Bile Acids With Differing Hydrophilic-Hydrophobic PropertiesDo Not Influence Cytokine Production by Human Monocytes

and Murine Kupffer Cells

ALBERTO BERGAMINI,1 LUCIANA DINI,3 LEONARDO BAIOCCHI,2 LUISA CAPPANNOLI,1 LAURA FALASCA,3

FRANCESCA BOLACCHI,2 MARCELLA CAPOZZI,1 EMANUELA FAGGIOLI,1 ALESSANDRA NISTRI,2 AGATA SALANITRO,1

LAURA VENTURA,1 GIOVANNI ROCCHI,1 AND MARIO ANGELICO2

conclusion, bile acids with widely different hydrophobi-Bile acids have been proposed to exert immunologicalcities are incapable of influencing the release of IL-6 andeffects of potential pathogenic or therapeutic relevance,TNFa by monocytes and Kupffer cells, provided they areyet the experimental evidence remains preliminary. Westudied at noncytotoxic concentrations and in the pres-reexamined the effects of a variety of bile salts with dif-ence of physiological amounts of proteins. (HEPATOLOGYfering hydrophilic-hydrophobic properties on the pro-1997;25:927-933.)duction of interleukin-6 (IL-6) and tumor necrosis factor

a (TNFa) from monocytes and Kupffer cells. Monocytesfrom healthy human donors and Kupffer cells from 5- Infectious complications1,2 and endotoxemia3,4 occur fre-week-old mice were incubated for up to 18 hours with quently in patients with severe cholestasis. The underlyingor without varying concentrations of bile salts and lipo- mechanisms are not fully understood, yet cholestasis may bepolysaccharide (LPS). Monocyte viability was¢95% with the consequence of the impairment of cell-mediated immu-up to 250 mmol/L sodium ursodeoxycholate and °90% nity5-7 and macrophage function.8 Endotoxins are lipopolysac-with 200 mmol/L chenodeoxycholate, decreasing sharply charides (LPS) present in the wall of gram-negative bacteria.at higher concentrations. Kupffer cells were more vul- Endotoxemia may derive from both systemic infections withnerable, particularly to chenodeoxycholate (viabilities gram-negative bacteria or from absorption of LPS from theof 25% and 0% at concentrations of 100 mmol/L and 200 gut. Normally, LPS is cleared by liver macrophages (Kupffermmol/L, respectively). In monocytes incubated in the cells), but when their phagocytic function is impaired, as inpresence of 20% fetal calf serum, neither ursodeoxycho- cholestasis, or an excessive amount of LPS is being deliveredlate and chenodeoxycholate, nor a variety of other un- from the intestine, endotoxins may spill over into the periph-conjugated and conjugated bile acids, tested up to their eral circulation. LPS stimulates both circulating and tissuemaximal noncytotoxic concentrations, influenced the macrophages to release several cytokines,9,10 known to be po-IL-6 and TNFa production, at any level of LPS stimula- tent inflammatory mediators,11 as well as activators of a com-tion. Similar to monocytes, incubation of murine Kupffer plex immuno-metabolic cascade,12-14 mimicking the effects ofcells with ursodeoxycholate and chenodeoxycholate did acute infections.15 This establishes a vicious cycle, aggravat-not influence cytokine release. In contrast, the addition ing the original cholestasis and often leading to multiple or-of 10 nmol/L dexamethasone to monocytes significantly gan failure. On the other hand, during bacterial sepsis, thedecreased TNF-a and IL-6 release (69 { 11% and 48 { liver is the main organ responsive to LPS,16 presumably be-15%, respectively). When monocytes were incubated cause Kupffer cells are the major source of inflammatorywith 200 mmol/L chenodeoxycholate in the presence of cytokines.9lower concentrations of fetal calf serum (10% and 5%, Cholestasis is always characterized by enhanced bile acidrespectively) a significant inhibition of cytokine release concentrations in blood as in the liver,17 hence bile acids havewas observed, whereas incubation with ursodeoxycho- been proposed to play a causative role in the impairmentlate did not cause any effect. Flow cytometry using fluo- of cell-mediated immunity and macrophage function. Recentresceinated LPS showed that chenodeoxycholate does studies have shown that both chenodeoxycholic acid (CDCA),not interact with the CD14 receptor, thus excluding the an endogenous hydrophobic bile acid, and ursodeoxycholicpossibility of an interference with the LPS uptake by acid (UDCA), an exogenous hydrophilic bile acid, show somemonocytes. Incubation with [14C]-chenodeoxycholate immunosuppressive properties,18 including the inhibition ofshowed that the intracellular bile acid uptake was in- inflammatory cytokine production by LPS-stimulated humanversely related to the concentration of fetal calf serum, monocytes.19-21 As most biological properties of bile acids arebeing negligible (õ3 fmol/cell) at the highest level. In known to depend on their physical chemical properties,22

which greatly differ between CDCA and UDCA,23 we reexam-ined their in vitro effects on the release of interleukin-6 (IL-6) and tumor necrosis factor a (TNF-a) from human blood

Abbreviations: LPS, lipopolysaccharide; CDCA, chenodeoxycholic acid; UDCA, ursodeoxy- monocytes and also studied murine Kupffer cells. In addition,cholic acid; IL-6, interleukin-6; TNF-a, tumor necrosis factor a; RPMI, Roswell Park Memo-

we explored the effects of a variety of other bile acids, uncon-rial Institute; FITC-LPS, fluorescein isothiocyanate-lipopolysaccharide.From the Divisions of 1Infectious Diseases and 2Gastroenterology, Department of Public jugated as well as glycine- and taurine-amidated, encom-

Health, University of Tor Vergata, Rome, and the 3Department of Biology, University of passing a wide range of hydrophilic-hydrophobic properties.Lecce, Italy. Fax: (396) 2020-799.

Received November 25, 1995; accepted January 13, 1996. MATERIALS AND METHODSAddress reprint requests to: Mario Angelico, M.D., Cattedra di Gastroenterologia, Dept.

of Public Health & Cell Biology, Tor Vergata University, Building F, Torre Nord, Room Chemicals576, Via di Tor Vergata 135, 00133 Rome, Italy. Fax: 396-2020-799.

