236

Altered Kupffer cell function inbiliary obstructionRebecca M. Minter, MD,a,b Ming-Hui Fan, MD,b Jianmin Sun, BS,b Andreas Niederbichler, MD,b

Kyros Ipaktchi, MD,b Saman Arbabi, MD, MPH,b Mark R. Hemmila, MD,b Daniel G. Remick, MD,c

Stewart C. Wang, MD, PhD,b and Grace L. Su, MD,a,d Ann Arbor, Mich

Background. An altered Kupffer cell (KC) response is thought to be responsible for the characteristicphenotype observed after biliary obstruction: a phenotype marked by a defect in the hepatic reticulo-endothelial system and a hypersensitivity to endotoxin. Few studies, however, have directly examined KCfunction. We have sought to define the specific alterations in function and phenotype that occur in theKC after biliary obstruction.Methods. KCs were isolated from female C57BL/6 mice 4 days after a sham or common bile duct ligation(CBDL) operation. Phagocytosis, oxidative burst potential, and intracellular bacterial killing weremeasured as markers of reticuloendothelial system function. The KC response to endotoxin was assessedby measuring tumor necrosis factor alpha and interleukin 6 levels in the media after stimulation withlipopolysaccharide (LPS) or with LPS plus LPS-binding protein (LBP).Results. CBDL KCs demonstrated a significant increase in phagocytic ability and significantly decreasedbaseline oxidative stress, compared with Shams. The oxidative burst potential, however, was equivalentor higher for CBDL KCs. CBDL KCs also demonstrated increased numbers of viable intracellularbacteria after infection; however, it is unclear if this finding represents impaired intracellular bacterialkilling or increased phagocytosis of bacteria. With respect to the KC response to endotoxin, CBDL KCswere found to be less sensitive to the stimulatory effects of LPS alone but were exquisitely sensitive to theeffects of LBP. LBP levels were found to be significantly elevated in CBDL animals, and CBDL KCsdemonstrated a dose-dependent, exaggerated tumor necrosis factor alpha and interleukin 6 response toLPS administered with LBP.Conclusions. KC function is clearly altered after biliary obstruction. Phagocytic ability is actuallyincreased, although the ability of CBDL KCs to kill bacteria within the phagosome remains ill defined.CBDL KCs are exquisitely sensitive to the effects of LBP, and LBP levels are elevated after biliaryobstruction. LBP may be responsible for the increased proinflammatory response observed after endotoxinchallenge in animals with biliary obstruction. (Surgery 2005;138:236-45.)

From the Veterans Administration Ann Arbor Healthcare Systems,a Departments of Surgery,b Pathology,c andMedicine,d University of Michigan Medical School, Ann Arbor

DESPITE USE OF broad-spectrum antibiotic therapyand improvement in operative technique, patientswith cholestatic liver disease continue to experi-ence a high incidence of postoperative morbidityand mortality,1-3 with numerous clinical and ex-perimental studies demonstrating an increasedincidence of bacteremia, sepsis, and death inassociation with biliary obstruction.3-8 Animal stud-

Presented at the 66th Annual Meeting of the Society ofUniversity Surgeons, Nashville, Tennessee, February 9-12, 2005.

Reprint requests: Rebecca M. Minter, MD, University of Mich-igan, Dept of Surgery, 1500 E Medical Center Dr, TaubmanCenter TC2920H, Ann Arbor MI 48109-0331. E-mail: rminter@umich.edu.

0039-6060/$ - see front matter

� 2005 Mosby, Inc. All rights reserved.

doi:10.1016/j.surg.2005.04.001

SURGERY

ies have demonstrated a characteristic phenotypeafter common bile duct ligation (CBDL), markedby a defect in the hepatic reticuloendothelialsystem (RES),9-12 and a hypersensitivity to endo-toxin or bacterial challenge.13-15 The RES dysfunc-tion is thought to be related to a defect in the fixedhepatic macrophage population, Kupffer cells.After intraperitoneal Escherichia coli challenge inrodents previously having undergone CBDL, bac-terial clearance from the liver is reduced, com-pared with sham-operated animals, with aconcomitant increase in bacteremia.9,11,12,16 It isunclear whether this Kupffer cell defect is charac-terized by an impairment in phagocytic ability, aninability to kill intracellular bacteria, or a combi-nation of both because conflicting data exist.10,16

In addition to the RES dysfunction, a hypersen-sitivity to endotoxin challenge also exists, which

SurgeryVolume 138, Number 2

Minter et al 237

leads to an exaggerated proinflammatory responsein jaundiced animals. This response is marked byan increased production of tumor necrosis factoralpha (TNF-a), interleukin (IL)-1, and IL-6 inCBDL versus Sham animals,13-15,17,18 and is associ-ated with increased markers of end-organ injury(aspartate aminotransferase and creatinine) anddeath.15 Concurrently, blockade of Kupffer cellfunction in jaundiced animals undergoing endo-toxin challenge with gadolinium chloride has beenshown to improve survival and to suppress thesystemic proinflammatory response.17,19

