the ability to induce the expression of hepatic cytokines in response to the bile acid-containing atherogenic diet is responsible for the divergent phenotypic response exhibited by C57BL/6 and C3H/HeJ mice. Thus, C57BL/6 mice display an induction of hepatic cytokine expression (Fig. 1B) and a repression of CYP7A1 (Fig. 1A), whereas C3H/HeJ neither display an induction of hepatic cytokines nor a repression of CYP7A1. In essence, rosiglitazone treatment caused C57BL/6 mice to display the C3H/HeJ phenotype in regard to resistance to repression of CYP7A1 by dietary bile acids.
Our findings do not exclude other mechanisms, independent of cytokines, acting to repress CYP7A1 in response to bile acids. In experimental models other than inbred mice and L35 cells, bile acid repression of CYP7A1 has been shown to involve BARE (17), activation of protein kinase C (18), and activation of FXR (16,19). The structure of the bile acid influences its ability to repress CYP7A1 (12) and activate FXR (16,19,32). Our additional findings showing that while unconjugated CDCA activated THP1 cells to produce cytokine repressors of CYP7A1 (Fig. 2), its taurine and glycine conjugates were without effect (data not shown) are consistent with the distinct physiochemical and physiological properties of unconjugated and conjugated bile acids. Conjugated bile acids require cell type-specific bile acid transporters to be efficiently taken up by cells in the ileum and liver (33). In contrast, unconjugated bile acids can enter cells lacking bile acid transporters via diffusion (33). Our combined results suggest that the unconjugated, hydrophobic bile acid CDCA entering the enterohepatic circulation via intestinal absorption may initiate CYP7A1 repression via the activation of regulatory cytokines by resident macrophages. Therefore, it is likely that the composition and concentration of the bile acid pool within the enterohepatic circulation determines which of several possible mechanisms will be invoked in regard to regulating the expression of CYP7A1.
FIG. 3. The PPAR␥ agonist rosiglitazone blocks the repression of CYP7A1 mRNA caused by feeding C57BL/6 mice the bile acidcontaining atherogenic diet. Female C57BL/6 mice (n ϭ 24) were fed either chow or the bile acid-containing atherogenic diet. Half the mice in each feeding group were given either vehicle (0.25% Tween 80, 1% carboxymethylcellulose) alone or vehicle containing 1 mg/ml of rosiglitazone daily by oral gavage. After 3 weeks, mice were sacrificed, and the relative content of rat CYP7A1 mRNA compared with GAPDH was determined. Each value represents the mean Ϯ S.D of six separate mice. The asterisk denotes a significant difference between the values for the rosiglitazone-treated chow-fed mice and the rosiglitazone mice fed the bile acid-containing atherogenic diet, p Ͻ 0.01.