The bile acids used in this study were obtained from the followingCopyright q 1997 by the American Association for the Study of Liver Diseases.0270-9139/97/2504-0023$3.00/0 sources: (CDCA), glyco-cholate, and glyco-deoxycholate from Calbio-

927

AID Hepa 0013 / 5p1d$$$241 03-12-97 08:29:57 hpta WBS: Hepatology

928 BERGAMINI ET AL. HEPATOLOGY April 1997

chem (San Diego, CA); hyodeoxycholate, cholate, deoxycholate, and Assay of Bile Acid Toxicitytauro-deoxycholate from Sigma (St. Louis, MO); UDCA, tauro-

In addition to the use of the standard trypan blue exclusionUDCA, glyco-UDCA, glyco-CDCA, tauro-CDCA, and tauro-cholatemethod, bile acid toxicity in monocytes was evaluated by measuringfrom Prodotti Chimici ed Alimentari, (Basaluzzo, Italy); tauro-hyo-the effect of CDCA and UDCA (200 mmol/L) on [35S]-methionine in-deoxycholate from Drug Research (Milan, Italy). [14C]-carboxyl-corporation, taken as an index of cell function. In brief, just afterCDCA was obtained from New England Nuclear (Boston, MA). Mostpurification, monocytes were metabolically labelled with 1 mCi ofunlabeled bile acids were obtained as sodium salts from the manufac-[35S]-methionine in methionine-free medium for 18 hours, with orturer. Those obtained in acid form were converted to sodium saltswithout bile salts or dexamethasone. After cell lysis, the radioactivityin our laboratory by potentiometric titration with 5N NaOH andincorporated in the acid insoluble material was determined as de-then recrystallized twice from diethylether/methanol. All the bilescribed by Amici and Santoro.28

acids used had a purity greater than 98.5%, as determined by thin-layer chromatography24 and high-performance liquid chromatogra-

Assay of Cytokine Production by Monocytesphy analyses.25 Stock solutions (20 mmol/L) of each bile acid underand Kupffer Cellsstudy were prepared in phosphate buffer at pH 7.4, with 150 mmol/

L NaCl. The bile acid solutions were filter-sterilized using 0.2-mm At a first approach, approximately 150,000 monocytes were incu-pore size filters (Sterile Acrodisc, Gelman Sciences, Ann Arbor, MI). bated for 18 hours in 48-well plates (Costar, Cambridge, MA) con-Direct measurement of bile acid concentration by HPLC25 in the taining complete medium, supplemented with 20% heat-inactivatedfiltrates excluded the possibility of nonspecific bile salt adsorption fetal calf serum, in the presence or in the absence of LPS and in theduring the filtration procedure, the difference between the measured presence or in the absence of different bile acids and/or dexametha-and the calculated concentrations being õ2% at all times. Sterile sone. In subsequent experiments aimed at evaluating the influencebile acid solutions were stored at 0207C until used. of fetal calf serum, similar experiments were also performed using

LPS from Escherichia coli 0111/B4 (Sigma Chemical, St. Louis, 5% and 10% heat-inactivated calf serum, respectively, in the pres-MO) was diluted in phosphate buffer to yield stock concentrations ence of UDCA or CDCA.of 1 mg/mL. Aliquots were stored at 0207C until used and then Kupffer cells adherent to dishes were incubated in complete me-thawed and diluted in culture medium (Roswell Park Memorial Insti- dium, supplemented with 5% heat-inactivated fetal calf serum, intute [RPMI] 1640) to yield final concentrations ranging from 0.01 duplicate, for 18 hours, in the presence of LPS and in the presenceto 100 ng/mL. Fluorescein isothiocyanate-lipopolysaccharide (FITC- or in the absence of bile acids. At the end of each incubation period,LPS) was obtained from Sigma Chemical (St. Louis, MO) and [3H]- monocyte and Kupffer cell supernatants were collected, centrifuged,thymidine from Amersham International (Buckinghamshire, UK). and stored at 0807C until used for cytokine assays.Dexamethasone 21-phosphate (disodium salt) was obtained from La- Human and murine Il-6 and TNF-a concentrations were measuredboratorio Farmacologico Milanese (Caronno P., Italy). It was diluted by enzyme-linked immunosorbent assay (Genzyme, Milan, Italy).in phosphate-buffered saline to yield stock concentrations of 1 mg/ The human Il-6 and TNF-a assay had a detection limit of 15 pg/mL;mL. Aliquots were stored at 0207C until used and then thawed and the murine assay had a detection limit of 100 pg/mL. Each enzyme-diluted in culture medium to yield final concentration of 10 nmol/L. linked immunosorbent assay was specific for the cytokine it mea-H-collagenase, pronase, DNAase, and metrizamide were purchased sured, and correlated well with its specific bioassay. Preliminaryfrom Boehringer (Mannheim, Germany). [35S]-Methionine was ob- tests showed that the bile acids under study did not interfere with thetained from Amersham. All other materials and reagents were ob- assay. Total cellular proteins were measured by the Biorad Proteintained from Merck (Darmstadt, Germany), or from Sigma Chemical. assay.29 Each data point represents the average of duplicate determi-

nations.Cells

Flow CytometryHuman Monocytes. Peripheral blood obtained from healthy blood This technique was used to study the potential interaction of bile

donors was enriched for mononuclear cells by centrifugation over acids with the LPS-CD14 binding receptor.30 In brief, one millionFicoll Hypaque. Monocytes were obtained by adherence of peripheral monocytes were incubated at 377C for 30 minutes, in the presenceblood mononuclear cells to plastic, as previously described.26,27 Cells or absence of 200 mmol/L CDCA with 100 ng/mL of FITC-LPS, prein-obtained by this method are more than 95% pure and viable, as cubated for 15 minutes at room temperature in RPMI 1640. Experi-determined by nonspecific esterase activity27 and by trypan blue ex- ments were performed using 5% and 20% heat-inactivated fetal calfclusion. Each set of experiments was performed using monocytes serum. After incubation, the cells were washed three times in PBSobtained from a single donor. containing 0.37% paraformaldehyde and analyzed with a cytofluo-