Similar to findings in animal models, patientswith cholestatic liver disease also exhibit an alteredresponse to infectious stimuli. Several authors havereported an increased rate of postoperative sepsisand mortality in patients with biliary obstruc-tion.1,2,20,21 In addition to an elevation of bothpro- and anti-inflammatory mediators in theplasma of these patients, studies have also docu-mented a significant increase in circulating lipo-polysaccharide-binding protein (LBP) in patientswith biliary obstruction and cirrhosis from othercauses.22-24 In a prospectively analyzed cohort ofcirrhotic patients with ascites, an elevated plasmaLBP level was the only factor independently asso-ciated with the development of a severe bacterialinfection.23 Another group of investigators hasdocumented an increase in serum LBP levels inpatients with malignant obstructive jaundice,which decreases after internal biliary drainage.24

To date, the majority of studies evaluating thenature of the inflammatory response and the im-mune defect present in biliary obstruction havebeen in vivo experiments and clinical studies. Thealtered cytokine responses and increased suscepti-bility to infection documented in biliary obstructionhave been attributed to a Kupffer cell defect; how-ever, these conclusions have been largely derivedfrom in vivo studies. In the present report, we havesought to define the specific alterations in functionand phenotype that occur in the Kupffer cell afterbiliary obstruction. Our findings suggest that whileintrinsic Kupffer cell function is altered after CBDL,local factors present in vivo are likely responsible fordriving the phenotypic alterations we observed inthese animals.

MATERIALS AND METHODS

Animal preparation. Specific pathogen-free fe-male C57BL/6 mice (20-25 g) were obtained fromHarlan Laboratories (Indianapolis, Ind) and werehoused in the University of Michigan Unit forLaboratory Animal Medicine’s facility with unlim-

ited access to chow and water for the duration ofthe experiments. All animal studies were approvedby the Animal Care and Use Committee of theUniversity of Michigan. The laboratory adheres tothe ‘‘Guiding Principles of Laboratory AnimalCare,’’ as promulgated by the American Physiolog-ical Society.

On day zero, mice were anesthetized with 35mg/kg body weight of intraperitoneal sodiumpentobarbital, and a CBDL was performed thro-ugh a transverse right subcostal incision. Doubleligation of the common bile duct with a 4-0 silksuture was performed, with care taken not toinadvertently ligate or injure the hepatic artery orpancreatic duct. Abdominal wall closure wasachieved with interrupted 3-0 silk sutures, andthe skin was closed with staples. Sham animalsunderwent laparotomy and identical dissectionwithout bile duct ligation. On postoperative day4, Kupffer cells were isolated from both CBDL andSham animals as described below. Blood was alsoobtained from additional animals on day 4 formeasurement of bilirubin and transaminase levels.

Kupffer cell isolation and culture. Kupffer cellswere isolated from mice as previously de-scribed25,26 by using the modified methods de-scribed by Knook and Sleyster.27 Briefly, livers wereperfused retrograde through the inferior vena cavawith Gey’s balanced salt solution (GBSS; GibcoBRL, Gaithersburg, Md) followed by GBSS with0.2% pronase E. The liver was then excised andminced before incubation with GBSS/pronasesolution with continuous stirring at 37�C for 60minutes. DNase (0.8 lg/mL) was added to preventcell clumping. The liver slurry was filtered throughgauze mesh, washed with phosphate-buffered sa-line (PBS) with DNAse (0.8 lg/mL), and centri-fuged twice at 600g for 5 minutes each. Cells werethen further purified with the use of a discontin-uous Percoll gradient of 25% and 50% Percoll, asdescribed in detail by Pertoft and Smedsrod.28

Kupffer cells were enriched by differential adher-ence to tissue culture plates. Cells (2.0 or 4.0 3 105

cells/well) were plated on 96-well tissue cultureplates at 37�C for one-half hour in serum-freemedia. This media, and nonadherent cells, werethen removed and replaced with media containing5% fetal calf serum for incubation overnight. Thefollowing morning the cells were washed 3 timeswith serum-free media before experimentation. Allexperiments were then performed in serum-freemedia. Kupffer cell viability was assessed withtrypan blue, and their purity was confirmed bytheir ability to ingest latex beads. Cell viability wasgreater than 90% as assessed by trypan blue.

SurgeryAugust 2005

238 Minter et al

All experiments were then performed in the 96-well tissue culture plate the morning after isolationand culture. RES function was measured as out-lined below. Sensitivity to endotoxin and LBP wereassessed by stimulating the isolated Kupffer cellswith increasing doses of LPS (0 ng/mL-100 ng/mL) and LBP (0-1.3 lg/mL) in serum-free mediafor 6 hours. The media was then removed from thecells and assayed for TNF-a and IL-6 (see Cytokinemeasurement).