Mouse Kupffer Cells. Kupffer cells were isolated from 5-week-old rimeter (FACScan Flow Cytometer, Becton Dickinson, Mountainmale Swiss mice (Morini, Reggio Emilia, Italy), weighing 20 to 30 g. View, CA). The cell population selected for the analysis consisted ofThe animals, maintained at a 12-hour day-night rhythm, were fed 97% CD14 positive cells (i.e., macrophages) and 3% CD3-positivea standard diet ad libitum. They received care in compliance with cells (i.e., lymphocytes).the National Research Council’s criteria for animal care. Kupffercells were isolated by an enzymatic perfusion technique. The livers Assay of Bile Salt Uptake by Monocyteswere perfused in situ via the portal vein with pronase (0.1% in RPMI

To measure the apparent bile acid capacity to penetrate mono-1640) at 377C, followed by the same solution supplemented withcytes, the cells were incubated for 18 hours at 377C in completecollagenase (0.05%). The livers were then removed, minced in smallmedium, using different concentrations of fetal calf serum (5%, 10%,pieces and incubated at 377C for 30 minutes with 0.1% pronase,and 20%), with 5 mCi of [14C)-carboxyl-CDCA and in the presence of0.05% H-collagenase, and 0.001% DNAse. The suspension was fil-40 mmol/L or 200 mmol/L of unlabeled CDCA. Monocytes incubatedtered through nylon gauze, centrifuged at 300g for 10 minutes atwith 5 mCi of [3H]-thymidine (11.6 nmol/L, final concentration) were47C and washed with RPMI 1640. The cells were then centrifugedused as control cells. After incubation, aliquots (10 mL) of the super-on a metrizamide gradient at 1,400g for 15 minutes at 257C. Cellsnatant were collected and counted for radioactivity using a Wallacin the interface layer, essentially devoid of hepatocytes, were washedbeta-counter (LKB, Bromma, Sweden). The cells were then washedin RPMI 1640 containing penicillin (50 U/mL), streptomycin (50 mg/three times in phosphate-buffered saline containing 0.37% paraform-mL), 2 mmol/mL L-glutamine (complete medium, pH 7.4), supple-aldehyde and lysed in 1% sodium dodecyl sulfate. The lysate radioac-mented with 5% heat inactivated fetal calf serum, adjusted totivity was measured as the supernatant’s. Intracellular [14C]-CDCA500,000 cell/mL, and incubated overnight in 35 1 10-mm tissue cul-and [3H]-thymidine moles were calculated multiplying the counts perture dishes (Falcon 3801, Becton Dickinson Labware, Lincoln Park,minute of the lysate to the total CDCA moles present in the sampleNJ). Nonadherent cells were removed by washing the dishes severaldivided by the total CPM (lysate / supernatant) of the sample andtimes with RPMI 1640 at 377C and the adherent cells (Kupffer cells)normalized to one cell.were left attached. In general, 80,000 to 150,000 viable cells re-

mained attached to each dish as detected by trypan blue exclusion. Electron MicroscopyViability of the cells obtained by this technique was greater than95%. More than 90% of the adherent cells were able to phagocytose For electron microscopic analysis, monocytes and Kupffer cellsIndia ink and had the morphological characteristics of Kupffer cells were processed as described above. After incubation with bile salts,

cells were fixed with 2.5% glutaraldehyde, detached by gentlewhen examined under transmission electron microscopy.

AID Hepa 0013 / 5p1d$$$242 03-12-97 08:29:57 hpta WBS: Hepatology

HEPATOLOGY Vol. 25, No. 4, 1997 BERGAMINI ET AL. 929

incorporation method. This showed similar average intracel-lular radioactive counts in control monocytes (12,800 { 1,400cpm) and in cells exposed to 200 mmol/L UDCA (13,200 {1,000 cpm), but significantly lower (P õ .01) counts in thoseexposed to 200 mmol/L CDCA (10,900{ 800 cpm), correspond-ing to 15% mortality.

In contrast to human monocytes, murine Kupffer cells weremuch more vulnerable to bile acid incubation, particularlywith CDCA (Fig. 1B). At bile acid concentrations of 200 mmol/L, Kupffer cell mortality was already 20% for UDCA andreached 100% for CDCA. The maximal noncytotoxic concen-tration whereby it was possible to measure the influence ofCDCA on cytokine release by Kupffer cells was 66 mmol/L.Figure 2 shows the electron microscopic appearance of Kupf-fer cells obtained after 18 hours incubation in the absence(Fig. 2A) or in the presence of 100 mmol/L CDCA (Fig. 2B).The latter shows severe damage, including loss of microvilli,disarray of organelles, and formation of large vesicles.

Effect of LPS on IL-6 and TNF-a Production by HumanMonocytes. The LPS-stimulated production of IL-6 andTNF-a by human monocytes is illustrated in Fig. 3. In theabsence of LPS, cytokine release was negligible. The additionof LPS to the culture medium caused a dramatic increase ofIL-6 and TNF-a release, which was already remarkable fromLPS concentrations of 0.1 ng/mL. Further increases of LPSconcentrations resulted in additional increments of cytokinerelease up to 100 ng/mL and in some loss of cell viabilityabove this level.

Effect of UDCA and CDCA on IL-6 and TNF-a Productionby Human Monocytes. The effect of UDCA and CDCA on theproduction of the two cytokines was examined in monocytesnot stimulated by LPS, as well as at various levels of LPS-stimulation. These initial experiments were performed incu-bating the cells in the presence of 20% fetal calf serum in themedium. UDCA and CDCA were tested up to their maximalnoncytotoxic concentrations (250 mmol/L and 200 mmol/L, re-spectively). In the absence of LPS-stimulation, UDCA andCDCA did not induce any cytokine release, which excludedthe possibility of bile acid contamination with LPS. In LPS-

FIG. 1. (A) Percent mortality of human monocytes incubated in the pres- stimulated monocytes, neither the addition of UDCA nor ofence of increasing concentrations of UDCA or CDCA. (B) Percent mortality of

CDCA induced significant changes in the production of IL-6murine Kupffer cells incubated in the presence of increasing concentrations ofand TNF-a compared with control values (i.e., without bileUDCA or CDCA. Mortality was calculated by the trypan blue dye assay.salts), at any level of LPS-stimulation and of bile acid concen-tration. The influence of 200 mmol/L of UDCA or CDCA onthe release of IL-6 and TNF-a from monocytes at different

scraping, collected by low-speed centrifugation, and postfixed in 1% levels of LPS-stimulation is reported in Table 1. Similar non-OsO4. After dehydration, specimens were embedded in Epon resin.significant changes were obtained when monocytes were pre-Thin sections were stained with uranyl acetate and lead nitrate andincubated for 1 and 24 hours with UDCA and CDCA beforeobserved under a Philips electron microscope (Kassel, Germany).LPS-stimulation. Moreover, none of the results on the effect

Statistics of bile acid on IL-6 and TNF-a release by monocytes differedsignificantly from controls (without bile acids) when the ex-Each experiment was performed at least in triplicate and the re-perimental data were expressed as a function of the totalsults are presented as means { SD. In most studies with bile acids,cellular protein concentration (data not shown).cytokine production is expressed as the percentage of control values

(i.e., without bile acids). Statistical differences were tested by the Effect of Other Unconjugated and Conjugated Bile Acids onANOVA and the Student’s t test for paired or unpaired data. LPS-Stimulated IL-6 and TNF-a Production by Human Mono-

cytes. A variety of other bile acids were tested for their poten-RESULTS tial effects on IL-6 and TNF-a release by monocytes. The

following bile salts (200 mmol/L) were examined in triplicateEffect of UDCA and CDCA on Human Monocytes and MurineKupffer Cell Viability. To test whether potential effects of experiments: glyco-CDCA and tauro-CDCA; glyco-UDCA and

tauro-UDCA; cholate, glyco-cholate, and taurocholate; deoxy-bile acids on cytokine release were independent of toxic ef-fects, cell viability was first examined by trypan blue exclu- cholate, glyco-deoxycholate, and tauro-deoxycholate; and hy-

odeoxycholate and tauro-hyodeoxycholate. With the excep-sion as a function of increasing concentrations of UDCA andCDCA. As illustrated in Fig. 1A, monocytes were consider- tion of deoxycholate, which caused 100% of monocyte

mortality, cell viability was not significantly influenced byably resistant to bile acid exposure: cell mortality was absentor minimal (°5%) in the presence of UDCA up to a concentra- any of the above compounds. Under these experimental con-

ditions, none of the noncytotoxic bile acids listed above signif-tion of 250 mmol/L, and approximated 10% in the presenceof CDCA at a concentration of 200 mmol/L. Monocyte viability icantly reduced the production of either IL-6 or TNF-a by

monocytes with respect to control values, whether the datadecreased sharply at higher concentrations of CDCA, pre-cluding further experiments. These results were confirmed were normalized or not for total cellular protein concentra-

tion.when cell toxicity was examined using the [35S]-methionine

AID Hepa 0013 / 5p1d$$$242 03-12-97 08:29:57 hpta WBS: Hepatology

930 BERGAMINI ET AL. HEPATOLOGY April 1997

Effect of Dexamethasone on IL-6 and TNF-a Production byHuman Monocytes. To rule out the possibility that inappro-priate levels of LPS-stimulation were being used in our study,we determined the effect of adding dexamethasone, a knownreference immunodepressant,31 as a positive control on cyto-kine production. As shown in Fig. 4, dexamethasone (10 nmol/L) caused a significant reduction of IL-6 and TNF-a releasefrom LPS-stimulated monocytes incubated for 18 hours, bothwhen tested alone (61.7 { 9% and 72.5 { 12% of controlvalues, respectively) and in combination with 200 mmol/LUDCA or CDCA.

Effect of Bile Acids on Cytokine Production by Murine Kupf-fer Cells. The effect of UDCA, CDCA, and cholic acid on cyto-kine production by LPS-stimulated murine Kupffer cells isshown in Table 2. These bile acids, studied at their maximalnoncytotoxic concentrations (i.e., 66 mmol/L for CDCA and100 mmol/L for UDCA and cholic acid, respectively) did notcause any significant change in the production of IL-6 andTNF-a compared with control cells.

Effect of Decreasing Concentrations of Fetal Calf Serum onCytokine Production by Human Monocytes. Because our re-sults, as reported above, were partly in contrast with thoseof recent reports,20,21 we tested whether the effects of bileacids on cytokine release from monocytes could be influencedby differences in the incubation media. For this purpose, ad-ditional experiments were performed by incubating mono-cytes in media with decreasing concentrations of fetal calfserum (20%, 10%, and 5% dilutions). The preparation usedhad an initial total protein concentration of 3.5 g/dL, of which2.4 g/dL represented by albumin. It was not possible to carryout experiments in the absence of fetal calf serum as mono-cytes do not respond properly to LPS-stimulation under theseconditions. The results of these experiments are reported inTable 3. The data indicate that, both at a concentration of10% and, even more, 5% of fetal calf serum in the medium,the addition of 200 mmol/L of CDCA significantly decreasedthe release of IL-6 and TNF-a from LPS-stimulated mono-cytes. In contrast, the addition of 200 mmol/L of UDCA wasineffective at any given concentration of fetal calf serum.

Flow Cytometric Analysis of LPS Binding to Monocytes. Wenext investigated whether the inhibitory effect of CDCA oncytokine release observed at 5% (but not at 20%) concentra-tion of fetal calf serum was caused by a direct effect of CDCAon the interaction of LPS with the CD14 receptor, the majormechanism in triggering monocytes activation.30,32 In the ab-sence of CDCA, incubation of monocytes with FITC-LPS

FIG. 2. (a) Transmission electron microscopy (7,0001) showing normal mu-rine Kupffer cells in phagocytic activity (arrow), incubated in the absence ofbile acids. (b) Transmission electron microscopy (6,2001) showing severelydamaged murine Kupffer cells after 18 hours of incubation with 100 mmol/LCDCA. The cell surface is smooth, with loss of microvilli. Cytoplasmic organ-elles are disarrayed and large vesicles are present (arrows).