RES function. Phagocytosis: Vybrant PhagocytosisAssay Kit (V-6694) (Molecular Probes, Inc, Eugene,Ore) was used according to the manufacturer’sinstructions in a 96-well format. Briefly, isolatedKupffer cells (23105 cells/well) from Sham andCBDL animals were incubated with heat-killedfluorescein isothiocyanate (FITC)-E coli (20 bacteria:cell ratio). After 120 minutes, trypan blue wasadded to all wells to quench the signal fromexternally bound FITC. Internalized FITC-E coliwere then quantified with the use of a fluorescenceplate reader measuring fluorescence emission at520 nm with excitation at 480 nm.

Oxidative burst: OxyBurst Assay (O-13291 andF-2902) (Molecular Probes, Inc) was modified foruse in a 96-well format and used according to themanufacturer’s instructions. To measure extracel-lular oxidative products such as H2O2, we incu-bated isolated Kupffer cells (23105 cells/well)from Sham and CBDL animals with OxyBurstGreen H2HFF Reagent (O-13291) with and with-out LPS 10 ng/mL and LBP 0.3 lg/mL. The cellswere incubated with the OxyBurst Green H2HFFReagent and LPS/LBP or PBS control, and fluo-rescence was measured with the use of fluores-cence plate reader measuring fluorescenceemission at 520 nm with excitation at 480 nm.Intracellular oxidative burst was similarly mea-sured with the use of Fc OxyBurst Green assayreagent (F-2902) for fluorescent detection of Fcreceptor–mediated phagocytosis pathway. Briefly,dichlorodihydrofluorescein (H2DCF) is attachedto bovine serum albumin (BSA), then complexedto a rabbit polyclonal anti-BSA antibody. Afterbinding of this complex to a cell’s Fc receptor,the nonfluorescent immune complex is internal-ized and then oxidized to the fluorescent DCFwithin the phagosome. Intracellular fluorescencewas then measured at an emission of 520 nm andexcitation of 480 nm.

Intracellular bacterial killing: Intracellular bacte-rial killing was measured as previously described.29

Briefly, the morning after Kupffer cell isolation,the cells were incubated for 30 minutes withSalmonella typhi (ATCC 14028 in log-phase growth)

bacteria at a multiplicity of infection of 10 bacte-rium:1 Kupffer cell. The plate was then spun for10 minutes at 850g, and the cells were incubatedfor a total of 30 minutes with the bacteria at 37�C.Gentamicin (100 lg/mL) was then added to themedia to kill any remaining extracellular bacteriain the wells. The media was removed at 30 minutesand 6 hours after infection, and the cells werewashed with PBS to remove all remaining genta-micin. Cells were then lysed by the addition of50 lL of 10% Triton X-100 in sterile water. Tenminutes later, 450 lL of cold sterile PBS was addedto the wells. Samples were then collected in trip-licate from the wells and serially plated on Luriaagar plates. Bacterial colonies from the cell lysateswere counted 24 hours later.

Protein analysis. Cytokine measurement: MurineTNF-a (R&D Systems, Minneapolis, Minn) and IL-6 (Pharmingen, San Diego, Calif) levels weremeasured by sandwich enzyme-linked immunosor-bent assay (ELISA) with the use of commerciallyavailable reagents. Absorbance was read on aBiotek automated plate reader (Bio-Tek Instru-ments, Winooski, Vt) at 465 and 590 nm.

Markers of hepatic injury: Plasma bilirubin andtransaminase levels (aspartate aminotransferaseand alanine aminotransferase) were measuredaccording to the manufacturer’s instructions withthe use of colorimetric assays purchased fromSigma Diagnostics (St. Louis, Mo) before themerger of Sigma-Aldrich.

LBP preparation and measurement. We previ-ously produced a recombinant rat LBP (rLBP)protein using a baculovirus expression system thathas functional activity and cross reactivity withmouse cells.25 Hepatic levels of LBP were mea-sured by real-time polymerase chain reaction (RT-PCR) quantification. Briefly, total RNA was isolatedwith the TRIzol reagent per manufacturer’s in-structions (Life Technologies, Inc). Semiquantita-tive PCR was performed with the use of the iCyclerReal Time PCR machine (Biorad) and SYBR GreenOne Step RT-PCR kit (Qiagen). Primers weredesigned by using the reported sequences inGenbank for LBP. Primer sequences are as follows:5# ACTTCAAGATCAAGGCCGTGG and 3#CACC-GATGGAAGAGTCAGAGA. The specificity of theprimers was verified by analyzing the PCR producton ethidium bromide–stained agarose gel electro-phoresis as well as by direct sequencing of the PCRproduct. All RT-PCR runs included a melting curveanalysis to ensure the specificity of the productwith each PCR reaction.

The validity of the semiquantitative method isconfirmed by a consistent log linear correlation of

SurgeryVolume 138, Number 2

Minter et al 239

r2 > 0.96 between the starting template RNA con-centration and the threshold cycle.

Statistical analysis. Data are presented as themean ± SEM. For each experiment utilizingKupffer cells, cells were isolated and pooled from3 mice in each treatment group (CBDL and Sham)and repeated in triplicate. For the plasma studies,n = 7 animals per group. Student t test was used foranalyses when 2 different groups of samples werecompared. A 2-way analysis of variance was used toevaluate differences between treatment and doseor time, and a post hoc pairwise comparison of themean responses to the different treatment groupswas undertaken with a Tukey test. Statsisticalsignificance was considered to be achieved forP < .05.

RESULTS

Kupffer cell reticuloendothelial cell functionafter biliary obstruction. The majority of in vivostudies have demonstrated decreased bacterialclearance from the liver of CBDL animals aftersystemic delivery of bacteria; thus, hepatic RESdysfunction was presumed to exist.9,11,30 Thesedata are derived, however, and few investigatorshave directly examined the function of isolatedKupffer cells after biliary obstruction. We, there-fore, first investigated the reticuloendothelial cellfunction of isolated Kupffer cells 4 days afterCBDL; a time point at which we have verified thepresence of a significant cholestatic liver injury(Table). Phagocytosis and intracellular bacterialkilling were first examined as measures of reticu-loendothelial cell function.

Interestingly, Kupffer cells from CBDL animalsdemonstrated a significant increase in phagocyto-sis of heat-killed FITC-E coli, compared with ShamKupffer cells, P < .001 (Fig 1). This is in contrast tothe findings of the majority of in vivo studies.Intracellular bacterial killing, however, of live Sal-monella typhi (ATCC 14028) was impaired in theCBDL Kupffer cells at 30 minutes and 6 hours afterinfection, with increased numbers of live bacteriarecovered from lysed CBDL Kupffer cells as com-

Table. Markers of liver injury*

AST ALT Bilirubin

Sham 22.5 ± 1.5 54.7 ± 10.5 0.23 ± 0.03CBDL 279.1 ± 29.4y 494.0 ± 39.6y 5.0 ± 0.6yAST, Aspartate aminotransferase; ALT, alanine aminotransferase;CBDL, common bile duct ligation.*Significant liver injury is present 4 days after CBDL, as evidenced byincreased levels of bilirubin and serum transaminase levels, comparedwith Sham animals, yP < .001.

pared with Sham Kupffer cells, P < .005 (Figure 2).To determine if impaired oxidative burst afterphagocytosis of bacteria may explain the observeddecrease in intracellular bacterial killing, we nextexamined oxidative stress and oxidative burst po-tential of CBDL and Sham Kupffer cells. Extracel-lular and intracellular baseline oxidative stresswere first measured with the use of OxyBurst Assayreagents Green H2HFF reagent and Fc-H2DCF-BSA immune complexes. As shown in Figures 3, Aand B, baseline extracellular and intracellularoxidative stress measurements were significantlydecreased in the CBDL Kupffer cells, comparedwith Shams, P < .001 and P < .05, respectively.However, when these same cells were stimulatedwith 10 ng/mL of LPS and 0.8 lg/mL LBP,

Fig 1. Phagocytosis. CBDL Kupffer cells demonstrate in-creased phagocytic activity compared with Sham Kupffercells, as evidenced by the increased fluorescence presentin CBDL Kupffer cells after ingestion of FITC-E coli,*P < .001.

Fig 2. Intracellular bacterial killing. Increased numbersof viable bacteria were recovered from CBDL Kupffercells 30 minutes and 6 hours after infection with Salmo-nella typhi bacteria (ATCC 14028), *P < .005.

SurgeryAugust 2005

240 Minter et al

oxidative burst potential (total oxidative burstproducts after LPS/LBP stimulation – baselinelevels of these products) was either the same (Fig4, A) or actually increased for CBDL Kupffer cells,P < .001 (Fig 4, B).

These results demonstrate that isolated Kupffercells from CBDL animals actually possess an in-creased, rather than decreased, phagocytic ability.Thus, the in vivo findings of others may reflectlocal environmental influences in a cholestaticliver on bacterial phagocytosis, rather than anintrinsic defect in the Kupffer cell. The intracellu-lar bacterial killing and oxidative burst data aremore difficult to interpret. The increased numbersof live bacteria present in CBDL Kupffer cells 30minutes and 6 hours after infection may representimpaired bacterial killing, or may simply reflectincreased phagocytosis of bacteria by the CBDLKupffer cells. Likewise, the decreased baseline

Fig 3. Baseline oxidative stress. Baseline extracellular(A) and intracellular (B) oxidative stress, as measuredusing OxyBurst Green H2HFF Reagent and Fc-H2DCF-BSA immune complex, respectively, were significantlydecreased in the CBDL Kupffer cells, compared withShams, *P < .001 (A) and *P < .05 (B).

oxidative stress present in these cells is interesting;however, when they are stimulated, the cells areable to mount an increased intracellular oxidativeburst response.