FIG. 3. Production of IL-6 and TNF-a by human monocytes incubated inthe absence and in the presence of increasing concentrations of LPS in theculture medium. Results were obtained using 20% fetal calf serum in themedium and are expressed as cytokine concentrations at the end of 18 hoursof incubation.

AID Hepa 0013 / 5p1d$$$242 03-12-97 08:29:57 hpta WBS: Hepatology

HEPATOLOGY Vol. 25, No. 4, 1997 BERGAMINI ET AL. 931

TABLE 2. Effects of Bile Acids on IL-6 and TNF-a Production in LPS-TABLE 1. Effect of UDCA and CDCA Acids on IL-6 and TNF-aProduction by Human Monocytes at Different Stimulated Murine Kupffer Cells

Levels of LPS-StimulationIL-6 Production TNF-a Production

Bile Salt (pg/mL) (pg/mL)IL-6 and TNF Production (% of control) in Macrophages Stimulatedby LPS Concentration (ng/mL)

None 5,700 { 1,100 3,800 { 6000.1 1 10 100 UDCA (100 mmol/L) 5,450 { 1,200 3,700 { 900

Cholic acid (100 mmol/L) 6,200 { 1,500 4,100 { 1,000IL-6 UDCA 90 { 12 115 { 9 93 { 8 104 { 7 CDCA (66 mmol/L) 5,400 { 900 4,100 { 1,200

CDCA 93 { 9 96 { 14 118 { 12 97 { 11TNF-a UDCA 101 { 15 92 { 12 112 { 13 110 { 9 NOTE. Data are means { SD of triplicate incubation experiments lasting

CDCA 101 { 11 112 { 13 94 { 15 93 { 10 18 hours. LPS (0.1 ng/mL), UDCA and CDCA and cholic acid were addedto the culture media containing 5% heat-inactivated fetal calf serum before

NOTE. Data are means { SEM and are expressed as percentages of the incubation. For details see Materials and Methods. None of the differencescontrol values obtained in the absence of bile acids. No significant differences between individual bile salts and controls (without bile acids) reached statisti-from controls were observed. Different concentrations of LPS and UDCA or cal significance.CDCA (final concentrations 200 mmol/L) were added to the culture media con-taining 20% fetal calf serum at the beginning of monocyte incubation, whichlasted for 18 hours. For details see Materials and Methods.

DISCUSSION

In chronic biliary obstruction bile acids are extensively dis-tributed outside the enterohepatic circulation, reaching con-caused a significant shift in the main fluorescence intensitycentrations of 50 to 150 mmol/L in peripheral blood and possi-compared with untreated cells, both using 5% and 20% fetal bly greater in the liver.17 Bile acids may thus come intocalf serum in the incubation medium (Fig. 5). The binding of contact with a number of cells that are normally not exposedFITC-LPS to monocytes was not influenced by the addition to them, including lymphocytes and macrophages. Kupfferof 200 mmol/L CDCA at both concentrations of fetal calf se- cells are likely to be exposed to even higher bile acid concen-rum, as shown by similar patterns of fluorescence obtained trations compared with circulating macrophages. Because ofin the presence or in the absence of CDCA. These data indi- the impairment of the enterohepatic circulation, bile acidscate that CDCA does not interfere with the uptake of LPS have long been suspected to play a causative role in the im-by monocytes. munosuppression characterizing cholestatic conditions5-8 and

Uptake of Bile Acids by Monocytes. The uptake of [14C]- experimental data have been generated to support thisCDCA (40 mmol/L and 200 mmol/L) by monocytes incubated view.18-21,23,33-34

at different concentrations of fetal calf serum is illustrated In this study, CDCA and UDCA were selected as two proto-in Table 4, in comparison with that of [3H]-thymidine as a type bile acid molecules with divergent physico-chemicalcontrol. The data show that CDCA uptake, both at 40 mmol/ characteristics,35 to test their ability to influence the produc-L and 200 mmol/L concentration of the bile acid, was inversely tion of IL-6 and TNFa from monocytes and Kupffer cells.related to the concentration of fetal calf serum in the medium, In agreement with the study by Calmus et al.,20 UDCA andbeing markedly inhibited at the highest level of the latter. CDCA showed a modest toxicity against human monocytesIn contrast, [3H]-thymidine uptake was independent from fe- in concentrations up to 200 mmol/L, yet CDCA was clearlytal calf serum. The actual amount of CDCA entering each more cytotoxic than UDCA. In contrast to human monocytes,monocyte during the incubation period was in the order of mouse Kupffer cells were much more vulnerable to both thefew fentomoles per cell and was directly related to the CDCA bile acids. This is in agreement with studies showing thatconcentration in the medium, indicating a passive diffusionof the bile acid.

TABLE 3. Influence of Fetal Calf Serum Concentration and CDCA onIL-6 and TNF-a Production in LPS-Stimulated Human Monocytes

TNFaLPS Concentration CDCA (or UDCA) IL-6 Production Production

(ng/mL) (mmol/L) (pg/mL) (pg/mL)

FCS: 20%10 0 8,500 { 1,700 7,200 { 80010 200 8,500 { 700 6,600 { 900

100 0 10,800 { 200 9,765 { 1,200100 200 11,600 { 1,800 8,600 { 1,100

FCS: 10%10 0 3,200 { 100 6,400 { 80010 200 600 { 100* 1,500 { 300*

100 0 8,600 { 900 8,669 { 1,100100 200 1,100 { 300* 2,600 { 500*

FCS: 5%10 0 2,100 { 300 1,100 { 20010 200 õ15* 300 { 40*

100 0 3,812 { 730 6,600 { 900100 200 250 { 50* 400 { 100*100 200 (UDCA) 4,900 { 600 5,900 { 900

NOTE. Data are means { SD of triplicate incubation experiments lasting18 hours. LPS and CDCA were added to the culture media containing 20%,FIG. 4. Production of IL-6 and TNF by LPS-stimulated human monocytes10%, or 5% fetal calf serum (FCS), respectively, before incubation. For detailsincubated in the presence of UDCA (200 mmol/mol/L), CDCA (200 mmol/L) andsee Materials and Methods.dexamethasone (DXM, 10 nmol/L), with or without bile acids, in the culture

* Significantly different (P õ .01) from the corresponding data obtainedmedium. Results were obtained using 20% fetal calf serum in the medium andare expressed as percent of control values. without CDCA.