Kupffer cell response to endotoxin after biliaryobstruction. In vivo studies have demonstrated anincreased proinflammatory response after endo-toxin challenge in CBDL animals, marked by anincrease in systemic TNF-a and IL-6 levels.13-15 Invivo blockade of TNF-a production with the ad-ministration of gadolinium chloride or TNF-aantibodies has been shown to lead to improvedsurvival in CBDL animals challenged with endo-toxin.17,19,31 The source of these proinflammatorycytokines has been presumed to be hepatic macro-phages, or Kupffer cells; however, examinationof actual Kupffer cell cytokine production hasbeen limited. We, therefore, sought to define the

Fig 4. Oxidative burst potential. CBDL Kupffer cell oxi-dative burst potential, as measured by the difference be-tween total oxidative burst products after LPS/LBPstimulation and baseline levels of these products, wasthe same when measured using the extracellular com-pound (A), or actually increased when measured withthe intracellular compound (B), *P < .001.

SurgeryVolume 138, Number 2

Minter et al 241

exact nature of the Kupffer cell response toendotoxin.

Isolated CBDL Kupffer cells stimulated withboth 1 ng/mL and 10 ng/mL concentrations ofLPS demonstrated significantly decreased levels ofproduction of both TNF-a (Fig 5, A) and IL-6 (Fig5, B), compared with Shams, P < .001. This findingwas quite surprising and directly contrary to thefindings present in the in vivo studies outlinedabove. However, if Kupffer cells were stimulatedwith LPS (10 ng/mL) in the presence of LBP (0.3lg/mL), we observed a marked and significantincrease in TNF-a (Fig 6, A) and IL-6 (Fig 6, B)production from CBDL Kupffer cells, comparedwith Shams, P < .001. These results suggest thatKupffer cells from CBDL animals are less sensitiveto the stimulatory effects of LPS alone than areSham Kupffer cells, but are exquisitely sensitive tothe effects of LBP. In fact, it appears that CBDL

Fig 5. Cytokine production after stimulation with endo-toxin alone. CBDL Kupffer cell production of TNF-a (A)and IL-6 (B) after stimulation with LPS at 1 ng/mL and10 ng/mL was significantly decreased, compared withSham Kupffer cell production, *P < .001. CBDL,Common bile duct ligation; TNF-a, tumor necrosisfactor alpha; IL-6, interleukin 6; LPS, lipopolysaccharide.

Kupffer cells require the presence of LBP tomount a significant response to LPS.

Kupffer cell response to LBP. Elevated circulat-ing LBP levels have been documented in patientswith biliary obstruction24 and cirrhosis;22,23 how-ever, the effect of LBP on Kupffer cell function inbiliary obstruction has not been directly examined.We first examined LBP levels present in the liverafter CBDL and found that LBP levels are signif-icantly higher in CBDL animals, compared withShams, P < .001 (Fig 7). We next performed anLBP dose-response study to determine how CBDLand Sham Kupffer cells responded to increasingdoses of LBP delivered with a constant dose of LPS(10 ng/mL). TNF-a and IL-6 levels were measuredin the supernatant. Interestingly, CBDL Kupffer

Fig 6. Cytokine production after stimulation with LPSand LBP. Kupffer cell stimulation with LPS in the pres-ence of LBP led to a significant increase in the produc-tion of TNF-a (6A) and IL-6 (6B) by CBDL Kupffer cellsas compared with Sham, *P < .001. The addition of LBPsignificantly altered the CBDL Kupffer cell response toLPS (10 ng/mL), while having only a minimal effecton the Sham Kupffer cell response. CBDL, Commonbile duct ligation; TNF-a, tumor necrosis factor alpha;IL-6, interleukin 6; LPS, lipopolysaccharide; LBP, LPS-binding protein.

SurgeryAugust 2005

242 Minter et al

cells demonstrated a dose-dependent response toLBP, with increasing doses of LBP leading toincreased TNF-a and IL-6 production. In contrast,Sham Kupffer cells demonstrated an initial rise incytokine production at lower doses of LBP, fol-lowed by inhibition of production at higher doses.CBDL Kupffer cell production of TNF-a and IL-6were also significantly higher than Sham produc-tion at all doses of LBP, P < .001 (Fig 8, A and B).These results show that LBP levels are elevatedafter CBDL and that Kupffer cells from theseanimals demonstrate a dose-dependent, exagger-ated proinflammatory response to LPS in thepresence of elevated levels of LBP. LBP may rep-resent a potential target for prevention of thesystemic inflammatory response syndrome–typeresponse that we observe in CBDL animals chal-lenged with endotoxin or gram-negative infection.

DISCUSSION

An immune defect is clearly present after biliaryobstruction, which results in increased septiccomplications, multisystem organ failure, anddeath after surgical interventions.1-3 The exactnature of this immune defect, however, remainsunclear. Numerous experimental studies havesuggested that impaired RES function is responsi-ble for the increased septic complications ob-served;9,11,16 however, the majority of theseconclusions have been derived on the basis ofquantification of bacteria present in the liver andother organs after infection, and of the rate ofclearance of bacteria or various particles from thesystemic circulation. Additionally, others have

Fig 7. Relative levels of hepatic LBP messenger RNA(mRNA). RT-PCR quantification of hepatic LBP mRNAlevels demonstrated a significant increase in LBPmRNA in the livers of CBDL animals, compared withSham counterparts, *P < .001. CBDL, Common bileduct ligation.

demonstrated that bacterial translocation is in-creased in biliary obstruction and also likely con-tributes to the increased infectious complicationsobserved in these patients.8,12,30 Lastly, to furthercomplicate matters, in vivo studies have alsoyielded conflicting results, with the number ofstudies demonstrating decreased bacterial trap-ping in the livers of CBDL rodents11,12,16 equalingthe number of those demonstrating increased orequivalent numbers of bacteria in the livers ofCBDL and Sham animals after infection.9,30,32

Those investigators demonstrating increased bac-terial content in the livers of CBDL animals aftersystemic delivery of bacteria have suggested thatthe RES defect is one of impaired bacterial killingafter bacterial trapping in the liver.

Fig 8. Effect of LBP on cytokine production. Increasingdoses of LBP delivered with low-dose LPS (10 ng/mL)led to increased production of TNF-a (A) and IL-6 (B)by CBDL Kupffer cells. In contrast, Sham Kupffer cellsdemonstrated only a modest increase in TNF-a (A)and IL-6 (B) production initially, but higher LBP dosesactually inhibited cytokine production, *P < .001.CBDL, Common bile duct ligation; TNF-a, tumor necro-sis factor alpha; IL-6, interleukin 6; LPS, lipopolysaccha-ride; LBP, LPS-binding protein.

SurgeryVolume 138, Number 2

Minter et al 243

The present study has directly examined theRES function of Kupffer cells. With regard tophagocytic activity, we have shown that CBDLKupffer cells actually demonstrate increased phag-ocytic ability, compared with Sham Kupffer cells.This is actually consistent with the few otherstudies that have examined phagocytosis directlyin Kupffer cells. Sheen-Chen et al33 demonstrateda significant increase in phagocytosis of heat-killedCandida albicans by isolated Kupffer cells fromCBDL rats, compared with Sham, and Tomiokaet al12 demonstrated an equivalent phagocyticindex (mean number of E coli/Kupffer cell)between isolated CBDL and Sham Kupffer cells.Lastly, although not a quantitative assessment ofphagocytosis, Ball et al10 examined electron micro-graphs of CBDL and Sham Kupffer cells, and foundno morphologic differences between the phago-cytic vesicles of CBDL and Sham Kupffer cells.

Phagocytic killing of bacteria by CBDL Kupffercells has also been hypothesized to be impaired.Our results also suggested the possibility of im-paired bacterial killing by CBDL Kupffer cells ofSalmonella typhi bacteria at 30 minutes and 6 hoursafter infection. However, another possible inter-pretation of the increased numbers of bacteriafound in the lysed CBDL Kupffer cells is that thesecells simply phagocytosed greater numbers of bac-teria than did the Sham Kupffer cells. Unfortu-nately, it is not possible to discern which of thesepossible explanations is correct with this metho-dology. It is, however, interesting to consider thatthe in vivo data, which has demonstrated increasednumbers of bacteria present in the liver aftersystemic infection,9,30,32,34 may reflect increasedbacterial trapping in the liver, rather than im-paired bacterial killing.

It has also been suggested that perhaps CBDLKupffer cells lack the ability to kill bacteria withinthe phagosome because of impaired oxidativeburst capabilities. In this study, the finding ofdecreased baseline oxidative stress is interestingand is consistent with the findings of Sung et al,35

who demonstrated a significant reduction in super-oxide production by CBDL Kupffer cells, com-pared with Sham Kupffer cells. However, whenthese same cells are stimulated with endotoxin andLBP, they mount an equivalent or increased oxi-dative burst response. Thus, it is difficult to inter-pret the importance of these baseline differencesin oxidative stress. Unfortunately, with the cur-rently available studies and the methodology used,it is not possible to definitively determine if anintrinsic Kupffer cell RES defect is present inbiliary obstruction. The importance of local envi-

ronmental factors must also be considered, andfurther studies will be needed to truly understandthe exact nature of the Kupffer cell RES defectpresent in biliary obstruction.

In addition to the RES defect described inCBDL Kupffer cells, a hypersensitivity to endotoxinhas also been described, with production of largeamounts of TNF-a observed after LPS administra-tion to CBDL animals.13,14,17,19 In contrast to thevaried results observed in the RES Kupffer cellstudies, the response to endotoxin has been con-stant and highly reproducible. The source of thehypersecretion of TNF-a after endotoxin challengein CBDL animals is thought to be the Kupffer cell,the primary source of cytokine production in theliver. This hypersensitivity to endotoxin has beenlinked to increased mortality; blockade of Kupffercell function with gadolinium chloride13,17,19 orantibody inhibition of TNF-a31 in CBDL animalsreceiving endotoxin has resulted in improvedsurvival. This marked proinflammatory responseto endotoxin in the setting of biliary obstructioncarries significant clinical relevance because it ishypothesized that circulating levels of endotoxinare increased in patients with biliary obstructionrelated to increased bacterial translocation andmanipulation of the biliary tree in patients withextrahepatic biliary obstruction.17 To date, only afew studies have actually examined cytokine pro-duction by isolated CBDL Kupffer cells after stim-ulation with endotoxin.