AID Hepa 0013 / 5p1d$$$243 03-12-97 08:29:57 hpta WBS: Hepatology

932 BERGAMINI ET AL. HEPATOLOGY April 1997

both UDCA and CDCA, starting from concentrations as lowas 50 mmol/L, on the production of IL-1 and, to a lesser extent,of IL-6. Yet, the effects of the two bile salts clearly differedwith regard to TNFa production, UDCA virtually having noeffect. Similar differing results on TNFa production were re-ported by Greve et al.19 using UDCA and CDCA concentra-tions between 25 mmol/L and 125 mmol/L. Although in thesestudies a cellular system similar to ours was used, the au-thors employed LPS concentrations greater than those usedin most of our studies. Greve et al.,19 for example, used in-creasing concentrations of LPS, up to 3000 ng/mL, yet themaximal TNF-a production in control cells was already ob-tained with the lowest concentration tested (1 ng/mL). Toexclude possible biases because of the use of suboptimal LPS-stimulation in our assay, we examined the effects of UDCAand CDCA under a wide range of LPS concentrations (from0.1 to 100 ng/mL), yet at all of them we failed to demonstratesignificant effects of bile acids on cytokine release.FIG. 5. Flow cytometric analysis of fluorescein FITC-LPS binding to hu-

man monocytes. The black histograms represent the fluorescence of untreated Furthermore, the validity of our monocyte culture systemcells. The white histograms represent the fluorescence of cells exposed to FITC- to detect changes in LPS-stimulated cytokine production wasLPS. (A) Cells incubated with LPS with 5% FCS in the absence of CDCA. (B) checked by determining the effect of adding dexamethasone,Cells incubated with LPS with 5% FCS in the presence of 200 mmol/L CDCA.

a known reference immunodepressant, as a positive control.31The data indicate that CDCA does not affect the uptake of FITC-LPS by mono-cytes (duplicate experiments). Dexamethasone markedly inhibited IL-6 and TNF-a produc-

tion, irrespective of the presence or absence of UDCA andCDCA. Thus, bile acids do not even interfere with the mecha-nisms of steroidal immunosuppression, an effect thought tobile salts may alter cultured Kupffer cells functionally,36 but operate at the level of gene transcription.41

in apparent conflict with Van Bossuyt et al.,37 who did not We next noted, however, that all previous studies19-21 onfind changes in the viability of rat Kupffer cells exposed to this subject employed lower concentrations of fetal calf serumhigher concentrations of taurocholate. Differences between in the incubation media than that used in our initial studies.rat and mouse Kupffer cell and the use of a poorly cytotoxic Thus, in additional experiments using lower concentrationstaurine-conjugated bile acid instead of unconjugated com- of fetal calf serum we were able to detect a marked inhibitionpounds38-39 may perhaps explain the latter contradiction. of cytokine release from monocytes induced by CDCA (200In the present study, in contrast with earlier reports onmmol/L), but not by UDCA. Table 5 summarizes our resultshuman monocytes,19-21 none of the bile salts under study in comparison with those of the other studies that used ashowed significant effects on cytokine production by human similar experimental design. Furthermore, using radioactivemonocytes, provided that they were examined at noncytotoxic CDCA we found that the above phenomenon was associatedconcentrations and in the presence of 20% of fetal calf serum with significant cellular uptake of the bile salt. In contrast,in the medium. Nor did they influence the release of IL-6 and the latter was negligible (°6 fmol/cell) when cells were incu-TNF-a by murine Kupffer cells. bated at the highest level of fetal calf serum (20%), whichAt first, the reason why our results on cytokine release corresponds to a final albumin concentration of 0.48 g/dL.differed from previous studies was not obvious. Bile salt pu- Thus, bile acid uptake by monocytes is inversely related tority was high and all the compounds were used as sodium the concentration of proteins (albumin) in the incubation me-salts both in the present as in previous reports; and there dium. Extrapolation of these findings to an in vivo situation,were no apparent differences in the pH of the incubation where albumin concentration is at least one order of magni-buffers (pH 7.4). This is important because our findings using tude greater, suggests that bile acids are unlikely to exertradioactive CDCA indicated that bile acid uptake by mono- inhibitory functions on circulating or resident macrophages,cytes likely occurs through passive diffusion, a process poten- if present in noncytotoxic concentrations.tially influenced by pH-induced variations in bile acid proton- An alternative hypothesis to be considered is that bile acidsation.40 This also appears in keeping with the concept that may influence cytokine release indirectly, through modula-monocytes and Kupffer cells do not have active transport

systems for bile acids.On the other hand, data from Lacaille and Paradis21 were

TABLE 5. Comparative Effects of CDCA and UDCA Acids on IL-6 andnot easily comparable with ours, as these authors used phyto-TNF-a Production by Human Monocytes in Different Studies Usinghemoagglutinin-stimulated peripheral blood mononuclear

Similar Experimental Designscells in their assay. Calmus et al.,20 using LPS-stimulatedFetal IL-6 TNF-amonocytes, reported a dose-dependent inhibitory effect ofCalf Release Release

Bile Salt Serum (% of (% ofAuthor Concentration (%) control) control)

TABLE 4. Uptake of CDCA and (3H)-Thymidine by Monocytes Greve et al.19 CDCA, 125 mmol/L 20% n.d. 37%UDCA, 125 mmol/L 10% n.d. 86%Fetal Calf Serum Concentration in Cell Medium

Calmus et al.20 CDCA, 100 mmol/L 10% 28% 30%5% 10% 20% CDCA, 200 mmol/L 10% 2% 4%