Interestingly, the present study and the 2 priorstudies that measured TNF-a production by iso-lated CBDL Kupffer cells stimulated with endo-toxin have demonstrated equivalent or decreasedproduction, compared with Sham Kupffercells.32,33 Abe et al32 reported a significant increasein TNF-a production by stimulated CBDL perito-neal macrophages, compared with Sham perito-neal macrophages, with a concomitant decrease inproduction by CBDL Kupffer cells versus ShamKupffer cells. Because gadolinium chloride is acrude antagonist of all macrophage populations,these findings are not inconsistent with the find-ings of the in vivo studies, it may just be that thesource of the TNF-a in CBDL animals is actually amacrophage population other than Kupffer cells.In addition to TNF-a, IL-6 production was alsomeasured in the current study and mirrored thatof TNF-a, with a significant decrease in productionby endotoxin-stimulated CBDL Kupffer cells, com-pared with Sham Kupffer cells. These findings,coupled with the observation in a recent clinicalreport of elevated LBP levels in the plasmaof patients with biliary obstruction,24 led us to

SurgeryAugust 2005

244 Minter et al

explore the possible role of LBP in the CBDLKupffer cell response to endotoxin.

As shown in this study, LBP levels are signifi-cantly elevated after biliary obstruction, and CBDLKupffer cells are exquisitely sensitive to the aug-menting effects of LBP to low-dose endotoxin.Additionally, these cells appear to have a differen-tial response to LBP, compared with Sham Kupffercells, with increasing doses of LBP leading toincreased TNF-a and IL-6 production by CBDLKupffer cells, while higher doses of LBP actuallyinhibit Sham Kupffer cell production of thesesame cytokines. These data are very interesting,particularly when considered with the recentreports of Albillos and colleagues23 showing thatelevated serum LBP levels predict the developmentof severe bacterial infections in cirrhotic patients.Although additional studies are clearly needed,LBP may represent a potential target for preven-tion of the systemic inflammatory response syn-drome we observe in animals and patients withbiliary obstruction, who are challenged with gram-negative infection or endotoxin stimulation.

CONCLUSION

We have shown that Kupffer cell function isclearly altered after biliary obstruction. This alter-ation appears to involve variation in the RESfunction; however, more work is needed to definethe exact nature and implications of these alter-ations with respect to the development of infec-tious complications. Kupffer cells also have avariable response to endotoxin and LBP afterbiliary obstruction. This study suggests that theincreased TNF-a response observed in vivo inCBDL animals after endotoxin challenge mayactually be related to elevated LBP levels augment-ing their response, rather than to a direct effect ofLPS on the CBDL Kupffer cells. Future studies areneeded to explore this possibility further.

REFERENCES

1. Nomura T, Shirai Y, Hatakeyama K. Impact of bactibilia onthe development of postoperative abdominal septic com-plications in patients with malignant biliary obstruction. IntSurg 1999;84:204-8.

2. Thompson JN, Edwards WH, Winearls CG, Blenkharn JI,Benjamin IS, Blumgart LH. Renal impairment followingbiliary tract surgery. Br J Surg 1987;74:843-7.

3. Pitt HA, Cameron JL, Postier RG, Gadacz TR. Factorsaffecting mortality in biliary tract surgery. Am J Surg 1981;141:66-72.

4. Dixon JM, Armstrong CP, Duffy SW, Davies GC. Factorsaffecting morbidity and mortality after surgery for ob-structive jaundice: a review of 373 patients. Gut 1983;24:845-52.

5. Blamey SL, Fearon KC, Gilmour WH, Osborne DH, CarterDC. Prediction of risk in biliary surgery. Br J Surg 1983;70:535-8.

6. Su CH, P’Eng FK, Lui WY. Factors affecting morbidity andmortality in biliary tract surgery.World J Surg 1992;16:536-40.

7. Tanaka N, Ryden S, Bergqvist L, Christensen P, Bengmark S.Reticulo-endothelial function in rats with obstructive jaun-dice. Br J Surg 1985;72:946.

8. Deitch EA, Sittig K, Li M, Berg R, Specian RD. Obstructivejaundice promotes bacterial translocation from the gut. AmJ Surg 1990;159:79-84.

9. Ding JW, Andersson R, Norgren L, Stenram U, Bengmark S.The influence of biliary obstruction and sepsis on reticulo-endothelial function in rats. Eur J Surg 1992;158:157-64.

10. Ball SK, Grogan JB, Collier BJ, Scott-Conner CE. Bacterialphagocytosis in obstructive jaundice. A microbiologic andelectron microscopic analysis. Am Surg 1991;57:67-72.

11. Katz S, Grosfeld JL, Gross K, et al. Impaired bacterialclearance and trapping in obstructive jaundice. Ann Surg1984;199:14-20.

12. Tomioka M, Iinuma H, Okinaga K. Impaired Kupffer cellfunction and effect of immunotherapy in obstructive jaun-dice. J Surg Res 2000;92:276-82.

13. Kennedy JA, Clements WD, Kirk SJ, et al. Characterizationof the Kupffer cell response to exogenous endotoxin in arodent model of obstructive jaundice. Br J Surg 1999;86:628-33.

14. O’Neil S, Hunt J, Filkins J, Gamelli R. Obstructive jaundicein rats results in exaggerated hepatic production of tumornecrosis factor-alpha and systemic and tissue tumor necrosisfactor-alpha levels after endotoxin. Surgery 1997;122:281-6;discussion 286-7.

15. Harry D, Anand R, Holt S, et al. Increased sensitivity toendotoxemia in the bile duct-ligated cirrhotic Rat. Hepa-tology 1999;30:1198-205.

16. Hoshino S, Sun Z, Uchikura K, et al. Biliary obstruction re-duces hepatic killing and phagocytic clearance of circulatingmicroorganisms in rats. J Gastrointest Surg 2003;7:497-506.

17. Kennedy JA, Lewis H, Clements WD, et al. Kupffer cellblockade, tumour necrosis factor secretion and survivalfollowing endotoxin challenge in experimental biliaryobstruction. Br J Surg 1999;86:1410-4.

18. Fujiwara Y, ShimadaM, Yamashita Y, et al. Cytokine character-istics of jaundice in mouse liver. Cytokine 2001;13:188-91.

19. Lazar G Jr, Paszt A, Kaszaki J, et al. Kupffer cell phagocytosisblockade decreases morbidity in endotoxemic rats withobstructive jaundice. Inflamm Res 2002;51:511-8.

20. Armstrong CP, Dixon JM, Taylor TV, Davies GC. Surgicalexperience of deeply jaundiced patients with bile ductobstruction. Br J Surg 1984;71:234-8.

21. Kuzu MA, Kale IT, Col C, Tekeli A, Tanik A, Koksoy C.Obstructive jaundice promotes bacterial translocation inhumans. Hepatogastroenterology 1999;46:2159-64.

22. Albillos A, de la Hera A, Gonzalez M, et al. Increasedlipopolysaccharide binding protein in cirrhotic patientswith marked immune and hemodynamic derangement.Hepatology 2003;37:208-17.

23. Albillos A, de-la-Hera A, Alvarez-Mon M. Serum lipopoly-saccharide-binding protein prediction of severe bacterialinfection in cirrhotic patients with ascites. Lancet 2004;363:1608-10.

24. Kimmings AN, van Deventer SJ, Obertop H, Rauws EA,Huibregtse K, Gouma DJ. Endotoxin, cytokines, and endo-toxin binding proteins in obstructive jaundice and afterpreoperative biliary drainage. Gut 2000;46:725-31.

SurgeryVolume 138, Number 2

Minter et al 245

25. Su GL, Klein RD, Aminlari A, et al. Kupffer cell activation bylipopolysaccharide in rats: role for lipopolysaccharide bind-ing protein and toll-like receptor 4. Hepatology 2000;31:932-6.

26. Su GL, Goyert SM, Fan MH, et al. Activation of human andmouse Kupffer cells by lipopolysaccharide is mediated byCD14. Am J Physiol Gastrointest Liver Physiol 2002;283:G640-5.

27. Knook DL, Sleyster EC. Separation of Kupffer and endo-thelial cells of the rat liver by centrifugal elutriation. ExpCell Res 1976;99:444-9.

28. Pertoft H, Smedsrod B. Separation and characterization ofliver cells. New York: Academic Press; 1987.

29. Weiss DS, Raupach B, Takeda K, Akira S, Zychlinsky A. Toll-like receptors are temporally involved in host defense.J Immunol 2004;172:4463-9.

30. Ding JW, Andersson R, Soltesz V, Willen R, Bengmark S.Obstructive jaundice impairs reticuloendothelial function

and promotes bacterial translocation in the rat. J Surg Res1994;57:238-45.

31. Beierle EA, Vauthey JN, Moldawer LL, Copeland EM 3rd.Hepatic tumor necrosis factor-alpha production and distantorgan dysfunction in a murine model of obstructive jaun-dice. Am J Surg 1996;171:202-6.

32. Abe T, Arai T, Ogawa A, et al. Kupffer cell-derivedinterleukin 10 is responsible for impaired bacterial clear-ance in bile duct-ligated mice. Hepatology 2004;40:414-23.

33. Sheen-Chen SM, Chau P, Harris HW. Obstructive jaundicealters Kupffer cell function independent of bacterial trans-location. J Surg Res 1998;80:205-9.

34. Scott-Conner CE, Grogan JB. The pathophysiology of biliaryobstruction and its effect on phagocytic and immunefunction. J Surg Res 1994;57:316-36.

35. Sung JJ, Go MY. Reversible Kupffer cell suppression inbiliary obstruction is caused by hydrophobic bile acids.J Hepatol 1999;30:413-8.