UDCA, 100 mmol/L 10% 97% 113%CDCA, 40 mmol/L 1860 (3.7) 1220 (2.8) 768 (2) UDCA, 200 mmol/L 10% 67% 85%CDCA, 200 mmol/L 8800 (18) 8092 (16) 3096 (6) Present study CDCA, 200 mmol/L 20% 97% 93%Thymidine, 116 nmol/L 8609 (0.05) 8111 (0.05) 8314 (0.05) CDCA, 200 mmol/L 10% 13% 30%

CDCA, 200 mmol/L 5% 7% 7%NOTE. The data are expressed as cpm in 500,000 cells and represent the UDCA, 200 mmol/L 20% 104% 110%

average of duplicate experiments. The numbers in parenthesis indicate the UDCA, 200 mmol/L 5% 128% 89%actual cellular concentrations (fmol/cell).

AID Hepa 0013 / 5p1d$$$243 03-12-97 08:29:57 hpta WBS: Hepatology

HEPATOLOGY Vol. 25, No. 4, 1997 BERGAMINI ET AL. 933

H. Mechanisms of cholestasis. 6. Bile acids in human livers with or withouttion of LPS ingestion by macrophages, or by LPS bindingbiliary obstruction. Gastroenterology 1972;63:837-845.or disaggregation.42 Van Bossuyt et al.35 reported that the 18. Calmus Y, Weill B, Ozier Y, Chereau C, Houssin D, Poupon R. Immunosup-

phagocytic activity of cultured Kupffer cells is decreased by pressive properties of chenodeoxycholic and ursodeoxycholic acids inmouse. Gastroenterology 1992;103:617-621.several bile acids, but this effect occurred only at bile acid

19. Greve JW, Gouma DJ, Buurman WA. Bile acids inhibit endotoxin-inducedconcentrations greater than 600 mmol/L, clearly toxic underrelease of tumor necrosis factor by monocytes. HEPATOLOGY 1989;10:454-the experimental conditions used in the present study. Oth- 458.

ers reported that bile acids inhibit the pinocytosis of colloidal 20. Calmus Y, Guechot J, Podevin P, Bonnefis M-T, Giboudeau J, PouponR. Differential effects of chenodeoxycholic and ursodeoxycholic acids oncarbon by cultured hamster Kupffer cells, but only at concen-interleukin 1, interleukin 6 and tumor necrosis factor-a production bytrations above 2 mmol/L.34 Our flow cytometric studies usingmonocytes. HEPATOLOGY 1992;16:719-723.FITC-LPS did not provide evidence for an interaction of 21. Lacaille F, Paradis K. The immunosuppressive effect of ursodeoxycholic

CDCA with the CD14 receptor, both at high or low concentra- acid: a comparative in vitro study on human peripheral blood mononuclearcells. HEPATOLOGY 1993;18:165-172.tion of fetal calf serum, thus excluding the possibility of a

22. Hofmann AF, Roda A. Physicochemical properties of bile acids and theirbile acid interference with the LPS uptake process by mono-relationship to biological properties: an overview of the problem. J Lipidcytes. On the other hand, it is unlikely that bile acids would Res 1977;25:1477-1489.

bind and/or disaggregate LPS in concentrations below their 23. Gianni L, Di Padova F, Zuin M, Podda M. Bile acid-induced inhibition ofthe lymphoproliferative response to phytohemagglutinin and pokeweedcritical micellar concentration.mitogen: an in vitro study. Gastroenterology 1980;78:231-235.In conclusion, our study does not support the proposed view

24. Batta AK, Shefer S, Salen G. Thin layer chromatographic separation ofthat bile acids, be they hydrophilic or hydrophobic, inhibit conjugates of ursodeoxycholic acid from those of lytho-, chenodeoxy-, deoxy-the release of IL-6 and TNF-a by cultured human monocytes, and cholic acids. J Lipid Res 1981;22:712-714.

25. Cantafora A, Di Biase A, Alvaro D, Angelico M. An improved methodnor by murine Kupffer cells, provided that they are tested atfor measuring the glycine and taurine conjugates of bile salts by highnontoxic concentrations and in the presence of physiologicalperformance liquid chromatography with tauro 7-alpha, 12-alpha, dihy-

amounts of proteins. However, the failure of our study to droxy-5-beta-cholanic acid as internal standard. J Chromatogr 1987;386:detect significant effects of bile acids on cytokine release in 367-370.

26. Perno CF, Yarchoan R, Cooney DA, Hartman NR, Webb DSA, Hao Z,vitro does not exclude the possibility of other in vivo immuno-Mitsuya H, et al. Replication of immunodeficiency virus in monocytes.modulatory effects. These may occur through other mecha- Granulocyte/macrophage colony-stimulating factor (GM-CSF) potentiates

nisms, such as effects on lymphocyte proliferation,23,31 or on viral production yet enhances the antiviral effect mediated by 3*-azido-2*3*-dideoxythymidine (AZT) and other dideoxynucleoside congeners ofKupffer cell activation and death, as indirectly suggested bythimidine. J Exp Med 1989;169:933-951.the vulnerability of murine Kupffer cells to even low bile acid

27. Fischer DG, Koren HS. Isolation of human monocytes. In Adams DO,concentrations. Edelson PJ, Koren S, eds. Methods for Studying Mononuclear Phagocytes.San Diego: Academic Press, 1981:41.

28. Amici C, Santoro MG. Suppression of virus replication by prostaglandinREFERENCESA is associated with heat shock protein synthesis. J Gen Virol 1991;72:1855-1877.1. Pain JA, Cahill CJ, Bailey ME. Perioperative complications in obstructive

29. Bradford MM. A rapid and sensitive method for the quantitation of micro-jaundice: therapeutic considerations. Br J Surg 1985;72:942-945. gram quantities of protein utilizing the principle of protein dye binding.2. Armstrong CP, Dixon JM, Taylor TV, Davies GC. Surgical experience of Anal Chem 1972;72:248-254.deeply jaundiced patients with bile obstruction. Br J Surg 1984;71:234- 30. Gallay P, Barras C, Tobias PS, Calandra T, Glauser MP, Heumann D.238. Lipopolysaccharide (LPS)-binding protein in human serum determines the3. Bailey ME. Endotoxin, bile salt and renal function in obstructive jaundice. tumor necrosis factor response of monocytes to LPS. J Infect Dis 1994;Br J Surg 1976;63:774-778. 170:1319-1322.4. Wilkinson S, Moodie H, Stamatakis JD, Kakkar VV, Williams R. Endotoxi- 31. Baybutt HN, Holsboer F. Inhibition of macrophage differentiation and

nemia and renal failure in cirrhosis and obstructive jaundice. Br Med J function by cortisol. Endocrinology 1988;127:476-480.1976;2:1415-1418. 32. Yang Z, Khemiani LS, Dean DF, Carter CD, Slauson DO, Bochsler PN.

5. Cainzos M, Potel J, Puente JL. Anergy in jaundiced patients. Br J Surg Serum components enhance bacterial lipopolysaccharide-induced tissue1988;75:147-149. factor expression and tumor necrosis factor-alfa secretion by bovine alveo-

6. Fox RA, Dudley FJ, Samuels M, Milligan J, Sherlock S. Lymphocyte trans- lar macrophages in vitro. J Leukoc Biol 1994;55:483-488.formation in response to phytohemoagglutinin in primary biliary cirrhosis: 33. Keane RM, Gadacz TR, Munster AM, Birmingham W, Winchurch RA.the search for a plasma inhibitory factor. Gut 1973;14:89-93. Impairment of human lymphocyte function by bile salts. Surgery 1983;95:

7. Roughneen PT, Drath DB, Kulkarni AD, Rowlands BJ. Impaired nonspe- 439-444.cific cellular immunity in experimental cholestasis. Ann Surg 1987;206: 34. Calmus Y, Podevin P, Robert A, Poupon R. Differential effects of cheno-578-582. deoxycholic and ursodeoxycholic acids on expression of procoagulant activ-

8. Drivas G, James O, Wardle N. Study of the reticuloendothelial phagocytic ity by human monocytes. J Hepatol 1994;20:466-472.capacity in patients with cholestasis. Br Med J 1976;1:1568-1569. 35. Heuman DM, Hylemon PB, Vlahcevic ZR. Regulation of bile acid synthesis.

9. Luster MI, Germolec DR, Yoshida T, Kayama F, Thompson M. Endotoxin- III. Correlation between biliary bile salt hydrophobicity index and theinduced cytokine gene expression and excretion in the liver. HEPATOLOGY activities of the enzymes regulating the cholesterol and bile acid synthesis1994;19:480-488. in the rat. J Lipid Res 1989;30:1161-1171.

10. Michie HR, Manogue KR, Spiggs DR, Revhaug A, O’Dwyer S, Dinarello 36. Takiguchi S, Koga A. Effects of bile acids and endotoxin on the functionCA, Cerami A, Wolff SM, Wilmore DW. Detection of circulating tumor and morphology of cultured hamster Kupffer cells. Virchows Archiv B Cellnecrosis factor after endotoxin administration. N Engl J Med 1988;318: Pathol 1988;54:303-311.1481-1486. 37. Van Bossuyt H, Desmaretz C, Gaeta GB, Wisse E. The role of bile acids

11. The role of cytokine mediators in septic shock. Adv Surg 1990;23:21-56. in the development of endotoxinemia during obstructive jaundice in the12. Van der Poll T, Romijun JA, Endert E, Borm JJ, Buller HR, Sauerwein rat. J Hepatol 1990;10:274-279.

HP. Tumour necrosis factor mimics the metabolic response to acute infec- 38. Heuman DM, Pandak WM, Hylemon PB, Vlahcevic ZR. Conjugates oftions in healthy humans. Am J Physiol 1991;261:E457-E465. ursodeoxycholic acid protect against cytotoxicity of more hydrophobic bile

13. Starnes HF Jr, Warren RS, Jeevanandam M, Gabrilove JL, Larchian W, salts: in vitro studies in rat hepatocytes and human erythrocytes. HEPA-Oettgen HF, Brennan MF. Tumour necrosis factor and the acute metabolic TOLOGY 1991;14:920-926.response to tissue injury in man. J Clin Invest 1988;82:1321-1325. 39. Ota S, Tsukahara H, Terano A, Hata Y, Hirahishi H, Mutoh H, Sugimoto

14. Morita M, Watanabe Y, Akaike T. Inflammatory cytokines up-regulate T. Protective effects of tauroursodeoxycholate against chenodeoxycholate-intercellular adhesion molecule-1 expression on primary cultured mouse induced damage to cultured gastric cells. Dig Dis Sci 1991;36:409-416.hepatocytes and T-lymphocyte adhesion. HEPATOLOGY 1994;19:426-431. 40. Igimi H, Carey MC. pH-solubility relations of chenodeoxycholic and urso-

15. Tracey KJ, Beutler B, Lowry SF, Merryweather J, Wolpe S, Milsark I, deoxycholic acids: physical-chemical basis for dissimilar solution and mem-Hariri RJ, et al. Shock and tissue injury induced by recombinant human brane phenomena. J Lipid Res 1980;21:72-90.cachectin. Science 1986;234:470-474. 41. Narumi S, Hamilton TA. Dexamethasone selectively regulates LPS-in-

16. Freudemberg MA, Freudemberg N, Galanos C. Time course of cellular duced gene expression in murine peritoneal macrophages. Immunophar-distribution of endotoxin in liver, lungs and kidneys of rats. Br J Exp macology 1990;19:93-101.Pathol 1982;63:56-65. 42. Shands JW, Chun PW. The dispersion of gram-negative lipopolysaccharide

by deoxycholate. J Biol Chem 1980;255:1221-1226.17. Greim H, Trulzsch D, Czygan P, Rudick J, Hutterer F, Schaffner F, Popper

AID Hepa 0013 / 5p1d$$$243 03-12-97 08:29:57 hpta WBS: Hepatology